-L

HARVARD UNIVERSITY

Wm

LIBRARY

OF THE

Museum of Comparative Zoology

s.

■1 J a-^jl — iisi

MEMOIRS

OF THE

Mus. mu?. im.

mumm

MUSEUM OF COMPARATIVE ZOOLOGY

AT

HARVARD COLLEGE.

VOL. XLII.

CAMBRIDGE, MASS., U. S. A.
PRINTED FOR THE MUSEUM.
1915.

/■

MEMOIRS

OF THE

MUSEUM OF COMPARATIVE ZOOLOGY

AT

HARVARD COLLEGE.

VOL. XLII.

CAMBRIDGE, MASS., U. S. A.
PRINTED FOR THE MUSEUM.
1915.

The Cosmos Press:
Edward W. Wheeler, Cambridge, U. S. A.

/IDemoics of tbe /iDuseum ot Comparative Zo'61oq\]

AT HARVARD COLLEGE.
Vol. XLTT.

REPORTS ON THE SCIENTIFIC RESULTS OF THE EXPEDITION TO THE
EASTERN TROPICAL PACIFIC, IN CHARGE OF ALEXANDER AGASSIZ,
BY THE U. S. FISH COMMISSION STEAMER "ALBATROSS," FROM
OCTOBER, 19(M, TO MARCH, 1905, LIEUT. COMMANDER L. M. GARRETT,
U. S. N., COMMANDING, AND OF OTHER EXPEDITIONS OF THE "ALBA-
TROSS," 1891-1899.

XXIX.

THE SPONGES.

3. HEXACTINELLIDA.

By ROBERT VON LENDENFELD.

WITH ONE HUNDRED AND NINE PLATES.

TEXT.

(Published by permission of Hugh M. Smith, U. S. Commissioner of Fish and Fisheries].

CAMBRIDGE, U.S.A.:

printeD for tbe HDuseum.

June, 1915.

m. GOMP. Z09L

umm

HKIVESSin

^0 ixrJacc

/TU^U -

CONTENTS.

REPORTS on the Scientific Results of the Expedition to the Eastern Tropical Pacific,
in charge of Alexander Agassiz, by the U. S. Fish Commission Steamer "Albatross,"
from October, 1904, to March, 1905, Lieut. Commander L. M. Garrett, U. S. N.,
Commanding, and of other expeditions of the "Albatross," 1S91-1S99. XXIX. The
Sponges. 3. Hexactinellida. By Robert von Lendenfeld. 397 pp. 109 Plates.
June, 1915.

TABLE OF CONTENTS.

Page.

I. INTRODUCTION 11

II. METHODS 11

1. Soft parts 11

2. Skeleton 12

3. Spicules 12

4. Graphic representation 13

5. Measuring 13

6. Biometry 16

7. Nomenclature of the spicules 17

III. DESCRIPTION OF THE SPECIES COLLECTED BY THE ALBATROSS . 19

Hexactinellida 19

Hexasterophora 19

Edplectellidab 20

Edplectellinae 20

Holascus 20

H. edwardsii (Plate 18, figs. 15-26; Plate 19, figs. 1-24; Plate 20, figs. 1-20) 21

Holascella 29

H. taraxacum (Plate 21, figs. 1-13; Plate 22, figs. 1-41; Plate 23, figs. 1-3) 29

H. ancorata (Plate 23, figs. 4-25; Plate 24, figs. 1-9) .... 37

H. euonyx (Plate 24, figs. 10-17; Plate 25, figs. 1-24) .... 44

Caolophacidae 47

Caulophacus 47

C. schulzei (Plate 7, figs. 20-31; Plate 8, figs. 1-29; Plate 9, figs. 1-33;

Plate 10, figs. 1-29; Plate 11, figs. 1-17) ... 48

Caolophacella 64

C. tenuis (Plate 12, figs. 1-19) 64

Calycosilva 67

C. cantharellus helLx (Plate 1, figs. 1-8, 20-24; Plate 2, figs. 3, 7-13, 15;
Plate 3, figs. 1-5, 8-30; Plate 4, figs. 23, 24; Plate 5,
figs. 1, 2, 4, 5, 7-9, 11-15, 18-20; Plate 6, figs. 5-21, 24-

34; Plate 7, figs. 1-10, 12-14, 16, 17) ... 67
C. cantharellus simplex (Plate 1, figs. 9-19, 25-29; Plate 2, figs. 1, 2, 4-6, 14,
16; PlateS, figs. 6, 7; Plate 4, figs. 21, 22; Plate 5, figs.
3,6,10,16,17; Plate 6, figs. 1-4, 22, 23; Plate 7,

figs. 11, 15, IS) 68

C. cantharellus megonychia (Plate 4, figs. 1-20; Plate 5, fig. 21; Plate 7,

fig. 19) 68

Doubtful caulophacid (Plate 32, figs. 10-12) 92

rossellidae .............. 93

Rossellinae ............. 93

Bathydorus 94

B. laevis 94

B. laevis spinosissimus (Plate 14, figs. 1-32; Plate 15, figs. 1-22; Plate 16,

figs. 1-24) 94

Landginellinae 103

Lanugonychia 103

L. flabellum (Plate 12, figs. 20-34; Plate 13, figs. 1-28) .... 103

Acanthascinae . 112

8

CONTENTS.

Staurocalyptus

S. hamatus (Plate 16, figs. 25-43; Plate 17, figs. 1-25; Plate 18, fig-s. 1-14)

EUBETIDAE .......

Fakrba .......

F. occa scutclki (Plate 25, figs. 25-29; Plate 26, figs. 1-21; Plate 27, figs
1-17) .

F. sp.? (Plate 32, figs. 1-3) .

F. sp.?

EURETE .......

E. ei-ectum (Plate 30, figs. 1-17; Plate 31, figs. 1-28)

E. erectum A-D .....

E. spinosum (Plate 29, figs. 1-26)

Euretid from Station 4641 (Plate 106, figs. 1-3)

Euretid (?) from Station 4651 (Plate 32, figs. 4-6)

Euretid (?) from Station 4685

Euretitl (?) from Station 4695

COSCINOPORIDAB

Chonelasma

C. sp. (Plate 32, figs. 7-9) .
Tretocalycidae ......

Hexactinella ......

H. monticularis (Plate 28, fig.s. 1-28) .
H. sp. mdet. (Plate 32, figs. 13-15)
Ajmphidiscophora ......

Hyaloivematidae ......

Hyalonema

Hyalonema

H. obtusum gi-acilis (Plate 33, figs. 1-24; Plate 34, figs

1-37; Plate 36, figs. 1-45; Plate 37, figs. 1-22; Plate
38, figs. 1-S; Plate 39, figs. 1-10
H. obtusum robusta (Plate 39, figs. 11-41 ; Plate 40, figs. 1-22) .
H. agassizi (Plate 41, figs. 1-14; Plate 42, figs. 1-59; Plate 43

Plate 44, figs. 1-30; Plate 45, figs. 1-64; Plate 46
1-16; Plate 47, figs. 1-13)
H. polycaulum (Plate 53, figs. 1-17; Plate 54, figs. 1-45)
H. placuna (Plate 63, figs, 29-51; Plate 64, figs. 1-19; Plate 65, figs. 1-23

Plate 66, figs. 1-5)

H. sp. from Station 4656 (Plate 68, figs. 26-33; Plate 69, figs. 1-5)

H. tenuifusum (Plate 67, figs. 1-26; Plate 68, figs. 1-25)

H. tylostylum (Plate 69, figs. 6-25; Plate 70, figs. 1-10) .

H. grandancora (Plate 78, figs. 16-45; Plate 79, figs. 1-26)

H. sp. from Station 3684 (Plate 80, figs. 1-16) ....

Leptonema

H. campanula (Plate 81, figs. 1-26) ......

Prionema

H. agujanum tenuis A (Plate 72, figs. 17-21, 23-25, 27; Plate 73, figs. 1-6;

Plate 74, figs. 1^, 8; Plate 75, figs. 1-13, 15, 17, 19-27,

29-37; Plate 76, figs. 1-7, 11, 12, 15-36)

H. agujanum tenuis B (Plato 72, figs. 10, 22, 26; Plate 73, fig. 7; Plate 74,

figs. 6, 7, 9; Plate 75, figs. 14, 16, 18, 28; Plate 76, figs.

8-10, 13, 14)

H. agujanum lata (Plate 77, figs. 1-10; Plate 78, figs. 1-15)

H. azuerone (Plate 56, fig. 1; Plate 57, figs. 1-23; Plate 58, figs. 1-22)

H. spinosum (Plate 48, figs. 1-31; Plate 49, figs. 1-23; Plate 50, figs. 1-5)

H. crassum (Plate 106, figs. 4-37; Plate 107, figs. 1-20; Plate 108, figs. 1-17)

H. pinulifusum (Plate 70, figs. 11-24; Plate 71, figs. 1-11; Plate 72, figs.

1-15)

H. fimbriatum (Plate 59, figs. 1-6; Plate 60, figs. 1-34; Plate 61, figs. 1-11;
Plate 62, figs. 1-45; Plate 63, figs. 1-28)

1-19; Plate 35, fig;

1-7

Page.
112
112
119
119

119
125
125
126
126
135
135
139
140
141
141
142
142
142
143
143
144
149
150
150
150
152

1.53
153

172
201

207
220
222
229
235
243
245
245
250

251

251
251

266
273

278

284
294

CONTENTS.

OONEMA ..............

H. bianchoriituin pinuliiia (Plate 82, figs. 1-34; Plate 83, figs. 1-68; Plate

84, figs. 1-32; Plate 85, figs. 1-8)
H. henshawi (Plate 97, figs. 1-36; Plate 98, figs. 1-7)
H. crassipinulum (Plate 92, figs. 1-23; Plate 93, figs. 1-10; Plato 94, figs

1-33)

H. deiLsum (Plate 94, figs. 34-42; Plate 95, figs. 1-20;

H. sequoia (Plate 85, figs. 9-21; Plate 86, figs. 1-36

Plate 88, figs. 1-13; Plate 89, figs

1-10; Plate 91, figs. 1-6)

Phialonema ......

H. brevancora (Plate 55, figs. 1-37)
H. pateriferum (Plato 50, figs. 6-15; Plate 51
Skianema .......

H. acquatorialc (Plate 99, figs. 1-37; Plate

1-3) .
H. umbraculuin (Plate 101, figs. 4-17; Plate 102,
1-36) .
Thallonema ......

H. gemiiiatum (Plate 103, figs. 37-62; Plate
1-14) .
IV. LIST OF STATIONS

1-28
100, figs

104,

figs
figs-

Plate 96, figs.

Plate 87, figs

1-36; Plate 90

1-14)

1-7

, figs

Plate 52, figs
-12; Plato 101
1-8; Plate 103

1-14; Plate 105

1-29)

• figs

figs

figs

Page.

307

308
327

332
341

349
358
358
362
376

377

383
390

390
397

HEXACTINELLIDA.

I. INTRODUCTION.

In this Report the HexactineUida collected during the Albatross cruises
of 1899-1900 and 1904-1905 in the Tropical Pacific under the direction of
Alexander Agassiz are described.

Mr. Agassiz's hberality has enabled me to employ methods of research and
graphic representation not hitherto used and to describe the material very
fully.

• II. METHODS.

1. The soft parts.

The deep-sea HexactineUida which come into the hands of specialists are
generally in such a condition that very Uttle can be made out, by the ordinary
methods of sectioning and staining, of their very tender soft parts. Tliis is
due to their mixing with the deep-sea ooze during the passage of the dredge
over the bottom and to the pull and pressure acting on them in the long haul
to the surface. After many experiments I finally found the following method
best suited to this kind of material: — a piece of the specimen, I to 1 cm. in
diameter, with intact surface is imbedded in paraffin and cut into thick radial
sections. These are not stuck on the sUde but placed free, first in xylol, then in
alcohol, where much of the deep-sea ooze, which has got into the sponge during
capture, and many of the fragments of spicules spUntered in cutting fall out of
the section, so that it becomes fairly clean. These loose sections are then

12 METHODS.

passed into absolute alcohol, in which magenta or another aniline dye soluble
in alcohol is dissolved. In this solution the sections very rapidly become well
stained. They are not washed after this, but immediately transferred into
xylol, in which the magenta, azur, etc., are insoluble, and then mounted in
balsam. By this method the canals and the flagellate chambers can be made
out in many a perfectly hopeless looking specimen.

2. The skeleton.

For the study of the arrangement of the spicules, and of the skeleton in
general, thick radial sections made in the manner described above, but not
stained, gave the best results. Such sections even of hard forms with a con-
tinuous skeleton-net, like the Euretidae, can be cut without difficulty.

3. The spicules.

My method of fractional sedimentation with final centrifugation has
also been employed in the examination of the Hexactinellida. On account of
the great amount of foreign siliceous material (skeletons of Radiolaria, etc.)
in many of the specimens these spicule-preparations are, however, often not
so clean as one would wish. To obtain clean preparations of the larger spicules
I made a heap of spicules of sediment (I) by boiling a piece of the sponge in nitric
acid, allowing it to settle a short time and drying in the usual manner. From
this I, or rather my wife, who in time grew exceedingly expert, picked out under
the microscope the spicules wanted. A fine needle, the point of which was
rendered sticky with Schellibaum's inixture of collodion and clove-oil, was used
in tills work. These spicules were then regularly arranged on a slide, also
covered with a thin layer of Schellibaum's mixture. To this they adhere, and
can be inunersed in balsam and covered with a cover-glass without becoming
disarranged.

The preparations of the smaller spicules of sediment (II), etc., and the cen-
trifugated ones were heated till all the chloroform used for dissolving the balsam
had evaporated and only the previously boiled balsam, which is quite hard at
ordinary temperatures, was left. They were then, whilst cooling, pressed between
the leaves of a book. In this way preparations are obtained which are much
clearer than unpressed ones, and which can be examined with the highest powers
much more conveniently.

METHODS. 13

4. Graphic representation.

All the figures on the plates in tliis Report are photographs. These photo-
graphs were taken partly with ordinary, and partly with ultraviolet (wave
length 280 ^i) light ; with the same apparatus and in the same way as those illus-
trating my Report on the Geodidae (Mem. M. C. Z., 1910, 41, p. 12 ff.) where
the photographic methods employed are described. I found it very difficult
to obtain good photographs of fioricomes and other small microhexaster-f orms ;
chiefly because it is hardly possible to get good clean preparations of intact
spicules of this kind, either in balsam (for photography with ordinary light),
or in chloral-hydrate glycerine (for photography with ultraviolet light). My
hexaster-photographs are consequently not nearly so attractive as the drawings
of them in the papers by other authors — but they accurately represent what
one actually sees.

To facihtate comparison the figures representing the systematically most
important spicules of the same kind in the different species are given in the same
magnification throughout. To these commensurate figures others, in other mag-
nifications, are added where necessary. The uniform magnifications selected
for the commensurate figures are such that the smallest forms observed come
out just large enough to allow their main characters to be distinctly made out.
They are : — for the pinules .300 ; for the microhexactines, the hexasters and
their derivates, and the amjihidiscs .500. The photographs of parts of the
spictfles and of whole small spicules showing minute details were all taken with
ultraviolet light and are magnified 2000.

5. Measuring.

Every exact description m\ist be based on measured dimensions. The
dimensions of organisms and their parts are inconstant and vary in various ways.
To obtain dimensional data sufficient for use as premises for a systematic or
any other biological conclusion it is therefore necessary to ascertain the range
and biometrical character of the variation in the extension in space of the parts.
In the case of such organisms as the Hexactinellida the smaller spicules at
least should be studied biometrically. They can be most easily and accurately
measured and are considered by all authors as the most important part from a
systematic (phylogenetic) point of view. It would have been quite impossible,

14 METHODS.

within a reasonable space of time, to take the many thousand measurements
necessary for this by means of the methods hitherto employed.

I therefore cast about for a better method and finally worked out a new
micromeasuring apparatus, which Mr. Agassiz's Uberality enabled me to set
up. The plan of this apparatus (Fig. 1) has proved most useful. The greater
part of it is also represented on Plate 109. The light of a powerful, self-regulat-
ing constant-current arc-light (Fig. la) passes a system of lenses and cooler
(Fig. lb) and enters a microscope (Fig. Ic) with the optical axis placed horizon-
tally. A movable mirror 1.5 X 1.5 m. in size (Fig. Id) is so placed in front
of the microscope that the image produced is reflected on to a vertical glass-
plate (Fig. le) frosted on the side turned towards the mirror, and measuring
2 X 2 m. Lamp and microscope are so placed that the latter stands at the
side of and close to the frosted glass-plate. The observer sitting in front of
the latter can comfortably work both the screws moving the sHde to right
and left and up and down (Fig. Ig), and that focussing the microscope (Fig. Ih).
The horizontal optical axis of the microscope is oblique (not vertical) both to
the mirror and the frosted glass. The mirror, however, is placed so that the
axis of the cone of Ught reflected from it abuts vertically on the frosted glass-
plate. This arrangement insures the image, thrown on to and visible on the
frosted glass, being perfectly true, and not in any way distorted. By means
of the screws moving the slide, everything on it can easily be passed in review.
When a spicule, or anything else that is to be measured, comes into view, it
is focussed and measured.

When working with this apparatus, I placed lamp, microscope, mirror, and
frosted glass always in the same position; and every time I commenced I tested
the correctness of the position of the parts by projecting and measuring the
micrometer-slide. For each combination of objectives and eyepieces used I
made a special scale which was drawn on a ribbon of tracing-cloth. These
ribbons (tapes) were fixed to canes, like strings to bows (Fig. 2). It is easy
with these bow-string tapes to measure rapidly the distance between any two
points in a plane vertical to the optical axis of the microscope.

The observer dictates the dimensions thus measured, and anything else
notable he may observe. His assistant sits behind him at a table (Fig. li)
with a shaded (Fig. Ik) light (Fig. 11). It is possible, if the preparation is a
good one, to write down the dimensions at the rate of six to ten per minute.

This method is not only convenient and rapid, but also exceedingly accurate.
The measurements taken with it when using high powers are exact to 0.1 m-

oCZ!r

Fig.l

Fig. 2
Figs. 1, 2. — Projection measuring apparatus. 20 : 1.

16 METHODS.

The exactness of these measurements is indeed so great, that I detected, in
working with this apparatus, certain sUght errors in the micrometers employed,
which had been obtained from a firm of excellent standing. These errors this
firm itself found after having, at my request, reexamined the micrometers.

6. Biometry.

To utilize the measurements taken biometrically, all those of the same
dimension in different individuals must be arranged in groups of suitable extent.
In each group all the measurements lying between certain limits are placed,
and the number of dimensions lying between these limits ascertained. These
numbers are. then plotted on equidistant ordinates in a graph and the points
thus obtained connected by a line. This line is the biometrical frequency-curve
of the dimension studied.

In the method generally employed the groups of dimensions represented
by the ordinates of the graph are equal in range. That is to say, the mean
dimensions of the groups to which the ordinates correspond form an arithmetrical
progression hke 1, 2, 3, 4, .... n. This method involves a systematic error wliich
makes the resulting biometrical curve wrong and misleading. When a dimen-
sion examined varies between limits small in comparison to itself, that error
is slight and generally overlooked. When, however, as is the case in the amphi-
discs of the HexactinelUda for instance, the dimensions examined vary so much
that the largest may be twenty-five times as great as the smallest, the error
leads to results so glaringly wrong that it is noticed at once. This error is caused
by the equality of the extent of the successive groups and by their mean dimen-
sions, which are represented by the ordinates of the graph, forming an arith-
metrical progression. For it is obvious that a difference, say of 10 /x, in the length
of objects only 10-20 /i long must be of far greater biological importance than
the same difference of 10 fi in the length of objects 500-510 n long. To avoid
this error I divide the measurements taken into groups of uniformly increasing
range. The increment selected is such that the extent of each successive group
is 10 % greater than the extent of the preceding group, so that the ordinates,
which are also placed by me at equal distances in the graph, represent a series
of mean-dimensions of groups which form the geometrical progression 1.1, l.P,
1.1\ 1.1^ 1.1°.

The biometrical curves obtained in this manner express identically the
character of dimensional variation of all the individuals compared, however

METHODS. 17

large or small they may be, and are therefore biometrically much more correct
than those obtained by the method generally in use. In this Report such fre-
quency-curves, which are biologically more correct, are extensively made use of.

7. Nomenclature of the spicules.

I use the same names for the spicules as those employed by F. E. Schulze
and other authors. The few new names given are explained where they first
occur. I find F. E. Schulze's di\asion of the amphidiscs into the three groups
macramphidiscs, mesamphidiscs, and micramphidiscs by no means universally
applicable and have divided the different kinds of amphidiscs found in each
species according to their morphological and biometrical characters, independ-
ently of and without regard to the arrangements of them in other species. I
have, however, retained Schulze's names, because only a very small fraction
indeed of the Amphidiscophora actually growing on the sea-bottom are
known, and it would be premature to propose a new general arrangement of
these spicules, and to replace Schulze's names by others.

III. DESCRIPTION OF THE SPECIES COLLECTED IN THE PACIFIC
OCEAN BY THE ALBATROSS.

HEXACTINELLIDA O. Schmidt.

Siliceous sponges with hexactine (triaxon) spicules, and derivates of such.

The Albatross collection of Pacific hexactineUids comprises, besides a
number of small, quite irrecognizable fragments and isolated hyalonematid
stalk-spicules, from Stations 3G84 (A. A. 17), 3685 (A.A. 25), 3689 (A.A. 134), 4630,
4631, 4649, 4651, 4656, 4685, 4709, 4711, 4721, 4732, 4736, 4742, which are not
further referred to in this Report, 124 more or less complete specimens and 130
fragments sufficiently large for study and at least approximate identification.

The examination of this material has corroborated the correctness of F. E.
Schulze's ' division of the order Hexactinellida into the two suborders Hexas-
terophora and Amphidiscophora.

The collection contains representatives of both suborders.

Hexasterophora F. E. Schulze.

HexactineUids generally (or always) with hexasters, always without amphi-
discs. The spicules are either all free, or some of them are joined by
secondarily deposited silica to form a firm supporting skeleton-net.

The collection comprises sixty-seven more or less complete specimens and
124 fragments of Hexasterophora.

The examination of these sponges does not make necessary any alteration
in F. E. Schulze's most recent arrangement of the Hexasterophora in ten fami-
lies ; - all of them find a place in these families.

' F. E. Schulze. Revision des systemes der Hyalonematiden. Sitzungsb. Akad. Berlin, 1893, no.
30, p. 541; Amerikanische Hexactinelliden, 1899, p. 93.

- F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 172.

20 HOLASCUS.

The collection contains representatives of the following families: — - Euplec-
tellidae, Caulophacidae, Rossellidae, Euretidae, Coscinoporidae, and Treto-
calycidae.

EUPLECTELLIDAE (Gray) Ijima.

Tubular, cup-shaped or massive Hexasterophora attached by a stalk or a
tuft of basal spicules or sedent. Generally with numerous separate oscules.
The dermal skeleton is composed of hexactines, the proximal ray of which is the
longest. Without hypodermal pentactines.

The collection comprises four more or less complete specimens and six frag-
ments of specimens of Euplectellidae.

Ijima 1 and F. E. Schulze- distinguish two subfamilies, Euplectellinae and
Corbitellinae. The collection contains representatives of the former.

Euplectellinae Ijima.

Euplectellidae which are attached by a tuft of basal spicules.

The collection comprises four more or less complete specimens and six
fragments of Euplectellinae. These belong to the two genera, Holascus and
Holascella. The latter is new.

HOLASCUS F. E. Schulze.

Tubular Euplectellidae (Euplectellinae) with terminal sieve-plate, with
root-tuft, and without parietal apertures in the body-wall. The chief support
of the body-wall is a network composed of large tetractines, pentactines, or
hexactines, held together by slender comitals. Oxyhexasters and graphiocomes
are always present. Discohexasters and floricomes are absent. Hexactines
with equal rays, calicocomes, and sigms occur in some species and are absent in
others. The hypodermals are hexactines with short and thick, spiny distal ray,
to which slender comital rhabds are attached. The anchoring spicules of the
root-tuft are diactine rhabds with oblique backwardly directed spines and a
distal tyle, from which anchor teeth-like spines arise. The morphological
centre (axial cross) of these spicules is situated a considerable distance above
their terminal anchor-tyle.

The collection contains two more or less complete specimens and two
fragments of this genus. All belong to the same species, which is new.

' /. Ijima. Studies on the Hexactinellida. III. Journ. Coll. soi. Tokyo, 1903, 18, p. 26, 27.
- F. E. Schulze. Loc. cit., p. 173.

HOLASCUS EDWARDSII. 21

Holascus edwardsii, sp. nov.
Plate 18, figs. 15-26; Plate 19, figs. 1-24; Plate 20, figs. 1-20.

Two somewhat fragmentary specimens and two separate root-tufts of this
species were trawled in the Milne Edwards Deep, off the coast of central Peru,
at Station 4672, on 21 November, 1904; Palominos Light House, N. E., 163 km.
(88 miles) ; 13° 11.6' S., 78° 18.3' W.; depth 5203 m. (2845 f.) ; they grew on fine,
green clay; the bottom-temperature was 35.2°.

Shape and size. One of the two specimens is faii'ly large, the other small.
The large specimen (Plate 20, fig. 4) appears as a fairly straight, somewhat conic
tube about 180 mm. in length. Originally this tube probably had a circular
transverse section. Now it is flattened, one side touching the other. The tube
is about 90 mm. in circumference at the upper end, and attenuated below to a
circumference of about 50 mm. Its upper margin has a lacerated appearance,
and is not to be considered as the true termination, but as a line of fracture
along which the upper end of the- sponge has been torn off. Below, this tube
gradually passes into the root-tuft, the upper part of which appears as a compact
stalk, circular in transverse section and 15 mm. in diameter. This root-tuft is
about 100 nam. long, considerably and uniformly curved, slightly attenuated in
the middle, and spread out distally to form a somewhat irregular spicular mass.

The wall of the tube is 4-5 mm. thick. Its outer dermal face (Plate 20,
fig. 4) is very rough and irregular, an appearance probably due, to some extent
at least, to the indifferent state of preservation of the sponge. The inner, gastral
face (Plate 20, fig. 3) is perforated by numerous more or less circular apertures.
Two kinds of such apertures, large and small ones, can be distinguished. The
large apertures are 1.5-2.3 mm. wide in the central part, half way up the tube.
Toward both the upper and the lower ends of the tube they become smaller.
These apertures are very regularly arranged in one spiral line, or in a succession of
ring-shaped transverse rows. Within the spiral (the rings) they are close to-
gether, separated by walls of tissue, usually only 0.5-1.5 ram. broad. The spiral
turns (rings) themselves are farther apart, separated from each other by zones
3-4.5 mm. broad. The small apertures are mostly circular, 0.3-0.4 mm. in
diameter, and scattered in considerable numbers between the large ones.

The small specimen is similar but only 42 nun. long, and also destitute of
the upper end. Its tubular part is not collapsed, circular in transverse section,
and 6 mm. in diameter. The root-tuft is bent quite round so as to form a semi-
circle.

22 HOLASCUS EDWARDSII.

The larger of the two separate root-tufts is rectangularly bent near the
middle of its length. One limb, which evidently formed the stalk of the sponge,
is 60 mm. long, cylindrical, straight, and throughout about 14 mm. thick. The
other limb, which formed the root, is 70 mm. long, conic, and attenuated to a
fine point. The smaller separate root-tuft is similar but more uniformly curved.

The colour of the sponge-body proper, that is the tube, is, in spirit, nearly
dark brown. The root-tufts are colourless.

The skeleton. A network with rectangular meshes composed of large, stout-
rayed pentactines, held together by slender-rayed comitals, forms the main
support of the tubular sponge-body. The large pentactines have a short apical
and four long lateral rays; two of the latter, the two opposite ones, are usually
markedly longer than the other two. These pentactines lie side by side in a
single layer in the choanosome of the tube-wall; their apical rays are directed
radially outwards; their lateral rays extend paratangentially, the longer ones
longitudinally, the shorter ones transversely. The distances between the centres
of these pentactines are much smaller than the length of their lateral rays,
which consequently cross each other repeatedly. Slender-rayed diactine to
hexactine comitals accompany these pentactines in large numbers. As the
rays of these spicules closely adhere to the rays of the pentactines, and as dif-
ferent rays of the same comital are often attached to rays of different pen-
tactines, the latter are firmly held together and in position by the former.

Small hexactine megascleres, rhabds, microscleres, and siliceous skeletons
of foreign organisms also occur in the choanosome.

The small choanosomal hexactine megascleres appear to be much more
abundant in the lower than in the upper parts of the tube-wall. Quite low down,
in the region where the tubular body passes into the root-tuft, they form dense
masses.

Of choanosomal rhabds other than the diactine comitals of the large
pentactines I have observed two kinds, centrotyles and exceedingly slender,
thread-like rods. The centrotyles are of varying size, and the large ones usually
accompanied by smaller ones arranged round them comital-fashion. The
slender thread-like rhabds were found only in the choanosome of the small
specimen.

The microscleres are oxyhexasters, hemioxyhexasters, microoxyhexasters,
graphiocomes and (?) ring-shaped sigms. Of the three first named, which must
be considered as different varieties of the same kind of spicule, the oxyhexasters
are by far the most numerous; the hemioxyhexasters are rather, the micro-

HOLASCUS EDWARDSII. 23

oxyhexasters very scarce. The graphiocomes are also rather rare and nearly
always destitute of end-rays. The ring-shaped sigms are very numerous, both
in the centrifuge spicule-preparations and in the sections, but in spite of this I
am not at all sure that they are proper spicules of the sponge. They may,
Uke the masses of other siliceous skeletal structures found in the sponge, be
altogether foreign to it.

Below the outer surface of the tube-wall hypodermal hexactines with two
radially situated, differentiated rays occur. The distal, differentiated, some-
what protruding ray is short, stout, and spined. It raises the dermal membrane
conule-fashion. The four not differentiated (lateral) rays extend paratangen-
tially. The proximal ray is elongated. To the distal rays of these hexactines
slender, simple or centrotyle, comital diactines are attached, which, when
numerous, form a sort of mantle around it. Below the inner surface similar,
hypogastral, hexactines are situated. The distal rays of these spicules are, how-
ever, more slender and destitute of comitals.

The root-tuft consists chiefly of very long diactine anchor-spicules. A few
spined styles and tylostyles, with the blunt end situated distally, are also found
in it; these may, however, be foreign to the sponge.

The choanosomal centrotyle rhabds (Plate 19, figs. 22-24) are 290 ii~l.7 mm.
long and, near the middle, 5-47 n thick. The small ones, under 400 ^ in length,
are fairly numerous, the larger ones rare. The tyle is usually more (Plate 19,
fig. 23) or less (Plate 19, fig. 22) toward one end, more rarely situated centrally
(Plate 19, fig. 24). It consists of four ray-rudiments, which are, however,
so small in some, particularly the large centrotyle rhabds, that they can hardly
be individually distinguished. The tyle measures 14-60 n in transverse diame-
ter. In the small centrotyle rhabds it is relatively large, the proportion between
the tyle-diameter and the thickness of the adjacent parts of the spicule being
here 2:1 to 3.5 : 1. In the large centrotyle rhabds the tyle is relatively small,
this proportion being here 1.27: 1 to 1.5: 1. The two rays are conic or cylindric
and sharp-pointed or, more frequently, blunt. The largest centrotyle rhabds
appear to be quite smooth. The small ones are spiny, particularly near their
ends. The degree of spinulation is on the whole in inverse proportion to the
length of the rhabd.

The slender, thread-like rhabds observed in the small specimen are under 1 n
thick and relatively very long.

The rhabd cornitals of the distal rays of the hypodermal pentactines are straight
or slightly curved, simple or centrotyle, and generally 200-400 yu long and 2-2.5 n
thick.

24 HOLASCUS EDWARDSII.

The rare styles and tylostyles of the root-tuft, which, as above stated, may
be foreign to the sponge, are covered with spines, and near the distal, rounded
end 12-17 n thick. The distal end itself is either simply rounded off or, more
frequently, thickened to a terminal tyle, with a maximum transverse diameter
of 32 M.

The anchoring spicules (Plate 20, figs. 5-20) are anisoactine rhabds. I
did not observe any long intact ones. The longest fragments observed were
45 mm. in length. The morphological centre, the position of which is clearly
marked by a well-developed axial cross (Plate 20, figs. 5-8a) is only 137-200 m
distant from the distal end of the spicule. Thus, whilst the proximal ray may
attain a length of over 40 mm., the distal ray is usually less than 0.2 mm. long.
Proximally the spicule is gradually attenuated to a fine point. Distally it
thickens, and it attains its maximum thickness some distance beyond the middle
of its length, long before the morphological centre (axial cross) is reached.
Beyond, it again becomes thinner, and near the distal end, at the thinnest point
between the morphological centre and the terminal anchor, is 7-11 m thick,
about two thirds to three quarters of what it measures in the middle. At the
distal end the spicule is thickened to a terminal tyle.

The proximal part of the spicule (Plate 20, fig. 11) is perfectly smooth.
Somewhere about the middle of its oblique length, backwardly directed spines
begin to make their appearance; these usually enclose an angle of 20-30° with
the axis of the spicule. At first (Plate 20, fig. 12) these spines are very small
and far between. Farther on (Plate 20, figs. 13, 14) they become larger and
more numerous, and they continue to increase in number and size quite up to
the morphological centre (axial cross). On the distal ray the spicule has four
to seven spines every 100 ^ (Plate 20, figs. 5-10, 15-20). In the middle of the
spicule the spines are uniformly scattered and not arranged in groups (Plate 20,
figs. 13, 14). Towards the distal end they tend to form verticillate clusters
(Plate 20, figs. 5-8, 15-20), two of which are particularly pronounced, one situ-
ated at the morphological centre (axial cross) (Plate 20, figs. 5-8a), the other
at the end. Together with the terminal tyle this second cluster of spines forms
the anchor.

The large spines of the distal part of the spicule are 10-30 m long and 4-7 yu
thick at the base. The terminal ones, which form the anchor-teeth, are similar
to the others, but somewhat stouter.

The anchor appears as a conspicuous terminal thickening with an outline
closely resembling an inverted gothic arch. From the proximal side of this

HOLASCUS EDWARDSII. 25

thickening arise backwardly directed teeth usually six to eight in number. The
anchor-teeth of the same anchor being more or less unequal in size, shape, and
position, the anchors themselves appear more (Plate 20, figs. 5, 6, 16, 18) or less
(Plate 20, figs. 7, 8, 19, 20) irregular. The anchor, that is the terminal style to-
gether with its teeth, is 45-72 ix long and 32-50 ^ broad.

The axial thread of these spicules extends quite to the end of the terminal
anchor-tyle. Within this tyle it is thickened in a spindle-shaped manner and
here measures 2.5-4 /^ in maximum transverse diameter (Plate 20, figs. 5-10).
This terminal thickened part of the axial thread is granular, and irregular in
outline. From it arise a number of branchlets, up to 1 m long, usually very
oblique, strongly inclined towards the end of the anchor. According to F. E.
Schulze's figures ^ in other species of Holascus the axial thread of the anchoring
spicules in the terminal anchor-tyle is not thicker than elsewhere. In describ-
ing H. tenuis this author, however, says^ "Der Achsenkanal durchsetzt den
Kolben" (that is, the terminal anchor-tyle) "bis dicht an seine untere Spitze
und erfahrt hier zuweilen eine kleine terininale Verbi'eiterung oder Zerteilung
in ein schmales Buschel mehrerer Endauslaufer."

The thickening of the distal end of the spicule and of its axial thread
in the anchor-spicules of Holascus edwardsii is doubtlessly correlated to the
shortening of tlie distal ray. I think that the influence which prevented the
distal ray from obtaining a length commensurate with the length of the proximal
ray, also caused the thickening of the ends of the distal ray and its axial thread,
and the formation of the oblique branchlets of the terminal swelling of the latter.
This influence may be inherent, arising naturally at a certain period of develop-
ment in the spicule-building cells themselves, or it may be due to the resistance
which the distal ray encounters at its tip whilst it is being pushed outward
(downward) in consequence of the continued longitudinal growth of the proximal
ray. The latter alternative seems a priori the more probable, but I am rather
inclined to favour the former since young anchoring spicules, the distal ends
of which do not protrude over the surface and have not yet reached the deep-sea
deposit (ooze) into which they are afterwards driven, already possess a distal
anchor-tyle.

The terminal thickening of the axial thread with its branchlets in the
anchor-tyle is in many respects similar to certain structures found in the cladomes

1 F. E. Schulze. Rept. Voy. Challenger, 1887, 21, pi. 16, figs. 11, 13 H.fibulalus; Hexactinelliden des
Indischen Oceanes. II. Abh. Akad. Berlin, 1895, 1896, taf. 1, fig. 6 H. robustus; Ergeb. Deutsch. tiefsee-
exped., 1904, 4, taf. 1, figs. 4, 6 H. tenuis.

2 F. E. Schulze. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 6.

26 HOLASCUS EDWARDSII.

of the anatriaenes of Thenea valdiviae ' and other tetraxonid sponges. In these
the branchlets, however, appear to be rudiments of primary axial threads and
morphologically equivalent to the axial threads of the rhabdome and the fully
developed clades. Here in Holascus edwardsii they can only be considered as
(secondary) axial thread-branches equivalent to the axial thread-branches in
the end-clades of dichotriaenes.

The large choanosomal pentactines (Plate 18, fig. 22a; Plate 20, figs. 1, 2)
have straight or slightly curved conic rays, which are 75-145 m thick at the
base. The lateral (paratangential) rays are long and form the edges of a low
quadrangular pyramid, from the apex of which the short apical (distal) ray
arises. The lateral rays, which extend longitudinally, are 13-19 ^ long, the lat-
eral rays, which extend transversely, 7-11 mm. long. The apical ray has a
length of about 1.5 mm. The terminal parts of the rays bear scattered, small,
broad, and blunt spines. The other parts of the rays are smooth.

In the distal part of the lateral rays of these pentactines the axial thread is
thickened at frequent intervals. These thickenings are, on the whole, conic
and consist of verticils of short, rod-like, axial thread-branches diverging only
slightly from the axis of the ray, extending backward centripetally and a little
outward. The slightly granular substance, of which these rods consist, is
apparently the same as the substance composing the axial thread. Sometimes
a small cap, with the convex side turned towards the distal end of the ray, is
found within the conic rod-verticil, just below its apex. These axially situated
caps consist of a substance with a refractive index very different from that of
the substance of the axial thread and the silica-layers of the spicule, and are
consequently, in spite of their small size, very conspicuous. They look as if
they were portions of tissue rich in water, entrapped by the growing spicule
(axial thread). The conic verticils of rods (axial thread-branches) and these
caps indicate that the growth of the lateral rays of the pentactines is inter-
mittent; the rod-verticils and caps marking the positions of the ray-tips at the
times of suspension of longitudinal growth. Each node of the ray between two
adjacent thickenings of the axial thread is doubtlessly produced by the unin-
terrupted work of a spicule-builder or a set of spicule-builders. The secession
of work by the cell or cells causes the interruption of growth. After a time
the same spicule-builder or same set of spicule-builders or a fresh one or fresh
set recommences or commences work, whereupon the growth again goes on.

The comitals (Plate 18, figs. 15, 16, 22b, 23) which hold the large pentactines

' B. V. Lendenfeld. Die Tetraxonia. Ergeb. Deutsch. tiefsee-exped., 1907, 11, p. 200 ff.

HOLASCUS EDWARDSII. 27

together are slender-rayed hexactines. The most frequent forms are triactines
with a central thickening. Usually the rays are either well-developed and
very long, or reduced to mere knobs arising from the centre. Rays intermediate
between these extremes (Plate 18, fig. 23) are rare. The rays have a maximum
length of 7 mm., are very slender, only 6-14 /^ thick, usually nearly cylindrical
and terminally rounded, more rarely considerably attenuated towards the end.
The central thickening, which is composed of the knob-like rudiments of the
aborted rays, is 15-32 ^ in diameter.

The small choanosomal hexadine megascleres have straight, conic, pointed
rays, 0.17-1 mm. long and 12-18 fx thick at the base.

The hypodermal hexactines (Plate 18, figs. 19-21, 24-26) have a more or less
curved proximal ray, usually 1.2-1.5 mm. long and 5-7 ^ thick at the base.
The lateral rays ate fairly straight, have the same basal thickness, and are
usually 180-240 ^ long. The distal ray is 180-260 tx long, straight, and 6-13 ^
thick at the base. It is more or less club-shaped, thickened above, and abruptly
pointed. At the thickest point, which is usually about 50 /x from the end, the
distal ray measures 13-23 m, on an average (of twelve measurements) 18 n in
transverse diameter. The proportion of the basal to the maximum thickness
is 1 : 1 to 1 : 3, usually about 1 : 2. The distal ray is covered with spines. These
are small and scarce below but become larger and more numerous above, towards
the distal end. The spines are broad, conic, and pointed, with a maximum length
of 2.5 iJL and are directed obliquely upward, towards the end of the ray. On
account of their relatively great breadth and their obliquity, they appear as
nose-like protuberances of the ray.

The hypogastral pentactines (Plate 18, figs. 17, 18) are similar to the hypo-
dermal ones but their distal rays are distally much less thickened. The
maximum thickness of their distal rays is only 7-20 ix, on an average (of twelve
measurements) 13 m- The proportion of the basal to the maximum thickness
is 1:1 to 1:2.

The abundant oxyhexasters, the rare hemioxyhexasters, and the still rarer
microoxyhexasters (Plate 19, figs. 1, 2) are obviously all different forms of the
same kind of spicule. They measure 108-180 ^ in total diameter. A difference
in the size of forms with simple and with branched rays could not be detected.
The main-rays, which are, in the same spicule, equal, and enclose right angles
with their neighbours, are 8-12 fx long and 2-4 ^ thick. Each one bears from
one to four end-rays. The number of end-rays on the six main-rays of the same
spicule is usually unequal; but the difference is generally only one, main-ray

28 HOLASCUS EDWARDSII.

with only one end-ray being usually associated with bifurcate, bifurcate with
trifurcate, and trifurcate with quadrifurcate ones. In the simple rays, that is
in those consisting of a main-ray and a single end-ray, the point of demarcation
between main- and end-ray is clearly marked by a thickening of the distal end
of the former (Plate 19, figs. 4, 5). I consider this thickening a rudiment of the
other, reduced, end-rays. The end-rays arise steeply from the main-ray, but
immediately curve outward, and are, farther on, usually nearly straight. Occa-
sionally the proximal part of the end-ray, beyond the basal curve, is irregularly
bent. The end-rays are conic, gradually attenuated to a fine point, 57-83 n
long and 1.6-2.8 ^ thick at the base. The bases of the simple rays have the
same thickness as the main-rays. All parts of the spicule are perfectly smooth
(Plate 19, fig. 11).

The graphiocomes (Plate 19, figs. 12, 13) have main«rays which enclose
angles of 90° with their neighbours and are equal in the same spicule. The
main-rays are 11-17 m long and 2.5-3.5 ti thick. The single end-ray brush
measured was 15 m long.

The ring-shaped sigms (Plate 19, figs. 14—21), which, as above stated, may
be skeletal structures foreign to the sponge, are rods, 1-2 ju, rarely 2.8 ti, thick in
the middle, attenuated at both ends to fine points, regularly and uniformly curved
so as to form a whole low spiral turn or, more rarely, a part of such a spiral.
Lying flat they usually appear as circular rings with an interruption at one point.
The rings formed by them are 17-57 pt in diameter. The ends are usually simple
and sharp-pointed (Plate 19, fig. 14) ; rarely they bear on the concave side one
or two small, cylindrical, terminally rounded spines (Plate 19, fig. 18). Near
the middle of the rod a slight irregularity can usually be made out, but this does
not appear to be a thickening which could with any probabihty be considered
as the rudiment of another ray.

Although the upper end is missing in all the specimens and it must therefore
be left undecided whether they possessed terminal sieves or not, I think that the
want of parietal apertures, the spiculation, and the other characters described
above show clearly that they belong to the genus Holascus. From the nine
hitherto described species of this genus they differ by possessing ring-shaped sigm
microscleres. Since, however, these ring-sigms may not be homologous to the
sigms of H. fibulatus, but foreign to the sponge, I shall not consider them in the
following systematic discussion.

Apart from this, Holascus edwardsii differs from five of the nine Holascus
species by the absence of calicocomes. Of the remaining four, one, H. undulatus,

HOLASCELLA TARAXACITM. 29

is distinguished from it by the possession of discohexasters ; another, H. stellatus,
by the possession of oxyhexasters (heinioxyhexasters) with strongly curved rays;
and a third, H. fibulatus, which also has sigm microscleres, by the absence of oxy-
hexasters. The fourth, H. obesus, which appears to differ from H. edwardsii
only by its thicker body-wall and by having hypodermal hexactines with some-
what longer distal ray, seems to be more closely aUied to it. But the material
on which F. E. Schulze bases this species was very fragmentary and his descrip-
tion of it is somewhat incomplete. Therefore quite apart from the absence of
ring-sigms in H. obesus and their presence in H. edwardsii, I should hesitate
pronouncing these sponges, found respectively off Enderbyland in the Antarctic
and off Peru in the Pacific, as specifically identical.

Tubular EuplecteUidae (Euplectellinae) with root-spicule bundles and
(probably) without parietal apertures. The body-wall is supported by a net-
work of stout hexactines, pentactines, or tetractines held together by slender
comitals. To discohexasters or microdiscohexactines, other hexasters, micro-
hexactine forms, and pentactine and tetractine derivates of these may be added.
The hypodermals are hexactines with spiny distal ray. The root-spicules are
long, smooth shafts (rhabds or the long radial rays of pentactine anchors, the
distal ends of which have been lost) and monactines with oblique, backwardly
directed spines and a distal tyle, from which arise similar spines, representing
anchor- teeth. The morphological centre (axial cross) of these spicules is situ-
ated in the terminal anchor- tyle.

The collection contains two more or less complete specimens and four frag-
ments, which belong to three species, all of which are new.

Holascella taraxacum, sp. nov.
Plate 21, figs. 1-13; Plate 22, figs. 1-41; Plate 23, figs. 1-3.

One specimen, the upper end of which is missing, but which is otherwise
fairly complete, and three fragments of this species were trawled in the Eastern
Tropical Pacific at Station 4649, on 10 November, 1904; 5° 17' S., 85° 19.5' W.;
depth 4086 m. (2235 f.); they grew on a bottom of fine, sticky, gray mud; the
bottom-temperature was 35.4°.

The specific name refers to the sinularity of the abundant discohexasters
to the seed-balls of Taraxacum.

30 HOLASCELLA TARAXACUM.

Shape and size. The specimen (Plate 21, fig. 8), which is fairly complete
apart from the missing upper end, consists of a nearly straight tube, open at
both extremities, from the lower, somewhat attenuated end of which arise three
dense bundles of root-tuft spicules. The tube is about 120 mm. long and has a
circumference of 70 mm. at the upper end and of 40 mm. at the lower. It is
now, although rather rigid, considerably compressed and flattened. In the fresh
state it probably had a circular transverse section. The wall of the tube, that
is the body proper of the sponge, is, for the most part, 2.5-3.5 ram. thick, and
perforated by numerous apertures. These apertures are more or less circular
in outline, 0.3-1.5 mm. wide, and quite irregularly distributed. Besides these
apertures radial canals of similar width, but covered on the outer side by rem-
nants of tissue, are observed in the tube-wall. For this reason, on account of
their quite irregular distribution, and because the open apertures are destitute
of a special marginal membrane, and all the larger and most of the smaller ones
are traversed by rays of choanosomal spicules, I do not think that they can be
considered as true parietal apertures. I believe myself justified in assuming
that the tube-wall is, in the living sponge, continuous and destitute of parietal
apertures, and that the openings now observed in it are post mortem artifacts,
produced by the shrinkage and partial maceration of the soft parts, and the
loss of extensive tracts of the dermal membrane.

The three root-spicule bundles are very dense, 80-120 mm. long, con-
siderably and uniformly curved, and attenuated distally to quite fine points.
Proximally they widen out paratangentially and they pass, by the divergence
of the spicules composing them, gradually into the lower end of the tubular
body.

Of the three fragments, one is the lower end of a tube similar to the one
described above. It is *45 mm. long, circular in transverse section, slightly
attenuated below, and open at both ends. Above it has a diameter of 14, below
of 12 mm. From its lower end three root-spicule bundles arise. The other two
fragments appear to be parts of tubular bodies.

The colour of the body proper is, in spirit, dirty brown; the root-spicule
bundles are colourless.

Skeleton. The chief support of the body consists of longitudinal and
transverse bars, which form a paratangentially extending net with rectangular
meshes. This net is composed of the paratangential rays of large stout principal
spicules, held together and in position by slender comitals. Most of the prin-
cipal spicules are hexactines, a few pentactines and tetractines. Each node of

HOLASCELLA TARAXACUM. 31

the net is occupied by the centre of one of these spicules. The two rays of the
large principal hexactines, which extend longitudinally, are considerably longer
than the other four. The two rays extending transversely are intermediate in
size. The two rays extending radially are the shortest, the proximal one,
pointing towards the axis of the tube, being the shorter of the two. The para-
tangential rays of most of the principal pentactines and tetractines are simi-
larly differentiated. The single racUal ray of the pentactines points outward.
Most of the comitals are centrotyle rhabds, a few tri-, pent-, or hexactines.

Besides these spicules, there have been found in the body of the sponge
hexactines intermediate in size, very long and slender, longitudinally extend-
ing rhabds, minute rhabds, micro-tetractines, -pentactines, and -hexactines,
oxyhexasters, discohexasters, onychhexasters, (calicocomes) , and the central
parts (main-ray crosses) of graphiocomes. The oxyhexasters, onychhexasters,
graphiocome-centres, and minute rhabds are rare. One or the other of these
kinds of spicules may possibly be foreign to the sponge. The other spicule-
forms mentioned are abundant and doubtlessly proper to the sponge.

Hypodermal and hypogastral hexactines with two axes (four rays) extending
paratangentially and one axis (two rays) extending radially (vertically to the
surface) are found below the dermal and the gastral surfaces. The proximal
ray of these spicules is elongated, the distal ray spined and more or less thick-
ened. Hexactines of this kind with greatly, and with only sUghtly, thickened
distal ray are indiscriminately mingled both in the outer dermal and the inner
gastral face of the tube-wall. The hypodermal and the hypogastral hexactines
are \^ery similar. The only difference between them which I could detect is that
in some of the hypodermals the distal ray attains a greater length than in any
of the hypogastrals, and that in some of the hypogastrals the lateral rays attain
a greater length than in any of the hypodermals. It also appears that the
distal rays of the hypodermals of the lower part of the sponge are on the whole
thicker than those of the upper part.

The root-spicule bundles are composed of numerous large, smooth rhabds,
broken off below, and a few spined monactine anchors.

The rays of the large principal hexactines (Plate 22, figs. 5, 6, 9, 10, 36;
Plate 23, fig. 1) are shghtly and irregularly curved (Plate 22, fig. 7) or, more
rarely, angularly bent (Plate 22, fig. 9), blunt, and usually conic. In very long
rays (Plate 22, fig. 7) the thickest point is often some distance from the base,
and such rays are somewhat spindle-shaped. Rarely one of the rays is abnor-
mally reduced in length and terminally thickened (Plate 22, fig. 6), or divided

32 HOLASCELLA TARAXACUM.

at the end into two branches (Plate 22, fig. 8). The rays are 100-160 m thick
at the base; the longitudinal ones are 6-22.5 long, the transverse ones 2-10,
the distal one 1.5-2.5, and the proximal one about 1 mm.

In the proximal part of large rays a homogeneous central part, about 40 n
thick, and a conspicuously stratified superficial part can usually be distin-
guislied. In the axis of the distal part of such large rays structures are observed
somewhat similar to those described above in the corresponding spicule-rays
of Holascus edwardsii. The axial thread is quite thin in the proximal part of
the ray; in the distal part it is considerably thickened, and interrupted by caps
composed of a substance of different refractive index from the axial thread and
the silica-layers surrounding it (Plate 23, fig. 1). These caps are usually 4-6 n
broad and so situated that the convex side lies distally. These caps are irregu-
larly distributed along the axis and are very numerous. Sometimes quite a
number of them follow in close succession. From the margin of most of these
caps a distinct limit between successive silica-layers arises. These limits extend
proximally, are conic in shape, and pass uninterruptedly into the limits between
the silica-layers forming the outer, clearly stratified zone of the proximal part
of the ray. These limits represent former surfaces of the spicule, whilst the caps
mark the positions of the tip of the ray at various times. There can be little
doubt that here, as in Holascus edwardsii, the growth of these spicules is intermit-
tent, interrupted by periods of rest. Every time the longitudinal growth of the
rays recommences after such an interruption a cap is formed.

It has been stated above, that in some of the large principal hexactines one
of the rays is reduced in length and terminally thickened. In the centre of the
terminal thickening of such shortened rays the central, unstratified zone of the
spicule ends in the shape of a slender, pointed cone. The terminal thickening
itself is formed exclusively by the clearly stratified superficial zone, each layer
of which is here markedly thickened.

The few large principal pentactines and tetractines (Plate 22, figs. 7, 8, 11) are
similar to the principal hexactines described above. Most of them differ from
the latter only by the absence of one (the pentactines) or both (the tetractines)
the radial rays. In some of them also the difference of the longitudinally and
transversely extending rays is less pronounced than in the principal hexactines.

The axes of the intermediate hexactines are not differentiated and, although
the rays are in the same spicule often more or less unequal, they are apparently
equivalent. The rays are 140-300 fi long, usually cylindrical, 7-12 n thick, and
rounded and often thickened at the end. The tips of the rays are spiny. The
other parts of the spicule are smooth.

HOLASCELLA TARAXACUM. 33

The comital rhabds (Plate 22, figs. 29-36, 38-41) are diactines with a dis-
tinct thickening Ijong more or less centrally. They are 5-15 mm. long and the
two developed rays are, at their proximal end, near the centre of the spicule, 11-
45 II thick. They taper distally and measure, at their thinnest point, which is
usually situated a short distance from the end, 6-20 ^ in transverse diameter.
The end itself is usually thickened, more rarely conic and pointed.

The two ends of the same spicule usuallj^ differ considerably from each other.
The terminal tliickening is oval or club-shaped and 20-47 fj. in diameter (Plate 22,
fig. 38-40). The ends of these spicules are sUghtly spiny, all the other parts
smooth.

The more or less centrally situated tyle consists of four rudimentary rays,
the axial threads of which can always be distinguished as an axial cross within
it. The degree of reduction of these four rays is subject to considerable varia-
tion, and not infrequently the four rudimentary rays of the same spicule are
reduced to a very different degree. A series of forms representing different
degrees of ray-reduction is reproduced (Plate 22, figs. 29-35).

The tri-, pent-, and hexactine comitals are rather rare. The triactine forms
are similar to the diactines above described and differ from them only by being
smaller and by one of their four reduced rays being much longer than the others.
The rays of these spicules are 17-25 m thick and the longest is 1.5-2.6 mm. long.
The pentactines and hexactines have rays 1-2.5 mm. long and 13-20 fj. thick.

The long slender rhabds are centrotyle and similar to the diactine comitals
above described, so that one might consider them as giant forms of these. They
attain a length of 36 mm. and a thickness of 27 /i. The central tyle has a maxi-
mum thickness of 44 /u. Some of them have a very large terminal tyle, sometimes
70 ju in diameter, at one end. Rays thus terminally greatly thickened are cor-
respondingly reduced in length. The axial cross, which lies in the central thick-
ening, is in some of these spicules very irregular, the axial thread-rudiments
composing it enclosing angles very different from 90° with the axis of the two
developed rays.

The rare minute rhabds, which may perhaps be comitals of the distal ray of
the hypodermal hexactines, but which were never seen in situ in this position,
are about 260 ^u long and 1.5 mm. thick.

The proximal and lateral rays of the hypodermal and hypogastral hexactines
(Plate 22, figs. 1-4, 12-17) are 5-11 n thick at the base, cylindrical or only
slightly attenuated towards the end, and abruptly pointed or blunt. The proxi-
mal ray is 0.8-1.8 mm. long, tKe lateral rays are 0.2-1 nmi. The distal ray is
160-500 /u long, at the base 5-18 /i thick, and thickened more or less above. At

34 HOLASCELLA TARAXACUM.

its thickest point, which Ues only a very short distance below the end, the distal
ray is 16-60 ii thick. The proximal and the lateral rays are often curved; the
distal ray is straight. The proximal and lateral rays are smooth apart from their
ends, which are often slightly spined. The basal part of the distal ray is smooth,
or only slightly spined; its thickened end is covered with spines, situated very
oliliquely and directed upwards toward the tip of the ray. These spines are
quite numerous and close together, have a maximum length oi b ix, and are
about 4 n thick. They appear as oval protuberances, the ends of which are
drawn out to sharp and slender points. The tip of the ray is free from spines
for a distance of about 10 ti. In the distal rays of medium thickness the
tip appears as a broad cone, in the very thick ones as a broad round dome.
The proximal part of the axial thread of the distal ray is quite thin, its distal
part thickened, and about 1.5 ;u broad.

I have observed a few spicules in the spicule-preparations which also appear
to be hypodermal or hypogastral hexactines, but which differ from the spicules
above described by one, two, or even three of their lateral rays being thickened
and spined like the distal ray.

The smooth root-spicules are all broken off at the lower, distal end. The
longest fragments measured were 150-160 mm. long. Proximally these spicules
are gradually attenuated to a fine point. Their thickest portion is about 120
mm. from the proximal end; here they are 100-340 m thick.

The spined, anchor-like root-spicules (Plate 22, figs. 26, 37; Plate 23, figs.
2, 3) are remarkably scarce. All those seen were broken so that I cannot give
their length. To all appearance they are much shorter than the smooth root-
spicules. Near their distal end these anchor-spicules are 12-17 ^ thick. The
end itself is thickened to a terminal tyle, 48-56 fi broad, 54-70 ^ long, and in
shape like a blunt, inverted cone with convex sides or a rotation-paraboloid.
From the shaft of the spicule and from the margin of the upper, basal face of the
terminal tyle arise conic, obliquely situated, backwardly directed spines 7-17 fj.
long. The axial cross (morphological centre) of the spicule lies in the terminal
tyle (Plate 23, figs. 2, 3). These spicules are not, like the similarly shaped
anchors of the species of Holascus, diactines, but monactine tylostyles.

Among the micro-oxyhexactines, -oxypentactines, and -oxytetractines {staur-
actines) (Plate 22, figs. 20-25), the hexactine forms appear to be the most
abundant. The rays of these spicules enclose angles of 90° with their neighbours
and are equal in most of the hexactines and stauractines. In the pentactines
and some of the hexactines (Plate 22, figs. 20, 21) a differentiation of the three

HOLASCELLA TARAXACUM. 35

axes is to be noted, two rays of such hexactines lying in one axis, and the ray of
the pentactines which has no opposite being longer than the four rays lying in
the two other axes. The rays are straight, conic, pointed or blunt, 120-180 m
long and 3-8 ix thick at the base. With the exception of the base and the extreme
tip, which are smooth, the rays are covered with spines, 1-1.5 ^ long. The distal
spines are distinctly recurved (Plate 22, figs. 18, 19), the proximal ones arise
nearly vertically.

The rare oxyhexasters (Plate 21, figs. 1, 2, 9} are about 95 m in diameter.
Their equal and regularly arranged main-rays are straight, fairly smooth, 19 ^
long, 4 yu thick at the base, and slightly attenuated towards the distal end. Each
main-ray bears a terminal verticil of usually three end-rays, enclosing angles of
about 45° with the continuation of the main-ray. The end-rays are perfectly
straight, 37 /z long, 2 ^ thick at the base, conic, sharp-pointed, and covered with
minute spines.

The rare onychhexasters (Plate 22, figs. 27, 28) are 98-105 n in diameter and
have a thickened centre, 4-5 ^ in diameter. The main-rays are regularly
arranged, in the same spicule faii'ly equal, straight, on the whole cylindi-ical,
8-11 n long and 1.5-2.3 ju thick. They bear from one to four, usually three,
end-rays, and sometimes also one or a few irregular knob-like protuberances
on their sides. The end-rays are 30-50 /u long and 0.6-1 ix thick at the base.
Distally they taper gradually to about 0.3 ix. The end-rays arise nearly verti-
cally from the main-ray and are cm-\-ed in an S-shaped manner, their proximal
part strongly concave towards the continuation of the main-ray, their distal
part sUghtly in the opposite direction. This curvature is different in different
end-rays and the degree of divergence of the chords of the end-rays from the
continuation of the main-ray is variable. Each end-ray bears several terminal
spines. These generally arise at nearly right angles, are curved, concave towards
the centre of the spicule, slender, and 2-5 m long. In view of the shape of the
end-rays these onychliexasters might also be termed cahcocomes.

Of graphiocomes only a few centres (main-ray crosses) have been observed.
The main-rays are regularly arranged, equal, 11-13 m long and 2.5-4 ix thick.

The abundant discohexasters (Plate 21, figs. 3-7, 10-13) are regularly spherical
and measm-e 180-290 ix in total diameter. Thek main-rays are regularly ar-
ranged, in the same spicule equal, perfectly smooth, about 14 ^ long, 3.5-5 ai
thick in the middle, and thickened at both ends ; proximally to the centre of the
spicule, distally to a stout, lens-shaped, transverse disc from the margin and
distal face of which the end-rays arise (Plate 21, fig. 10). The end-rays are

36 ' HOLASCELLA TARAXACUM.

so numerous that it is exceedingly difficult to count them. So far as I could
make out 23-27 end-rays arise from the terminal disc of each main-ray. The
end-rays arising from the central part of the distal face of the terminal main-
ray discs are nearly straight throughout, and extend in a radius from the centre
of the spicule. The end-rays become longer, more curved and concave toward
the continuation of the main-fay axis the farther they are situated from the
centre of the disc.

This curvature is restricted to the basal part; the middle- and end-parts are
always straight. This increase of length and curvature towards the margin
of the disc is such that the tips of all the end-rays are nearly equidistant and lie
in the surface of a regular sphere, and that the straight middle- and end-parts
of all the end-rays lie in radii from the centre of the spicule. In consequence
of this, and because the crowd of end-rays hides the main-rays, the whole spicule
appears as a regularly spherical aster composed of numerous straight, concentric,
and equidistant radial rays. The end-rays are 80-140 y. long and 2.5-3.5 ^
thick at the base. Towards the middle of their length they are attenuated to
1.5-2.5 m; farther on they again become thicker, and attain a transverse diameter
of 3.2-5 M at their distal end. At the base and just below the tip the end-rays
are quite smooth for a short distance. For the remaining greater part of their
length they are covered with oblique, backwardly directed and backwardly
curved, conic spines, 1-2.5 m long. From the end arises a terminal verticil of
about fifteen recurved spines. The basal parts of these spines are joined to
form a disc with strongly convex distal face, from the margin of which their ends
protrude freely for a distance of 2-3 ix. The terminal spine-verticils (end-discs)
measure 7.5-12 ju in transverse diameter.

The general structure and spiculation of the sponges above described clearly
show that they are Euplectellidae, whilst the presence of root-spicule bundles
assign them to the Euplectellinae. Since, however, the upper part is not present
in any of the specimens, and the state of their preservation is insufficient to deter-
mine whether the wall of their tubular body is perforated by parietal apertures
or not, it is somewhat difficult to decide in which genus they should be placed.
Whether the upper end of the tubular body was open or covered by a sieve-plate
of course cannot be decided. About the parietal apertures, however, we may
with some confidence say, for the reasons above given, that the holes now ob-
served in the body-wall are post mortem artifacts produced by shrinkage and
maceration and that the sponge possesses no parietal apertures in the fresh state.

At present three genera, Euplectella, Holascus, and Malacosaccus are dis-

HOLASCELLA ANCORATA. 37

tinguished in the Euplectellinae. The certain presence of discohexasters and
the probable absence of parietal apertures preclude the sponges described above
being placed in Euplectella. From the known species of Malacosaccus, which
are soft, flexible, and sac- and cup-shaped, they differ by being hard and brittle
narrow tubes. From all the known species of Holascus, except Holascus undu-
latus F. E. Schulze ^ and the species collected by the Challenger and mentioned
by F. E. Schulze- as Holascus sp., they differ by possessing discohexasters.
The spicules of H. undulatus described by F. E. Schulze {loc. cit., 1899, p. 17) as
discohexasters differ, however, considerably from the true discohexasters found
in the sponges described above and have by F. E. Schulze himself lately ^ been
declared to be calicocomes, and not discohexasters, so that this species also does
not appear to be aUied to the sponges above described. Their only closer allies
appear to be the species of Holascus referred to and the new Pacific species
described as Holascella ancorata, and H. euonyx.

As they differ from all the hitherto described and named species of Holascus
by possessing discohexasters, hemidiscohexasters, or microdiscohexactines, and
as the absence or presence of such spicules should be considered as a difference
sufficient for generic distinction, I name the new genus Holascella, on account
of its similarity to and historic derivation from Holascus.

From Holascus sp. Schulze and the sponge here described as Holascella
ancorata, Holascella taraxacum differs by being destitute of floricomes, and
from the latter also and from the sponge here described as Holascella euonyx by
the absence of discohexactines and hemidiscohexasters with large anchor-Uke,
terminal spine-verticils. From H. ancorata and H. euonyx it is also distin-
guished by its principals being mostly hexactines.

Holascella ancorata, sp. nov.
Plate 23, figs. 4-25; Plate 24, figs. 1-9.

One specimen of this species was trawled in the Eastern Tropical Pacific
at Station 4649 on 10 November, 1904; 5° 17' S., 85° 19.5' W.; depth 4086 m.
(2235 f.) ; it grew on a bottom of sticky, gray mud; the bottom-temperature was
35.4°.

It has discomicroscleres with long, strongly recurved terminal spines
not joined at the base to a terminal tyle ("disc"). The end-rays (rays) of

1 F. E. Schulze. Amerikanische Hexactinelliden, 1899, p. 15, taf. 3, figs. 1, 2.

= F. E. Schulze. Kept. Voy. Challenger, 1887, 21, pi. 15, figs. 14-23.

* F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped. , 1904, 4, p. 130, 131.

38 HOLASCELLA ANCORATA.

these spicules are exquisitely anchor-shaped in consequence. To this the name
refers.

Shape and size. The single specimen (Plate 23, fig. 9) is a conic tube 40
mm. long. It is circular in transverse section, broken off at both ends, at one
end 11 mm. in diameter, at the other 7 mm. Its wall is continuous, not perfo-
rated by parietal apertures, and about 2 mm. thick. To the narrower end a
root-tuft appears to have been attached.

The colour in spirit is dirty brown.

The skeleton. The chief support of the tubular body is a paratangential
network of principal spicules held together and in position by slender comitals.
The principals have from three to five, rarely six rays. Two opposite rays
extend more or less longitudinally. One or both of these longitudinal rays
are longer than any of the others. All the rays of the triactines and tetractines
and four rays of the pentactines and hexactines lie paratangentially ; one ray of
the pentactines and two rays of the hexactines extend radially. These radial
rays are always shorter than the others. The comitals, which are attached to
the rays of these principals, are diactines, triactines, and tetractines. Besides
these spicules a few tetractine and a good many hexactine megascleres, with spined
rays, much smaller than the principals of the supporting network, occur in the
choanosome. Hypodermal and hypogastral hexactines, with the two (opposite)
rays of one of the axes differentiated, occur below the dermal and the gastral
surface. One of these differentiated rays is elongated, the other thickened and
more or less spined. The axis of the two differentiated rays is situated radially;
the elongated ray points inwards, the thickened and spined ray outwards. A
few spined anchoring spicules have been found in the narrower part of the tube.
They are probably root-tuft spicules of the sponge. In addition to the spicules
mentioned, rods and tetractines to hexactines with very short, stout rays,
probably foreign to the sponge, have been observed in the spicule-preparations.
Of microscleres spined microhexactines, floricomes, onycho- and discomicro-
scleres, and a few main-ray crosses without end-rays have been observed. Among
the onycho- and discomicroscleres microhexactines and hemihexasters are
much more frequent than true hexasters. Some of the main-ray crosses observed
are the central parts of the floricomes; others may be centres of graphiocomes.
The discomicroscleres are very numerous and doubtlessly proper to the sponge.
All the other microscleres are rather rare and one or the other of them may be
foreign to the sponge.

Among the large triactine to hexactine principal spicules (Plate 23, fig. 4;

HOLASCELLA ANCORATA. 39

Plate 24, figs. 3, 8) the triactines to pentactines are much more numerous than
the hexactines. Many of these spicules are very irregular, the rays, also opposite
ones, frequently differing very greatly in length, and the longer rays being invari-
ably more or less curved. Most of the triactine principals consist of two oppo-
site longer rays lying in the same axis longitudinally, and one lateral (transverse)
shorter ray, more or less vertical to the rhabd formed by the other two. In
some of the principal spicules the rays are not only unequal but seem also to be
irregular in position, to enclose angles other than 90° with their neighbours.
A closer inspection, however, shows that the axial threads of the basal parts of
the rays of such spicules are also regularly disposed at right angles (Plate 24,
fig. 8), their apparent irregularity of position being due merely to strong curva-
tures near their basal part. The rays are smooth and blunt-pointed. The
shorter ones are simply conic and gradually attenuated to the end; in the long-
est ones the thickest point often lies a short distance from the base, so that these
rays appear somewhat spindle-shaped. Such rays are at the thickest point about
7% thicker than at the base. The principal spicules are 18^2 mm. long, their
longitudinally extending rays measuring 10-21 mm. in length, their transverse
paratangential rays 3-15 mm. The rays are 70-160 n thick at the base. The
basal thickness of the rays is, on the whole, proportional to their length.

The rays of these large principals are, like those of the principal spicules
of Holascella taraxacum, composed of a nearly homogeneous axial and a very
clearly stratified superficial zone. In the tetractine (Plate 24, fig. 8) the axial
zone is 18 M in diameter near the centre of the spicule, whilst the clearly strati-
fied superficial zone has here a thickness of 56 ix. The layers of the latter are
very unequal in thickness; distally they terminate in cones, the apices of which
lie in the axial thread.

In some of these spicules distinct signs of their having been broken at some
time during the period of growth are to be noticed. In the portion of a ray of a
principal spicule (Plate 23, fig. 4) a fracture is visible, which shows that the
tip of this spicule-ray was broken off at a point where it was about 25 fj. thick,
and that the ray continued to grow, not only in thickness but also in length, after
this breakage. It is clearly to be seen that a new axial thread, lying exactly
in continuation of the old broken one, was formed after the fracture. This new
axial thread is widened proximally to a cone, which encloses the tip of the old
broken axial thread. The new axial thread is a regenerate, the existence of which
shows that the elements attached to the tip of a growing spicule-ray are not the
only ones that can build up an axial thread,

40 HOLASCELLA ANCORATA.

The svialler, spined hexactine and tetractine megascleres (Plate 24, figs. 1, 2)
are 1-4 mm. in maximum diameter. Their rays are unequal, often curved,
up to 1.7 mm. long, 12-17 ^ thick at the base, and rounded at the end or blunt-
pointed. The bases and often also the tips of the rays are smooth, the other
parts show sparse, broad, sharp-pointed spines.

The comital spicules (Plate 24, fig. 9) are di- to tetractine. Their rays are
straight or irregularly curved, gradually attenuated distally, and terminally
rounded. The end-part is usually somewhat thickened and spined. The other
parts of the spicule are smooth. The rays attain a very considerable length.
Measurements of this dimension cannot, however, be given since all the long
rays observed were broken off. The longest intact ones seen were 1.5 mm. long.
The rays are 8-28 ^ thick at the base and attenuated distally to 5-8 tx. The
spined end-part is 7-10 m thick. In the tetractine and triactine comitals two
opposite longitudinal rays lie in a straight line and are longer than the transverse
ones (one) . In the triactine forms the centre is markedly thickened on the side
opposite the single transverse ray (Plate 24, fig. 9). The diactine forms are cen-
trotyle. The central tyle measures 14-36 m in diameter. The proportion of the
basal thickness of the rays to the transverse diameter of the tyle is 1 : 1.4 to 1 : 3,
usually about 1: 1.6. The two rays of these spicules are usually unequal in
length and sometimes one of them is reduced to a mere knob. Such excessive
longitudinal reduction is always associated with a considerable thickening.
In an extreme form of this kind one ray was observed to be over 2 mm., the
other only 44 m, long. The central tyle measures 40 ^ in diameter; the long ray
is 16 ju thick and nearly cylindrical. The short ray is 33 ^ thick at the base and
farther on it is 44 /x thick. This knob-like rudimentary ray is covered with small
spines down to within a short distance of its base.

The proximal and lateral rays of the hypodermal and hypocjastral hexactines
(Plate 23, figs. 12, 13) are 5-9 m thick at the base. They are cylindrical or
slightly attenuated distally, and usually rounded at the end, rarely pointed.
Their tips are generally spined, their other parts smooth. The proximal ray
is 0.9-1.5 mm. long, the lateral rays 370-450 m- The distal ray is 220-450 ^
long, at the base as thick or somewhat thinner than the other rays, and distally
thickened. At its thickest point, which lies near the distal end, it measures
17-40 n in diameter. The proximal part and the extreme tip are smooth, the
other parts of it more or less spined. The spines increase in size and number
distally. They arise very obliquely and point towards the tip of the ray. The
hypodermals are similar to the hypogastrals. Hexactines with thick strongly

HOLASCELLA ANCORATA. 41

spilled, and with thin only slightly spined, distal rays occur among both. The
distal rays of the hypodermals appear to attain a greater length than the distal
rays of the hypogastrals, the former being usually over, the latter under, 400 ^
long.

The few root-tuft anchors observed are monactines. Their axial cross lies
in their terminal anchor-tyle. The shaft is covered with very irregularly dis-
tributed, backwardly directed spines 17 ii thick just above the terminal anchor-
tyle. The terminal anchor-tyle is similar to that of Holascella taraxacum. It
is, with the spines, 57-65 m broad and 74-90 m long. Its spines, the anchor-
teeth, are very irregular.

The microoxyhexactines (Plate 23, fig. 8) measure 112-195 m in total chame-
ter. Their rays are regularly arranged, in the same spicule fairly equal, straight,
conic, pointed, 55-105 /x long and 3-5 m thick at the base. Their length is not
in proportion to their basal thickness, the shorter rays of smaller microoxyhex-
actines being often thicker than the longer rays of larger ones. The rays are
rather sparsely spined. The spines are sharp, not over 1 yu long, and directed
obliquely backwards.

The onychomicroscleres (Plate 23, figs. 10b, 11, 14b, 15, 16) measure 65-90 fi
in total diameter, and have one to three end-rays. Many are microonychhex-
asters with only one end-ray on all the main-rays. Others are hemionychhex-
asters, with one end-ray on some, and two or, rarely, three end-rays on the other
main-rays. A few are true onychhexasters with two to three end-rays on each
main-ray. The main-rays are regularly arranged and, in the same spicule,
fairly equal. They are cyhndrical, smooth, about 5 m long and 1.5-2 ^ thick.
The end-rays are straight or only slightly curved, 28-50 ^ long, conic, at the
base about 1 m thick, and at the end 0.5-0.8 fi. They bear exceedingly minute
spines along their length, and at their end there are several, usually three or
four, large, more or less vertical spines. These terminal spines are not o\'er
7 M long, very slender, and curved, either simply, concave to the centre of the
spicule, or in an S-shaped manner. \Yhen two or three end-rays arise from a
main-ray, they enclose angles of 30° to 40° with its continuation; when there is
only one end-ray it Ues in the continuation of the main-ray, and usually passes
into it so gradually that main- and end-ray together appear as a simple, conic
hexactine ray. Sometimes a slight irregular thickening or change of direction
indicates the point where the main-ray passes into the end-ray. Such simple
rays are 33-35 m long.

The discomicrosckres (Plate 23, figs. 5-7, 10a, 14a, 17-25) measure 130-220 /*

42 HOLASCELLA ANCORATA.

in total diameter. They generally have only one, sometimes two, very rarely
three end-rays. Most of them are microdiscohexactines with one end-ray on
each main-ray ; some hemidiscohexactines with one end-ray on some main-rays
and with two end-rays on others. A few are true discohexasters with two end-
rays on all or with two end-rays on some and three end-rays on the other main-
rays. The main-rays are regularly arranged and, in the same spicule, fairly
equal. They are smooth, about 5 n long and 2.5-3.7 n thick. The end-rays are
straight or slightly irregularly curved, 50-110 yu long, thinnest some distance
below the distal extremity, and thickened at both ends. The proximal end is
2-3 M thick, the thinnest point 0.7-1.5 n, and the distal end 2.6-4 fi. The end-
rays bear very minute spines on their sides and a verticil of large, anchor-teeth
like, strongly recurved spines at their end. These terminal spines are conic,
8-10 M. long and 1.2-1.6 n thick at the base. They are not joined at the base
to a terminal tyle or disc, and together form an exquisite anchor, 9-12 n broad
and about as high. When two or three end-rays arise from a main-ray, they are
usually arranged somewhat irregularly and enclose angles of 20°-45° with the
continuation of the main-ray. When, as is the rule, there is only a single end-
ray, it lies in the continuation of the main-ray, and usually passes into it so
gradually that main- and end-ray together appear as a simple hexactine ray.
Such simple rays are 65-115 ti long.

Axial threads are found only in the main-rays. They appear as thin, fairly
straight rods, are about 5 m long, and terminate abruptly at the point where the
main-ray divides into the two or three end-rays (Plate 23, fig. 7, right), or passes
into the single end-ray (Plate 23, figs. 6, 7, left, upper and lower). The simple
rays with only one end-ray consequently possess an axial thread only in their
basal (main-ray) part.

The few main-ray crosses observed, which may be centres of graphiocomes ,
consist of regularly arranged, equal, straight main-rays 10 m long and 3.5 n thick,
from the ends of which large numbers of end-rays arise.

The floricomes (Plate 24, figs. 4-7) measure 48-60 m in total diameter.
Their main-rays are regularly arranged, in the same spicule equal, cylindrical,
straight, 6-7 yu long, and about 1.5 ^ thick. Each main-ray bears a verticil of
about twelve end-rays. All the end-rays arise at exactly the same level, 1-1.5 ^
below the distal end of the main-ray, which protrudes for that distance in the
shape of a rounded knob beyond the ring-shaped line of their insertion. The
end-rays of the same verticil are exactly equal in size, shape, and position, rela-
tive to the main-ray from which they arise. They are, measured along their

HOLASCELLA ANCORATA. 43

chord, 20-23 m long and strongly curved in an S-shaped manner. Their basal
part is directed outwards and slightly backwards, their central part upwards
and slightly outwards, and their distal part again outwards and slightly back-
wards. They are exceedingly thin at the base, but thicken distally and attain,
a short distance from the end, a maximum transverse diameter of about 1.3 ft.
The end-rays are smooth on the inner side, that is the side turned towards the
continuation of the main-ray. On the opposite, outer side their thicker distal
part bears fairly large spines.

As far as the fragmentary condition of the specimen allows one to judge, the
only species more closely allied to it is the specimen referred to by F. E. Schulze '
as Holascus sp. and those described in this paper as Holascella taraxacum and
H. euonyx. It is very clearly distinguished from Holascus sp. Schulze and
Holascella taraxacum by the terminal spines of its discomicroscleres. In the
sponges described above these are long, slender, strongly recurved, and isolated
quite down to the base. In Holascella taraxacum they are certainly, and, to judge
by the figures, in Holascus sp. Schulze most probably, short, divergent, and
basally joined to form terminal tyles (' ' end-discs ") . From the former H. ancorata
also differs by the principals, which are in the sponge above described chiefly
tri- and tetractines, in H. taraxacum chiefly hexactines; by the discomicroscleres,
which have few end-rays in the former and very numerous end-rays in the latter;
and by the floricomes which are present in the former and appear to be absent
in the latter. There can, therefore, be no doubt that H. ancorata is specifically
distinct from H. taraxacum. Whether it is also distinct from Holascus sp.
Schulze, of which no adequate description exists, is not so easy to say, the
figures of this sponge given make it highly probable, however, that it belongs
to a different species.

It appears to be more closely related to these species than to the sponge
here described as Holascella euonyx. From this it differs by the superficial
hexactines, the distal rays of which are much thicker and more club-shaped in
H. ancorata than in H. euonyx; by the discohexactines and hemidiscohexasters,
the rays (end-rays) of which are more spiny and bear much smaller terminal
anchor-teeth in the former than in the latter; by the onychhexactines and hemi-
onychhexasters, which have much shorter terminal spines in the former than in
the latter, and by the presence of floricomes in the former and their absence in
the latter.

1 F. E. Schulze. Kept. Voy^ ChaUenger, 1887, 21, p. 86, 87, pi, 15, figs. 14-23.

44 HOLASCELLA EUONYX.

Holascella euonyx, sp. nov.
Plate 24, figs. 10-17; Plate 25, figs. 1-24.

A fragment of this species was trawled nearly under the equator in the
Eastern Pacific at Station 4742, on 15 February, 1905; in 0° 3.4' N., 117° 15.8'
W.; depth 4243 m. (2320 f.); it grew on very hght, fine, Globigerina ooze; the
bottom-temperature was 34.3°.

It is characterized by possessing hemionychhexasters and onychhexactines
with very long terminal spines (end-claws). To this the name refers.

Shape and size. The single fragment (Plate 25, fig. 17) is a very slightly
cylindrically curved plate which may have formed part of a wide tube. It is
51 mm. long, 15 mm. broad, and about 1.5 mm. thick.

The colour in spirit is brown.

The internal skeleton is composed of parallel bundles of spicule-rays, and of
loose spicules. Near the surface special superficial (dermal, gastral) hexactines
occur. The bundles are composed of stout principals, for the most part tetrac-
tine, and slender comitals likewise chiefly tetractine. The loose parenchymal
spicules consist of numerous large and a few small simple hexactines ; a few hemi-
onychhexasters ; numerous onychhexactines; numerous small discohexasters
with many end-rays; very few large hemidiscohexasters with few end-rays;
and numerous large discohexactines. The special superficial hexactines have a
differentiated distal ray.

Besides these spicules numerous small hexactines with curved rays, a few
pinules, and a good many large amphidiscs have been observed both in the sec-
tions and the spicule-preparations. These spicules are in all probability foreign.

The distal ray of the superficial (dermal, gastral) hexactines (Plate 25, figs.
14, 15, 21-24) is fairly straight, 235-270 n long and about 6-10 ^ thick at the
base. Towards the distal end it is thickened more or less, the end itself being
abruptly pointed. At its thickest point, which lies a short distance below the
end, the distal ray measures 9-16 yu in transverse diameter. The basal part of
the ray is smooth, the distal part covered with stout, very oblique spines 1-2 n
long. These spines are somewhat curved, concave to the axis of the ray, and
point towards its distal end. These distal rays consequently somewhat resemble
wheat-ears. The axial thread extends quite to the tip of the ray, its end is not
covered with sihca (Plate 25, figs. 22, 24). The proximal and the lateral rays
are curved, cylindroconic or conic, at the base about as thick as the basal part
of the distal ray, smooth in their proximal part, and covered with minute oblique

HOLASCELLA EUONYX. 45

spines, inclined towards the end, in their distal part. The lateral rays'are 215-
420 (U long and often thickened at the end. The proximal ray attains a length
of 400-530 fi. The fragmentary state of the specimen renders it impossible to
determine which of the superficial hexactines observed are dermal and which
gastral.

The tetractine principal spicules (Plate 25, fig. 16) have two long rays extend-
ing longitudinally and two shorter transverse rays. The four rays do not he
in one plane. The rays are 80-140 m and more thick at the base. About their
length I cannot be definite, since the larger spicules of this kind were invariably
broken. The longest longitudinal ray-fragment observed was 19 mm. long.

The tetractine comital spicules (Plate 25, fig. 18) are similar to the principal
ones, but have rays usually only 9-17 ix thick.

The large loose hexactines (Plate 25, figs. 19, 20) have straight or curved,
equal or unequal rays, which arise from a distinct central thickening, 38-50 n
in diameter. The rays are 0.3-1.7 mm. long, at the base 10-35 yu thick, usually
10-15 n, and smooth or, more frequently, covered with sparse fairly stout, low
spines.

TheswioZ? hexactines measure 120-150 m in diameter, and have straight, conic
rays, 6-7 ix thick at the base, and densely covered with rather large spines.

The onychhexactines and hemionychhexasters (Plate 24, figs. 13, 14; Plate
25, figs. 1, 6-9, 13b) are both derivates of onychhexasters, and there is no differ-
ence between them, except that in the former (Plate 25, figs. 7, 9) all the main-
rays bear only one end-ray, while in the latter (Plate 25, fig. 8) one or two of the
main-rays bear two end-rays. When, as is the rule, only one end-ray is present,
this either extends exactly in the continuation of the main-ray (Plate 25, fig. 9),
or there is a slight, abrupt curvature at the point where the main-ray passes into
the single end-ray (the upper ray, Plate 25, figs. 7, 8). In any case the main-
and the single end-ray together form a ray, simple in outer appearance. That
these apparently simple rays are in truth composed of a main-ray and a (single)
end-ray is, however, clearly shown by the axial thread, which is only 7-8 fi long,
and present in the basal (main-ray) part of the ray only. These onychhexac-
tines and hemionychhexasters measure 53-95 n in total diameter. Their simple
rays are 25^5 m long. The main-rays which bear two end-rays are, like the
axial threads of the simple rays, 7-8 m long. The simple rays are 2-3 fi thick at
the base and taper distally to 1-1.5 m- They are either smooth throughout, or
slightly roughened bj^ exceedingly minute spines in their basal and middle-parts.
Each ray (end-ray) bears at its end a verticil of four or, more rarely, three large

46 HOLASCELLA EUONYX.

curved, conic spines, 7-15 n long. The basal part of these spines is directed
outward, slightly upward, and usually encloses an angle of 105°-102° with the ray.
Their ends are bent downwards, towards the centre of the spicule. These spines
are regularly arranged and, when four in number, form a regular cross.

The small discohexasters (Plate 24, figs. 10-12, 13b, 14b, 15-17; Plate 25,
fig. 13a) measure 38-44 fi in total diameter. They have a centrum 3.3-4 ^
in diameter, from which six equal and regularly arranged main-rays arise. The
main-rays are cylindrical, 6.5-9 /j. long, 1-1.4 /^ thick, and simply rounded off at
the end. About 1 n below the end each main-ray bears a high frill, which appears
as a round, subterminal, transverse disc 5-7 n in diameter. From the margin
and the upper distal face of this disc very numerous diverging end-rays arise,
which together form a short and broad bunch, at the distal end 19-25 /; in diame-
ter. The individual end-rays are, at the base, curved, concave to the continua-
tion of the main-ray axis, and in their distal and middle-parts straight. They
are 13 m long, throughout about 0.2 n thick, covered with exceedingly minute,
recurved spines along their length, and crowned at the end with a verticil of
similar but larger spines. These terminal spines together form a kind of end-
disc, generally a little less than 1 ;u in transverse diameter.

The large hemidiscohexasters and discohexadines are very similar and
differ from each other only in that one of the main-rays bears two end-rays
(Plate 25, figs. 2-5, 10, 11) in the former, whilst all six main-rays bear only
one end-ray in the latter. The large discohexactines measure 173-232 n in
total diameter, usually 194-215 /i. Their six simple rays are fairly equal and
regularly arranged, straight or slightly and uniformly curved, and sometimes
just perceptibly abruptly bent at the point where the short basal part, which is
the main-ray, passes into the long distal part, which is the single end-ray. The
short basal (main-ray) part of the ray contains an axial thread 6-7 n long;
6-7 tx is accordingly the length of the main-ray. The long distal (end-ray)
part is destitute of an axial thread. The rays of the large discohexactines are
95-110 n long and thickened at both ends. They measure at the base 4.5-6 n,
at the thinnest point, which lies somewhere near the middle of their length,
2.4-5 M, and at the distal end 5-6.5 n in transverse diameter. Along their
length these rays are either quite smooth or bear a few minute, recurved spines.
The end is crowned by a terminal verticil of usually five or six recurved spines,
7-12 M long, and 1.8-4 fi thick at the base. These spines are conic, uniformly
recurved and rather sharply pointed; together they form an exquisite anchor
15-22 n broad and 10-16 fx high.

CAULOPHACUS. . 47

The nearest ally to this sponge appears to be Holascella ancorata. From
this it is distinguished by its superficial hexactines having more slender distal
rays, by its discohexactines having smoother rays and larger terminal anchor-
teeth, by the terminal spines of its onychhexactines and hemionychhexasters
being much longer, and by possessing no floricomes. On account of its general
similarity to Holascella ancorata I assign it to the genus Holascella. It must,
however, be borne in mind that the fragmentary condition of the specimen
precludes the possibiUty of saying with certainty whether it really belongs to
this genus, for if the sponge of which it formed part should have been destitute
of a root-tuft, which is quite possible, it would have to be placed in Corbitella or
another genus of the Corbitellinae. In this respect it is noteworthy that its
discohexactines are rather sinular to the discohexactines of Corbitella (Eudictyon)
elegans Marshall.^

CAULOPHACIDAE F. E. Schulze.

Wineglass- or mushroom-shaped Hexasterophora with a firm stalk; soli-
tary or forming branched colonies. With dermal pinules and large hypodermal
pentactines.

The collection comprises thirty more or less complete specimens and eighty-
three fragments of specimens of this family. The position of three of the latter
is doubtful. The others belong to the three genera Caulophacus, Caulophacella,
and Calj^cosilva; the last two of these are new.

CAULOPHACUS F. E. Schulze.

Mushroom-shaped Caulophacidae with hollow stalk, discohexasters, and
microdiscohexactines .

There are twenty-eight more or less complete specimens and forty-nine
fragments and stalks of Caulophacus, all of which belong to the same species.

' W. Marshall. Untersuchungen iiber Hexactiuelliden. Zeitschi-. wiss. Zool. Suppl., 1875, 25,
p. 211, 212, taf. 16, figs. 66 a-1. /. Ijima. Studies on the Hexactinellida. II. Journ. Coll. sci.
Tokyo, 1902, 17, p. 11-16, pi. figs. 13-15.

48 CAITLOPHACUS SCHULZEI.

Caulophacus schulzei Wilson.
Plate 7, figs. 20-31; Plate 8, figs. 1-29; Plate 9, figs. 1-33; Plate 10, figs. 1-29; Plate 11, figs. 1-17.
Mem. M. C. Z., 1904, 30, p. 43; Plate 4, figs. 1, 3, 5-10; Plate 5, figs. 1-6, 8-10.

All the specimens referred to this species were trawled at Station 4651 off
northern Peru on 11 November, 1904; 5° 41.7' S. 82° 59.7' W. ; depth 4063 m. (2222
f.); they grew on sticky, fine, gray sand; the bottom-temperature was 35.4°.

Apart from peculiarities due to differences of age and preservation, all
these sponges are fairly identical. The nearly complete specimens are mush-
room-shaped, composed of a discoid body and a stalk attached to the lower
face of the disc. The fragments appear to be parts of similar sponges. Six
specimens have been selected for detailed study, and to these all the figures
on the plates refer. These specimens are marked A-F. A, B, and C are small
specimens with discs 27-31 mm. in diameter. D, E, and F are large specimens.
D had a disc 60 mm. in diameter. E was probably still larger, but is too frag-
mentary for exact measurement. F is a detached stalk which appears to have
belonged to a specimen with a disc also about 60 mm. in diameter.

Shape and size. In the smallest nearly complete specimen, the disc-shaped
body is oval in outline, 19 mm. long, 16 mm. broad, and 2.5 mm. thick in the
middle. Towards the margin it thins out. The central part pf the upper, gas-
tral face is flat, its marginal part shghtly convex. The stalk is eccentric, obhque,
2 mm thick close to its point of insertion to the sponge-body (disc), and atten-
uated below. In seven of the nearly complete specimens the disc is fairly flat,
circular to oval in outline, 24-35 mm. in maximum diameter, and 5-7 mm. thick
in the middle. One of these small specimens is represented on Plate 9, fig. 30.
In these specimens the central part of the upper, gastral face is flat, slightly
concave or slightly convex, the marginal part usually distinctly convex. The
proximal end of the stalk is 2-5 mm. thick. The eccentricity of its point of
insertion varies considerably and is in one of the specimens so great that its
distance from the farthest point of the margin is thrice that of its distance from
the nearest. In one small specimen (Plate 9, fig. 29) the disc is a little over
30 mm. in diameter, 7 mm. thick in the middle, and folded in above. The upper,
gastral face is, apart from the remarkable infolding, nearly flat in the middle
and strongly convex towards the margin. The lower, dermal face is convex in
the middle and flat near the margin. The margin itself is very clearly defined
and sharp (Plate 8, figs. 28b, 29b). The stalk is, close to its point of insertion,
4.5 mm. thick; 8 mm. lower, where it is broken off, it is only 2.2 mm. thick.

CAULOPHACUS SCHULZEI. 49

The remaining nineteen specimens, one of which is represented on Plate 9,
fig. 28, are larger. Their irregularly o^^al discs are 40-64 mm. long, 34-54 mm.
broad, and 7-12 mm. tliick in the middle. The more or less eccentric and
oblique stalk is, near its point of insertion, 2.5-7.5 mm. thick and quite rapidly
attenuated below. The disc is flat, slightly convex or concave. The greater
part or the whole of the marginal portion of the upper face is convex, so that
the margin appears shghtly bent down.

The stalk is not intact in any of the specimens, but there are among the
fragments three rather long stalks with intact lower (distal) end. These are
30-40 mm. long, curved, particularly near the base, and 2 mm. thick (at the
lower end) to 3.3 mm. (at the upper end). One of these stalks (Plate 9, fig. 28)
appears to have been torn off the larger specimen. In the photograph this
stalk is artificially attached to it.

The specimens examined by Wilson (loc. cit., p. 43, Plate 4, fig. 3) were
smiilarly composed of a calyculate, flat, or somewhat convex disc-shaped body,
22-50 mm. in diameter, and a stalk invariably broken.

The colour of all the specimens in spirit is brownish gray.

General structure. Remnants of a superficial membrane supported by the
lateral pinule-rays can be made out both on the dermal and the gastral faces
of the sponge. This membrane lies on both sides, 70-100 ju above the level
occupied by the lateral rays of the hypodermal and hypogastral pentactines.
In the intervening space shreds of tissue are observed, indicating that in life
this zone was occupied by a network of trabeculae. Below the level marked by
the lateral pentactine rays subdermal and subgastral cavities occur, which lead
into canals extending more or less transversely, often through the greater part
of the thickness of the whole disc (Plate 8, figs. 28, 29; Plate 9, fig. 32). The
entrances to these canals are clearly visible, both on the dermal and the gastral
face of the disc-hke body. WTiere the superficial membrane is still present,
they are covered by it; where tliis membrane has been lost, as is the case on
nearly the whole of the surface in most of the specimens, they are freely exposed.
The apertures of the dermal face resemble in shape and arrangement those of
the gastral face, but are on the whole somewhat larger. The largest are always
formed on the central part of the disc. Towards the margin they become
smaller. Their distance from each other is in proportion to their size; the
marginal ones he much closer together than the central ones. The largest cen-
tral apertures are 0.8-4 mm. wide, their width being, on the whole, in proportion
to the size of the specimen. Apertures over 3 mm. in diameter have been

50 CAULOPHACUS SCHULZEI.

observed only in specimens with discs more than 50 mm. long. Apart from this
it is also to be noticed that the smaller these apertures are, the better the speci-
men is preserved. Their great width in indifferently preserved specimens is
probably due to extreme post mortem shrinkage of the soft parts.

The canals into which these apertures lead are 0.2-1.5 mm. wide in the best
preserved specimens (Plate 8, figs. 28, 29). In specimens not so well-preserved
the largest attain, probably in consequence of excessive shrinkage of the soft
parts, a width of 3 mm. (Plate 9, figs. 32, 33).

The spaces between these canals are occupied by a dense, readily stained
tissue in which are observed traces of oval cavities 120-140 yu long and 70-90 ^
broad, which may be remnants of the walls of the flagellate chambers.

The stalk is hollow (Plate 9, figs. 27, 33a). I failed to find any open com-
munication between the cavity of the stalk and the wide canals of the body
proper.

The stalk is supported by a tubular network (Plate 9, fig. 27; Plate 10,
figs. 8, 13, 14) composed of many longitudinal and a few transverse rhabds and
other megascleres, joined by apposed silica, which solders these spicules together
where they come in contact, and which forms short rods connecting adjacent
spicules. It is to be noted also that pinules are embodied in this network (Plate
10, fig. 8a). The longitudinal rhabds, of which this skeleton-net is chiefly com-
posed, are in the outer part of the tube situated longitudinally. Towards
its inner surface their position becomes on the whole more oblique, and here
transverse rhabds also occur. The (mostly longitudinal) beams of the network
are usually 20-60 yu thick, their (mostly transverse) secondary connections
usually 4-12 m- Above, where the stalk passes into the body of the sponge, the
longitudinal rhabds of its skeleton become free.

The megascleres of the body are chiefly pinules, hypodermal and hypo-
gastral pentactines, hexactines, and rhabds. Besides these a few large, not
hypodermal or hypogastral, pentactine hexactine-derivates have been observed.
The hexactines and pentactine hexactine-derivates are scattered throughout
the choanosome. Some of the rhabds are isolated, most of them form bundles,
which traverse the interior obliquely and extend paratangentially some distance
below the surface. Hypodermal and hypogastral pentactines, with paratan-
gential lateral rays and an apical ray directed radially inward, are noticed under
the dermal and gastral surface. The whole of the surface is occupied by
pinules, the lateral rays of which extend paratangentially in the superficial
membrane. The gastral pinules and the dermal pinules of the body and the

CAtJLOPHACUS SCHULZEI. 51

stalk are quite similar. Nearly all the pinules are regularly hexactine. There
are numerous forms of microscleres. These can be classified in two groups not
connected by transitions. The first group comprises hexasters, hemihexasters,
and hexactines, the rays (end-rays) of which are, when young, smooth, and
sharp-pointed, when adult covered with numerous large lateral spines and
crowded with a verticil of terminal spines. The young forms of this group
appear as oxy-, the adult forms as disco-hexasters, -hemihexasters and -hexac-
tines. The second group comprises discohexasters with generally smooth
main-rays, from the ends of which arise regular verticils or bunches of slender
end-rays. The end-rays are densely covered with small lateral spines, and
crowned with a verticil of terminal spines. The spicules of this group appear to
replace, in this and other species of Caulophacus, the plumicomes of Sympagella
and Calycosilva. For this reason and because they differ very considerably from
the discohexasters of the other group of microscleres I think it better not to
describe them as discohexasters, as previous authors have done, but to give them
another name, discocomes.

The discohexasters, etc., occi^^y the choanosome in dense masses. One
of the rays of those situated in the walls of the large choanosomal canals is
usually directed canalwards and protrudes into the canal-lumen. The walls
of these canals therefore appear somewhat spiny and the spicules rendering them
so might be considered, to a certain extent, as canalaria.

The discocomes are met with chiefly in the subdermal and the subgastral
region, and here occasionally form clusters in which large and small ones are
irregularly intermingled.

The rhabds of the stalk (Plate 10, figs. 11, 12) are 14-28 fi thick near the end.
The end itself is more or less thickened. This terminal thickening is greater
in the stout, than in the slender rhabds. When great it gives to the rhabd-
termini the appearance of oval tyles. The thickened end-part (tyle) measures
18-38 n in transverse diameter, and is 4-12 yu thicker than the adjacent parts of
the spicule. This more or less thickened end-part is densely covered with small
spines; the remainder of the spicule is smooth. The terminal region occupied
by the spines is 4^60 fx long.

The rhabds of the body proper (Plate 10, figs. 1-7, 9, 10) are more or less,
sometimes very considerably cm"ved, slightly attenuated toward the rounded,
usually somewhat anisoactine ends, centrotyle, and everywhere smooth, except
at the ends. The end-parts are covered with small spines, and sometimes slightly
thickened. These rhabds are 1.2-4.3 nmi. long, measured along the chord

52 CAULOPHACUS SCHULZEI.

connecting their ends, and 7-26 m thick near the middle. The central tyle
measures 11-32 fi in transverse diameter and is 2-9 ti, on an average about 4.4 yu,
thicker than the adjacent parts of the spicule. In the central tyle a transverse
cross is observed, composed of four rudimentary axial tlireads, about 1 ii long.
At the ends, these rhabds are 0-4 ju thinner than near the centre. Their aniso-
actinity is not great, the difference in the thickness of the two ends usually not
exceeding 2-3 m- The terminal spiny regions are 20-45 n long. The spines in
them stand close together, attain a length of about 1 ti, and arise vertically.
They are either straight or slightly curved backwards, towards the middle of
the rhabd, at the end.

In comparing the measurements of the rhabds of the body of the small
specimen (with a disc about 30 mm. in diameter) with those of the rhabds of
the large specimen D (with a disc about 60 mm. in diameter), I found no per-
ceptible difference in their length, but a well-pronounced difference in their
thickness, the body-rhabds of the 30 mm.-specimen B being 10-21 /x thick and
having central tyles 15-26 n in diameter; those of the 60 mm.-specimen D being
10-26 M thick and having central tyles 16-32 /i in diameter.

Wilson {loc. cit., p. 45) states that in the specimens examined by him the
rhabds were 1-4 mm. long, usually 1.5-2.5 mm.; 8-12 yu thick, exceptionally
24//; and " subterminally roughened with microtubercles." To me the sub-
terminal protuberances appear as sharp-pointed spines and I should not call
them "microtubercles."

The hexadine megasderes of the dioanosome (Plate 7, figs. 20-31) measure
1.2-3.2 mm. in total diameter. The rays of the same hexactine are fairly equal
in thickness but differ more or less, often very considerably, in length. In
many hexactines the longest ray is two to three times as long as the shortest.
The rays are 250 /i-1.4 mm. long, conic, blunt, 25-74 yu thick at the base, and
7-18 fi just below the end. They are more or less, often considerably, curved,
rarely angularly bent (Plate 7, fig. 28). The long rays are invariably smooth
and attenuated toward the end. The rays reduced in length are either conic,
pointed, and spiny (Plate 7, fig. 21) or, more rarely, cylindrical, terminally
thickened, and smooth (Plate 7, fig. 29).

In the shortened conic and spined rays there is a correlation between the
number and size of the spines on the one hand, and the degree of longitudinal
reduction of the ray on the other, the development of the spines being in pro-
portion to the degree of reduction. Here, as in the similar case of Calycosilva
cantharellus, this correlation between spine-development and reduction in length

CAULOPHACUS SCHULZEI. 53

is probably due to the potential energy of the silicoblasts building the short rays
being partly diverted from their normal use of forming long rays and converted
into the work of producing spines. The longitudinal reduction of the rare,
smooth, terminally thickened, shortened rays is obviously of a different nature,
the potential energy of the silicoblasts being in this case diverted in another
direction. The difference of these two kinds of reduction is probably attribut-
able to a difference in the cause of the reduction.

There is no perceptible difference in the dimensions of the hexactine choano-
somal megascleres of the small (B, 30 mm.-disc) and large (D, 60 mm.-disc)
specimen.

Wilson {loc. cit., p. 44) found the hexactine rays 0.7-1.2 mm. long, and
28-48 IX thick at the base. He occasionally observed hexactines with spines
on all rays, but does not mention the forms with spines on the reduced ray only.
In his figm-e {loc. cit., Plate 5, fig. 10) all the hexactines are drawn with stout,
straight, and equal rays. My photographs (Plate 7, figs. 20-31) show that
in the material examined by me their appearance is different. Since, however,
the figure (Plate 5, fig. 10) of Wilson is a general view of a section, I believe
myself justified in assuming that this difference is not real but merely due to the
hexactines in the figure cited having been di-awn schematically.

The rare pentactine hexactine derivates (Plate 8, figs. 23, 24) are, apart
from the suppression of one of the rays, similar to the hexactines. Some of
them bear spines on all the rays. These pentactines are 1.1-3.1 mm. in
diameter. The longest of their usually unequal rays, which may be the un-
paired apical one or another, is 0.8-1.6 mm. long, their shortest ray 0.4-1.4 mm.
The rays are 30-55 m thick at the base.

The hypodermal and hypogastral pentactines (Plate 8, figs. 1-7, 12-22, 25-27)
are very similar. Theu- lateral rays are either all properly developed and
fairly equal (Plate 8, figs. 19-21, 25-27), or one, two, or three of them are more
or less reduced in length, shorter than the others (other), and also, if more than
one, unequal in length among themselves (Plate 8, figs. 12-14, 18, 22). The
properly developed lateral rays are straight or slightly curved, conic, and very
blunt. Their proximal part is either smooth or it bears a larger or smaller num-
ber of spines. Farther on, and up to a short distance from the end, they are
nearly always smooth. The end itself is either also smooth or densely covered
with small spines. The proximal spines extend, when present, from over a
quarter to nearly a half of the length of the ray. They are low, broad, pointed,
and conic. In regard to their number the lateral rays, even of the same spicule,

54 CAULOPHACUS SCHULZEI.

are very unequal (Plate 8, figs. 13-15). Sometimes there are a good many,
sometimes but a few, occasionally only one, and not infrequently none at all.
When the spines are numerous a spiral arrangement of them can occasionally
be made out quite distinctly (Plate 8, figs. 13, 15).

When the lateral rays are markedly reduced in length, their ends are usually
thickened (Plate 8, figs. 12, 14, 18, 22). The terminal thickening is either spiny
(Plate 8, fig. 14) or smooth (Plate 8, fig. 12).

The lateral rays are 250 /x-l.l mm. long, and 25-65 ij. thick at the base.
The properly developed long ones taper gradually to the rounded end which
is 5-12 IX thick. The ends of the longitudinally reduced and terminally thick-
ened lateral rays measure 15-25 n in transverse diameter. In some specimens,
as for instance in the small one with 30 mm.-disc, the lateral rays of the hypo-
dermal pentactines are slightly shorter than those of the hypogastrals, those
of the former sometimes measuring in this specimen 0.9 mm. in length, those of
the latter 1 mm. In the other specimens, as for instance in the large one D
with 60 mm.-disc, the lateral rays of the hypodermal pentactines are consider-
ably longer than those of the hypogastrals, those of the former measuring, in
this specimen up to 1 mm., those of the latter only up to 670 m in length.

The proximal ray is straight or shghtly, often irregularly, curved, and tapers
gradually to the blunt or rounded end. The basal part of the proximal ray is
smooth for a short distance, then follows a region which usually bears broad
and low, conic spines. The number of these spines is variable. Sometimes
(Plate 8, figs. 7, 16, 17) there are a good many, sometimes only few (Plate 8,
figs. 1, 5), and sometimes none at all (Plate 8, fig. 6). When these spines are
numerous, the portion of the proximal ray bearing them is usually more or less
thickened. The distal and the middle-part of the ray are generally smooth.

The proximal ray is 600ju-l.l mm. long, and 30-55 yu thick at the base.
The maximum thickness of the spiny part is 1-4, rarely as much as 7 ju greater
than the thickness of the base of the ray.

The difference in the length of the proximal ray of the hypodermal and
hypogastral pentactines is similar to, but not so great as, the difference in the
length of their lateral rays. In the small 30 mm.-disc specimen B the proximal
ray of the hypodermals is not over 0.95 mm. long, that of the hypogastrals not
over 1.1 mm. In the larger 60 mm.-disc specimen D the proximal ray of the
hypodermals is not over 1.1 mm. long, that of the hypogastrals not over 0.96 mm.

Also in the specimens examined by Wilson {loc. cit., p. 46) the hypodermal
and the hypogastral pentactines were very similar. Wilson {loc. cit., p. 46, 47)

CAULOPHACUS SCHULZEI. 55

remarks that there is no trace of a (sixth) distal ray. In the pentactines of the
specimens examined by me, such a trace occurs as a short continuation of the
axial thread of the proximal ray beyond the centrum. The measurements
given by Wilson {loc. cit., p. 47) are: — lateral rays 0.4-0.75 mm. by 36-48 m,
proximal ray 0.78-1 nmi. by 50-60 m- In his drawing (loc. cit., Plate 4, fig. 9)
of a pentaetine the spines are much smaller than in the pentactines examined by
me vdth. distinctly spined proximal ray (Plate 8, figs. 5, 7, 16, 17).

The pinnies (Plate 11, figs. 1-16, 17a) are nearly always regularly hexactine.
Only exceptionally a pinule with a rudimentary proximal ray or some other
abnormity is met with. Apart from certain differences in their dimensions,
which will be dealt with below, the dermal pinules of the upper part of the stalk,
the dermal pinules of the body, and the gastral pinules are identical. The
lower parts of the stalks at my disposal are denuded of their pinules, so that I
am unable to say what these may be Uke. Probably they are similar to those
of the upper part of the stalk, but smaller.

The distal pinule-ray is straight, 140-390 fi long, and 8-23 ^ thick at the
base. Above the ray thickens, and it attains its maximum transverse diameter
a httle below the middle of its length, where it is usually a third to twice as thick
as at the base. In four pinules measured, the thickness of the distal ray was: — •

at the base 14, at the thickest point 25 /x
" " " 16, " " " " 28 m

" " " 18, " " " "28 m

" " " 20, " " " " 35 m

Beyond the thickest place it is attenuated, at first slowly, then rapidly, to a
rather sharp point. It consequently appears spindle-shaped. It is covered
throughout with sharp-pointed, conic spines. The proximal spines are very
small. Farther up they become larger and they increase in size to the middle
of the ray, where they measure 18-22 m by 5-8 m- From here onwards they
again become smaller, but only ^'ery sUghtly, the uppermost spines still being
quite large. The proximal spines arise vertically, those farther up obUquely,
and the nearer we approach the middle of the ray the more inclined towards
its tip do their basal parts become. The small proximal spines are straight,
those farther up slightly curved towards the tip of the ray. The maximum thick-
ness of the distal ray, together with its spines, is 25-70 m-

The proximal ray is, when normally developed, straight, 70-145 m long,
8-17 M thick at the base, and attenuated towards the blunt end, at first gradually.

56 CAULOPHACUS SCHULZEI.

then abruptly. Its distal part bears nmnerous fairly large spines, its proximal
part fewer and smaller ones. Sometimes this part of the ray is nearly smooth.
Exceptionally the proximal ray is reduced or hypertrophied. When reduced
it is 7-40 ;u long, more or less cylindrical, as thick throughout as the normal
proximal ray at its base, and terminally rounded. A hypertrophic proximal
ray observed (Plate 11, fig. 13) was 150 ti long, considerably thickened in the
middle, attenuated to a rather sharp point, and densely covered with large
spines. It measured at the base 16 ^ and at its thickest point 25 ix in transverse
diameter. Another still more hypertrophic one in all respects resembled the
(opposite) distal ray.

The lateral rays are similar to the proximal ray, but more frequently smooth
in their basal part. They are 80-157 ix long, and 8-18 y. thick at the base.

The lateral rays of the same spicule are usually fairly equal. But in one
quite abnormal pinule which I found in the large specimen D they were very
unequal. In this remarkable spicule two adjacent lateral rays were hyper-
trophic, covered with long spines, and similar to the distal ray of an ordinary
pinule, whilst the other two laterals were normal and the spinulation of the
distal ray so much reduced that it resembled the (normal) proximal ray.

I measured a good many pinnies of four different specimens (B, C, D, and E).
The results of these measurements are tabulated on p. 57. Specimen E, which
was taken for detailed study because it appeared to be a part of a specimen
larger than any of the nearly complete ones, was too fragmentary to allow of
a distinction between its dermal and gastral faces. The dermal and gastral
pinules of this specimen are therefore not distinguished in the table.

From this tablp the following conclusion concerning the differences in the
dimensions of the pinules can be drawn. There is, in specimens of similar
dimensions (B and C), a not inconsiderable variation in the dimensions of the
pinules, particularly the length of the distal ray. In the large specimen D all
the rays of the body-pinules attain a greater thickness and the proximal and
lateral rays also a greater length than the corresponding rays of the correspond-
ing pinules (dermal and gastral) of the small specimens B and C. The rays
of the dermal pinules are thicker at the base than the corresponding rays of
the gastral pinules of the same specimen. This applies also to the maximum
thickness of the distal ray with its spines in specimens C and D, but not in B.
The distal rays of the gastral pinules attain a considerably greater length than
the corresponding rays of the dermal pinules of the same specimen. The other
rays are in some specimens longer in the dermal, in others longer in the gastral

CAULOPHACUS SCHULZEI.

57

PINULES.

Di.stal apical ray

Proximal apical
ray

Lateral rays

Length

'a

d

a

Is

SB
■^ a

If

Length

<»

i

■3
'3

Length

03

i

dermal body

limits ti

215-
340

10-19

25-40

100-
120

10-12

90-
140

10-14

Specimen B

average of the
largest three ii

289

16

37

112

12

128

13

(30 mm.-disc)

gastral

limits It

240-
390

10-15

26^5

90-
110

8-13

115-
125

10-11

average of the
largest tlu-ee m

337

13

38

103

10

120

11

dermal body

limits M

190-

228

10-16

32H10

95

13

125

12

Specimen C

average of the
largest three m

216

13

38

95

13

125

12

(30 mm.-disc)

gastral

limits 11

280-
300

8-12

28-32

90-
100

average of the
largest three n

298

12

32

100

der-
mal

upper
part

of :5

stalk

limits M

140-
230

8-17

30-50

70-
105

80-
120

average of the
largest three m

217

12

45

97

110

Specimen D

body

limits fi

210-
310

11-21

30-68

75-
124

10-17

90-
170

9-15

(60 mm.-disc)

average of the
largest three m

2S7

21

65

115

15

150

14

gastral

limits li

216-
365

13-19

37-57

SO-
US

8-15

100-
165

9-15

average of the
largest three n

332

18

56

127

14

138

13

Specimen E (disc probably ,. .
over 60 mm.); body **

190-
385

10-23

32-70

90-
133

8-17

90-
157

8-18

58 CAULOPHACUS SCHULZEI.

pinules. The dermal pinules of the upper part of the stalk are in all dimensions
smaller than the dermal pinules of the body of the same specimen. This differ-
ence is greatest in respect to the basal thickness of the distal ray.

In the specimens examined by Wilson {loc. cit., p. 45) the proximal and lateral
pinule-rays measure about 100 by 8-10 /jl. These rays, particularly the laterals,
are, according to this, considerably smaller in Wilson's specimens than in those
examined by me. The distal pinule-ray is, according to Wilson (loc. cit., p. 45),
covered with narrow scales not over 16-20 fx long. His measurements agree
with mine, but it does not seem correct to call these structures scales. As my
photographs (Plate 11, figs. 15, 16) show, they are ordinary conic spines with a
fairly circular transverse section. Wilson {loc. cit., p. 46) gives sets of measure-
ments of the distal pinule-rays of two specimens, one in which they are slender
and one in which they are stout. The distal ray of the dermal pinules is (together
with its spines) in the first 240-320 /x by 36-40 n, in the second 210-240 ^ by
44-56 fi ; the distal ray of the gastral pinules in the first is 260-360 ju by 32-36 n,
in the second 280-320 a by 36-40 a.

The oxyhexasters, hemioxyhexasters, and small oxyhexactines, found in small
numbers in several specimens, I at first took for skeletal elements sui generis.
A careful search, however, revealed the presence of a few spicules connecting them
with the discohexasters, hemidiscohexasters, and discohexactines, and the
examination of these transitional forms made it clear that these oxyhexasters,
etc., are young forms of the discohexasters, etc.

The young oxyhexaster-like etc. of discohexasters etc. (Plate 9, figs. 14-16)
measure 60-190 n in total diameter, and have six rays. These are 2-4 ^ thick
at the base, conic, sharp-pointed, perfectly smooth, and either simple or provided
with one or two, rarely three, branch-rays. The spicules of this kind with all
six rays simple, which appear as small oxyhexactines, are very rare; in the
majority one or more (Plate 9, figs. 14, 15) or, less frequently, all rays (Plate 9,
fig. 16) bear branches. The simple rays are straight. Those bearing branches
diverge a little above the branching point shghtly in a direction opposite to
that in which the branch lies, but are straight apart from this divergence. The
branches arise at a distance of 12-20 fi from the centrum of the spicule, steeply,
sometimes nearly vertically, from the rays, but very soon curve sharply outward
and then again become fairly straight, remaining so to the end. Like the main-
rays themselves, these branches (branch-rays) are conic, sharp-pointed, and
perfectly smooth. When a ray bears more branches than one, they usually
arise at the same point and diverge in different directions. Exceptionally I

CAULOPHACUS SCHULZEI.

59

have seen two branches arising at different points and extending in nearly the
same direction.

Wilson {loc. cit.) does not mention such spicules as occurring in the specimens
examined by him.

In the rare transitional (adolescent) forms described above, connecting the
oxyhexasters, etc., with the fully developed discohexasters, etc., the rays and
branches are quite smooth along their length but crowned at the end by small
verticils of recurved spines. By an increase in the thickness of all parts, by a
growth of the terminal spines, and by the addition of lateral spines along the
length of the rays and branches, these adolescent forms become adult discohex-
asters, etc.

The adult discohexasters, hemidiscohexasters, and discohexactines (Plate 8,
figs. 10, 11; Plate 9, figs. 1-7, 9-13, 17-26; Plate 10, figs. 27a, 28a, 29a) have,
including the branches, six to seventeen rays and measure 139-264 fi in total
diameter. In the large specimens D and E these spicules attain a larger size
(diameter of largest 264 and 260 /j respectively) than in the smaller specimens B
and C, where the largest measured were only 240 fi in diameter. The basal
thickness of the rays is 6-15 m- The total diameter, and to a certain extent also
the basal thickness, of the rays are, as the subjoined table shows, in inverse
proportion to the number of rays (branches) .

DISCOHEXASTERS, HEMIDISCOHEXASTERS, AND DISCOHEXACTINES.'

Number of rays

Diameter of the spicules

Basal thickness

and branches

limits It

average of the
largest three fi

of the rays n

6

170-264

254

7-15

7-8

170-250

245

7-15

9-12

140-230

238

7-15

13-17

139-195

161

6-8

In the large specimens D and E the discohexactines with six simple un-
branched rays are the most frequent. In the small specimens B and C on the
other hand the majority of these spicules are hemidiscohexasters and disco-
hexasters with branches on one to all six rays.

' This table is based on the measurements of the discohexasters, etc., of all the specimens examined.

60 CAULOPHACUS SCHULZEI.

Except in the vicinity of the branching points, the main- and branch-rays
are fairly straight. All the rays, both main and branch, are conic and gradually
attenuated to the end, which is 2-4 n thick. The simple stem-like basal part
of the branch-bearing rays is smooth. Apart from this the rays and branches
are entirely covered with stout, pointed, and strongly recurved spines 2-4 yu
long. These spines are quite uniformly scattered along the length of the rays
(branches) and congregated at their ends, where they form terminal verticils
or bunches 7-10.5 fi in transverse diameter. Particularly when viewed with
lower powers the terminal spine-verticils more or less resemble convex discs with
deeply serrated margin. The lateral spines decrease in size toward the distal
ends of the rays (branches). The most distally situated spines, which form the
terminal verticil or bunch, are much larger than the adjacent lateral ones, about
as large as the basal lateral ones. Exceptionally one single hypertrophic ter-
minal hook-like spine replaces the verticil or bunch.

The distance of the branching point of the rays from the centrum of the
spicule, that is the length of the simple stem of the branch-bearing rays, is 15-22 m-
These stems are very short accordingly, compared to the size of the whole spicule.
Generally there is only one branch on a ray, but two or three are also frequently
met with. More than three are rare. The largest number of branches on one
ray observed was six. When there are more branch-rays than one, they arise
either at the same or at different levels. When the number of branch-rays is
great the latter is the rule. In most cases the main-ray is clearly distinguished
from the branch-ray or branch-rays by its slighter divergence from the continua-
tion of the axis of the stem, and by its greater length (Plate 9, fig. 21). Some-
times, however, there is no such distinction, the distal part of the main-ray
being as long and diverging as much as the branch, and the stem appearing to
divide into equal branches (Plate 9, fig. 18). The angle between the distal
part of the main-ray and the branches is variable, most frequently about 45°.

Wilson {loc. cit., p. 48) states that in the specimens examined by him the
rays of the discohexactines were 80-110 ^ long and 8 n thick at the base, their
terminal spine-verticils or branches being 10-12 ^ in diameter and appearing
as watch-glass shaped end-discs. To me these groups of spines, which are
correctly represented in Wilson's figures (loc. cit., Plate 5, figs. 4, 5, 9), do not
appear as watch-glass shaped end-discs.

The discocomes (Plate 10, figs. 15-26, 27b, 28b, 29b) normally consist of
six main-rays joined at right angles, each of which bears a terminal verticil or
bunch of end-rays. One discocome I saw had seven main-rays. The discocomes

CAULOPHACUS SCHULZEI. 61

measure 44-312 n in total diameter. The main-rays and end-ray verticils or
bunches of the same spicule are equal. Exceptionally (Plate 10, fig. 17) one
or two end-rays arise from a main-ray below the terminal verticil or bunch.
In one discocome one of the main-rays bore a stout branch, which was crowned,
like a main-ray, by a bunch of end-rays. In respect to these irregularities the
rays of the same spicule are unequal. The main-rays are straight, 20-65 ^
long, cylindrical, and 1.5-7 yu thick. At their distal end they are abruptly thick-
ened to an inverted cone or convex disc, 7-16 ^ in diameter, from the distal
face of which the end-rays arise. The main-rays are generally perfectly smooth,
exceptionally they bear a few rather large spines. The axial thread of the
main-ray ends abruptly in the terminal thickening and does not give off branches
for the end-rays; the latter appear to be destitute of axial threads. The number
of end-rays on each main-ray is from six to eighteen or more. When their num-
ber is great, it is exceedingly difficult to count them. When few in number
they form a verticil, when more numerous, a bunch or brush, in which they are
fairly equidistant. The terminal verticils or bunches formed by the end-rays
appear as inverted cones with apical angles of 25-100°. The individual end-rays
are straight or curved in an S-shaped manner with outwardly directed distal end.
Measured along their chord they are 18-103 fi long. They are 2-3 n thick at
the base and taper gradually to 1-1.5 yu at the distal end. They are densely
covered with lateral spines all along their length, and crowned by a verticil of
terminal spines at the end (Plate 10, figs. 20, 21, 26). The largest spines are the
terminal ones. The verticil formed by them somewhat resembles a convex
end-disc with serrated margin and has in transverse diameter a maximum of
4.5 M- The lateral spines usually decrease in size very considerably towards
the proximal end of the end-ray, but in some of these spicules the proximal end-
ray spines are quite large (Plate 10, figs. 15, 16). All the spines arise obliquely,
their basal parts being inclined towards the centre of the spicule; their ends are
bent down in the same direction. They, therefore, appear strongly recurved
(Plate 10, figs. 15, 16). All the spines on the end-rays of the large discocomes
and the distal spines on the end-rays of all but the very smallest discocomes are
clearly visible. The spines on the proximal parts of the end-rays of the smaller
discocomes and all the spines of the smallest are, however, too minute to be
clearly discernible as such; the presence of these spines is indicated only by
(he rough appearance of the end-rays.

The individual discocomes differ very considerably in size. A cursory
examination shows that the relative dimensions of the main- and end-rays, and

62 CAULOPHACUS SCHULZEI.

the number, position, arrangement, and shape of the latter are not the same
in the large and small discocomes. To obtain an insight into the correlation of
the various peculiarities of the discocomes of different size, I measured thirty-
three of these spicules all from the same large specimen (E). The smallest of
which all the dimensions were taken was 70 yu in diameter, the largest 312.
Discocomes less than 70 m in diameter are rare, the smallest observed was 44 ix,
and these small ones are probably only young forms. As I was unable to take
with sufficient exactitude all the measurements required of these small disco-
comes I have not taken them into consideration in the correlations of the several
characteristics. The difference of the diameters of the smallest (70 n) and the
largest (312 /u) completely measured discocome is 242 ii. This 242 m represents
the range of variation in size (diameter) of the thirty- three discocomes studied.
The fourth part of it was taken, 242:4 = 60.5, and thus the variation-range
itself divided into four equal parts, each extending over 60.5 ix of diameter-
variation. To the first of these four parts belong all the discocomes 70-130.5 m
in diameter, to the second all 130.5-191 m, to the third all 191-251.5 ju, and to
the fourth all 251.5-312 /x. The discocomes belonging to the same quarter of
the diameter-variation range were considered as forming a group and their meas-
urements combined. The subjoined table (p. 63), gives the measurements of
the four groups of differently sized discocomes.

The table indicates that the small discocomes (of group I) are more numer-
ous than the larger, and that among the latter those of group III (191-251.5 ix
in diameter) are scarcer than those of groups II and IV. It further shows that
the relative length of the main-rays is in inverse proportion and the relative
length of the end-rays is in true proportion (total diameter) of the spicule, that
is, the larger the discocome the relatively longer the main-rays and the relatively
shorter the end-rays.

The number of end-rays on each main-ray and the degree of their divergence
(the width of the apical angle of the bunch or verticil formed by them) are in
inverse proportion to the size of the spicule; largest in the smallest (group I)
and smallest in the largest (group IV). Besides these differences of the relative
dimensions of the parts of larger and smaller discocomes, also differences in
the shape and arrangement of the end-rays are to be noticed. In the small
discocomes (Plate 10, figs. 17, 18, 24, 25) the end-rays are more or less curved
in an S-shaped manner, in the large discocomes (Plate 10, figs. 19-21) they are
straight. In the small discocomes, where they are more numerous, the end-rays
form brush-like bunches, whilst in the large discocomes, where they are fewer
in number, they are arranged in a simple verticil. The terminal thickening of

CAULOPHACUS SCHULZEI.
DISCOCOMES.

63

Dimensional groups
(diameter)

I

(70-131.5)

II
(131.5-191)

III

(191-251.5)

IV

(251.5-312)

Number of discocomes of
each group measured

15

7

3

8

Diameter

limits n

70-130

136-186

200-250

270-312

average m

97.1

162.3

225.4

296.5

limits M

20-35

30-50

45-65

44-65

Length of

average ^

27.1

40

54.4

54.9

main-ra}'s

average per-
cent of total
diameter

27.9

24.6

24.1

18.4

num-
ber on
each
main-
ray

limits

9-18
and more

6-11
and more

8

6-9

aver-
age M

about 16

8.5

8

8

End-rays

length

limits

18-35

25-18

55-60

70-103

aver-
age yu

20.7

41.2

58.3

93.8

aver-
age
per-
cent of
total
diam-
eter

22.1

25.4

25.9

31.6

Apical
angle of
terminal

limits

40-100

28-60

30-35

25-35

bunch or
verticil of
end-ray

average

65

38

33

29

the main-rays, from which the bunches or verticils of end-rays arise, is very
(different in discocomes of different size (Plate 10, figs. 17-25) ; long and conical
in the large ones (Plate 10, figs. 19-21), and short and more disc-shaped in the
small ones (Plate 10, figs. 22-25).

In consequence of all these differences the large discocomes differ from the
smaU ones very considerably in appearance; they nevertheless represent a fairly
continuous series of forms.

Some of the small discocomes here considered, like those less than 70 /^ in

64 CAULOPHACELLA TENUIS.

diameter, may be young forms of the large ones. Most of them, however, can
not be considered as such, since their end-ray bunches could not be converted
into end-ray verticils like those of the large discocomes by means of the apposi-
tion of silica-layers and since no silico-clastic process is known to occur in sponges,
which would make possible a conversion (development) of the end-ray bunches
of the small discocomes into the end-ray verticils of the large ones.

Wilson {loc. cit., p. 48, 49) is inclined to consider the large and the small
discocomes as distinct forms, although he has himself observed transitions
between them. The discocomes observed by him had main-rays 16-50 ^ long,
and on each main-ray five to twenty or thirty end-rays 16-100 ^ long.

Wilson {loc. cit., p. 43) examined fourteen specimens from Albatross Stations
3382 and 3399. Both lie off Panama, 3382 6° 21' N., 80° 41' W., 3399 1°
7'N., 81°4' W. The depth at these Stations, 3279 m. (1793 f.), 3182 m. (1740 f.),
is considerably less than at Station 4651 (2222 f.) where my specimens were
obtained.

There can be no doubt that the specimens described above belong to
Caulophacus schulzei Wilson. Their similarity among themselves, and to
those examined by Wilson, is indeed remarkably great. This great similarity
makes it probable, first, that this species is very constant in character, and,
secondly, that the conditions of life are very similar at the three Stations 3382,
3399, and 4651.

CAULOPHACELLA, gen. nov.

Caulophacidae with oxyhexasters, without any other kind of microsclere.
The collection contains one fragmentary specimen of this genus, which
belongs to a new species.

Caulophacella tenuis, sp. nov.
Plate 12, figs. 1-19.

One fragmentary specimen of this sponge was trawled in the Eastern
Tropical Pacific, southwest of the Garrett Ridge, at Station 4732, on 21 Janu-
ary, 1905; 16° 32.5' S. 119° 59' W.; depth 3679 m. (2012 f.); it grew on a bot-
tom of Globigerina ooze; the bottom-temperature was 34.8°.

The specimen is a thin lamella. To this the specific name refers.

The only specimen in the collection is a fragment, measuring 15 by 8 mm.,
of a flat lamella, about 1 mm. thick.

The colour in spirit is nearly dark brown.

CAITLOPHACELLA TENUIS. 65

General structure of the skeleton. Both faces of the lamella are covered
with pinules occupying the usual position. The pinnies on one side are much
larger than those on the other. The face covered with the larger pinules I
consider the dermal, the opposite face the gastral. Pentactines, with para-
tangentially extending lateral rays, and an apical ray directed radially inwards,
occur under both surfaces. The pentactines underneath the face with the larger
pinules are much larger than those underneath the opposite face. The former
are considered as hypodermal, the latter as hypogastral. Numerous slender
and some large rhabds, a few large hexactines, and dense masses of oxyhexasters
and hemioxyhexasters occur in the interior (Plate 12, fig. 13). The oxyhexasters
are much more numerous than the hemioxyhexasters.

Small hexactines with rays strongly curved at the end ; large sword-shaped
hexactines with stout and spiny sword- handle ray; middle sized hexactines with
cylindrical, terminally rounded, strongly curved raj^s; and a few other forms
have also been observed in the spicule-preparations. I consider these spicules
as foreign to the sponge.

The rhabds are 3-17 mm. long, 10-34 n thick, rarely 50 m, and quite sharply
pointed.

The rare large hexactines have straight, conic rays, 0.5-1 mm. long and
20-45 M thick at the base. The rays are smooth for the greater part of their
length, their tips only being covered with small tubercles.

The hypodermal and hypogastral pentactines differ only in regard to their
size. Their lateral rays are straight, conic, and rather sharply pointed. Those
of the former are 500-1100 ^ long and 14-32 m thick at the base; the correspond-
ing measurements of the latter are 110-330 n and 7-14 ix.

The dermal and gastral pinules (Plate 12, figs. 1-8, 14, 15, 19) also differ only
in size. In both the distal ray is straight, stout at the base, and only slightly
thickened above. Sparse, small spines arise from its basal part. Towards
the end the spines become more crowded and larger, the largest attaining a
length of 7 M in the dermal pinules and a length of 5 /i in the gastral. These
spines are sharp-pointed, directed obliquely upward towards the tip of the ray,
and also curved in this direction. Their basal part is inclined at an angle of
about 60° to the ray ; farther on they bend, usually somewhat abruptly, towards
the ray ; so that the angle between their end-part and the ray is 45° or less.
Usually the spines of the same region are fairly uniform. Sometimes, however,
adjacent spines differ considerably in position. Occasionally I have observed
pinules in which the distal spines all tended to one side as if bent by a lateral

66

CAULOPHACELLA TENUIS.

current. As a rule (Plate 12, figs. 1-4, 7, 8) their curvature is simple, the spines
extending in planes which pass through the axis of the ray. In a few cases, how-
ever, the spines (Plate 12, figs. 5, 6) are curved doubly, and spirally twisted round
the ray.

The lateral rays are conic, straight or slightly curved, and rather uniformly
and densely covered with small spines. The proximal ray is rudimentary, very
considerably shortened, cylindrical, terminally rounded, and also covered with
small spines, some of which arise from its apex. The dimensions of the dermal
and gastral pinules are tabulated below.

PINULES.

Dermal
pinules

Gastral
pinules

length n

270-373

115-180

Distal ray

thickness at
base M

6-13

5-7

maximum trans-
verse diameter
(with the spines)
above /n

14-22

8-15

Proximal ray

length M

5-16

5-10

thickness jj

6-13

5-7

Lateral rays

length n

120-232

85-130

thickness at
base M

5-12

5-6

The oxyhexasters and hemioxyhexasters (Plate 12, figs. 9-13, 16-18) measure
100-125 M in total diameter, rarely 137 n. The main-rays are in the same spi-
cule equal, and regularly arranged, each one enclosing right angles with its four
neighbours. They are straight, cylindrical, 8-11 ix long, and 2.7-5 m thick, rarely
as much as 6 /z. Each main-ray bears from one to three end-rays. The number
of end-rays is by no means always the same on the six main-rays of the same
spicule. Most frequently oxyhexasters are observed with three end-rays on
some main-rays and two end-rays on the others, and with two or three end-rays
on all the main-rays. More rarely one or two main-rays bear only one end-ray,
which is either clearly distinguished as such or gradually passes into the main-
ray. These spicules, which appear as hemioxyh'exasters, are, apart from their

CALYCOSILVA CANTHARELLTTS. 67

hemioxyhexastrose character, identical with the oxyhexasters in structure.
The basal part of the end-rays is usually directed obliquely outward and encloses
an angle of 40°-45° with the continuation of the main-ray. A short distance
from then- origin the end-rays are curved rather abruptly toward the continua-
tion of the main-ray, whereupon they straighten out, their middle- and end-
parts being straight, or only slightly curved. The end-rays are 42-50 ^ long,
and 1.5-3 fx thick at the base. They are conic and taper very uniformly to an
extremely fine point. The main-rays are smooth, the end-rays covered with
rather numerous slender, oblique, backwardly directed spines, which attain a
length of 0.3-0.5 m-

The dermal and the gastral pinules and the spiculation generally indicate
that the tliin lamellar fragment above described formed part of a caulophacid
sponge. It differs, however, from all the known forms of the Caulophacidae
which comprise Caulophacus F. E. Schulze, Sympagella O. Schmitt, Aulascus
F. E. Schulze (identical, according to Ijima, with Sympagella), and Calycosilva
Lendenfeld (comprising part of F. E. Schulze's Calycosoma). The absence of
discohexasters, hemidiscohexasters, and discohexactines makes it impossible to
place it in the genus Caulophacus, the absence of plumicomes excludes it from
Sympagella, Aulascus, and Calycosilva. The new genus is named Caulophacella
on account of its similarity to Caulophacus.

CALYCOSILVA, gen. nov.

Stalked, calyculate or mushroom-shaped Caulophacidae. The choanosomal
megascleres are rhabds and hexactines. Hypodermal pentactines are always pres-
ent. Hypogastral pentactines are present or absent. The surface is covered with
hexactine pinules which are similar on the dermal and gastral side of the body.
The proximal ray of some of the pinules may be reduced. The microscleres are
onychhexasters and plumicomes to which oxyhexasters and helonychhexasters
may be added. Without discohexasters and discohexactine microscleres.

The collection contains one complete specimen and thirty-one fragments
referred to this genus, all of which belong to a new species.

Calycosilva cantharellus, sp. nov.
helix, var. nov.

Plate 1, figs. 1-8, 20-24; Plate 2, figs. 3, 7-13, 15; Plate 3, figs. 1-5, 8-30; Plate 4, figs. 23, 24; Plate 5,
figs. 1, 2, 4, 5, 7-9, 11-15, 18-20; Plate 6, figs. 5-21, 24-34; Plate 7, figs. 1-10, 12-14, 16, 17.

68 CALYCOSILVA CANTHARELLUS.

simplex, var. nov.

Plate 1, figs. 9-19, 25-29; Plate 2, figs. 1, 2, 4-6, 14, 16; Plate 3, fig.'^. 6, 7; Plate 4, figs. 21, 22; Plate
5, figs. 3, 6, 10, 16, 17; Plate 6, figs. 1-4, 22, 23; Plate 7, figs. 11, 15, 18.

megonychia, var. nov.
Plate 4, figs. 1-20; Plate 5, fig. 21; Plate 7, fig. 19.

All the specimens of this species were trawled at Station 4651 off northern
Peru, on 11 November, 1904; 5°41.7' S., 82°59.7' W.; depth 4063 m. (2222 f.);
they grew on sticky, fine, gray mud; the bottom-temperature was 35.4°.

The complete specimen shows that the sponge is, in outer appearance, similar
to the mushrooms of the genus Cantharellus, and to this the specific name refers.
It possesses spirally twisted onychhexasters, which I name helonychhexasters.
Such spicules have not been found in any of the other (more or less fragmentary)
specirnens. In some of the latter the average and the maximum size of the
onychhexasters is considerably greater than in the others. On account of this
and other differences between them I distinguish three varieties within this
species : — var. helix, with helonychhexasters (the complete specimen) ; var.
megonychia, without helonychhexasters, with larger onychhexasters (six frag-
mentary specimens) ; and var. simplex, without helonychhexasters, with smaller
onychhexasters (twenty-five more or less fragmentary specimens). Twenty-
four of the specimens of var. simplex are identical and obviously parts of the
body proper of the sponge. These are designated C. c. var. simplex (A). One
corresponds to the basal part and the stalk of the complete specimen. This is
designated C. c. var. simplex (B).

Shape and size. The complete specimen of C. c. var. helix (Plate 6, fig. 18)
appears as a horizontally expanded plate, from near the centre of the lower side
of which a slender stalk arises. The stalk is 52 mm. long, nearly circular in
transverse section, and at the lower end, where it was attached to the sea-bottom,
4 mm. thick. It gradually thickens above and measures at its upper end, where
it gradually passes into the body proper of the sponge, 7 mm. in transverse diame-
ter. Its lower portion is inarkedly bent and has the appearance of having first
grown somewhat obliquely and later vertically. The plate, which is to be con-
sidered as the body proper of the sponge, is irregularly oval in outline and
measures 68 by 92 mm. Its central part, to which the stalk is attached, is 6 mm.
thick. Towards the margin it thins out. The plate is somewhat bent in an
undulating manner and at one place strongly curved inwards. In the figure

CALYCOSILVA CANTHARELLUS. 69

(Plate 6, fig. 18) this involuted portion of the body, which extends quite to the
centre, lies in front.

The lower side of the plate-like body is the dermal, the upper, the gastral.
They are identical in structure and both formed by a transparent membrane
destitute of larger apertures. The entrances to the large afferent and efferent
choanosomal canals are seen through this membrane. In some places where
the superficial membrane has been lost these canal-entrances are bare.

The six specimens of C. c. var. rnegonychia are fragmentary plate-like parts
of the body proper of the sponge. Parts of some of these attain a thickness of
8 mm., and these thin out to a rather fine margin at one side. The largest of
these fragments is 49 mm. long and 35 mm. broad.

The twenty-four fragmentary specimens of C. c. var. simplex (A) are parts
of plates with a maximum length and breadth of 50 mm., and are at their thickest
point 4-6 nmi. thick. One of these fragments formed a central part of a sponge;
to this the upper part of a stalk is attached. The surface has the same character
as in C. c. var. helix, the only difference being that much more of the superficial
membrane has been lost and that some slender spicules protrude from it to dis-
tances of 10 mm. or more. I am incUned to consider these hair-Uke spicules as
foreign.

The single specimen of C. c. var. simplex (B) (Plate 5, fig. 10) has the shape
of a pipe. It is traversed by the fragment of a large foreign spicule, probably
a root-tuft spicule of a hyalonematid. This foreign spicule, which forms the
base of attachment is — for a length of 39 mm. — coated by a thin layer of the
sponge. Thus a cylinder 39 mm. long and 2-3 mm. thick, appearing as the stem
of the pipe, is formed. This stem is to be considered as the stalk of the sponge.
From one end of this stalk, which probably lay horizontally on the sea-bottom,
a structure 9 mm. thick and 15 mm. long, resembling the bowl of the pipe,
arises at an angle of about 106°. This part of the specimen is to be considered
as the upper end of the stalk and part of the body proper of the sponge.

Colour. All the specimens are grayish brown. C. c. var. helix has a more
grayish colour, C. c. var. rnegonychia and simplex are more brownish. The speci-
mens of C. c. var. megonychia are rather darker than the others.

General structure. A fine superficial membrane uniformly covers the
dermal surface of the stalk and the dermal and gastral surfaces of the body
proper. This membrane is supported by the paratangential rays of the pinules
and perforated by pores which lead into a superficial cavity, 60-90 yu high (radial
dimensions), and traversed by numerous fine trabeculae. This cavity is limited

70 CALYCOSILVA CANTHARELLUS.

distally by the superficial membrane containing the paratangential pinule-rays,
proximally by another perforated membrane or, to speak more correctly, a net-
work of paratangential trabeculae, containing the paratangential rays of the
(hypodermal and hypogastral) pentactines. The subdermal (Plate 5, figs. 1, 4f)
and subgastral (Plate 5, figs. 1, 4b) cavities extend below this membrane or
network. These are identical in shape, 300-600 ^i high (radial dimensions), and
traversed by radial columns connecting their roof with their floor. Each of these
columns consists of a proximal ray of a (hypodermal or hypogastral) pentactine,
enveloped in a mantle of soft tissue. In the formation of many of them also the
distal end of a transverse choanosomal rhabd takes part. The columns are on
an average 250 fx apart, and usually 30-40 n thick. Numerous fine, thread-like
trabeculae arise from the columns and join to form close reticulations which sur-
round them like trellis-work. Above and below these reticulations are very
extensive, and join to form continuous networks extending along the roof and
floor of the cavity. In the middle they appear to be less extensive. In the sec-
tions large empty spaces are observed between the trellis of trabeculae surround-
ing the columns in this region. I think it quite likely that in life these cavities
are also traversed by trabeculae, and am inclined to ascribe their emptiness in
the sections (Plate 5, figs. 1, 4) to the trabeculae having here been torn and lost
through shrinkage when the sponge was captured and preserved.

The floor of the subdermal and subgastral cavities is traversed by bundles
of rhabds (Plate 5, figs. 1, 4c, f) and perforated by numerous apertures. These
are more or less circular, both on the dermal and the gastral side, and in the best
preserved specimens (parts) are sometimes 1.5 mm. wide. In specimens (parts)
not so well-preserved and more strongly shrunken some of them attain a diameter
of 3 mm. (Plate 4, fig. 20). On the gastral side their distance from each other
nearly equals their diameter, on the dermal side they are farther apart. The
apertures on the lower, dermal side lead into the afferent, those on the upper,
gastral side into the efTerent canals. The afferent and effei'ent canals are, in
well-preserved parts of the plate-like body proper of the sponge, 0.75-1.5 mm.
wide, and extend in a direction more or less vertical to the surface (Plate 5, figs.
1, 4, 16). Their length is on the whole proportional to their width. The widest
reach to within a short distance of the floor of the (subdermal or subgastral)
cavities on the opposite side. The stalk is hollow, with an eccentric, not axial
cavity, and walls, which in the middle of the stalk of the specimen of C. c. var.
helix are 2.5 mm. thick on one side and 0.7 mm. on the other. Vertical, appar-
ently efferent canals of considerable width (Plate 5, fig. 16) leading up to the

CALYCOSILVA CANTHARELLUS. • 71

central part of the gastral face of the sponge are observed where the stalk is
attached to the body proper.

Some canals are traversed by thread-like or membraneous trabeculae, others
appear to be destitute of such. It is difficult to say whether the latter are empty
in the living sponge, or whether they have lost their trabeculae after capture.
I think, liowever, that the latter assumption is more likely to be correct than the
former.

The trabeculae traversing the cavity of the stalk are distinctly membrane-
ous, and stouter and more distant than the ones spread out in the canals of the
body proper of the sponge.

The flagellate chambers (Plate 2, fig. 7a) form a continuous layer interA^en-
ing between the afferent and efferent canals. They are more or less oval, wide-
movithed sacs and measure 70-100 ii in transverse diameter. Their length varies
considerably, from 120 to 220 /x.

More or less spherical bodies 2-5 m in diameter are met in various parts of
the sponge. These lie either singly or in groups, and stain strongly with aniline-
blue and magenta. The largest of the groups formed by them attain a maxi-
mum diameter of 36 yu, and are composed of forty or more such bodies. Some of
them, particularly the larger single ones, are enclosed in spherical envelopes,
about 9 Ai in diameter, which are often very distinct and appear as cell-walls.
The space between the highly stained body and the envelope is occupied by a
colourless (unstained) and transparent substance. The highly stained body is
not always stained uniformly throughout. One can often distinguish within it
a not so strongly stained ground-substance, and a very strongly stained, irregu-
larly branched, apparently chromidial mass. On a few of the envelopes enclos-
ing these bodies was observed a circle, 2 n in diameter; this appeared as the
margin of a round aperture, perhaps covered by an operculum. In one of
these the stained body did not lie altogether within the envelope but had partly
emerged from it through this aperture and filled it up like a phig.

The spicules taking part in the formation of the skeleton are: — large, stout-
rayed hexactines; derivates of these hexactines with less than six rays, mostly
diactines (in C. c. var. swwpfex only) ; ordinary rhabds; slender, rectangulai'ly
bent diactines (in C. c. \a,Y. helix only); large, slender-rayed triactines (in
C. c. var. megomjchia only) ; pentactines ; hexactine pinules with a fully developed
or a reduced proximal ray; a few microhexactines (in C. c. vars. simplex and helix
only) ; a series of forms of regular onychhexasters ; a few irregular onychhexasters ;
onychhexaster-derivate oxyhexasters (in C. c. vars. helix and megonychia only) ;

72 CALYCOSILVA CANTHARELLUS.

helonychhexasters (in C. c. var. helix only) ; and plumicomes (exceedingly scarce
in C. c. var. megonychia) .

In the body proper of the sponge all the spicules are isolated and free.
In the stalk the ordinary choanosomal rhabds are joined to form a dictyonal net-
work to which also a few hexactines may be attached.

In C. c. var. simplex (B) the skeleton-net of the stalk (Plate 5, figs. 3b, 10)
closely sm-rounds the hyalonematid root-tuft spicule (Plate 5, figs. 3, 10a)
which forms the base of attachment. The thinner distal end of this envelope,
which corresponds to the lower end of the stalk, consists of a network with
beams 12-35 /x thick, and irregularly triangular or polygonal meshes, on an
average about 100 ^ wide. In this network main longitudinal and secondary
transverse beams cannot be distinguished. Farther on, towards the upper end
of the stalk, the network becomes more regular and more and more distinctly
composed of longitudinal main beams (15-65 fi thick, usually 20-45 m), joined
by short, transverse connections to a ladder-like structure. In consequence
of the main beams not being quite parallel, and the transverse beams very irregu-
larly distributed, the meshes of this part of the network are very unequal in size,
5-50 /J. and more broad, 30-200 yu and more long. However different their
size may be, in shape and position these meshes are very much alike, always
elongated, oval, or rectangular with strongly rounded corners, and arranged with
their long axis extending longitudinally. In some parts of this network the
beams are smooth, in others covered by small, low, sharp spines. At the upper
end of the stalk the transverse connections become less numerous and the net-
work dissolves itself into a sheaf of longitudinal rhabds.

Ends of the rhabds taking part in the formation of the net in many places
freely protrude from it. These free rhabd-termini, which are rather scarce
below, but become quite frequent above, are blunt-pointed or rounded, and
just below the end, for a distance of 50-70 n, densely covered with fairly large
spines. In the blunt-pointed ones the end itself is free from spines. In the
terminally rounded ones the spines cover the end also. The spined part below
the end, particularly in the blunt-pointed forms, is considerably thickened, club-
shaped, and measures 15-28 ^ in transverse diameter.

The skeleton-net in the stalk of C. c. var. helix (Plate 5, figs. 5, 7) is similar,
but smooth and still more ladder-like. Its beams are 10-53 m, usually 20-40 /i,
the free ends of the rhabds taking part in its formation, 15-30 fi thick. At the
upper end of the stalk the transverse beams become scarcer and the character-
istically tubular network dissolves itself into a hollow sheaf of isolated longi-

CALYCOSILVA CANTHARELLUS. 73

tudinal rhabds. On reaching the body proper of the sponge this hollow sheaf
of rhabds opens out in a calyculate manner and divides into numerous rhabd-
bundles (Plate 5, figs, le, 4e, 16), which extend in the floor of the subdermal
cavities paratangentially and more or less radially towards the margin of the
plate-like sponge-body. Occasionally anastomosing they here form a kind
of loose paratangential network with radially elongated meshes.

On the gastral side, in the floor of the subgastral cavity, a similar network
of preponderantly radially extending rhabd-bundles (Plate 5, figs. Ic, 4c) is
observed.

Besides these paratangential rhabd-bundles in the floors of the subdermal
and subgastral cavities numerous isolated rhabds and loose bundles of them,
situated obhquely or transversely (Plate 5, figs. Id, 4d), are found in the choano-
some. The ends of many of the transverse rhabds adhere to proximal rays of
hypodermal and hypogastral pentactines. In the centre of the sponge-plate,
near its junction to the stalk, some oblique rhabds, similar to these but very
much larger, have been observed (Plate 5, fig. 16).

At the point of junction of the stalk to the body proper of the sponge some
long and slender diactines with actines enclosing an angle of about 90° (ortho-
monaenes) have been observed in C. c. var. helix.

The hexactines lie scattered rather irregularly in the choanosome. The
thickness of their rays and the size of their spines are subject to considerable
variations. The shortest rayed and most strongly spined are found in the
centre of the body at its junction with the stalk. Towards the margin of the
sponge-plate the rays of the hexactines become more slender and less spiny.

Tetractine and triactine hexactine-derivates have been found in small
numbers, chiefly in C. c. var. megonychia. The diactine hexactine-derivates
are not numerous, and have been observed only in C. c. var. simplex in the
region of the junction of the stalk to the body proper of the sponge.

The gastral and dermal surfaces of the body proper and the surface of the
stalk are uniformly covered by a dense pinule-fur (Plate 2, figs, la, 8a, 13a;
Plate 4, figs. 21-24; Plate 5, figs, la, g, 4a, g, 11a, g, 16a, g). The two kinds of
pinules, with long, well-developed, and pointed proximal ray, and with short,
rudimentary, rounded proximal ray, which are not, or hardly at all, connected
by intermediate forms, are quite indiscriminately scattered, and although the
former are relatively more numerous on the body and the latter predominate on
the stalk, both kinds appear every^vhere to be intermingled. Apart from the
dermal pinules of the body being on the whole slightly larger and having slightly

74 CALYCOSILVA CANTHARELLUS.

larger distal rays than the gastrals, there seems to be no difference between them
(Plate 4, figs. 21-24; Plate 5, figs. 1, 4). The pinules of the stalk are consider-
ably smaller than those of the body proper.

The crosses formed by the four lateral rays of the pinules lie paratangentially
in the superficial membrane (Plate 2, figs. 8a, 13a). Their centres are about as
far apart as their rays are long. In some places they are arranged regularly,
two rays of any two adjacent ones lying parallel and close together (part of the
lower half of Plate 2, fig. 13). In other places their arrangement is not so regular.
The (smaller) pinules of the stalk are much closer together than the (larger)
pinules of the body.

The apical distal ray is much longer in the larger pinules of the body
than in the smaller pinules of the stalk and, as stated above, on the whole in the
dermal body-pinules slightly longer than in the gastrals. But also apart from
this, the distal pinule-ray is very variable in length, and we find everywhere
pinules with long and with short apical distal ray intermixed indiscriminately.
This renders the fur formed by these pinule-rays very shaggy (Plate 4, figs.
21-24).

In the outermost region, which is occupied by the superficial cavities and, as
above stated, is 60-90 m thick, no skeletal elements other than proximal pinule-
rays are met with.

A membrane or network extends parallel to the surface below this region
and separates it from the subdermal and subgastral cavities, forming the roof
of the latter. The centres and the paratangentially extending lateral rays of the
(hypodermal and hypogastral) pentactines are situated in this membrane.
The apical rays of these spicules are situated radially and directed inward. The
crosses formed by the lateral rays of the pentactines are for the most part regu-
larly arranged. The distances between the centres of adjacent pentactines are
in the same region fairly equidistant and a little shorter than the length of their
lateral rays. Two lateral rays of adjacent pentactines are parallel and lie
close together (Plate 2, fig. 13b). The hypodermal and hypogastral pentactines
of the body are quite similar and nearly equal in size; the hypodermal pentac-
tines of the stalk are considerably smaller. The body-pentactines are accordingly
also farther apart than the stalk-pentactines. The distances between the
centres of these spicules being shorter than the length of their lateral rays, the
tips of the lateral rays of each pentactine extend beyond the centres of the four
adjacent ones. This renders the quadratic reticulations formed by the lateral
pentactine rays quite firm. In some places small (probably young) pentactines

CALYCOSILVA CANTHARELLUS. 75

have been observed. The lateral rays of these spicules extend paratangentially
like those of the larger ones, but are, apart from this, more or less irregularly
disposed and Ue anyhow in the meshes of the quadratic network formed by the
lateral rays of the large pentactines.

In most of the pentactines the apical ray is weU-developed, longer than the
lateral rays; in some it is reduced and considerably shorter than the laterals.
The pentactines with short apical ray appear to be scattered indiscriminately
among the pentactines with long apical ray.

Most of the apical (proximal) rays of the pentactines penetrate and extend
beyond the paratangential membranes or networks forming the floors of the sub-
dermal and subgastral cavities. The end-part of many transverse rhabds are
parallel to and in close contact with proximal pentactine rays (Plate 2, fig. 12b).

The microhexactines are very rare and have been found only in the regions
of the subdermal and subgastral cavities.

The regular onychhexasters are abundant in the choanosome and in the
floors of the subdermal and subgastral cavities (Plate 2, fig. 3; Plate 5, fig. 1).
Some also occur in the proximal parts of the columns and threads which traverse
these cavities. They are not confined to the body proper of the sponge and also
occur in the stalk. These onychhexasters form a series, one end of which is
represented by onychhexasters with short and stout end-rays, the other by
onychhexasters with long and slender end-rays. The former are found in the
proximal parts of the subdermal and subgastral regions of C. c. vars. simplex
and helix, but appear to be absent in C. c. var. megonychia. The latter are,
in all varieties, plentiful in the interior. Intermediate forms are met with
wherever onychhexasters occur.

The oxyhexasters occur in small numbers in the choanosome of C. c. var.
helix and somewhat more frequently in C. c. var. megonychia.

The helonychhexasters, which occm* only in C. c. var. helix, are met with
in fairly large numbers in the floors of the subdermal and subgastral cavities-
and are also found in the proximal parts of the columns and threads traversing
these cavities. These spicules are not uniformly distributed throughout this
region, but in some parts of it are much more numerous than in others.

The plumicomes are confined to the columns and threads which traverse
the subdermal and subgastral cavities and are more numerous in their distal
than in their proximal parts. Their paratangential distribution is fairly uni-
form. They are quite abundant in C. c. vars. simplex and helix, but very rare
in C. c. var. megonychia.

76 CALYCOSILVA CANTHARELLUS.

The ordinary rhabds (Plate 1, figs. 1-4; Plate 2, figs, lb, 12b; Plate 5, figs.
1, 2, 4, 8, 9, 11-16) are more or less, sometimes very considerably curved, usually
in a somewhat wavy manner. They attain a length of 2.2-9.1 mm. and a
thickness of 5-80 yu. The ordinary stout rhabds, which are found in small
numbers in the region of the junction of the stalk to the body proper, are in
C. c. var. helix 45-55 ju thick, in C. c. var. simplex 55-80 ^u. The ordinary slender
rhabds, which form the paratangential bundles in the floors of the subdermal
and subgastral cavities and which traverse the choanosorae obliquely and trans-
versely in large numbers, are 2.2-6.2 mm. long and 5-23 ^ thick near the middle.
They are in C. c. vars. megonychia and simplex on the whole somewhat thicker
than in C. c. var. helix. Forms intermediate between the stout rhabds mentioned
above and these slender ones occur, but they are rare. The longest rhabd ob-
served, which measured 9.1 mm. in length, Iselongs to these intermediate forms.

In these rhabds a longitudinal main axial thread is observed, which termi-
nates just below the two ends in the adult spicules, but opens out freely, with a
funnel-shaped widening, in some at least of the young. Besides this there are
two short rudimentary axial threads, forming a cross. The two rudimentary
axial threads are usually 1.5-4 /^ long. Very distinct rounded protuberances
arise over most of the slender rhabds (Plate 1, figs. 3, 4; Plate 5, fig. 9). These
protuberances are generally very low, lower than broad, and in that case the
spicule appears as a centrotyle. Exceptionally they attain a greater length, and
in that case the spicule appears as a tri- to hexactine, with two long rays, and
from one to four perfectly smooth, terminally rounded, cylindrical, rudimentary
rays. The longest rudimentary ray of this kind observed measured 38 n in
length. In some of the slender rhabds the central tyle is so small as to be
hardly or not at all discernible (Plate 5, fig. 8). Hardly or not at all centrotyle
rhabds are much more frequent among the thicker than among the slender
rhabds and most of the thickest are not centrotyle at all. Twenty slender rhabds
of C. c. var. helix which I measured were 7-20 m thick in the middle, close to the
central tyle, which measures 10-23 yu in diameter. In these spicules the central
tyle was 1-7 n thicker than the adjacent parts of the spicule. This difference is
correlated to the thickness of the spicule only in so far as it is on the whole some-
what smaller in the thicker than in the thinner rhabds.

Most of the ordinary rhabds are more or less anisoactine amphistrongyles
or very blunt amphioxes (Plate 5, figs. 2, 12, 13). Their ends are usually about
a third to a half as thick as their central part. Sometimes a slight spindle-shaped
thickening is observed just below the end. Occasionally (Plate 5, figs. 14, 15)

CALYCOSILVA CANTHARELLUS. 77

one end is considerably thickened to a more or less spherical tyle, 24-50 ii in
transverse diameter. Only a few rhabds are smooth throughout. In most a
spined zone, 10-90 fi in extent, is observed at, or just below, the ends. In the
terminally thickened tyle-ends, this spined zone is short and situated terminally,
the spines being sometimes restricted to a small patch on the apex of the terminal
tyle. .\lso in the cylindrical strongyle rhabd-termini the spines often extend
quite to the end. In the tapering rhabd-termini, which are usually slightly thick-
ened in a spindle-shaped manner just below the end, the extreme tip is usually
quite smooth. In such rhabd-termini the smooth terminal zone is sometimes
27 fi long, the spined zone appearing as a belt below the end. The spines are
conic, simple, 1.5-4.5 yu long, and crowded quite closely in the spined zones.

Abnormal rhabds are rare. In one there were two distinct centres, 5 ^
apart, each with a cross of rudimentary transverse axial threads and a tyle.
In another, one end was abruptly bent. Several show short rudimentary branch-
rays, each with an axial thread, at one end.

Some rhabds are corroded and have partly lost one or more of the super-
ficial silica-layers composing them. In one, which had lost a part of its outer-
most layer, a perfectlj^ regular spiral split, forming six close turns, traversed the
part of its still remaining portion bordering on the Une along which the rest had
broken away.

The rectangularly bent diadines (Plate 5, fig. 20) have been found only in
C. c. var. helix, and here also they are very rare. The two actines are straight
or slightly bent, perfectly smooth, 1.. 3-2.3 mm. and more long and 32-35 m
thick at the base. The angle enclosed by them is 89-99°.

The large slender-rayed triactines (Plate 5, fig. 21) have been found only in
C. c. var. megonychia, and here also they are very rare. Their rays are smooth
and at the base about 24 n thick. Two lie in a straight line, from which the third
arises vertically.

The hexactines (Plate 1, figs. 14-24; Plate 2, figs. 4, 6, 9, 11, 14-16) of
the two varieties are similar in shape and size. They measure 0.9-3.4 mm. in
total chameter. Their rays are 0.16-2.2 nun. long, straight or slightly curved,
very rarely angularly bent, and, on the whole, conic. At their base the rays are
18-67 ju thick and, for a distance of 40-160 fi, smooth. Farther on they are cov-
ered with spines for a distance of 140-450 n. In this spined region the rays are
thicker than at their base, and attain a thickness of 22-75 fi. In the rays reduced
in length the spiny region extends quite to the end. In the normal long rays a
distal part of considerable length, which is either smooth throughout or provided

78 CALYCOSILVA CANTHARELLUS.

only with a few small spines close to the end, follows the spiny region. This
distal part of the ray is conic and tapers gradually to a rounded end, 5-17 ii
thick. This has been observed only rarely in the hexactines of C. c. var. megony-
chia. One of the rays is thickened at the end, the terminal tyle attaining a trans-
verse diameter of 35 m-

In some hexactines the rays are nearly equal in length. In most an often
very considerable difference in length of the individual rays is to be noticed. This
irregularity is usually due to one ray (Plate 1, figs. 16, 17, 22) or two rays
(Plate 1, figs. 14, 15) being more or less reduced in length. In the slender-rayed
hexactines, which are probably young forms, the rays thus shortened are similar
to the long ones. In the stout-rayed hexactines, which are certainly full-grown
forms, this difference in ray-length is associated with and obviously correlated
to a difference in the arrangement and shape of the spines, which renders the
appearance of the shortened rays often very different from that of the long ones.

The spines of the spiny regions of the long, not reduced, hexactine rays
(Plate 1, figs. 14-18; Plate 2, figs. 4, 6, 11, 16) are conic, not very sharp-pointed,
and 5-35 ju long. They arise quite or nearly vertically and are not very close
together, on an average about 50 m apart. In some cases they seemed to bs
arranged spirally, but I could not verify this and was indeed unable to prove the
existence of any kind of regularity in their arrangement. The spines of the short,
reduced, hexactine rays (Plate 1, figs. 14, 15, 16; Plate 2, figs. 2, 14) are much
closer together, often in contact with each other at the base, and occasionally
branched. The branched spines (Plate 2, figs. 2, 14) consist of cylindroconic
stems, the ends of which are split up into from two to four stout, conic, obliquely
diverging, secondary spines. These spines somewhat resemble the protruding
rays of the sterrasters of the Geodidae.

The silicoblasts building the rays of the hexactines possess, when they start
work, a certain amount of potential energy, E. This is expended in building the
ray and in forming the spines. The production of the former requires the work
Wi, the production of the spines the work W2. When their task is done the whole
of E will have been converted into work, W, and this W will be equal to Wi + W).
Under normal conditions there is a certain proportion between Wi and W2.
When, however, a spicule or some other obstacle prevents the silicoblasts from
producing a ray of the normal size, less than the usual proportion of W is ex-
pended on Wi so that, W being = Wi + W2, more remains for W2. This leads
to the hyperdevelopment of the spines actually observed on the shortened rays.

The ray being much shorter and the spines more numerous and on the whole

CALYCOSILVA CANTHARELLUS. 79

larger, there is no room on the ray for the development of a spineless distal part,
and there is not even on the whole ray space sufficient for the spines to be placed
at so great a distance from each other as in the spined regions of the normally
developed long ones. It seems very probable that this crowding may lead to a
partial concrescence of two or more adjacent spines and thus to the formation
of the apparently branched structures above referred to, which I am inclined to
consider as more or less coalesced groups of as many spines as they bear terminal
spinelets.

Of hexactine-derivates with less than six rays pentactine, tetractine, and tri-
actine foi-ms have been observed in all varieties, diactine ones, however, only
in C. c. var. simplex. Several pentactine to triactine forms have been observed,
in C. c. var. megonychia. In the two other varieties they are exceedingly rare.
Apart from the smaller number of their rays, they do not differ from the hexac-
tines above described. The diactine forms are much more frequent than those
in C. c. var. simplex.

The diactine hexactine-derivates (Plate 1, figs. 25-29; Plate 2, fig. 5), which
I have found only in C. c. var. simplex, appear as straight, or slightly curved, or
angularly bent, blunt, usually isoactine amphioxes. They are 2.6-3.3 imn.
long and 40-90 n thick in the middle, where a slight thickening is sometimes dis-
cernible. Some taper from here uniformly towards both ends, in others each
actine is thickened some distance from the centre. The transverse diameter
of these thickened parts is in such spicules 10-15 yu greater than that of the centre.
These hexactine-derivate amphioxes bear spines, the size, number, and arrange-
ment of which are subject to considerable variation. A part of the spicule, situ-
ated at or near the middle of its length, is always free from spines (Plate 1, figs.
25-29). Farther on the two actines bear spines, which are either sparsely and
irregularly scattered (Plate 1, figs. 25-27) or restricted to distinct belts, one on
each actine, within which the spines stand rather close together (Plate 1, figs. 28,
29 ; Plate 2, fig. 5) . In shape and size the spines of these amphioxes resemble the
spines of the hexactines above described. A few of the spines are branched
(bifurcate) .

The pentactines have very much the same shape and size in the three varie-
ties, and there seems to be hardly any difference between the hypoderraal and
the hypogastral pentactines of the body. The pentactines of the stalk are
smaller and have a relatively shorter apical (proximal) ray.

The four lateral rays of the hypodermal (Plate 6, figs. 1-8) and hypogastral
(Plate 1, figs. 5-13) pentactines of the body proper are 250-770 fx long, usually

so CALYCOSILVA CANTHARELLUS.

300-700 II, 16-47 m thick, usually 19-32 ^ at the base, and on the whole conic.
They taper gradually to the rounded end, which is 1-11 ^ thick, usually 2-4 /i.
In two belt-like regions, one a short distance from the base and the other a short
distance from the end, each lateral ray usually bears small, low spines, the spines
of the proximal belt being larger and more numerous than those of the distal
belt. The base, the middle, and the end of the lateral ray are usually smooth.

The angles between the lateral rays are always nearly 90°. In this respect
the crosses formed by them are regular. The length of the lateral rays of the
same spicule is, however, by no means the same. In this respect the crosses are
irregular. Among all the many pentactines I measured I found not one with
equally long lateral rays. The difference in the length of the longest and short-
est lateral ray of the same spicule amounts to 20-320 n.

The apical (proximal) ray is similar to the lateral rays in shape, but very
variable in length, 0.15-1.37 mm. long, and at its base usually somewhat thicker
than the lateral rays. In most of the pentactines the proximal ray is well-
developed and longer, in some reduced, as long as or shorter than the lateral
rays. Some of these reduced proximal rays are truncate at the end as much as
15 ix thick. A correlation between this occasionally occurring reduction of the
proximal ray and the development of the lateral rays does not seem to exist.
The influences (obstacles) which prevent the silicoblasts building the former
from properly executing their task of producing a proximal ray of normal length,
do not appear to affect in any way those building the latter.

A crowding of the spines is observed in the reduced apicals of the pentactines
similar to that in the reduced hexactine rays, described above; it is not, however,
so marked. The cause of this crowding is doubtless in both the same.

The lateral rays of the hypodermal pentactines of the stalk (Plate 6, figs. 9-12,
13a) are 230-520 ^ long and at the base 17-42 yu thick, usually 18-32 //• They
are conic, have blunt ends, and are either quite smooth or provided only with
very small spines. The crosses formed by them are, like those of the pentac-
tines of the body, regular in respect to the angles between the rays, which are
always about 90°, but irregular in respect to their length, which always differs
more or less. The difference between the length of the longest and shortest
lateral ray of the same spicule is in these pentactines 10-80 ti. The apical (proxi-
mal) ray is similar to the laterals in shape and usually about as long or only a
little longer.

As mentioned above there are two kinds of pinules; pinules with properly
developed proximal and apical rays, and pinules with such rays rudimentary and
not at all, or but slightly, connected by intermediate forms. Both kinds occur

CALYCOSILVA CANTHARELLUS. 81

both on the body and on the stalk. The dermal and gastral pinules of the body
are very similar, and both differ from the pinules of the stalk. The pinules wath
rudimentary proximal ray do not differ from those in which this ray is properly
developed in other respects. I shall, therefore, describe the pinules in two
groups : — the dermal and gastral body-pinules with long and short proximal
ray, and the dermal stalk-pinules with long and short proximal ray.

The four lateral rays of the (dermal and gastral) pinules of the body (Plate 2,
fig. 13; Plate 4, figs. 21-24; Plate 6, figs. 14-17, 19-25; Plate 7, figs. 6-19)
have in all three varieties the same shape and size. They are 50-144 ii long,
enclose right angles, and form crosses 120-280 ^ in diameter (Plate 2, fig. 13a;
Plate 6, fig. 23; Plate 7, figs. 16-18). The lateral rays of the same pinule are not
equally long, but their differences in length are usually not great, the longest
lateral ray being only 2-20 'fi, rarely as much as 45 ju, longer than the shortest.
The lateral rays are straight, 4-10 ^ thick at the base, nearly always conic, and
pointed at the end. Pinules with one or more lateral rays reduced in length and
terminally rounded (Plate 7, fig. 8), or very thick at the base and abruptly attenu-
ated to a thin conic end-part, are exceedingly rare. The lateral rays bear vertically
arising spines, which are very small and close together near the end but more
distant and larger (1.5-3 yu long) in the middle and proximal parts of the ray.
These spines are remarkably slender, even the longest is not much over 1 n
thick at the base.

The apical proximal ray is, when properly developed (Plate 6, figs. 19-22,
25; Plate 7, figs. 6, 7, 9, 10, 19), similar to the lateral rays in shape, spinulation,
and size. When reduced (Plate 6, fig. 24; Plate 7, figs. 12-15) it is as thick,
but in C. c. vars. simplex and helix is only 5-10 n, in C. c. var. megonychia 5-20 yu
long, and terminally rounded. On such proximal rays the spines are generally
as far apart as on the long ones, and not crowded. One or two spines often arise
from the rounded end of the ray.

The apical distal ray of the body-pinules is longest in C. c. var. helix,
shorter in C. c. var. simplex, and still shorter in C. c. var. megonychia. It is in the
dermal body-pinules of all varieties longer than in their gastral ones, being

in the dermal body-pinules of C. c. var. helix 130-385 fi

" " " " " " " " " simplex 100-290 M

" " " " " " " " " megonychia 100-180 n

" " gastral " " " " " " helix 120-270 m

" " " " " " " " " simplex 100-240/.

" " " " " " " " " megonychia lOO-lQB tJtlong.

82 CALYCOSILVA CANTHARELLUS.

This ray in all varieties is at its base 5-13 n thick, usually a little thicker
than the other rays. It thickens above and attains its maximum thickness about
a quarter of its length from the centrum of the spicule. From its thickest point
it tapers gradually towards the stout, blunt-pointed, distal end (Plate 6, figs. 14-
17). It bears numerous spines. These have the ordinary conic shape and are
not at all broadened and flattened like scales. The spines arising from the basal
part of the ray are vertical, quite distant, short, and straight (Plate 6, figs. 21, 22,
24; Plate 7, figs. 6, 8-10). Farther up they become more numerous, longer, and
more inclined toward the ray, their ends pointing obliquely upward. This oblique
direction is attained partly by the spines of this region arising obliquely, partly
by their being more or less abruptly bent in their basal portion. About two
thirds or three quarters of the way up the ray these spines attain their maximum
size. They are in this region about 15 n long, 1.5-2 ,u thick at the base, and arise
at an angle of about 70° from the ray. They are bent abruptly upwards 1-2 n
froni their base, the axis (chord) of their conic, slightly and somewhat irregularly
curved end-part enclosing an angle of about 23-30° with the axis of the ray.
Towards the end of the ray the spines gradually become smaller, those arising
nearest its freely protruding tip being only 5 /^ long, or still shorter. The distal
pinule-ray, together with its spines, resembles the tail of a mammal or wheat-ear
more than the cone of a fir-tree. Its maximum breadth is 15-32 /j. In respect
to this dimension there is no perceptible difference between the dermal and
gastral body-pinules and the body-pinules of the three varieties.

The (dermal) pinnies of the stalk (Plate 6, figs. 26-34) have the same struc-
ture as the body-pinules but differ from them in their dimensions and the preva-
lence of forms with reduced proximal ray. Their lateral rays are only 54-100 ^
long and the crosses formed by them 115-196 ^ in diameter. Their proximal ray
is, when properly developed, 50-90 fx long, when reduced, 4-10 ju. These rays
are 3-7 m thick at the base. The distal ray is 55-115 fi long, and 6-10 /u thick
at the base. The maximum breadth of this ray, with its spines, is 13-30 ix.
On the whole these pinules decrease in size from the upper to the lower end of the
stalk.

I have above drawn attention to the fact that there are no, or hardly any,
intermediate forms connecting the pinules with reduced proximal ray with those
in which this ray is properly developed. In fact I have not observed a single
body-pinule with a proximal ray 21-59 /x long, the nearest approach to an
intermediate form being the pinule (Plate 7, fig. 8) in which the proximal ray is,
although 60 m long, nearly cylindrical and terminally rounded.

CALYCOSIU^A CANTHARELLUS. 83

This absence of intermediate forms, and the fact that the spines are no
closer together on the reduced than on the long proximal rays, show that the
reduction of the short ones cannot be due merely to obstacles which impeded
their longitudinal growth. It is, therefore, probable that we have here to deal
with two distinct kinds of pinnies, one with long, and one with short proximal
ray, even though the pinules with short proximals are locally aggregated only
in so far as they are relatively much more numerous on the stalk than on the
body.

The scarce microhexactines (Plate 3, fig. 1), which have been found only in
C. c. vars. simplex and helix, measure 16-19 /x in total diameter, and consist of
six equal, cyUndrical, terminally rounded rays, (without centrum) G-8 m long,
and 3-8 m thick, which enclose right angles. The rays bear numerous larger or
smaller spines on their sides and on their ends.

At first I took these spicules for central remnants of hexasters which had
lost their end-rays, like those described by F. E. Schulze^ and Ijima'- in other
Hexactinellida. But since I found no indication of these spicules having once
possessed end-rays, I tliink this view hardly tenable.

Of hexasters two main groups can be distinguished : — one represented by
the regular onychhexasters, the onychhexaster-derivate oxyhexasters, their
regular onyclihexasters, and the helonychhexasters; the other by the plumi-
comes.

The regular onychhexasters (Plate 2, fig. 3a; Plate 3, figs. 21-30; Plate 4,
figs. 1-19) form a series commencing with small ones with stout end-rays and
recurved terminal spines, and ending with large ones with slender end-rays
and terminal spines directed obliquely outward. In C. c. vars. helix and simplex
the whole of this series of onychhexasters is met with; in C. c. var. megonychia
I have observed the large forms with slender end-rays only. The onychhexasters
of C. c. var. helix are 39-88 ^ in total diameter, those of C. c. var. simplex 48-
106 n, and those of C. c. var. megonychia 80-130 m- They consist of a centrum,
5-6 IX in diameter, from which arise the six concentric and equal main-rays, the
axes of which enclose angles of 90°, with trumpet-shaped, proximal extensions.
The main-rays are smooth, 2-A n thick, (without the centrum) 1.5-6 ^ long, and
thickened in a trumpet-shaped manner and divided into from two to five end-
rays at the distal end. The end-ra^^s of the same spicule are fairly equal in shape

' F. E. Schulze. Amerikanische Hexactinelliden, 1899, p. 31, taf. 5, fig. 8.

- 1. Ijima. Studies on the Hexactinellida. I. Jouni. Coll. sci. Tokyo, 1901, 15, p. 198, 292, pi. 4,
fig. 20.

84 CALYCOSILVA CANTHARELLUS.

and size; the number of them on each of the six main-rays is, however, by no
means always the same. The end-rays are 1.5-2 fi thick at the base, and arise
steeply, sometimes at nearly right angles, from the main-rays. Farther on
they curve inward, towards the continuation of the axis of the main-ray to which
they belong. Distally this curvature rapidly decreases and the end-part is
for a smaller or greater, usually a very considerable length, either quite straight
or only slightly curved, or irregularly bent like an oak-branch and knotty in
appearance. Onychhexasters with end-rays thus bent have been chiefly found
in C. c. var. rnegonychia (Plate 4, figs. 2-4, 14, 17). In all regular onychhexasters,
whether large or small, the centrum, the main-rays, and the proximal parts of the
end-rays are nearly identical in shape and have the dimensions given above ; the
great differences in these hexasters observed are entirely due to differences
in the degree of longitudinal development of the distal straight end-parts of
the end-rays. In the smallest onychhexasters observed (Plate 3, fig. 21) this
distal straight part is quite insignificant and hardly distinguishable. The
larger the onychhexaster is, the longer and the more conspicuous does this part
of the end-ray become (Plate 3, figs. 22-27; Plate 4, figs. 2-7). The end-rays
are cylindroconic, attenuated distally. This attenuation is slight and very
much the same in all end-rays, however long they may be. The consequence
of this is that the thickness of their distal ends is in inverse proportion to the
length of the end-rays; greatest in the shortest, and smallest in the longest.
In the small onychhexasters, 39-45 yu in diameter, of C. c. var. helix, the end-rays
are 15-18 /u long and 1-1.8 m thick at the end; in the largest onychhexasters,
80-88 IX in diameter, of the same variety the end-rays are 34-41 ^ long and only
0.8-1 M thick at the end. In the larger onychhexasters of C. c. vars. simplex
and megonychia the same inverse relation between the length and terminal thick-
ness of the end-rays is observed. The angles between the chords of end-rays
arising opposite each other from the same main-ray are correlated and in inverse
proportion to the size of the spicule and the length of the end-rays. In the small
onychhexasters, 39-45 m in diameter, of C. c. var. helix, these angles are 70°-90° ;
in the large ones, 80-88 m in diameter, of the same variety 59°-77°.

The end-rays bear numerous small recurved spines along their length
(Plate 3, fig. 28; Plate 4, figs. 9, 10, 16) and one to five large spines at the end.
The former are largest and most conspicuous in the smallest onychhexasters
(Plate 3, fig. 22) ; in the large onychhexasters they are smaller. Their size is, on
the whole, in inverse proportion to the length of the end-rays and the size of
the whole spicule. In the smallest onychhexasters the terminal spines are 2-3 ^

CALYCOSILVA CANTHARELLUS. 85

long and generally strongly recurved, so that these end-rays become anchor-
like. With the increase in size of the onychhexaster (length of the end-rays) the
terminal spines become longer. In the largest onychhexasters they are 2-8 n
long. At the same time they change their shape and their position relative to
the end-ray from which they arise, generally being the more directed outward
the longer the end-ray is. In the medium-sized onychhexasters (Plate 3, figs.
24, 25, 28-30) they are usually more or less vertical to the end-ray, their end
being shghtly bent inward (Plate 3, figs. 28, 30) or outward (Plate 3, fig. 29).
In the large onychhexasters (Plate 3, figs. 26, 27; Plate 4, figs. 2-4, 13-19) they
are generally directed obliquely outward. This clearly pronounced correlation
between the length of the end-rays and the position of the terminal spines is very
remarkable.

The oxyhexasters (Plate 3, figs. 4, 5; Plate 4, fig. 1) are not numerous and
have been found only in C. c. vars. helix and megomjchia. They measure in the
former 90-94 ix in total diameter, in the latter 100-133 n. From a centrum
5-7 n in diameter arise four smooth main-rays, 2-6 /x long, 2-2.5 ^ thick in the
middle, and thickened at each end. The main-rays of the same spicule are
equal and their axes enclose angles of 90°. Each main-ray bears two to four
end-rays, 40-60 n long and 1.5-2 fi tliick at the base. The end-rays arise steeply
from the main-rays. Their proximal end is curved inwai'ds, towards the con-
tinuation of the axis of the main-ray to which they belong. Their distal and
middle-parts are nearly straight. The chords of opposite end-rays of the same
main-ray enclose angles of about 70°. The end-rays bear along their length a
few very small spines, are conic, and taper gradually to a fine point.

This description shows that these oxyhexasters are very similar to the largest
onychhexasters and distinguished from them only by the tips of their end-rays
being destitute of terminal spines. In some hexasters (Plate 4, fig. 1), similar
in every other respect to the oxyhexasters above described, a slight angular
bend is to be noticed 4-8 m below the tip in one or more of the end-rays. In
others again (Plate 4, figs. 16-18) this angular bend is more pronounced, the bent
end-part diverging strongly from the continuation of the middle-part of the
end-ray. In others again only some of the end-rays are simply pointed, the
others bearing terminal spines, similar to those of the large onychhexasters.

From these observations I conclude that the oxyhexasters above described
are to be considered as onychhexaster-derivates. I think theu" appropriate
place is in a continuation of the onychhexaster-series beyond the end represented
by the large ones with long end-rays and outward-directed terminal spines.

86 CALYCOSILVA CANTHARELLUS.

It seems very probable that they have been produced by a further development
of the onychhexasters in the direction of small forms with recurved terminal
spines, or large forms with upward directed terminal spines. I think that the
forms described above, in which the end-rays appear to be angularly bent near
the end, have been developed out of large onychhexasters by a reduction of the
number of the terminal spines to one, and by a further increase of the angle
at which this single remaining spine arises from the end-ray. The bent terminal
part which appears as the distal end of the ray is, according to this, not a part
of the end-ray at all, but a terminal spine. When, by a further development
in this direction and a further increase of the angle between the terminal spine
and the end-ray, this angle becomes 180°, an apparently true oxyhexaster is the
result.

That the oxyhexasters are to be considered as such ultra-end forms of the
onychhexaster series is corroborated by the fact that they are larger than the
largest regular onychhexasters found in the same variety.

In the not spirally twisted irregular onychhexasters (Plate 3, figs. 2, 3, 6, 7),
which are very rare, the end-rays only or both the end- and the main-rays may
be irregular. The onychhexaster (Plate 3, figs. 6, 7) is an example of the former
case. In this spicule, which was found in C. c. var. simplex, the main-rays are
regularly disposed, equal, abnormally stout, 7 ^ long, and 5 m thick. Each
main-ray bears only one or two somewhat irregularly curved end-rays, which
are also abnormally stout, being 2-3 fx thick at the base. The terminal spines
are 3 m long and recurved. The whole spicule measures 74 n in maximum diame-
ter. The onychhexaster (Plate 3, figs. 2, 3) is an example of the latter case.
In this spicule, which was found in C. c. var. helix, two opposite main-rays, lying
in a line, are considerably longer than the other four, and the end-rays are not,
as is invariably the case in the regular onychhexasters, arranged in a verticillate
manner at the end of each main-ray, but arise from them at various points.
The main-rays are 4 ix thick, the end-rays basally 2 ti. The terminal spines are
irregularly disposed, and 3-5 m long. The whole spicule is 89 // long and 64 ii
broad. I consider these rare, not spirally twisted, irregular spicules as mere
pathological abnormities.

The helonychhexasters are onychhexasters in which most end-rays or all of
them are spirally twisted. To this spiral twisting the name I have given these
spicules refers. The helonychhexasters are quite abundant in C. c. var. helix,
but absent in the other two varieties.

The helonychhexasters of C. c. var. helix (Plate 2, fig. 3b; Plate 3, figs. 8-20)

CALYCOSILVA C'ANTHARELLUS. 87

have a ceijtrum 4-6 m in diameter, from which six main-rays arise. These are
very similar to the main-rays of the regular onychhexasters, enclose right angles
with their neighbom-s, are smooth, 2-4 // thick, and (without the centrum)
2-5 Ai long. Forms with long and slender end-rays and outward directed termi-
nal spines, forms with short end-rays and recurved terminal spines, and inter-
mediate forms, corresponding to the different forms of the regular onychhexasters
described above, are met with also among the helonychhexasters.

The twisted end-rays of the same spicule are always curved in the same
direction and describe evolvent (spiral) curves extending in planes parallel to
each other and vertical to one of the three axes of the spicule. The two (oppo-
site) main-rays of the spicule representing this axis, lie in the axis, the four others
in a plane parallel to the spirals accordingly. Each main-ray bears from one to
four end-rays. When only one end-ray is present the main-ray usually passes
into it gradually. The end-rays 10-35 yu long arising from the main-rays, which
lie in the axis of the spiral, often do not participate in the general twisting and
are usually either irregularly curved throughout, or curved only basally and
nearly straight distally, like the end-rays of the regular onychhexasters. The
end-rays arising from the four main-rays parallel to the plane of the spiral twist
are generally all affected by the torsion. At the base, where they are most
strongly curved, their radius of curvature is about 7 m. Farther on their curva-
ture, being a spiral or evolvent one, decreases. Still farther, at a smaller or
greater distance from the distal end, the curvature is usually reversed, the
terminal part of the end-ray being fairly straight and arising radially or obliquely
from the convoluted central mass of the spicule. Depending, as it does, on the
variable position of the point of recurvature, the length of this end-part is veiy
variable.

The transverse diameters parallel to the plane of spiral twist, which pass
through the centrum of the spicule, represent the breadth of the spicule, while
the diameter along the axis of torsion can be considered as its length. Taking
breadth and length in this sense, we find that the spicule is 33-58 n long, whilst
its central convoluted mass measures 15-22 m, and the whole spicule 16-67 n
in breadth. Numerous small, recurved spines, uniformly scattered along the
length of the ray, and two to five larger terminal spines 1.7-4 fi long originate
from the end-rays arising from the main-rays lying in the torsion-axis and also
from those of the others in which the straight end-part attains a greater length.
Accordingly the spinulation of these end-rays is very similar to that of the end-
rays of the regular onychhexasters. The spinulation of the end-rays spirally

88 CALYCOSILVA CANTHARELLUS.

twisted for a greater part of their length is much more irregular. Here, the spines
scattered along the length of the end-ray are not uniformly distributed and often
restricted to its outer, convex side; they are also unequal in size, some of them
attaining a very considerable length. Such end-rays usually have only one ter-
minal spine, sometimes 6.5 ^ in length, which is directed obliquely outward.
The angle between this terminal spine and the adjoining part of the end-ray
usually is rather obtuse and not infrequently becomes 180°. In this case the
terminal spine appears as the tip of the end-ray, and the end-ray itself becomes
simple, as in the oxyhexasters above described.

Intermediate forms with somewhat curved but not properly spirally twisted
end-rays, connecting the helonychhexasters with the regular onychhexasters,
have occasionally been found, but they are exceedingly rare.

Spicules twisted spirally like the helonychhexasters of C. c. var. helix have
repeatedly been noticed in Hexactinellida.

Oxyhexasters and oxyhexaster-derivates with a reduced number of spirally
twisted rays have been found by F. E. Schulze in Holascus stellatus,^ Holascus
ridkyi,'^ and Rhabdocalyptus mollis^ and by Ijima in the last named species,*
in Hyalascus giganteus^ and in Staurocalyptus pleorhaphides.^

Discohexasters with the verticils of end-rays twisted spirally round the
continuations of the axis of the main-rays from which they arise, have been
found by F. E. Schulze in Hertwigia falcifera,'' Rhabdopedella tintinnus,^ and
Saccocalyx pedunculata ^.

Clavules with branch-rays twisted spirally round the shaft have been
found by F. E. Schulze in Farrea convolvulus '° and by jWilson in Farrea occa
claviformis ^^.

The spirally twisted oxyhexasters of Holascus stellatus, Holascus ridleyi,
Rhabdocalyptus mollis, Hyalascus giganteus, and Staurocalyptus pleorhaphides
are similar to ordinary, not twisted oxyhexasters occurring in the same sponge
and more or less connected with them by intermediate forms. The same applies

' F. E. Schulze. Kept. Voy. ChaUenger, 1887, 21, p. 86, pi. 14, figs. 10-12.
2 F. E. Schulze. hoc. cit, p. 90, pi. 17, fig. 7.
' F. E. Schulze. hoc. cit, p. 157, pi. 64, fig.s. 10, 11.

* I. Ijima. Studies on the Hexactinellida, IV. Journ. Coll, sci. Tokyo, 1904, 18, p. 266, pi. 20, fig. 9.
*j/, Ijima. hoc. cit., p, 106, pi. 8, fig. 16
« /, Ijim^. hoc. cit., p, 229, pi. 16, fig. 8.

' F. E. Schulze. Amerikanische Hexactinelliden, 1899, p. 23, taf. 3, fig. 8.
» F. E. Schulze. Rapt, Voy. ChaUenger, 1887, 21, p. 108, pi. 12, fig, 8,

'' F. E. Schulze. Hexactinelliden des Indischen Oceanes. II. Abh. Akad. Berlin, 1895, 1896, p. 55,
taf. 5, figs. 4, 9, 10.

i» F. E. Schulze. AmerikanLsche Hexactinelliden, 1899, p. 72, taf, 16, figs. 1, 2,
" H. V. Wilson. Mem. M. C. Z,, 1904, 30, p 59, pi. 7, fig, 3

CALYCOSILVA CANTHARELLUS. 89

to the clavules with twisted branch-rays of Farrea occa claviformis and the
helonychhexasters above described. The discohexasters with spirally twisted
end-ray verticils of Herlwigia falcifera, Rhabdopedella tintinnus, and Saccocalyx
pedunculata and the clavules with similarly twisted branch-rays of Farrea con-
volvulus on the other hand do not appear to be associated with regular, not
twisted spicules of the same kind.

In studying the question how these spirally twisted spicules have been
produced I gained the impression that the parts of the living mass ^ which built
them must have changed their relative positions in a torsional manner during
the growth of those rays or portions of rays which are spirally twisted in the
full-grown spicule. In the case of the oxyhexasters, helonychhexasters, and
clavules with spirally twisted branch-rays there is only one torsion-axis corre-
sponding to one of the axes of the spicule, and in these the torsion seems to
have affected the whole living mass uniformly. In the case of the discohexasters
there are six torsion-axes of this kind, corresponding to the axes of the six main-
rays,, and six different torsional systems in the living mass.

In speaking of the spirally twisted oxyhexasters of Rhabdocalyptus mollis
which were found in some, but not in all specimens, Ijima ^ says : "I am therefore
disposed to consider them as of inconstant occurrence in the species. Possibly
they are produced only under certain abnormal conditions." Also in Calyco-
silva cantharellus they have been found in one specimen only. Since, however,
this was obtained together with the others destitute of these spicules in the
same locahty, at a considerable depth, where doubtless the environment was
very monotonous, it is hardly to be supposed that the external influences acting
on it could have been in any way different from the influences acting on the
others. A spiral twisting of some of the spicules is, as the above statement
shows, if not a frequent, still a widespread occurrence in hexactinelUds. It
seems therefore improbable that the spiral twist is produced through the influence
of abnormal conditions, and to be considered as an abnormity. Neither can it
be ascribed to obstacles preventing the (twisted) rays from growing in the
usual direction, because, in the first place, there are no such obstacles, and
because, in the second place, their presence could not affect all the actually
twisted rays of a spicule in the same way and induce them to curve round spirally
in the same direction.

' I use the expression " living mass," because I do not know whether these spicules are built by
distinct cells, and if so, by how many, or by syncitia, and if so, how many nuclei or chromidia or other
centres of vital action, these syncitia contain.

2 /. Ijima. Studies on the Hexactinellida. IV. Journ. CoU. sci. Tokyo, 1904, 18, p. 266, 267.

90 CALYCOSILVA CANTHARELLITS.

In view of these circumstances I consider these spirally twisted skeletal
elements as spicules sui generis, and the difference between them and the similar,
not spirally twisted spicules, as similar in nature to the difference between the
simply curved and spirally twisted horns of closely allied forms of Bovidae.

The plumicotnes (Plate 2, fig. 12c; Plate 7, figs. 1-5) are quite abundant in
C. c. vars. simplex and helix, but exceedingly rare in C. c. var. megonychia.
Those of C c. var. helix are 47-55, those of C. c. var. simplex 50-69 m in diameter.
Those of C. c. var. megonychia seem smaller than those of the two other varieties.
Measurements I cannot give, because I saw only few and these were broken.
Four equal main-rays, which enclose angles of 90° with their neighbours, arise
from a slight central thickening about 2.5 m in diameter. These main-rays are
straight, cylindrical, smooth, 9-12 n long, 0.9-1.6 n thick, and simply rounded
at the end. Some distance below the end each main-ray is thickened to a spheri-
cal or oval tyle, 2-3 ^ in diameter. The distal part of the main-ray, lying beyond
this tyle, is 2-4 n long. A considerable number, twenty or more, branch-rays
arise from the tyle of each main-ray. Their points of origin are fairly equidistant,
but irregularly scattered over the surface of the tyle. The end-rays are very
thin and strongly curved in an S-shaped manner. They terminate with fine
points. A few small spines are occasionally observed on the concave side of their
distal part. The end-rays are equal and regularly arranged so as to diverge
above in a plumose manner.

I tliink there can be no doubt about the close relationship of the sponges
above described. All the specimens were found at the same station; the frag-
ments appear to be parts of sponges similar in shape to the complete specimen;
no difference could be detected in their soft parts; and the shape and arrange-
ment of most of their spicules are the same in all. Still there are differences in
the spiculation of the thirty-two specimens, according to which they fall into
three groups. The chief differences between these groups, which I describe as
distinct varieties, are tabulated on p. 91.

The nearest alhes of the sponges described above as Calycosilva cantharellus
are the species assigned by previous authors to the genera Sympagella, Calyco-
soma, and Aulascus: — Sympagella mix O. Schmidt 1870 (F. E. Schulze 1887,
1897, 1899, 1900; Topsent 1904); Aulascus johnstoni F. E. Schulze 1887 (F. E.
Schulze 1897); Calycosoma validum F. E. Schulze 1899; Sympagella anomala
I. Ijima 1903; and Calycosoma gracile F. E. Schulze 1903. All these, with the
single exception of Aulascus johnstoni F. E. Schulze differ from Calycosilva
cantharellus by being destitute of hypogastral pentactines. Aulascus johnstoni

CAIATOSILVA CANTHARELLUS.

91

C. c. var.
helix

C. c. var.
simplex

C. c. var.
megonychia

Rectangularly bent diactine

megascleres

present

absent

absent

Large, slender-rayed triactines
with two rays in a straight line
and the third arising vertically
from this

absent

absent

present

Amphiox, diactine hexactine-
derivates

absent

present

absent

Body-pinules

dermal

distal ray
120-270 M long

distal ray
100-240 M long

distal ray

100-180 M long

gastral

distal ray
120-270 M long

distal ray
100-240 M long

distal ray
100-195 M long

Onychhexasters

39-88 M
in diameter

48-106 M
in diameter

80-130 M
in diameter

Oxyhexasters

present

absent

present

Helonychhexasters

present

absent

absent

F. E. Schulze differs from it by its shape and by some of its spicules. In Aulascus
johnstoni all the pinules have a properly developed proximal ray, and discohex-
asters and discohexaster-derivate discohexactines occur. In Calycosilva can-
tharelliis many pinules have a reduced proximal ray, and discohexasters and
discohexactines are absent. Also the plumicomes are somewhat different.
Among the above mentioned related forms without pentactine hypogastralia,
Calycosoma gracile F. E. Schulze, which is not very different in shape and has
very similar onychhexasters and plumicomes, appears to be most closely allied
to Calycosilva cantharellus. The differences between the latter and the most
similar of the aUied forms {Aulascus johnstoni and Calycosoma gracile), let alone
the others, are so considerable as to necessitate the establishment of a new
species for its reception.

Whilst I experienced no difficulty in coming to this decision about the estab-
lishment of a new species, I found it exceedingly difficult to decide whether this
new species should be assigned to one of the three genera mentioned, and if so,
to which one. Ijima ^ attaches little systematic importance to the presence
or absence of hypogastral pentactines and accordingly proposes to unite the

1 1. Ijima. Studies on the Hexactinellida. III. Journ. Coll. sci. Tokyo, 1903, 18, p. 96.

92 CAULOPHACID.

species referred by F. E. Schulze to the two genera Sympagella, without,
and Aulascus, with hypogastral pentactines in one genus, which should, since
this has priority, be named Sympagella. According to the diagnosis of this
genus given by Ijima the species belonging to it possess discohexasters. Since
discohexasters are entirely absent in Calycosilva cantharellus, we cannot, if we
accept Ijima's classification place this sponge in this genus or in Aulascus, united
with it by him. If we follow Ijima's example of not considering the presence or
absence of hypogastral pentactines of systematic importance sufficient for generic
distinction, we must place the sponges above described in the same genus as Caly-
cosoma gracile F. E. Schulze. According to Ijima ^ the Caulophacidae are distin-
guished from the Rossellidae i. a. by the former possessing and the latter being
destitute of true pinules with a well-distinguished distal ray. If this be accepted
the two species {validum and gracile) placed by F. E. Schulze in Calycosoma must
be generically separated, because only one (C. validum) has no true pinules and
can be retained in the Rossellidae, where F. E. Schulze '' although doubtful about
it himself, places Calycosoma; whilst the other {C. gracile) jjossesses pinules
with well-distinguished distal ray and must be assigned to the Caulophacidae.
The first described of these two species is the rossellid Calycosoma validum.
This must therefore be considered as the type species of the genus and for this
the generic name Calycosoma must be retained. The other species, Calycosoma
gracile, with which the sponges described above might be generically united, has,
according to this, to be excluded from Calycosoma, and a new generic name
has to be found for it.

Under these circumstances I establish a new genus for the sponges described
above, in which also Calycosoma gracile F. E. Schulze 1903 might be placed.
The name Calycosilva denotes its origin from Calycosoma on the one hand, and
on the other indicates that the sponges are covered by a forest of pinules with
well-distinguished distal ray.

Doubtful Caulophacid.

Plate 32, figs. 10-12.

The collection contains three fragments of skeleton-nets collected with the
tangles at Station 3689 (A. A. 134) on 28 October, 1899; 18° 06' S., 142° 24' W.;
depth 1476 m. (807 f.) ; they grew on a bottom of fine coral-sand and manganese
nodules; the bottom-temperature was 37.6°.

» /. Ijima. Studies on the Hexactinellida. III. Jouin. Coll. sci. Tokyo, 1903, 18, p. 79,80, 112, 114.
2 F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 174, 176.

ROSSELLINAE. 93

The largest fragment (Plate 32, fig. 10) is apparently part of a stalk. It is
99 mm. long., 12 mm. thick at the base, and 17 mm. at the upper end. At the
thinner, probably the lower, end it is solid for a distance of 18 mm. ; farther on it
is tubular. The tube thus formed widens above in a calyculate manner and has
a wall about 3 mm. thick. The two other fragments which are 57 mm. and 49
mm. long respectively are curved lamellae, also about 3 mm. thick. They
apparently formed parts of lamellar, calyculate sponges.

The skeleton-net (Plate 32, figs. 11, 12) consists of main-beams 100-300 n
thick, which extend obliquely longitudinally or, more rarely, transversely. These
main-beams form more or less distinct bundles, within which they lie quite close
together and connected by numerous short, somewhat thinner, transverse beams.
The bundles are on an average about 0.6 nmi. thick and form a network with
elongated spindle-shaped meshes, usually 0.4-0.6 mm. broad and 1-3 mm. long.
The indi\ddual beams are either smooth or bear a few scattered, low, blunt
spines. Here and there the spines are more crowded.

In some respects these specimens resemble the skeleton-nets of the stalk
and the lower parts of the body of caulophacid hexactinellids, and they may have
formed parts of such.

ROSSELLIDAE (F. E. ScmiLZE) Ijima.

Hexasterophora with special diactine to hexactine dermal, and pentactine
hypodermal spicules, which latter sometimes project a considerable distance
beyond the surface, their lateral rays then forming a veil covering the sponge.

The collection contains twenty-two more or less complete specimens and
twelve fragments of this family.

F. E. Schulze ^ distinguishes three subfamilies in this family, all of which
are represented in the collection.

Rossellinae F. E. Schulze.

RosseUidae without discoctasters and plumicomes.

The collection contains twenty more or less complete specimens and twelve
fragments of specimens belonging to this subfamily. All are referred to the
new subspecies Bathydorus laevis spinosissimus.

' F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 176.

94 BATHYDORUS LAEVIS SPINOSISSIMUS.

BATHYDORUS F. E. Schulzb.

Thin-walled, sac-shaped, calyculate or lamellar Rossellidae (Rossellinae)
with oxyhexasters and sometimes also hemioxyhexasters but no other micro-
scleres in the choanosome. With hypodermal pentactines. The dermal
spicules are usually chiefly tetractines (stauractines), but forms with fewer (one
to three) or more numerous (five or six) rays may also occur, the forms with
fewer rays sometimes predominating. The gastral spicules are hexactines, the
distal ray of which may be differentiated so as to render these spicules somewhat
pinule-like.

Bathydorus laevis F. E. Schulzb.

F. E. ScHULZE, Abh. Akad. Berlin, 1895, 1896, p. 57, taf. 6, figs. 1-10. F. E. Schulzb, Sitzungsb. Akad.
Berlin, 1S97, p. 535. I. Ijima, Annot. zool. Jap., 1898, 2, p. 47. F. E. Schulzb, Indian Triaxonia,
1902, p. 78, pi. 14, figs. 1-10. H. V. Wilson, Mem. M. C. Z., 1904, 30, p. 51, pi. 5, figs. 11-13; pi.
6, f\gs. 1, 2.

Bathydorus laevis spinosissimus, sulisp. nov.
Plate 14, figs. 1-32; Plate 15, figs. 1-22; Plate 16, figs. 1-24.

All the specimens of this subspecies were trawled off northern Peru, Station
4651, on 11 November, 1904; 5° 41.7' S., 82° 59.7' W.; depth 4063 m. (2222 f.);
they grew on sticky, fine, gray sand; the bottom-temperature was 35.4°. Three
specimens distinguished as A, B, and C were examined in detail.

These sponges are related to Bathydorus laevis F. E. Schulze (loc. cit., 1896,
p. 57). Within this species Wilson (loc. cit., p. 51) has distinguished the sub-
species spinosus, which differs from the typical B. laevis by its spicules, particu-
larly its dermals, which are much more spiny. In the specimens here described
the dermals are still more spiny, and the pentactines, which are smooth in B.
laevis spinosus, are also usually covered with spines. The name of this new sub-
species refers to this further development of the spinulation.

Shape and size. The best preserved specimens, one of which C (Plate 14,
fig. 13), show calyces with rather broad bottom and a thin undulating wall.
The margin, which is much torn, appears to have been lobose in the living sponge.
In the calyculate specimens the two halves of the calyx-wall are quite flattened
and pressed against each other. In the fresh state these calyxes were, no doubt,
open. The other, fragmentary specimens are lamellae, which appear to have
formed part of calyxes similar to those described above. The walls of the
calyculate specimens are mostly 1-2 mm. thick, and thin out towards the margin.
The fragmentary lamellae are 1-2.5 mm. thick. The largest calyculate speci-

BATHYDORUS LAEVIS SPINOSISSIMUS. 95

men (C) is 39 mm. high and of varying breadth, 7 mm. below, 45 mm. above.
This specimen has a slender protuberance 11 mm. long and 2.5 mm. thick,
which arises from the outer (lower) convex side of the broad rounded bottom of
the calyx. Another calyculate specimen is more slender, 28 mm. high, 8 mm.
broad below increasing to 25 mm. above. The lamellar fragments measure
25-56 mm. in maximum diameter.

Pores, mostly 200-400 m in diameter, on both sides of the calyx-walls
are observed. These are now open. In the living sponge they were probably
covered by (dermal and gastral) sieves. From the inner and the outer surface
large and small prostal spicules protrude. Most of these, particularly the larger
ones, are very slanting and enclose small angles with the surface. Pores are
observed also on the surface of the fragmentary lamellae, but these no doubt in
consequence of post mortem shrinkage and maceration are much wider than in
the better preserved calyculate specimens and have a maximum diameter of
1.5 mm. These fragmentary, lamellar specimens show but little of the protrud-
ing prostal spicules.

The colour in spirit is light dirty brown with a greenish tinge.

General structure. The superficial pores above referred to lead into canals,
in specimen A 300-400 n wide, which traverse nearly the entire thickness of the
lamella in a somewhat obUque direction (Plate 14, figs. 14, 15). Indications of
flagellate chambers can be made out in parts of the sections of this specimen.
It seems that they are small, spherical or shghtly oval, and 70-120 n in diameter.

The skeleton of the interior consists chiefly of rhabds and oxyhexasters.
The former are exceedingly variable in size and extend paratangentially and
obliquely. The proximal parts of the prostals, which are imbedded in the
choanosome, also take part in the formation of its skeleton. Beneath the dermal
and the gastral membranes paratangentially situated rhabds form loose reticula-
tions. Fairly numerous hypodermal pentactines with rather long lateral rays
arranged in the usual manner are observed below the dermal membrane. Hypo-
gastral pentactines appear to be absent. The dermal membrane is occupied
by dense masses of spicules, the rays of which are on an average about twice as
long as the distance between their centres. These spicules are mostly regular
tetractine stauractines, but similar spicules with one or two shortened or entirely
suppressed rays (irregular stauractines, triactines, and diactines) occur. Similar
spicules with five or six rays (pentactines and hexactines) also occur in the dermal
membrane. The gastral membrane is occupied by more slender-rayed pinule-
like hexactines, generally with one more or less differentiated, outwardly directed

96 BATHYDORUS LAEVIS SPINOSISSIMUS.

ray. These gastral spicules do not lie quite so close together as the dermals.
The prostals which protrude from both surfaces are large and small rhabds.

Besides the spicules described above, which I have observed in situ in the
sections, some other forms, which I am inclined to consider as proper spicules of
the sponge, also occur in the spicule-preparations. These are: — hemioxy-
hexasters; angularly bent diactine megascleres; hexactine megascleres with
fairly equal rays; hexactine megascleres, with one ray much longer than the
other four; and pentactine megascleres, with relatively short lateral rays. The
hemioxyhexasters, which are similar to the oxyhexasters, doubtlessly form part
of the skeleton of the interior. The angularly bent diactines, and the hexactines
with rays fairly equally long, may also take part in the formation of the interior
skeleton. According to Schulze ^ such hexactines occur in the choanosome of
the type of Bathydorus laevis. About the hexactines with one long ray and the
pentactines with long proximal ray I have my doubts. Wilson - says that
hexactines with one elongated ray, 10 mm. long, occur in Bathydorus laevis
spinosus and that these spicules are here so situated that their elongated ray
protrudes freely beyond the surface, prostal-fashion. The hexactines with one
elongated ray observed by me were much smaller and made rather the impres-
sion of being derivates of hypodermal pentactines, with a short apical distal
ray. The pentactines with short lateral rays are probably also hypodermal.

The rhabds (Plate 14, figs. 1-10) vary exceedingly in size, and a continuous
series of intermediate forms connects the smallest with the largest. They are
1-21 mm. long, and 5-105 fi thick at the thickest point. The small rhabds are
distinctly centrotyle (Plate 14, figs. 5, 6), many of the large ones without a
central tyle. The four rudimentary rays which compose the tyle are often very
clearly distinguished, particularly in the small rhabds. Not infrequently they
are unequal in length, in which case the tyle formed by them appears eccentric
(Plate 14, fig. 5). The tyle may measure 22 fj. more in transverse diameter than
the adjacent parts of the spicule. This difference is not only relatively but also
absolutely greater in the small and slender than in the large and stout rhabds.
Differences of over 11 /^ in thickness of tyle and adjacent parts of the spicule
were only observed in rhabds less than 20 m thick.

The end-parts of the rhabds are conic (Plate 14, fig. 7), cylindroconic
(Plate 14, figs. 7, 9), or cylindrical (Plate 14, figs. 1, 3, 10), and terminally

' F. E. Schulze. Hexactinelliden des Indischen Oceans, II. Abh. Akad. Berlin, 1895, 1896, p. 58,
taf. 6, fig. 2; Indian Triaxonia, 1902, p, 79, pi. 14, fig. 2.
« H. V. Wilson. Mem. M. C. Z., 1904, 30, p. 52.

BATHYDORUS LAEVIS SPINOSISSIMUS. 97

rounded or, rarely, sharp-pointed (Plate 14, fig. 2), or first thickened and then
attenuated to a blunt point (Plate 14, fig. 4). The ends of the normal rhabds,
with rays not differing very much in length, are 3-45 ix thick, which is a sixth to
three quarters, sometimes nearly quite as thick as their middle-part. The small
rhabds are on the whole less attenuated towards their ends than the large ones.

The two rays composing the rhabds usually differ more or less in respect to
the shape and thickness of their ends, and also in respect to their length. One
end is often much more blunt than the other, and the difference in the thickness
of the two ends is sometimes so great that one end is more than twice as stout
as the other. The difference of the two rays in length is usually inconsiderable ;
occasionally, however, one ray is reduced very considerably in length, and then
this difference is great. In four rhabds of this kind one (the normal ray) meas-
ured over 2 mm. long, the other (the reduced ray) only 100-290 n. These
greatly shortened rays are terminally thickened either gradually or abruptly,
in which latter case their end appears as a terminal tyle, the transverse diameter
of which may be nearly twice as great as that of the base.

In some of the large rhabds, particularly the large prostals, one or more
thickenings are observed some distance below the end (Plate 14, figs. 7-9).
Occasionally such thickenings also occur near the centre of the spicule.

The end-parts of all the small and most of the large rhabds for a distance of
about 80-100 y. are covered with vertically arising spines 1-2 /x high (Plate 14,
figs. 1-4, 8-10). Some of the largest rhabds appear to have smooth end-parts
(Plate 14, fig. 7). Apart from their end-parts the rhabds are perfectly smooth.
On the end-parts of some rhabds the spines are more numerous than on the end-
parts of others. Also in this respect the two rays of the same rhabd often differ.
The ends of the rays reduced in length are always densely spined.

Two kinds of angularly bent diactines (Plate 16, fig. 19) can be distin-
guished, one with an obtuse angle between the two rhabd-rays, the other with
an angle of 90° or less. The former are similar to the rhabds described above,
from which they differ by the angle between the rays being only about
120° instead of 180°. I consider these diactines as derivates of the ordinary
rhabds. In the latter the angle between the two rays is usually 75°-90°. The
rays of these spicules are 0.3-1.5 mm. long, 7-20 ^ thick at the base, straight, and
rather unequal in length. At the point of junction of the two rays the spicule is
thickened to a conspicuous tyle. I have not observed any angularly bent diac-
tines with angles of 90°-120° which might be considered as transitions between
the two kinds of these spicules observed, and I am not sure whether the forms

98 BATHYDORUS LAEVIS SPINOSISSIMUS.

with angles of 90° or less are to be considered as rhabd-derivates like those with
obtuse angles.

In all these diactines the two rays are of equal thickness at the base. I
found, however, also an angularly bent diactine with an angle of 73°, in which
one ray, 460 n long and 7 n thick at the base, was curved concave towards the
other, the other being straight, 850 n long and 23 ix thick at the base.

The pentactines and apparently pentactine-derivate hexactines. Among the
pentactines two groups can be distinguished, one with relatively long, and the
other with relatively short lateral rays. The hexactines appear to be derivates
of pentactines belonging to the first group.

The pentactines with relatively long lateral rays (Plate 16, figs. 4-8, IG, 17,
20-24). The proximal ray is 0.5-1.5 mm. long, usually straight, conic, and 20-
38 fi thick at the base, and gradually attenuated to the blunt end, which meas-
ures 4-8 n in transverse diameter (Plate 16, figs. 4-6, 8). Rarely the proximal
ray is either curved, or nearly cylindrical, rounded, and slightly thickened at the
end (Plate 16, fig. 7). The lateral rays are generally straight, conic, and blunt,
those of the same spicule being usually not very different in length (Plate 16,
fig. 4). Sometimes, however, one of the lateral rays is either shghtly curved
(Plate 16, fig. 16) or greatly reduced in length, cylindrical, and terminally rounded
(Plate 16, fig. 17). The longest lateral ray of the pentactine is 270-750 ju long,
the shortest 60-700 /^. The lateral rays are usually about 2 n thinner at the base
than the proximal ray of the same spicule. Their ends are 2.5-15 n thick. The
lateral rays enclose angles of 76°-97°, usually considerably less than 90°, with the
proximal ray.

In many of these pentactines all the rays are rather densely covered with
spines throughout their whole length (Plate 16, figs. 15, 20-23). These spines
are conic, sharp-pointed, about 1 ii high and, on an average, 4 n apart. Towards
the ends of the rays the spines usually become smaller and less numerous. The
ends themselves, however, generally bear considerably larger spines, which either
pass gradually into the smaller ones, or are separated from them by a distinct
limit, situated a short distance from the end of the ray. In some pentactines
the spiculation is not so great, portions of the rays appearing quite smooth. In a
few hardly any spines, or no spines at all, could be detected.

When these spicules are slightly heated, the superficial silica-layers partly
spht off and it is then clearly to be seen (Plate 16, figs. 20, 21) that the limits
between the outermost and the next silica-layers are perfectly smooth. From
this it follows that the spines are not formed until after the spicule has attained

BATHYDORUS LAEVIS SPINOSISSIMUS. 99

its full size. The partly and wholly smooth pentactines, above referred to,
should, I think, therefore be considered as not completely developed, adolescent
spicules, in which the spines are not yet, or as yet only partly, formed.

The rays of the rare apparently pentactine-derivaiive hexactines (Plate 16,
fig. 3) are 12-25 n thick at the base. One of them is elongated and 570 n-2 mm.
long. This ray corresponds with and is similar to the proximal ray of the pentac-
tines above described. The four rays vertical to this elongated ray are, in the
same spicule, more or less unequal in length, the longest being 160-500 ^u long,
the shortest 135-225 m- They correspond with and are similar to the lateral
rays of the pentactines. The sixth ray, which lies in the continuation of the axis
of the elongated one, is straight, conic, blunt, and 88-420 n long.

The pentactines with relatively short lateral rays (Plate 16, iigs. 1, 2) have an
apical (probably proximal) ray 780 ^-2.7 mm. long and 13-22 /x thick at the base.
This ray is generally more or less curved. It is nearly cylindrical in its proximal
part and gradually attenuated to a blunt end. The lateral rays of the same
spicule usually differ in length, the longest being 200-290 n, the shortest 145-
221 n long. They are at the base about as thick as the proximal ray, cylindrical,
and blunt. The rounded end is usually one to two thirds as thick as the base of
the ray. The lateral rays enclose angles of considerably less than 90° with the
apical (probably proximal) ray, and are usually curved, concave to the latter.
The lateral rays of these pentactines exhibit the saine spinulation as the pentac-
tines with long lateral rays described above. The proximal ray is less spiny,
sometimes apparently quite smooth.

The rare regular hexactines with fairly equal rays measure 0.6-2 mm. in
diameter and have mostly smooth, rather straight, cylindroconic, terminallj'
rounded rays 0.35-1.1 mm. long and 15-40 n thick at the base.

Besides these regular ones I have found a few irregular hexactines, one of
which is represented on Plate 16, fig. 18. This spicule has rays 250-830 m long.

The dermal spicules are di- to hexactine, by far the greater number of them
being tetractine (stauractine) . Most of these stauractines are fairly regular,
having four properly developed, straight rays differing only slightly in length
and enclosing equal angles with their neighbours. Besides these a few staurac-
tines occur in which either one, two, three, or all four rays are greatly reduced
in length, or one or more rays are strongly bent, or the angles between the rays
are unequal.

The regular stauractines (Plate 14, fig. 11; Plate 15, figs. 1, 2, 19; Plate 16,
figs. 13, 14) generally measure 80-215 m in diameter. In specimen A, I have

100 BATHYDORUS LAEVIS SPINOSISSIMUS.

found besides the ordinary ones also, however, a few 220-320 n in diameter.
Their rays do not lie in one plane but form the edges of low obtuse pyramids
with quadratic bases. In consequence of this and the fact that the rays are,
in the same spicule, nearly equally long, the ray-length is a little more than half
the diameter of the spicule. The basal thickness of the rays is in the regular
stauractines 80-215 y.; the diameter is 4.5-9, generally 5-7 ju. The rays of the
giant stauractines 220-320 yu in diameter, above referred to, are 6-15 y. thick at
the base. The rays are generally terminally rounded and either cylindrical, at
the end as thick as at the base; or, more frequently, cylindroconic, at the end
only one to two thirds as thick as at the base; very rarely the ends are pointed.
In respect to the degree of attenuation towards the end the rays of the same
spicule are often unequal.

The whole of the spicule is densely and uniformly covered with sharp conic
spines, its central part being quite as spiny as the distal parts of its rays. The
proximal spines are nearly vertical, the distal ones directed more or less obliquely
outward. Large and smaller spines are irregularly intermingled; the largest
are sometimes 4 n long.

Stauractines with one or more rays reduced in length (Plate 15, figs. 5, 9, 10,
18, 21, 22) are quite frequently met with. Apart from the ray-reduction these
spicules resemble the regular stauractines above described. When two of their
rays are reduced, these reduced rays may be either adjacent (Plate 15, fig. 21)
or opposite (Plate 15, fig. 5). A stauractine in which all the four rays are
reduced is represented (Plate 15, figs. 9, 10). This spicule is only 30 ix in diame-
ter, and has cylindrical, terminally rounded rays 8 n thick.

Irregular stauractines with unequal interactine angles (Plate 16, fig. 12) or
with curved rays (Plate 15, fig. 11) are met with much more rarely. Apart
from the irregularities characteristic of them, they also resemble the regular
stauractines above described.

Among the dermal spicules with less than four rays, which are doubtlessly
to be considered as stauractine-derivates with reduced ray-number, triactine and
diactine forms occur. Most of the triactine stauractine-derivates (Plate 15,
figs. 4, 6) are straight or curved rhabds, 83-125 m long, from the central part of
which arises a ray-rudiment 9-12 n long. Some of them, however, appear as
more or less regular triactines with rays nearly equally long, enclosing fairly
equal angles with their neighbours. The diactine stauractine-derivates are
straight, or slightly curved, or strongly angularly bent. The latter resemble
more or less widely open compasses. In regard to the thickness of their rays

BATHYDORUS LAEVIS SPINOSISSIMUS. 101

and their spinulation the triactine and diactine stauractine-derivates resemble
the regular stauractines above described.

The dermal spicules with more than four rays, which I am inclined to con-
sider as stauractine-derivates with increased ray-number, are pentactines and
hexactines. The pentactine forms (Plate 15, fig. 20), which are met with rather
frequently, have four fairly equal rays similar to those of the regular stauractines,
and a fifth shorter ray vertical to the plane of the tips of the four others. The
hexactine forms (Plate 15, fig. 3) are very rare. They appear either as fairly
regular hexactines with nearly equal rays, enclosing angles of 90° with their
neighbours; or they are irregular, having rays unequal in length and irregular
in position. These spicules are 100-200 ij. in diameter. In regard to the thick-
ness of their rays and their spinulation they resemble the regular stauractines
above described.

The more or less pinule-like gastral hexactines (Plate 14, fig. 12; Plate 15,
figs. 7, 8, 12-17; Plate 16, figs. 9-11) have five quite similar and one differentiated
ray, which latter corresponds to the distal ray of true pinules. This differen-
tiated (distal) ray is straight (Plate 15, figs. 12, 17; Plate 16, figs. 9, 10); or,
much more frequently, curved (Plate 15, figs. 7, 8, 13-16; Plate 16, fig. 11), its
curvature often being very considerable (Plate 15, figs. 8, 14, 16). It is at the
base 2.5-6 n thick and, measured along its chord, 70-145 n. It is attenuated
uniformly towards the end, or cyhndrical in its proximal and conic in its distal
part, or even sUghtly thickened near the middle, and always terminates in a fine
point. It bears spines along the whole of its length. The spines are small,
rather scarce, and nearly vertical on its basal part. Farther on they become
more numerous, larger, and inclined towards the end of the ray. They attain
their maximmii length of 3-8 m about half way up. Beyond this point the
spines again become smaller, but retain their incUnation towards the tip of the
ray. The large spines are usually somewhat curved, concave towards the end
of the ray. At its thickest (most bushy) point the distal ray is, together with the
spines, 6-15 ^ in transverse diameter. In some of the gastral hexactines of
specimen A the distal ray is stouter and more bushy than in the gastral hexac-
tines of specimen B, where its basal thickness does not exceed 4.5 n, and its
maximmn transverse diameter, with the spines, 10 /x.

The four rays vertical to the differentiated one, which correspond to the
four lateral rays of true pinules, are straight, usually rather abruptly pointed,
43-94 n long, and 3-6 m thick at the base. They are covered with spines 1.5-
2.5 n long, which either arise vertically, or are incUned towards the tip of the ray.

102 BATHYDORUS LAEVIS SPINOSISSIMUS.

These four otherwise quite similar rays sometimes differ considerably in regard
to their spinulation. Thus the right lateral ray of the hexactine (Plate 15,
fig. 7) bears more numerous, larger, and more inclined spines than the left one.
The sixth ray, which lies in the continuation of the axis of the differentiated one
and which corresponds with the proximal ray of the true hexactine pinule, is
similar to the four (lateral) rays above described in size and spinulation, but is
slightly, sometimes considerably, curved (Plate 15, fig. 8). Also these rays are
on the whole stouter in the gastral hexactines of specimen A than in those of
specimen B.

The oxyhexasters and (rare) hemioxyhexasters (Plate 14, figs. 16-32) measure
65-135 M in diameter. The main-rays are cylindrical, smooth, 4-12 yu long, and
2-4 M thick. They enclose angles of 90° with their neighbours. The main-rays
of the same spicule are equal. From the distal end of each main-ray a verticil
of from two to four end-rays, rarely only a single end-ray, arises. The main-rays
are often unequal in respect to the number of end-rays which they bear. A
single end-ray is found only on one or two of the main-rays of the spicule, the
others bearing more than one. The spicules with a single end-ray on one or two
of the main-rays, which must be designated as hemioxyhexasters, resemble the
true oxyhexasters in every respect except in regard to the end-ray number.
The end-rays arise very steeply from the main-rays but very soon curve outward,
that is towards the continuation of the main-ray axis, and then straighten out,
their distal and middle-parts being only slightly curved, or quite straight. The
angle between the chord of the end-ray and the continuation of the main-ray
axis is, on an average, about 45°. The end-rays are 30-60 ^ long and, at the
base, 1.6-2.5 /x thick, rarely as much as 3 yu. They are conic and taper gradually
to a very fine point. The end-rays are covered with rather sparse, backwardly
directed, slender spines, which decrease in size from the base to the tip of the
ray (Plate 14, figs. 24-32). The largest of these spines are 0.5-2 ^ long.

The description given above shows that these sponges are most closely
allied to Bathydorus spinosus F. E. Schulze ^ and Bathydorus laevis F. E. Schulze
(Schulze and Ijima, loc. cit.) within which latter Wilson (Mem. M. C. Z., 1904,
30, p. 51, pi. 5, figs. 11-13; pi. 6, figs. 1-2) has distinguished the subspecies B. I.
spinosus. According to F. E. Schulze {loc. cit., 1897, p. 535) B. laevis and B.
spinosus are very similar and may be specifically identical. Judging from

1 F. E. Schulze. Rept. Voy. ChaUenger, 1887, 21, p. 153, pi. 50, figs. 0-9. Sitzungsb. Akad. Berlin,
1897, p. 534. /. Ijima. Annot. zool. Jap., 1898, 2, p. 46. E. Topsent. Res. Voy. Belgica, 1901,
p. 36, taf. 1, fig. 1.

LANUGONYCHIA FLABELLUM. 103

E. Topsent's description (loc. cit., 1901, p. 36) the B. spinosus examined by him.
resembles B. laevis F. E. Schulze and also Wilson's subspecies B. laevis spinosus.
Of all the descriptions cited, Wilson's of B. laevis spinosus {loc. cit., 1904, p. 51)
agrees best with the sponges here under discussion. The latter differ, however,
from that subspecies, and also from B. laevis and B. spinosus, by the pentactines
being generally densely covered with small spines. For this reason, and because
there are also other, minor differences, I establish a new systematic unit for
them, which should, I think, be a subspecies of Bathydorus laevis, equivalent to
Wilson's B. laevis spinosus.

However should future studies prove, as seems probable, that Bathydorus
laevis and B. spinosus are identical, the latter having priority, the name Bathy-
dorus spinosus spinosissimus would prevail.

The specific name of Bathydorus laevis is variously given as laevis and levis.

Lanuginellinae F. E. Schulze.

Rossellidae with plumicomes, but without discoctasters.
The collection contains one specimen of this subfamily, a new species of the
new genus Lanugonychia.

LANUGONYCHIA, gen. nov.

Rossellidae (Lanuginellinae) with onychhexasters, discohexasters, and plu-
micomes. Without octasters. The superficial skeleton consists of hexactines
in which all the six rays or only five, four, three, two or one are normally
developed, the others being reduced to terminally rounded protuberances. The
unreduced forms (true hexactines) preponderate in the gastral, the strongly
reduced forms (hexactine-derivate diactines and monactines) are restricted to
the dermal membrane.

Lanugonychia flabellum, sp. nov.
Plate 12, figs. 20-3-4; Plate 13, figs. 1-28.

The unique specimen of this species was found northeast of Easter Island
at Station 4695 on 23 December, 1904; 25° 22.4' S., 107° 45' W.; depth 3694
m. (2020 f .) ; it grew on fine, Ught brown ooze.

On account of its being fan-shaped I name it flabellum.

Shape and size. The single specimen is somewhat fragmentary and macer-
ated. It appears (Plate 13, fig. 8) as a flat, elongated, irregularly triangular

104 LANUGONYCHIA FLABELLUM.

lamella about 60 mm. long, 40 mm. broad, and 1 mm. thick, from the sharpest
angle of which arises a uniformly curved stalk about 90 mm. long and 2-3 mm.
thick. Whether the complete sponge, of which the specimen formed a part,
was also fan-shaped, is hard to say. It may have been calyculate or even tubu-
lar, but it certainly was thin-walled and stalked.

The colour in spirit is rather dark reddish brown.

General structure. The lamellar body is reticulate in structure (Plate 13,
fig. 14), composed of a network of bands, mostly 0.2-0.5 mm. thick, which enclose
round meshes, 1 mm. wide. These meshes are partly covered by remnants of
superficial membranes.

The skeleton of the stalk (Plate 13, fig. 7) consists chiefly of longitudinal
beams 20-90 ^ thick (usually 40-80 m), about equally far apart, and joined at
frequent intervals by short transverse bars. The latter are thickened, and
trumpet-shaped at the base. The meshes of the whole ladder-like network
formed by the beams and bars are rounded, usually oval, 35-210 fi long and 25-
90 n broad. Most of the longitudinal beams are rhabds; some appear to be
elongated rays of pentactines and hexactines.

The interior of the lamellar body is occupied by rhabds and great numbers
of onychhexasters, and large and small regular discohexasters. Plumicomes
and a few irregular discohexasters with primary and secondary end-rays have
also been found in it. Besides these spicules several large amphiasters have been
observed. These are, however, most likely foreign to the sponge. Very proba-
bly small hexactines also occur in the choanosome. I am not, however, certain
about these spicules ; the ones observed may in truth have been gastral or dermal
and brought down into the choanosome accidentally. Below the surface pentac-
tines are met. The superficial (dermal and gastral) skeleton consists of hexac-
tines and pentactine to monactine hexactine-derivates with only five to one
properly developed and one to five reduced rays, which latter appear as short,
terminally rounded protuberances of the centre of the spicule. The tetractine
forms are stauractines; the diactine forms are mostly centrotyle amphioxes; a
few of them appear compass-shaped. On one side of the lamella true hexactines,
with all six rays fully developed, greatly predominate, spicules with only five or
four fully developed rays (pentactines and stauractines) being rare, and spicules
with only three or still fewer (triactines to monactines) absent altogether. On
the other side of the lamella hexactine-derivates with fewer than six fully
developed rays are more frequent than true hexactines and here also triactine to
monactine forms with only from three to one fully developed rays are frequently

LANUGONYCHIA FLABELLUM.

105

met, the diactines being particularly abundant. Judging l)y analogy I should
say that the surface, the skeleton of which consists chiefly of true hexactines, is
gastral, the other dermal.

The hexactine and pentactine forms are orientated in such manner that
four of their rays extend paratangentially whilst one protrudes vertically out-
ward. The stauractines, triactines, diactines, and monactines are usually
extended wholly paratangentially.

The rhabds are 4-20 mm. long and 5-140 n thick near the middle. Those
5-50 M thick are usually 4-7 mm. long. The slender ones are always distinctly
centrotyle, the tyle being 1-6 ^ thicker than the adjacent parts of the spicule.
In the stout rhabds the central tyle is only slightly developed, inconspicuous,
and often altogether absent. The axial cross is equally developed in the stout
non-centrotyle and the slender centrotyle rhabds. The smallest rhabds are
nearly cyUndrical and rounded at the ends. The rhabds 20^0 /i thick in the
middle taper gradually to 5-18 n towards the ends, which are usually unequally
stout and simply rounded off. The large, stout rhabds generally have blunt,
somewhat irregular, conic termini and are, just below the end, considerably
thinner than the small slender rhabds. The measurements of five rhabds,
tabulated below, indicate that these spicules are the more centrotyle and the
more cylindrical the smaller they are, and vice versa.

RHABDS.

Tyle

Thickness

Length
mm.

transverse
diameter ai

difference be-
tween transverse
diameter of tyle
and thickness
of adjacent parts
of the spicule

of the spicule
close to the tyle.
near the middle

of one end m

of the other
end M

5 19

4

15

15

12

5

23

3

20

11

5

5.7

56

3

53

12

8

6.3

53

1

52

18

14

19

80

(no tyle)

SO

5

4

The ends of the rhabds are, for a short distance, covered with small spines.
Apart from tliis these spicules are smooth. The .spiculation of the end-parts is
more conspicuous in the small than in the large rhabds.

106 LANUGONYCHIA FLABELLUM.

The (hypodermal and hypogastral) pentadines (Plate 13, figs. 10, 12, 13,
16b) have an apical (proximal) ray 0.6-1 mm. long, and lateral (paratangential)
rays 400-800 /x. The lateral rays of the same spicule are more or less unequal,
the longest usually being 150-250 m longer than the shortest. All the rays are
straight, conic, blunt, and 20-40 yu thick at the base. The end-parts of the
lateral rays bear quite numerous sharp-pointed spines. Proximally these spines
become more blunt, lower, and less numerous, and they pass gradually into
slight, hardly perceptible, flattened protuberances, finally disappearing alto-
gether. The proximal parts of the lateral rays are smooth.

Pentadines with very long apical rays (Plate 13, figs. 9, 16a) have also been
observed. The apical (proximal) ray is in these spicules 3-9 mm. long. The
lateral rays are usually broken; one intact one (Plate 13, fig. 9) was 1.85 mm.
long and curved. These spicules may be foreign. Some of them are strongly
corroded.

A few large sword-like hexadines with the rays of one axis differently devel-
oped from the rays of the other two axes have also been observed. The two rays
in the differentiated axis represent the blade and the handle of the sword. The
former is very long and broken off in the spicules observed. The latter is 165 m
long and covered with spines. At the base it is 24-30 n thick and either cylindri-
cal or terniinally thickened, club-shaped. The other four rays, which represent
the guard of the sword, appear to be long and ec|ual among themselves. They
were all broken off in the sword-like hexactine observed. These spicules seem
to take part in the formation of the skeleton of the stalk ; it is possible, however,
that they are foreign.

The small hexadines and hexadine-derivates (Plate 12, figs. 24-34; Plate 13,
figs. 5c, 28) always have fairly straight rays, but are, apart from this, remarkably
variable and irregular. In the first place the angles between adjacent rays are
not, as is generally the case in hexactinellid spicules, invariably 90°. In a good
many of the tetractine (stauractine) (Plate 12, fig. 33), the triactine, and particu-
larly the diactine (Plate 12, fig. 26; Plate 13, fig. 5c) forms, other than right
angles are enclosed by them. This angular irregularity is particularly pro-
nounced in some diactines which appear as variously opened compasses (Plate 12,
fig. 26; Plate 13, fig. 5c). In the second place one to five of the rays may be
reduced to mere terminally rounded protuberances arising from the centre of
the spicule. Finally the reduced rays and, to a certain extent, also the fully
developed rays of the same spicules are frequently unequal among themselves.
In spite of this variability there are, however, absolutely no transitions between
the reduced and the properly developed rays.

LANUGONYCHIA FLABELLUM.

107

These hexactines and hexactine-derivates measure 134-318 /x in total
diameter. Their fully developed rays are 83-180 fj. long and 5-14 ju. thick at the
base. They are usually regularly conic and sharp-pointed, rarely cyhndro-
conic and somewhat blunt. The reduced rays are 6-25 yu long, 7-16 ^ thick,
cyhndrical, and terminally rounded. The length and thickness of the properly
developed rays is, as the subjoined table shows, in the monactine to pentactine
forms in inverse proportion to their number. The hexactine forms apparently
do not conform to this rule. Since, however, the state of preservation of the
specimen renders it impossible to ascertain clearly whether the not numerous
larger hexactines are choanosomal or superficial, it might be assumed that these
large hexactines are choanosomal spicules and do not belong to the series repre-
sented by the dermals and gastrals, to which that rule applies. The small
hexactines conform to the rule, and some at least of these are certainly superficial.

HEXACTINES AND PENTACTINE TO MONACTINE HEXACTINE-DERIVATES.

Pully developed rays

Length

Basal thickness

Number

limits 11

average of the
longest three fi

limits M

average of the
tliickest three m

6

83-140

132

5-10

9

5

90-118

114

5-9

8

4

95-125 s

120

6-9

8

3

100-142

141

5-11

10

2

105-lGO

152

0-11

10

1

123-190

185

10-14

13

Both the fully developed and the reduced rays are covered with spines.
On the basal parts of the fully developed rays the spines are somewhat sparse
and here they arise vertically. On their distal parts the spines are more numer-
ous and here they point obliquely outward. The extreme tip of the ray is
usually free from spines for a distance of 4 or 5 fi. The spines are conic, sharp-
pointed, and 0.5-2 n high. Their size is in proportion to the thickness of the ray
from which they arise, the stoutest rays bearing the largest spines.

The cylindrical, terminally rounded, more or less knob-like, reduced rays
are covered with similar spines, which are either smiilarly distributed as on the
fully developed rays or more crowded.

108 LANUGONYCHIA FLABELLUM.

It has been stated above that some of the fully developed rays are more
cylindroconic and less sharply pointed than the great majority of rays. Such
blunt rays have only been observed in the hexactines and pentactines, and it is
always the distal protruding ray of these spicules which exhibits this pecuUarity.
This differentiation is interesting, since it would, if further developed and associ-
ated with an increase in the size of the spines, convert these superficial hexactines
and pentactines into pinules.

The onychhexasters (Plate 13, figs. 6b, 15b, 27) measure 86-135 n in total
diameter. Their main-rays, which are regularly arranged and enclose angles
of 90° with their neighbours, are 5-8 m long, thickened at both ends, and, in the
middle, where thinnest, 2-3 n in transverse diameter. From the end of each
main-ray several, most frequently four, branch-rays arise. These are con-
siderably curved, convex to the centre of the spicule, at the base, but soon
straighten out, often, however, exhibiting slight bends farther on. They are
48-60 II long, 1.3-2 n thick at the base, and gradually attenuated to 0.3-0.6 n
at the end. They bear along their length sparse, minute, backwardly directed
spines, and on their ends two to four, most frequently three, slender terminal
spines, 3-5 n long. These usually enclose angles of 90-120° with the end-ray,
and are curved, concave towards the centre of the spicule, or nearly straight.
They generally arise from the same point, quite terminally, and form a verticil.
Sometimes, however, one is situated a little below the end of the end-ray.

All the discohexasters (Plate 13, figs. 1-4, 5a, 6a, 15a, 17-26) have very
much the same shape, but they differ quite considerably in regard to their size
and the number of their end-rays. I measured 22 of them and found that

was under 80

2 were 81-100
6 " 101-120

3 " 121-140
was 141-160

2 were 161-180
6 " 181-200

3 " 201-220
was over 221 ju

in total diameter.

This gives the following frequency-curve (Fig. 3).

From this remarkably regular double curve I conclude that two kinds of
discohexasters are to be distinguished, a large kind over 150 n in diameter and a
small kind under 150 /i.

LANUGONl'CHIA FLABELLUM.

109

. 6

>

O

CO 4

3

CO

S 2

X!
13 1

; i 1 1 ; ; ;

4 -i 1 1 L jU -j.

61 ^80*^
81 — 100-
101 — 120-
121 — 140-
141 160-
161 180 -
181 200-
201 — 220-
221 240

Fig. 3. — Diiscohexasters.

The large discohexaskrs measure 165-220 ju in total diameter. Their main-
rays, which are regularly arranged and enclose angles of 90° with their neigh-
bours, are 8-10 n long and 4.5-7 ^u thick. Each main-ray bears a terminal
verticil of usually four end-rays, which arise steeply from the main-rays, but at
once curve outwards, and are quite straight, apart from the short, curved, basal
part. The basal curvature is such that the distal straight and middle-parts of
all the end-rays become fairly concentric with the centre of the spicule, and also
fairly equidistant ; the whole discohexaster in consequence appearing as a quite
regular rosette. The end-rays are 90-105 fi long and 3.5-G fi thick at the base.
They are attenuated distally and are 1.5-3 m thick at their thinnest point, a
short distance below the end. From here they again thicken and measure, at
the end itself, 3-5 m in transverse diameter. Below the thinnest point the end-
rays bear minute backwardly directed spines. The spines are rather sparse at
the base of the ray but become very nimierous distally towards its thinnest point.

110 LANUGONYCHIA FLABELLUM.

The distal part of the end-rays, beyond the thinnest point, is smooth. At the
end each end-ray bears a verticil of seven large anchor-teeth, like recurved spines
with a maximum length of 11 /i. The basal parts of these terminal spines
coalesce to form a kind of convex terminal disc. The transverse diameter of
these terminal spine- verticils is 12-16 fi. The constancy of the number (seven)
of these terminal spines seems very remarkable, since this number is apparently
in no way connected eitlier with the triaxon (hexactine) ground plan of all hexac-
tinellid spicules, or the physical (crystallograpliic) properties of the silica of
which they consist.

The small discohexasters differ from the large ones described above only in
regard to their size and the number of their end-rays. They measure 82-140 fi
in total diameter, and have main-rays 5-8 /z long and 2.7-.5 yu thick. Each
main-ray usually bears seven or eight end-rays 36-62 /i in length. These
measure in thickness 1.5-3 ju at the base, and 0.5-1.2 n at the thinnest point near
the end, and 1-1.5 /x at the end itself. The terminal spine- verticils measure 6-
11 /i in transverse diameter.

The plumicomes (Plate 12, figs. 21-23) have a central thickening about
3.5 M in diameter and regularly arranged main-rays, enclosing angles of 90° with
their neighbours. The proximal part of the main-rays is cylindrical, and 1-1.5 n
thick. Near their end they are thickened to an oval knob, 2-3 fi in transverse
diameter, from which the end-rays arise. A terminal cylindrical rod, 0.8-1.4 /x
thick, 2-4 fi long, and rounded at the end, arises from each knob. This rod,
which lies in line with the proximal part of the main-ray, appears as its termina-
tion. The total length of the main-rays (including the terminal rod) is 10-14 ii.
The end-rays, of which there may be about twenty on each main-ray, are curved
in an S-shaped manner, and are 30-40 fj. long.

The irregular discohexasters with primary and secondary end-rays (Plate 12,
fig. 20) are very rare. I found only three. These spicules may be malformed
discohexasters. Since, however, the three observed are very much alike and
since no intermediate forms connect them with the other hexaster-forms, they
may also be spicules sui generis.

They measure 120-140 ^ in total diameter. Their main-rays, which are
regularly arranged and enclose angles of 90° with their neighbours, are 5-11 n
long and 3-7 ix thick. Each main-ray bears two or three basally curved, but for
the greater part of their length fairly straight, strongly spined, primary end-rays.
These are 50-60 fi long, 3-4 ^ thick at the base, and about 2 ;u at the end. The
ends of many of them are divided into short and stout, irregularly bent, trans-

LANUGONYCHIA FLABELLUM. Ill

verse branches. Slender secondary end-rays 8-17 m long arise from the sides
and ends of the primary end-rays and their terminal branches. The basal parts
of these are directed obhquely backwards towards the centre of the spicule, but
they at once curve strongly outward, their distal and middle-parts being fairly
straight and directed obliquely outwards. Each of these secondary end-rays
bears a terminal verticil of relatively large, recurved spines, which appears as a
terminal disc with strongly serrated margin. These terminal spine-verticils,
which measure as much as 10 /x in transverse diameter, closely resemble the
terminal spine-verticils of the discohexasters above described. In examining
these remarkable spicules I gained the impression that their secondary end-rays,
the basal parts of which are in exactly the same position relative to the primary
end-rays as the spines, might be considered as hypertrophic spines.

The amphiasters, which, as stated above, I beheve to be foreign, have
a shaft about 13 n long and 1.2 ^ thick, from each end of which arise three
branch-rays, sometimes 23 m long. These branch-rays bear secondary branches
at the end.

The known species most closely allied to the sponge described above are
Mellomjmpha velata (Wyv. Thoms.), Lanuginella pupa O. Schm., and certain
rossellinas. It differs from all these by its spiculation to such an extent, how-
ever, that a new species must be established for it. About this there can be no
doul)t. It is more difficult to decide in which genus this species should be placed.
Is it to be assigned to one of the ah-eady established genera and if so to which
one, or is a new genus to be established for it?

In regard to its internal microscleres and to its large pentactines Lanugony-
cliia flahellum resembles most closely Mdlonympha velata, the only species of
Mellonympha. Since, however, its body is lamellar and thin, since its dermal
spicules are reduced hexactines, mostly with only from one to four fully developed
rays, since it is very doubtful whether the large pentactines observed in it pro-
trude beyond the surface to form a veil, and since ordinary small, not protruding
hypodermal pentactines certainly occur in it, I hardly think it advisable to place
it in the same genus as this ovoid sponge with its large, freely protruding velar
hypodermal pentactines and its pentactine dermals.

Lanuginella pupa, the only species of Lanuginella, although also differing
from Lanugonychia very considerably in shape, resembles it more closely in
regard to its dermal and gastral spicules. It is, however, destitute of onych-
hexasters, spicules which are very abundant in Lanugonychia flabellum. Ijima '

1/. Ijima. Studies on the Hexactinellida. IV. Journ. Coll. sci. Tokyo, 1904, 18, p. 12.

112 STAUROCALYPTUS HAMATUS.

has indeed observed small and delicate oxyhexaster-like spicules in rare instances
in Lanuginella pupa. Since, however, he considers these spicules as young stages
of the discohexasters, this observation does not invalidate the correctness of
F. E. Schulze's statement ^ that the absence of onychhexasters (to which kind of
spicules F. E. Schulze considers the onychhexasters to belong) is characteristic
of Lanuginella. There being therefore no reason for altering this characteristic
of Lanuginella I accept it and am consequently unable to place the sponge above
described in Lanuginella.

Since the otherwise similar species of RosselUnae differ from Lanugonychia
flahellum by the absence of pluinicomes, and since F. E. Schulze and I. Ijima
consider the absence or presence of plumicomes in the Rossellidae as a difference
sufficient for generic distinction, I do not think it advisable to place Lanugony-
chia flahellum in any of the described genera. As it seems to be most closely
allied to Lanuginella and as it differs from this genus chiefly in that it possesses
onychhexasters, I propose Lanugonychia, the type and, at present, only species
of which is the Lanugonychia flahellum.

Acanthascinae F. E. Schulze.

Rossellidae with discoctasters.

The collection contains one specimen of this subfamily, a new species of
Staurocalyptus.

STAUROCALYPTUS Ijima.

Rossellidae (Acanthascinae) with oxyhexasters, small discohexasters, and
discoctasters, and with hypodermal pentactines the lateral rays of which are des-
titute of long curved spines.

Staurocalyptus hamatus, sp. nov.
Plate 16, figs. 25-43; Plate 17, figs. 1-25; Plate 18, figs. 1-14.

One specimen of this species was trawled at Station 4642 on 7 November,
1904; 1° 30.5' S., 89° 35' W.; depth 549 m. (300 f.); the bottom was composed
of broken Globigerina and molluscan shells; the bottom-temperature was
48.6°. It is characterised by the possession of numerous oxyhexactines and a
few hemioxyhexasters with hook-hke rays (end-rays). To this the name refers.

Shape and size. The specimen has the shape of a shallow, inverted cup.

• F. E. Schulze. Revision des systemes der Asconematiden und Rosselliden. Sitzungsb. Akad.
Berlin, 1897, p. 548.

STAUROCALYPTUS HAMATUS. 113

Its lower, concave side fits the dorsal side of a crustacean, apparently a species
of Dicranodroniia, whicK firmly holds the sponge on its back by the dorsally
du-ected, last pair of thoracic extremities. In its original position the sponge
completely covered the Dicranodroniia dorsally (Plate 18, fig. 14). Seen from
above (Plate 18, fig. 5) or below (Plate 18, fig. 6) the sponge appears oval in
outline, with a protuberance at one end. It is 35 mm. long and 28 mm. broad.
The wall of the inverted cup, formed by it, is about 3 mm. thick. Scattered
pores are observed both on the free upper convex side and the lower concave side
which rested on the back of the Dicranodromia. Those of the upper side are
mostly oval, with a maximum measurement of 1 mm. in length and 0.5 in breadth.
Those of the lower side are relatively broader, more nearly circular, and reach
1.5 mm. in diameter. Large prostal rhabds protrude both from the upper and
the lower side.

The colour in spirit is light brown.

General structure. I found a few remnants of a dermal membrane both on
the concave, lower, and the marginal part of the convex, upper side. Of a
gastral membrane no trace could be detected. The remnants of the soft parts
in the interior indicate that the sponge has sac-shaped flagellate chambers, 80-
100 M long and 50-70 ^ broad.

Skeleton. Spicule-bundles, 40-200 ^ thick, traverse the sponge. These
bundles appear to be most numerous just below the lower, concavq face of the
sponge, where they extend chiefly paratangentially. They are composed of
rhabds — of small rhabds only, or of a large rhabd accompanied and more or less
enveloped by numerous, coinital, small rhabds. Besides the rhabds forming
the bundles, isolated rhabds also occur in large numbers. Oxyhexasters, hemi-
oxyhexasters, and oxyhexactines with straight rays and end-rays, oxyhexactines
with terminally curved, hook-like rays, and discoctasters of various size are very
numerous. The last appear to be much more frequent in the interior than near
the surface of the sponge. Small discohexasters, and hemioxyhexasters with
rays, either all hook-like or partly hook-like and partly straight, are met with
in smaller numbers. Hypodermal pentactines and a few triactine megascleres
occur at, or just below, the surface. On those parts of the surface where rem-
nants of the dermal membrane are left, spiny rhabds are observed. Most of
these are simple diactine rhabds. Some are centrotyle, and a few possess,
besides the two properly developed rays, short rudiments of one or two further
rays. These spicules and a few angular diactines and stauractines, similar in
regard to size and spinulation, found in the spicule-preparations, I consider as

114 STAUROCALYPTUS HAMATUS.

the dermal spicules of the sponge. Spicules which might be considered as gas-
trals were not observed.

The choanosomal and prostal rhabds (Plate 16, figs. 25-38, 39a, b; Plate 18,
fig. 13) are usually more or less curved, and exceedingly variable in size. They
are 0.67-13 mm. long, and 5-175 m thick at the thickest point. The rhabds
under 3 mm. in length are less than 50 m thick, those 3-9 mm. in length are 40-
100 fi thick, those over 9 mm. in length, usually 100-160 /i. Although there is,
as this shows, on the whole, a certain correlation between thickness and length,
the proportion between these two dimensions is nevertheless very far from being
constant and varies between 50 to 1 and 122 to 1. The thickest point of the
rhabd may be situated at or near the middle of its length (Plate 16, figs. 29, 39a),
or it may be more (Plate 16, fig. 30) or less (Plate 16, fig. 34) approximated to
one of the ends. A tyle is met with only exceptionally. It is, when present, in
the small rhabds 4-6 m more in transverse diameter than the adjacent parts of
the spicule, and may be situated near the middle or nearer one end. Occasion-
ally it lies quite terminally, in which case the spicule appears as a tylostyle. In
the large rhabds the axial thread is usually somewhat thickened (Plate 16,
fig. 36) at several points, but an axial cross can only rarely be made out. In the
small rhabds an axial cross can generally be found. When a tyle is developed
the axial cross generally Ues in its centre. In the large rhabds the two rays taper
towards the end and are usually abruptly and bluntly pointed (Plate 16, figs. 27,
33, 35, 37), rarely rounded or sharp-pointed. In these spicules the ends are
5^-3 as thick as the thickest portion of the middle-part. In the small rhabds the
ends are cylindroconic or quite cylindrical and terminally either abruptly and
bluntly pointed, like the ends of the larger rhabds (Plate 16, fig. 26), or more
rounded (Plate 16, figs. 25, 27). In these spicules the ends are from half as thick
to quite as thick as, or even slightly thicker than, the thickest portion of the
middle-part. In the rhabds in which the thickest part lies near one end, this end
is conic and stout (Plate 16, fig. 33), the other being cylindrical and slender
(Plate 16, fig. 31).

The wh(jle of the rhabd, with the exception of the two ends, is smooth.
The ends are covered with broad, conic, vertically arising spines 0.5-1 m, rarely
2 IX long. The terminal spiny region is 40-230 ju long and passes, as the spines
become scarcer and lower, gradually into the smooth middle-part of the spicule.

In some of the rhabds an abrupt step-like attenuation occurs at a shorter
or longer distance from one of the ends. Of other rhabd-irregularities noticed
I mention slight transverse grooves which give to the contour an indented ap-
pearance. As the figure (Plate 18, fig. 13) of such a spicule clearly shows, these

STAUROCALYPTUS HAMATUS. 115

indentures are not restricted to the outer surface but affect the whole of its
superficial, clearly stratified part, down to the more homogeneous central part,
the surface of which also shows the indentures.

The hypodermal pentadines (Plate 18, figs. 8-10) have a straight or slightly
curved proximal ray, which is 0.5-2.2 mm. long and 9-22 n thick at the base.
The lateral rays are vertical to the proximal ray and in the same spicule often
unequal, the longest being 120-210 fi, the shortest 80-170 fx long. All these rays
are blunt. The tips of the lateral rays are spiny.

A triactine with one longer and two shorter rays, opposite in the same straight
line (axis), which enclosed an angle of 70° with the axis of the long ray was
observed in the spicule-preparations. The long ray of this spicule was 860 m
long, the two short rays were 260 and 280 fi. The distal parts of all the rays
were spined.

The dermal spicules (Plate 16, figs. 40^3) are usually simple, straight or
slightly curved, diactine rhabds (Plate 16, figs. 42, 43), 335-470 n long, and 7-
13 M thick at the thickest point, which is usually situated near the middle. An
axial cross can usually be discerned at or near the middle. The two rays are
cyhndroconic and terminally rounded. Their ends are usually a half to a third
as thick as the thickest portion of the middle-part of the spicule. Sometimes,
however, they are thinner than that, down to a quarter of the maximum thick-
ness of the middle, or thicker, up to nine tenths of this, or even slightly more.
The two ends of the same spicule are usually somewhat unequal, one being 1 ii
or so thicker than the other. The whole spicule is covered with conic, vertically
arising spines, 0.5-2 n long. The spines are more numerous at the ends than in
the middle. This difference in the degree of spinulation of the different parts is
the more clearly pronounced the longer the spicule is.

Besides these simple diactine dermal rhabds similar ones with a tyle, situated
either more or less centrally or, rarely, terminally, are met with. The tj^le may
be a simple thickening, and concentric with the axis of the spicule, or it may be
one sided (Plate 16, fig. 41), or composed of two protuberances (Plate 16, fig. 40).
These protuberances, which are obviously ray-rudiments, are up to 10 yu long and
covered with spines like the other parts of the spicule.

I found in the spicule-preparations a few tetractines (stauractines) and
angularly bent diactines with rays similar in regard to their spinulation to those
of the rhabds above described. The former have rather unequal rays 160-230 n
long and 9-10 ix thick at the base. One of the latter had rays 48 /n long, 8 m
thick at the base, and 5 ix at the end.

The oxyhexasters, hemioxyhexasters, and oxyhexactines with straight rays

116 STAUROCALYPTUS HAMATUS.

(Plate 16, fig. 39c; Plate 17, figs. 5-8, 9b, 10b) measure 96-165 /x in diameter
and have from one to four end-rays. The forms with partly simple and partly
bifurcated rays, that is the hemioxyhexasters with two end-rays on the branched
main-rays, appear to be the most frequent. The true oxyhexasters usually
have two or three, rarely three or four end-rays. The size of the spicule is, on
the whole, in inverse proportion to the number of end-rays. The oxyhexactines
and the hemioxyhexasters and oxyhexasters with two end-rays are 110-165 m in
diameter, the oxyhexasters with more than two end-rays on all or some of the
main-rays 96-130 fi in diameter. The main-rays (and simple end-rays) enclose
angles of 90° with their neighbours. The simple rays are 54-84 ti long, 3-4.5 tx
thick at the base, and conic. Their end is very slender and they terminate in an
exceedingly fine point. The basal part of the ray is for a short distance smooth.
Farther on it bears slender, straight, A-ery oblique spines, which point backwards
towards the centre of the spicule. The proximal spines are the largest and attain
\ ix'm length. Farther on they rapidly become smaller and on the distal part of
the ray no spines at all can be detected. This decrease of the size of the spines
towards the ray-end is either gradual throughout, or there is a step-like, abrupt
decrease a short way up. The rays of these spicules, particularly those in which
there is such an abrupt decrease of the size of the spines, resemble the threads of
exploded cnidoblasts of certain hydroids. I consider these simple rays as main-
rays with a single end-ray; their proximal smooth part is their main-ray, their
middle and distal spined part, their end-ray. The main-rays which bear end-
rays are smooth and very short, only 4-8 n long, and 3-5.5 /x thick. The end-
rays arise very steeply, often nearly vertically, from the main-rays and at once
curve outwards, so that their nearly straight distal and middle-parts enclose
angles of 30-35° with the continuation of the main-ray axis. Apart from their
basal curvature these end-rays resemble in shape and spinulation the middle
and distal spined part of the simple rays above described. The end-rays are
spined quite down to the base, are 37-75 n long and 2.5-4 n thick at the base.

Rarely hemioxyhexasters are met with some rays (end-rays) straight and
others hook-like (Plate 17, fig. 4). These spicules appear as transitions between
the straightrrayed spicules described above and the spicules with hook-like rays
to be described below. The transitional hemioxyhexaster represented (Plate 17,
fig. 4) measures 170 /z in diameter, has two hook-like simple rays, two straight
simple rays, and one main-ray with two straight end-rays.

The oxyhexactines with hook-like rays (Plate 16, fig. 39d; Plate 17, figs. 1-3,
10c) measure 140-227 m in diameter. The rays of the same spicule may be equal

STAUROCALYPTUS HAMATUS. 117

or unequal. The rays are, measured along the chord, 75-120 ix long, and 2.5-
8 ix thick at the base. They are conic and gradually attenuated to a fine point.
The proximal parts of the rays are straight and regularly arranged so as to enclose
angles of 90° with their neighbours. At a distance from the centre usually equal
to from one half to three quarters of the length of the chord of the whole ray,
the rays begin to curve either gradually or more often abruptly with a distinct
angular bend. The distal part of the ray, beyond this point, is uniformly
curved through an angle of at least 90°, usually more. Sometimes the curvature
is so great that the end points directly backwards and the end-tangent becomes
nearly parallel to the axis of the basal part of the ray (Plate 17, fig. 1). Excep-
tionally the curved end-part forms nearly a whole turn (Plate 17, fig. 10c, the
upper ray). In such cases it is clearly to be seen that the curvature is spiral,
and it seems probal:)le that it is of this nature also in those cases where the curved
part of the ray is shorter, and the true nature of its curvature not so clearly
discernible.

Like the simple rays of the straight-rayed hemioxyhexasters and oxyhexac-
tines described above, the rays of these spicules are smooth at the base, and
farther on covered with slender, oblique, backwardly directed spines, which
decrease in size distally, so that the end-part appears merely roughened or nearly
smooth.

Of hemioxyhexasters with hook-like rays I found only two or three. These
had one bifurcate and five simple rays. One of these spicules measured 210 m
in diameter; its simple rays were 3 n thick at the base.

The small discohexasters (Plate 17, fig. lOe; Plate 18, figs. 1-4, 7, lib, 12b)
measure 20-23 n in total diameter. The main-rays of the same sjncule are equal
and enclose angles of 90° with their neighbours. A central thickening, 3-4 jx in
diameter, can clearly be made out. The main-rays are smooth, 3.5-4.5 n long,
1.2-1.6 M thick in the middle, and thickened at both ends, proximally to the
centrum, distally to the somewhat extended base, from which the end-rays arise.
Each main-ray bears about 16 end-rays. The end-rays are curved, concave to
the continuation of the main-ray axis, quite considerably at the base, but only
very slightly, or not at all, towards the end. They are 7-8 m long, about 0.2 ^
thick at the base, and attenuated towards the end, which bears a thickening
about 0.8 n in transverse diameter. This terminal thickening is certainly
broader than high and convex on the outer side. However, in consequence of
its small size more cannot be made out about its shape. This thickening may
be, and, judging by analogy, probably is, a verticil of terminal, recurved spines.

118 STAUROCALYPTUS HAMATUS.

The discodasters (Plate 16, fig. 39e; Plate 17, figs. 9d, lOd, 11, 12, 13d, 14-
25) measure 58-320 fi in diameter, usually 70-260 m- They consist of six short
and stout main-rays, each of which bears several, in the regular forms, four
end-rays. Eight groups of three of these (24) end-rays, belonging to three
different main-rays, usually coalesce to as many single rays, which are divided
distally into verticils of about six terminal branches. The main-rays in the
same spicule are equal and their axes enclose angles of 90° with those of their
neighbours. They are distally rounded, 6.5-9 n long and about as thick. The
six main-rays together appear as a compact central body from which arise six
dome-shaped protuberances, placed in the positions of the corners of an octa-
hedron. Seen from above this structure appears, when standing upright (on a
corner of the octahedron), as a cross with short stout arms (Plate 17, figs. 11, 19-
23) ; when lying on one of the sides (of the octahedron) it is six-lobed in shape
(Plate 17, figs. 16-18). The eight coalesced end-ray groups of three arise from
the eight depressions between the dome-shaped tips of the main-rays, at points
corresponding to the eight faces of the octahedron. These coalesced end-ray
groups, which might be designated as pseudomain-rays, are 16-49 m long and
3-10 M thick. They are on the whole, cylindrical, but usually somewhat irregu-
lar, thickened here and there (Plate 17, figs. 14-16). The terminal branches
of these pseudomain-rays, which may be designated as secondary end-rays, are
sUghtly curved, convex to the continuation of the pseudomain-ray axis, and
diverge from it at angles of 12-16°. They are 16-1 15 ^ long, 0.7-2 ^ thick at the
base, and attenuated towards the end, where they measure 0.4-1.5 m in transverse
diameter. They bear, along their length, very obliquely situated, backwardly
directed spines, which are sometimes 1.5 m long and somewhat curved. Their
end is crowned by a terminal verticil of similar but stouter and more divergent
recurved spines, which together form a sort of terminal disc with serrated mar-
gin, 1-2.5 M in transverse diameter (Plate 17, fig. 24).

The great differences in the size of the discoctasters is due chiefly to differ-
ences in the length of the secondary end-rays, 15 n in the smallest, 115 ^ in the
largest, and to a small extent also to differences in the length of the pseudomain-
rays, 16 M in the smallest, 49 m in the largest. The main-rays are in the largest
discoctasters only 3 ^ longer than in the smallest.

Not infrequently (Plate 17, figs. 13d, 14) a simple ray (end-ray), curved at
the base and straight farther on, arises directly from the central mass composed
of the main-rays, between the pseudomain-rays. These simple main-rays are
27-31 M long, and 1.5-2 fx thick at the base. They are attenuated distally and

FARREA OCCA SCUTELLA. 119

provided with lateral spines and a terminal verticil ("disc") of such, like the
secondary end-rays. I consider these simple rays as ordinary end-rays which
have not coalesced with others to form pseudomain-rays and which are not
divided distally into branches (secondary end-rays).

I am inclined to consider the specimen above described as the basal part of a
higher, perhaps cup-shaped, sponge, the upper parts of which may have been
either nipped off by the Dicranodromia, which used it as tent and shield, or torn
off during capture.

Since hexactine megascleres are absent and since the sponge possesses
hypodermal pentactines, mostly diactine spiny dermals, oxyhexasters, hemioxy-
hexasters, microoxyhexactines, small discohexasters, and discoctasters, 1 tliink
there can be little doubt that it belongs to Staurocalyptus, although its gastral
spicules are unknown. It differs from all the species of this genus hitherto
described by the possession of oxyhexactines and hemioxyhexasters with hook-
like rays. This and other minor differences necessitate the establishment of a
new species for it.

EURETIDAE Zittel.

Hexasterophora the body of which is calyculate or composed of ramified or
anastomosing, thin-walled tubes. With a firm reticulate supporting skeleton-
net. Among the free spicules are always uncinates and either scopules or
clavules. With oxyhexasters or discohexasters or both.

The collection contains nine more or less complete specimens and twenty
fragments of this family. The generic position of two specimens and twelve
fragments is doubtful. The others belong to the two genera Farrea and Eurete.

FARREA BOWERBANK.

Euretidae with clavules, without scopules.

There are four more or less complete specimens which represent a new
variety of Farrea occa Bowerbank. Eight fragments apparently belong to two
distinct forms which, however, cannot be specifically determined.

Farrea occa scutella, var. nov.
Plate 25, figs. 25-29; Plate 26, figs. 1-21; Plate 27, figs. 1-17.

The collection contains four more or less fragmentary specimens of this
sponge, all trawled off the southern coast of western Panama at Station 4621 on
21 October, 1914; 6° 36' N., 81° 44' W.; depth 1067 m. (581 f.); they grew on
green mud and rock ; the bottom-temperature was 40.5°.

120 FARREA OCCA SCUTELLA.

They resemble portions of wine-glasses with stems. To this the name of
the new variety refers.

Shape and size. From an extensive basal plate, which at one time was
obviously attached to something hard on the sea-bottom, a short stem arises,
which spreads out above to form a thin, curved, lamellar body (Plate 26, figs.
16-21). One of the specimens has two basal plates and two stems (Plate 26,
figs. 18, 19). This is probably the product of a concrescence of two specimens,
originally distinct, which grew side by side.

The basal plate measures 5-17 mm. in maximum transverse diameter, is
1-2 mm. thick near the middle, and thins out towards the somewhat irregular
lobose margin. The stem is 4-7 mm. broad and 2-3 mm. high. It consists of a
vertical curved lamella, about 1 mm. thick, which appears as a portion of the
wall of an upright cylindrical tube cut through longitudinally or obliquely.
Above it is generally curved outward and abruptly extended into the lamella
which forms the body proper of the sponge. This lamella is elegantly curved
in a cylindroid or saddle-shaped manner and at the base, where it arises from the
stem, is about 1 mm. thick. Towards the margin it gradually thins out. In all
the specimens this lamella is more or less fragmentary. In the largest it is 19
mm. long and 18 mm. broad, measured along the chord.

The colour in spirit is light brown.

The skeleton consists of a network and loose hexactines, pentactines,
uncinates, oxyhexasters, clavules with large teeth, and clavules with small
teeth.

The skeleton-net (Plate 25, figs. 25, 27-29; Plate 26, figs. 8-14, 16-21)
pervades all parts of the sponge. On the lower side of the basal plate (Plate 26,
figs. 10, 11) it is very dense and consists of smooth beams, 8-20 n thick, which
enclose round meshes 10-40 fi in diameter, so that this part of it appears as a
perforated plate. On the upper side of the basal plate and in the stem (Plate 26,
figs. 12, 13) it is composed of more or less spiny beams, 6-35 fi thick, which
enclose irregular, square, or triangular meshes 30-180 ^ wide. In this region
numerous small hexactines are attached to the beams of the network (Plate 25,
figs. 25, 27-29; Plate 26, figs. 12, 13) with one thickened ray. These attached
tree-like hexactines are 75-135 m high. In some places other similar hexactines
are soldered to these attached ones, whereby rudiments of a slender secondary
network are here and there formed. In the proximal part of the lamellar body
proper of the sponge the skeleton-net consists of an inner, regular layer with
square, rectangular meshes (Plate 26, fig. 17), and an outer, irregular layer, with

FARREA OCCA SCUTELLA. 121

chiefly triangular meshes (Plate 26, fig. 16). The marginal and middle-parts of
the skeleton-net of the body-lamella (Plate 26, figs. 8, 9, 14) consist of a single
layer composed of longitudinal and transverse beams. The former are in some
places curved, in others straight, and spread out towards the margin of the
lamella in a fan-shaped manner. Here and there they divide into two equal
branches, which, at first, diverge at an angle of about 30°, but very soon become
parallel ; thus the number of the longitudinal beams increases towards the margin
of the body-lamella. The transverse beams are vertical to the longitudinal ones
and accordingly also in some places curved, in others straight. All the beams
of this network are quite smooth. The longitudinal ones are mostly 73-80 ^
thick, the transverse 75-90 //. The meshes are mostly square and rectangular,
more rarely quadratic, and exceptionally (where the longitudinal beams branch)
triangular. The rectangular ones are 280-510 ^ long and 200—100 n broad. In
some places this network is remarkably regular (Plate 26, fig. 8). From each
node of this network two thorns, 32-45 n thick at the base, arise in opposite
directions. Both are vertical to the surface in which the network extends.
One is directed dermally, the other gastrally. These thorns are fairly straight,
either conic or thickened near the end, and covered with protuberances. At the
base these thorns are broad, rounded, and 6-8 /n high; towards the end they be-
come smaller and much more slender.

Of the loose spicules the uncinates and clavules with short teeth are very
rare and also the hexactines rather scarce. The other kinds of loose spicules,
particularly the oxyhexasters, are abundant (Plate 26, fig. 8). The fragmentary
condition of the specimens renders it difficult to ascertain the position of these
spicules in the sponge. I can say, however, that there is no reason to assume
that they are arranged otherwise than in the type of this species where their
position has been described by Schulze.^

The loose hexactines (Plate 25, fig. 26) are 110-190 ij. in total diameter, and
have straight, conic, spined rays usually 3.5-4 ^ tliick at the base.

The pentactines (Plate 26, figs. 8a, 15; Plate 27, fig. 6a) have regularly
arranged lateral rays, usually 180-255 yu long. The lateral rays of the same
spicule are as a rule somewhat unequal. The difference in length between the
longest and shortest is usually 15-30 /x. Very rarely one lateral ray is greatly
reduced in length, only 120 m long, and terminally tliickened. Wlien that is the
case this difference is of course much greater. The lateral rays are straight or,
more frequently, sUghtly and uniformly curved, concave to the proximal ray.

' F. E. Schulze. Kept. Voy. ChaUenger, 1SS7, 21, p. 277 ff., pi. 71-73, 76, figs. 1-3.

122 FARREA OCCA SCUTELLA.

They are on the whole cyhndroconic, about 9 fx thick at the base, and attenuated
distally to 4-6 n. The end is rounded off. Frequently a slight thickening is
observed just before the end. The lateral rays are spiny. On the basal and
middle-jjart of the rays the spines are 2-4 ^ high and arise vertically ; on the end-
part they are 1-1.5 n high and obliquely inclined towards the end of the ray.
The spines of the lateral rays are larger in the dermal pentactines than in the
gastral. In the former they are larger and much more numerous on the outer
side of the rays than elsewhere, the inner side being often nearly destitute of spines.
On the lateral rays of the gastral pentactines the concentration of spines on the
outer side is not so pronounced. The axial thread traverses the lateral rays
quite to their ends. The proximal ray is straight, 180-260 m long, and usually
bears only small spines near the end. In most of the pentactines, particularly
the dermal ones, a rudiment of the sixth distal ray is present. This is 14-17 ix
long, and as thick as the other rays. It bears a few large, upwardly directed
spines. Sometimes only a single terminal spine is present. In this case the
distal ray (together with the spine) appears a sharp-pointed, conic thorn.

The uncinates are very rare and I cannot positively assert that those ob-
served in the preparations really belong to the sponge. An intact one was
straight, pointed at both ends, and measured 1.6 mm. long and 10 ^ thick near
the middle. Its spines were slender and 8 ^ long.

The oxyhexasters (Plate 26, figs. 1-7, 8c; Plate 27, fig. 6c) are 105-140 /x in
total diameter. Their main-rays enclose angles of 90° with each other and are,
in the same spicule, usually equal ; sometimes, however, considerable inequalities
are observed in them, the proportion of the length of the shortest to that of the
longest sometimes being 3:5. The main-rays are 22-37 /x long, straight, cyhn-
droconic, 2.8-3.8 IX thick at the base, and attenuated distally to 2-2.7 /x. They
are perfectly smooth and traversed by an axial thread, which terminates below
the end and does not give off branches for the end-rays. Of end-rays there are
one to four, usually two or three. The end-rays are slightly curved. Concave to
the continuation of the main-ray at the base, and farther on usually fairly straight,
rarely considerably and irregularly curved. They are conic, uniformly attenu-
ated to a fine point, 30-44 ii long, 1.3-2.2 ij. thick at the base, destitute of axial
threads, and, like the main-rays, perfectly smooth. When only one end-ray is
present, it extends in the continuation of the axis of the main-ray to which it
belongs. When there are two they usually enclose an angle of about 60° and lie
in or near a plane which passes through the main-ray from which they arise.
The planes in which such end-rays extend are usually oblique to the two axial

FARREA OCCA SCUTELLA. 123

planes, passing through the axis to which these forks belong and either of the
two other axes of the spicule. The end-ray forks of opposite main-rays do not
lie in the same plane. As far as I could make out the planes of such forks are
opposite, and usually symmetrical, in such manner that the angle enclosed by
them with either of the two axial planes above mentioned are supplementary;
added together they give 180°. When there are three or four end-rays the most
divergent usually enclose an angle of about 90°.

The clamles with large teeth (Plate 27, figs. 1-5, 6b, 7-11, 13-17) are gener-
ally 300-370 M long; a few are shorter, down to 210 ^ in length. They consist
of a centrum, from the lower end of which there arises a shaft, and from the
opposite, upper end of which arises a verticil of recurved teeth. The centrum is a
short cylinder, 6.5-12.5 m, usually 9-12 n, in transverse diameter, which generally
bears one or a few spines at its lower end. These spines are oblique, inclined
towards the shaft, and 0.5-2.3 m long. Their size seems to be in inverse propor-
tion to their number; the solitary ones are the largest. At the base, where it
arises from the centrum, the shaft is 4-8 m thick; its basal part is conic; farther
on it becomes nearly cylindrical; just before the end it is 2.5-4.5 m thick. The
end is abruptly and bluntly pointed and frequently slightly thickened. The
proximal and middle-parts of the shaft bear oblique spines, incUned towards
its end. These spines are similar to those on the centrum, but smaller. The
end-part bears stouter, vertical spines, 0.6-1.5 ^ long. The number of these
spines is variable. Their size appears to be in inverse proportion to their num-
ber. A smooth belt sometimes intervenes between the middle region with
obhque, and the terminal region with vertical spines. There are usually nine,
more rarely ten, recurved teeth which form the verticil at the upper, distal end
of the centrum. They are fairly equal in the same spicule, and regularly ar-
ranged, the angle between adjacent ones being the same. The verticils formed
by these teeth measure 39-53 m in transverse diameter. The indi\'idual teeth
are conic, 5-7 n thick at the base, and uniformly attenuated towards the sharp-
pointed end. They are uniformly curved, concave to the centrum, and their
chords usually enclose angles of 55°-63° with the axis of the centrum and shaft.
The teeth generally bear spines, sometimes 0.7 fx. long, some distance below their
ends. These spines are confined to a median line following the outer, convex
side of the teeth. Usually they form short saw-like rows on the upper margin.
Sometimes they are very conspicuous (Plate 27, figs. 13, 14), sometimes so small
as to be hardly visible (Plate 27, figs. 16, 17). The apex of the tooth-verticil is
generally smooth and dome-Uke (Plate 27, figs. 1-5, 6a, 7, 8, 11, 13, 14, 16, 17).

124 FARREA OCCA SCUTELLA.

Sometimes a continuation of the shaft extends beyond it, forming an apical,
distally rounded, smooth protuberance, 6-7 m long and 4.5-6 n thick (Plate 27,
figs. 9, 10, 15).

The rare clavules with short teeth (Plate 27, fig. 12) are, apart from their
teeth, similar to but smaller than the large-toothed ones above described. Their
teeth are very short, hardly at all recurved, and the verticils formed by them
only 18 M in diameter. Whether these clavules are young forms of the large-
toothed ones, or a distinct Idnd of spicule, I cannot say.

Their spiculation assigns these sponges to Farrea. Their shape, however,
does not accord with F. E. Schulze's diagnosis ^ of the Euretidae to which Farrea
belongs, for in this diagnosis it is stated that these sponges are tubular. E.
Topsent ^, who has studied a sponge very similar to the one described above, says,
concerning this part of Schulze's diagnosis, "II ne faut evidemment pas prendre
ce caractere trop a la lettre" and places these sponges of his, in spite of their
non-tubular shape, in Farrea. I also am disinclined to attach any great system-
atic importance to that difference of shape and therefore also place the sponges
above described in Farrea.

Of all the known species Farrea occa Bowerbank is obviously most closely
related to them. A great many specimens, by no means identical in structure
and appearance, have been assigned by various authors to this species, and for
some of them distinct varieties and subspecies have been established by Topsent
and Wilson. Although it seems to me very doubtful whether all the sponges
assigned to Farrea occa are really specifically identical and belong to this spe-
cies, and although I think that the forms described as varieties and subspecies
of it might very well be considered as distinct species, I provisionally accept
this arrangement, because it would lead much too far to reinvestigate all these
sponges, and if we accept this arrangement, we must assign to this species so
wide a range of variation that the sponges described above find a place in it.
Among the sponges described as Farrea occa, those for which Topsent '* estab-
lished the variety F. o. var. foliascens are obviously most closely allied to F. o.
var. scuteUa. From these they differ by the abundance of clavules, the scarcity
and size (or absence, vide supra) of the uncinates, and the larger dimensions of
the superficial pentactines. Although these differences are not very great,
they are, in my opinion, quite sufficient for varietal distinction particularly

' F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 177.
^ E. Topsent. Farrea occa (Bowerbank) var. foliascens n. var. Bull. Mus. oci'anogr. Monaco,
1906, no. 83, p. 4.

= E. Topsent. hoc. cit., 1906, p. 1.

FARREA. 125

when held together with the fact that the specimens of F. o. foliascens were
trawled in the tropical Atlantic, whilst the sponges described above come from
the eastern Pacific.

Farrea sp.?
Plate 32, figs. 1-3.

There are in the collection one large and three small fragments of skeleton-
nets of this sponge, all trawled off the southern coast of western Panama, at
Station 4631, 3 November, 1904; 6° 26' N., 81° 49' W.; depth 1415 m. (774 f.) ;
they grew on green sand; the bottom-temperature was 38.0°.

The large fragment (Plate 32, fig. 1) is 36 mm. long and appears as a part of
the skeleton-net of a tube nearly circular in transverse section and about 10 mm.
wide. Very short branch-tubes about 6 mm. wide arise from this tube, which
can be considered as a main-tube. Attached to both sides of this skeleton-net
are portions of network which form short tubular covered ways about 3 mm.
high and broad.

The skeleton-net (Plate 32, figs. 2, 3) of the main-tube and its branches
forms a single layer and chiefly consists of smooth, longitudinal, and transverse
beams, mostly 80-140 ^ thick. Here and there a short oblique beam of similar
thickness is observed. The meshes are mostly square, rectangular, 350-600 n
long, and 180-240 ix broad. A few are triangular. From each node of this net-
work two thorns arise, one directed towards the inner gastral surface, the other
towards the outer dermal surface. These thorns are conic, vertical to the sur-
face, about 50 M thick at the base, and covered with very blunt spines. The
gastral ones attain a considerable length. The skeleton-net composing the
walls and roofs of the covered ways above mentioned is irregular and has
mostly triangular meshes.

A large number of hexactines 80-140 m in diameter are attached, with one
ray, to the beams of these networks. In places, other similar hexactines are
soldered to these, forming here and there a fine net.

The sponges to which these skeleton-nets belonged can be assigned with a
considerable degree of certainty to Farrea.

Farrea sp.?

There are in the collection four slightly curved, small fragments, the largest
19 mm. long, of simple skeleton-nets extending in two directions (one surface)
only. These skeleton-nets were trawled in the southeastern Pacific, at Station

126 EURETE ERECTUM.

4685, on the 10 December, 1904. 21° 36.2' S., 94° 56' W. ; depth 4033 m. (2205 f .) ;
they grew on dark brown clay; the bottom-temperature was 35.3°.

These skeleton-nets are very regular and composed of smooth longitudinal
and transverse beams, 40-60 /i thick, which enclose square rectangular meshes
about 750 n long and 200-350 m broad.

The sponges to which these skeleton-nets belonged can be assigned with a
considerable degree of certainty to Farrea.

EURETE Semper.

Euretidae composed of anastomosing tubes without central calyculate
structure. With scopules, without clavules.

The collection contains three specimens of this genus which belong to three
species, one of which is new.

Eurete erectum F. E. Schulze.
Plate 30, figs. 1-17; Plate 31, figs. 1-28.

Eurete erectum F. E. Schulze, Amerikanische Hexactinelliden, 1899, p. 72, taf . 17, figs. 1-3.

Etirete erectum subsp. lubuliferum H. V. Wilson, Mem. M. C. Z., 1904, 30, p. 63, pi. 7, figs. 9, 12; pi. 8,figs.

1-3, 6.
Eurete erectum subsp. gracile H. V. Wilson, Mem. M. C. Z., 1904, 30, p. 69, pi. 8, figs. 4, 5, 8, 9; pi.

9, figs. 1, 3, 5.

Two specimens of this species, a fairly complete larger and a fragmentary
smaller one, were trawled off the southern coast of western Panama, at Station
4622 on 21 October, 1904; 6° 31' N., 81° 44' W.; depth 1067 m. (581 f.); they
grew on green sand and rock.

Shape and size. The larger specimen (Plate 30, fig. 16) is a tube with quite
regular circular transverse section. This tube is .slightly spirally twisted, 67 mm.
long, and throughout about 14 mm. in (outside) diameter. Its wall is 1-1.5 mm.
thick and perforated by seven apertures. These are circular, arranged in a
regular spiral, about 10 mm. wide, and surrounded by slightly protruding rims.
The rims are in some places 5 mm. high and above strongly curved outward.
They appear as rudiments of wide calyculate branches of the main-tube. The
smaller specimen is a fragment of a similar but wider tube. It is 30 mm. long
and the main-tube, of which it formed a part, must have been about 17 mm. in
diameter.

A thin, membranous alcyonarian colony, the outer surface of which extends
in the level of the tips of the distal pinule-rays, covers large tracts of the outer
dermal surface of the sponge.

EURETE ERECTUM. 127

The colour in spirit is light yellowish brown. When the tube-wall is ob-
served by transmitted light, numerous small dark brown spots, about 1 mm.
apart, make their appearance in it. These appear to be accumulations of deep-
sea ooze in the bottoms of wide, vertical, sacular canals which lead from the
outer surface into the deeper parts of the tube-wall.

Canal-system. The flagellate chambers (Plate 30, figs. 7c, 10c, 17c) are
spherical or short oval, and measure GO-80 m in diameter.

Skeleton. A special dermal and a special gastral skeleton are developed
besides the internal. The internal skeleton consists of a supporting network
and loose spicules; the dermal and gastral skeletons are exclusively composed
of loose spicules.

The supporting skeleton-net (Plate 30, figs. 4-6, 10-12, 17; Plate 31, fig. 24)
appears as a lamella corresponding in shape to the tube-wall, but thinner than
this. It is composed of smooth beams, 30-105 fi thick. In its outer part
(Plate 30, fig. 4) the meshes are irregular, mostly triangular, the larger ones
generally a little under 200 m wide. Its inner part (Plate 30, figs. 6, 11) is more
regular, composed chiefly of longitudinal and transverse beams enclosing square,
rectangular meshes, mostly 370-400 m long, and 170-400 m broad. Here and
there small hexactines, 80-120 ^ in diameter, are attached vertically to the beams
of the net by one of their rays. From both faces of the lamella formed by the
skeleton-net large thorns protrude. These thorns arise from the superficial
nodes of the net, point outwards, and are nearly vertical or, more rarely, oblique
to the surface. They are straight or slightly curved, and quite regularly conic,
pointed or, rarely, inflated at the end, and covered with broad and low, terminally
rounded spines, which decrease in size distally. The thorns on the outer, dermal
side (Plate 30, figs. 7e, lOe) are mostly 140-340 ^ long, and 20-50 ix thick at the
base. The thorns on the inner, gastral side (Plate 31, fig. 24g) are larger, 230-
430 II long, usually 270-400 fi, and 35-60 /x thick at the base.

The loose spicules of the choanosome are uncinates and discohexasters. Tlie
former are fairly abundant, the latter rather scarce.

The dennal skeleton is composed of hexactine pinules and small scopules.
The dermal pinules are very numerous and form a continuous layer on the outer
surface. Their lateral rays (Plate 30, figs. 7a, 10a, 12a, 17a; Plate 31, fig. 22)
extend paratangentially and together form a network, usually with more or less
quadratic meshes (Plate 31, fig. 22). Their proximal and distal apical rays are
situated radially (Plate 30, figs. 7d, lOd, 12d, 17d). Their centres are on an
average 130 n apart. The dermal scopules are situated radially. Most of them

128 EURETE ERECTUM.

lie below the pinule-layer and their end-ray bunches do not, as a rule, protrude
beyond the surface. They are not numerous.

The gastral skeleton consists of hexactine pinules, regular and irregular,
derivates of such with reduced distal apical (pinule) ray, and scopules. The
pinules and pinule-derivates are very numerous and irregularly intermingled.
They form, like the corresponding dermal spicules, a continuous superficial layer.
Their lateral rays (Plate 30, figs. 12b, 15b, 17b; Plate 31, fig. 24b) extend para-
tangentially, their apical proximal and distal rays (Plate 30, figs. 12h, 17h;
Plate 31, fig. 24h) radially. The gastral scopules are situated radially. Most
of them protrude a considerable distance beyond the zone of the lateral pinule-
rays, and the end-ray bunches of many lie at a considerably higher level than the
tips of the distal pinule-rays. The gastral scopules are much more numerous
than the dermal.

The dermal pinules (Plate 30, figs. 7a, d, 10a, d, 12a, d, 17a, d; Plate 31,
figs. 2-5, 22) have a straight distal ray, 85-145 m long, usually 105-140 fi, and
at the base 8-18 ix thick, usually 10-12 ix. This ray is thickened above in a
club-shaped manner and rounded distally. Its proximal part is smooth, its
(thickened) middle- and end-parts covered with large spines. The maximal
thickness of the distal ray (together with the spines) is 30-50 n, usually 40-48 n.
The proximal spines are 5-8 ^ long, and nearly vertical to the ray, directed only
slightly upwards. Distally the spines increase in size and become more and
more inclined towards the tip of the ray; those arising from its summit are
parallel to its axis. Half way up the spines attain the largest size. Here they
are 8-13 n long. The proximal ray is usually straight. In its basal and middle-
parts it is attenuated only slightly, at the end abruptly, towards the pointed end,
like a Roman sword. It bears small spines near the end. The other parts of it
are smooth. The proximal ray is 78-222 ^i long, usually 110-200 n, and at the
base 6-13.5 n thick, usually 7-11 /x. The lateral rays enclose angles of 90° with
each other and are, in the same spicule, fairly equal. They are similar to the
proximal ray in shape and spiculation, 108-152 ix long, usually 110-142 n, and
at the base 6.5-15 n thick, usually 8-10 n.

The gastral pinules (Plate 30, figs. 1, 2, 9, 12h; Plate 31, fig. 24h) have a
straight distal ray, 70-130 n long, and 11-17 ju thick at the base. Its proximal
part is smooth, its middle- and end-parts covered with short spines 10-17 n long.
The number of these spines is variable and never great. Sometimes there are
only a few. The spines point obliquely upward and are rather irregularly
distributed. The fewer there are, the more marked does this irregularity of

EURETE ERECTUM. 129

their arrangement become. The proximal ray is straight, or slightly curved,
and attenuated, proximally and medially very gradually, distally very abruptly
towards the pointed end. It is 160-235 m long, rarely as much as 290 n, and
8-13 M thick at the base. Its basal and middle-parts are smooth. Near the end
it bears small spines. The lateral rays are in the same spicule usually fairly
equal. They are straight or very sUghtly curved and generally not extended
in a plane, but just perceptibly bent downward towards the proximal ray. The
angles enclosed between their chords and the proximal ray are consequently
somewhat smaller than 90°, those between them and the distal ray somewhat
larger. Apart from this they are regularly arranged, their projections on a
plane vertical to the axis of the apical rays enclosing angles of 90° with each other.
The lateral rays are 187-240 m long, 10-15 m thick at the base, and slightly at-
tenuated to the rounded end. Their middle- and end-parts bear small and
pointed (Plate 30, figs. 1, 9) or large and blunt spines (Plate 30, fig. 2). The
number of these spines is never great and on the whole in inverse proportion to
their size.

The gastral pinule-derivates are connected with the gastral pinules above
described by transitional forms, but these are remarkably rare. Most of them
are fairly regular pentactines with an apical knob, the reduced distal ray; some
are irregular.

The regular pentactine-like gastral pinule-derivates (Plate 30, figs. 8, 13, 14).
The proximal ray is usually straight, 250-320 n long, and 13-16 n thick at the
base. In regard to shape and spinulation it resembles the proximal ray of the
gastral pinules above described. The reduced distal ray is a rounded apical
protuberance, usually 7-12 m high, 14-18 m broad, and beset with a few large
spines. The lateral rays of the same spicule may be fairly equal or very unequal.
In extreme cases the largest are 30% longer than the smallest. The lateral rays
are shghtly inchned toward the proximal ray and also a little curved in this
direction (concave to the proximal ray) ; sometimes they are curved also in a
transverse direction. The projections of their basal parts on a plane vertical to
the axis of the proximal ray, however, always enclose angles of 90° with each
other. The lateral rays are 200-328 m long, and 14-23 ^ thick at the base.
Distally they *aper gradually and they are, at the rounded end, 7-13 m thick.
They bear thick, usually quite blunt, vertically arising spines, 4-10 ^ long. In
the middle-part of the ray these spines are large and sparsely scattered; towards
the end they become smaller, particularly more slender, and more numerous, the
end itself often being quite crowded with spines. I had the impression some-

130 EURETE ERECTUM.

times that the spines were arranged in elongate spiral rows; in other cases no such
spiral arrangement could be made out. Often the spines are restricted to the
distal and lateral sides of the rays ; sometimes, however, they are also found on
the proximal side.

The irregular gastral pinule-derivates (Plate 30, fig. 3) are similar to the regu-
lar pentactine-like ones and differ from them only in one or two of their lateral
rays resembling the distal rays of pinules.

The uncinates (Plate 31, figs. 13, 14) are slightly curved or nearly straight,
pointed at both ends, 0.5-1.6 mm. long, and 4-9 yu thick. Their spines are 7-27;u
long, and 0.6-1 n thick at the base. They either diverge considerably (Plate 31,
fig. 14) or are nearly parallel to the shaft (Plate 31, fig. 13). Their tips are 1.5-
4 ju distant from the shaft. This elevation of their tips is by no means always in
proportion to their length.

The discohexasters (Plate 31, figs. 15, 18, 21) measure 50-70 m in total
diameter. Their main-rays are regular, smooth, straight, 6-10 ^ long, and 1.0-
3 IX thick. Each main-ray bears from one to four end-rays. These are usually
curved, concave to the continuation of the main-ray at the base, and nearly
straight farther on. They are 18-26 /j long, 1.2-2 n thick at the base, and
attenuated distally to 0.8-1.5 n. The end-rays bear along their length minute
recurved spines, and at the end a terminal verticil of similar but larger spines,
which together form a kind of terminal disc with deeply serrated margin 2.5-4 m
in transverse diameter.

It is possible that there are two kinds of discohexasters similar in size, but
differing in respect to the end-rays, one with more slender and less spiny, the
other with stouter and more spiny end-rays. Since, however, these asters are
scarce I was unable to decide whether they all belong to the same series of forms,
or whether two distinct varieties of them, as indicated above, should be distin-
guished.

The dermal scopules (Plate 31, figs. 16b, 17, 19) are 200^20 ^ long and
consist of a centrum 4-10 ^ long and 5.5-1 1 .5 ^ broad, from which arises at one end
(the inner) a simple shaft, and at the opposite (the outer) a bunch of end-rays.
The centrum is not well-defined, often it passes quite gradually into the
shaft. It and the proximal part of the shaft are densely covered with minute
spines. The shaft is straight, cylindroconic, 170-330 fx long, 3-6 yu thick at the
base, and pointed at the end. Sometimes, particularly in the dermal scopules
with only two end-rays, this spinulation extends quite to the end of the shaft.
Some of the dermal scopules have four end-rays, others only two, and a few have
three. The dermal scopules with only two end-rays are fork-Uke. The end-rays

EURETE ERECTUM. 131

are 20-76 ^ long and 1-3.5 /x thick at the base. They are usually attenuated
towards the end, more rarely of uniform thickness throughout. The end itself
is pointed, blunt, rounded or sUghtly thickened to a terminal "disc," which,
however, is always small, only rarely over 3 ^ in transverse diameter. The end-
rays are usually curved in an S-shaped manner, rather strongly concave to the
continuation of the axis of the shaft at the base, and \-ery slightly, in the opposite
direction, in their middle- and end-parts. These cm•^'atures, particularly the
basal, are subject to considerable variation. The breadth of the bunches formed
by the end-rays is 11-25 ix. The end-rays are uniformly covered by densely
crowded minute spines. The terminal "disc" is, when present, composed of
similar but slightly larger spines.

The gastral scopules (Plate 30, fig. 15i; Plate 31, figs. 1, 6-12, 16a, 20, 23, 24i,
25-28) are 0.6-1.18 mm. long, and consist of a centrum, from one (the inner) end
of which arises a sunple shaft, and from the opposite (outer) a bunch of end-rays.

The centrum is sometimes (Plate 31, fig. 27) rather clearly defined, some-
times it passes gradually into the shaft. It is 5-18 m long, 6.5-17 n broad, and
bears small backwardly directed spines, like those on the adjacent parts of the
shaft and the end-rays. An axial cross, composed of six axial threads regularly
arranged in the usual manner, can always be detected in the centrum. One of
these axial tlireads is long and continued in the axial thread of the shaft. The
one opposite this one is short, and terminates a considerable distance below the
distal end of the centrum, without giving off branches for the end-rays. The
other four axial threads are still shorter and equal among themselves. Some-
times four very sUght elevations arise from the sides of the centrum over them.

The shaft is 0.52-1.05 mm. long, straight or slightly curved, and 3-11 fx
thick at the base, where it arises from the centrum. In some gastral scopules
it tapers toward the end, in its basal and middle-part, very gradually, in its distal
part rapidly. In most, however, its middle-part is cylindrical or thickened
and is 1-3 n, sometimes 13.5 m thicker than the base in transverse diameter.
The shaft terminates in a sharp point and is traversed throughout by an axial
thread. At the base it is covered with a greater or smaller number of minute
recurved spines, similar to those on the basal parts of the end-rays and on the
centrmn. Farther on it bears a few minute, isolated, vertical spines or is quite
smooth. A little distance below the end larger vertical spines are observed.

Of end-rays there are usually four; in some gastral scopules, however, three,
five, or six have been observed. The end-rays are 75-133 m long, and 2-8 n
thick at the base. Generally the end-rays become thicker toward the distal end
(Plate 31, fig. 1) ; sometimes they are of uniform thickness throughout (Plate 31,

132

EURETE ERECTUM.

figs. 10-12) . Just below the distal end they measure 4-8 m in transverse diameter.
In these measurements the fact finds its expression that the basally thin end-rays
are distally thickened, whilst the basally stout ones are of uniform thickness
throughout. The end-rays are destitute of axial threads and usually rather
densely covered with minute recurved spines, which increase in size from the
base, where they are about 0.7 yu long (Plate 31, fig. 27), to the end, where they are
1.5-3 n long (Plate 31, figs. 6-9, 26). The end of the end-ray is thickened to a
tyle, 12-17 ju in transverse diameter. This is particularly conspicuous in the
end-rays which are thin at the base and thickened distally. The distal, apical
face of the tyle is dome-shaped and usually quite smooth (Plate 31, figs. 6-9).
Its sides are densely covered with spines, directed obliquely downwards. The
spines nearest its apex are small, farther down they rapidly increase in size, and
the lowest attain 2 ;u or more in length. The spines of the tyle are, like those on
the other parts of the end-ray, distinctly curved downwards. The end-rays are
curved in an S-shaped manner, strongly, concave to the continuation of the axis
of the shaft at the base, and slightly in the opposite direction in their distal
and middle-parts. This second (outward) curvature is sometimes so light that
the distal part of the end-ray appears straight. The degree of divergence of the
end-rays is variable. The bunch formed by them is 60-102 ^ broad.

As examples the measurements of three gastral scopules of various dimen-
sions are tabulated below.

Total length

Z'

760

920

1180

Breadth of the bunch of end-rays

M

64

SO

62

length

M

642

789

1050

Shaft

thickness

in the middle

f

8.5

9

12

at the base

M

6.5

8 •

9

Centrum

length

Z*

13

18

10

thickness

A»

12

15

13

length

M

105

113

120

thickness

at the base

M

3.5

5

5

End-rays

just Ijelow the
terminal tyle

I'

4.5

8

5

transverse diameter of the
terminal tyle

M

15

16

12

EURETE ERECTUM. 133

The statements given above show that the sponges here described are very-
similar to Eurete erectum F. E. Schulze.^ Wilson has established three subspecies
of this species: — iubuliferum,- mucronatum,^ and gracile.'^ One of these, E. e.
mucronatum, differs from the sponges above described, and also from Schulze's
type, and from the other two of Wilson's subspecies, by possessing oxyhexasters
instead of discohexasters. This difference is in my opinion of such systematic
importance that I consider it distinct from the other sponges placed in Eurete
erectum.

.•yter the exclusion of this subspecies, Schulze's Eurete erectum, Wilson's
E. e. tubuliferum, Wilson's E. e. gracile, and the sponges described above, remain
as forms of one species. A comparison of these shows, that, although similar
in the main, they differ from each other in several minor points. The tubular
body of the sponge is in Schulze's type dichotomously branched, in the three
others simple. This tube is in Wilson's E. e. tubuliferum and in my specimens
14-17 mm. wide, in Schulze's type and in Wilson's E. e. gracile 8-12 mm. The
distal rays of the dermal pinules are in Wilson's E. e. gracile 50 n thick, in Schulze's
type and in Wilson's E. e. tubuliferum only 35-40 n. In the specimens examined
by me, dermal pinules occur together with distal rays as stout as those of E. e.
gracile and as slender as those of the other two. In my specimens the lateral
and proximal rays of the gastral pentactine-like pinule-derivates are considerably
larger than the corresponding rays of the gastral pinules proper. In the other
three no such difTerence occurs, their gastral pinules and pinule-derivates being
about as large as the gastral pinules of my specimens. The greatest differences
between these sponges are met with in their scopules. To facilitate a comparison
between the scopules of these sponges, short descriptions of them are tabulated
on p. 134.

In respect to their other characters, particularly the shape and size of the
uncinates and discohexasters, the four groups of forms appear to agree quite
closely. Schulze's type was collected at Albatross Station 2819, near the Gala-
pagos Islands, depth 717 m.; Wilson's E. e. tubuliferum at the Albatross
Stations 3358 and 3359, off the south coast of western Panama, depth 875 and
1015 m.; Wilson's E. e. gracile at Albatross Station 3380, Gulf of Panama,
depth 1693 m. ; and the specimens examined by me at .Albatross Station 4622,
off the south coast of western Panama, depth 1067 m. The differences between

• F. E. Schuke. Amerikanische Hexactinelliden, 1899, p. 72, t.af. 17, figs. 1-3.

2 H. V. Wilson. Mem. M. C. Z., 1904, 30, p. 63, pi. 7, figs. 9, 12; pi. 8, figs. 1-3, 6.

3 H. V. Wilson. Lac. cit., p. 68, pi. 8, fig. 7.

< H. V. Wilson. Lac. cit., p. 69, pi. 8, figs, i, 5, 8, 9, pi. 9, figs. 1, 3, 5.

134

EURETE ERECTUM.

SCOPULES OF THE DIFFERENT FORMS OF EURETE ERECTUM.

A
In Schulze's type

B

In E. e. tubuliferum
Wilson.

c

In E. e gracile
Wilson.

D

In the specimens ex-
amined by me.

Dermal

scopules

Total length 200-
600 m; 4-6 end-rays
with pointed, re-
curved spines and
terminal tyle with
larger spines on lower
side.

Two kinds. In one
3-4 cylindrical end-
rays, 40 by 2 M, with
minute sharp dentic-
ulations and small
smooth terminal tyle;
centrum distinct;
shaft 200-240 by 4 m-
In the other 4-10
end-rays, 60-100 by
2-3 Mi with minute
sharp denticulations
and a terminal tyle
5-8 M in diameter,
sometimes with re-
curved spines; shaft
a little longer than
in the other form, 6 m
thick.

Two kinds. In one
4 end-rays, 50-70 by
4-6 M, tapering dis-
tally, with minute
denticulations bas-
ally, smooth distally,
without tyle or a
very small terminal
tyle; shaft 300 by
6-8 M- In the other
4-6 cylindrical end-
rays, 70-100 by 3-5 m,
slightly roughened,
with terminal tyle
6-12 M in diameter,
and spines. Total
length 600-700 m-

Total length 200-
420 M- 2 or 4, rarely
3 end-rays. 20-76 by
1-3.5 M at the base,
attenuated distally
or cylindrical, the
end pointed, rounded,
or slightly thickened,
densely covered with
minute spines. Shaft
3-6 M thick.

Gastral
scopules

Similar to the dermal
but larger on the
whole and with more
divergent end-rays,
these more frequently
angularly bent.

4-6 end-rays 70-80 m
long, either smooth,
2 M thick at base, and
thickened to 4 m dis-
tally, with terminal
tyle 12 M in diameter
with recurved spines
on the lower side; or
cyUndrical, with mi-
nute denticulations,
with terminal tyle
8 M in diameter; or
transitions between
these; shaft 300 by
5 m.

.3-6 end-rays, 100-
120 M long, either (cy-
lindrical, 4-5 M thick,
with terminal tyle,
12 M in diameter with
recurved spines; or
12 M thick at base
and tapering distally,
without terminal tyle;
or transitions be-
tween these. Total
length 0.6-1.5 mm.;
shaft 8-16 M thick.

Total length 0.6-
1.18 M- 3-6, usually
4 end-rays, 75-133
by 2-S M at base,
thickened distally or,
rarely, cylinflrical.
Densely covered with
minute recurved
spines, with semi-
spherical terminal
tyle 12-17 m in diam-
eter, with large re-
curved spines below.
Shaft at base 4-11 m
thick.

the specimens examined by Schulze and Wilson and those described in this
paper indicate that the former differ from the latter quite as much as the latter
differ among themselves. This is particularly noticeable in that the former
possess dermal scopules with only two end-rays, which are absent in the latter,
and that the gastral pentactine-like pinule-derivatives of the former are much
larger than the corresponding spinules of the latter. The general agreement of
all these sponges, the localities from which they were obtained, and particularly
the fact that the differences between them appear to be virtually confined to the
superficial spicules, which are of course most liable to be influenced by the
environment, make it very doubtful, however, whether they should be considered

EURETE SPINOSUM. 135

as distinct subspecies. To me it seems that a subdivision of the species into four
local forms (A, B, C, and D) adapted to different surroundings, but congenitally
hardly at all different, would more correctly express the relation between them.
Their distinctive features are the following : —

Eurete erectum A. {Eurele ereclum F. E. Schulze, 1899).

Main-tube dichotomous. One kind of scopule with 4-6 end-rays with
terminal tyle. Total length of scopules 400-600 m-

Eurete erectum B. (Eurele ereclum subsp. tubuliferum Wilson, 1904).

Main-tube simple. Several kinds of scopules with 3-10 end-rays, all with
terminal tyle. Total maximum length of scopules 400 fx. Distal ray of dermal
pinules under 40 n thick.

Eurete erectum C. (Eurete erectum subsp. gracile Wilson, 1904).

Main-tube simple. Several kinds of scopules with 3-6 end-rays. These
in some with terminal tyle, in others distally attenuated and without tyle.
Total length of longest gastral scopules 1.5 mm. Distal ray of dermal pinules
50 M thick.

Eurete erectum D.

Main-tube simple. Several kinds of scopules. Some small dermal ones
with only 2 end-rays without terminal tyle. The others with 3-6 end-rays.
These either distally thickened and with terminal tyle or, more rarely, cylindrical
or attenuated distally, without terminal tyle. Total length of largest gastral
scopules 1.18 mm.

Eurete spinosum, sp. nov.
Plate 29, figs. 1-26.

One fragmentary specimen of this species was trawled off northern Peru,
west southwest of Aguja Point, at Station 4656 on 13 November, 1904; 6° 54.6'
S., 83° 34.3' W.; depth 4062 m. (2222 f.) ; it grew on fine, green mud mixed with
gray ooze; the bottom-temperature was 35.2°.

The lateral rays of its superficial pentactines bear exceedingly large spines.
To this the name refers.

Shape and size. The single specimen is a lamellar fragment 25 mm. long,
20 mm. broad, and 2 nun. tliick. It is curved in one direction, the radius of

136 EURETE SPINOSUM.

curvature being about 20 mm., and may originally have formed part of a cylindri-
cal tube about 40 mm. in diameter.

The colour in spirit is dirty brown.

The skeleton consists of a continuous net, which pervades the whole lamella,
and of loose pentactines, hemioxyhexasters, and scopules. Long, slender
rhabds have also been observed, but it is doubtful whether they belong to the
sponge. The pentactines form a continuous layer on the intact parts of the
surface. Their lateral rays extend paratangentially, their apical ray points
inward. The hemioxyhexasters are exceedingly numerous and appear, so far
as can be judged by the fragmentary specimen, to occur in dense masses in all
parts of the choanosome. The large (perhaps foreign) rhabds lie more or less
parallel to the surface.

The skeleton-net (Plate 29, figs. 18, 19, 23-25) is by no means uniform in
structure throughout the thickness of the lamella. In the dermal zone (Plate 29,
figs. 18, 23) it is rather irregular, composed of beams 8-40 fi thick, usually 20-
35 n, and here its meshes are triangular or irregularly square, not rectangular,
and 0.2-0.4 mm. wide. In the gastral zone (Plate 29, figs. 19, 25) on the other
hand the network is very regular, composed of longitudinal and transverse
beams. The former are 18-50 ^ thick, and on an average about 0.23 mm. apart;
the latter are 8-36 m thick, and individually usually extend obliquely but col-
lectively from zones which are 0.8-1 mm. apart, and extend transversely, vertical
to the longitudinal beams, quite across the whole specimen. With the exception
of a few, usually thin ones, which are quite smooth, the beams of the skeleton-net
are covered with conic spines, 2.5-8 m high, mostly 5-6 m- The spines of the thin
and thick beams are nearly equal in height l)ut differ, often very considerably,
in breadth, those on the thicker beams being usually much stouter than those on
the thinner beams. Freely terminating rays of the hexactines, by whose concres-
cence the network seems chiefly to be formed, arise from the beams in many
places. These spine-Uke protuberances are thinner than the beams of the net-
work and are only 4 n thick. Here and there local thickenings are observed in
the beams. Cylindroconic, terminally rounded spines attaining 25 n in length
and 9 Ai in thickness arise from these thickenings. These spines are parallel to
the surface of the sponge and the thickenings from which they arise also chiefly
extend in this direction. The thickenings with their spines have a cockscomb-
like appearance (Plate 29, fig. 24). The comparison of a number of these struc-
tures has convinced me that they are in truth hemioxyhexasters which have been
soldered to the growing skeleton-net and the rays of which have been secondarily

EURETE SPINOSUM. 137

thickened by the apposition of silica-layer, together with the beams of the skele-
ton-net which had, as it were, incorporated them.

The rhabds, which may, as above mentioned, be foreign, are long, smooth
centrotyles. They are about 15 yu thick near the centre. The tyle measures
about 17 ju in transverse diameter.

The superficial ■pentactines (Plate 29, figs. 20-22) usually have fairly equal,
straight, conic, terminally rounded lateral rays, which enclose angles of 90° with
their neighbours (Plate 29, fig. 20). Rarely (Plate 29, fig. 21) the lateral rays
are cyUndroconic, curved, and irregularly arranged. The lateral rays are 150-
270 M long, 12-22 /i thick at the base, and covered throughout with vertically
arising spines. The spines on the proximal part of these rays are 8-12 ii long;
distally they become smaller. The apical (proximal) ray is smaller than the
lateral rays and destitute of large spines.

The hemioxyhexasters (Plate 29, figs. 9-17, 26b) measure 80-122 n in total
diameter, usually 90-110 m. Two of their rays, which extend in the same axis
and lie opposite each other, are usually conic, short and simple, and only excep-
tionally bear an end-ray. These two rays I designate the apical. The four
other rays, which lie in a plane vertical to the axis of the other two, nearly always
bear end-rays. These four rays I designate the lateral. The simple stems
(main-rays) of these lateral rays always enclose angles of 90° with their neigh-
bours and are, in the same spicule, usually fairly equal (Plate 29, fig. 10) ; only
exceptionally they differ in length (Plate 29, fig. 12). In the ordinary regular
hemioxyhexasters the lateral main-rays are 16-24 fi long; in the rare irregular
forms the shortest is sometimes only 14 fx long, or still shorter. The lateral
main-rays are cyUndroconic, at the base 3.5-6.5 m thick, usually about 4 fi, and
uniformly attenuated towards the end, the transverse diameter of which is about
three quarters of that of the base. The thickness of these main-rays is not in
proportion to their length, the thickest not being longer, often indeed shorter
than the thinner ones. The lateral main-rays bear minute spines which decrease
in size proximally. In the thinner ones these spines can only be made out in the
distal part, and also here only in the u. v. photographs (Plate 29, fig. 9). The
thick ones are covered with clearly visible spines throughout (Plate 29, fig. 12).

Each lateral main-ray bears three regularly disposed end-rays which lie in
the plane of the four lateral main-rays. One of them extends in the same direc-
tion as the main-ray to which it belongs, and appears as a continuation of the
latter. The other two lie symmetrically on the two sides of this central one.
These regularly disposed end-rays are, in the same spicule, usually equal (Plate

138 EURETE SPINOSUM.

29, fig. 10). In the few hemioxyhexasters, however, in which the main-rays are
unequal, a corresponding irregularity is also observed in the end-rays (Plate 29,
fig. 12). The regularly disposed end-rays are conic, sharp-pointed, and covered
with minute, backwardly directed spines (Plate 29, fig. 14). In a few hemioxy-
hexasters with exceptionally thick lateral main-rays the regularly disposed end-
rays are reduced in length and terminally rounded. In the normal, regular
hemioxyhexasters the end-rays are 25-44 ii long, and 1.8-3.7 ^i thick at the base.
In the irregular forms the shortest are sometimes only 13 ix long. The central
end-ray of each group of three is straight throughout; the two lateral ones are
either also straight throughout (Plate 29, fig. 12) or, much more frequently,
straight only in their middle- and end-parts, but curved at the base, concave to
the central end-ray. The chords of the lateral end-rays enclose angles of 47-52°
with the central end-ray.

Besides these regularly disposed end-rays other end- or branch-rays some-
times arise from the lateral main-rays. Occasionally one or two supernumerary
end-rays are added to the regularly disposed three. These additional end-rays
extend, like the latter, in the plane of the lateral main-rays. More frequently a
branch-ray is seen arising some distance below the end of the lateral main-rays.
These branch-rays extend more or less vertically to the plane of the lateral
main-rays, and are parallel to the apical rays. In size and spinulation the
supernumerary end- and branch-rays are similar to the regularly disposed end-
rays; they are, however, more frequently irregularly curved.

The axis of each lateral main-ray is occupied by an axial thread (Plate 29,
figs. 14-17, 19). This terminates at the end of the main-ray and does not send
branches into the end-rays. The latter are destitute of axial threads.

The scopules (Plate 29, figs. 1-8, 26a) are 140-288 m long. They consist of a
centrum, 3.8-7 m in diameter, usually 4-5 m, from one side of which arises a shaft,
and from the other arise four or, rarely, five or si>{ end-rays. The shaft is conic,
straight, or slightly curved, 115-261 n long, and 1.9-2.4 ^ thick at the base.
Near the distal end and often also near the base it bears minute spines. The end-
rays diverge distally and together form a brush-like verticil 9-18 /i broad at the
end. The individual end-rays are very slightly cur\'ed, concave to the continua-
tion of the shaft at the base, and nearly straight in their remaining part. They
are 20-44 fi long, 0.9-1.5 fi thick at the base, very slightly attenuated towards the
end, and densely covered with minute, backwardly directed spines. At the end
they bear a verticil of larger, recurved spines, which together form a kind of
terminal disc with strongly serrated margin (Plate 29, figs. 7, 8).

EURETID. 139

Although the specimen at my disposal is but a small fragment there can be
little doubt that it belongs to the group of sponges represented by Eurete bower-
bankii F. E. Schulze and Eurete marshalli F. E. Schulze'. Since, however, it
differs from these species by its superficial pentactines, which are much more
spiny than in either E. boiverbankii or E. marshalli, and since its hemioxyhexasters
have relatively longer end-rays than those of E. bowerbankii - and relatively
shorter end-rays than those of E. marshalli,^ it cannot be assigned to either of
them and must be considered as a new species.

EuRETID FROM STATION 4641.
Plate 106, figs. 1-3.

The supporting skeleton-nets of three euretids, one large fairly intact, and
two small fragmentary ones, were trawled near Chatham Island, Galapagos, at
Station 4641 on 7 November, 1904; 1° 34.4' S., 89° 30.2' W.; depth 115 m.
(633 f.); they grew on a light gray Globigerina ooze; the bottom-temperature
was 39.5°.

The larger supporting skeleton-net (Plate 106, fig. 3) is 47 imn. long and
consists of a tube, 7 mm. wide, with nearly circular transverse section, which
rises vertically from the base of attachment. This tube is straight for the greater
part of its length, but bent abruptly to one side a httle below its free upper end.
Eighteen tubular branches, with a maximum length of 7 mm., and about as wide
as the main-tube, arise from this tube. These branch-tubes are arranged in a
spiral line. Some of them are distinctly widened distally, funnel-shaped. The
basal part of the main- tube, and the lowest branch- tubes have walls about 2 mm.
thick. Distally the walls become thinner, the uppermost being about 1 mm.
thick. In the smaller specimens the main-tube is shorter and a little wider.

The beams composing these skeleton-nets are, in the middle-part of the
length of the main-tube of the largest specimen, mostly 40-80 ^ thick. The
meshes of the network are, in the inner, gastral parts of the tube-waUs, 100-300 n
wide and square with strongly rounded corners (Plate 106, fig. 1). In the outer,
dermal parts of this portion of the skeleton-net the meshes are mostly 80-350 n
wide and more frequently triangular with rounded corners (Plate 106, fig. 2).
The axes of the rays of the spicules, through the concrescence of which these

' F. E. Schulze. Rept. Voy. ChaUenger, 1887, 21, p. 297.
2 F. E. Schulze. hoc. cil., pi. 79, fig. 13.
' F. E. Schulze. Loc. cit., pi. 79, fig. 3.

140 EURETID.

skeleton-nets hsive been produced, are very distinct. In many places small
hexactines, attached by the tip on one of their rays, arise vertically from the
beams of the skeleton-net.

I think there can be no doubt that these skeleton-nets belong to a euretid
sponge, but since no loose spicules were found in them, I am unable to say to
which genus they should be assigned.

Euretid (?) from Station 4651.

Plate 32, figs. 4-6.

There are in the collection a fairly complete skeleton-net and three lamellar
fragments of this sponge, all trawled off the coast of northern Peru, at Station
4651 on 11 November, 1904; 5° 41.7' S., 82° 59.7' W.; depth 40G4 m. (2222 f.);
they grew on sticky, fine, gray sand; the bottom-temperature was 35.4°.

The fairly complete skeleton-net (Plate 32, fig. 4) consists of a dense basal
mass with digitate processes, some of which are 10 mm. long and 6 mm. thick,
from which arises a broad and low calyculate, funnel-shaped lamella. The margi-
nal parts of the funnel are, for the most part, broken off. What remains of it is
65 mm. in maximum transverse diameter. Proximally, where it arises from the
basal mass, the lamella forming the funnel is about 1.5 mm. thick. Towards
the mai'gin it thins out to 1 mm.

The skeleton-net of the basal mass is very dense and irregular. Its beams
are mostly 15-100 ^ thick, and its meshes 15-220 fi wide. The small meshes
are round, the large ones triangular or irregularly square. The outer (dermal)
zone of the skeleton-net of the funnel (Plate 32, fig. 6) is irregular, composed of
beams 20-180 ^ thick, which enclose mostly triangular meshes up to 700 fx. wide.
The inner, gastral zone (Plate 32, fig. 5) is more regular, but does not attain such
a degree of regularity as is often observed in the corresponding zone of the skele-
ton-nets of the Euretidae. It is chiefly composed of smooth longitudinal and
transverse beams, but a fair number of usually spined, oblique beams also occur
in it. The longitudinal beams are 50-100 ^ thick, the transverse beams are
sometimes 160 n thick. The oblique beams are much thinner, usually only
15-30 M thick. The meshes are square or, less frequently, triangular. The
square ones are usually somewhat irregular, not rectangular, 600-900 fi long,
and 190-550 m broad.

These skeleton-nets, which are similar to the ones from Station 4695, proba-
bly belonged to a euretid.

EURETID. 141

EURETID (?) FROM STATION 4685.

There are in the collection three small, flat, lamellar fragments about 1 mm.
thick, the largest of which is IG mm. long, trawled in the southeastern Pacific at
Station 46S5 on 10 December, 1904; 21° 36.2' S., 94° 56' W.; depth 4033 m.
(2205 f.); they grew on dark brown clay; the bottom-temperature was 35.3°.

These lamellae are skeleton-nets composed on one face of longitudinal and
transverse beams, mostly 40-50 ix thick, which enclose square, rectangular
meshes, generally 300-500 m long and 200-250 yu broad; on the other face of
considerably thinner beams, which enclose smaller, irregularly triangular meshes.
The beams are mostly spined. The spinulation is more developed in the irregu-
lar than in the regular part of the network. Numerous hexactines, 100-150 p.
or more in diameter, are attached by one ray to the beams of this network.

These skeleton-nets probably belonged to a euretid.

EURETID (?) FROM STATION 4695.

There are in the collection four fragments of skeleton-nets of this sponge
trawled northeast of Eastern Island, at Station 4695 on 23 December, 1904;
25° 22.4' S., 107° 45' W.; depth 3694 m. (2020 f.); they grew on fine, light
brown ooze.

The largest and least incomplete is 32 mm. high, and appears as a tubular
stalk, extending above to a thin-walled funnel 22 mm. in diameter. The stalk
is about 10 mm. long, and in the middle, where it is somewhat attenuated, of
oval, transverse section, 6.5 mm. broad and 4.5 mm. thick.

The skeleton-net of the stalk is irregular, composed of longitudinal and oblique
spined beams, the former about 90 ^ thick, the latter 15-50 m. In places the
stout longitudinal beams of this part of the net bear numerous, vertically arising
thorns, 6-10 m thick at the base, and of varying length. The meshes of this
network are irregular, generally 50-200 ^ wide. The skeleton-net of the funnel
is more regular, chiefly composed of longitudinal and transverse beams. Oblique
beams, however, also occur in it, particularly in its outer zone. The beams of
this network are smooth and 50-130 m thick, the meshes in the inner zone
square, rectangular, in the outer zone more frequently triangular. The rec-
tangular meshes of the inner zone are mostly about 600 m long and 300-400 n
broad. Verticil thorns, directed towards the funnel-cavity, arise from the nodes
of the inner part of this network.

142 CHONELASMA.

These skeleton-nets may have belonged to a euretid sponge. They are
similar to those described above from Station 4651.

COSCINOPORIDAE Zittel.

Lamellar, calyculate, or more complicated Hexasterophora consisting, if
lamellar, of a simple plate; if calyculate or more complicated, of a rather thin
wall enclosing a wide cavity. This plate or wall is traversed by straight, conical,
blindly ending, sac-shaped afferent and efferent canals. With a firm supporting
reticulate skeleton and uncinates and scopules.

The collection contains one specimen of this family, which belongs to a
species of Chonelasma.

CHONELASMA F. E. Schulze.

Funnel-shaped or lamellar Coscinoporidae.

Chonelasma sp.
Plate 32, figs. 7-9.

There is in the collection a rather large skeleton-net of this sponge, collected
in the Paumotu Islands at Station 3689 (A. A. 134) on 28 October, 1899; 18°
06' S., 142° 24' W.; depth 1476 m. (807 f.); they grew on a bottom of fine
coral-sand and manganese nodules; the bottom-temperature was 37.6°.

This skeleton-net (Plate 32, fig. 7) is a curved plate, 92 mm. long, 51 mm.
broad, and 9-11 mm. thick. The sponge to which it belonged may have been
tubular or calyculate; probably it was of large size. The convex, probably
outer (dermal) zone of the skeleton-net (Plate 32, fig. 8) is on the whole smooth.
It is composed of skeleton-net lamellae vertical to the surface, extending in-
discriminately in all directions and crossing each other irregularly. These
lamellae form a network, the meshes of which are represented by short vertical
canals round or polygonal in transverse section and 0.5-2 mm. wide. The
concave, probably inner (gastral) zone of the skeleton-net (Plate 32, figs. 7, 9)
has some outgrowths. Most of these are quite small. One is 8 mm. high.
Apart from a curved, obliquely transverse band 3-5 mm. broad, where the net-
work is so dense as to appear nearly solid to the naked eye, the zone of the
skeleton-net bordering on this inner concave, probably gastral surface is com-
posed of skeleton-net lamellae, vertical to the surface and extending longitudi-
nally. These lamellae are about 0.7 mm. apart and connected by numerous

HEXACTINELLA. 143

transverse beams, whieh, to a certain extent, also form skeleton-net lamellae.
These transverse lamellae are, however, not nearly so compact and so regularl}^
arranged as the longitudinal ones. Together the longitudinal and the transverse
skeleton-net lamellae form a network with meshes about 0.7 nnn. broad and
0.7-1.5 mm. long.

The skeleton-net lamellae of the outer zone, that is the one bordering on the
convex side (Plate 32, fig. 8), are composed of a network of beams mostly 450-
650 ij. thick, which enclose roundish irregular meshes, usually 1.5-2.5 mm. wide.
The beams of this network are covered with large, rounded protuberances. Its
meshes are either ciuite empty or contain only slight traces of a fine secondary
network, similar to that in the inner zone, described below. The skeleton-net
lamellae of the inner zone, that is those bordering on the concave side (Plate 32,
fig. 9), are composed of a primary network in the meshes of which a fine secondary
network is spread out. The primary network consists of smooth, longitudinal,
transverse, and oblique beams. The longitudinal beams are situated either
singly or in bundles of two or thi-ee. Those of the same bundle are connected
at frequent intervals by short transverse beams. Here and there they even
coalesce to form irregular stems sometimes 350 ix thick. The individual longi-
tudinal beams are usually about 130 ^ thick, the transverse and oblique 60-110 yn-
The meshes are very irregular and are sometimes more than 1 mm. long. Thorns
about 200 fi long, 40 /i thick at the base, and provided with low, rounded protu-
berances arise from some of the nodes of this network. The secondary network
extends in the meshes of the primary and in the transverse band above referred
to, and also occupies the interstices between the lamellae. It is composed of
])eams, 5-10 m thick, which enclose square, rectangular, or, more rarely, irregular
meshes 50-130 ^ wide.

TRETOCALYCIDAE F. E. Schulze.

Hexasterophora with ramified afferent and efferent canals. With a firm
reticulate supporting skeleton and uncinates and generally also scopules.

The collection contains one specimen and three fragments of this family,
which belong to Hexactinella.

HEXACTINELLA Carter.

Tretocalycidae which are calyculate or composed of simple, ramified, or
anastomosing tubes; with firm reticulate supporting skeleton, uncinates, scopules,

144 HEXACTINELLA MONTICULARIS.

and discohexasters or oxyhexasters or tylehexasters or two of these forms. With-
out microonychhexactines and tylostyles with slender branch-rays, bearing end-
discs, on the tyle.

The collection contains one specimen and three fragments of this genus.
The specimen is insufficiently preserved for specific distinction. The three
fragments all belong to a new species.

Hexactinella monticularis, sp. nov.
Plate 28, figs. 1-28.

Three fragments of the skeleton of this sponge were trawled south of Chat-
ham Island, Galapagos, at Station 4642 on 7 November, 1904; 1° 30.5' S., 89°
35' W. ; depth 549 m. (300 f.); they grew on broken Globigerina shells; the
bottom-temperature was 48.6°.

From the surface broad and truncate conic protuberances arise and to these
the name refers.

Shape and size. The three fragments measure 16, 17, and 20 mm. in maxi-
mum diameter respectively. All appear to be parts of an irregular massive
sponge with stout, truncate, conic protuberances. One of these protuberances,
which is about 4 mm. high and 8 mm. broad at the base, is represented (Plate
28, figs. 23, 28).

The colour in spirit is brown.

The skeleton consists of an internal and superficial network and loose hexac-
tines, pentactines, uncinates, discohexasters, and scopules.

The internal skeleton-net (Plate 28, figs. 23, 24, 26, 28) forms meandric
lamellae, mostly nearly 0.5 mm. thick, which appear as the walls of tubes, with
lumina more or less circular in transverse section and about 1 mm. wide. In the
interior of the sponge these tubes are variously curved and irregular in their
course. On ajjproaching the surface they straighten out. On the whole they
extend chiefly radially and longitudinally from the base to the upper and lateral
parts of the surface, where they open out. The openings are fairly equidistant
and uniformly distributed, as numerous on the summits and the sides of the
monticular processes as on the other parts of the surface. Since most of the
tubes reach the surface obliquely their superficial openings are more or less oval
(Plate 28, figs. 23, 28). It is to be presumed that the tubes form two systems,
one afferent, vestibular ("Epirhysen"); the other efferent, preoscular ("Apo-
rhysen").

HEXACTINELLA MONTICULARIS. 145

The lamellae separating these tubes consist of a network of beams, mostly
40-100 ti thick, with meshes 100-200 m wide. Some parts of this network are
quite irregular, others more regular, with more or less quadratic meshes. The
beams generally bear small, broad, sharp-pointed, conic spines (Plate 28, fig. 22).
Large, freely terminating, conic protulserances, which are hexactine rays and may
be designated as thorns, arise from the beams in many places. In the inner part
of the lamellae these thorns are not numerous; they are small, usually 90-200 ^
long (Plate 28, fig. 22). In their superficial part they are more numerous, more
or less vertical to the surface of the lamella, and larger, 120-360 n long, and about
GO M thick at the base. These superficial thorns are covered with spines similar
to those on the beams, but on the whole larger and more densely crowded.

The superficial skeleton-net (Plate 28, figs. 21, 27), remnants of which have
been found in several places, extends paratangentially on the surface. It is
rather loose and irregular, and consists of pentactines the lateral rays of which
have been more or less soldered together.

The loose hexactines (Plate 28, figs. 17, 18) found in the interior are probably
destined to be soldered together to form the internal skeleton-net. The small,
probably young forms have straight, or slightly curved, nearly smooth rays, 70-
100 M long and 3 m thick at the base. In the larger, probably older ones (Plate
28, figs. 17, 18) the rays are 100-260 tx and more long, nearly cyhndrical, and
10-14 n thick. They are, in the same spicule, often unequal and always covered
with spines. Most of these spines are small, whilst some, which lie irregularly
scattered between the small ones, attain a very large size and measure 10-50 n
in length. These large spines, of which each ray bears from five to ten or more,
increase in size towards the distal end of the ray. The largest of them bear
small secondary spines. Several, usually three, are situated terminally. These
are always the largest. The large spines along the length of the rays arise nearly
vertically, the terminal ones usually point obliquely outward.

The pentactines (Plate 28, figs. 19, 21, 27) are situated superficially. Their
lateral rays, which form the superficial net, are 120-200 n long, 4-10 n thick at the
base, and slightly attenuated towards the end. They are covered throughout
with vertically arising spines. Young, still free, superficial pentactines (Plate 28,
fig. 19) have slender rays and very small spines. Older ones, already incor-
porated in the superficial net (Plate 28, figs. 21, 27), have stouter rays and longer
spines.

Of uncinates two kinds, a smaller and a larger, can be distinguished.

The smaller uncinates (Plate 28, fig. 10), which are very numerous and

146 HEXACTINELLA MONTICULARIS.

doubtlessly proper to the sponge, attain a length of 225-420 n. They are
centrotyle and anisoactine, the tyle, which marks the morphological centre,
being situated much nearer the end from which the spines diverge than the
other. The proportion between the length of the two actines is 2 : 3 to 1 : 3.
Close to the tyle these uncinates are usually 2-3 // thick, the tyle itself being
about 0.7 more in transverse diameter than the adjacent parts of the spicule.
The spines are numerous, very oblique, and so thin that it is impossible to see
them with ordinary light. The u. v. photographs, however, show them clearly
enough (Plate 28, fig. 10). I should say that these spines are scarcely thicker
than 0.1 fi.

The large uncinates are rare and may be foreign to the sponge. All those
observed were broken. The largest fragments were 600-800 ij. long and about 5 n
thick. Their spines are strongly inclined, nearly parallel to the shaft, and
exceedingly thin.

Two kinds of discohexasters, a larger and a smaller, can be distinguished.
These are, it is true, connected by intermediate forms, but the latter are so rare
that the distinction between them is quite clearly pronounced.

The large discohexasters (Plate 28, figs. 12, 15, 16, 25) measure 52-62 n in
total diameter, usually about 60 fi, and have equal and regularly arranged, fairly
smooth main-rays, 5-6 ju long and about 1.8 m thick. Each main-ray bears four
rather strongly divergent end-rays. The end-rays are curved, concave to the
continuation of the main-ray at the base, and straight or slightly curved in an
irregular manner farther on. The end-rays are about 23 /i long, 1.2-1.3 m thick
at the base, and attenuated distally to 0.7-1 ij.. They bear along their whole
length numerous minute, backwardly directed spines and at the end a verticil
of larger, recurved spines, which together form a kind of convex terminal disc
with strongly serrated margin, 1.5-2.2 m in transverse diameter (Plate 28, fig. 12).

The small discohexasters (Plate 28, figs. 11, 20) measure 30-47 fi in total
diameter, and have equal, regularly arranged, fairly smooth main-rays, 4.5-0.5 ^
long and 1-1.6 yu thick. Each main-ray bears four, exceptionally five, end-rays.
These are curved at the base, concave to the continuation of the main-ray, and
nearly straight farther on. In these small discohexasters the basal curvature
usually extends farther than in the large discohexasters. The end-rays are
&-18 IX long, 0.5-1 IX thick at the base, and attenuated distally to 0.4-0.7 ju.
They are covered along their whole length with numerous minute, backwardly
directed spines, and usually bear at the end a verticil of four or more larger
recurved spines, which, when seen in profile, together appear as a convex terminal

HEXACTINELLA MONTICULARIS. 147

disc, 1-2.5 n in transverse diameter (Plate 28, figs. 11, 20). Sometimes these
spines are so small that the end-rays appear terminally rounded and destitute
of terminal discs.

The scopules (Plate 28, figs. 1-9, 13, 14) are 220-400 ^ long. They consist,
of a stout centrum, from one side of which arises a simple shaft, and from the
opposite a verticil of end-rays. Sometimes one or two end-rays are also attached
to the sides of the centrum.

The centrum is 6.2-9.6 ^ broad, 4-9 m long, and has four lateral protulaer-
ances arranged regularly crossways. Wlien small these protuberances appear
as slight rounded elevations (Plate 28, figs. 7-9), when large as short, cylindrical,
terminally rounded ray-rudiments, equaling the shaft in thickness (Plate 28,
figs. 5, 6). The centrum and its protuberances are uniformly and densely
covered with minute spines.

The shaft is 200-345 m long, straight or. rarely, curved. It is nearly
cylindrical for the greater part of its length, and rather abruptly attenuated to a
sharp point. It is 3-4.5 m thick at the base, where it rises from the centrum;
in the middle of its length it is slightly thinner, or, not so frequently, as thick or
slightly thicker than basally. The middle-part of the shaft is nearly smooth.
The proximal part, for a distance of about 30 n from the centrum, is, like the
centrum, densely covered with minute spines. In a belt which is 10-20 m broad
and situated a short distance from the distal end, larger, particularly broader
sparsely scattered spines occur.

Of end-rays there are from five to nine, most frequently seven. As men-
tioned above, these generally aU rise from the apex of the centrum, that is the
face opposite the shaft. These terminal end-rays are slightly curved, concave
to the continuation of the axis of the shaft for a short distance, and, for the
remainder of their length, straight or curved slightly in the opposite direction
(outwards). They diverge above more or less and together form a stouter or
more slender, brush-like or calyculate verticil, 9-22 fi broad at the distal end.
These end-rays are 1 1-05 ju long, most frequently 1 1-30 ix. Of 33 measured : —

inc;

was

under 10 n

long.

1

was

36-40 M lor

5 were 11-15 ^

41-45 M "

9 "

16-20 M

1

40-50 M "

7 "

21-25 n

51-55 M "

7 "

26-30 M

56-60 M "

2 "

31-35 M

1

61-65 M "

was

over

66 m

long

148 HEXACTINELLA MONTICULARIS.

These end-rays are, at the base, 1-2 n thick, very rarely 2.6 n, and attenuated
towards the distal end to 0.6-1 m, rarely 1.4 n. They are densely covered with
minute, backwardly directed spines, and usually bear a terminal verticil of
larger, recurved spines which together form a kind of convex terminal disc with
deeply serrated margin, 1.2-2.5 n in transverse diameter. Sometimes the termi-
nal spines are so small that no disc-shaped terminal thickening at all can be
detected.

The exceptionally occurring lateral end-rays are more divergent, more
curved, and shorter than the terminal ones above described, which they resemble
in all other respects.

From a point in the middle of the centrum six axial threads extend in three
straight lines vertical to each other. One of these is long and well-developed.
This one is continued in the axis of the shaft, which can be traced quite to the
end of the latter. The other five axial threads are short, rudimentary, and
terminate within the centrum. The one in line with and opposite the axis of
the shaft is directed towards the terminal end-ray verticil, and ends before
reaching it without giving off branches. The end-rays are destitute of axial
threads. The other four axial threads terminate in the four lateral protuber-
ances of the centrum.

The shape of the scopules and the arrangement of their axial threads indi-
cate: — that the upper part of the centrum, from which the end-ray verticil
arises, is, as far as it is traversed by the axial thread, an end-ray bearing main-
ray; that the shaft is a well-developed, simple ray; that the four lateral pro-
tuberances of the centrum are rudimentary simple rays, and that the end-rays
are homologous to hexaster end-rays. Thus the whole scopule appears as a
hemihexaster. Since its end-rays bear the terminal verticils of recurved spines
characteristic of the discohexasters and hemidiscohexasters, these scopule-
hcmihexasters are discohemihexasters.

In view of this I think it not unlikely that the scopules of the Hexactinellida
generally are to be considered as apically highly differentiated hemihexasters,
the scopules of the sponge here described being not quite so far advanced in this
development and not so far removed from the ancestral form as the scopules
destitute of lateral protuberances of the centrum of other hexactinellids.

The comparability of the scopules with hexasters was first noticed by F. E.
Schulze who says' concerning their end-rays, "I should be more inclined to
compare them with the terminal rays of the rosettes." But this author does not

1 F. E. Schulze. Kept. Voy. Challenger, 1887, 21, p. 34.

HEXACTINELLA. 149

draw the same conclusion as I should concerning then- origin from this com-
parability and their general structure, and expresses ^ his inclination to consider
them "as diacts or monacts."

In spite of the fragmentary condition of the specimens they can, with a
sufficient degree of certainty, be assigned to Hexactinella. Of all the known
species only two, H. ventilabrum Carter and H. labyrinthica Wilson, have, like
them, discohexastrose microscleres. From both of these species the sponge
above described differs by the scopules, which have four end-rays in the former,
and usually seven end-rays in the latter.

Hexactinella sp. indet.
Plate 32, figs. 13-15.

A skeleton-net probably a species of Hexactinella was trawled off the south-
ern coast of western Panama at Station 4631, on 3 November, 1904; 6° 26' N.,
81° 49' W.; depth 1415 m. (774 f.); they grew on a bottom of green sand; the
bottom-temperature was 38°.

This skeleton-net (Plate 32, fig. 13) has the shape of a funnel 30 mm. high
and 52 mm. in maximum breadth above. The funnel-wall is 4 mm. thick.
Both the upper marginal part and the lower end, which latter may have been
attached to a stalk, are broken off. The funnel-wall consists of skeleton-net
lamellae extending radially and longitudinally from the base towards the margin.
These lamellae are mostly a little over 1 mm. apart and joined to each other by
groups of oblique beams, which, on the inner side of the funnel, form a honey-
comb-like net (Plate 32, fig. 15) composed of lamellae vertical to the surface
and enclosing short, Ukewise vertical canals, round or polygonal in transverse
section, and mostly 1.5-2.5 mm. wide.

The skeleton-net of these lamellae consists of smooth beams, on an average
about 100 M thick, which in some places extend longitudinally and transversely
with rather large square, rectangular meshes, but which are generally, particu-
larly in the inner honeycomb zone, so variable in theii- dii'ection, so crowded,
and joined at so frequent intervals, that they form a quite irregular and very
dense network.

' F. E. Schulzc. Loc. cil., p. 35.

15U HYALONEMA.

Amphidiscophora F. E. Schulze.

Hexactinellida the spicules of which are always isolated; with amphidiscs;
without hexasters.

Of the two famihes into which F. E. Schulze^ divides this suborder, one,
the Hyalonematidae, is represented in the collection.

HYALONEMATIDAE (Gray) F. E. Schulze.

Amphidiscophora in which the afferent apertures all lie in one area, the
gastral face.

The collection contains fifty-seven more or less complete specimens and six
fragments of this family. iVU belong to the genus Hyalonema.

HYALONEMA Quay.

Hyalonematidae with gastral cone, without conuli-like protuberances on
the dermal face; with one or, exceptionally, several stalks composed of long
intertwined anchoring spicules; with acanthophores in the lower end-part of
the body.

Two specimens cannot be specifically determined. The other fifty-five,
and the six fragments, belong to twenty-four species, twenty-two of which are
new.

Hitherto fifty well-defined species of Hyalonema have been described. To
these twenty-two are added in this Report, so that there are now seventy-two
valid species of Hyalonema. The number of species being so great I en-
deavoured to arrange them in subgenera. In attempting to do this I first
thought it might be possible to fall back on F. E. Schulze's ' original division of
the genus into the subgenera Hyalonema (with a special gastral sieve-mem-
brane) and Stylocalyx without such a structure. I found, however, as Schulze
himself did on reconsideration,'* that this could hardly be done with advantage.
Then I tried to attain my object with the help of the key given in Schulze's
Valdivia report,^ but this also helped me only to a small extent. I therefore

' F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 181.

F. E. Schulze. Kept. Voy. Challenger, 1S87, 21, p. 189.

^ F. E. Schulze. Revision des i3y.stemes der Hyalonematiden. Sitzungsb. Akad. Berlin, 1893, no.
30, p. 554.

' F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 163.

HYALONEMA. 151

propose a new arrangement, based on the results of my examination of the
twenty-four Pacific species.

These results have led me to think that certain characters of the amphi-
discs could be utilised for this purpose. It is true that the numerous very differ-
ent forms of these spicules are to a great extent connected by transitions ; there
are, however, in spite of this, some amphidisc-forms not so connected.

The anchor-teeth of the amphidiscs of most of the Pacific Hexactinellida
have smooth margins. In five of them, however, there occurs a particular kind
of amphidiscs with serrated anchor-teeth. For these I establish the subgenus
Prionema. Of the fifty species previously known there are, I beUeve, only two,
//. poculum F. E. Schulze ' and H. validum F. E. Schulze,^ in which amphidiscs
with serrated teeth have been noticed and described. I think it highly probable,
however, that such amphidiscs occur in others also, as for instance in H. lusita-
nicum Bocage, and H. cupressiferum F. E. Schulze, where they have not been
mentioned either because they were overlooked — they are generally small and
clearly visible only with high powers — or because the authors who studied these
sponges did not consider them of importance.

Most of the species of Hyalonema examined by me in which the anchor-
teeth of all the amphidisc forms are smooth-margined, generally have hyper-
bolic, semispherical, or bell-shaped anchors and measure from about a quarter
to a third of the whole spicule in length. In some of them, however, the amphi-
disc-anchors are of other relative dimensions and often also of another shape.
In five of the Pacific species examined, one of which had been previously de-
scribed, the anchors of a certain kind of amphidiscs are more or less semi-
spherical and about half as long as the whole spicule, so that the two anchors
of the same spicule nearly or quite meet in the middle. For these species I
establish the subgenus Oonema. Of the species previously described there are,
besides the one in the A. Agassiz Pacific collection above referred to, four {H.
tenerum F. E. Schulze, H. rohustum F. E. Schulze, H. globiferum F. E. Schulze,
and H. pedunculatum Wilson) which can certainly, and one {H. ovuliferum
F. E. Schulze) which can perhaps, be assigned to this subgenus.

In two of the Pacific species examined by me, one of which had been previ-
ously described, the anchors of the largest amphidiscs are small and relatively
very short and broad. For these species I establish the subgenus Phialonema.

' F. E. Schulze. Rept. V03'. Challenger, 1SS7, 21, p. 208. (This serration is not shown in the
figure of a macramphidisc of this species. Loc. cil., plate 33, fig. 4).

= F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 82, taf. 34, fig. 8.

152 HYALONEMA.

Of the species hitherto described there is, besides the one reexamined by
me which is referred to above, one {H. pellucidum Ijmia) at least, probably
several, which can be referred to this subgenus.

In two of the Pacific species examined one kind of amphidisc has broad and
rather low, umbrella-like amphidisc-anchors. For these I establish the sub-
genus Skianema.

In one of the Pacific species examined I found a peculiar kind of amphidisc
with from one to three branches on the convex side of some or most of its anchor-
teeth, which give to the anchors the appearance of being doubled. For this
species I establish the subgenus Thallonema.

The remaining species of Hyalonema, in which none of the different kinds of
peculiar amphidiscs referred to above occurs, can be divided, in accordance with
the primary division used in F. E. Schulze's key, into those in which the largest
amphidiscs are stout and have a thick shaft; subgenus Hyalonema, and into
those in which these amphidiscs are slender and have a thin shaft; subgenus
Leptonema.

Nine of the Pacific species examined by me, two of which were insufficient
for exact description and for naming, and the great majority of the species of
Hyalonema previously described, belong to the subgenus Hyalonema.

One of the Pacific species examined by me, and at least five previously
described species {H. poculum F. E. Schulze, H. solutum F. E. Schulze, H.
urna F. E. Schulze, H. divergens F. E. Schulze, H. depressum F. E. Schulze)
belong to the subgenus Leptonema.

Possibly H. lusitanicum Bocage and H. cupressiferwn F. E. Schulze men-
tioned above as probably belonging to the subgenus Prionema, and H. ovuliferum
F. E. Schulze assigned to the subgenus Oonema may also belong to the subgenus
Leptonema.

HYALONEMA (Gray) Lendenfeld.

Species, the amphidiscs of which have hyperbolical, semispherical, or bell-
shaped terminal anchors from about one fourth to one third of the whole spicule
in length; without amphidiscs of any other kind. The largest amphidiscs are
stout and have a thick shaft.

The collection contains twenty-three more or less complete specimens and
three fragments of this subgenus. Two of the specimens, apparently represent-
ing two distinct forms, could not be specifically determined; the twenty-one
others and the three fragments belong to seven different species, all of which
are new.

HYALONEMA (HYALONEMA) OBTUSUM. 153

Hyalonema (Hyalonema) obtusum, sp. nov.
gracilis, var. nov.

Plate 33, figs. 1-24; Plate 3-t, figs. 1-19; Plate 35, figs. 1-37; Plate 36, figs. 1-45; Plate 37, figs. 1-22;

Plate 38, figs. 1-8; Plate 39, figs. 1-10.

robusta, var. nov.
Plate 39, figs. 11-41; Plate 40, figs. 1-22.

Two specimens were trawled at two stations in the Tropical Pacific : —
Hyalonema (H.) obiusum var. robusta at Station 3681 (A. A. 2) on 27 August, 1899;
28° 23' N., 126° 57' W. ; depth 4330 m. (2368 f .) ; it grew on light brown volcanic
ooze; the bottom-temperature was 34.6°. H. (H.) o. var. gracilis at Station 3684
(A. A. 17) on 10 September, 1899; 0.50' N., 137° 54' W. ; depth 4504 m. (2463 f.) ;
it grew on light yellow-gray Globigerina ooze. These sponges are distinguished
from their nearest allies by the stout truncate or terminally rounded spines on
their macramphidisc-shafts. To these the name refers.

Although on the whole very similar in their spiculation, these two sponges
differ in respect to their external appearance and certain characters of their
skeletal element so that I consider them distinct varieties. The spicules of the
specimen from Station 3681 (A. A. 2) are generally speaking stouter, those from
Station 3684 more slender. I therefore name the former H. (H.) o. var. robusta,
and the latter H. (H.) o. var. gracilis.

Shape and size. The specimen of var. robusta is rather fragmentary, its super-
ficial parts having to a great extent been lost. It consists (Plate 39, fig. 33) of a
flattened body, 65 mm. long, 12 mm. thick, and 42 mm. broad above. Below
it becomes narrower, and there protrudes from its rounded lower end a bundle
of stalk-spicules. This bundle, where it arises from the sponge-body, is about
2.6 mm. thick. The stalk-spicules forming it are broken off at a distance of
35 mm. from the lower end of the sponge.

The specimen of var. gracilis is well-preserved, but destitute of the stalk ; the
sponge-body having apparently been pulled off the stalk-spicules by the trawl.
It has the shape of a short and broad spindle or top (Plate 33, fig. 15), is 47 mm.
long (high), and has a maximum transverse diameter of 30 mm. The lower end,
from which in life the large stalk-spicules arose, is now simply rounded off.
The upper end consists of a gastral cone closely enveloped by the thin, frill-like
margin of the wall surrounding the gastral cavity. The cone (Plate 33, fig. 16a)
is 9 mm. high, nearly cylindrical, circular in transverse section, terminally
rounded, 6 mm. thick at the base, and 4 mm. at the end. Its end is sUghtly

154 HYALONEMA (HYALONEMA) OBTUSUM.

bent to one side (Plate 34, fig. 3c). The frill .surrounding it terminates with a
fairly circular margin which lies in the level of the summit of the cone. The
gastral cavity appears as a narrow fissure 5-12 mm. deep but only 0.4-1 mm.
wide (Plate 33, fig. 16; Plate 34, fig. 3b) separating the gastral cone from the
marginal part of the sponge-body.

The surface of the cone, and the inner face of the upper tubular marginal
part of the wall surrounding the gastral cavity are smooth and destitute of aper-
tures of any kind, the efferent openings being restricted to the bottom of the
fissure-like gastral cavity. The intact parts of the outer surface exhibit a fine
reticulate structure with meshes about 0.7 mm. wide (Plate 33, fig. 15).

The colour of the specimen of var. robusta in spirit is rather dark reddish
brown, that of var. gracilis light greenish brown.

Canal-system. The state of preservation of the specimen of var. robusta
renders it impossible to say anything about the canal-system. In the specimen
of var. gracilis subdermal cavities (Plate 34, figs, lb, 3, 4, 19c), mostly 0.3-0.7
mm. high and 0.2-0.5 mm. broad, are spread out below the dermal membrane
(the outer surface) . These cavities are generally separated from each other by
thin partitions. From most of them small afferent canals take their origin;
some are directly continued in large afferent canal-stems, 0.3-0.7 mm. wide,
which extend somewhat tortuously towards the interior, and ramify in the
central part of the sponge. Occasionally junctions of two such afferent canal-
stems have been observed. The choanosome, that is the region occupied by the
flagellate chambers, does not extend, for the most part, beyond the level of the
floors of the subdermal cavities. In a few places only broad, conical groups of
flagellate chambers rise between adjacent subdermal cavities, up to a distance
of only 0.1 mm. from the outer surface.

The individual flagellate chambers appear to be broad oval or nearly
spherical, and attain a maximum diameter of 60-100 /z (Plate 34, fig. 2). The
efferent canals join to form canal-stems up to 1.2 mm. wide, which, as above
mentioned, open out into the bottom of the narrow, fissure-like gastral cavity.
The larger of these canals are considerably contracted at the mouth.

The skeleton of var. gracilis. The outer surface is covered with dermal
pinules, micramphidiscs, and small macramphidiscs. Most of the pinules are
pentactine, some hexactine. Their paratangentially extending lateral rays lie
in the dermal membrane; their radially extending and freely protruding distal
rays form a fur about 150 ^ high (Plate 35, fig. 24). The micramphidiscs
are, in some places at least, exceedingly numerous. They seem to be quite

HYALONEMA (HYALONEMA) OBTUSUM. 155

irregularly situated. The small macramphidiscs are also numerous and often
arranged in groups (Plate 34, figs. 1, 19b; Plate 35, fig. 24b). Their shafts ex-
tend radially or obUquely and their distal parts protrude freely beyond the
surface.

The dermal membrane is supported by hypodermal pentactines very
variable m size. In the upper parts of the sponge the large pentactines greatly
predominate, at the base the small ones are more numerous. The centres of the
large hypodermal pentactines are about 0.7 mm. apart. The apical rays of
these spicules are directed radially inward (Plate 34, fig. Ic); their lateral rays,
which are markedly incUned towards the apical ray, extend nearly paratangen-
tially in the beams of the superficial network above referred to. Uncinate
amphioxes, situated for the most part radially or obliquely, are met with in the
subdermal region. The superficial part of the choanosome underlying the
dermal surface is occupied, down to a depth of about 2.5 nmi., by hexactine
megascleres, rather regularly arranged in several paratangentially extending
layers. These hexactines are situated so that two of their rays extend radially
(inwards and outwards) , two longitudinally (upwards and downwards) , and two
transversely (to the right and left). The distance between the centres of these
spicules is less than the length of their rays, and the opposite rays of adjacent
ones usually extend for some distance side by side and close together (Plate 34,
fig. 19d). These hexactine megascleres, therefore, form a three-dimensional
network with fakly regular, somewhat cubic meshes. These spicules vary
greatly in size ; the larger are situated proximally, the smaller distally.

Numerous rhabd-megascleres and a few angularly bent diactines of similar
dimensions occur in the choanosome. Most of the rhabds are blunt amphioxes
or amphistrongyles, but styles and tylostyles also occur. Some of these rhabds
are isolated; most of them, however, form loose strands. In the central (axial)
part of the choanosome, the rhabds extend for the most part longitudinally; in
the other parts of the choanosome they are mostly dii-ected obliquely upwards
and outwards, and generally lie in the walls of the canals. The styles and tylo-
styles are situated so that their rounded (thickened) end points downward and
inward, their pointed end upward and outward. The choanosome is rich in
microscleres. Large numbers of micramphidiscs are imbedded in the canal-walls
and throughout it are scattered some macramphidiscs, masses of microhexac-
tines (Plate 34, fig. 2), and a few microhexactine-derivates, chiefly with only
two opposite rays fully developed and the others more or less, sometimes entirely
reduced.

156 HYALONEMA (HYALONEMA) OBTUSUM.

As above stated the sponge-body was in life obviously attached to a bundle
of stalk-spicules, which have, however, been pulled out of it. Empty tubular
spaces, sometimes 0.9 mm. wide (Plate 36, fig. 26a), the walls of which are
formed by fine, highly stainable membranes, mark the places where the upper
ends of the largest of these stalk-spicules were situated. These spaces lie in the
axial part of the sponge-body. They are conical, attenuated above, and extend
upwards to within a distance of 2 mm. from the summit of the gastral cone.
In the lower part of the sponge-body these spaces are surrounded by a kind of
cement, composed of dense masses of stout, one- to five-rayed, most frequently
tetractine or diactine acanthophores (Plate 36, fig. 26). In this cement a
few microhexactine-derivate pachymicrohexactines also occur. Quite at the
bottom, a short distance below the dermal membrane, numerous slender-rayed
spicules with long spines, which I consider as slender acanthophores, form a
kind of felt. These spicules are mostly tetractines, but a good many triactines
and a few diactines, pentactines, and hexactines also occur among them. Transi-
tional forms, connecting these spicules on the one hand with the stout acantho-
phores above referred to, and on the other with the dermal pinules, are also found
in this part of the sponge.

The thin marginal part of the circular wall which surrounds the gastral
cavity, and forms the boundary between the dermal and the gastral parts of the
surface, contains numerous, longitudinally situated, diactine pinules, the distal
rays of which protrude freely beyond the surface.

The gastral surface, that is the surface of the gastral cone, and the inner
surface of the wall surrounding the fissure-like gastral cavity are covered with
micramphidiscs and gastral pinules. The micramphidiscs are situated irregu-
larly, and in some places are so numerous as to form dense masses. The gastral
micramphidisc-layer does not terminate at the openings of the efferent canal-
stems into the gastral cavity, but is continued in the walls of these canals and
their branches quite down into the innermost parts of the choanosome. The
gastral pinules are mostly pentactine, but hexactine forms also occur. Their
centres are 30-100 m apart. Their lateral rays extend paratangentially in the
gastral membrane; their distal apical rays arise vertically from the surface, and
protrude freely beyond it, forming a dense fur about 125 m high (Plate 35, figs. 1,
3, 16). This pinule-fur is not, like the micramphidisc-layer, continued down the
efferent canals, but terminates at their mouths.

Small hypogastral pentactines, similar in position to the hypodermal pen-
tactines above referred to, occur below the surface of the cone.

HYALONEMA (HYALONEMA) OBTUSUM. 157

Strands of longitudinal rhabds enter the waU of the gastral cavity and the
gastral cone from below. The rhabds of the gastral wall for the most part
follow the gastral membrane, and here form a dense and distinct subgastral
layer. The rhabds of the cone lie partly superficially, partly axially. The
superficial rhabds of the cone are more slender than the axial ones. The axial
rhabds are congregated in strands which together form a loose column extending
quite to the summit of the cone (Plate 34, fig. 3). A few hexactine megascleres
apparently with long longitudinal and shorter transverse rays also take part in
the formation of this column. The micramphidiscs not only form a dense layer
on the outer surface of the gastral cone, but also extend for some distance into
its interior. Farther down, at a level about 0.3 mm. below the surface, micro-
hexactines and microhexactine-derivates, similar to those of the choanosome,
make their appearance in the cone. These spicules extend down to a depth of
about 0.8 mm., thus occupying a zone about 0.5 mm. thick. The inicrohexac-
tines are very numerous in this zone, the microhexactine-derivates rare. The
central part of the cone, in which the axial column of longitudinal rhabds ex-
tends, is destitute of microscleres.

The skeleton of var. robusta appears to be on the whole similar. The micro-
hexactine-derivates are more various ; the uncinates attain a larger size, and reach
down to greater depths of the sponge. The upper ends of the large stalk-spicules
are still present, and the felt formed by the slender acanthophores in the basal
part of the sponge-body is denser and more extensive (Plate 39, figs. 22-24).

The dermal pinules (Plate 35, figs. 23, 24a, 25, 29-37; Plate 40, figs. 4, 5)
are mostly pentactine, but hexactine forms also occur, differing from the pentac-
tine ones only bj^ possessing a sixth, proximal, apical ray. The distal apical ray
is generally straight, rarely angularly bent below the middle of its length (Plate
35, fig. 25). In the dermal pinules of var. gracilis 137-165 n, the ray is usually
143-154 /i long and 4-5 yu thick at the base. Farther up it thickens and it gener-
ally terminates with a stout, broad, and blunt cone 8-11 n tliick. Rarely it is
terminally rounded, dome-shaped, and has a maximum thickness of 12 yu (Plate
35, fig. 31). The proximal part and the terminal cone (or dome) of the distal
ray are smooth, its middle-part bears sharp, conic spines. The lowest of these
spines are sparse, short, and strongly divergent. Farther up they become more
numerous. The size of the spines increases up to the middle of the length of the
distal ray and then again decreases; the inclination of the spines towards the
tip of the ray increases continuously quite to the end. The largest spines attain
a length of 13-19 n, and are 2-4 n thick at the base, usually about 3.5 yu. The

158 HYALONEMA (HYALONEMA) OBTUSUM.

maximum transverse diameter of the distal ray, together with the spines is
18-32 IX. The distal ray of the dermal pinules of var. rohisia is 140-172 ^
long and 5-8 n thick at the base. The maximum transverse diameter of this ray,
together with its spines, is 25-40 /x.

The proximal apical ray, of the hexactine forms of var. gracilis, (Plate 35,
figs. 29, 30) is straight, 10-42 n long, and 3.7-4.5 m thick at the base. It is
cylindroconic, generally abruptly and sharply pointed, and covered with minute
spines.

The four lateral rays of the same spicule are usually fairly equal, the greatest
difference of length observed not exceeding 4 /i. The lateral rays are straight;
in the dermal spinules of var. gracilis they are 20-40 /j. long, rarely up to 50 n,
and 3-5 n thick at the base; in those of var. robusta they are sometimes (Plate 40,
fig. 4) much shorter, 10-36 m long. They are cj'lindroconic, rounded or, more
rarely, abruptly pointed, and covered with minute spines.

The gastral pinules (Plate 35, figs, la, 2, 3a, 4-9, 16; Plate 40, fig. 3) are, like
the dermal pinules, mostly pentactine; hexactine forms, however, also occur dif-
fering from the pentactine only by possessing a sixth apical proximal ray. The
distal ray is straight; in the gastral pinules of var. gracilis, it is 73-145 m long,
usually 77-135 m, and 4.5-5.5 /x thick at the base, in the gastral pinules of var.
robusta 94-140 m long and 5-7 n thick at the base. It is thickened above, and
attains its greatest thickness a little way beyond the middle of its length ; then
it again becomes thinner, and it ends in a usually sharp-pointed, rather slender,
terminal cone, which does not exceed the proximal end of the ray in thickness.
The proximal end-part and the distal cone of the ray are smooth; its middle-part
bears spines. The spines on the proximal half of the distal ray are very sparse,
point obliquely upward, and are strongly divergent, the angles enclosed between
them and the ray being 40°-55°. The spines on the distal half of the ray are
smaller, more crowded, and less divergent, their size decreasing and their inclina-
tion increasing towards the end of the ray. They attain a length of 15-25 m
and a basal thickness of 3-4.5 m- The maximum transverse diameter of the
distal ray, together with the spines, is in both varieties 26-45 n.

The proximal ray (of the hexactine forms) (Plate 35, fig. 4; Plate 40, fig. 3)
is straight, 43-74 n long, 4.5-5.5 m thick at the base, conical, pointed, and spiny.

The four lateral rays of the same spicule are fairly equal or rather unequal
in size; the maximum difference observed in their length was 15 m- The lateral
rays are straight or, more rarely, sUghtly curved. Their length is subject to
considerable variation. They are in the gastral pinules of var. gracilis 35-90 ju

HYALONEMA (HYALONEMA) OBTUSUM. 159

long, usually 45-70 n, at the base 4.5-5.5 ix thick, rarely as much as 6 m, conic,
pointed or, more rarely, rounded at the end, and covered with spines, which are
more conspicuous in the distal than in the proximal portion of the rays. In
the gastral pinules of var. rohusta the lateral rays are on an average somewhat
longer, they measure here 47-75 y. in length.

The marginal pinules (Plate 36, figs. 10-13, 26-28) have been found only in
var. gracilis; in the specimen of var. rohusta they appear to have been lost. In
these pinules only the distal and proximal rays are properly developed, the lat-
eral rays being altogether rudimentary, and together forming merely a tyle.
These spicules consequently appear as centrotyle diactines. The outer, distal
one of their two properly developed rays, which corresponds to the distal apical
ray of the hexactine and pentactine pinules, is 304-300 ^ long, rarely as much
as 450 IX, fairly straight, and 5-10 tx thick at the base. Its proximal part, and its
distal, conic, sharp-pointed end-part, are smooth. Its middle-part bears spines,
which are rather strongly inclined towards the end of the ray, attain 6-10 ai in
length, and are 1.5-2 n thick at the base. The maximum transverse diameter of
this ray, together with the spines, is 17-20 //.

The opposite, inner, proximal one of their two properly developed rays, which
corresponds to the proximal apical ray of the hexactine pinules, is usually 490-
665 n long, fairly straight, at the base about as thick as the distal ray, and
attenuated towards the end. Sometimes it is greatly reduced in length, only 50 ix
long, cylindrical, and thickened at the end to a terminal tyle 13 ix in diameter.

The central tyle, which is all there is left of the reduced lateral rays, is 3-7 ix
thicker than the adjacent parts of the spicules, and measures 8-15 m in transverse
diameter.

The hypodermal pentactines of the outer surface (Plate 33, figs. 5-14, 17, 24;
Plate 34, fig. Ic; Plate 39, figs. 31, 32, 40, 41). The proximal apical ray is
straight or, rarely, slightly curved, and usually properly developed, conic, and
blunt-pointed, occasionally reduced, cylindrical, and terminally thickened
(Plate 34, fig. Ic). It is in the large hypodermal pentactines, which greatly pre-
dominate in the upper parts of the specimen of var. gracilis, when properly de-
veloped, 0.7-1.86 mm. long and, at the base, 30-75 /x thick, rarely 90 ix. When
reduced it retains its thickness throughout, but is less than half as long. In the
small hypodermal pentactines, which occur chiefly in the lower part of this
sponge, the proximal ray measures 0.27-0.6 mm. by 10-23 ix. The lateral rays
are inclined towards the proximal ray, and enclose with it angles of 73°-84°.
The four lateral rays of the same spicule are usuallj^ very unequal in length ; the

160 HYALONEMA (HYALONEMA) OBTUSUM.

longest is sometimes more than twice as long as the shortest. The greatest
difference in length between the lateral rays of the same hypodermal pentactine
observed was 310 m- The lateral rays are straight, conic, and blunt. They are,
in the large hypodermal pentactines of var. gracilis, 240-730 fj. long and 32-60 ^
thick at the base. In the small ones they measure 135-440 ix in length. The
small hypodermal pentactines accordingly have, relative to the proximal ray,
considerably longer lateral rays than the large ones. At the end the lateral rays
are usually from one fifth to one third as thick as at the base, and here measure
5-22 IX in transverse diameter. The hypodermal pentactines of var. robusta
have a proximal ray 0.47-1.3 mm. long, and 40-80 yu thick at the base. The
lateral rays are on the whole attenuated towards the distal end less than in the
hypodermal pentactines of var. gracilis. They are, when not reduced, 250-750 n
long and, at the base, about as thick as the proximal ray.

The end-parts of the lateral rays of these spicules exhibit remarkable irregu-
larities of external shape and internal structure. These irregularities are the
more conspicuous the thicker (the more blunt) the rays. Such an irregular
lateral ray-end of a hypodermal pentactine of var. gracilis is represented (Plate
33, fig. 17). A rudiment of a branch-ray, arising a short distance from the tip
of the main-ray, and a marked irregularity in the axial thread and the stratifi-
cation of the siliceous body of the latter are noticeable in this spicule. I am
inclined to ascribe these irregularities to the influence of the obstacles — other
spicules — met by these rays during their longitudinal growth. The cells
building the tips of these rays were forced to act in an abnormal manner; being
prevented by other spicules from adding to the length of the axial thread and from
depositing silica around it in a normal and regular manner, they produced the
irregular structures observed. The obstacles (other spicules) which thus cause
these irregularities are probably the stout proximal rays of adjacent hypodermal
pentactines.

The hypogastral pentactines of the gastral cone of var. gracilis have straight
proximal apical rays, usually 240-400 n long, and about 12-20 n thick at the
base. The lateral rays are slightly inclined towards the proximal ray, straight,
and generally 180-250 n long. In the specimen of var. robusta the parts contain-
ing these spicules appear to have been lost.

The hexactine megascleres of the distal part of the choanosome (Plate 34,
figs. 5-18, 19d) are in both varieties very variable in size and have a maximum
diameter of 350 /i - 2 mm. The rays of the same spicule are often unequal. The
greatest difference of length observed in them was 400 /*. The rays are 130-950 ix

HYALONEMA (HYALOXEMA) OBTUSUM. 16l

long, straight, conic, rather blunt-pointed, and 7-35 /u thick at the base. Their
basal thickness is roughly speaking in proportion to their length.

Very young stages of these hexactines appear as spheres, 20 ix in diameter,
perforated by six axial cylinder threads, 5 m thick, which are joined at right
angles in the centre. Wliere these axial cylinder threads reach the surface of
the sphere this is elevated in the shape of very thin-walled tubes rising about
10 II over the surface of the sphere (Plate 39, fig. 5).

The hexadine megascleres of the loose axial spicular column, which were found
only in var. gracilis, appear to be larger than the more superficially situated, but
since I have not been able to find any intact ones, I can only say that their
longitudinally extenchng rays appear to be much longer than their transverse
rays, and that their rays are, at the base, about 40 ^ thick.

The stout acanthophores (Plate 36, figs. 1-25, 27-45; Plate 39, figs. 17-21,
34-38) of the basal part of the sponge-body range from pentactine to monactine.

The peritactines are rare. The few observed in var. gracilis were 225-530 m
in diameter, and had rays, at the base, 12-29 ii thick.

The tetractines (Plate 36, figs. 1-25, 27, 28; Plate 39, figs. 18-20) generally
have more or less unequal rays. The inequality of the rays is often very con-
siderable. The rays are exceedingly variable in size, curvature, shape, and
spinulation, but constant and uniform in so far as their basal parts always form
a fairly regular, rectangular cross, and as the rays themselves always appear to
extend nearly in one plane. The tips of the rays are nearly always more or less
spiny, only quite exceptionally (Plate 39, fig. 20) entirely smooth. In both
varieties these spicules measure 180-840 /x in total diameter. Among the irregu-
lar ones all sizes between these limits are met. The regular ones never appear
to exceed 500 ^ in diameter. The rays are generally wavy in outline, cylindro-
conical or cylindrical, and chstally thickened, or, more rarely, without a thick-
ening at or near the end (Plate 36, fig. 1; Plate 39, fig. 20). The ray either
terminates with the distal thickening and then appears simply rounded off at the
end (Plate 36, figs. 22, 23, 25; Plate 39, fig. 18), or it is continued beyond the distal
thickening in the shape of a terminal cone (Plate 36, fig. 7). The rays of these
spicules are in var. gracilis 35-380 fi long and 12-35 fi thick at the base; in var.
robusta, where they are more irregular and stouter, 40-500 n long and, at the base,
20-50 n thick. The distal thickening is in the tetractines of var. gracilis 10-40 n
in chameter, in those of var. robusta 10-60 /u.

The thickness of the rays is not in proportion to their length, and varies in
the rays of all lengths between similar limits. We consequently find among the

162 HYALONEMA (HYALONEMA) OBTUSUM.

short rays relatively much stouter ones than among the long rays. The rays
most strongly reduced, that is those under 55 ai in length, are by far the relatively
stoutest. None of the rays as short as this was in gracilis under 25 ix, several of
them were here 35 n thick.

In both varieties the tips of the rays bear broad, conic, vertical spines with a
maximum length of 4 /i (Plate 36, figs. 27, 28). On the distal thickening these
spines are usually densely crowded (Plate 36, figs. 6, 7, 9), more rarely sparsely
scattered (Plate 36, fig. 4, fig. 18, the upper and lower ray, fig. 23, the upper and
left ray). They are usually confined to the distal thickening, the proximal part
of the ray and the conic tip (when present) protruding beyond it being quite
smooth (Plate 36, figs. 1, 2, 4-25, 27, 28). Sometimes, however, the spines cover
the whole ray (Plate 36, fig. 3) in greater or smaller numbers.

It is to be noted that the axial thread is in many of these tetractines, particu-
larly in the slender-rayed (perhaps young) ones, remarkably wide (Plate 36,
fig. 27) , sometimes as much as half as thick as the ray itself.

The triactines (Plate 39, fig. 21) are obviously tetractine-derivates, in which
one of the rays has quite disappeared. They are more frequent in var. robusta
than in var. gracilis, measure in both 330-760 ju in total length, and have rays
about 37-42 n thick in the former, and about 20 /x tliick in the latter.

The diactine and monactine rhabds are of two kinds: — shorter and stouter
tetractine-derivates, and longer and more slender derivates of the ordinary
rhabds of the choanosome.

The tetractine-derivate rhabds of the cement-mantles (Plate 36, figs. 31,
40-45; Plate 39, figs. 17, 34-38) are generally diactine and sUghtly and irregu-
larly curved, rarely (Plate 39, fig. 17) strongly angularly bent. Such strongly
bent, compass-like spicules have only been found in var. robusta. In var.
gracilis some small spicules, also apparently belonging to this group, have been
observed, in which the curvature is so great that one half of the spicule forms a
complete circle (Plate 36, fig. 45). The tetractine-derivate rhabds in var.
gracilis are 170-950 m long and 11-39 n thick near the middle; in var. robusta
450 ,u-1.42 mm. long and 6-50 ti thick near the middle. In the shorter spicules
of this kind a central tyle is usually present, but the longer ones are often without
it (Plate 39, figs. 34, 35, 37). When present the central tyle is, in var. gracilis,
as much as 10 m, and, in var. robusta, as much as 20 fi, thicker than the adjacent
parts of the spicule; in transverse diameter they measure in the former 13-39 fi,
and in the latter 25-70 ii. It either passes gradually into the body of the spicule
(Plate 36, figs. 40, 44; Plate 39, figs. 36, 38) or it is set off from it more or less
distinctly (Plate 36, figs. 31, 41-43). Most of these spicules are fairly isoactine,

HYALONEMA (HYALONEMA) OBTUSUM. 163

some distinctly anisoactine (Plate 36, fig. 31 ; Plate 39, fig. 38). Their end-parts
are thickened more or less to spherical tyles (Plate 36, figs. 31, the left one, 41,
44; Plate 39, figs. 34, 35, 36, the right one, 37) or spindle-shaped tyles (Plate 36,
figs. 40, 42, 43). These distal thickenings are in var. gracilis usually smaller,
rarely (Plate 36, fig. 44, the right one) stouter, than the central tyle. In var.
robusta they attain much larger dimensions and have a maximum diameter of
90 IX. The middle-part of the spicule is smooth. The two ends are generally
covered with spines, for a shorter or longer distance (Plate 39, fig. 38, the left
one); they are rarely smooth (Plate 36, fig. 41; Plate 39, fig. 37).

Although doubtless derived from the tetractines (triactines) among which
they occur, these rhabds are hardly at all connected with them by transitional
forms, and therefore readily distinguishable from them.

The acanthophore rhabds, which are to be considered as derivates of the
ordinary rhabds of the choanosome, are in var. gracilis (Plate 36, figs. 29, 30, 32-
39), where they appear to be more numerous than in the other variety, 0.6-2 mm.
long and 10-18 ju thick in the middle. Most of them are rather uniformly curved
throughout (Plate 36, figs. 29, 30, 32, 34-36), some are irregularly curved (Plate

36, fig. 33), and a few strongly angularly bent near the middle (Plate 36, fig. 39).
Some of them are fairly isoactine (Plate 36, figs. 30, 37-39), others distinctly
anisoactine (Plate 36, figs. 29, 32-36). i\ll these spicules are more or less thick-
ened at both ends. In the isoactine forms both terminal thickenings are slight,
spindle-shaped, and situated a short distance below the end (Plate 36, figs. 30,

37, 38). In the anisoactine only one of the distal thickenings is of this nature,
the other being stouter, 25-45 m in diameter, spherical or oval, and situated
terminally (Plate 36, figs. 29, 32-36). The spindle-shaped thickening usually
passes gradually into the body of the spicule; sometimes it is distinctly set off
from it (Plate 36, fig. 32). The shaft or body of the spicule is smooth. The ends
are sometimes also smooth (Plate 36, fig. 33); usually, however, one (Plate 36,
figs. 29, 36) or both (Plate 36, figs. 30, 32, 34, 35) of them bear spines. The axial
thread is widened in the spindle-shaped distal thickenings (Plate 36, figs. 37, 38)
and extends quite to their end. In the ray-ends thickened to a stouter, spherical
or oval terminal tyle, the axial thread does not extend quite to the end. The
silica-layers of the isoactine forms therefore appear as tubes open at lioth ends,
those of the anisoactine forms as tubes open at one end only.

An abnormal stout acanthophore 220 fj. in diameter was found in var. gracilis.
Its rays are straight, cylindrical, terminally rounded, and very unequal in length,
but all about 10 ^ thick.

The intermediate transitional acanthophores (Plate 39, figs. 1, 6, 11, 12)

164 HYALONEMA (HYALONEMA) OBTUSUM.

have rays which are in both varieties 4-6 m thick and covered with stout, blunt
(Plate 39, fig. 1) or pointed (Plate 39, figs. 11, 12), usually curved, oblique spines
2-4 M long.

The slender acanthophores (Plate 39, figs. 2-4, 13-16, 22-24) are mostly
tetractine, triactine or diactine tetractine-derivates ; a few hexactine and pentac-
tine forms appear to be pinule-derivates. In both varieties the rays of these
spicules are sometimes 200 yu long and, at the base, in var. gracilis 1.3-1.5 n, in
var. robusta 1.5-3.5 m thick. They are usually curved more or less in an irregular
manner and bear sparse, irregularly distributed spines. In both varieties these
spines reach 6 /x in length and are usually more or less curved. The spines
arising from the end-parts of the rays are usually directed backwards and re-
curved; the others are either also recurved, or vertical, or directed outwards.
The basal parts of the rays appear always to retain their original, regular, relative
position. In the tetractines these parts of the rays form regular rectangular
crosses, in the triactines a T, and in the diactines usually a right angle. When a
ray entirely disappears, a large spine usually takes its place (Plate 39, fig. 16).

The rhabds of the choanosome and gastral cone are for the most part blunt
amphioxes or amphistrongyles, but a few styles and tylostyles are also found
among them.

The hlunt amphioxes and amphistrongyles of the choanosome and cone (Plate
33, figs. 1, 2, 18, 19, 21-23; Plate 39, fig. 39) are in both varieties nearly straight,
slightly curved, or angularly bent, and usually provided with a more or less
prominent central tyle. They are perfectly smooth. Their end-parts are
generally somewhat wavy in outline. The amphioxes (amphistrongyles) of the
upper part of the choanosome and the wall of the gastral cavity are 0.5-1.55 mm.
long and 4-20 m thick near the central tyle. The tyle is 2-8 m thicker than the
adjacent parts of the spicule, and measures 11-28 ^ in transverse diameter.
When there is an angular bend the apex of the angle invariably lies in the central
tyle (Plate 33, fig. 2). In the basal and the axial parts of the sponge these
amphioxes (amphistrongyles) attain a larger size. They are here 1-3.5 mm. and
more long and 8-50 m thick.

The styles and tylostyles of the choanosome and gastral cone (Plate 33, fig. 20)
are in both varieties shorter than the isoactine rhabds above described, usually
only 0.9-1.6 mm. long. The largest terminal tyles of the tylostyles observed
were 30 m in diameter.

The uncinate amphioxes of the superficial parts of the sponge (Plate 33, figs. 3,
4; Plate 39, figs. 25-30) are straight, or slightly curved, and sharp-pointed at
both ends. They are in var. gracilis 580 m long, and 4.5-9 n thick in the middle.

HYALOx\EMA (HYALOXEMA) OBTUSU.M. 165

In war. robusta they are larger, sometimes 1.1 mm. long and 20 ^ thick. A slight
central thickening (tyle) with an axial cross in the interior can usually be made
out, particularly in the smaller uncinates. This is generally about 0.5, rarely
as much as 1 yn thicker than the adjacent parts of the spicule. The spicule is
covered with slender spines, all strongly inclined in the same direction. Near
the end from which these spines diverge, they are rather numerous, towards the
other end they become very scarce. So far as I could make out these spines
consist of a rather broad conic basal part and a fine, exceedingly slender, needle-
like end-part. The basal conic part arises steeply from the shaft and bends
round above, where it passes into the fine end-part, so that the latter comes to
lie nearly parallel to the shaft.

The large stalk-spicules (Plate 40, figs. 21, 22), in var. roMista 8 mm. below
their upper ends, where they are all broken off, are 40-720 ix thick. The empty
spaces previously occupied by them in var. gracilis have a maximum width of
900 fi. The upper ends of these spicules of var. robusta are curved, the curvature
increasing towards the (upper) end. The axial thread is for the most part 3-4 ^
thick. It does not lie centrally, but describes a spiral line around the mathe-
matical axis of the nearly cylindrical spicule. It is by no means a simple cylindri-
cal thread. Some parts of it (Plate 40, fig. 21) are uncinate-like, covered with
strongly inclined spine-like processes directed upwards, others (Plate 40, fig. 22)
are thickened, quite irregular, and attain 20 fi in transverse diameter.

In both varieties the regular microhexactines (Plate 35, figs. 14, 15, 17a, 18,
19; Plate 40, figs. 6, 7, 20b) measure 42-80 // in diameter. The six rays of the
same spicule are fairly equal, and regularly arranged. The chords of the rays
are 20^3 m long. The rays themselves are 1.5-2.2 ^ thick at the base, gradually
and uniformly attenuated distally to a fine point, and covered with very minute,
vertically arising spines. The basal parts of the rays are nearly straight, the
distal parts strongly curved through an angle usually a little over 90°. The
direction of curvature of the end-part of each individual ray is generally opposite
to that of the end-part of the ray opposite it in the same axis.

The microhexactine-derivates (Plate 35, figs. 20-22; Plate 40, figs. 8-15, 20c)
represent two series of forms. One begins with microhexactines in which the
two rays lying opposite in the same axis are longer than the other four, and ends
with centrotyle diactines. The other begins with micropentactines with equal
rays, and ends with style monactines. In var. robusta forms of both series are
rather frequent ; in var. gracilis hardly any but diactine forms, with the two fully
developed rays opposite in the same axis, have been observed.

First series of microhexactine-derivates. One of the microhexactines, with

166 HYALONEMA (HYALONEMA) OBTUSUM.

two opposite longer, and four shorter rays, with which the first series commences,
is represented in Plate 40, fig. 9. This spicule is 117 ^ in length. Forms still
farther removed from the regular microhexactine are produced by a further
reduction of the four shorter rays of such a spicule. The reduction of the four
shorter rays is either unequal or more or less equal. In the first case pentactines,
tetractines, and triactines (Plate 40, fig. 11) with two opposite longer rays, and
three, two or only one shorter, are produced ; in the second case forms like those
represented on Plate 35, figs. 20-22, and Plate 40, figs. 12, 13, and 20c. In the
extreme forms of this series all that remains of the shorter rays is a slight tyle
(Plate 35, fig. 22; Plate 40, fig. 12). It is to be noted that a distinct increase
in size of the two opposite, developed rays is, in these spicules, associated with
the reduction of the four other rays. Such diactine microhexactine-derivates
are, particularly in var. gracilis, more numerous than .any of the others. They
are in both varieties 156-204 m long, but in var. robusta considerably stouter than
in var. gracilis, the basal parts of their properly developed rays being in the former
1.5^ M, while in the latter only 1.5-2.5 m thick. The fully developed rays of
these spicules are gradually attenuated to fine points, straight in their basal part
and curved at the end. The reduced ones are straight throughout, cylindrical
or cyhndroconic, terminally rounded, and reach 6 ^i in length. The terminal
curvature of the fully developed rays is not so great as in the rays of the regular
microhexactines, nor is its direction generally opposite.

To the second series of microhexactine-derivates belong the spicules repre-
sented on Plate 40, figs. 8, 10, 14, and 15. The first (fig. 8) of these is a pentactine
with equal rays, 100 m in diameter. The second (fig. 10) is a compass-shaped
diactine. It consists of two fully developed rays, 47 m long, the basal parts of
which enclose a right angle; and the insignificant rudiments of two other rays
opposite to the two fully developed ones. The third and fourth (figs. 14, 15)
are monactines. Such monactines are more frequent than the other forms of
this series. They are 73-86 m long. Their single fully developed ray is 2.5-4 m
thick at the base and tapers gradually to a fine point. It is straight in its basal
part but strongly curved, through an angle of about 120°, in its distal part.

These spicules are, like the regular microhexactines, covered with minute
spines. In the larger ones the spines are more conspicuous than in the smaller
ones, the size of the spines being, generally speaking, proportional to the thick-
ness of the ray from which they arise.

The pachy7nicrohexactines (Plate 39, figs. 7-10) are rather rare, and have
only been found in the basal part of var. gracilis. I consider them as hypertrophic

HYALONEMA (HYALONEMA) OBTUSUM. 167

microhexactines. They consist of six fairly equal rays joined at right angles,
and measure 52-SO n in total diameter. Their rays are cylindrical, of nearly
uniform stoutness throughout, and rounded at the end. They are 26-42 fi long,
5-15 n thick, and generally quite smooth. Their basal part is straight, their
end-part either straight (Plate 39, fig. 7) or more or less curved (Plate 39, figs.
8-10). Axial threads, terminating however a considerable distance below their
ends, can be easily made out in the rays of these spicules.

Among the amphidiscs of var. gracilis two main groups can be distinguished
morphologically: — large forms, the largest of which have broad and rather short
terminal anchors and a stout, spiny shaft; and small forms the largest of which
have long and very slender terminal anchors and a slender shaft with very small
spines. In each of these main groups, which I name macramphidiscs and micr-
amphidiscs respectively, two subgroups can be distinguished: — in the macram-
phidiscs larger forms with relatively shorter, and smaller forms with relatively
longer, terminal anchors; in the micramphidiscs larger forms with long and
slender anchors, and smaller forms with shorter and broader anchors.

The biological length frequency-curve of these amphidiscs exhibits (Fig. 4)
a gap between the lengths 54.76 m and 66.26 ai. The amphidiscs, to which the
part of the cur\-e to the right of the gap pertains, are the amphidiscs referred to
abo\'e as macramphidiscs ; those to the left of the gap as micramphidiscs. Each
of these two parts of the curve exhibits a conspicuous depression dividing it into
two distinct elevations. These elevations correspond to the smaller and larger
kinds of the macramphidiscs and the micramj^hidiscs, which are, as above stated,
also distinguished from each other morphologically by the shape of their terminal
anchors.

Thus both the morphological and the biometrical qualities of these amphi-
discs show that /our kinds of these spicules are to be distinguished in var. gracilis :
— large macramphidiscs, small macramphidiscs, large micramphidiscs, and small
micramphidiscs.

The amphidiscs of ^•ar. robusta also fall into these four groups.

The large macramphidiscs of var. gracilis (Plate 37, figs. 20-22) are some-
what infrequent. They are 250-356 ^ long, most frequently about ' 264 n, and
have a straight shaft 8-14 ii thick. This is thickened slightly and gradually
to 14-22 fi at the ends, and abruptly in its middle-part to a central tyle 15-18 m
in diameter. The tyle never appears to lie quite in the middle ; the difference

'This phra-se " most frequently about " refers, throughout the descriptions, to the summit of that
part of the length frequency-curve of the graph which pertains to the amphidiscs in question.

168

HYALONEMA (HYALONEMA) OBTUSUM.

Q.

B

<

X)

e

Micramphidiscs

Macramphidiscs

' O ^ CM O 00 l-^ ^ r^ C^ Ovl O r-H Cj I^ <

i;£>ot>-ooooa)0^cM^LOi^coo

^.-1l-l,-lr-(,-(,-ICVll

to 3
f~ o

Length of Amphidiscs (^).
Fig. 4. — Length frequency-curve.

between its distances from the two ends is 7-37 n, 3%-15% of the length of the
whole spicule. In the shaft an axial thread is distinctly visible (Plate 37, fig. 21).
This thread appears to be quite simple and it certainly does not give oE branches
in the central tyle. From the central tyle, and also from other parts of the shaft,
spines arise. The spines of the central tyle are generally five to eight in number
and arranged in a more or less regular verticil. They are conical, blunt-pointed,
or truncate, 9-14 m long, exceptionally 18 n, and 5-7 m thick at the base; the
truncate ones bear minute secondary spinelets on their ends. The other spines
are irregularly scattered over the middle-part of the shaft. The thickened
conical end-parts of the shaft are free from spines. These spines are not very
numerous; often there are a good many more spines on one side of the central
tyle than on the other. Most of these spines are vertical to the axis of the shaft ;
a few of them are, however, oblique, inclined toward the centre of the spicule.
These scattered spines are similar in shape and about as stout, but, as a rule,
are shorter than the spines of the central tyle. The spines are destitute of
axial threads.

HYAL0NE:\IA (HYALOXEMA) OBTUSUM. 109

The two anchors of the same spicule are equal or slightly unequal in size.
The greatest differences between them in length and breadth observed were 12
and 8 yu respectively. The anchors are 70-100 n, that is generally a little less
than a third of the whole spicule in length, and 70-95 tx broad. The proportion
of length to breadth in these anchors is 100 to 60-108, on an average 100 : 93.2.

Each anchor consists of eight recurved teeth. The teeth of the same anchor
are fairly equal and regularly arranged. Their axes extend in planes passing
through the axis of the shaft and enclosing angles of 45° with their neighbours.
The individual teeth arise nearly vertically from the shaft, and then curve con-
cave towards it. This curvature in the basal part of the tooth is for some
distance fairly uniform, but it decreases distally. Towards its end the tooth is
hardly at all concave to the shaft, or straight, or even slightly convex to the
shaft. The whole curvature amounts to about 90°, the end-parts of the teeth
being nearly parallel. Seen en face (from above) the individual teeth are elon-
gate oval in appearance, 10-16 /i broad in the middle, and rounded or pointed at
the end. Seen in profile they appear stoutest at the base and are at first gradu-
ally, and near the end abruptly, attenuated to a sharp point. Each anchor-
tooth somewhat resembles a T-iron. It consists of an outer band-shaped part,
and an inner keel. The outer band-shaped part is broadest in the middle, at-
tenuated towards both ends, and bent, concave to the shaft, both transversely
and longitudinally. Its transverse convexity increases, and its longitudinal
convexity decreases, towards the distal end. The inner keel is highest at the
base of the tooth; towards its distal end it becomes lower, at first gradually,
then more abruptly.

The large macramphidiscs of var. robusta (Plate 40, figs. 1, 2, 19) are similar
to those of var. gracilis, but larger and provided with somewhat thicker shafts
and broader anchors. They are 235-335 m long, most frequently about 297 fi.
The shaft is 10-17 m thick. Its central tyle measures 15-18 m in diameter. The
anchors are 70-100 ^ in length, that is about a third of the whole spicule, and
90-110 fj. broad. The proportion of the length to the breadth of these anchors
is 100 to 95-150, on an average 100 : 117.6.

The S7nall macramphidiscs of var. gracilis (Plate 34, fig. 19b; Plate 35,
fig. 24b; Plate 37, figs. 12-19; Plate 38, figs. 4-8) are 86-212 m long, most fre-
quently about 164 n. The shaft is 2.5-9 fi thick, and thickened in or near the
middle to a central tyle 4-12 m in diameter. The shaft bears spines similar to
those on the shafts of the large macramphidiscs. These spines are 5-12 ^ long
and 2-3.5 /x thick. The terminal anchors are 32-72 n in length, usually a little

170 HYALONEMA (HYALONEMA) OBTUSUM.

more than a third of the whole spicule, and 16-69 /x broad. The proportion of
the length to the breadth of the anchors is 100 to 56-96, on an average 100 : 81.2.
The individual anchor-teeth are strongly curved in their basal part. Distally
their curvature decreases and their end-parts, a third to a half of their total
length, are nearly straight and parallel.

The small macramphidiscs of var. robusta are similar but stouter and pro-
vided with broader terminal anchors. Their measurements are: — length, 146-
205 fi, most frequently about 176 n; thickness of shaft 4-11 n; anchor-length
45-80 fi, usually a little over a third of the length of the whole spicule; proportion
of anchor-length to anchor-breadth, 100 to 72-112, on an average 100 : 81.2.

The large micramphidiscs of var. gracilis (Plate 37, figs. 6-11) are 28.5-68 /x
long, most frequently about 32.4 fi. The shaft is O.S-1.5 m thick. It always
bears a few spines in or near the middle, and usually some also elsewhere. The
central spines do not form verticils. The spines are 0.4-1 n long, about as thick,
and usually cylindrical and terininally rounded, or truncate.

The two anchors of the same spicule are very similar, the greatest difference
observed in thek lengths and breadths being 2 n and 1 fj. respectively. The
anchors are 11-24 yu in length, that is a little over a third of the whole spicule,
and 6-10 n broad. The proportion of the length to the breadth of the anchors
is 100 to 44-64, on an average 100 : 52.6. The individual anchor-teeth of the
smaller forms of these spicules are strongly cur\-ed in their basal parts and fairly
straight in their distal parts, their total curvature being such that their tips
are nearly parallel (Plate 37, figs. 6, 7). In the larger forms the teeth are
relatively longer, and the curvature of their basal part stronger, whilst their
end-parts are slightly curved in the opposite direction, convex to the shaft.
The tips of the teeth of these amphidiscs are parallel or convergent, and the
anchors themselves at the end sometimes as much as 3 /i narrower than in their
broadest, more proximal part. This gives to these spicules quite a peculiar
appearance (Plate 37, figs. 8-11).

In var. robusta no amphidiscs have been observed similar to the larger forms
of the large micramphidiscs with distally attenuated anchors; all the large
micramphidiscs of var. robusta are similar in shape to the smaller forms of the
large micramphidiscs of var. gracilis. The dimensions of the large micramphi-
discs of var. robusta are: — length 27-64 ^, most frequently about 45.5 ji;
anchor-length 11-25 fi, about two fifths of the length of the whole spicule;
anchor-breadth 7-26 /z; proportion of anchor-length to anchor-breadth 100 to
64-104, on an average 100:84.

HYALONEMA (HYALONEMA) OBTUSUM. 171

The small micramphidiscs of var. gracilis (Plate 35, fig. 17b; Plate 37,
figs. 1-5; Plate 38, figs. 1-3) are 13-26 ju long, most frequently about 16.7 ti.
The shaft is straight and 0.5-1.2 yu thick. It bears in its central part a few short
and broad, terminally rounded protuberances. The terminal anchors are 5-9 ix
in length, that is a third to two fifths of the whole spicule, and 4.2-8 m broad.
The proportion of the length to the breadth of the anchors is 100 to 55-120, on
an average 100 : 90.

The small micramphidiscs of var. robusta (Plate 40, figs. 16-18, 20a) are
similar but somewhat smaller. Their measurements are: — length 12-23 n,
most frequently about 16.2 i^; anchor-length 3.8-7 ju, that is about a third of the
length of the whole spicule ; anchor-breadth 4-5 yu ; proportion of anchor-length
to anchor-breadth 100 to 72-125, on an average 100 : 90.

A few young forms of micramphidiscs were observed in var. robusta. These
spicules (Plate 40, fig. 18) have a centrotyle and spiny shaft thickened at both
ends. Their anchors appear as terminal verticils of small vertically arising and
shghtly recurved teeth.

The two specimens described are in respect to their spiculation similar
enough to be considered the same species. Their skeletal elements, however,
differ in detail, the spicules generally, and particularly both the stout and the
slender tetractine and tetractine-derivate acanthophores, having much stouter
rays. The uncinates are larger and the anchors of both kinds of macramphi-
discs and of the large micramphidiscs are relatively considerably broader in
the specimen from Station 3681 (A. A. 2) than in the specimen from Station
3684 (A. A. 17). For this reason and because the two specimens differ con-
siderably in outer appearance and come from localities a good distance (over
3000 km.) apart, I think it advisable to consider them as two distinct varieties.

The only species of Hyalonema which appears to be at all closely allied to
these sponges is the one described in this report as Hyalonema (H.) agassizi. From
this they differ chiefly by their macramphidiscs and large micramphidiscs having
more strongly curved and less divergent teeth, by their microhexactines being
smaller and having more strongly curved rays, by the spicules of their acantho-
phores being larger, and by the rays of the slender acanthophores having longer
spines.

172 HYALONEMA (HYALONEMA) AGASSIZI.

Hyalonema ^HyaloneIna) agassizi, sp. nov.

Plate 41, figs. 1-U; Plate 42, figs. 1-59; Plate 43, figs. 1-7; Plate 44, figs. 1-30; Plate -45, figs. 1-64;

Plate 46, figs. 1-16; Plate 47, figs. 1-13.

Ele\'en more or less complete specimens and three fragments of this species
were trawled in the Tropical Pacific at five stations. One of these sponges is a
very fine specimen, the best in the collection. It is therefore appropriate to
name this new species after the leader of the several Albatross expeditions which
brought home the material here reported on.

Two of the five specimens from Station 4742 are cake-shaped, the three
others more elongate, pear- or top-shaped. The general appearance and the
spiculation of the fragments indicate that they are parts of similar pear- or top-
shaped sponges. The two cake-shaped specimens from Station 4742 appear to
be identical in structure among themselves, but to differ from all the rest. The
same is to be said of the three pear- or top-shaped specimens and the fragments
from the same station, and of the three specimens from Station 4740. The speci-
mens from the three other stations all differ from each other and from the rest. I
shall, for the reasons given below, describe these six different kinds as distinct
"forms": —

A, the one taken at Station 4656 on 13 November, 1904; 6° 54.6' S., 83° 34.3'

W. ; depth 4063 m. (2222 f.); bottom composed of fine, green mud mixed
with gray ooze; the bottom-temperature was 35.2°.

B, the one taken at Station 4651 on 11 November, 1904; 5° 41.7' S., 82° 59.7' W.;

depth 4063 m. (2222 f.) ; bottom a sticky, fine gray sand; bottom-tempera-
ture 35.4°.

C, the three from Station 4740 taken on 11 February, 1905; 9° 2.1' S., 123° 20.1'

W. ; depth 4429 m. (2422 f .) ; bottom composed of dark gray Globigerina
ooze; bottom-temperature 34.2°.

D, the one taken at Station 3684 (A. A. 17) on 10 September, 1899; 0° 50' N.,

137° 54' W.; depth 4504 m. (2463 f.); bottom light yellow-gray Globi-
gerina ooze.

E, the two caked-shaped specimens, and

F, the three pear- or top-shaped specimens and the fragments, all from Station

4742, on 15 February, 1905; 0° 3.4' N., 117° 15.8' W.; depth 4243 m. (2320
f.); bottom composed of very light, fine Globigerina ooze; bottom-
temperature 34.3°.
Shape and size. The single specimen of form A (Plate 41, fig. 2) is well-

HYALONEMA (HYALONEMA) AGASSIZI. 173

preserved. Its Ijody is broad top- or spindle-shaped, has at every level a nearly
circular transverse section, and measures 66 mm. in length and 48 mm. in maxi-
mum transverse diameter. At its apex the rounded summit of the nearly cylin-
drical gastral cone is visible. The cone is surrounded by a circular wall which
terminates in a narrow frill, the margin of which appears as a circle 8 mm. in
diameter. The circular wall is separated from the cone by a circular fissure
about 1 mm. wide. This fissure is the gastral cavity. The outer surface of the
sponge-body is quite smooth and continuous; apertures, visible to the naked eye,
do not occur in it. From the lower, rounded end of the body the stalk arises.
.\t its origin this is about 5 mm. thick, thickens slightly below, and measures a
little over 40 cm. in length. The spicules composing it are all broken off at the
lower end; in life the stalk was probably considerably longer. Its lower and
central parts are quite straight. Its upper part is strongly and uniformly bent
through an angle of about 60°.

The single specimen of form B (Plate 41, fig. 1) is not so well-preserved.
Of its dermal membrane only a few patches are left and the upper part is much
torn. It is massive, pear- or club-shaped, 81 mm. long and 61 mm. broad. The
stalk is straight, 7 mm. thick at the point of origin and broken off at a distance
of 9 cm. from the sponge-body. Although, as above stated, the upper part of
the sponge is much torn, one can make out in the middle of it a nearly cylindrical,
terminally rounded gastral cone about 10 mm. thick, connected by four radial,
vertical, membraneous plates joining it with the gastral wall. The surface
appears very rough and uneven. This is doubtlessly due to the loss of the dermal
membrane.

The three specimens of form C are cake-shaped. Of their dermal mem-
branes and the stalks only slight remnants remain. The best preserved one
(Plate 41, figs. 13, 14) is a stout, marginally rounded disc, broad-oval, nearly
circular in outline. It measures 66 mm. in length, 60 mm. in breadth, and 27 mm.
in thickness (height). The lower face is nearly flat, the upper convex. The
centre of the latter is occupied by a gastral depression 20 mm. in diameter,
nearly circular in outline, and surrounded by a circular wall on the margin of
which remnants of a thin frill can be made out. Where this frill is best preserved
it appears to be turned outward. A low, dome-shaped, gastral cone about 6 mm.
thick arises from the centre of the depression. This cone is connected with the
gastral wall by four vertical, radial membraneous plates. The wide spaces
between these radial plates appear as diverticular parts of the gastral cavity,
which are continued down into the interior of the sponge. A few stalk-spicules,

174 HYALONEMA (HYALONEMA) AGASSIZI.

broken off 1-3 cm. from the sponge, arise from the centre of the flat lower face.
Many parts of the surface, particularly of the upper side, exhibit a reticulate
appearance, caused by the presence of a superficial network with irregularly
square meshes, the centres of which are about 1 nun. apart (Plate 41, fig. 14).
The centre of each mesh is occupied by the circular entrance to an afferent
canal. The other two specimens of this form are similar. One of them is about
as long and broad as the one above described, but thicker (higher) ; the distance
between the summit of its gastral cone and the slight protuberance on its lower
face, from which the (absent) stalk arose, being 40 mm. The other is smaller,
has no trace of a stalk, and measures 45 by 37 by 12 mm.

The single specimen of form D is also indifferently preserved. A few small
remnants of the dermal membrane and some stalk-spicules, broken off short,
are, however, still present. This sponge (Plate 41, fig. 12) is also cake-shaped.
It measures 51 mm. in length, 46 mm. in breadth, and 17 mm. in thickness
(height) .

The larger and more complete of the two specimens of form E is cake-
shaped, 36 mm. long, 28 mm. broad, and 25 mm. thick (high). In the middle of
its flattened upper face there is a gastral depression, surrounded by a thin circu-
lar wall with sharp margin. This margin is nearly circular and measures 13 mm.
in diameter. In the centre of the depression a low gastral cone is situated, from
which radiate several somewhat irregularly disposed vertical plates. Between
tliese plates wide diverticula of the gastral cavity extend downwards to a dis-
tance of about 14 mm. A protuberance 5 mm. high is observed in the middle
of the lower, more convex face of the sponge. The holes in it indicate that,
in life, the spicules forming the stalk arose from this protuberance. The other
specimen of this form is very similar. It measures 34 by 28 by 21 mm.

The most complete specimen of form F is laterally compressed and appears
as an irregularly triangular plate about 5 mm. thick. The plate is 30 mm. broad
above and narrows below to 5 mm. A bundle of stalk-spicules arises from the
lower end. The other two specimens of this form are more fragmentary and
consist only of the central and the attenuated basal part of the sponge-body,
and the upper part of the stalk. One is (without the stalk) 37 mm. long, the
other 35 mm. The largest of the fragments of this form is 28 mm. long.

The colour of the specimen of form A in spirit is a rather rich coffee-brown,
of form B a dirty light greenish gray, of form C a light reddish brown, of form
D a light dirty brown, and forms E and F are whitish.

Canal-system. The canal-system of form A (Plate 45, figs. 18, 23) seems to

HYALONEMA (HYALONEMA) AGASSIZI. 175

be similar to that of Hyalonema (Hyalonema) oblusum xax. gracilis; the chief
(HfTerence apparently being that the former is more dense and has narrower
subdermal cavities and canals. The flagellate chambers are elongate and 50-80 ^
broad. In one of the specimens of form C the afferent canals are very clearly
visible. They here appear as tubes, about 0.5 mm. wide, which lie parallel side
by side and extend vertically down into the interior of the choanosome. In
this form, and in the forms B, D, and E, the gastral cavity is divided by radial
vertical plates into diverticula. The plates are, in several of these specimens,
four in number and regularly arranged in a cruciate manner. The diverticula
extend downward, are tubular, very wide above, attenuated below, and nearly
circular in transverse section. Their walls are perforated by numerous efferent
apertures, many of which attain considerable dimensions.

Skeleton. The whole of the outer surface of form A, and the (small) parts of
it, in the other forms where the dermal membrane is still present, are covered
by a dense pinule-fur (Plate 42, fig. .36; Plate 45, fig. 23a). Certainly in form A
and probably also in the other forms, the pinules of all parts of the outer surface
are similar, with the exception of the part close to the origin of the stalk. They
are in all forms for the most part pentactines ; a few, however, possess a more
or less developed sixth, proximal ray, and appear as hexactines. Between the
lateral rays of these dermal pinules a few micramphidiscs lie scattered on the
outer surface. From the thin, upper, free margin of the wall surrounding the
gastral cavity (fissure or depression) the distal rays of diactine pinules protrude.
The gastral surface, that is the inner surface of the wall surrounding the gastral
cavity, and the surface of the gasti-al cone are likewise covered by a pinule-fur.
The pinules composing it are chiefly pentactines, more rarely hexactines, excep-
tionally diactines. A few minute spiny pentactines have also been observed
here. On these gastral surfaces also micramphidiscs occur. These spicules are
here, however, much more abundant than on the outer surface, and in places
form dense masses. Below, where the gastral cavity passes into the large effer-
ent canal-stems, the pinule-fur ends; the coating of micramphidiscs, however, is
continued along the walls of these canals quite down to the innermost parts of the
choanosome. The micramphidiscs of the outer, dermal surface and of the
surfaces bordering on the upper part of the gastral cavity and enclosing the inner,
proximal parts of the efferent canals, are all or nearly all small ones. Those on
the surfaces surrounding the lower proximal part of the gastral cavity and the
mouths of the large efferent canals on the other hand are, certainly in form A,
and probably also in the other forms, in great part large macramphidiscs.

17G HYALONEMA (HYALONEMA) -AGASSIZI.

Just below the level in which the lateral rays of the dermal pinules of the
outer surf?<;e extend, the paratangentially situated lateral rays of hypodermal
pentaetines are met (Plate 42, fig. 37a; Plate 45, fig. 23). In form C these
lateral pentactine rays extend in the beams of the superficial network above
described. The apical rays of the hypodermal pentaetines extend radially in-
ward. In form A a superficial zone about 0.6 mm. thick, underlying the dermal
membrane, is occupied by dense masses of more or less radially arranged unci-
nates and irregularly scattered microhexactines and microhexactine-derivates
(Plate 45, fig. 23). This zone contains no spicules besides these and the proxi-
mal rays of the hypodermal pentaetines, which traverse it. Below this zone
hexactine megascleres begin to make their appearance. Those lying nearest
the surface are quite small, towards the interior they increase in size. Though
often irregularly disposed in the sections, these spicules are, in the living sponge,
in all probability regularly arranged in such a manner that two opposite rays
extend longitudinally upward and downward, two radially outward and inward,
and two paratangentially and laterally to the right and left. In most of the large
and in a good many of the smaller hexactines the two opposite longitudinally
extending rays are longer than the other four. Masses of large macramphidiscs
are met with a little below the level where the hexactines begin to make their
appearance. In some places these form but a thin layer, in others they extend
a considerable distance, 2 mm. or more, into the interior of the choanosome.

The inner parts of the sponge are occupied by the large hexactine mega-
scleres referred to above, and also by rhabd-megascleres, uncinates, microhexac-
tines, microhexactine-derivates, amphidiscs, and spheres.

The large inner hexactines usually have two opposite, longitudinally extend-
ing, greatly elongated rays and four shorter transverse rays. The rhabds of the
axial part of the sponge are situated longitudinally and form a kind of axial
column, which extends upwards to the summit of the gastral cone. Loose
strands of rhabds diverge from this axial column and extend upwards and out-
ward. Below, in the interior of the choanosome, these diverging rhabd-strands
dissolve into scattered, obliquely situated, isolated rhabds; above they join to
form distinct layers lying below the dermal and the gastral surfaces of the thin
frill-like marginal part of the gastral wall. In the forms B, C, and D masses of
longitudinal rhabds also occupy the vertical radial plates connecting the gastral
cone with the gastral wall. Most of these rhabds are very blunt amphioxes or
amphistrongyles ; but sharp-pointed amphioxes, amphityles, styles, and tylo-
styles also occur among them. In the axial column of form A both large and

HYALONEMA (HYALONEMA) AGASSIZI. 177

small rhabds are met. Outside the axial column, however, only the smaller
ones have been observed.

In the interior the uncinates are not very numerous and are irregularly
scattered. Of amphidiscs both macramphidiscs and micramphidiscs occur in
the interior. The former are very scarce, the latter, which appear chiefly to
occupy the walls of the efferent canals, exceedingly numerous. The micro-
hexactines and their derivates are, in the interior, rather frequent, but not nearly
so abundant as in the superficial region. The spheres appear to be restricted
to the axial column, where they occur singly or, more rarely, in clusters. They
are rather numerous in form A and have also been found in form D.

In the specimens of form A, B, and in two of the specimens of form F the
stalk is more or less intact. It is in these forms composed of stouter and more
slender rods, broken off at the lower, distal end. In the specimen of form A
the stalk is over 40 cm. long and now consists of twenty spicules; in life there
may have been more. The spicules composing it are very distinctly spirally
twisted, like the strands of a rope and also similarly entwined. The twist has
the same direction in all. Progressing from the proximal to the distal end the
spiral curvature is in the direction of the movement of the hands of a watch.

The stalk-spicules extend for some distance upwards into the sponge-body,
and they are, in the basal part of the latter, surrounded by masses of acantho-
phores. These are stout-rayed, usually terminally spined tetractines (staurac-
tines), derivates of these spicules, more or less spiny pentactines, modified
pinules, and modified rhabds with spiny ends. In the basal part of some of the
specimens spheres also occur. In the specimens of all the forms with the excep-
tion of those of forms E and F, slender-rayed, long-spined spicules with four to
six rays also occur just below the surface of that part of the body from which
the stalk arises. Their absence in forms E and F is probably due to the frag-
mentary nature of the specimens of these forms.

The dermal pinules (Plate 42, figs. 20-23, 25-30, 37b, 42b) are nearly all
pentactine; only a few are hexactine. Their distal ray, in form A (Plate 42,
figs. 25-28, 35, 36), is straight, 93-110 m long, usually 94-107 m, on an average
100.4 fi; and, at the base, 4.4-7 m thick, usually 4.5-6.7 m- Above it thickens,
and it ends with a well-developed, smooth, terminal cone 6.5-11 n thick, usually
8.5-9 n. The proximal, basal part of the ray and its terminal cone are free from
spines; the rest of it, usually about 60% of its length, is covered with spines.
The most proximal spines diverge strongly, and are often nearly vertical to the
ray. Distally they become more inclined towards the tip, and the uppermost

178 HYALONEMA (HYALONEMA) AGASSIZI.

spines, which surround the terminal cone, are nearly parallel to the ray-axis.
The lowest spines are straight and quite short. Distally they become slightly
curved, concave towards the tip of the ray. Up to the middle of the length of
the ray they increase in size; beyond they again become smaller. The largest
spines on the middle-part of the ray are 10-20 n long and 2-4 n thick at the base.
The maximum transverse diameter of the distal ray, together with its spines, is
18-32 fi, usually 20-30 n, on an average 23.6 m- The lateral rays of the same
spicule are usually fairly equal (Plate 42, fig. 35), sometimes considerably unequal
(Plate 42, fig. 36). They are 21-32 ^ long, straight, nearly cylindrical in their
basal part, attenuated toward the end in their distal part, and blunt-pointed or
terminally rounded. The proximal parts of the lateral rays are usually rather
smooth; their distal parts bear sparse small spines. A sixth proximal ray is
observed very rarely and, when present, is short and rudimentary.

The dermal pinules of form B (Plate 42, fig. 29) are very similar but have
a shorter and more bushy distal rays and longer lateral rays. The distal ray
is 85-97 M long, on an average 93 fi, and 3.5-6 ^ thick at the base. Its maximum
transverse diameter, together with the spines, is 28-34 ai, on an average 29.2 fi.
The lateral rays are 25-30 n long.

The dermal pinules of form C (Plate 42, figs. 20-23) are even more similar
to those of form A, but have a slightly more bushy distal ray and longer lateral
rays. The distal ray is 95-114 fi long, on an average 104.9 /i, and 4-6 ^ thick
at the base. Its maximum transverse diameter, together with the spines, is
22-32 fi, on an average 27 n. The lateral rays are 25-35 fi long.

The dermal pinules of form D (Plate 42, figs. 30-34, 42) differ from those of
the other forms by the distal ray being not so long, having a shorter and stouter
terminal cone, and being covered with more numerous and crowded spines.
The distal rays of the dermal pinules of this form therefore appear, when com-
pared with those of the other forms, more stunted, stout, and dense. The distal
ray is 82-101 fi long, usually 87-95 fi, on an average 89.1 yu, and at the base
4.5-8 M thick, usually 4.5-6 fi. Its maximum transverse diameter, together with
the spines, is 23-32 yu, on an average 26.2 /j. The terminal cone is 8-1 1 n thick.
The lateral rays are 17-30 fi long, exceptionally up to 38 fi.

The dermal pinules of form E have a distal ray 70-94 fi long, on an a^'erage
86.7 fi. Its maximum thickness, together with the spines, is 22-32 fi. The
lateral rays are usually 18-23 fi long.

The dermal pinules of form F ha^'e a distal ray 85-97 n long, on an average
90 fi. Its maximum thickness, together with the spines, is 23-25 fi. The lateral
rays are usually 23-43 m long.

HYALONEMA (HYALONEMA) AGASSIZI. 179

Peculiar, very variable, modified pinnies (Plate 42, figs. 38-41, 43-45, 47,
48) occur in the basal region, where the large stalk-spicules enter the sponge-
body. These spicules appear to be dermal pinules changed in shape and in part
pushed into the interior by the stresses arising in this region from the resistance
of the embedded upper ends of the stalk-spicules to the weight of the sponge-body,
and to such passive movements of it as may be caused by the impact of moving
deep-sea animals. Transitional forms appear to connect these modified pinules
with the slender-rayed, long-spined basal tetractines and other spicules described
below.

The modified basal pinules of form A (Plate 42, figs. 38-41) are pentactine
or hexactine. The distal ray is straight, 78-120 n long, and 2.5-6 fi thick at the
base. It bears, in its middle- and end-parts, sparse, strongly divergent spines,
which are sometimes irregularly distributed, and are more numerous on one side
than on the other (Plate 42, fig. 40). These spines are slender, conic, pointed,
straight or slightly curved, simply or in an S-shaped manner; they are 15-29 n
long and 1.3-3 n thick at the base. The maximum transverse diameter of the
distal ray, together with the spines, is 18-42 ii. The lateral rays of the same
spicule are equal or unequal. They are straight, 30-57 n long, conic, usually
pointed, and covered with spines directed obliquely outward. The largest
of these spines are 1.5-2 /i long. The proximal ray (of the hexactine forms)
is 42-55 M long; in shape and spinulation it resembles the lateral rays.

The modified basal pinules of form B (Plate 42, figs. 47, 48) are similar but
have shorter and stouter rays. The distal ray is 83-94 ai long and 4-6 m thick
at the base. Its maximum diameter, together with the spines, is 28-35 /i.
The spines of the distal ray are 18-24 /x long and 2.5-2.7 m thick at the base.
The lateral rays are 25-32 fj. long.

Some of the modified basal pinules of form C (Plate 42, figs. 43, 44) are
considerably larger than those of the other forms. The distal ray is 101-135 ^
long and 4.5-5 m thick at the base. Its maximum transverse diameter, together
with the spines, is 13-33 fx. Its spines are 14-15 yu long and about 2.3 n tliick
at the base. The lateral rays are 20-80 fi long.

In form D, E, and F, I found only few modified l)asal pinules (Plate 42, fig.
45). One of form D which I measured had a distal ray 82 ^ in length and,
together with the spines, 28 n in maximum thickness. Its largest spines meas-
ured 18 by 2.5 M.

The frill on the margin of the gastral wall, containing the diadine -pinules,
is preserved in a satisfactory manner only in the specimen of form A. The
diactine marginal pinules of this form (Plate 44, figs. 1-5) have a total length

180 HYALONEMA (HYALONEMA) AGASSIZI.

of 350-640 IX. The distal ray is 148-245 m long, fairly straight, 6-9 m thick at
the base, and thickened above. It ends with a smooth, rather slender, sharp-
pointed terminal cone. All parts of it, with the exception of its basal portion
and its terminal cone, are covered with spines strongly inclined towards the tip.
The largest spines are situated about a third of the length of the distal ray from
the tip. From here they decrease in size both distally and proximally. The
largest spines are 6-7 n long and 2-3 yu thick at the base. The maximum trans-
verse diameter of the distal ray, together with the spines, is 15-26 n. The lateral
rays are generally reduced to mere rounded protuberances, only exceptionally as
much as 9 M high (Plate 44, fig. 2, the left one). Together they form a central
tyle 11-21 fx in diameter. The proximal ray is straight or slightly curved,
175-400 M long, and, at the base, as thick as the chstal ray. It usually bears a
few spines and a number of very low and broad rounded protuberances which
render the appearance of its outline somewhat wavy.

I have observed a few transitional forms which appear to connect these
diactine pinules with ordinary, centrotyle, amphiox megascleres. The ray corre-
sponding to the distal ray of the diactine pinules of one of these spicules, which
I measured, was perfectly smooth, 680 n long, and 22 n thick at the base, and
thickened above the middle of its length to 26 fi. Its central tyle measured
30 ;u in transverse diameter.

The gastral pimdes (Plate 42, figs. 1-8, 10-19, 24). In form A, where
the gastral pinules lioth on the cone and on the inner face of the gastral wall
could be conveniently measured, I found the distal rays of the former markedly
longer than the distal rays of the latter, and also noticed that the distal rays of
the pinules of the gastral wall decreased in length towards the upper, free margin.

The gastral pinules of the cone of form A (Plate 42, figs. 1-8, 10-13) are for
the most part pentactine; a few, however, are hexactine (Plate 42, figs. 1, 2)
and one that I observed was diactine (Plate 42, fig. 13). The distal ray in these
pinules, when normally developed, is 97-135 n long, usually 100-134 n, on an
average 118.2 fx, and 3.5-9.5 m thick at the base. One (Plate 42, fig. 8) that
had apparently been broken off during growth and then partly regenerated
was only 65 n long. The distal ray-ends with a smooth, l)lunt, terminal cone
4. .5-9 IX thick. This and the basal part of the ray are destitute of spines. The
remaining parts of it bear somewhat sparse spines. The proximal spines are
strongly divergent, only sUghtly inclined, and curved towards the tip of the ray.
Distally they become more inclined in this direction, but are, on the whole, much
more divergent than those of the dermal pinules. The spines attain their great-

HYALONEMA (HYALONEMA) AGASSIZI. 181

est length near the middle of the ray and from here decrease in size both proxi-
mally and distally. The largest spines are 10-21 n long and 2.5 fi thick at the
base. The maximum transverse diameter of the distal ray, together with the
spines, is 21-41 ju, usually 30-40 m, on an average 34.1 ti. The lateral rays
of the pentactine and hexactine gastral cone-pinules are, in the same spicule,
fairly equal (Plate 42, figs. 1, 10, 11) or more or less unequal (Plate 42, figs. 3, 5,
12), some of the lateral rays of the same spicule often being short, cylindrical,
and terminally rounded, the others long, conic, and pointed. The individual
lateral rays are 20-85 n long, straight, cylindrical, and terminally rounded, or
conic and pointed at the end. Their basal part is usually quite smooth, their
distal part for two thirds or more of their length is covered with more or less
conspicuous spines. The proximal ray in the hexactine forms is 17-76 n long,
gradually attenuated towards the end, or abruptly pointed. It is smooth
throughout, or covered with spines in its distal part. The proximal ray of the
diactine cone-pinule (Plate 42, fig. 13) is 95 n long.

The gastral pinnies of the inner surface of the gastral wall of form A are
similar to those of the cone, but have distal rays only 91-112 /x long.

The gastral cone-pinules of form B (Plate 42, figs. 14, 15) are pentactine or,
more rarely, hexactine and similar to those of form A. They have, however,
shorter distal rays with more divergent and longer spines. The distal ray is
in these spicules 83-109 ^ long, on an average 98.2 fi, and 5-7 fx thick at the base.
Its maximum transverse diameter, together with the spines, is 27-45 yu, on an
average 34.4 ix. Its largest spines attain a length of 25 m- The lateral rays are
30-48 fi long, the proximal ray (of the hexactine forms) is 12-27 n long.

The gastral cone-pinules of form C (Plate 42, figs. 16, 17) have a longer
distal ray than those of form B and resemble those of form A very closely.
The distal ray in these spicules is 110-127 n long, on an average 118.5 m, and
4.5-5.5 M thick at the base. Its maximum transverse diameter, together with
the spines, is 24-29 /x. The lateral rays are 35-50 m long.

The distal rays of the gastral pinules of the inner surface of the gastral wall
of form C (Plate 42, figs. 18, 19) are similar in size but have more divergent
spines and consequently, together with the spines, a greater maximum trans-
verse diameter. The divergence of the proximal spines from the tip of the ray
is so great that some of them stand vertical, and some even point the opposite
way. The distal ray is 103-126 m long, on an average 115 fx, and 5-6 /x thick at
the base. Its maximum transverse diameter, together with the spines, is 28-35 ^l.
The lateral rays are 36-50 ix long.

182 HYALONEMA (HYALONEMA) AGASSIZI.

The gastral pinnies of the cone of form D (Plate 42, fig. 24) resemble those
of form A rather closely. They have a distal ray 98-113 ju long, on an average
106.7 M, and 6.5-8.5 m thick at the base. Its maximum transverse diameter,
together with the spines, is 25-38 fi. The lateral rays are 26-64 /^ ; the proximal
ray (of the hexactine forms) is 38-50 /i long.

The gastral pinules of form E have a distal ray 69-103 m long, usually 83-
99 n, on an average 90.5 m, and 4.5-7 ix thick at the base. Its maximum thickness,
together with the spines, is 21-37 m- The lateral rays are 21-57 /^ long. In two
hexactine gastral pinules of this form measured, the proximal ray was 42 and 44 yu
long respectively.

The gastral pinules of form F have a distal ray 100-153 n long, on an average
129 n, and 5-6 ^ thick at the base. Its maximum thickness, together with the
spines, is 25-40 jj.. The lateral rays are 50-62 /x long.

Minute pentactines with spiny rays (Plate 42, figs. 9, 50) were found in small
numbers in the spicule-preparation of the gastral cone of form A, and the basal
part of forms A and B. These pentactines have straight, conic, blunt-pointed
rays, smooth at the base, but covered with conspicuous spines in their distal part.
Their apical ray in form A is 36-80 fi long, in form B 43-65 m, and is in both
3-6 n thick at the base. The lateral rays of the same spicule are equal. In
form A they are 40-50 ix long, in form B 25-42 n, and about as thick as the apical
ray.

The hypodermal pentactines of form A (Plate 41, figs. 3-11 ; Plate 45, fig. 23)
have a fairly straight, conic, and Islunt apical (proximal) ray, which measures
0.3-1.5 mm. in length, and 18-90 ^ in thickness at the base. The lateral rays of
the same spicule are fairly equal or more or less unequal. Among the small
hypodermal pentactines forms with equal lateral rays predominate, but among
the large ones forms with unequal lateral rays are the more numerous. The
lateral rays are more or less oblique and enclose angles of 80°-88° with the apical
(proximal) ray. They are usually somewhat curved, conic, and rounded at the
end. The longest lateral ray is 0.25-1.3 mm. long. The ends of the lateral
rays (Plate 41, fig. 9) are, as in Hyalonema obtusum, usually irregular, and proba-
bly for the same reason (c/. p. 160).

In the forms B, C, and D, in which the greater part of the dermal membrane
is lost, only few hypodermal pentactines were found. All those observed in
forms C and D were similar to those of form A. In form B spiny pentactines
of similar dimensions were found, in addition to the ordinary smooth ones of the
other forms. For the reasons given below (p. 183) I consider these spiny pentac-
tines as foreign spicules.

HYALONEMA (HYALONEMA) AGASSIZI. 183

The hexactine megascleres of form A (Plate 45, figs. 6-13) measured were
0.4-6 mm. in maximum diameter. The rays of the same spicule are in the smaller
ones either equal or unequal, in the largest ones always unequal in length, two
opposite ones being in these much longer than the other four. The four shorter
rays are often also unequal among themselves. The rays arise from a central
thickening 30-90 ^ in diameter, are smooth, conic, 10-58 m thick at the base,
and blunt or rounded at the end. They are in the small hexactines straight,
in the large usually slightly curx-ed. The longest ray is 220 ;u-3.2 mm. long.

The hexactines of forms C and D are similar. In form B I found, besides
hexactines similar to those of form A, one 11 mm. in maximum diameter with
rays 70 /x thick at the base, and only slightly attenuated to the rounded ends.
In this form also spined hexactines, 2-5.5 mm. in diameter, occur. Although
these are quite numerous and found in the depth of the choanosome, I do not
believe that they really belong to the sponge. They are, Uke the large spined
pentactines referred to above, identical with the spined hexactine and pentac-
tine megascleres of Calycosilva cantharellus (Plate 1, figs. 5-24; Plate 6, figs. 1-12),
a large number of specimens of which were trawled at the same station. Some
of the spined hexactines and pentactines of these sponges may therefore have got
accidentally into the sponge.

In the basal part of the body, from which the stalk arises, slender acantho-
phores, usually with four, more rarely with five or six rays (Plate 42, figs. 49, 51-
59), are met in all the forms except E and F. In form A these spicules (Plate 42,
figs. 49, 51, 52) are 95-170, usually 110-135 m in diameter, and generally consist
of four rays lying in the same plane and enclosing angles of 90° with their neigh-
bours. Sometimes a fifth ray, vertical to the other four, is present. The rays
of these spicules are fairly straight, at the base 2.5-^ m thick, rarely 5 m, conic, and
sharp-pointed. They bear numerous slender oblique spines inclined towards
the tip of the ray. The largest spines are 4-12 fi long.

In form B these spicules (Plate 42, figs. 53, 54) are similar, measure 85-150 n
in diameter, and have rays 2-4.5 m thick at the base. Here only tetractines were
observed.

In form C these spicules (Plate 42, figs. 55, 56, 58) are larger, 120-210 ^ in
diameter, and have four or, more rarely, five fairly straight or considerably
curved rays, 3.5-5 m thick at the base.

In form D some of these spicules (Plate 42, figs. 46, 57, 59) attain a still
larger size. They measure here 100-230 m in diameter and have usually five,
more rarely four or six rays 2.8-6 ^ thick at the base.

Transitional forms were found quite frequently in the basal part of the

184 HYALONEMA (HYALONEMA) AGASSIZI.

sponge-body, particularly in form C, apparently connecting these slender-rayed
acantliophores with the modified basal pinules described above on the one hand,
and the stout acanthophores described below, on the other hand.

The stout acanthophores surrounding the proximal end-parts of the large
stalk-spicules (Plate 45, figs. 1-4, 14-17, 24, 25, 35-39) are mostly tetractines
and cUactine tetractine-derivates. However, similar pentactines, triactines, and,
exceptionally, also monactines occur among them. Occasionally one meets
with tetractines and pentactines of this kind with all the rays greatly reduced
in length. These spicules appear as transitions, leading to the spheres described
below.

The rays of the same spicule are always more or less, and sometimes very
unequal. They generally join at angles of about 90° or 180°, and are straight
or curved, and cylindrical and terminally rounded, or conic and either blunt or
pointed at the end. The diactine ones are either centrotyle or simply cylindrical
in the centre, straight, slightly angularly bent in the middle, or, rarely, strongly
curved. One 11 m thick, which I observed in form A, formed a complete ring
65 M in diameter. Sometimes the rays bear rudiments of branch-rays. The
basal parts of the rays are usually smooth or only sparsely spined ; their end-parts
bear numerous, rather large, generally nearly vertical spines, which stand close
together. The smooth proximal part is usually a little longer than the spined
distal part.

In form A the larger, normal acanthophores are 200-690 ix in maximum
diameter and have rays 14-40 ^ thick at the base. The small ones transitional
to the spheres (Plate 45, figs. 24, 25, 38) are 46-115 in diameter and have rays
9-14 M thick.

In the other forms these spicules appear to be similar. Form D possesses
mon- to pentactine spicules of this kind 195-550 n in maximum diameter with
rays 15-35 /x thick. The monactines are very rare. One that I measured was
195 M long, and at the rounded, somewhat thickened end, 12 m in transverse
diameter.

In the preparations of one of the specimens of form F the stout acanthophores
are particularly numerous. The triactine and tetractine forms here measure
120-640 M in diameter, usually 420-590 m, and have rays 20-40 fi thick at the
base. The diactine forms are usually fairly straight, rarely strongly angularly
bent in the middle so that the two rays enclose an angle of 90° or less. The
fairly straight diactines are 120-550 fj. long. Their rays have the same thickness
as those of the triactines and tetractines.

HYALONEMA (HYALONEMA) AGASSIZI. 185

In the spicule-preparations of the basal part of this form numerous small,
hollow, cross-Uke siUceous bodies were observed. The smallest of these are about
20 fi in diameter, and consist of four somewhat conic rays, 10 m long, about IG m
thick at the base, and hoUowed out by cylindrical axial canals about 8 ^ wide.
These smallest crosses are connected with the large normal stout-rayed tetractines
above described by an uninterrupted series of spicules intermediate in size.
The axial canals of these spicules are usually 5-9 m wide. The axes of the rays
of the full-sized, stout-rayed basal spicules are occupied either by a fine axial
thread, or a more or less widened axial canal. The broadest axial canals in these
spicules were about 9 m in diameter. In cylindrical, terminally rounded rays
these axial canals are closed at the end ; in conical and pointed rays they usually
open out freely.

The wide axial canals are regular or irregular. The regular ones are either
cylindrical throughout or widened distally. Distal widenings occur both in the
terminally open and in the terminally closed axial canals. The irregular ones
are of two kinds. In some the axial canals bear short, irregular, branch-Uke
diverticula, which usually arise near the distal end, and are vertical or oblique,
directed outward or, more rarely, inward. Others possess backwardly directed
diverticula, which arise from their basal part and occupy interstices between
adjacent silica-layers.

It is obvious that the small forms of this series are to be considered either
as the yoimg of the full-sized ones, or as the last remnants of full-sized ones which
have in great part been dissolved. The general appearance of the whole series
seems to me to be in favour of the latter alternative. I accordingly assume that
the stout acanthophores with wide axial canals are spicules in process of decay
(solution), that this decay or solution is the further advanced the smaller the
spicules are, and that the dissolving agency acts on the silica-layer both from
the inner (the axial threads) and the outer side (the surface). No doubt the sea-
water can and does dissolve the silica of the spicule in this way when the protect-
ing organic or semiorganic sheath is lost, but it must not be overlooked that the
living sponge-tissue of the sponge itself might possibly also attack and dissolve
the silica in spicules which have become superfluous, and use the material thus
obtained for building up other spicules.

The spheres of form A (Plate 45, figs. 26-34) are irregularly nodular or
spherical and measure 18-57 /u in diameter. Most of them are smooth (Plate 45,
figs. 26-29, 33, 34), some more or less spiny (Plate 45, figs. 30-32). They consist
of concentric layers of silica. The centre around which these silica-layers are

186 HYALONEMA (HYALONEMA) AGASSIZI.

deposited may be a simple point, a short rod, or a cross. The spheres with a
cross-shaped centre (Plate 45, fig. 29) lead to those short-rayed tetractines
(Plate 45, figs. 24, 25) which have been referred to above as transitions between
the normal long-rayed, stout, basal spicules and the spheres; I am inclined to
consider the spheres as derivates of these spicules.

I have not seen any spheres in the preparations of form B and C, but I
found some, similar to those of form A, in form D.

The microhexactines and their derivates form a series commencing with
regular equal-rayed hexactines and ending with diactines and monactines. They
fall into two groups: — 1, regular and irregular microhexactines proper, and
2, diactine and monactine microhexactine-derivates.

The microhexactines -proper (Plate 44, figs. 15, 16, 17b, 18-23, 25-30) have
regularly disposed rays which enclose angles of 90° with their neighbours. The
rays are conic and pointed. Their basal part is straight, their distal part nearly
straight or curved more or less, sometimes considerably. In the forms E and F,
where the microhexactines with the most strongly curved rays are found inter-
mingled with the other, more straight-rayed forms, the degree of curvature
appears to be in inverse proportion to the size of the spicule. The end-parts
of opposite rays are usually curved in opposite directions. The rays of these
spicules are beset with small backwardly directed spines. These are largest
and most numerous on the middle-part of the ray; proximally they decrease
in number, distally in size. It is also to be noted that these spines are on the
whole much larger in the large (and straight-rayed) than in the small (and more
curved-rayed) microhexactines.

The microhexactines proper of form A (Plate 44, figs. 16, 17b, c, 18-20, 22,
23, 30) are 50-144 ^ in diameter and have rays 1.7-4 m thick at the base. The
irregular forms are larger (longer) and have thicker rays than the regular. The
difference in the length of the rays of the irregular forms is sometimes very con-
siderable, the length of the shortest ray being occasionally only a ninth of that of
the longest.

In the other forms the microhexactines proper are similar and also in these
the irregular ones are larger than the regular. The maximum diameter of the
microhexactines proper measured was in form B (Plate 44, figs. 21, 25, 27)
66-145 M, in form C (Plate 44, fig. 26) 44-130 m, in form D (Plate 44, figs. 28, 29)
48-114 n, in form E 53-157 n, and in form F 52-160 n.

The diactine and monactine microhexactine-derivates are by no means fre-
quent. The diactine microhexactine-derivates of form A (Plate 44, fig. 24) are

HYALONEMA (HYALONEMA) AGASSIZI. 187

more or less centrotyle spiny rods, pointed at both ends. Their middle-part is
straight, their end-parts are slightly curved. These spicules are 84-240 fi long
and 3.6-5 n thick near the centre. The central tyle measures 5.8-15 n in trans-
verse diameter and, when large, clearly shows that it is composed of four ray-rudi-
ments.

The monactine microhexactine-derivates are very rare. One of form A
which I measured was 75 m long, and 3.5 ^ thick at the rounded end.

Apart from the diactine pinules and the diactine and monactine derivates
of the stout-rayed basal tetractines and the microhexactines above described,
three kinds of rhabds can be distinguished : — ordinary rhabds of the choanosome
and axial column, modified rhabds of the basal part of the sponge-body, and
uncinates.

The ordinary rhabds of the choanosome and axial column (Plate 45, figs. 19-
22) are smooth, slightly curved or, rarely, angularly bent, 0.3-7 mm. and more
long, as some long fragments observed indicate, and 7-50 ix thick, rarely 95 n.
Most of them are blunt amphioxes or amphistrongyles, but amphityles, styles,
and tylostyles also occur among them. The smaller amphioxes, amphistrongyles,
and amphityles are generally distinctly centrotyle. Remarkably regular cylin-
drical amphityles occur in the marginal part of the gastral wall of form A. These
are mostly 0.7-1 mm. long and 26-30 m thick; their spherical terminal tyles
measure 40-50 ^ iii diameter. A short somewhat spindle-shaped style 75 m
thick at the rounded end and 95 m at the stoutest point was observed in the axial
column of this form. The axial canals (threads) of the small rhabds are usually
quite fine, those of the large ones on the other hand generally wide, sometimes
5 M or more in transverse diameter.

The modified basal acanthophore rhabds (Plate 45, fig. 5) are centrotyle, usu-
ally slightly curved, smooth in the middle, and strongly spined at the ends, which
are generally somewhat thickened. The terminal thickenings are either spheri-
cal or spindle-shaped, and in the same may be either similar or dissimilar, one
end-thickening frequently being spherical, the other spindle-shaped. These
spicules are 0.8-2.6 mm. long, and 7-22 ju thick near the centre. The central
tyle is « 12.5-27/:* in transverse diameter, the terminal thickenings 12-30 m-
The spherical terminal thickenings are stouter than the spindle-shaped. The
spiny end-parts are 80-260 m long. The axial canals (threads) of these spicules
are often very wide. They are usually closed in the rounded ends and open in
the spindle-shaped.

The uncinates are mostly diactine, but monactines also occur.

188 HYALONEMA (HYALONEMA) AGASSIZI.

The diaciine uncinates (Plate 44, figs. 6-14, 17a) are generally straight oi'
slightly curved, simple amphioxes; considerably curved and centrotyle ones,
however, also occur. The ordinary amphiox uncinates in form A are 330-800 ^
long, 5.5-12 n thick in the middle, and beset with splines. As far as I could make
out these spines are 0.7-1.5 m long, and about 1 n thick at the base. Sometimes
it appeared as if they were continued in a fine terminal filament which was,
however, too thin to be distinctly projected even with the 280 n light. At one
end of the spicule these spines are numerous, rather close together, and strongly
inclined toward the opposite end. Toward the other end they become much
scarcer and less inclined. Some of the spines nearest the latter end are vertical
or even inclined in the opposite direction. In the centrotyle uncinates the cen-
tral tyle is 15% to 45% thicker than the adjacent parts of the spicule.

The monactine uncinates appear as tylostyles. In form A they are 260-
293 fi long and 9-12 n thick just below the rounded end. The rounded end itself
is thickened to a more or less spherical tyle 14-16 n in transverse diameter.

The large stalk-spicules of form A (Plate 41, fig. 2; Plate 43, figs. 1-7) have
a maximum length of 42 cm. and all are broken off at the lower distal end. Wliere
they arise from the sponge they are 0.05-0.95 mm. thick; 30 cm. lower, where
most of them are stoutest, they are 0.5-1.3 mm. thick.

One (Plate 43, fig. 1), which I studied in detail, is 160 m thick at the upper
end, and rapidly increases in thickness to 730 m at 7 cm. from the end; it then
gradually thickens down to 28 cm., where it attains its maximum thickness
of 1050 n. Farther on it again becomes thinner, and at the lower end, 42 cm.
from the tip, is 760 ix thick. Its axial thread is for the most part thin. It is
thickened, however, here and there in an irregular manner. The silica is very
clearly stratified. The surface of the upper, proximal part of the spicule is quite
smooth. Where the spicule attains its maximum thickness fine transverse lines
(Plate 43, fig. 7) make their appearance on its surface, and 1 cm. above the distal
end its surface, for a short distance, has quite a peculiar structure (Plate 43,
fig. 4). Here a silica-layer is exposed which consists of lamellae overlapping
like tyles, and composed of parallel rods about 10 yu thick and lying close to-
gether. These rods extend nearly but not quite paratangentially and longitudi-
nally. They deviate slightly both radially and laterally from the direction of the
axis of the spicule. The radial deviation is due to their forming the overlapping
lamellae, and like the lamellae themselves they slightly diverge from the axis
below. The lateral deviation is due to their lying somewhat obliquely in elon-
gated spirals.

HYALONEMA (HYALONEMA) AGASSIZI. 189

Most of the other stalk-spicules exhibit, in their lower portion, the same
transverse lines as the one described above, and in six of them the same spiral
rods, combined to form tyle-like lamellae, are visible on portions of the surface
near the end.

The transverse lines may be considered as fissures in the superficial silica-
layer. In the six spicules where it was observed, (and probably also in the
others) the portions showing the superposed rows of spiral rods indicate that
there are one or more silica-layers (composed of thin, spirally extending
rods) quite different in structure from the rest. These layers are rendered
visible where the disintegration (solution) of the spicule (which proceeds
from the surface downwards) has just reached them; and their structure is
probably brought out so clearly by the silica joining the rods having been partly
dissolved.

Being composed of layers differing in structure, one or more of which consist
of superimposed rows of spirally arranged rods or threads, the stalk-spicules
may, in respect to their internal structure, be compared to cables.

No traces of backwardly directed spines or of terminal anchors could be
found in any of the spicules.

The amphidiscs (Plate 44, fig. 17d; Plate 45, figs. 40-64; Plate 46, figs. 1-16;
Plate 47, figs. 1-13). The biological length frequency-curve of the amphidiscs
of Hyalonema agassizi, form A, shows (Fig. 5), that, as regards the frequency of
those of different length, these spicules fall, like those of Hyalonema obtusuni,
into four groups: — large macramphidiscs, small macramphidiscs, large micram-
phidiscs, and small micramphidiscs. The second and third of these groups are,
in respect to their length frequency, not as clearly distinguished from each other
as from the first and fourth respectively. The parts of the curve pertaining to
the large macramphidiscs and the small micramphidiscs each ha\'e two culmi-
nations, a principal, and a secondary. The measurements and examination of
the amphidiscs of various length of the three other forms show that these also
fall into the four groups mentioned, and that, at least in two of them (B and C) ,
the gap between the small macramphidiscs and large micramphidiscs is not so
distinct as in the others. In the forms B and C these two kinds of amjihidiscs,
which can be readily distinguished by differences in their shape, slightly overlap
in respect to their length.

The large macramphidiscs of form A (Plate 46, figs. 2-5, 9, 12, 13; Plate 47,
figs. 1, 2, 5, 6, 10) are 134-242 ^ long, most frequently about 200 m- The shaft
is straight, cylindrical, 7-13.5// thick, and thickened abruptly at some point

190

HYALONEMA (HYALONEMA) AGASSIZI.

Micramphidiscs

Macramphidiscs

c o

CTl O

i O ^

I CO CO

O •* to

Tf .-H M

i>; .-^ in

OO Tf 'a'

I I I I I I I I I I I
ootO'^tDa>oooc^jcN)ai<r3^i'mc^tD

t>;t^(NMoqi-HCMOt^CO"-IOCMOOO

cri'S-oidcviooot^uDo^aioitxi.-Ia^

rtlO^OC^QOOOa^O»-HC^-rtLOt^OO

O <M

00

r-l (b

in r^

<>4 M

a> on

O CO

M I I I I I I I I I I I I I I I I I I M I I M M I M I I I

►J -1

IM CD
■^ 00

in o>

CO
rt CVJ
i-J CO
CM CSl '

O O -*

p -^ 1-;

Tt t^ T-H

CO CO ■^

(OOOt£>-^tOCriOOOCMCMCT>CO'5tLnt^

CMt>;i>^c^CMoq'-Hcgp^--cOf-;ocMoq
Lriai-^cD*ocNiooQt^ot-^c^cM'tdT-H
-^•■^iniotoooooooorHCM-^Lnt^

00 o

—I CM

00 r-i

CM in

CM CM

I— t <Ti
00 ■*

t> o

CN CO

Fig. 5. — • Form A. Amphidiscs.

near its centre to 8-20 n, and gradually at both ends to 8-15 yu- From the
central thickening arises a regular, or irregular, verticil of conic truncate spines
which is 2-8 ju long, and 2-5 n thick at the base. The ends (terminal faces) of
these spines bear clusters of small secondary spinelets. The other parts of the
shaft usually bear a few, rarely a good many, very low protuberances covered
with clusters of secondary spinelets. These protuberances are circular in outline
and agree in breadth and secondary spinulation with the spines on the central
thickening. The shaft is traversed by a fine axial thread which, where it passes
through the central thickening, is slightly thickened to a small but well-defined
point. Traces of branch-rays of the axial thread (an axial cross) could be
detected neither here nor elsewhere. The terminal anchors are composed of
eight to twelve recurved teeth. They are 40-80 ^ in length, that is about a third,
generally a little less than a third, of the whole spicule, and 41-86 ju broad. The
proportion of the length to the breadth of the anchors is 100 to 91-120, on an

HYALONEMA (HYALONEMA) AGASSIZI. 191

average 100 : 103. The anchor-teeth are T-shaped in transverse section. The
upper part appears as a curved band, for the greater part of its length 8.7-11.5 /u
broad, and attenuated to a point or, more rarely, rounded off at the end. The
teeth arise steeply and are uniformly curved, concave to the shaft. The curva-
ture is not very great and never suffices to give the end-parts of the teeth a
direction parallel to the shaft. The tangents on these end-parts enclose, in the
adult large amphidiscs, angles of about 5° with the axis of the shaft. In young
spicules of this kind (Plate 46, fig. 9) this angle is, of course, much greater.

The large macramphidiscs of form B (Plate 46, fig. 15) are 126-310 ix long,
most frequently about 240 fi. Their shafts are 8-14 /z thick, and their anchors
are 46-67 // long and 48-76 m broad. These spicules are longer and have smaller
anchors than those of form A.

The large macramphidiscs of form C (Plate 46, figs. 10, 11) are 110-290 n
long, most frequently about 240 /j. Their shafts are 6-10 m thick, and their
anchors 40-63 n long and 35-75 yu broad. They are very similar to those of
form B, and are like them longer and provided with smaller anchors than those
of form A.

The large macramphidiscs of form D (Plate 46, figs. 6-8, 16) are mostly
regular and similar to those of the other forms. They are 132-282 ti long, most
frequently either about 240 or 180 m, with shafts 9.5-12 /i thick, and with anchors
52-73 n long and 48-66 n broad. In respect to their length the regular large
macramphidiscs of form D are intermediate between forms B and C on the one
hand and form A on the other. Their anchors are similar in size to those of
forms B and C. In the specimen of form D I found an irregular large macram-
phidisc (Plate 46, fig. 7) 192 ix long, with a shaft 4.4 n thick. The two anchors
of this spicule are both about 26 /x broad but very unequal in length, one being
44 n long, the other considerably shorter. The shaft is beset with numerous
large pointed spines, all strongly bent towards one end. These spines increase
in size towards the end from which ihey diverge; the largest is 16 ju long.

The large macramphidiscs of form E are 150-280 yu long, u.sually 170-256 n,
most frequently about 200 ^u. Their shaft is 6-12 yu thick. Their anchors are
51-67 ti long and 51-80 ij. broad.

The large macramphidiscs of form F are 112-260 /i long, usually 150-232 ix,
most frequently about 200 ti. Their anchors are 36-65 ix long and 34-70 tx
broad. In a preparation of one of the specimens of tliis form I found an ab-
normal large macramphidisc, in which the central spine- verticil was replaced
by an anchor similar to but slightly smaller than the terminal anchors.

192

HYALONEMA (HYALONEMA) AGASSIZI.

To attain a clearer insight into the range and character of the variation of
the length of the large macramphidiscs in the four forms of this sponge I drew
the following graph, in which the frequency of the large macramphidiscs of
various lengths of all the four forms is represented. To make the curves in it
commensurate I calculated the relation of the number of large macramphidiscs
actually measured to 100, that is its percentage in each form, and multiplied all
the numbers of amphidiscs belonging to the same category, in respect to length,
by this number. These percentages are represented by the curves in Fig. 6.

Frequency
(%)

--- CO

a. o

•5

C

.2

c o

r— (

T— (

1

CM

1—1

1

CM
1

1

1

1

CO

s

t-^

cn

Fig. 6. — l^arge macramphidiscs.

The above curves, expressing the relative frequency of the large macramphi-
discs of different lengths in the six forms, are based on 247 measurements. The
curves pertaining to all the forms, except D, have one main elevation, the curve
pertaining to form D has two. The second main elevation of this curve coin-
cides with the main elevations of the curves of forms B and C, and with the
considerable right secondary elevation of form E. All four lie at a point corre-
sponding to amphidiscs about 240 n long. The main elevations of the curves
of forms A, E, and F lie at points representing spicules about 200 m long. The

HYALONEMA (HYALONEMA) AGASSIZI. 193

first elevation of the curve of form D corresponds to still shorter amphidiscs, of
about 180 M in length. The curves of forms A, D, and F have a small secondary
elevation at about 290 fjt. The curve of form B has a secondary elevation at
about 164 m; the curve of form F has also an additional slight secondary eleva-
tion at about 112 yu.

These curves indicate that A, E, and F have, on the whole, smaller large
macramphidiscs than the other forms; that D possesses two nearly equally
numerous varieties of these spicules, a larger and a smaller one; and that in C
the small forms of the large macramphidiscs are much rarer than in the others.
They further show that the large macramphidiscs of all the six different forms
differ in respect to the character and range of the variation of their length. It is
also to be noted that none of the curves are similar to a mathematical proba-
biUty curve; for these spicules do not, in respect to their length, vary uniformly
round a mean.

The small viacramphidiscs of form A (Plate 46, fig. 1) are 62-115 n long,
most frequently about 93 /x, and have a cylindrical centrotyle shaft 3-8 /x thick.
The central tliickening bears a verticil of rather blunt spines 1.3-1.5 fi long and
about 0.5 ju thick. Numerous similar spines are found on the other parts of the
shaft. The anchors are 22-43 ix long, that is about a third, generally a Httle
more than a tliird, of the whole spicule, and are 16-42 fi broad. The proportion
of their length to their breadth, in the smaller forms under 80 fi in length is, 100 to
73-84, on an average 100 : 79; in the larger forms, over 80 fi in length, it is 100 to
76-105, on an average 100 : 90. The shape of the individual anchor-teeth is,
on the whole, similar to that of those of the large macramphidiscs; but it is to be
noted that their curvature, in the smaller forms of these spicules, is considerably
greater.

In the specimen of this form (A) I found a remarkable hexactine spicule
116 yu in diameter, composed of four fully developed and two rudimentary rays.
The four developed rays are cylindrical, 7.5 n thick, and bear at their ends verticils
of large recurved teeth which together form somewhat irregular anchors 36 m
long and about 64 fj. broad. One of the rudimentary rays is a short, terminally
rounded cyhnder; the other is bifurcate and slightly longer at the end. The
whole spicule appears as a cross formed by two small macramphidiscs joined
in the middle, from the centre of which two protuberances arise.

Spicules of this kind have occasionally, but very rarely, been observed in
other species of Hyalonema, as H. tenerum} The only hyalonematid in which

' F. E. Schulze. Kept. Voy. Challenger, 1887, 21, pi. 31, fig. IS.

194 HYALONEMA (HYALONEMA) AGASSIZI.

they are more abundant is Monorhaphis dives, ^ where they usually have six
anchor-bearing rays. F. E. Schulze {loc. cit.) named these spicules hexadiscs.
The spicule above described and others similar, like the one found in Hyalonema
tenerum {luc. cit.), might, in an analogous manner, be called tetradiscs, or stauro-
discs.

In the other forms of Hyalonema agassizi only a few small macramphidiscs
have been observed. Those seen were similar but smaller than those of form A.
The small macramphidiscs of form B (Plate 45, fig. 53) are 53-101 n long. Their
shafts are 2-4 ix thick, and their anchors 18-27 ^ long and 18-33 ^ broad. Those
of form C are 53-80 ^ long. Their shafts are 2-3.3 /x thick, and their anchors
20-24 ft long and 16.5-20 /j. broad. In form D only a single small macramphi-
disc was found. This was 84 n long.

In form E the small macramphidiscs are 48-90 ^ long, most frequently about
51 M, and have anchors 10-33 m long and 10-25 m broad. In form F 1 found only
two such spicules. These were 45 and 70 ju long respectively.

In form A the large micramphidiscs (Plate 45, figs. 46-49; Plate 47, figs. 11-
13) are very numerous. They are here 42-60 n long, most frequently about
52.3 /J. The shaft is 1.1-2 n thick, cylindrical, or slightly and very gradually
thickened, in a spindle-shaped manner, in or near the middle, but without a
sharply defined central tyle. It is covered with numerous, slender, cylindrical,
vertical or, more rarely, oblique spines, sometimes 2 ^ long. The anchors are
relatively slender. They are 14.4-22 yu long, that is a little more than a third
of the whole spicule, and 8-14 ix broad. The proportion of their length to their
breadth is 100 : 53 (in one of the smallest), 100 : 71 (in one of the largest), on
an average 100:65. The individual anchor-teeth are about 1.5 ^ broad and
strongly curved, so that their end-i)arts lie nearly parallel to the shaft or converge
towards it.

In form B, where they are rather scarce, the large micramphidiscs appear
to be similar to those of form A, and measure 41-59 ^ in length, most frequently
about 54.8 ju-

In form C (Plate 45, figs. 59-61), where they are still more numerous than
in form A, they measure 36-64 ju in length, most frequently, as those of form A,
about 52.3 n in length, and have spined shafts 1-1.7 ^ thick. Their anchors are
17-23 M long and 10-13 m broad. One in which I was able to count the anchor-
teeth had fourteen. The individual teeth in the larger forms are about 2.5 ^
broad.

1 F. E. Schulze. Hexactinellida. Ergeb. Deutsch. tiefsee-exped., 1904, 4, p. 124, taf . 43, figs. 1, 6, 7.

HYALONEMA (HYALONEMA) AGASSIZI. 195

In form D, where they are also rather frequent, the large micramphidiscs
(Plate 45, figs. 63, 64) measure 35-55 ;u in length, most frequently about 47.5 n.
Their shafts are spiny and 1.5-1.7 ^ thick; their anchors are shorter than in the
other forms, 15-17 fi long, and 10-11 n broad.

In the specimens of form E the large micramphidiscs are exceedingly abun-
dant. They are here 40-69 n long, most frequently about 48 n, and have anchors
13-21 M long and 8-14 ti broad.

In the specimens of form F the large micramphidiscs are not nearly so
numerous. They are here 37-57 n long, most frequently about 52 m, and have
anchors 14-20 m long and 9-14 m broad.

According to the frequency of those of different length, three kinds of small
micramphidiscs can be distinguished in form F, and two kinds in fornis A, B, C,
and E. The small inicramphidiscs of form D are all of the same kind. The
smaller (A, B, C, E) or smallest (F) kind is invariably the most abundant. The
spicules belonging to the larger (A, B, C, E) or largest (F) kind have very slender
anchors and appear as transitions between the (broad-anchored) small and the
(slender-anchored) large micramphidiscs. Judged morphologically, by their
shape alone, the larger (largest) kind of small micramphidiscs should, indeed,
be considered as belonging to the large inicramphidiscs. Since, however, in
the smaller (A, B, C, E) or smallest and intermediate (F) kinds of small micram-
phidiscs the relative breadth of the anchors decreases with the increase in
the size (length) of the spicules, since in a few exceptional spicules of this kind
the anchors are quite as slender as in the larger (largest) kind, and since they are,
in the forms where they occur, separated biometrically much more clearly from
the larger micramphidiscs than from the smaller kind of the small micramphi-
discs, I provisionally place them in the latter group.

The small micramphidiscs of form A (Plate 44, fig. 17d; Plate 45, figs. 40-
45; Plate 47, figs. 3, 4, 7-9) are 15-36 yu long, most frequently about 18.3 n.
The limit between the larger and the smaller kind lies at about 29 ju. The shaft
is straight or rarely bent, 0.6-1.4 n thick, cyhndrical throughout, or slightly and
gradually thickened in the middle. It is either quite smooth, or it bears near
the centre an irregular cluster of a few spines, not over 0.5 /x long, or it is covered
with sparse, \'ertical, more rarely oblique, spines throughout. The anchors are
4-13 fi long, that is a quarter to a third of the whole spicule, and 4.7-9 n broad.
They consist of fifteen or sixteen recurved teeth. In the larger kind of small
micramphidiscs the proportion of the length to the breadth of the anchor is
100 to 58-75, on an average 100 : 64; in the smaller kind 100 to 75-135, on an

196

HYALONEMA (HYALONEMA) AGASSIZI.

average 100 : 93. It is to be noted that in the smaller small micramphidiscs the
relative breadth of the anchor is, on the whole, in inverse proportion to their size
(length) . The curvature of the individual teeth is such that the anchors appear
rather broad above and that the end-parts of the teeth come to lie parallel or
nearly parallel to the shaft.

The small micramphidiscs of the other forms appear to be very similar to
those of form A. Those of form B (Plate 45, figs. 50-52) are 13.5-31.5 m long,
most frequently about 20.1 ix. The limit between the larger and smaller kinds
lies, as in form A, at about 29 m- Those of form C (Plate 45, figs. 54-58) are
14-30 IX long, most frequently about 18.3 m- The limit between the larger and the
smaller kind lies here at about 25.5 m- Those of form D (Plate 45, fig. 62) are
13-28 M long, most frequently about 20.1 ix. The small micramphidiscs of form
E are 15-34 m long, most frequently about 22.2 fi. The limit between the larger
and smaller kind lies at about 24.4 /j. The small micramphidiscs of form F
are 15-33 ix long, most frequently about 18.3 fi. The limits between the smallest,
the intermediate, and the largest kind of small micramphidiscs of this form lie
at about 20.2 n and 25.4 m-

To obtain a clearer insight into the range and character of the variation
of the length of the small micramphidiscs in the four forms of this sponge I

Frequency

i Mm

1— 1

1

.—1

1

I— 1
1

»— 1

1

rH
1

1

1

1
S3

1

1
1— 1

1

1
00

1
in

1

o

t—*

CO

rH

-*

in

t— 1

I— 1

CO
(M

B

o

05

g

9

CO
CM

1

lO
CM

1

CO

c^^
1

1

1

1

1
I— 1

1

CO

1
in

lO

1

o

1

1

cq

CO
Cv)

lO
CM

S

o

CO

^

Form

3 A

^ B

I -- C

o D

^ E

" ^

Fig. 7. — Small micramphidiscs.

HYALONEMA (HYALONEMA) AGASSIZI. 197

drew (by the method already described) Figure 7, in which the relative
frequency of the small micramphidiscs of various lengths of all the four forms is
represented.

The above curves, expressing the relative frequency of the small micramphi-
discs of different lengths in the six forms, are based on 381 measurements. All
have one main elevation; those of forms A, C, and E have one secondary eleva-
tion, the cm-ves of forms B and F have two. The main elevations of the forms
A, C, and F correspond to amphidisc-lengths of about 18.4 m, those of forms B
and D to amphidisc-lengths of about 20.1 m, those of form E to amphidisc-
lengths of about 22.2 ix.

The first and principal secondary elevation of form F, which is very con-
siderable, coincides with the main elevation of form E at about 22.2 m- The
first secondary elevation of form B, which is quite insignificant, lies at about
24.4 fi. The first secondary elevation of forrii E and the second secondary
elevation of form F, which are both very well-pronounced, lie at about 26.8 m-
The single secondary elevation of form C, which is inconsiderable, is situated
at about 29.5 ix. The second secondary elevation of form B and the single
secondary elevation of form A both lie at about 32.5 m- The former of these is
very well-pronounced, the latter insignificant.

These curves clearly show that the small micramphidiscs of forms A, C,
and F are on the whole relatively small, those of forms B and D intermediate,
and those of form E relatively large, and further that all the six forms differ in
respect to the range and character of the variation of the length of their small
micramphidiscs.

The description given above shows these sponges to be so similar that there
can be no doubt about their belonging to one and the same species. They differ,
however, more or less by their external shape, the structure of their gastral cavity,
and the shape and size of their spicules. The variable spicule-characters which -
could be accurately ascertained in a sufficient number of spicules in all the forms
are : — the length and maximum thickness (together with the spines) of the distal
ray of the dermal pinules and gastral cone-pinules, the nature of the spinulation
of the former, the diameter of the microhexactines, and the length of the large
macramphidiscs and small micramphidiscs. In the following discussion I -have
considered only these spicule-dimensions, the shape of the sponge, and its gastral
cavity.

The specimens from Station 4651 and 4656 and some of the specimens from
Station 4742 are massive, spindle-, pear-, top-, or club-shaped, the specimens from

198 HYALONEMA (HYALONEMA) AGASSIZI.

Station 3684 (A. A. 17) and Station 4740 and the other specimens from Station
4742 are flattened, cake-shaped. In the specimen from Station 4656 the gastral
cavity is a narrow fissure, uninterrupted by radial plates; in all the specimens
frona Stations 4651 and 4740, in the cake-shaped specimens from Station 4742,
and probably also in the specimen from Station 3684 (A.A. 17), the gastral cavity
is quite wide and divided into separate diverticula by radial plates. The dermal
pinules of the specimens from Stations 4656 and 4740 have longer distal rays than
the others. The dermal pinules of the specimen from Station 4656 and the pear-
shaped specimens from Station 4742 have more slender distal rays (together with
the spines) than the others. The spines of the distal rays of the dermal pinules
of the specimen from Station 3684 (A.A. 17) are more crowded and form a more
compact structure than those of the others. The distal rays of the gastral cone-
pinules are of five sizes. Those of the cake-shaped specimens from Station 4742
are on an average only 90.5 n long, those of the specimen from Station 4651

98.2 n, those of the specimen from Station 3684 (A.A. 17) 106.7 m, those of the
specimens from Stations 4656 and 4740 128.2 ^1-128.5 n, and those of the pear-
shaped specimens from Station 4742 129 ^ long. Those of the specimens from
Station 4740 are (together with the spines) narrower than the others. The
microhexactines are relatively large in the specimens from Stations 4651, 4656,
and 4742, smaller in the specimens from Station 4740, and still smaller in the
specimen from Station 3684 (A.A. 17). In the specimens from Stations 4651 and
4740 the large macramphidiscs are of one kind and most frequently about 240 /j
long. In the specimens from Stations 4656 and 4742 these spicules may also
be said to be of one kind, and they are here most frequently about 200 m long.
In the specimen from Station 3684 (A.A. 17) a smaller kind, most frequently
about 180 IX long and a larger kind most frequently about 240 /i are nearly equally
abundant. The small micramphidiscs in the specimens from Stations 4656 and
4740 and the pear-shaped specimens from Station 4742 are most frequently about

18.3 M long, those of the specimens from Stations 4651 and 3684 (A.A. 17) most
frequently about 20.1 n, and those of the cake-shaped specimens from Station
4742 most frequently about 22.2 m-

This shows that the specimens of this species differ in respect to the follow-
ing ten accurately determinable qualities: — a, the external shape, b, the nature
of the gastral cavity, c, the length of the distal rays of the dermal pinules, d, the
maximum thickness of the distal rays, together with the spines, of the dermal pin-
ules, e, the density of the spinulation of the distal rays of the dermal pinules, /,
the length of the distal rays of the gastral cone-pinules, g, the maximum thickness

HYALONEMA (HYALONEINIA) AGASSIZT.

199

of the distal rays, together with the spines, of the gastral cone-pinules, h, the
diameter of the niicrohexactines, i, the length of the large macramphidiscs, and
k, the length of the small micramphidiscs. The following table, arranged in
pairs, shows which of these qualities the forms A to F have in common.

<a

a

o

J3

4656 (A) and 4651 (B)
" 4740 (C)
" " 3684 (A.A. 17) (D)
" " 4742 cake-sh. (E)
" 4742 pear-sh. (F)

4651 (B) and 4740 (C)

" 3684 (A.A. 17) (D)
" 4742 cake-sh. (E)
" 4742 pear-sh. (F)

4740 (C) and 3684 (A.A. 17) (D)
" 4742 cake-sh. (E)
" 4742 pear-sh. (F)

3684 (A.A. 17) (D) and 4742 cakc-sh. (E)
" 4742 pear-sh. (F)

4742 cake-sh. (E) and 4742 pear-sh. (F)

4

4

1

03

4

^

-4-3

7

o

-M

4

+3

CO

s

5

.£

&

6

+3

(U

03

1-,

5
3

3

.a

o-

<u

4

1
5
4
4

a e g h
c e f k

g

e g h i

a b d e g h k

b d e i

b c d g k

b c d e g h

a c e g h

abd

a b d e

e

a b c d g

c gi k

c e g h

These affinities are shown in Figure 8.

Of the five stations where these sponges were trawled, two. Stations 4651
and 4656, lie near together ofT the Peruvian coast. The other three. Stations
4740, 4742, and 3684 (A.A. 17), are a considerable distance apart in the central
Pacific and are far from the two Peruvian stations. The degree of similarity
of the specimens separated as the six kinds of Hyalonema agassizi stands in no
relation to the distances of their localities from each other. Thus the cake- and
the pear-shaped specimens from Station 4742 agree only in respect to four of the
ten quaUties, and the pair from Stations 4651 and 4656, which lie very near
each other, also agree only in respect to four qualities. The pairs which agree
most are the pear-shaped specimens from Stations 4656 and 4742, which agree
as to seven quaUties, and the cake-shaped specimens from Stations 4651 and
4742, which agree in respect to six. The units of the pairs of stations from which
these come are very far apart.

These and the other differences between the six kinds of Hyalonema agassizi
are not systematically important individually; I believe, however, that several
of them together demand recognition. Of the ten varying qualities here under
discussion, nine are different only in two pairs from Stations 4656, 3684 (A.A. 17),
and Stations 4740, 4742 pear-shaped. xAll the other pairs differ by from three
to seven of these qualities. Since the units of the two mentioned strongly diver-
gent pairs are connected in other ways, and since, as has been shown above,
there appears to be no correlation between the degree of difference and the
distance of the localities, I do not think that these differences warrant the

200

HYALONEMA (HYALONEMA) AGASSIZI.

Fig. 8.

establishment of separate varieties. The distinction of different foryns within
the species, designated A, B, C, D, E, and F, is a sufficient division.

The species composed of these six forms is most closely allied to the Hya-
lonema (Hyalonema) obtusum and the Hyalonema (Hyalonema) polycaulum.
The outer appearance of the variety gracilis of H. (H.) obtusum is indeed nearly
the same as that of form A. These latter differ, however, from H. (H.) obtusum
in the following respects: — the dermal pinules are considerably longer in both
varieties of H. (H.) obtusum than in any of the forms described above; the

HYALONEMA (HYALONEMA) POLYCAULUM. 201

slender-rayed basal spicules are about twice as large, and the inicrohexactines
are smaller, and composed of more strongly curved rays in the former than in
the latter; the large macramphidiscs of H. (H.) obtusum reach 356 ^ in length,
have shafts bearing large spines along their whole length and possess anchors
the end-parts of whose teeth are parallel to the shaft ; the large amphidiscs of H.
(H.) agassizi are not over 310 yu long, have shafts destitute of large spines outside
the centre, and possess anchors the end-parts of whose teeth diverge ; the end-
parts of the anchor-teeth of the large micramphidiscs are in the former far more
curved than in the latter.

Hyalonema {Hyalonema) polycaulum is in outer appearance, apart from its
polycaule nature, similar to the forms described as C, D, and E. It differs from
this species, however, by its large macramphidiscs, its pinules, and its mode of
attachment to the sea-bottom. The large macramphidiscs are in the sponges
described above considerably shorter, have relatively narrower anchors and
anchor-teeth much more strongly curved in their distal part and less divergent,
than in Hyalonema (Hyalonema) polycaulum. The distal rays of the pinules
of the former are considerably thickened above the middle and have a stout
terminal cone. In those of the latter such a thickening above the middle is
either absent or very insignificant, and the terminal cone is much more slender.
It is also to be noted that the distal rays of the pinules of the sponges described
above bear more numerous spines than those of Hyalonema (Hyalonema) poly-
caulum. The former is attached by a single stalk; the latter by several stalks.

Hyalonema (Hyalonema) polycaulum, sp. nov.
Plate 53, figs. 1-17; Plate 54, figs. 1-45.

One specimen of tliis species was trawled in the eastern part of the Tropical
Pacific at Station 4721 on 15 January, 1905; 8° 7.5' S., 104° 10.5' W.; depth
3811 m. (2084 f.) ; bottom composed of Ught brown Globigerina ooze. It appears
to have possessed four distinct stalks. To this the specific name refers.

Shape and size. The single, somewhat fragmentary specimen (Plate 53,
fig. 4) is oval, 54 mm. long, 48 mm. broad, and somewhat flattened. Only slight
remnants of the dermal membrane are left, the specimen appearing very porous
in consequence. A group of large cavities, separated by thin plates, occupies
one of the flat faces. A thickening at the joining hne of these plates, which,
however, does not project freely, is, as its internal structure shows, a gastral cone.
The large cavities around it are parts (diverticula) of the gastral cavity. On the

202 HYALONEMA (HYALONEMA) P()LYCAULUM.

opposite flat face of the sponge the superficial tissue is, in four places, consider-
ably harder than elsewhere. These harder patches protrude more or less and
appear as superficial knobs. They are distant from each other and rather uni-
formly distributed over the face opposite the gastral. From two of these knobs
a few broken stalk-spicules protrude.

The colour in spirit is brown.

Skeleton. Of the dermal pinule-fur only insignificant remnants are left; the
gastral pinule-fur, however, is preserved in places. The dermal pinnies have
much shorter lateral rays than the gastral. Here and there, where the super-
ficial parts of the sponge are still present, pentactine megascleres occur. The
hard superficial knobs contain dense masses of tetractine (tetractine-derivate)
stout, and diactine (diactine-derivate) more slender acanthophores. Here also
slender-rayed, long-spined acanthophores are met. Large quantities of micro-
hexactines, some microhexactine-derivates with fewer than six rays, and a good
number of more or less pinule-like pentactines, which may be tubular pinules,
and amphidiscs occur in the choanosome rhabd and hexactine megascleres.
Certainly from two, probably from all the four hard superficial knobs bundles of
rather large spicules, broken off at the surface of the sponge, extend towards the
interior. One of these bundles leads up to the gastral cone above referred to.
The spiculation of these knobs indicates that certainly from two and probably
from all four there arise in life stalks composed of bundles of spicules.

A good many anatriaenes with long and slender anchors, very large dicho-
triaenes, and large spiny metasters, aU foreign, and apparently belonging to some
species of Thenea, are found in the sponge. A few hexactinellid anchor-
spicules, similar to those of the holascids, were observed in the hard superficial
knobs. These spicules are 9-14 // thick, and have terminal tyles about 30 m in
diameter, beset with short anchor-teeth. I consider these spicules as foreign.

The dermal pinules (Plate 54, figs. 35, 38-40) are generally pentactine,
very rarely hexactine. The distal ray is 110-130 fi long, and 4.5-9 m thick at
the base. It is straight, thickened above only very slightly or not at all, and
ends with a slender and sharp-pointed terminal cone, only about 4 n thick at the
base, that is at the point of insertion of the uppermost spines. The basal part
and the terminal cone of the distal ray are smooth, its middle-part bears sparse
spines. The lowest spines are short, arise steeply, and are usually slightly
inclined towards the tip of the ray. Sometimes they are more strongly inclined
in this direction, sometimes vertical, and sometimes even inclined towards the
base. Distally the spines at first increase in size, and then again become smaller.

HYALONEMA (HYALOXEMA) POLYCAULUM. 203

Their inclination towards the tip generally increases uniformly towards the end
of the ray. The individual spines are conic, pointed, and curved more or less,
generally concave towards the tip (Plate 54, figs. 35, 40), sometimes in the
opposite direction (Plate 54, fig. 38, the lowest) and occasionally in an S-shaped
manner (Plate 54, fig. 39). The largest spines are 15-24 n long and 2.5-4 /u
thick at the base. The maximum thickness of the distal ray, together with the
spines, is 24-40 yu. The lateral rays are blunt-pointed or terminally rounded
and 24—40 n long. They bear rather sparse, conspicuous spines. The proximal
ray does not, when present, exceed the lateral rays in length.

The gastral pinules (Plate 54, figs. 41-45) are also usually pentactine; hexac-
tine forms are, however, not so rare among them as among the dermal pinules.
The distal ray is 110-133 ^ long and 6-9 ju thick at the base. It is straight,
simply conic throughout, or cylindrical basally and conic distally, or very slightly
thickened below the middle of its length. It ends with a slender and very sharp-
pointed terminal cone, 20-35 n long, and 3-5 ^ thick at the base, that is at the
point of insertion of the uppermost spine. Nearly the whole of the distal ray,
with the exception of the terminal cone, is spined; a spineless basal region being
absent altogether, or quite insignificant. The spines are sparse. The largest
arise from the middle-part of the ray. From here they decrease in size both dis-
tally and proximally. The basal spines are vertical or slightly inclined, either
towards the tip or the base of the ray. Distally they become more and more
inclined towards the tip of the ray. The individual spines are conic, sharp-
pointed, straight or slightly curved, either uniformly concave towards the tip
of the ray, or, more rarely, uniformly concave in the opposite direction, or in
an S-shaped manner. The largest spines are 12-16 n long and 3-4 ^ thick at
the base. The maximum thickness of the distal ray, together with the spines,
is 25-40 M- The lateral rays are 40-60 ^ long, usually conic, sharp-pointed, and
in their distal two thirds or to a farther extent, often quite down to their base,
they are covered with somewhat sparse, conspicuous spines. The proximal ray
is, when present, similar to the lateral rays in shape and spinulation and has a
maximum length of 50 fj..

The pinule-like pentadines, which may be canalaria (Plate 54, figs. 34, 37),
have straight, conic, sharp-pointed rays, rather densely covered throughout
with small spines. Their apical ray is 110-135 ix long, their lateral rays about
65 Ai. The rays are 5-6 /j. thick at the base.

The (hypodermal and ? hypogastral) pentactine megascleres have fairly
straight rays. The rays of the smaller forms are conic, very blunt-pointed.

204 HYALONEMA (HYALONEMA) POLYCAULUM.

smooth in their basal and spined in their distal part. Those of the large forms
are often nearly cylindrical, rounded at the end, and entirely destitute of spines.
In the smaller forms the apical (proximal) ray is usually about 300 ^ long, the
longest lateral ray being 220-340 n. In the large forms the longest lateral ray
attains a length of 650-1050 m- The basal thickness of the rays is 20-32 m in the
former, and 40-80 fi in the latter.

The hexadine megascleres (Plate 53, fig. 8) have conic and blunt, usually
fairly straight rays. In many, two opposite rays are longer than the other four.
The intact hexactines ob.served were 0.6-1.6 mm. in maximum diameter, and
had rays 15-35 m thick at the base. Some fragmentary ones had rays as much
as 40 n thick.

Among the stout-rayed tetractine and tetractine-derivate acanthophores (Plate 54,
figs. 1-15), which form the principal part of the skeleton of the hard superficial
knobs, forms occur with four, three, two, and one ray. Those with four rays,
that is the tetractines (or stauractines) , are the most frequent. The rays of
these spicules are on the whole cylindrical or cylindroconic, and rounded or,
more rarely, abruptly pointed at the end. They are straight or slightly curved,
and generally somewhat irregular in outline. The end-part of the ray is always
densely spined, the proximal part, usually one half to two thirds of it, is smooth.
Sometimes the spines extend farther proximally, occasionally quite down to the
centre of the spicule. The tetractines (stauractines) and triactines (tauactines)
(Plate 54, figs. 1-10) are 150-650 n in diameter, and have rays 10-37 m thick at
the base. The diactines (Plate 54, figs. 11, 12) are usually straight (Plate 54,
fig. 12) or more rarely abruptly rectangularly bent at the morphological centre
(Plate 54, fig. 11). The morphological centre is always thickened to a central
tyle, which is generally smooth, or more rarely spined. These spicules are
usually 0.4-1 mm. long, and 20-25 m thick near the middle. The central tyle
measures 30-50 m in diameter. The monactines (Plate 54, figs. 13-15) are usu-
ally 250-450 fx long, more or less cylindrical, 10-20 n thick, and closely resemble
single rays of the other forms. The end corresponding to the morphological
centre is thickened to a more or less spherical terminal tyle 18-40 m in diameter.

The slender-rayed, long-spined acanthophores of the hard superficial knobs
(Plate 54, figs. 30-33, 36) are usually hexactine, pentactine, or more rarely
tetractine. The hexactines and pentactines form a series extending from regular
hexactines (Plate 54, fig. 32) to pinule-like pentactines (Plates 54, fig. 35).
The tetractines are to be considered as pentactine-derivates. These slender-
rayed spicules measure 80-180 m in diameter. Their rays are straight, joined

HYALONEMA (HYALONEMA) POLYCAULUM. 205

at right angles, 3-8 n thick at the proximal end, conic, pointed, and covered with
somewhat sparse spines quite or nearly down to the base. The largest spines
are 5-15 m long and 2-3 n thick at the base. The basal spines arise vertically;
the distal ones are inclined towards the tip of the ray.

The microhexactines (Plate 53, figs. 9-12, 14-16) measure 80-142 n in diam-
eter, and have fairly equal, straight, or onlj^ very slightly curved, conic, and
pointed rays, which -are joined at right angles in the centre of the spicule. The
rays are 2.5-4 n tliick at the base and covered along their whole length with
spines sometimes 0.4 fi long. The basal spines are sparse and vertical, the distal
more crowded and oblique, incUned backwards, towards the centre of the spicule.

The rare microhexactine-derivates appear as spined aniphioxes, from the
centres of which arise terminally rounded rudiments of tlie four reduced rays.
In respect to size and spinulation they agree with the largest regular micro-
hexactines.

The cfioanosomal rhabds are mostly centrotyle amphioxes, more rareh'
styles (tylostyles). They are mostly 1-2 mm. long and 10-20 ju thick. The
central tyle is relati^•ely much larger in the thin than in the stout amphioxes, and
measures 15-35 ju in trans\'erse diameter. The proportion of the thickness of
the parts of the spicule adjacent to the tyle and of the tyle itself is 100:110-
100:250. It is to be noted that the tjde, particularly in the slender rhabd, is
often very eccentric, the four rays, the remnants of which it represents, being
not all reduced to the same extent.

The diactine (diactine-derivate) rhabd acanthophores of the hard superficial
knobs (Plate 53, fig. 17; Plate 54, figs. 16-20) are simple or centr