^ !^ i^ ^ X 'p ^ •+• j»^ T H E JOURNAL OF THE COLLEGE OF SCIENCE, IMPERIAL UNIVERSITY OF TOKYO, J A I> A N . VOL. XVIII. m M -^ gl :;^ ^ fP ^T PUBLISHED BY THE UNIVEKSITY. TOKYO, JAPAN . 1903-1904. , MEiJi xxxvr-xxxvii. Publishing Committee. •^^ Prof. K. IVIitSukuri, Ph.B., Rir/alcuhakushi, Dh-ectov of the College {e.v officio) Prof. J. Sakurai, LL.D., Pdnakuhalaishi. Prof. I. Ijima, Ph.D., Pagakühakushi. Prof. F. Ömori, Puiakuhakusld. Prof. S. Walasé, Ph.D., Rirmkuhnkushi. 2^ iL CONTENTS. Art, 1. — Studies on the Hexactinellida. Contribution III. (Placosoma, a New Euplectellid ; Leucopsacidœ and Ccmlopliacida^j. (With 8 plates). By I. IjiMA-Publ. May 5tb, 1903. Art. 2.— Cretaceous Cephalopoda from the Hokkaido. Part I. Lytoceras, Gaudnjccras and Tetrar/oitites. (With 7 plates). By H. Yabe. — • Publ. June Stb, 1903. Art. 3.— On the Formation cf Anthocyan in the Petaloid Calyx of the Red Japanese Hortense. (With one plate). By T. IcHnirnA.-Pubb June 20tb, 1903. Art. 4. — On the Comparative Anatomy of the Cucurbitacese, Wild and Cultivated in Japan. (With 5 plates). By A. Yasuda. — Publ. Sept. I5tb, 1903. Art. 5. — Untersuchungen über die atmosphärischen Pilzkeime. (Mi" -5 Tafeln). Von K Saito.— Publ. Jan. •28th, 1904. Art. 6.— On Some Jurassic Fossils from Rikuzen. (With 2 plates). By M. Yokoyama— Publ. Feb. 1st, 1904. Art. 7.— Studies on the Hexactineîlida. Contribution IV. (Enssel liehe). (With 2:i plates). By I. IJIMA.— Publ. June 18th, 1904. Art. 8.— Compositae Formosanœ. (With 2 plates). By B. Hayata. - Publ. Aui^- 13tb, 1904. PRINTED BY THE " TOKYO PRINTING COMPANY, LIMITED." JOURNAL OF THE COLLEGE OF SCIENCE, IMPERIAL UNIVERSITY; TOKYO, JAPAN. VOL. XVIII., ARTICLE 1. Studies on the Hexactinellida. CONTRIBUTION III. {Placosoma^ a New Eiiplectellid ; Leiicopsachlw and Cffulo2>7iacif1œ). By Isao Ijima, ^''g-- -P/?- D-. Rig.- Hah., Professor of Zo()logy, Imperial TTniversitv, Tokyo. With 8 plates. In this contribution I propose first to describe a new stalked Euplectellid of an interesting structure, which I call Placosoma paradictyum. This form became known to me after I had provisionally consid(!red my studies of the Euplectellidœ at end. It may however be best to describe it now and to follow it with some comments on the family generally, before I enter upon a treatment of other families. The present contribution will further contain full descriptions of six other forms, partly likewise new and parth^ before described by me in brief, viz., Leucopsacus orthodocus Ij., L. scoliodocus Ij., Chaunoplectella cavernosa Ij., 61 spinifera n. sp., Caulophacus loli folium n. sp. and Synipagella anomala u. sp. These will be referred to two families newly conceived and proposed for introduc- 2 AET. 1. T. TJIMA : ITEXACTINELLIPA, III. tion into the system, viz., the Leucopsaeidœ and the Caulophacidse. The former group is in part identical with that which I have formerly regarded as a Eossellid subfamily under the name of Leucopsacinee (Ijima '98, p. 41). The latter corresponds in a large measure to F. E. Schulze's Asconematida^, which name will have to be dropped out. Clrounds for the above rearrange- ment will 1)6 enunciated in the propor ])laees. PLACOSOMA PARADICTYUM. N. c, n. sp. Pis. T. and II. This new genus and species are founded on a single specimen obtained by Ki'Ma in Outside Okinose, Sagami Sea, from a depth of between 274 and :M3 fathoms (501-572 m.). The sponge is excellently preserved in a dried state, except for a rent across the body ; this, however, in no wise prevents us from obtaining a correct conception of the shape. It has a long stalk, which is attached to a brecciated tufaceous substratum by a large, irregularly lobate, basal expansion. Total height of the sponge, 210 mm. The sponge-body, though probably cup-like ^vhen young, can scarcely be called so as it is, but may best be described as an irregular, but transversely elongate and laterally compressed, cake- like mass. It is of a light, soft and very delicate texture. A light touch of the finger is liable to leave behind an impression PLACOSOMA tARADICTYUM. ^ Oil its surface. Length of transverse axis, 243 mm. Breadth in vertical direction, 126 mm. Thickness in the middle, 92 ram. On account of the compression tlie Ixxly presents two sides, both with uneven, thougli on the whole convex, surfaces. The margin is bv no means thin, being in places quite thick and rounded, while in other places it shows an angular edge-line. Very remarkable and striking is the difference in appearance presented by the two sides of the sponge- body. Nevertheless, 1 consider the surfaces of both sides, in fact the entire external surface, as dermal, and therefore as adapted to the afferent passage of water. However, as judged from structural relations, the inflow evidently takes place with special activity, in fact principally, on the one side which may be called the front (PI. I., fig. 1), while the other side, the back (fig. 2), seems to allow it only in a relativelv insic;nificant deo;ree and is further characterized bv the presence of numerous oscula distributed over it. The front of the entire sponge is quite destitute (Ä oscula. So far as the smooth and undulating frontage of the l)ody proper is concerned, it is nearly, if not wholly, occupied l)y a specialized area of the dermal layer, the frontal lattice, which consists of open-meshed dermal and hypodermal latticeworks of exquisite beauty. The dermal latticework (PI. II,, fig. 13), in which the laths are supplied by the fine paratangential rays of dermalia, is exceedingly delicate and shows small, regularly quadrate meshes. These measure 150-240 fi in length of sides. In nature they are easily discernible with the naked eye, but not in the reduced fig. 1, PI. I., in which however the minutely tessellated pattern may in some parts be observed with the aid of a hand-lens. — The hypodermal latticework, plainly visible in the figure just 4 ART. 1. 1. IJIMA : HEXACTINELLIDA, III. referred to, is a much coarser structure and exhilnts nieslies of a rectangular, trapezoidal, triangular or irregular shape. Within a larger mesh bounded l)y the stronger beams are inclosed smaller meshes formed by the weaker beams, and all the hypodermal meshes, large and small, are alike overspread by the dermal (or autodermal) layer already described, similarly as in certain Rossellids, Caulophacids, etc. While weaker hypodermal beams appear as simple spicular bundles, stronger ones are in the form of laterally compressed bands. The latter, at the points of junction with the choanosome, expand into vertical plates, which go to form irregular pillars. The frontal lattice evidently gives suj)port to the ectosome only. It thus differs sharply from the sieve-plate of Eujüectella or of Hyaloneiiia^ which structure, in my 023inion, is a peculiarly modified part of the sponge-wall in its entire thickness and therefore contains not only the ectosome but also the choanosome (Contrib. L, pp. 38, 66). Moreover, while a sieve-plate always stands in connection with either the principal or the sole place of water exit, it is just the opposite with the frontal lattice, which is developed apparently for facilitating the inflow of water. Nowhere else is the subdermal space so distinctly and spacious- ly developed as under the frontal lattice, which fact contributes not a little to the peculiarity of this side. It is especially deep (10 mm. or more) in the middle. In conformity with the wide subdermal space, many of the incurrent canals leading from it are large, some measuring VI mm. across at the entrance. The peripheral border of the frontal lattice may practically be considered to coincide with the marginal edge of the body, though at certain parts where this is rounded, it may not reach up to, but stops a greater or less distance short of, the ridge-line. In TLACOSOMA PAKADlCTYUM. Ö some parts of the margin, the lattice simply becomes gradually unrecognizable as such, without a distinct demarkation to delhnit it ; but the more usual condition is that there exists a rather sharply defined boundary. This consists either in a low and narrow wall-like ridge which is more or less distinctly inclined towards the front, or in that the angular body-edge is gently curved forwards, the concave surface being lined by the lattice up to the very edge and the convex, by the dense-looking cover- ing of the back, presently to be described. The latter condition reminds one strongly of that of the oscular margin in many cup-like Hexactinellids with the outflaring rim. On the iMck of the body the surface is much wrinkled or puckered up (PL I., fig. 2). Here, as on the entire stalk, a lattice-like arrangement of dermal spicules is not perceptiljle ; the subdermal space is scarcely developed, or at any rate is (juite insignificant. The entrances into incurrent canals are, if at all developed, small and indistinct. The dermalia are intimately associated with spicular elements of other categories and serve, together with these, to give to the surface a closely interwoven, opaque appearance. The superficial tissue thus formed is soft and extremely delicate, resembling a pith both in appearance and texture. — Further, this side of the body is, as before indicated, peculiar on account of the presence of oscula. These occur even on the stalk, but are confined to the same side. I distinguish a large primary or main osculum and numerous much smaller oscula of probably secondary origin. That which I consider the primanj osculum, is a roundish opening situated on the upi)er, broad and irregularly rounded edge of the body, Avell without the border of the frontal lattice. é ART. l.-r-î. UniA : IIEXACÏINÈLLIDA, ilt. The opening, witli a part of tlie gastral cavity into which it leads, is seen in lig. 2, Pi. IL, near the upper edge and to the left of the median line. It measures 36 mm. or more in maximum diameter. The oscular margin is partly rounded and indistinct, partly thin and wall-like. The gastral cavity may be described as a shallow cup-like depression, not more than 15 mm. deep. Its wall, formed of irregularly latticed strands, presents a some- what powdery appearance, due to the presence of large numbers of a certain hexaster. On it open a number of large and small, oval or roundish, excurrent apertures, measuring up to 8 mm, in width. Kight close to the apertures, the excurrent canals are seen to freely and widely intercommunicate with one another ; then, they penetrate deeply into the sponge-mass. — The osculum and the gastral cavity here in question, I am inclined to assume, were formed in a very early period of postlarval development. For a time they must have given to the young sponge a cup-like shape. With subsequent general growth, the cup-wall must have increased so excessively in thickness as to have obliterated the original shape, at the same time breaking through new oscula on the one side, as the need arose. These secondarily formed, or we may say, accessory oscula then seem to be equivalent in a way to the parietal oscula of Euplectella, save in this unessential point that they open into the excurrent canal-system, instead of into a common gastral cavity. The secondary oscula are oval or roundish, and sometimes rather irregularly shaped openings of various sizes. The larger may measure 5 nun. or more across, while some are small per- forations about 1 mm. or less in diameter. They are irregularly distributed all over the rough-surfaced back, but somewhat more sparingly on the upper rounded edge of the body. In some rLACOSO^rV PAr.ADlfTYUM. / places tliey lie tolerably close together. The thin and soft oscular rim is as often as not slightly raised in a lip-like manner. The oscnla lead either directly into deep going excnrrent canals or into such spaces of the excurrent canal-system as lie covered over by only a thin layer of the pith-like sponge-tissue. More than one secondary osculum may open into such a superficial space. This should not be mistaken for a subdermal space ; the thin layer over it represents the entire thickness of the sponge-wall, consisting, as it does, of parenchymalia in addition to the dermal layer. If we suppose a number of oscnla to have opened through it in close proximity to one another, we should have a structure strictly comparable to the Euplectellid sieve-plate. I believe it contains the chamber-layer in an undulating disposition, in much the same way as we usually see it in the thin marginal rim of cup-shaped Hexactinellids ; all the chambers, I think, have apopyles directed away from the external surface and towards the canalar space mentioned above. But this point could not be de- finitely ascertained by direct observation, on account of the desiccated state of the specimen. 80 far as the outward appearance goes, there is a certain resemblance between the back-surface of the present species and the gastral surface of certain other Hexactinellids, the oscnla of the former simulating the apertures of excurrent canals opening on the latter. But, from the mode of origin of the gastral cavity, there should constantly exist an essential difference in the relation of the dermal and scastral surfaces to the outer or the inner aspect of the chamber-layer. In the case of a real gastral surface, the apopylar ends of chambers should invariably be turned towards it, — not away from it, as they should be if the surface were dermaL. 8 ART. 1. r. T.TTMA : HEX ACTINELLTDA, ITI. In general shape the present species may be said to be not unlike a Caulophacus sj^ecies, in which the gastral surface, correctly recognized as such, is so turned outwardly and exjiosed as to form a part of the external surface. Indeed, l)y a hurried in- spection of the present species, one might possibly be misled into thinking of the area of the frontal lattice as the gastral surface similarly exposed on the outside. ]^ut the observer will soon see that structurally the frontal lattice is comparable with the dermal, and not to the gastral, layer of Caulophacus, Further, the system of canals communicating with the exterior by openings which I have unhesitatingly called the oscula is completely separate from the other canal-system that leads directly from the space right under the frontal lattice. Unless it be an error to regard the openings just referred to as oscula, the latter canal-system needs but to be considered as incurrent, and it then logically follows that the frontal lattice is a structure calculated to allow the afferent passage of water, which is invarial)ly the character of the dermal latticework. The stalk belongs to the back-side of the body ; at any rate, it arises a good distance apart from, and behind, the lower border of the frontal lattice and is covered all over by a direct and uninterrupted continuation of the pith-like surface of the back. Moreover, as before mentioned, it bears a certain number of secondary oscular openings. I lay some importance on the above facts, as demonstrating at once the dermal nature of the back-side surface, since the stalk surface can not possibly be anything but dermal. Excepting the soft covering layer, the stalk is firm and compact. It expands above to join the body. It is longitudinally PLACOSOMA PARADICTYÜM. V ribbed, more prominently on tlie back than on the front. Tt may be described either as a hollow tu])e with the lumen running close under the back surface, or as being grooved on this side, which groove is covered only with a thin sheet of the pith-like tissue. The lumen or the groove is evidently a part of the excurrent canal-system. It opens externally by a limited number of secondary oscula, distributed irregularly on the thin covering .sheet and therefore only on the posterior side of tlie stalk (PI. T., fig. 2.).=^= The large, irregularly lobed, basal disc is likewise soft on the surface but internally quite firm. The covered groove of the stalk extends for some distance on the disc- surface, showing on the cover a few more oscular openings. A certain number of other similar but branching canals are seen to ramify on the disc-surface in a vein-like manner. The most remarkable features in the organization of the present species lie in the massive development of the body and in the differentiation of a part of the external surface into an area, the frontal lattice, more especially adapted to the reception and passing in of the w^ater than other parts of the same. As regards the former point, an analogous case seems to be presented by Malaeosaccus floricomatus, recently described by TopsEXT ('oi). In this interesting Euplectellid the body should be solid and provided with a number of orifices — evidently exhalant orifices or oscula — distributed over the external surface. The describer however leaves margin to allow assumption that perhaps the presence of a shallow hollow at the superior extremity, represent- * The small hole seen on tlie front sUle of the stalk (PI. T., fig. 1) is an artifirial perforation. 10 AlîT. ]. 1. T.TIMA : IIEXACTINELLTDA, III. ing a rudimentary gastral cavity, is to be ascribed to the species. If that be true, the species should be said to agree with Placosoma imradiclyum not only in the excessively thick development of the originally cup-like sponge-wall but also in the possession of numerous secondary oscula in addition to the single primary one. The development of the frontal lattice and the occurrence of oscula exclusively on the opposite side of the body in Placosoma paradicfyum i:)robably stand in relation to the physical circum- stances of the habitat favoring the ingress of water on the one side or its egress on the other, — possilily in that the former is exposed to, and the latter sheltered from, a f>i'evailing current. Here again we seem to have to do with a structural arrangement which does not stand quite isolated among the Hexactinellida. So, for instance, in Semper ella schuUzei, as is well known, the external (dermal) surface is differentiated into two sorts of tracts, the afferent and the efferent. The latter may indeed be said to occur here on all sides of the columnar body, but is relatively nuich more extensively developed on its one side than on the other, — a fact, which seems to have hitherto escaped the attention of observers, but which I liave found to hold good for all the specimens examined by me. A closer analogy to the condition that obtains in Placosoma paradiclyuin is apparently exhibited by Poliopogon amadou, as known to me through F. E. Schulze's representations in the Challenger Report. In this form the entire external surface is, in my opinion, to be regarded as dermal, the gastral surface being nowhere outwardly exposed. The discharge of water is effected through orifices which are situated exclusively on the one — the concave — side of the half-rolled lamella-like body. This concave surface has been called by F. E. Schulze gastral, which appellation however seems inadmissible in view of PLACOSOMA PAKAt)ICïYUM. 11 the fact that the subjacent chamber- layer shows the Inline! chamber- ends all directed towards it, similarly as on the convex side (Chall. Eep., PL L., iig. 1). It is then plain that the inflow of Avater takes place all over the surface of both sides, except ot course at the separate excurrent oritices on the concave side. Ihe condition uf this side then exactly corresponds to that on the back of Placosoma paradictyu/n, while the convex side may be said to correspond to the front, though it lacks a special structure like the frontal lattice. Spiculation. (PI. IL). First let me complete an account of the spiculation in the sponge- body proper. Of the parenchymalia the predominating form is a line diactin, generally 2-8 mm. in length. The breadth, not exceed- ing 15 y, remains nearly the same throughout ; not seldom however it slio'htlv increases towards both extremities. These are usually bluntly pointed ; the surface near them is nearly smooth or obsoletely rough on account of insignificant tubercles. The center of the diactins is generally quite smooth, and exceptionally marked by an annular swelling or by knobs in cruciate arrangement. The diactins are disposed partly irregular- ly or in loose indefinite groups and partly in long, compact and thread-like fascicles of variable strength (PL II., fig. 15). The hypodermal beams of the frontal lattice (see fig. 13) are nothing else than extensions of such parenchymal fascicles from the choanosonie. — Among the parenchymalia making up the 12 ART. 1. — i. IJIMA: HÉXACTiNELLlDA, lit. fascicles there are none Avhicli, on account of a specially einpliasized size or strength, may be distinguished as the principalia. Nor does there exist synaptieular fusion among any of the paren- chymalia, so far as those of the main body are concerned. — Besides the diactins there also exist, among the parenchymalia, a number of medium- sized hexactins in sparse distribution (see lig. 15). Occasionally these are represented by forms more or less closely approaching a pentactin or even a stauractin in form. Axial length up to nearly 1 mm.; thickness of rays near base up to 20 /^ ; the rays gradually taper towards the free end, which is faintly rough-surfaced and either sharply or l)luntly pointed. The rays in a spicule are often of unequal lengths. In some cases one of the axes was found to be prolonged more or less in comparison with the rest, and with that elongated axis the spicule took part in the composition of a parenchymal fascicle ; but the rule is that the hexactins stand in no delinite relation with other parenchymal spicules as regards the situation of their rays. The parenchymal hexactins are, in point of size and shape, not always shar]:)ly distinguishable from those hexactins which may be called the canalaria. These are found sparingly and isolatedly in irregular distribution on the canalar surface as well as on the hypodermal beams of the frontal lattice. Two of them are seem in the lower part of fig. 13, PI. II. They are re- cognizable as such only when one of the six rays is considerably shorter than the rest and projects freely, from l)ase to tij), into the canalar lumeii or the subdermal space. The free ray shows, though not always, a further specialization in that it has a rounded termination, instead of being jiointed like the other rays. 1'LACOÖOMA l'AUAt)tCTYUM. lo The dermalia and tlic gastralia are likewise spicules whose shape and size present points of approach to the parenchymal hexactins. The dermalia are hexactins, in part somewhat sword-like in shape as in the generality of the Euplectellid members, but in part much flattened owing to an extensive shortening of the proximally directed blade-ray. In details of characters they differ somewhat according as they are located on the front or on the back of the sponge-body. First, the dermalia of the frontal lattice (PL II., figs. 1 and 2). The rays measure about 9,« in breadth near the spicular center. The distal free ray is ahvays very short, being only A')- ÖÖ I'- and seldom 'do <)■ in length ; it usually broadens slightly towards the distal end, which is rounded and has a surface supplied with quite insignificant microtubercles. The paratangen- tial rays are 150-200/^ long, as measured from the center; they taper outwards in a l)arely perceptible degree, to terminate with rounded or bluntly conical tips ; the surface is sprinkled with obsolete microtubercles which are distributed more densely near the end of the rays. The proximally directed ray resembles the paratangential s in general appearance but is subject to great variation as to its length, according to the position of the dei'malia in the frontal lattice. Where there exists a liy^Dodermal beam directly underlying the dermal layer, the spicules of the latter have a more or less elongated proximal I'ay dipping right into the substance of the former. That ray is then considerably — at times even thrice — longer tluin a paratangential of the same spicule (PI. II., fig. 1). Whereas, within a mesh-area bounded by hypodermal beams, i. e., in parts where the dermal layer has no skeletal support underneath, the dermalia have their proximal 14 AET. 1. — I. IJIMA : HEXACTINELLIDA, III. rays greatly reduced in length (PL II., fig. 2) In extreme cases, this ray may be only about as long as the distal ray and similarly club-like in appearance. So far as the dermal layer is made up of such — so to say, flat — hexactins, it stretches freely over the wide suljdermal space without any spicular connection with the choanosome. Himilarly flattened dermalia, occurring under the same circumstances, have been described by me from RegadrcUa homeyamai among the Euplectellidie (Contrib. I., p. 2G1). The dermalia on the back of the sponge-body are all supplied with a proximal ray which is always the longest of the six, the dermal layer on this side being everywhere in close connection with the choanosomal mass. Under a general agreement in ap- pearance, they ditler slightly from those of the frontal lattice in being on the whole larger and in the rays being somewhat more tapering and having a nearly smooth surface except for a short space at the end. Breadth of rays up to 10/^- near the spicular center. iJistal rav 70-100 /'■ Ions; ; swollen towards the rounded end. Paratangentials 220-300 // long. Proximal ray may be twice as long, or longer. In the frontal lattice the })aratangentials of separate dermalia are as a rule closely apposed to one another to form the exquisite checker-like latticework (fig. 13). On the back of the sponge this arrangement is carried out to a certain extent but not with the same degree of regularity (fig. 14). Here the separate dermalia generally lie wider apart and at places show no order as to the relative disposition of their paratangentials. The gadralia (PI. II., figs. 3 and 16), found in irregular distribution over the gastral surface inside the main osculum, are hexactins. They resemble in a})pearHnce the dermali'a of the TLACOSOMA rARADICTYFM. 15 l)ack. Length (»f tlie longest — the radially directed — axis, 600- 800 ,«. Paratangentials 200-260 ,« long. The free gastral ray is of abont the same length, or shorter ; it is not always rounded at the end, but may be gradually tapering and conically pointed like the other rays. Thickness of rays near base 73-8 /^ The hexasters of the species l^elong all to one type, the discohexaster ; but this occurs in no less than three varieties of markedly different characterization, viz., the hexactinose discohexaster, the spherical discohexaster and the hexactinose codonhexaster. Floricomes and graphiocomes are not present. Bv far the most abundant and the most «ienerallv distributed are the small hexactinose d hcoliexasters (PI. II., figs. 4 and 7). They are very numerous about the hypodermal beams of the frontal lattice and everywhere in the parenchyma (see figs. 13- 16). The manner of their distribution in large numbers reminds one of that of micramphidiscs in certain Amphidiseophora. The size is subject to considerable variation. In most cases the diameter measures 30-60 !'-, on an average about 46 !'■ ; but oc- casionally, and especially near the l)ack surface of the sponge, the hexaster attains a much larger size, reaching 100 /-« in diameter. The axial filaments in the central node extend, as determined by direct observation, only to the base of the six moderately strong rays. Tlie watchglass-like or nearly hemisj^herical terminal disc measures 8-15 !>- across ; its well developed marginal teeth vary from 8 to 15 in number, according to the smaller or larger size of the spicule. In some rare instances I have found the terminal disc unusually small and toothless, a condition which is probably to be considered as representing a stage in its developmental history. Also a few cases of heraihexasterous forms, such as are 16 ART. ]. — T. T.TIM A : HEXACTTNELLTDA, III. shown ill figs. 5 and 0, came under my ol^servation. It seems that through these forms tlie small hexactinose variety of hexasters now in question passes over gradationally into the spherical discohexaster next to be described. On the other hand, there not infrequently occur, especially near the surface of the back and of the gastral cavity, such forms as bridge over the gap ])etween the present variety and the hexactinose codonhexaster (fig. 8). The intermediate forms just referi-ed to are generally somewhat larger than usual and possess terminal discs, which, by elongation of the mai'ginal teeth, have acquired a moi*e or less bell-like shape. The spherical discohexasters (PL IL, figs. 9-12) are large and of great beauty, most closely resembling those tliat arc known from Dictyaulus elegans F. E. Sen. I have found them only close to the dermal layer on the back of the sponge, where they occurred abundantly in some places (see fig. 14), but only occasional- ly in others. They mostly measure 160-240 /^ in diameter. From the expanded end of each short and stout principal there arise in the smaller rosette 4 oi- ö terminals and in the larger ones 12 or thereabout. They are not always arranged in a whorl, l)ut often one or more are seen to spring from a position inside the points of origin of tlie peripheral ones. The slender terminals thicken considerably at the outer end and are capped with a strongly convex disc, which may measure up to 23 i^- in diameter. Of the well developed marginal teeth tliei-e are 10-13, sometimes more (up to 17), to a disc. All the terminal discs in a rosette of the kind are approximately equidistant from one other, so that a spherical form is given to the entire spicule. — Special mention must be made of unusually small, normally developed, ])ut onlv occasionally occurring discohexasters which seem to lead PLACOSOMA PARADICTYUM. 17 the large spherical diseohexaster gradationally over to the small hexactinose diseohexaster l)efore described, especially to the hemihexasterous form of this rosette. We have here to do with spherical discohexasters of under 150/^- diameter, leading doAvii to 100/^- or less (even to 'jö/<). Fig. 9, PI. II., represents one such case, measuring about 100 /' in diameter. Two or three, some- times four, terminals belong to a principal ; the terminal disc is just the same as in a small hexactinose diseohexuster. — Xotewortliv also seem the cases — quite rare though these are — of certain moderately large (132/^ diameter) discohexasters in which 4 or Ü terminals, each ending in a bell-shaped terminal disc, belong- to each principal (see fig. 10). Here is apparently an approach of the normal discohexasters to the hexactinose codonhexaster next to be described ; but an intermediate hemihexasterous codonhex- aster has not been met -with. The hexactinose codonhexcisters (PI. IL, figs. 11, IG) occur in great abundance just under the gastral layer. The powdery appearance, before noticed, of that layer is due to their crowded presence. They are also found in the parenchyma generally, but only quite seldom and at Avide intervals. Diameter in most cases 110-176 !'. The slender hexradiate rays, arising from a small central node, are usually more or less bent. Here again the axial filament of the central cross extends, as a matter of fact, only a very short distance into the base of the ra^'S. The deeply bell-like terminal umbel may be 30 !'■ broad and 42/-« long; the long teeth of the sides 12-16 in lUTmber. I have already indicated that this category of the hexasters is connected with the small hexactinose discohexasters by forms that are intermediate both in point of size and of the shape of the terminal disc. 18 ART. 1. — I. IJIMA : HEXACTINELLTDA, III. Finally, as to the spiculation of the stalk and of the basal disc. In both these parts the main mass of the parenchyme is rigid and consists of long, synapticularly connected diactins, measuring 10-40/^ in thickness. These run in longitudinal bundles in the stalk, while in the disc they are disposed more or less parallel to the surface but otherwise in nil directions. Hexasters among the fused parenchymalia seem to be rarely present. Towards the external surface, in both the stalk and the disc, the synapticulœ cease to exist. The peripheral diactinic jiarenchymalia thus left loose and unconnected, together with the dermalia and the hexasters present, form a soft tissue which in a thin layer covers the entire external surface of the parts in question, besides constituting the substance of the thin oscula- bearing sheet which partially shuts off the large excurrent canals from the exterior (see p. 9). The dermalia of the stalk compare well with those of the body, except in often having the distal ray considerably reduced in length, sometimes even to a mere knob or a gentle swelling. And, lower down the sponge, i.e., on the disc-surface, this ray is completely lost in all, so that here the dermalia are pentactinic, — a condition which may possibly represent the state as assumably obtains in the entire dermal layer of the species in the earliest period of postlarval development, as I have found to be the case in young RegadreUa oJcinoseana (see Contrib. I., pp. 246, 247). In all the dermalia the proximal ray, j^enetrating into the underlying tissue, is always the longest. Seen on the surface, the paratangential crosses lie rather densely crowded, without any regularity as to their relative position. The hexasters in the said soft tissue of the ]ieriphery are OBSERVATIONS OX THE Eui^LECTÉtLTDi: GENERALLY. lÖ the small hexactinose and the larger spherical discohexasters, both of exactly the same characters as those of the body. The former are common everywhere, while the latter have been found scattered here and there in the stalk only. The large excurrent canal in the stalk is likewise lined with a tliin layer of fine diactinic parenchymalia, free of synapticular formations. No special canalaria exist here, but the hexasters are represented by the usual, small, hexactinose discohexaster and the hexactinose codonhexaster of a larger size — the former in excessive abundance ])ut the latter only sparsely. It may be pointed out that the presence of the codonhexasters along the canalar wall manifests a point of agreement, and indicates a direct continuity, between the surfaces of that canal and oi' the gastral cavity, in a manner as it is perfectly natural to find, as we did, the same spherical discohexaster under the conthuious dermal surfaces of the stalk and of the back of the sponge-body. Ohset'cations on the l^JuplecfeUidw yenefally* As to the systematic position of Placosoma paracUctyum, it is safe to say that it should be 2)laced under the Euplectellidic. In an attempt to determine, as nearly as possible, its position within that family, I have been led tu undertake a renewed study of all the membejs of the family as regards their known systematic characters, and this induces me to make here some observations concerning the family diagnosis and the division into subfamilies, and in this connection I take the opportunity to modify certain 20 ART. 1. — I. IJIMA : HEXACTINELLIDA, lit. statements that I made prematurely in Contribution II. with respect to the Eiiplectellid subfamily Corbitellinœ. It may be well to preface my remarks by an enumeration of all the genera that I consider as making up the family as it now stands. These are as follows : 1. Ilolascus F. E. Sch. 2. Blalacosaccus F. E. Hch. o. Eu2)leclella Owen. 4. Regadrella O. Schm. 5. Corhitella Gray. (). Heterotella Gray. 7. Walteria F. E. Sch. 8. Diclyaulus F. E. Sch. 9. Dictyocalyx F. E. Sch. 10. Hertwigia O. Schm. 11. Trachycaulus F. E. Sen. 12. Saccocalyx F. E. Sch. 13. Rhahdopectella O. Schm. 14. Rhabdodictyuiii O. Schm. 15. Hyalosiyhis F. E. Sen. 16. Placosoma I.i. The family diagnosis, as given by F. E. Schulze ('99, p. 1)7) in the latest period, runs as follows : '' Röhren-, sack- oder kelchförmige Hexasterophom, welche entweder mit einem basalen Nadelschopfe im Boden wui-zeln oder, sei CS direkt, sei es mittelst eines langen röhrenfürmigen Stieles, aufgewachsen sind. Die Dermalmembran wird gestützt durch hexactine Hypodermalia, deren Proximalstrahl in der Kegel verlängert ist." (AVith 8 species) . V J» 0 0 V " 13 V ?5 11 0 \ ?J 3 V ?' 1 V " 2 \ '> V " \ '? V >> V '> V J> l '> V " >> )• V n " )• OBSERVAÏIOlSfS ON THE EUPLECTELLID.Ë GENERALLY. 21 Excellent as tliis diagnosis is on tl^e whole, I think it may perhaps with advantage ])c somewhat remodelled in order the more sharply to characterize the family. In fact, this seems necessitated to a degree by the discovery of new forms since the above diagnosis was drawn np. In attempting the revision, it is to be borne in mind that of all the lyssacine families it is especially the Caulophacidœ (for which, vide a later paragraph in this Contribution), with which the Euplectellidse, as bearing closest resemblance in certain important systematic characters, require to be placed in contrast. Xow, the more important points not indicated or explicitly mentioned in the aljove diagnosis F. E. Schulze's but which appear to call for our attention here, seem to be the following. 1. The massive development of tlie body in JIalacosaccvs florlcomatus Tops, and Placosoma paradictywi/ Ij., neither of which can be said to be tubular, saccular or cup-like. 2. The presence of a distinctly stalk-like part in the body of JIalacosaccus florlcomatus, running out into a tuft of anchoring spicules at the inferior end. The stalk is then a thing, the occurrence of which is not confined to those Euplectellids which at base are firmly fixed to the hard substratum. 3. The EuplectellidcC, excepting a few insufficiently known forms {JIalacosaccus vastus, 31. luiguiculatus, Hertwigia, Hyalo- slyhis), as a rule exhibit on the parietes a large number of separate orifices (oscula) for the discharge of water. It is connnon to find them all (Ilolascus) or in part [Euplectella, Corhiteila, etc.) in close congregation (forming the sieve-plate meshes) at the superior end of the body, whereby is brought about a condition which simu- lates such other cases as show a single hirge tei-minal osculum with (e.g., Flacosoiita paradiclijuin) or without (e.g., JIalacosaccus 22 ARî. 1.— î. iJiMa : hexaOtinelltda, iit. unguiculalus) a greater or less number of additional oscula on the lateral wall. The multiplicity af oscules is certainly not a peculiarity of tlie Euplectellidie ; and moreover, cases are not wanting in this family which indicate that we have here to do with a very variable character, sometimes apparently of no more than specific, or at most generic, value, Kevertheless, it seems undeniable as a general fact that the tendency toward that character is, in the Euplectellidte, brought into expression with such a degree of constancy and accentuation as is foreign to any other lyssacine hexasterophorous family (Leucopsacidse, Caulopha- cidic, Rossellidœ). And further, it is a noteworthy feature that on this account the aj^pearance of individuality is not generally in the least impaired. Whereas, in the Caulophacidae, the formation of additional oscula in an individual is always in connection with a ])rocess of budding, thus imparting to it a more or less cormus- like or " polyzoic " appearance. 4. The dermalia (=F. E. Schulze's Hypodermalia) in the Euplectellidte may be said, as a general matter of course, to still remain in a comparatively incipient stage of morphological ditlerentiation."" Thus, they stand in direct and intimate relation with the parenchymalia in that the proximal ray dips right into the choanosome ; they are known in some cases even to intergi'ade in the characters of rays with certain hexactinic parenchymalia. On the other hand, in the Cauloj^hacidae — and, I may add, in the E-OssellidiE — the corresponding spicules [I.e., the autodermalia) apparently represent a more highly specialized category of spicules which are joined to the parenchymalia only through the inter- ■" Tliis condition of tlie dernialia is shared in a way also liy the I.eucopsacida?, in ■which ]io\vever the denualia are pentactiiis or are at any rate always wanting in distally directed rays. OBSERVATIONS OX THE EFPLECTELLID-E GENERALLY. J.) mediation of the hypodermalia, especially the large pentactinic hypodermalia. Amongst the Euplectellids, the liypodermalia are thus far known only in the frontal lattice of Placosoma paradiclyuni but in no case have hypodermal pentactins been found. Here is, I think, an important negative character which distinguishes the family from such others as are most liable to be confounded witli it. After what I have said al)ove, the diasrnosis mav be made so as to read somewhat as follows : Lyssacine Hexasterophora" of tubular, cup-like or massive body; sometimes stalked; either rooted by a tuft of basal spicules or firmly attached by compact base; generally possessing numerous separate oscula. Dermal skeleton composed of hexactinic dermal ia, the proximal ray of which is as a rule much longer than any other in the same spicule; no hypodermal pen- tactins. Hexasters various. * I adopt F. E. Schxtlze's ('99, p. 93) system of dividing the Hexactinellida — the living Hexactinellida, at least — into two great primary groups or suborders, the Aniphidisc- ophora and the Hexasterophora. Here, as elsewhere in this paper, the term "lyssacine" is used without implying the Lyssacina as a systematic group, but merely to denote the condition of those Hexactinellids, in which the parenchymalia are all or mostly free and unfused, or in which tiiese may sometimes be extensively ankylosed and then consist not exclusively of hexactins, but of hexactins and their derivative forms or of the latter only. The term " dictyonine," as employed by me, refers to the state of those Hexactinellids in which the parenchymalia, consisting of hexactins, and as a rule of hexactins only, undergo fusion among themselves from a very early period of their existence. A brief exposition of my views on the Hexactinellid phylogeny and system, incomplete and somewhat pro- visional as these necessarily are in the present stage of my studies, may help to clear up the points in question, and may at the same time serve to indicate the position of the different families described in this Contribution. For the ancestral Protohexactinellida is to be assumed a lysmclne form, in which the spicules consisted, mainly at least, of more or less regularly developed hexactins. Now, the Amphidiscophora siiould represent a very early differentiated branch of the Hexactinellida, which has remained tlioroughly lyssacine in character and has been, on that account, in a position to give rise to such manifold variations of triaxonic spicules as we see in that group. 24 ART. 1. 1. TJIMA : IIEXACTTXELLIDA, ITI. 'Now as to the division of the Euplectellidœ into subfamilies. Of these, three have been recognized by F. E. Schulze ('99, p. 97), viz., the Holascinœ, the EnplectelHnir and theTœgerina'. The Hexasterophora, it seems safe to say, should have had for its prototype a form which was firmly attached by its base to the hard substratum. Whenever a Ilexactinellid is thus tixedly seated, even though it be a s^o-called Lyssacina, it is quite generally true that certain small liexactinic spicules occurring at the base very early undergo fusion, beginning with those in direct contact with the substratum, thus bringing about a typically dictyonine framework at tlie part. This miglit luive taken origin in consequence of the sponge requiring a certain degree of firmness at the part of attachment. That rigid frame- work, so far as it occurs in lyssacine Hexactinellids, I have called tlie hanidiclj/nnalia (Contrib. I., pp. 186 [foot-note], 232 and 264). The same framework it is, I liold, which has reached the most extensive development in the so-railed Dictyonina and lias been called by F. E. Schulze the dirlyonalia. Tlie basidictyonalia and the dictyonalia I consider as genetically and morphologically identical. Botli retain a primitive character in that they alike consist, as a general rule, of hexactins only, although derivative forms of spicules with a smaller number of rays may secondarily come into fusion witli the beams. The true dictyonine skeleton as here specified, should be kept distinct from another kind of ankylosed framework in wliich the elements involved are not solely hexactins, but principally derivatives of tlie sime, such as stauractins, tauactins, diactins, etc. (as, e. g. in the rigid skeletal frame of the lateral wall or the stalk in the EupIectelUmr). The rigidity in the latter case is clearly a much later acquisition than that in the basidictyonalia or the dictyonalia. For, before the ankylosing process, starting at the base, has encroached upon the parts occupied by the derivative spicules indicated, these must have had a long period of loose existence during which they should have derived their^shnpe from the original hexactinic form,— which would be impossible had they been soldered together l)eforehand. From this standpoint, the well-known skeletal framework of f. i. Eiiplectella aapei-giUum, notwithstanding the completely ankylosed state of its component spicules, ought not to be confounded with the basidictyonalia or the dictyonalia of certain other Hex- actinellids but should fill under the lyssacine category, in which the ankylosis has but secondarily set in. But to return to the Hexasterophora prototype. I consider this to have already possessed at its base a basidictyonal mass in addition to the loose spicules of the body proper, which were directly inherited from the Hexactinellid ancestor. In other words, it may be said that the skeleton was partly dictyonine and partly lyssacine in character,— a dual condition, which, in general, may be said to obtain in all the recent Hexasterophora, if we except those forms which have apparently lost the basidictyonalia in secondarily adapting themselves to the special mode of attachment by means of anchoring spicules (Euplectellinîe, Lophocalyx, Mdonympha). During the phylogeny, the elements of the dictyo- nine portion have remained essentially unchanged in character, as they should owing to their rigidly fixed state. On the other hand, those of the lyssacine portion have been capable of adapting themselves in manifold ways to varied conditions of existence, analo- gously to the similarly circumstanced spicules of the Amphidiscophnra, the result being the multifarious hexactin-modifications— amongst them the h?xasters— which we meet with in the Hexasterophora. OBSERVATIONS OX THE EUPLECTELLID-E GENERALLY. 25 Whether or not tlie Holascinse, made up of the two genera Holascus and 3Ialacosaccus, and the Euplectellin», consisting of the single genus Eujjlectella, are to be kept separate, is, I should think, largely a matter of opinion. To me it appears that the two subfamilies had better be united into one, chiefly because the main distinctive character that has l)een assumed as existing between them, — viz., the absence or presence of orifices on the lateral wall, — has been discredited by the recent discovery of 3Ialacosaccus floricomatus Tops., in which a number of orifices, Whether or not the palîwzoic lyssaciue forms, put together by ScnBAMMEN ('02) under a distinct suborder, the Stauractinophora, are to be looked upon as really representing a phylum systematically nearly equivalent to the Amphidiscophora or the Hexasterophora, I prefer to leave undecided, owing to uncertainties that always attach to the fossil Hexactinellids in respect of the finer spiculation. The Hexasterophora I assume to have early split into at least three branches or tribes, to be here provisionally called A, B and C. Tribe A, which may be allowed to retain Zittel's name Lyssacina but in an altered sense, comprises ail the hexasteropliorous lyssacine families, of which I distinguish four, viz., Euplectellidse, Leucopsacidse, Caulophacidse and Rossellidœ. Tiie other two tribes are both dictyonine and together correspond to Zittel's Dictyonina, but are probably not to be put together under one such systematic group. Under Tribe B, which in scope nearly agrees with F. E. Schulze's Inermia, I place the family Dactylocalycidre (made up of the genera Daclylocalyx, Mai'fjariteUa, Ifyliusia, Aulocalyx and Euryplegma) as well as all the lychniscophorous forms, both recent and fo.ssil. Schrammen (1. c.) though essentially in agreement with F. E. Schulze and with me as regards the principles of classification, stands in practice at variance with the view here advanced in that tiie Lychniscophora SciiR. is made by him into a suborder distinct from another, the Hexactinophora Schr., wdiicli latter is made up of the Tribes Aniphidiscophora, Hexasterophora and Uncinataria. That writer evidently lays undue weight on the lychnisc. This in my opinion is formed simply by the addition of peculiaidy arranged synapticuloe around the central node of hexactins composing an ordinary dictyonal skeleton. The Lychniscophora then seems to me to be just as much a Hexactinophora as any form referred by Schrammen to this group. And furthermore, it is certainly a Hexasterophora, as is proved by what we know of the living lychniscophorous genus Aitlocystis. Tribe C is exactly identical with F. E. Schulze's Uncinataria. The spicules called uncinates, from which the tr'.be received tlie name just mentioned, are peculiar in that they can not be proved to be secondarily derived from a triaxon owing to the absence of the axial cross, though the axial canal is present. A noteworthy fact it is that the same spicules occur also in certain Amphidiscopliora, though not in all. Tliis seems indicative of a near phylogenetic relation between the Lncinataria and the Aniphidiscophora, but just how it is sea: cely possible to determine, 26 ART. 1. 1. IJIMA : HEXACTIx^ELLTDA, III. apparently fundamentally the same as those seen in Euplectella, are found distributed all over the l)ody. At any rate, the system suffers no disadvantage if the Euplectellinte, consisting-, as it does, of only a single genus, be deprived of its doubtful status as a distinct subfamily ; and moreover, the group resulting from tlie above amalgamation seems to ]je a perfectly natural and well defined one, representing a Euplectellid phylum which has adapted itself to a special mode of attachment to the soft or loose sea- bottom. For this new group or sul)family, the name Enplectellinœ may however 1)0 retained l)ut in a new sense, as follows : Euplectellidse rooted in the substratum by a tuft of 1)asal spicules. Setting aside the three genera that make up the Euplectellinœ as defined above, the remaining genera (13 in number, vide the list on p. 20) are, assumably all and without exception, those forms which are directly and fii-mly fixed at base to the hard substratum, — prol)al)ly the j^rimitive mode of attachment of the Euplectellidje. This assemblage of genera includes all those that were referred to the TtEgerinaî by F. E. Schi^lze, besides others which have been left by him unassigned to any of the subfamilies as being too insufficiently known (F. E. Sch., '87, p. 99). In my last Contribution (Tj., '02, p. 30) I had, for what I considered a necessity l)y nsage, sul)stituted the name Corbitellinœ for that of T^egerina^, to which change I still adhere ; but the scope I had given to the subfamil}^, though on the whole much more extended than before, was in one respect narrowed, viz.y in that the genera Ilertwigia, Trachycaidus and Saccoccdyx, all placed by F. E. Schulze ('99, pp. 90-98) among his Ticgerinjc, were not included in my list of tlie Corbitellintie then given (Lt., /. c). It may OBSERVATIONS ON THE EUPLECTELLlDiE GENEEALLY. li suppleiiientarily be explained that tlie omission was made under the impression that a distinct subfamily should be instituted to receive the genera mentioned, together with certain others. I have since come to see that this is not practicable ; in fact, all the firmly seated Euplectellids known at present now seem to me to form a group, which admits of no subdivision so as to repi'esent more than one phylum. This is especially on account of the intricate manner in which certain highly characterized hexasters are distributed among, and combined in, the different genera. So that, after all, they can not but be left to stand as one sul)- family, the Corbitellimc, in contrast with the only other subfamily, the Euplectellinte, of the family under consideration. The Cor- bitellinse may then be characterized simply as : Euplectellidie firmly attached to the substratum l)y compact base. It goes without saving that the definition of the Corbitellimc as it stands in my Contriljution 11. and the list of the species appended thereto are rejectable, the former as going too much into particulars and the latter as Ijeing incomplete. Inseparably linked together as the Corbitelline genera ap- parently are, the series may nevertheless be said to show a general tendency of development in two directions. At the one end of the series may be placed those forms, which, like Itcgadrella , Gorbitella and the like, are of a tubular shape having the sieve- plate and possessing Ijoth the floricome and the graphiocome among the hexasters, EuplecteUa, and with it the Euplectellinre, represents in all probability an off-shoot by special adaptation somewhere from this end of the Corbitelline series. At the other end may l)e placed those forms, e. g., Victyaulus, Hertwigia and Saceocalyx, which, under variable external shap.es, have given rise to such special hexaster-modifications as the discospiraster, the 28 AET. 1.— I. IJIMA : HEXACÏINELLIDA, 111. codonliexaster, the aspidoplumicome, or the drepanocome, while the graphiocome, alone hy itself or together ^Yith the floricome, has either been lost or has never come into develoj^meiit. Placo- som.a paradictijiuii, as being in possession of codoiihexasters and wanting in both Horicomes and graphioconies, seems to have its position somewhere in the latter end of the Corbitelliue series, though it is ditHcult to indicate to which of the genera it is inost nearly related. The division of the Corbitellime somewhat in the manner hinted at al)Ove may become possible and even necessary in the future ; for a far wider knowledge than we at present possess is to be taken in prospect, when we remember that most of the genera are now known only in solitary species and that too, in so many cases, in a single specimen. LÈucoPSACiDi;. 29 LEUCOPSACID^e. In '98 (p. 41) I introduced into the system a grou]) by the name of Lencopsacinas to which the status of a subfamily under the Rossellidae was given. It is herewith made a distinct family with some change in its scope and definition. To the Leucopsacinjç, as a Rossellid subfamily, I originally referred the following six genera : 1. Leucopsacus I J. (With 2 species, L. orthodocus I J. and L. scoliodocus Ij,). 2. Cliamiopledella Ij. (With 2 species, C cavenioda Ij. and G. spinifera n. sp.). 3. Caulocalyx F. E. Sch. (With 1 species, C. teuer F. E. SCH.). 4. Placopegtna F. E. Sen. (With 2 species, P. sohUiim F. E. ScH. and P. sp.). o. Aulocalyx F. E. Sch. (With 1 species, A. irregularis F. E. Sch.). 6. Euryplegma F. E. Sch. (With 1 species, E, auriculare F. E. Sch.). I now think that the two last named genera, viz., Aulocalyx and Euryplegma, had better be separated from the group in question as well as from the Rossellidœ altogether and placed under a dictyonine family, Dactylocalycida^ together with Dac- tylocalyx, JLargaritella, and Myliusia^- I am led to this conclusion chiefly by reason of an essential agreement in the characters of * With 3[y(iusia GliAV ii i)rubiil)ly 'KhnucullScknpIc^ma O. ScHM. 30 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. the dermalia and of the parenchynialia made up of hexactiiis, and of hexactins only, which are quite extensively ankylosed except in growing parts of the body. It may be remarked that, through the two genera herewith united to the Dactylocalycida?, this family is brought into close relationship Avith the lyssacine family which I shall call the Leucoj^sacidœ. The four remaining genera, viz., Leucopsacus, Ckaunopleciella, Chulocalyx and Flacopegma, are all lyssacine forms, but should likewise, I think, be removed from the EossellidiB on account of a somewhat marked difference in the nature of the dermal skeleton, a point to which we give, I'ightly I believe, much weight in distinguishing the families. The removal would be an advantage to the system in this respect, that the Rossellidoe is thereby left a group much more uniformly characterized than before, in that its dermal skeleton may then be said to be uniformly composed of well differentiated, small, rough-surfaced dermalia and of nuicli larger, supporting spicules, the hypodermalia, which are generally pentactins much less specialized in characters fiom certain ])ar- enchymalia. On the other hand, the four genera in question have all relatively large peniaclin'w dermalia with which are associated no spicules that may be called the hypodermalia. Owing to this character they can not, in my opinion, very well be received into any known lyssacine family. However it inay be said that in general features of the dermal skeleton, as in fact in general spiculation, they most closely resemble, and there- fore sliow nearest relationship to, the Euplectellidie, but of course with this difterence that the dermalia lack the distally directed, tsixtli rays which are always present in those of the hunily just mentioned. If they should perforce be united to any of the accepted families, the union nuist be with the EuplectellicUe and LETTCOPSACID.E. ol not with any other family. The absence of distal ra3's to the dermalia, it may be said on theoretical grounds, is simply due to a secondary loss, which might be easily conceived if we remember the great variability of the corresponding spicules within the family Rossellidie. It may bo that the pentactinic dermalia, as I'epresented in quite young Regadrella okiiw^ea/ni (Lt., 'oi, p. 240) probably as the result of adaptation to certain secondary circumstances and which in this Euplectellid are soon overcrowded by later formed hexactinic forms, have in the present cases become permanent under the continuance of the same adaptive conditions. It is certainly not to l)e excluded that hexactinic dermalia develope in certain limited parts of the body- surface ; thus, in Placopegma solutum, according to F. E. Schulze ('95» P- ^'-^)j the dermalia on the oscular margin are hexactins, instead of pentactins as on the general surface. At all events it appears justifiable to assume that the four genera under considera- tion, whether separately or as a group, were derived either secondarily from the Euplectellidiç or from an early prototype of the same. With the progress of our knowledge in the future, it may become necessary to incorporate them all or in part in the family just referred to ; but meanwhile, I consider it expedient to keep them separate in a distinct family, to be designated the Leucopsacidie, even if only to avoid disturbing the integrity of the Euplectellidœ as already defined. As regards the genera Leucopsacus and Cliaunoplectella, it seems nothing stands in the way of regarding them as forming a systematically coherent group which may have very early diverged from the Euplectellid Corbitelline phylum. But now, in associating with tliem the genera Caulocalyx and Placopegma in one taxonomic group, I am not without misgivings as to whether 32 AET. 1. 1. T.TTMA : HEXACTINELLTDA, TTT. a ]3olypliyletic character is not thereby given to it. As to Gaulocalyx, the presence of aspidoplumicomes (F. E. Sch,, '97, p. o49) — the same peculiar hexasters as those found in Herhoigia and Saccocalyx — indicates its affinity with these highly organized Corbitellinae and so suggests tliat it had an origin later than, and independent of Leiicopsaeus and Ckaunoplectella. With respect to Placopegma, should F. E. Schulze's ('95, p. Q5) assumption of tlie presence of a basal anchoring tuft prove correct, the genus must probably be regarded as derived from the Eu- plectellime, and not from the Corbitellinœ as is the case with the other genera. The barbed anchor-needles that were discovered iu P. solutum have been assumed by F. E. Schulze (/. c.) to be pentactinic, each of the four anchor-teeth being taken for a real spicular ray ; l)ut this seems to me highly improbable. The short transversely disposed axial filaments forming a part of the central cross in the said needle (/. c, Taf. vi, Fig. !(>) are far from extending into the anchor-teeth and appear much too abortive to allow of these being interpreted as real rays, but they are only of such a degree of development as we see in a diactin or a monactin of a similar strength ; so that, I think, the anchor- needle is essentially comparable to that of the Euplectellinœ, save in this relatively unimportant respect that the central axial cross is brought down into the inferior swollen end, instead of being situated some distance above it. Under the above circumstances it is with a certain degree of reserve that I place Placopegma, and Gaulocalyx also, although with somewhat less care, under the Leucopsacid* together with Leucoinacui^ and Ghaunoplectella. I think the family may be made to stand on the strength of the two last mentioned genera ï\lone, if it should become necessary to remove the other two leucopsacid.t:. 33 from it. For the present I consider it preferable to include all the four genera in the Leucopsacidie, which may then be diagnosed as follo\YS : Lyssacine Hexasteropliora of tliick-walled, cup-like or ovoid body; sometimes stalked; firmly attached by base (? or rooted by basal spicules). Dermal skeleton composed as a rule of moderately large pentactins with the unpaired ray directed proximad; hypo- dermalia not distinguishable. Hexasters represented mainly by discohexasters (no oxyhexaster). A key to the genera and species, wliich should also show in a way the structural peculiarities of each, may properly be appended here. II. — With hexactinose forms among the discohexasters. Gastralia lioxactinic, snnilar to parenchymal hexactlns. Sponge-body small Leucopsacns. «1. — Other discohexasters with terminals in a distinct group to each principal. Paren- chymal hexactin with straight rays. Bodystalked ■. Leucopsucus orthndocua. (Sagami Sea). /y^— Other discohexasters generally .spherical in shape. Parenchymal hexactin with more or less bent rays. Not stalked... LcucopMcus scoUodocm. (Sagami Sea). h. — Without hexactinose discohexasters. Sponge-body of a considerable size. c^. — The larger discohexaster with terminal prongs arranged in a whorl like anchor- teeth. Gastralia hexactins Chaunoplectella. a". — Dermalia of not only pentactins but also other forms without distally project- ing rays. Discohexaster may be as large as to measure 250-400 y. in diameter. Small and delicate sigmatocomes present on the tip of the outstanding ray of canalaria Chaunoplectella cavernosa. (Sagami Sea). Z(-.— Dermalia always pentactins, some of which have paratangentials sparingly supplied with bent spines on the outer surface. Discohexaster not larger than 114 ;j. diameter. Sigmatocome not found Ouiunopledella spinifera. (Sagami Sea). f^.— The only or the larger kind of discohexaster with convex terminal disc, the margin of which is serrated. C-. — With aspidoplumicome. Gastralia hexactinic. Paratangentials of dermalia with curved spines laterally. Body stalked Caulocalyx. — C. lener. (W. of Tristan d'Acunha). 34 AKT. 1. T. TJI!\rA : HEXACTINELLTDA, III. d''. — Without aspidoplumicome. Gastralia pentactinic. Discohexaster in one kind Placopegma. a'.— Paratangentials of dernialia beset with small erect spines on the outer surface. Witli s'eve-plate and anchoring needles. Discohexaster up to 100;/. in diameter Placopegma sohUum. (B. of Bengal). h^. — Paratangentials of dermalia without spines. Discohexaster spherical, 140 \i. in diameter Placd-pegma sj)., F. E. Sen. *99, p- 44. (SW. of Vancouver Is.). I now proceed to give fall descriptions of the genera and species that occur in the Sagami Sea. LEU COPS ACUS=^= Ij. Ijima, '98, p. 42. Leucopsacids with small, ovoid or spindle-like body, which may be stalked. Parenchymalia chiefly hexactins ; diactinic parenchymalia present, but play a subordinate part. Gastralia represented by hexactins similar to those of the parenchyma. Discohexasters in part hexactinose and in part hexasterous. lEUCOPSACUS ORTHODOCUS j Ij- PI. III., figs. 14-26. Leucopsacifi^ orthodocus, Ijima, '98, p. 42. The species is based on two specimens, obtained by myself at different times at and near Doketsba from a depth of, say, 214-429 meters (117-235 faths.). * >.£vxbi:, white ; <\ia.Y.aç, drop, ■j" ôp8bç, straight ; cov.hç, bearn, tEUCOtSACUS ORTHODOCÜS. 3Ô The first specimen (Sei. Coll. Mus., Sp. No. 230) was discovered among the trophies of a long-lining expedition to Döketsba, which I had undertaken on Aug. G th, 1894, together with Professor Mitsukuei. I consider the exact date as not unimportant owing to the fact that the specimen contained larvic in different stages of development. It was found attached to a dead Caryopliyllia-VikQ coral in company with a specimen of Lanuginella pupa, and is shown in PI. Ill, fig. 14, in natural size. The body is of an ovoid shape, with the inferior narrower end continued into a stalk of moderate length. Total height, 11 nnn.; greatest breadth, (j.ö mm. The upper rounded end bears, somew^hat to one side of the center, a roundish osculum of 1.5 nnn. diameter. The stalk is laterally compressed, measuring in breadth 2 mm. in one direction and only 1 mm. in another ; it expands at the lower end into a small basal disc. The sponge-wall is about 1.5 mm. thick at the part where the body is broadest ; the oscular margin is thin and simple (fig. 15). So far as can be made out by cutting open the wall, the gastral cavity seems to extend downwards for a considerable distance into the stalk. The second specimen (Sp. No. 438) was obtained also by me, Aug. 12 th, 181J5, at a spot in or near Döketsba l)ut which could not be precisely located on account of the foggy weather. It is a fragment representing the upper part of an individual probably similarly shaped, but somewhat larger than the first specimen. It bears a roundish osculum of 2 mm. diameter. The wall is nearly 2 mm. thick in the thickest part. In general appearance both specimens remind one of Lanu- (jincUa pupa, long known from the Atlantic and which also occurs in the Sagami sea. The external surface is perfectly smooth. Examined under a hand-lens the apertures of incurrent canals 36 ATA\ 1. — I. IJIMA : HEXACTINELLIDA, III. ai^pear as indistinct minute spots beneath a thin and clear dermal layer. On the gastral side the excurrent canals open apparently directly into the cavity ; many of the openings are considerably larger than the incurrcnt canalar a])ertures seen on the outside. The sponge is soft and delicate in texture. In all essential points of the spiculation the two specimens agree closely with each other. In view of their small size, I at first considered it possible that these were young specimens, but the presence of large archœocyte-congeries in both and the discovery, in one specimen, of larvse which apparently belonged to it, decidedly support the view that thev are mature and fidl-grown individuals. Spiculation. The parenchymalla, forming the main framework of the sponge-w^all, are for the most part legular oxyhexactins of moderately large size ; diactins occur in a relatively small number. The oxyhexactins have straight tapering rays which may reach nearly half a millimeter in length and 10 f- in thickness close to the central node. There are all sizes leading down to the dimensions, given later, of those oxyhexactins which I consider as gastralia. The rays appear smooth under a low magnification but are in fact insignificantly rough-surfaced on account of minute tubercles occurring at rather wide intervals all over them. The o-eneral manner of arransrement of the hexactins in relation to one another is such that, while being disposed in several layers with one axis directed radially, they have each of the rays placed, for LEUCOPSACUS' ORTHODOCÜS. 37 nearly its entire length, side by side with or in more or less close apposition to, a ray of adjacent hexactins (PL III, fig. 26). The result is a parenchymal framework which, whether seen in a transverse or a longitudinal section, presents approximately rectangular or quadrate meshes bounded by straight beams ; hence the name 1 have given to the species. The beams consist usually of two, but sometimes of three, spicular rays running alongside but not always placed compactly together. It is however by no means infrequent to find some hexactins, situated in an indefinite relation to their neighbors, thus disturbing the regularity of the framework. This is no doubt due in a measure to the free state of each separate spicule. The comparatively few diactins that occur as parenchymalia are small forms, probably never more than 1.5 mm. in length and 8 /'• in breadth near the center, wliich is always marked externally by an annular swelling. The surface is smooth, but towards the gradually attenuated ends, is roughened by micro- tubercles sparsely present. More especially are the oxydiactins to be found in the deeper parts of the wall, running either isolatedly or comljined in weak bundles. In the basal disc and directly against the foreign surface to which the sponge is attached, the parenchymal hexactins form a thin, irregularly meshed, basicUdyonal plate. This is of essentially the same structure and appearance as that I have described from Eegadrella okinoseana and homeyainai (Contrib. I., pp. 231-232, 2(34). The medium-sized, rough hexactins, which go to compose it bv being soldered together either directlv rav to ray or by means of irregular synapticuUe, may have rays as thick as 16//. 38 ART. 1.— i. IJIMA : HEXACTINELtlDA, III. Tlie gaüralia differ in no way from the ordinary parenchymal oxyhexactins in their characters, except in being on the whole smaller and in having proportionally thinner rays. Axial length 1^80,« and over. It would not he improper to say that they are here represented simply by those parenchymal oxyhexactins which, being situated in the deepest part of the wall, project one of their rays into the gastral space. They are found rather irreo-ularly scattered and are far from forming a continuous gastral layer. Like the gastralia the dermalia are only slightly dilïerentiated from the parenchymal hexactins, except in lacking a distally directed, sixth ray. To be more explicit, the dermalia are exclusively moderately large oxypentactins with the rays supplied with sparsely distributed, quite insignificant microtubercles similarly as in parenchymal oxyhexactins. The cruciate paratangentials up to 270 /'- in length and 15 /^- in Ijreadth near the central node, are in either a perfectly flat or a slightly outwardly curved plane. The straight, unpaired, proximal ray is longer than, — often fully twice as long as, — the jiaratangential in the same spicule ; it generally runs in association with a radially directed beam of the parenchymal framework. Observed from the surface, the centers of the paratangential crosses lie separated from one another by a space nearly equal to, or considerably less than, the length of the paratangentials, — by a distance of about 200/-« on an average. Occasionally there are seen two centers, placed close together. At the same time the paratangentials form the usual, quadrate-meshed, dermal latticework, the beams of which are composed usually of two, but sometimes of one or three, rays running together. As is usual, the dermal meshwork is LEUCOPSACrS ORTITODOCrS. 39 by no means everywhere uniformly and regularly developed but shows at jilaces a greater or less deviation from the regular pattern. The hei'aster>i of the species may be said to be of two kinds, viz., the larger hexaetinose and the smaller hexasterous discohex- asters, both found commonly in the tissues of thechoanosome. The hexaetinose discohexaster (PL III., fig. 16) is essentially similar to that known to occur in the Euplectellid genus Corbi- iella (Ij., Contrib. II.) or to the spicule figured by F. E. Schulze from Ros^ella antarctica (Chall. Eep. PI. lv., fig. 8) but regarded by him as perhaps extrinsic and intruded. In the present species it measures 110-168 y- in axial length. The six, slender, smooth and straight arms, arising from the central node in exactly the same way as the rays in a regular hexactin, thicken slightly towards the outer end, which bears a convexly arched, anchor-like umbel of 3-5, usually 4, strong, recurved and sharply pointed teeth. These are, as measured from tip to umbel center, 14/^ long on an average. Special examination of the spicule mounted in glycerine, showed the central cross of axial filaments in the central node, the filaments not extending themselves beyond the base of each arm, precisely in the manner indicated in fig. 30, PI. III. The hexasterous diseohexaster (PI. Ill, figs. 17-20), i.e., the form in which each principal bears more than one terminal in distinction from the hexaetinose form, is somewhat variable as regards size and certain other points, not only in different in- dividuals but also in one and the same individual. It may be said in general that the most usual size is 60-75/^ in diameter. Jilçich short principal is supplied with a bejl-shaped, outwardlv 40 ART. ]. — T. um A : hexactixellida, m. expanding tiift of 4-8 (most commonly 7 or 8) terminals, which arise not always in a single circle but sometimes so that one is surrounded by others in a whorl. The terminals are of moderate strength and terminate each with a small star-like disc having 5 or 6, minute, marginal teeth ; their surface is obsoletely rough. The general shape of the discohexaster is quite often not unlike that of a floricome. In one (Sp. No. 438) of the two specimens, on which the species is based, I find the range of variability of the discohexaster in question somewhat wider than in the other. (PL III., figs. 18- 20). In that specimen there are occasionally some discohexasters that are so small as to measure only 50/^ in diameter, while others not infrequently reach up to 88 /^. The terminals are slightly thinner but often, though not always, more numerous (up to 11 in a tuft) than in the other specimen (Sp. No. 230). By the side of such discohexasters rather copiously supplied with terminals, there exceptionally occur others of a comparatively large size such as is represented in fig. 20, in which each principal is seen to bear only 2 or 3 terminals. No such rosette was found in Sp. No. 230. Notes on the Soft Parts and the Larva. As the specimens were killed and hardened by means of corrosive sublimate, the soft parts are preserved, not so satisfactorily as might be desired, but in a sufiiciently good condition to enable me to make the following observations. As in EupledeUa marslialli (Contrib. 1., p. 123), the dermal LEUCOPSACrS ORTTTODOCrP. 41 membrane is usually not membranously develojted but is represented bv fine, irre2;ularlv l)rancliin2; and anastomosins;, cobweb-like tbrends spread over tlie raeshwork formed by the paratangentials of the dermalia. The "pores" or the gnps inclosed by the threads are accordingly not rounded and ]^ore-like, but quite irregular in shape and size. The dermal memln-ane is therefore in no way distinguishable from the more deeply situated trabeculre, with which it is in fact directly continuous. In the subdermal space, which is nowhere of any great extent, trabecuhc are present in moderate abundance. The nuclei, found here and there in the little stained granular substance of the trabeculpe, measure not more than 2 /'- in diameter. They are moderately strongly colored without showing chromatic con- tents with any distinctness. Archiçocytes occur in abundance as usual, especially on and close to the outside of the flagellated chambers, either solitarily or grouped together in exceedingly variable number. They are 2-3? JJ- large. A strongly stained and externally well defined cytoplasm is to be ascribed to the cells, though always sparsely present. The nucleus is of about the same size as the trabecular nucleus, but o-enerallv incloses a somewhat refractive chromatic mass ; it is about as well stained as the cytoplasm, though ap- pearing clearer at a certain focus of the microscope. In Sp. Xo. 230, but not in the other specimen, I have found a number of large, ovum-like cells (dia. 20-40«), apparently freely occuring in the trabecular spaces. A description of these peculiar cells has already been given by me in my Contribution I. ('oi), p. 182, and need not be repeated. A figure of one is now given in PI. III., fig. 21. As to their real nature and import, I am still in the dark. 42 ART. h T. IJIMA : HEXACTINELLIDA, III. The flagellated chambers are cup-like or thimble-like. The more elongated chambers occur in the periphery of the choano- some. The flagella are not preserved in mv preparations. The chamber-wall consists of the usual membrana reticularis, the open meshes of which may measure 7^^ /^ or less across. At places, a pale ill-defined nucleus may be defined toleraldy constantly at every nodal point of the reticulum. The larva (PI. III., figs. 23-25). — As announced in my Contribution I. (pp. 182, 187), Leucopmcus oriliodocus is one of the two Hexactinellid species in which I have discovered devel- opmental stages of larvse, or at any rate of certain reproductive bodies which may reasonably be interpreted as such, — the other species being VitrolluJa fertile, a Eossellid to be described in full iu a future Contribution. It was in only 8p. No. 230 of the above mentionod Leucopsacid that I have found the said reproductive bodies. These had to be searched for in sections of the sponge and were by no means so numerous, nor so favorably conditioned for observation, as was desirable ; hence, the frag- mentary nature of the account given below. As before indicated, there exist, in liotli the sj)ecimens examined, variously sized archœocyte-congeries (PI. III., fig. 22) between the chambers or in the narrow spaces between the eva- ginations of the chamber-layer. To judge from appearances, they grow in size by multiplication of the compactly crowded cells. A large congeries is spherical or approximately so and may be nearly or quite as large as the body, shown in PI. III., fig. 23 and which I consider as the larva in an early stage of development. The larva in this stage is spherical, measuring about 57 /^ in tEÜCOPSACÜS ORTHODOCUS. 43 diameter. It shows a moderately tlnck epitlielial covering, in which the nuclei ;ire indistinctly visible but the cell-outlines scarcely visible at all, owing in a great measure to the diffuse and strong staining of the cells as well as to the thickness of the preparations. However, to judge from the arrangement of the nuclei, we here evidently have to do with a single-layered cylindrical epithelium. The epithelium appears on the whole somewhat clearer than the internal cell-mass from which it seems to be distinctly delimited. Whether it extends over all the surface in equal thickness or distinctness could not be determined with certainty. Often in one and the same larva it can not in fact be perceived Avith jis much distinctness in one part as in another ; this may be due as well as not to certain defects in the manner of preparation. Frequently, but not ahvays, a clear, irregularly granular layer is seen to cover the extei'ual surface. I consider that layer to have been brought about by the deteriora- tion of the flagella belonging to the epithelial cells, as the result of the hardening process. — The internal mass is diffusely and very strongly colored ; it shows closely crowded nuclei, around which cell-outlines can not be defined with any degree of dis- tinctness. A further insight into the histoloo'v could not be obtained, all my sections being much too thick for that. Under a moderately high power of magnification, the internal cell-mass appears densely and uniformly granular, exactly agreeing in all points with a larger archœocyte-congeries. From the latter, it may be said, the larva in the early stage now being described differs only in the presence of the peripheral epithelial layer. And there exists nothing besides such archaeocyte-congeries to which the origin of the developing larva can be traced back with any degree of probaljility. For the enigma which follows this 44 ART. 1. — î. IJIMA : HEXACTINELLIDA, III. way of interpreting the source of tlie larva, the reader is referred to my enunciations in my Contribution I., pp. 185-190. By the time the larve has gro\Yn to a size of 60-70 1'- diameter, the first spicules make their appearance (PI. III., fig. 24). These are minute and delicate-rayed oxystauractins, — not liexactins, contrary to what might be expected on a priori grounds. It does not necessarily follow from this that stauractins represent the most primitive form of Hexactinellid spicules. I simply consider that the stauractinic form of the spicules develop- ing first in the ontogeny is due to the suppression of one of the three primitively present axes, in adaptation to a cei'tain secondary condition of the larva — assumably to circumstances of the space in which the spicules develop themselves, as seems to hold true in a general way of all triaxonic spicules with a reduced number of rays. A close investigation of the central cross of the axial filaments — which however can not be undertaken with the objects in hand — will presumably reveal an abortive third filament re- presenting the suppressed axis. The said embryonal oxystauractins are situated in the periphery of the central cell-mass, with the plane of the four rays disposed paratangentially to the surface of the larva ; they lie not in direct contact with the external epithelium but well separated from its internal limit by a few cells of the internal mass. To give a connected account of what I have seen in several larv{:e, the oxystauractins are at first scattered singly ; they are by no means numerous in number and show no definite rule as to the manner of relative distribution, except in that they always occur in a single layer. They grow in size, apparently without increas- ing in number ; all seem to have taken origin nearly simultaneously and are therefore of aj)})roximately the same size. They soon LÉUCOPSACÙS ORTHODOCCS. 45 come to intersect one another with their elonsiHted rnvs. After reachino- a. certain size, each sino-le oxvstauractin is distinctly outwardly convex, in conformity with the rounded external surface of the larva. The convexity becomes more and more pronounced as the spicule grows larger, which takes place with comparatively greater rapidity than the growth of the larva in general size. The entire skeleton, considered apart from the soft parts, represent a hollow spherical basketwork composed of the loose oxystauractins. (In figs. 24 and 25, PL III., the soft parts are drawn as seen in optical sections, while the spicules put in are all those that could be seen in one-half of the larva by focussing the microscope up and down). In the stage shown in fig. 24, PL III., in which the aj^proximately spherical body measures about 70/'- in diameter and the spicules reach up to 30 !>■ or thereabout in axial length, the soft parts still appear to retain the same histological character as before the formation of s])icules. At places favorably situated for the observation, there is to be seen on the surface a granular coating, indicating the presence of flagella to the external epithelial layer. Further advanced stages than those just referred to were not discovered except in a single case, which is showai in fig. 25. In this larva, the bodv, still approximately spherical in shape, may measure nearly 100/'- in diameter. An exact measurement can not be made since the larva lies so mixed up in the soft tissues of the choanosome that it is difficult to precisely determine its external limit. Tlie epithelial covering, which should not Ijc wanting until after the larva should have fixed itself on a foreign ol)ject after liberation from the mother body, is entirely concealed from view. Most plainly visible is the hollow skeletal basketwork. 46 ART. 1. 1. IJIMA : HEXACTINELLiDA, 111. The oxystauractins composing this have now greatly grown in dimensions ; their smooth, gradnally tapering rays may be -57 ,« long. All that I can say concerning the soft j^arts is that the cells (or nuclei) within the hollow of the skeletal system are now arranged, not compactly crowded as before, but in irregularly reticular tracts, evidently on account of the formation of vacant sinus-like spaces. The above account of the larval development, incomplete as it is, will be corroborated and in a measure also supplemented by the description, to l)e given in another place, of the same process in Vitrollula fertile. LEUCOPSACUS SCOLIODOCUS '' Ij. PI. III., %. 27-37. Le2icopsacus scoliodocus, Ijima, '98, p. 43. This species is now known to me in more than a dozen specimens, all from the Sagami Sea and a depth of 400 hiro (313 fms. =572 m.) or thereabout. They represent fusiform, ovoid or globular, moderately thick- walled sacs, up to the size of a small acorn or a hazel-nut. The ^particulars about the specimens are as follows : Specimen No. 233 of the Sei. Coll. Mus. (the largest of the three shown in PI. III., fig. 27) is globular but somewhat laterally compressed. Height 17 mm.; breadth in the middle 10-13 mm., wall 4mm. thick in the thickest part. The constricted * (T/.oAiô:, curved ; oo/o:, Leaiu. • LEUCOPSACUS SCOLTODOCUS. 47 base is ßi mm. broad. This expands below into a firm, thick, basal disc, 11 mm. in diameter. The osculum at the upper end is romidish, measuring 5 mm. in diameter and with a thin and simple oscular margin. The sjDecimen was found attached on an undescribed Hexacl'mella , which shall be described in a later Contribution under the name of //. Joriea. Locality : Outside Okinose by Iwado-line. A bottle, numbered 23.5 in the 8ei. Coll. jMus., contains no less than nine specimens of the present species from the same locality, all found attached likewise to a piece of Hexactinella lorica. Two of them are shown in natural size right and left of fig. 27, PI. III. All the specimens are small, being fusiform or ovoid in shape and round in cross- section. The smallest is not larger than a grain of rice, while the largest is 13 mm. long and 6 mm. broad in the middle. The upper end is occupied by a round and simple osculum ; the opposite end terminates in a firm basidictyonal mass or plate, which may be very thin or of a considerable thickness. Fig. 28, PI. III., represents the appear- ance of one of the specimens as seen in a stained longitudinal section. On still another Hexactinella lorica (Sp. No. 448) from the same spot and a depth of 572 m., I have found several L. scoliodocus of varying sizes (2-13 mm. in height), in company with Lanuginella pupa, Staurocalyptus pleorhaphides, etc. Finally I have to mention a specimen from the northern side of Onigase (Sp. Xo. 434). It has the shape of a somewhat laterally compressed spindle, 20 mm. long and 11m. broad in the midde, in which region the wall is about 2 mm. thick. The truncated oscular end as well as the base measures about 4 mm. in breadth. 48 ART. 1. — I. IJIMA : HEXACTTNELLIDA, III. The dermal surface in all the specimens is smooth. In the j)rofile edge of the body, there is seen a clear space right under the dermal layer and separating this from the opaque choanosome ; it is the relatively widely developed subdermal space. The surface of the choanosome presents a spongy appearance on account of the ill-defined, variously sized but on the whole small, apertures to incurrent canals. The gastral surface is not covered with a continuous gastral layer, l-ut there directly open excurrent canals which may measure 1 mm. or more across in the larger sj)ecimens. Such a small sjiecimen as is represented on the right of fig. 27 may, in the wet state, be said to be nearly translucent all over excepting only the basidictyonal mass which appears whitish. None of my specimens is in a sufficiently good state of preservation for a histological study. Nevertheless, thus much could be determined, viz., that the trabecuhe are scantily and thinl}^ developed both in and below the bounding surfaces, and that the shape and arrangement of the chambers are much the same as in L. orlhodocus. Spiculation. The iparencliymalia consist of oxyhexactins and diactins, the latter occurring only very sparsely (PI. III., fig. 37). The parenchymal oxyhexactins may be said to be of a moderately large size, tliough subject to much variation in this respect. A large one may measure 4 mm. or moie in axial length and about 13 <'■ in breadth of ray near the central node ; but such large dimensions are attained by the spicule in question LEÜCOPSACUS SCOLIODOCUS. 49 only in the larger specimens of the species. The larger paren- chymal oxyhexactins lead over by forms of intermediate sizes to the small and slender-rayed gastral oxyhexactins, which measnre generally only 2()0-340 /-'- in axial length. The rays gradually attenuate outwards and are usually smooth except near the finely pointed ends which are more or less rough-surfaced. Only the smaller oxyhexactins, notably those which I consider to be gastralia, have rays rough all over on account of very minute and sparsely scattered tubercles. Both the parenchymal and the gastral oxyhexactins in the present species are characterized by the fact that the rays are seldom straight luit for more usually curved to a greater or less degree — sometimes gently and at other times in an arch-like or a wavy manner — apparently in no definite plane or direction. Accordingly, when they combine to form a bundle, as they sometimes do, this takes an irregular sinuous course ; hence the name I have given to the species. In this respect the species is at once distinguishable from L. orthodocus in which the parenchymal hexactins have straight rays and give rise to a skeletal framework with rectangular meshes. The parenchymal diactins are present in such a small number that in many slide-preparations of the wall they have to be specially searched for. However, they are to be constantly found in tolerable abundance, either isolatedly or in bundles, in the basal region of the body, where they pass out from among the basidictyonal mass and upwards into the body- wall. In size and characters the diactins agree with those of L. orthodocus. The dermal skeleton again is very similar to that of the species just referred to. It consists exclusively of moderately large oxypentactins, amongst which none can be distinguished as 50 AET. ]. — I. IJIMA : HEXACTINELLIDA, III. hypodermalia ; they are themselves but little specialized from the parenchymal oxyhexactins. It may be said in general that in length of rays they are about equal to the larger oxyhexactins of the parenchyma in the same individual. In most of the different individuals examined, I have found the length of the cruciately disposed paratangential axis to fluctuate between 500 /^ and 800 jj- ; in the large specimen (No. 434) from Onigase, the same axis reaches up to le mm. in length and 20/^ in breadth of ray near the central node. The unpaired proximal ray is always longer than, and often fully twice as long as, the paratangential of the same spicule ; it is always straight and dips inwards, generally in association with the radially directed axis of a parenchymal oxyhexactin. The paratangentials, as seen in surface view, are straight ; the latticework formed by them is on the whole irregular, though in places an aj^proach is shown to the formation of rectangular meshes. In lateral views the paratan- gentials are either likewise straight or so curved as to accommodate themselves to the curvature of the external body-surface. As in parenchymal oxyhexactins, the rays taper towards their ends, near which the otherwise smooth surface is more or less roughened by the presence of obsolete microtubercles. The ff astral oxyhexactins, already described in passing, occur in abundance on the interual surface, without however showing any definite order in their relations with one another or forming a distinct layer by themselves. Exactly similar oxyhexactins often occur also as canalaria along the lumen of excurrent canals. The basidiclyonal plate or mass is composed, as usual, of synapticularly fused, thick-rayed hexactins, the rays of which are LEUCOPSACUS SCOLIODOCÜS. 51 beset with prickles on the surface. The hexactins may be so large as to measure 120/^ in axial leugth and 10/>« in thickness of rays. At places the prickles on the basidictyonal beams are elongated into stout, sharply pointed spines, as much as oO /^ in length. As constant hexasters of the species are to be mentioned disc- ohexasters of both the hexactinose and the hexasterous varieties. The hexactinose discohexasters (PI. III., figs. 29 and 30) are shaped exactly like those of L. orthodocus. They are of common occurrence everywhere in the body-wall ; only in the specimen (No. 434) from Onigase were they found scantily represented. Axial length, 100-180 /a Number of terminal anchor-teeth, usually 3 or 4, sometimes 5, in a whorl. The limited extent of axial threads in the spicule, as ascertained by special examination, is depicted in fig. 30. Of the hexasterous discohexasters, the most constant are the forms shown in PI. 111., figs. 32-34. These are especially abundantly met with in the periphery of the wall. In diameter they commonly measure 70 /^, though sometimes only tis much as 46 /^. In the larger discohexaster each very short principal usually bears 8 or more, rough-surfaced, moderately strong-looking terminals, while in the smaller one there may be only 4 terminals to a principal. The small, convex, terminal disc is provided with 5 or more, minute, marginal teeth. The terminals from all the six principals radiate in such a w^ay that the terminal discs are all situated nearly equidistant from one another, the result being the spherical shape of the entire spicule. Here I see another point of difference from L. orthodocus, in which the corresponding discohexaster has the terminals arising from each principal arranged in a separate perianth-like tuft. — As variations 52 ART. 1. 1. IJIMA : HEXACTINELLIDA, III. of the discohexaster under consideration are found occasionally such forms as resemble the two shown in figs. 35 and 36, PI. III. They grade over to the more usual spherical discohexasters through intermediate forms. Those, of which the one figured iu fig. 35 may be taken as a representative, are unusually large (up to 114/^ diameter) and have long, slender terminals, 7 or 8 in number to each principal. Spicules like fig. 36 are more rarely seen ; in these, each principal possesses usually three somewhat bent terminals, the discs of which have marginal teeth appre- ciably larger than usual. Of inconstant occurrence in the species is the small and delicate form of hexasters, represented in fig. 31, PI. III. AVe have here to deal with a rosette which is very much like a floricome in appearance but differs from it in the fact that the terminals end n insignificant pinhead-like knobs, instead of toothed plates. For the sake of reference we may call it a tylfloriconie. Diameter 38- 50 /^. Principal slender ; in length about 1 that of the entire ray, or shorter. Terminals very fine, slightly thickened towards the outer knobbed end ; 7-10 in a whorl to each perianth, which is narrow but outflaring at the outer end. I first became aware of the presence of the tylfloricomes in the specimen (No. 434) from Onigase, in which they are tolerably common, especially near the gastral surface. Not infrequently a tylfloricome is found shifted right to, and hanging on, the tip of the freely projecting ray of a gastralia, after the manner of floricomes on Euplectellid dermalia. A subsequent search in the Okinose specimens revealed the oc- currence of the same hexaster, though never more than sparsely, while in some cases it was entirely absent. CHAUNOPLECTELLA CAVERNOSA. 53 CHAUNOPLECTELLA == Ij. Ijima, 'g6, p. 250 ; '98, p. 43. Leucopsacids \Yitli moderately large, ovoid or vase-like and tliick-walled body, attached by short stalk-like base. Parenchy- malia cliiefly liexactiiis and diactins. Dermalia either all pent- actinic or with a varial)le number of rays, none of which however are distally outstanding. Gastralia represented by hexactins similar to those of the parenchyma. Discohexasters always hexasterous, the larger ones with terminal prongs arranged in a whorl like anchor-teeth ; with or witliout sigmatocome in addition. CHAUNOPLECTELLA CAVERNOSA Ij. PL IV., and PI. v., figs. 8-13. Cliauaopleclella cavernom, Ijima, '96, p. 200 ; '98, p. 43. Since my first description of this species which was based in 1896 on a single fragmentary specimen, I have liad opportunities to examine nearly a dozen specimens of various sizes, all preserved complete in the dried state. They all came from the Sagami Sea, though in only two cases can the collecting ground be more exactly specified ; viz., the vicinity of Okinose in one case, and Maye-no-Yodomi, 572 m. (313 faths.), in the other. * yauvo:, loose ; TJ.v/.-.rj;, interwoven. 54 AET. 1. — I. IJIMA : HEXACTINELLIDA, III. In general tbe sponge is ovoid and goblet-like or more elongate and vase-like. Total height, up to 200 mm. or more. Inferiorly it is attached to the hard substratum by a short, thick, stalk-like base. The wall is thick and consists of remarkably loosely interwoven tissues. It presents quite a cavernous appear- ance on account of the wide spaces within it. Hence the generic and specific name I have chosen for the sponge. To make special mention of a few specimens, the one shown in PL IV., fig. 1, reduced to one-third the natural size, is the fragment (Sei. Coll. Mus. No. 443) from which I originally described the species. It is a part of a large individual, with the wall as thick as 52 mm. and measuring not less than 160 mm. in diameter of body as judged from the curvature of the external surface. The piece includes a portion of the stalk-like base. In general shape the entire individual must have closely resembled the one shown in the woodcut given below. This complete and beautiftdly preserved speci- men belonged to Mr. Alan Owston (O. C.='= No. 106). i Total height, 185 mm. I Diameter in the middle of body, 126 mm. Wall in the middle, 30-44 nnn. thick. The osculum at the U])per end was oval, measuring 47 mm. by 63 A complete specimen Chaunnplec- • t . j_i .7/ //^ n AT 1^/^^ Ü 1 1 i^^ii^- 111 diameter; the ieila cavernosa (O. L. JNo. lUoj. i\euuceu ' to i natural size. margin was thick and * O. C. stands for Owston Collection. CHAUNOPLECTELLA CAVERNOSA. 55 rather obtusely edged, without fringing spicules. The gastral cavity was tubular, 105 mm. deep and somewhat narrowed at the bottom. The attachment-surface of the base was covered over with a compact basidictyonal plate, 2 mm. thick, from which the loose parenchymal spicules could be easily se2:)arated without injuring their ends. Another ovoid and comj^lete specimen I examined was 190 mm. high and 15 mm. broad. It was attached to a smooth loose stone. Some specimens show a more elongate and vase-like shape ; they may be cylindrical or more or less laterally compressed. Further, the basal end may sometimes be bent, probably as the result of the sponge happening to grow upon a perpendicular surface. — The tallest specimen I have seen was an erect, laterally compressed, vasiform individual, 200 mm. high. Another elongate but cylindrical specimen (190 mm. long) was remarkable for the fact that it grew with a bent base upon a piece of porcelain ware, a part of an old-fiishioned Japanese oil-burner, that had probably been dropped from a sea-faring junk. PI. v., fig. ]0, represents in half natural size a strongly laterally compressed, pouch-like specimen, exceptionally w^ell preserved as to the delicate spicular texture. It belonged to Mr. Alan Owston (O. C. No. 4386). Height, 156 mm. Breadth, 125 mm. on one side and 65 mm. on the other. Wall, 30 mm. thick in the thickest part. Osculum, irregularly oval, 60 mm. in longest diameter ; its margin, sharp but not thin. The stalk-like base is bent towards one side ; this bending, it wall be noticed, is in the sagittal plane of the compressed body, which fact is the rule wntli all lyssacine Hexactinellids having a laterally compressed body and bent base. 5ß AET. 1. 1. IJIMA : HEXACTINELLIDA, III. Here I must turn to a consideration of certain small and young specimens which seem to offer some points of interest. On that most remarkable skeleton of a dead Ghonelasma calyx, which I have had occasion to mention and figure in Contribution I. of this series of studies (Ij., 'oi, pp. 25, 31), I have found, amongst the host of other animals attached to it, five young of the present species in different stages of growth (Sei. Coll. Mus. No. 407). Four of these are shown in PI. V., figs. 8 and 9, in natural size. The three specimens to be seen in the latter figure are all globular in shape, each with a small roundish osculum at the upper end. The smallest of the lot measures only 3 mm. in body-diameter. They all sit together on a common, compact, basidictyonal plate of irregular outline and disproportionally large size. The undulating surface of the plate is finely granulated and shows some furrows in it ; below, its substance permeates the underlying Chonelasma skeleton to a depth of 2 mm. or more. There can be no doubt whatever that the plate in question belongs to, and is the product of, the little sponges growing on it. — The fourth specimen, which is not figured, is of an elongate ovoid shape, 9 mm. in length ; the base, only 2 mm. thick, stands out from the center of an irregularly expanded basidictyonal disc of about 9 mm. diameter. — The fifth and largest specimen is the one shown in fig. 8. It has the form of a thick- walled cup, 31 nnn. high. The broad base joins the basidictyonal plate — which clasps a parietal process of the Chonelasma and at one place stretches out fully 10 mm. beyond the apparent basal edge of the body proper — making a distinct line of demarcation, which is due to the circumstance that here the loose tissue of the sponge- wall passes abruptly into the rigid reticulum of the plate. — The relatively large size of the basidictyonal plate in the above specimens CHAUNOPLECTELLA CAVERNOSA. 57 leads one to the supposition that its growth takes place with eonsiderahle — so to say, precocious — activity in the early period of post-larval life. Later, the rate of its gT0\Yth seems to more closely coincide with that of the body, so that the early ratio in size of the two parts is not maintained in after-life. The external sur&ce of the sponge is rather uneven. The dermal layer covering it is of a somewhat unusual spicular composition. As seen under a hand-lens, it consists of irregularly interlaced beams, running in quite indefinite directions in the plane of the surface (PI. IV., fig. 7). The beams, though thin on the whole, are of various degrees of strength, the strongest being as much as i mm. in thickness. The thinnest beam con- sists of no more than a single spicular ray, while others are made up of a varying number of rays grouped together into a more or less compact-looking bundle. Xot infrequently, a beam, springing out from another as a branch, is seen to terminate freely, without reaching up to that towards which it is directed. Altogether the dermal latticework is irregular. The meshes often have sides of 1 mm. and more in length, and must be said to be on the whole coarse, besides being irregular in shape. — The meshes are covered with a sieve-like dermal membrane, made up of membranously flattened trabeculœ around and between the closely disposed, roundish pores (PI. IV., fig. 7). The want of regularity in the arrangement of dermal spicules and the coarseness of the meshes serve as a convenient mark of distinction between this species and certain Rossellids of a bewilderingly similar external appearance, e. g., Staurocalyptus glaber Ij. On the surface of the stalk-like base, the dermal laver is 58 ART. 1. — T. TJIMA : HEXACTINELLIDA, III. usually wanting, thus directly exposing the thick, wavy, soft and silky bundles of the parenchymalia. An unusually wide subdermal space separates the dermal layer from the choanosorae (see PL IV., fig. 4). In the larger specimens it may in places be nearly 10 mm. wide. Pillars of conical or irregular shape project from the choanosomal surface at intervals of 5-12 mm. or more ; these divide at the apex into a number of fibrous bundles which go to join those of the dermal layer. The oval or round entrances into the incurrent canals are more or less conspicuously visible through the dermal layer. They are on the whole very large, though by no means uniform in this respect, either in the same specimen or in differently sized individuals. In the specimen of fig. 1, PL IV., some of the incurrent canalar openings have a width of 15 mm. or even more ; in that shown in the woodcut on p. 54, the largest opening measured 10 mm. across. In all large specimens there exists a continuous gastral layer, covering up the wide apertures of the excurrent canals. This is of nnich the same irregularly reticular appearance as the dermal layer ; only it may be said that in general the meshss are somewhat wider, while the intersecting points of the beams strike the eye as small whitish knots due to accumulations of certain discohexasters. In the interapertural spaces the layer lies closely over, and is often indistinguishable from, the parenchymal tissues. In the medium-sized specimen of fig. 10, PL V., the gastral layer leaves a few of the excurrent apertures uncovered, seemingly not as a result of artificial disturbance. In small specimens, such as are depicted in figs. 8 and 9, PL V., all the excurrent canals open directly into the gastral cavity. The gastral layer then CHAUNOPLECTELLA CAVERNOSA. 59 appears to be a structure which begins to be formed at a com- paratively late stage of life. The excurrent canalar apertures as compared with the incurreiit, are on the whole somewhat larger but less numerous. This is evidently in correlation with the difference in extent between the inner and the outer choanosoraal surfaces respectively occupied by the two sorts of apertures. The fact just mentioned may be laid down as a rule holding good for a large number of Hexactinellids of similar shape. The fineness of all the spicules, the wide subdermal space and the broad canals — both incurrent and excurrent — separated by thin septal walls, all combine to give to the sponge a light, delicate and cavernous character, which is especially pronounced in specimens of larger size (fig. 4, PI. IV). Spiculation. The following account of the spiculation refers, unless other- wise stated, to large and full-grown individuals as represented chiefly by the fragment shown in PI. IV., fig. 1. It may be regarded as holding good also for young specimens, but these are not wdthout noteworthy points of difference, of which special mention is required. The parenchymalia are, it may be said, mainly hexactins and diactins, although those with 3-5 rays are not uncommon. They are also very variable in dimensions. The rays are smooth, gradually tapering and terminate either simply pointed or with a 60 AÈT. 1. — I. IJIMA : HEXACÏINELLIDA, III. slight siibterminal swelling. The surface for a sliort distance from the end is always roughened by the presence of niicrotu- bercles. A parenchymal hexactin may attain a considerable size. One of the largest I have picked out from a large specimen had rays as long as 7 mm. and 70 !'■ thick near the central node. But the majority are much smaller and more slender-rayed, leading down to such oxyhexactins as will later be described as the canalaria (PI. IV., fig. 8; PI. V., fig. 12). The hexactins, when of a small size, present a regular or nearly regular appearance, but are otherwise more or less irregular, not only in that the rays are bent — sometimes strongly bent — but often also in having rays of unequal lengths (PI. IV., fig. 6). This inequality may be sufficient to give a stump-like appearance to some lays in com- parison wdth the others in the same spicule. It even leads over to cases in which one or more rays are reduced to total atrophy ; so that, besides the hexactins and diactins there are to be met with among the parenchymalia such intermediate forms as are to be called pentactins, tetractins and triactins (PI. IV., fig. 5, h-d ; PI. v., fig. 12). The tetractins are represented either by' staur- actins or by those formed by suppression of two rays belonging to difterent axes. The triactins are usually tauactins, seldom the other form composed of rays representing three half-axes. — The abortive development as well as the crooked state of the rays in many of the above parenchymalia evidently stands in relation to the thinness of the choanosomal septa, which ill affords sufficient space for their free and natural development. Nevertheless, parenchymalia are occasionally found which project one or more of the rays l)eyond the septal surface and freely into canalar lumen. For instance, in fig. 12, PI. V., which represents the CHAUNOPLECTELLÄ. CAVERNOSA. 6l sj)iculation of the septal wall as seen in a section, a parenchymal tauactin is seen to send out beyond the surface its short unpaired ray, which, like the outstanding rays of canalar oxyhexactins, is supplied with a sigmatocome at the tip. In small specimens of the species, such as are shown in fig. 9, PI. v., all the parenchymal hexactins, which are certainly never so large as in full-grown individuals, are nearly regular in form. In most of them the rays are almost straight or but little bent, and do not show striking differences in their length. Spicules with 3-5 rays are not present or are, at any rate, quite scarce, so that the parenchymalia may be said to consist largely at least, if not entirely, of hexactins and diactins. The parenchymal diactins are represented by all sizes, from those of the dimensions of ordinary comitalia up to those with a length of 12 mm. and a breadth of 50/^- in the middle. But the thickness never reaches the degree attained by the rays in some hexactins. The center is sometimes externally marked by an annular swelling, but more generally it is not. The diactins are disposed either more or less isolatedly or in bundles, generally in company with other kinds of parenchymal spicules. They are relatively few in number in the main body of the sponge, where by far the greater part of the- parenchymalia consist of spicules with more than two rays. The relative proportion becomes however gradually reversed in the lower region of the body. Here, but more especially in the stalk-like base, the parenchymal fibers form anastomosing bundles of very considerable thickness and of exquisitely silky appearance (see PI. V., fig. 10). The elements of these bundles are preponderatingly diactins, inter- spersed amongst which ure found small slender-rayed hexactins. The diactins at the inferior end of the bundles are inserted 62 ART. 1. — I. IJIMA : HEXACTINELLTDA, III. into the interstices of the basidictyonal plate, without becoming soldered to the latter. The basidiclyonalia (PI. V., fig. 13) form a dense, irregularly meshed reticulum of beams, the surface of which is thickly beset with sharply pointed conical spines. The beams may be 40 1^ thick; the spines, as long as 15/^-. Notwithstanding the irregu- larity in the arrangement of the beams, it is not difficult to make out that these are fundamentally nothing else than synap- ticularly fused rays of hexactins and sometimes of pentactins also. The rays in these basidictyonal spicules do not exceed 135 /^ in length. Sometimes small, thick-rayed and nearly or quite smooth- surfaced "hexactins are met with, lying free in close proximity to the spiny beams. They seem to represent early stages in the formation of basidictyonalia, before the soldering together has set in. This takes place wherever the sj)icules come in contact with one another. The spiny processes on a beam may grow so as to touch and fuse wâtli another beam lying close by, thus transforming themselves into synapticulœ. (In fig. 13, PL V., is seen a spiny siliceous ring standing in connection with basidictyonal 1 beams. This is an accidental formation, without doubt due to the same siliceous secretion, as thai: which is added to the surface of the beams, having taken place around some round object which no longer remains in the preparation). As regards the dennalia, it would be well to mention first those of quite young specimens (PI. V., fig. 9). In these, they ai-e nearly all, if not exclusively, pentactins of approximately the same ray-length as an average-sized parenchymal hexactin in the same individual. The unpaired ray is of course directed proximad. CHAUNOPLECTELLA CAVERNOSA. 63 The paratangeiitials are nearly straight; sometimes they are seen to exhibit an uneven surface throughout their entire length, due to the presence of obsolete warty prominences. Tlius, in its composition and in the arrangement of the elements, the dermal layer is here essentially the same as I know it in Leucopsacus or in Ghaunoplectella spinifera. But a very remarkable difference in the spicular composition of the dermal Inyer is presented by large individuals. With the growth of the sponge, it seems the dermalia are constantly supplemented by spicules genetically belonging to parts directly underlying the dermal membrane. In other words, a large number of peripherally situated parenchymalia are apparently taken up, as it were, into the constituency of the dermal skeleton. The process seems to l)e not without analogy in other Hexactinellids. Thus, in some Euj)lectellids certain hexactinic spicules, which have taken their origin right among the parenchymalia, show indications of being shifted on to the surface, to be taken into the rank of the dermalia (Contrib. I., pp. 47, 74, 235) ; further, the so-called liypodermalia are in all cases spicules which are apparently most nearly related to parenchymalia but have gone by adaptation into the support of the dermal layer. But it must not be supposed from this that certain spicules in the dermal skeleton of mature Ckaunopleclella caveimosa are to be regarded as liypodermalia and the rest, as dermalia proper. The fact is, at any rate, that none of the dermalia in any stage of the sponge's growth can be distinguished as liypodermalia, a point common to all the Leucopsacids as well as to the Euplectellids in general. In surface-view preparations of the dermal layer taken from large specimens (PL lY., fig. 7), tlie spiculation resembles in a measure the parenchymalia as seen in choanosomal septa. This 64 ÄHT. 1. 1. IJIMA : HEXACTIXELLIDA, III. means about the same as to say that the dermalia are but slightly differentiated from the parenchymalia. Like these they are on the whole large, or moderately large ; quite variable in the number of rays ; and often irregular in shape, not only in having rays of different lengths in the same spicule but also in that these are more or less bent. A large dermalia may show dimen- sions nearly equal to those of the largest parenchymalia. As regards the number of mys, which in character quite agree with those of the parenchymalia, the dermalia are commonly pentactinic, tetractinic or triactinic, and sometimes even hexactinic or diactinic. Hexactinic dermalia are generally so situated that the central node lies a short distance below the dermal surface. Four of the rays, representing two axes, run paratangentially and usually associate, soon after their origin from the central node, with other dermalia to form bundles of varying strength. A fifth ray is directed proximad, while the opposite distal ray is either so short that it never projects beyond the dermal surface or is otherwise so bent as to pursue a paratangential course in the dermal layer. Such hexactinic forms occur only occasionally : they are of interest as occupying a position which may be said to be still j)artly parenchymal. — Pentactinic dermalia usually have the four cruciate rays disposed paratangentially, the fifth unpaired ray dipping inwards into the pillars. Occasionally tlie spicule may be so unnaturally flattened by the bending of its rays that all five are taken up into the dermal layer, the surface of which is thereby made more or less uneven. — When tetractinic, the dermalia are either stauractins or of the form which shows one complete axis and two half-axes. The stauractins generally lie in the plane of the layer with all the four rays. In the case of the other form of tetractins, either one of the unpaired rays may be directed CHAUNOPLECTELLA CAVERNOSA. 65 proxiraad and the rest run paratangentially, or all the four rays alike may support the dermal memhrane in that the t\YO unpaired rays are forced apart from each other so as to form an angle of more than 90° between them. — Triactinic dermalia are most generally in the form of tauactins. Seldom was the form represent- ed by three half-axes met with. Except that a ray is never directed distad, the triactins may lie in all sorts of positions as regards the directions of the rays. — Diactinic elements of the dermal spicules are of the uniaxial form. They are always slender and comitalia-like, occurring but occasionally as components of spicular bundles in the dermal layer. — Altogether the dermalia are irregular in shape and the latticework formed by them is likewise irregular in appearance. The gastral layer closely resembles the dermal in its spicular structure, except in the fact that hexactins with a freely projecting proximal ray are here of somewhat common occurrence. In the smaller specimens of the species, the hexactinic gastralia are below the medium size and have all the rays nearly equally long; they are comparable in all respects to the canalar oxyhexactins soon to be described. In the larger individuals, the same spicules are much larger, being about as large as the dermalia in the same specimen, and have rays of unequal length. The free proximal ray is always much shorter than most others in the same spicule. Oxyhexactinic canalaria of rather small size line the walls of both incurrent and excurrent canals in irregular distribution (PI. IV., fig. 8; PI. v., fig. 12). They are nearly regular in shape and measure 200 /-< and upward (mostly about 500 /^-) in 66 ART. 1. 1. IJIMA : HEXACTINELLÎDA, III. axial length. Rays thin, smooth throughout, gradually attenuat- ing, nearly straight or slightly bent. One of the rays always stands out freely from the septal surface, and ^Y]lere the septum is not sufficiently thick to inclose the entire length of the opposite ray, this may also project from its other surface to a greater or less extent. As before indicated, a sharp distinction can not be drawn between the canalaria and the hexactinic parenchymalia. The hexasters are, broadly speaking, of two kinds, viz., disc- ohexasters and sigmatocomes. The discohexasters occur in abundance everywhere in the body except in the dermal layer. Of them I may distinguish three varieties or forms which I shall designate with the letters a, h and G. All these mero'e into one another throueh forms of in- termediate shapes and sizes. They occur in different quantitative proportions and also show certain differences in the manner of relative distribution in the body, according to the size of the sponge. As will directly be more fully pointed out, the three forms seem to represent in a great measure different developmental stages of one and the same kind of discohexaster, — stages passed through by it during the post-larval growth of the sponge. Hence it may happen that a quite young sponge lacks discohexasters in the older phase of their development, and that a mature one, on the other hand, is either wanting in those representing their younger phase or shows these in but a limited number ; whereas, all the developmental phases are numerously and constantly met with in individuals of certain intermediate ages. Form a, to begin with that discohexaster-phase which seems to represent the earliest stage of development, comprises the smallest discohexasters of the species. Diameter, commonly 100- CHAUNOPLECrELLA. CAVERNOSA. 67 120 /^ but may lead down to 50/'. In general appearance the discohexaster is very much like those I have figured from Lanuginella ijvj)a in PL V., figs. 1, 2 and 4-ß, or from Chaunoplectella spinifera in PL V., figs, lö and IG. In fact the two last mentioned figures may just as well be considered as representing the discohexaster-form in question from Ch. cavernosa. From the end of each very short principal there arise three or four, obsoletely tubercled or nearly smooth terminals, which are each capped with a small convex terminal disc, pro\'ided with 7-9, minute marginal teeth. The terminals to each principal diverge in such a manner that they do not form a separate bunch but give to the entire rosette an approximately spherical shape. — The form occurs numerously in all parts of the body in the three small specimens (under 11 mm. height) shown in PL v., fig. 9. It is somewhat more scarce in the nut-sized specimen (ol mm. high) of PL \., fig. 8, though quite common in the meshes of its basidictyoual plate. In three much larger specimens (above 156 mm. in height) specially examined in respect of the quantitative proportion of différent rosettes, the form a was found missing, or at any rate exceedingly rare in the sponge-body proper, though still commonly present in the basidictyoual plate remaining on one (O. C. Ko. 106) of the said specimens. Form b represents intermediate phases between forms a and c. It is larger than form a, measuring on an average, say, 200 ,« in diameter. In general appearance it is quite like the rosette I have figured in fig. 14, PL V., from Chaunoplectella sjyinifera. From the swollen end of each stout but very short principal, there arise 2-4, long, slender and obsoletely rough or nearly smooth terminals, which so\^diverge as to give a spherical shape to the entire spicule. The terminal disc, it may be said, is made 68 AKT. 1. — I. IJIilA : HEXACTINELLIDA, III. lip of 4-8, radially arranged, recurved teeth, whieli give a w^atchglass-like or hemispherical oultine to the disc. — The form in question occurs sparsely in the small specimens of PL V., fig. 9. It is common in the nut-sized specimen of PI. V., fig. 8, while in still larger specimens it is either scarce again or is not found at all. Form c is the most characteristic and the most constant of the discohexasters in specimens of the species that have attained a growth beyond the nut-size. A good idea of its appearance may be obtained from PL IV., fig. 9. It is of a very large size, though subject to a considerable variation in this respect like the other forms of discohexasters. It commonly measures 240-340/^ in diameter ; in the larger specimens of the species it may sometimes be even so large as to measure 400 /^ in diameter. The short principals are usually, though not always, so thickened that each presents a rounded knob-like appearance. The long and slender terminals, 2-4 (usually 3) in number to each principal, are strongly divergent and not always straight in their course. They are smooth and perceptibly thickened towards both ends but somewhat more towards the outer end, whicli bears a reverted umbel of 4-G, long, anchor-arm-like prongs. This terminal umbel gives to the hexaster a very striking appearance. The cupola of the umbel is rounded. The sharply pointed prongs may be 30/^ long ; unlike those in a codouhexaster, they are often more or less bent in an irregular way and proceed divergingly backwards from the cupola, so that the umbel assumes the form of a bell with a flarins; rim. — The above discohexaster-form undoubtedlv represents the most advanced stage of development undergone by the discohexasters of the species. It is still undeveloped in the three smallest specimens depicted in PL V., fig. 1). With tolerable CHAÜNOPLECTELLA CAVERXOSA. 69 frequency it occurs in the nut-sized inclividiial of fig. 8 in the same plate. And in all still larger specimens it occurs very commonly. . In fact, it is at least the predominant, if not the only, discohexaster-form to be seen in the parenchyma of all full-grown specimens. It is usually most abundantly found among the beams of the gastral layer where it is often seen in patch- like congeries. The following table may serve to show the relative proportion in which the three above- described discohexaster-forms occur in differently sized individuals of the species : SpecimeD. Size of specimen. Discohexaster Form h. Form c. c. ^^^'^ "■ ,., Moderately large 1 Largest form Smallest form like f^,^.,,, 1^,^ 'tig. 14, like fig. 9, PI. figs. 4-6, PI. \ . PI. V. IV. The o snuill specimens shown in fig. 9, PI. V. 3-11 mm. Numerous in both , (Sei. Coli. Mus. No. ! liigli. ^»?tl^' ^}}^^ basi- Few. | ^ot found. 407). 1 dictyonalia. i i Tlie nut-sized speci- men shown in fig. S, PI. V. (Sei. Coll. Mus. Xo. 407). 31 mm. F*^^' though com- hio-li ™"" "^ basidictvo- Common. Common. " ■ , ualia. ' ; The specimen shown in fig. 10, PI. Y. (0. C. No. 4386). 1 156 mm. ^'ot found. (Basi- i ' hi"h. dictyonalia not , Not found. Very common. ° ' 1 preserved). \ 1 ! 1 The specimen shown in tlie woodcut on p. 54 (0. C. No. 106). 185 mm. high. Not found in body, but present in Rare. basidictyonalia. Very common. The specimen sliown , Vprv in fig. ],P1.IV. (Sei. 1 1^,4 Coll. Mus. No. 443). ° Not found. (Basi- dictyonalia not ' Rare. Very common, preserved). The above data seem to allow the following general observa- tions to be made. In an early stage of the sponge's growth the discohexasters are mostly of the form a, only a few of the form 70 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. h being found and as yet none of the most highly developed form c. AVith the growth of the sponge, a diminishes in number and finally disappears altogether in the parenchyma, apparently as the result of its transformation into b ; thougli it seems to persist in the original condition in the basidictyonal plate. At the same time, h in its turn is constantly developing into c, and when this development is quite or nearly completed as in all the larger specimens, the latter form becomes almost the only one that is to be met wdth in the parenchyma. The second kind of hexasters constantly present in the species may be called the sigmatocome (PI. IV., figs. 2 and 3; PI. V., fig. 11). Here we have to do with small and delicate-looking rosettes measuring only 50-64 p- in diameter. In general shape and in the proportion of parts they closely resemble Euplectellid flori- comes, except in tliat each perianth of terminals is somewhat more expanded and in that the terminals are conically pointed at tlie outer out-flaring end instead of having toothed plates. About a dozen or more, slender and distally gradually thickened terminals spring in a single whorl fi'om the margin of a plano- convex disc at the end of each moderately long principal. In some rosettes, assumably those in an early stage of development, both the principals and the terminals are considerably thinner than in others. — The sigmatocome is found even in the smallest specimens shown in PI. V., fig. 9, though not in so great abundance as in all larger individuals. In these it is of common occurrence in the parenchymal septa ; it occurs much less frequently in the gastral layer also. On the surfaces of ]mrenchymal septa, it is quite common to see the rosette shifted out to the tip of the free ray of canalar oxyhexactins (PI. IV., fig. 8 ; PI. V., fig. 12), CHAUNOPLECTELLA SPIXIFERA. 71 exactly after the manner of Euplectellid florieomes. I have found this to be of mucli more general occurrence on the excurrent, than on tlie incurrent, surface of parenchymal septa. Sometimes, as before mentioned, a sigmatocome has been found hanging on the end of an outstanding spicular ray which belonged to an indubitable parenchymalia. CHAUNOPLECTELLA SPINIFERA. N. sp. PI. v., figs. 14-17 and PL YL, figs. 1-S. In the Science College Collection I have discovered a specimen (No. 459) which bears close resemblance to Chaunopleciella cavernosa but seems to deserve erection into a distinct species. I propose to call it CltaunoplecteUa spinifera, in view of the spine-bearing character of some of its dermalia. Another, much smaller and evidently very young specimen (Sei. Coll. Mus. No. 430), which I am inclined to refer to the same species, has also come under my observation. As it differs in some respects from the type-specimen, it will be well to treat of it separately. I shall refer to it as the second specimen. The type-specimen (PI. VI., figs. 1-8) comes from Homba, Sagami Sea, where it was taken at a depth of about 572 m. ( = 313 fathoms). It is unfortunately incomplete in that it lacks the basal part, which had been torn off and lost. It represents a thick-walled sac of about the size and shape of a small plum, measuring, say, 30 mm. in diameter. The wall is 9 mm. thick in the thickest part; it thins but little towards the sharp- edged oscular margin, which is only partially preserved. The osculum 72 ART. 1.— I. IJIMA : HEXACTINELLTDA, IIT. must have been roundish, measuring about 1-5 mm. across. The texture of the sponge is quite h'ght, soft and fragile. The spicules are all very fine and loosely interwoven ; they do not combine to form bundles of a noticeable strength, except in the region of the severed base. Both the external and internal surfaces are covered with a loose, delicate and irregular interlacement of the dermalia and the gastralia respectively, beneath which are seen canalar aj^ertares not more than 2 mm. in diameter. The second specimen was found in the same bottle together with specimens of Lanuginella pupa. The locality is Outside Okinose ; depth unknown. The small body is barrel-shaped, measuring only 9 mm. in length and 4 mm. in breadth at the middle. Wall up to le mm. in thickness. An osculum of 1 mm. diameter occupies one end, while the other end is provided with a basidictyonal mass. Macroscopically the specimen presents no specially characteristic feature. Suffice it to say that, without studying its spiculation, it might easily pass for a young C/iauno- 'plectella cavernosa or for a Leucojpsacus or a Lanuginella. Spiculation. First, as to the spiculation of the type-specimen. As in general appearance of the sponge, so also in this respect there is observable a near approach to the young Cliaunoplectalla cavernosa. The jmrenchymalia consist of hexactins and diactins. No other forms of spicules have been found amongst them. The liexactinic parenchymalia are of quite variable dimensions, CHAUNOPLECTELLA SPINIFERA. 73 While many exhibit rays about 2 mm. long and 40/^ broad near the spicnlar center, others are smaller, leading down to those which are one-fourth the size just mentioned or even smaller. Occasionally ^xe meet with very small and fine-rayed hexactins, measuring 200 f^ or less in ray-length and only about 2/^ in breadth of rays near the center. In one and the same parenchymal hexactin the rays are frequently of markedly unequal length. They are sometimes nearly straight, at other times somewhat bent. They are usually smooth throughout and taper gradually towards the sharply pointed end. See PL VI., fig. 8. The diactinic parenchymalia play a comparatively less im- portant part in the composition of the parenchyma, though the}^ can not be said to be sparse in quantity. They mostly occur in small loose strands running in company with the rays of hexactinic parenchymalia. In the basal region of the sponge, however, the diactins combine to form bundles of a rather conspicuous strength and seem to constitute the principal mass of the parenchyma in that region. They are long and slender, being mostly under 14 /i in thickness, though there occasionally occur much thicker ones among the bundles of the base. The center is externally smooth or else is provided with an annular swelling, rarely with four cruciate knobs. Ends rough-surfaced, usually slightly swollen and conically pointed. The gadralia are variously sized oxyhexactiiis which in no way differ from those of the parenchyma. In forming the gastral layer they are loosely and irregularly interlocked with one another, but always projecting one of the rays into the gastral cavity. Diactins are not found in the layer. 74 ART. 1. — I. IJIMA : HEXACTINELLTDA, III. The äermalia are oxypentactins, which differ but little from parenchymal oxyhexctins except in having one ray less. All the rays are straight or nearly so. The cruciate paratangential rays are in most cases è-lè mm. long as measured from the center ; the unpaired proximal ray is often twice as long as the paratan- gential in the same spicule, and sometimes even longer. Uniformity of size can not therefore be attributed to the dermalia. Not infrequently we meet with exceptionally small and fine-rayed dermalia, which likely are still in an incomplete state of develop- ment. Now, what seems to constitute a characteristic feature of the species is the fact tliat the paratangentials in certain dermalia — not in all — are peculiarly spined (PI. VI., fig. 2). The slender and sharply pointed spines, sometimes small but sometimes 90/^ long, are situated in a row on the outer side of paratangentials, at wide but indefinite intervals (of 34-150/^). As seen in lateral views of paratangentials, the spines mostly start out nearly erect at base but are usually gently curved one way or the other. They never occur numerously, their number on a single ray being not more than six ; often there are only one or two to a ray. Dermalia thus spined are common among the medium-sized elements of the layer. The largest dermalia are, like the smallest, unspined. The paratangentials with spines are seen to run sometimes over, and sometimes under, those without spines. In fact, it seems there exists no rule as to the relative position in layers of the spiny and the smooth dermalia of various sizes. The thin dermal latticework formed by the intersecting of paratangentials at various angles is irregularly meshed. (In the upper part of fig. 8, PI. VI., the dermal latticework is represented, not in section, but as seen obliquely en face). CHAUNOPLECTELLA SPINIFERA. 75 The hexasters are discoliexasters (PL VI., figs. 3-7) of varied size and appearance, occurring very commonly in the paren- chyma. Of them I may distinguish at least four varieties, which however completely grade over into one another by intermediate forms, so that a sliarp demarcation can not be drawn between them. In the first place, there occur, especially abundantly in the peripheral part of the body, small and spherical or nearly spherical forms of discoliexasters, of which fig. o, PI. VI., may be con- sidered a typical representative. It closely resembles the form a of the discoliexasters of Ch. cavernosa (p. Q(S). Diameter, 64-90 /■«. Each principal, which can not be said to be very short, bears a bunch of 5, 6, or more terminals. These are smooth-surfaced and thicken slightly towards the outer end ; they so diverge that the terminal discs in the entire rosette are nearly equidistant from one another. The discs are small, watchglass-like and outwardly convex ; their margin shows 6-8 small teeth. Common in the deeper parts is a second variety of disco- liexasters, distinguishable from the first by its larger size and 1jy a tendency of the terminals to each principal to form a separate tuft (PI. VI., figs. 4 and 5). Diameter, 90-160 /^ The terminals, smooth and outwardly somewhat thickened, number 6- 9 to each principal. The discs at the ends have small marginal teeth, just like those exhibited by the variety first mentioned. A third variety of discoliexasters is made up of those which form the largest rosette of the species and in which each long and slender terminal is capped with a disc composed of 4-6, moderately large, recurved prongs arranged in a whorl (PI, IV., fig. 7). In appearance it is very much like those discoliexasters ill Ck. caocrnom, which I have called Form h. Diameter, up 76 ART. 1. — i. IJIMA : HEXACÏINELLIDA, III. to 230 1^. Siicli large discohexasters are not iincommonly fouiid together with the second variety. As a fourth variety of discohexasters may be mentioned the form shown in fig. 6, PL IV., which form is but rarely met with in the parenchyma. It is characterized by very slim terminals, grouped in separate, narrow and outwardly somewhat expanding tufts. The terminal discs are rudimentary and piuhead-like in appearance. The diameter, in one that I measured, was 144 /i. The delicate sigmatocome, which I have discovered in even the smallest specimens of Cli. cavernosa at my disposal, has not been found in the j^resent species. The second and much smaller specimen, which I refer to the present species, is, I should say, essentially the same in spicu- lation as the type, but with such jDoints of deviation as are indicated below. None of the dermalia show spines on the paratangential rays. This however I consider as due to the young state of the specimen. Probably the spines develope after the sponge has come nearer to maturity. Of the discohexasters, the commonest form (PL V., figs. 15 and 16), corresponding in general shape to that which I have called the first variety in the type s])ecimen, has a smaller number of terminals (usually 3 or 4) to each principal. Diameter, 54/>« and upward to 100,« or over. — The second variety of the type seems to be wanting here ; but ])erhaps it is to be considered as being represented by the larger of the rosettes that I have just now compared to the first variety. — The third variety is well represented though not in abundance ; an example of it is shown in fig. 14, PL V. Diameter, 200/^. The discohexasters thus far CHAUNOPLECTELLA SPlNlFERA. 77 indicated as being present in the second specimen are scarcely distinguishable from the like hexasters of Ch. cavernosa, and I should possibly have held the specimen to be a young indivi- dual of that species, had it not been for the total absence of the sigmatocome and for the not infrequent occurrence, near the gastral surface, of an exceedingly fine-rayed discohexaster-form (PL v., fig. 17), which is comparable to that which I have described as the fourth variety in the type specimen of Ch. spinife7'a. This discohexaster-form measures about 80/« or more in diameter ; 3-5, filamentous and obsoletely rough-surfaced terminals, each with a small and minutely toothed terminal disc, form a more or less distinct tuft to each principal. It seems to pass over gradationally into the third discohexaster-form, as this likewise does into the first. A thorough examination of the slide- preparations, into which the entire specimen was converted, revealed a single case of a strobiloplumicome being included in the tissues. I can not but think that this is extrinsic, — that it originally belonged to Lanuginella pupa, together with which the specimen had Ijeen thrown in the same bottle. After all, the peculiarities which the spiculation of tlic second specimen presents in comparison with the type, I consider as due merely to individuality. The basidictyonalia, preserved in the second specimen, presents much the same characters as that oîC/i. cavernosa (PL V., fig. 13). 78 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. CAULOPHACID.^. Tlie idea of this new family has been conceived by me for the reception of certain genera from the ranks of F. E. Schulze's Asconematidœ, all the rest of this old family being given over to the Rossellidœ. The genus Asconema being one of those thus removed, it is self-evident that the remaining Asconematidœ require a new denomination. The Asconematidos, when last treated of by F. E. Schulze ('99, p. 98), were made to comprise those " lyssacinen Hexas- terophora, deren Autodermalia und Autogastralia aus pentactinen oder hexactinen Pinulen mit vorragendem bedornten Radialstrahle bestehen." Seven genera were then referred by him to that family, viz., Asconema Sa v. Kent, Hyalascus Ij., Gaidophacus F. E. Sch., Aulascus F, E. Sch., Sympagella O. Schm., Calycosoma F. E. Sch. and Calycosaccus F. E. Sch. At the same time that experienced writer himself expressed, as on an earlier occasion, his conviction that the Asconematidie and the ßossellid* would not hold out long as separate and distinct families. That he kept up the former family, as he did, was due more to his careful consi- deration of our yet defective knowledge of the forms concerned, than to any other reason. After some deliberation I have come, as already indicated, to entertain the view that, while a total amalgamation of the two families seems not advisable, several of the genera mentioned above may even now be annexed to the Kossellidie with advantage to the system. CAULOPHACIDvE. 79 From the above assemblage of seven genera may be separated Caulophacus, Aulascus and Sympagella, as having a number of important systematic characters in common, by virtue of which common points they, as a group, seem to be clearly distinguishable not only from all the remaining genera, but also from the Rossel- lidaî. This group — an intimately coherent group — it is, which, I think, deserves institution into a distinct family, the Caulophacidre. Since Aulascus appears to me as unitable with Sympagella, this family may in fact be said to be made u]) of only the two genera, Caulophacus and Sympagella. In both these genera the body is probably always provided with a long and distinct stalk ;'•' the dermalia and gastralia are invariably pinules, in which the freely projecting pinular ray is always of a characterization markedly different from that of any other ray in the same spicule ; the rosettes are mainly discohexasters, while oxyhexasters may be said to be generally totally wanting/''* Whereas, what remain of the Asconematidae, viz., the four genera Asconema, Hyalascus, Calyeosoma, and Calycosaccus, consti- tute a rather heterogeneous assemblage. In contrast to the Caulo- phacidœ, they have a body which never seems to exhibit a long, distinct stalk ; the freely projecting ray of the dermalia and gastralia is either not at all or comparatively but little differen- tiated from other rays in the same spicule ; the rosettes always include an abundance of oxyhexasters, occurring in addition to either discohexasters or some other hexaster variety. The combination of the above-indicated common features suffices to * Only in Sympagella {Aulasain) johnstoni has tlie pi'osence of a stalk not been determined , the species being known from an incomplete specimen lacking the basal part. ** Oxyhexasters are known from Sympagella nux only, in which they occur but rarely and inconstantly; so that, they have been put by F. E. Schitlze ('99, P- 34) under the uncharacteristic hexaster varieties or aberrations of the species. 80 ART. 1. — I. IJIMA : HEXACTINELLTDA, III. distinguish every one of the genera in question from the Caulo- phacidœ {cß\ above), while at the same time it seems to bring them within the diagnostic scope allowable for the Rossellidie. F. E. Schulze ('gg, p. 101) looks upon the absence of a pinidar distal ray to dermalia as the principal family- character of the Hossellidre. Not that Kossellid dermalia should always lack a distal ray, but he in fact includes, and quite rightly I believe, the hexactinic form within the range of their legitimate variation, with the restriction that the one necessarily distally directed ray in such hexactins should not be pinularly developed but simple and similar in appearance to the other rays. Thus, in Aphorme horrida F. E. Sen.''' ('gg, p. 41), and in a manner also in Trichasterina borealis F. E. Sch, ('oo b, p. 103), the dermalia are said to be hexactins, while cases of other Rossellids with well-developed hexactins occurring sporadically among the dermalia are by no means rare (e. g., Rossella nuda Tops., R. racovitzce Tops., Rhabdocalyptns tener F. E. Sch., Acanthascus platei F. E. Sch.). I have expressly referred to the above point because the dermalia of the Rossellidœ were originally considered to be always without a distal ray (Chall. Rep., pp. 129, 374) in contrast to those of the Asconematidre, — a fact which apparently has had influence in maintaining the status quo of the latter group. * I must say that the small slender-rayed " oxy hexactins," taken by F. E. Schulze for the dermalia of Apharme hmrkia, are probably not real hexactins but oxyhexasters of hexactinose shape. I was strengthened in this belief on viewing under the microscope a preparation of the type-specimen, kindly shown me by Professor F. E. Schulze. The real dermalia of the species seem to be the stauractins called by him the hypodermalia, while these seem to be really represented in the large pentactins occurring as prostalia. A part of the hexactinose oxyhexasters had apparently secondarily taken up a position outside the dermal layer, a process probably analogous to the shifting of Euplectellid floricomes to the extreme outer end of dermalia. Notwithstanding the above fiicts, it is plain that F. E. ScnULZE is at one with me in the idea that forms with entirely hexactinic dermalia may be taken up under the Rossellidre, if other characters permit it. CAULOPHACID^. 81 i^ow, as regards Asconema (with the single species, A. seiii- balense), the unpaired ray of its pentactinic dermalia and gastralia is, to judge from F. E. Schulze's description and figure (Chall. Eep., p. 117 ; PL XXI., fig. 4), scarcely differentiated in appear- ance from the para tangential rays. It is exceedingly doubtful if it can at all be called pinular. The mere fact that the unpaired ray in those pentactins projects freely outwards and does not dip into the body-wall, appears to me a much too slight and unreliable ground for excluding the genus from the Rossellidœ. This I say, not only on a general consideration of the wide variability — ranging from hexactins down to diactins or even to monactins — exhibited by dermalia in that family, but also in view of the fact that in Lophocalyx spinosa F. E. Sch. ('oo, p. 37) we have a Kossellid in which there occui-, together with stauractinic dermalia, others that are pentactinic and have the unpaired ray directed outwards. I consider the dermalia of Asconema to have been directly derived, by atroph}^ of the proximal ray, fiom such simple hexactinic forms as are sometimes shown by certain Rossellids, — not from such hexactinic pinules as are possessed by Caulophacus, as the pentactinic pinules of Sympogella mix unquestionably are. The same should hold good mutatis mutandis for the gastralia also. Exceptional as the condition certainly is, the pentactinic dermalia and gastralia with the unpaired simple ray directed away from the body are, in my estimation of their bearing on the systematic, no farther removed from the original simple hexactinic form than are those — so commonly met with in the Rossellidaä — in which the unpaired ray is directed the other way. After all, I think that, if Asconema is to be kept separate from the Eossellid?e, it should rather be removed from association in the same family with Caulophacus and Sympagella. 82 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. Hyalascus is a genus instituted by me in '96 to receive the single species then known to me, H. sagamiensis Ij. In this species the dermalia are partly pentactins and partly hexactius, the former with the unpaired ray directed proximad and the latter with the distal ray in no way differently characterized from the other rays. In a second species, referable to the same genus which has since become known to me {H. giganteus n. sp.), the dermalia are mostly pentactins with the unpaired ray directed proximad and a distal sixth ray represented by a knob, and occasionally stauractins and simple hexactins. In '96 I was not quite certain as to which family H. sagarniensis should be referred and therefore had to satisfy myself with the remark that it was probably to be considered as a near ally of Ascone7na, without denying at the same time its close affinity to the Ros- sellidœ. I now see, in the nature of the dermalia or in any other respect of the spiculation, nothing that seems to stand in the way of placing the genus under the Kossellids, but much that indicates that it properly should be so placed. I think the same may fairly be said of Calycosoma and Calycosaccus. In both, it may be said collectively, the dermalia are either hexactins or pentactins, in the latter case the unpaired ray being outwardly turned, ►^o far the dermalia fall within the range of variability as assigned by me to those of the Rossellidœ {afr. supra). But the one point, which might possibly be consi- dered— as indeed it was considered by F, E. Schulze — to interfere with the introduction of the two genera among the Rossellids, is he fact that the distal ray of the said dermalia shows a spindle- like swelling and exhibits prickles somewhat more strongly developed than those on the other rays, thus causing its resem- blance to a pinular ray. But the resemblance can by no means CAULOtUACIDJî. 83 be called complete, and probably nobody will hold it, by itself, as a character adequate to base a family on. In all other respects the two genera in question are both quite like Eossellids ; the somewhat special development shown by the freely projecting ray of their hexactinic gastralia is of no moment, since the same feature is not infrequently observable in Rossellids of unquestion- able status (e. g., Rhahdocalyptus nodulosus F. E, Sch., Rh. teuer F. E. ScH., Rh. mirabilis F. E. Sch., etc.). With respect to Calycosaccîis, it has even been stated by F. E. Schulze the deseriber, ('99, p. 100), that he would not have hesitated to regard it as identical with the Rossellid genus Aulosaccus Ij., if only the unpaired ray of its pentactinic dermalia had been direct- ed proximad instead of distad. So that, while the two genera seem unitable to the Rossellidge, they may, for the reason already advanced, be kept sej)arate from the group that I have called the Caulophacidœ. Galycosoma (with the single species, C validum. F. E. Sch.) had probably best be received into the subfamily Lanuginellinse, espesially on account of the strobiloplumicome present in it. Whereas, Calycosaccus (likewise with a single species, C. rjimai) is to be placed in the subfamily Eossellinœ in direct proximity to, if not in amalgamation into Aulosaccus. I may here add that to the same subfamily should also belong Hyalascus and Asconema. The three genera herewith referred to the Rossellinœ show no sign of specially close bonds of relationship between them, w^hen considered in relation to other Rosselline genera. But to return to the Caulophacidœ. Its near affinity to the Euplectellidœ is undeniable, so much so that Saccocalyx pedun- culata F. E. Sch., now recognized as an Euplectellid, was at first 84 AßT. 1. — I. IJIMA : HEXACTINELLIDA, III. regarded as an Asconematid evidently on account of its resemblance in both general shape and spiculation to Caulophacus or Sympa- gella. On the other hand, a relationship, about equally close, to the Lanuginellin« under the Rossellidœ seems to be indicated by the fact that Sijriipagella is in possession of the strobilopluinicome, a peculiar form of rosette present in all the Lanuginelline genera. An intermediate position between the Euplectellidse and the Rossellidœ is therefore to be ascribed to the Caulophacidœ. I define the family as follows : Lyssacine Hexasterophora of cup-like or mushroom- disc-like body; always stalked and firmly attached at base; solitary or forming a small branched colony by budding. Dermal skeleton composed of small hex- actinic or pentactinic j^inular dermalia and of large j^entactinic hy jiodermalia. Hexasters represented mainly by discohexasters, either alone or in company with s t r 0 1 J i 1 0 p 1 u m i c 0 m e . The two genera composing the family are so joined to each other bv an interchan2;e of their characters that a statement of their differential points, which will hold good for all the cases of species, can only be made in the most meager terms as follows : A. — With strobilopliimicome among the hexasters Sipupaydla O. ScHM. B. — "Without strobilophimicome among the hexasteis Cauloplmcus F. E. ScH. CAULOPHACUS. 85 CAULOPHACUS F. E. Sch. Balanites, F. E. Sch., '86; '87, pp. 122,:372. Balanelht, F. E. Scir., '87 (postscrip>t). Caulophacus, F. E. Sch., '86; '87, pp. 124, 373; '97, p. 520. Caulophacids of cup-like or fungiform body ; solitary (always ?). Pareiicliymalia chiefly diactiiis but iiicludiug also large regular hexactins. Dermalia, liexactinic pinules ; on the stalk these may be replaced by pentactinic forms. Gastralia, either liexactinic or pentactinic pinules. Discohexaster j^resent in varying forms; with- out strobiloplumicome. Noteworthy is the fact that in most species (probably all except C. loiifolium n. sp.) of the genus the unpaired ray of the pentactinic hypode.rmalia and hypogastralia is beset with prickles throughout neaily its entire length. Further, in most species (the exception being again G. lotifolium) the discohexaster occurs in tw^o easily distinguishable types, which I shall call, for the sake of easy reference, the 'pachydiscohexaüev and the lophodiscohexaster. The former has thick, thorny or barbed, widely diverging termi- nals ; it is frequently hemihexactinose or hexactinose, and in some species occurs in the hexactinose form only. The latter has nmcli thinner terminals, which form a distinctly separate tuft of bell- like or elongate-conical shape to each principal. The follow^ing is a key to the five known species of this o;enus : 86 ART. 1. — I. IJIMA : HËXACTINELLIDA, III. «.—Body cup-like; or, the principals of hexasterous discohexasters (of both types) much longer than the terminals C. pipetta F. E. £cH. (Antarctic Ocean). b. — Body fungiform, with the ga?tral surface outwardly exposed ; or, the principals of hex- asterous discohexasters (of both types) nearly equal to or shorter tlian the terminals. «'. — Gastralia are predominantly pentactinic pinnies; or pachydiscohexasters represented by hexactinose forms only. a". — Pinular ray of dermalia mostly ovuid; or, the slender pinular ray of gastralia nearly 1 mm. or more in length ; or the terminal tuft of lophodiscohexasler nearly as long as the principal ; or, the unpaired radial ray of hypodermalia and hypogastralia echinated throughout nearly the entire length C. latus F. E. ScH. (South Indian Ocean). b". — Pinular ray of dermalia slender and sharply pointed at apex ; or, the slender pinular ray of gastralia less than i nun. in length ; or, the terminal tuft of lophodiscohexaster 3-4 times longer than the principal ; or, the unpaired radial ray, as also other raj's, of hypodermalia and hypogastralia echinated at base.... O. ayasaizl F. E. ScH. (Atlantic Ocean). i^ — Gastralia are hexaclinic pinnies ; or, pachydiscohexasters represented by hexasterous, hemihexactinose and hexactinose forms. c^.— The unpaired radial ray of hypodermalia and hypogastralia echinated through- out ; or, pachydiscohexasters over loOiJ. in diameter C. ek(jms F. E. ScH. (North Pacific Ocean). d". — The unpaired radial ray of hypodermalia and hypogastralia smooth, being rough-surfaced only at the end; or, pachydiscohexasters under 115 [j. in diameter. Lophodiscohexaster not present C lotifoUum Ij. (Sagami Sea\ Besides the five species ""' embodied in the above, a case of undeterminable Caulophacus from off the coast of Maryland is men- tioned in F. E. Schulz'es " Amerikanische Hexactinelhden " ('gg, p. 39). The specimen consisted of only the stalk and the basal disc, but the hexactinic pinular dermalia and the nature of the hexasters found in it attest the correctness of the generic identifi- cation. As to the completely macerated pieces of tubular stalk from the Antarctic mentioned by Topsent ('oi) as Qmlophacus sp., nothing can be said except that that writer was perfectly justified in attaching a query to the name. * The specimen from the Antarctic, on which F. E. Schulze based his Pleorhabdus oviformis (Chall. Eep., p. 121 and P. S.) is evidently a Caulophacid. Were it not for the fict that that genus and species was withdrawn from the system by the original descrlber ('97), I might have regarded Pleoihahdm, on the strength of the description and figures given in tlie Cliall. Rep., as identical with Caulophacus and thus added a sixth species to the list; but as the matter now stands, it has to be entirely left out of consideration. CAULOPHACüs lotifoliu:m. 87 Here let me give the description of the only sepcies that has as yet been obtained in tlie Sagami Sea. CAULOPHACUS LOTIFOLIÜM. N. sp. PI. YII. This species is founded on a unique specimen which was obtained by Kuma from a depth of about 572 m. (313 fms.) in Maye-no-Yodomi, Sagami Sea. It passed into the possession of Me. Alan Owston, while a small piece cut from it and thrown into alcohol on the spot was sent to me by the collector. The specimen, well preserved in the dried state, was subsequently purchased by Dr. Lampert of the Kgl. Naturalien Kabinet in Stuttgart, in which institution it is now preserved. In general shape and appearance the type specimen (PI. VIL, fig. 1) reminds one of a drooping lotus-leaf; hence the specific name. The apex of the inverted-conical body is continued below^ into a long crooked stalk. The latter expands at the lower end into an irregular basal mass, by means of which the sponge was attached to a tufaceous substratum. If the stalk had been straight, the entire specimen would have measured no less than 410 mm. in height, about three- fourths of tliis measurement appertaining to the stalk. The body proper measures 132 mm. across the broadest part. The stalk is 8-11 mm. thick in the middle part. As in C. eleyans, agassizi and kdus, the gastral cavity is turned entirely inside out, its surface thus forming the upper terminal surface of the inyerted-cQuical sponge-body. A not 88 AET. 1. 1. TJIMA : HEXACTINELLTDA, III. inconsiderable part of the periphery of this surface is reflected outwards and backwards, forming at one place an ear-like flap almost touching the lateral surface. The irregular gastral surface is much folded and creased, which may have occurred in the desic- cating process. It is covered all over with an extremely delicate gastral layer, which is supported on fine hypogastral fibers barely distinguishable to the naked eye. Through and close beneath the layer are seen the apertures of excurrent canals. These are mi- nute near the oscular rim but become larger towards the center, where thev mostlv measure about 2 mm. thou2;li son e are as much as 6 mm. in diameter. The larger apertures lie separated from one another by a space usually narrower than their own width. That the above considered terminal area of the sponge is the gastral surface, is placed beyond the reach of doubt by the fact ascertained by direct observation, that the chamber layer beneath it has the apopyles turned towards that surface. The oscular edge is sharp, though not thin. No prostalia maro-inalia are seen alono; it. The kiteral surflice of the sponge-body shows a dermal skeleton of exquisite beauty. The dermal layer is exceedingly fine and delicate ; its minute meshes are just discernible as such with the naked eye. It is borne on two sets of hypodermal latticework. Of these one is formed by the paratangentials of pentactinic hypodermalia (PI. YIL, fig. 19). This is quite delicate, present- ing small meshes which are generally regularly rectangular and measure onl}" about half a millimeter in the length of sides. The other hypodermal latticework is formed of fibrous bundles of varying caliber. It is the coarsely and irregularly meshed lattice- work that is prominently visible in fig. 1, PI, YII. In somç CAULOPHACUS LOTIFOLTUM. 89 places the bundles are fully 2 mm. thick ; in others they nre quite fine. The coarser bundles are seen to run from the narrow end of the body divergingly towards the oscular edge, interposed between the dermal layer and the choanosomal mass. On the way thither they frequently divide and unite or send out anasto- mosino' branches. The finer bundles stretch over the subdermal space entirely separate from the choanosome but in union with the dermal layer. Through the dermal layer and separated from it by the subdermal space which varies in width in different places, are seen the roundish or oval entrances into incurrent canals. These do not exceed 6 mm. in diameter. Towards the narrow end of the body and on the stalk, the dermalia lose the lattice -like arrano-ement and liccome so denselv crowded together as to form a smooth compact coating of frosted appearance. The coating easily falls off" and, in the greater part of the lower two- thirds of the stalk, is lost. The surface thus exposed is soiled and has given attachment to a creeping Hydroid stolon. Except in the upper end the stalk is quite firm, owing to an extensive synapticular fusion of spicules. This fusion is however not participated in by the dermal pinnies and the liypo- dermal oxypentactins, nor by the liexasters present in that region. Superiorly the fusion gradually disappears and at the junction with the sponge-body all the spicules are loose, the parenchymalia here running parallel and densely packed together but soon to split above, as far as seen on the surface, into the coarse hypodermal bundles before mentioned. The stalk is hollow, being axially traversed by a canal measuring 3 mm. or less in diameter. It may therefore be said to be a thick-walled tube. The lumen is evidently an extension 90 ART. 1. 1. IJIMA : HEXACTINELLIDA, III. of the excurrent canal-system. " There certainly does not exist any direct connexion between it and the gastral siir&ce. The texture of tissues in the sponge-body is markedly soft and delicate, which is due to the fineness of all the spicules. Spiculation. The parenchymalia consist mainly of diactins, intermixed with which are however a not inconsiderable quantity of regular hexactius. The diactins run either isolated or in loose or compact bundles. They are all short and thin, seldom exceeding 2 mm. in length and 15/^ in breadth. The spicular center is usually plain, only occasionally supplied with an annular swelling. The breadth generally remains nearly the same in the greater part of the rays. The terminal roughened portion is often slightly swollen before ending in a rounded or a conical point. In the compact wall of the stalk, the diactins, in the main longitudinally disposed, are generally somewhat longer and frequently as thick as 35/^. Here the rough ends are often swollen in a knob-like manner. There is nothing in the synapticular formations in the stalk requiring particular mention. The parenchymal hexactins have straight smooth rays, whi(;h gradually taper towards the sharply or bluntly oi- conically pointed, slightly rough-surfaced end. In these charactei's they agree well with both the hypodermal and the hypogastral pentactins ; so also in dimensions in the generality of cases, but sometimes they are considerably larger and stronger (figs. 18 & 21). The axial length may reach 1,6 mm. CAULOPtlACÜS LOTIFOLlÜM. ' 91 The liypodermal, anastomosing Innidles of fibers, which may be considered as a part of the parenchymal skeleton, seem to consist of the diactinic elements only. The hypodermal peniacii/is in the sponge-body measure com- monly between 300/^ and 600/'- in length of paratangential rays. The proximally directed, unpaired ray is usually longer than, but never so nuicli as twice, the length of the paratangential s in the same spicule. Thickness of rays near the center 23-30 /A The rays gradually attenuate towards the more or less conically pointed end. Their surface is smooth except for a very short space at the ends, which are rough. This holds true of all the rays, not excepting the proximal ray, which in all other known species of the genus seems to be echinated throughout almost its entire length. The paratangentials form the dehcate, more or less regularly quadrate-meshed, hypodermal latticework before mentioned (fig. 19). — In the stalk, especially in its lower part, the hypodermal pentactins (of which two are shown in fig. 20) are nmcli smaller than in the body. Length of paratangentials 100 ,« on an average ; the proximal ray 3-4 times as long, sometimes bearing on the surface a small number of minute prickle-like points. The pentactins are here irregularly scattered. As hypogastralia there occur pentactins similar in all respects to the hypodermalia of the body (fig. 21) ; only they are distri- buted without order as to the mutual relation of the cruciate paratangentials, so that these do not exhibit a checker-like arrangement. They may associate with some parenchymal diactins in forming the thin hypogastral strands. 92 AKT. 1. 1. IJIMA : HEXACTINELLIDA, III. The derinalia (fig. 3) are exclusively liexactiiiic pinnies, so far as those of the body proper are concerned. The pinular ray as a whole is spindle-shaped ; it is 120-140/^- long and 80-42,« broad in the middle, which is about the broadest part. In this part, the obliquely upwardly directed, elongate conical spines may be as long as 23 JJ-. 'J'he rhachis is smooth for a short distance at the base, wdiich is about 11/'- thick; its conically pointed, outer end forms the tij) of tlie pinular ray. The remaining five rays are all more slender, and gradually taper towards the conically or bluntly pointed end. They are beset with small, generally erect prickles, sparingly at the base but more numerously at the end. Length 88-120«. The proximal ray is usually slightly shorter than the paratangentials of the same pinule. As is usual, two paratangentials of every two adjoining pinnies lie side by side for nearly their entire length, the result being the fine, quadrate-meshed dermal latticework (fig. 19), in which each mesli has sides approximately equal in length to that of the rays concerned in inclosing it. As ascertained on sections, the said latticework does not lie in direct contact with the underlying hypodermal paratangentials but is separated from these by a space about as wide as, or slightly less wide than, the length of the dermal proximal rays (fig. 18). Here and there among the dermalia of the bodv I have met with such forms as may appropriately be regarded as early stages of their development. They are comparatively small and slender-rayed hexactins, in which the distally directed (the future pinular) ray is not at all or but little difterentiated from the other rays, being nearly as prickly as these. In the upper part of the stalk, the pinnies retain the characters described above ; but they are here much more closely packed CAULOPHACUS LOTIFOLIUM. 93 together, the checker-like arrangement of parataugentials totally disappearing. Lower down on the stalk, where the parenchynialia have undergone ankylosis, the hexactinic pinnies gradually pass over into a peculiarly modified, pentactinic form, in which the suppressed pi'oximal ray is however frequently indicated Ijy a small prominence (figs. 9-11). The persisting rays are much thicker than the corresponding rays in a normal pinule of the body, and what at once strikes the eye is the greatly swollen state of the unpaired free ray, especially at its distal end. In this way that ray acquires a club-like, or more generally a balloon-like, shape. The swollen end is densely covered with short but stout, conical tubercles, which become less numerous below on the stalk-like part of the ray, finally to disappear altogether at the base of that part. The terminal globular knobs often measure 9'3 !'■ in diameter, but are of various sizes leading down to a club-like shape of 35 ,« breadth. In surface view under a low power of the microscope (fig. 20), the dermal coating of the stalk appears as if consisting of rongh- surfaced spherules closely crowded together. A similar change in the character of dermalia on the stalk has been known in Caulophacus elegans F. E. ScH. ('87) and a agassizl F. E. Sch. ('99, p. 39). The gadraUa (figs. 2, 21) are likewise hexactinic pinnies ; rarely pentactinic, the aborted distal ray being represented by a mere knob. The free pinular ray, compared with the same of normal dermalia, is much longer and proportionally more slender. It is 220-300 />« long and about 30/^ broad as measuied across from tip to tip of the strongest lateral spines in about the middle of the ray. The axial rhachis, about 12/^ thick at base, gradually tapers towaids the sharply pointed free end. The five remain- 94 AKT. 1. 1. IJIMA : HEXACTlNELLTDA, III. ijig rays are exactly comparable to the corresponding rays of dermalia, except in the fact that they are perceptibly longer. Length of paratangentials 110-132/^; that of the distal ray 88- 100/^. The paratangentials are so arranged as to bring about a rectangularly meshed latticework, in the same manner as those of the dermalia. The hexader of the species consists of rough and thick-rayed discohexasters — evidently the pachydiscohexaster of other Caulo- phacus species — and their variations. Lophodiscohexasters are not present. As the starting point of our description may serve the spherical or nearly spherical, normally developed discohexasters shown in fju;s. 4 and 5. Such a form occurs, too;ether with certain varieties soon to be mentioned, in tolerable abundance in both the subdermal and subgastral regions, more commonly in the latter than in the former. It measures 60-92 /^ in diameter. In the larger and well developed cases (fig. 5), the principals are obsolete, making it difficult to determine the number of terminals borne on each. However, the usual number seems to be 3 or 4, perhaps occasionally 5. In smaller specimens of the hexaster (fig. 4) the piincipals are only just long enough to be distinguishable and each bears only 2, at most 3, terminals. In all these cases the terminals are thick (up to 4 /^- in breadth) and nearly straight or l)ut slightly bent ; they are profusely beset, throughout their length, with rather strong, retroverted prickles. The convex terminal disc exhibits 4-6, strong, recurved, marginal teeth. The form just described is occasionally hemihexactinose (fig. 6) and frequently hexactinose (figs. 7 and 8). The hexactinose CAULOPHACUS LOTTFOLIUM. 95 type is especially common in the choanosorae ; in feet, it may be said that almost all the discohexasters here present are of this form. The terminals are exactly comparable to those of normally developed discohexasters. When hexactinose, the axial length may reach 115 /', showing an increase in diameter over normal forms, — a flict which according to my experience, is generally observable in all Ivssacine Hexactinellids having a hexaster in the two varietal forms mentioned. Fig. 7 shows a case of the hexactinose discohexaster, in which one of the six rays is bent at base, i. e., at the junction of the terminal with the principal, indicating its derivation from a dilophoiis ray by loss of one of the two terminals. Fig. 17 shows the extent of the axial cross in a hexactinose discohexaster, it being limited in extent to the spicular center and the basal parts of the rays corresponding to the original principals. Further variations — or possibly early developmental stages — of the discohexaster are seen in unusually thin-rayed and obsoletely rough or nearly smooth surfaced forms, such as are represented in figs. 12-16. Transitional forms connect them with the thick- rayed discohexasters, and they occur, together with these, not uncommonly in both the subdermal and subgastral regions, but especially in the latter. Sometimes they have all the principals supplied with 2, occasionally 3, terminals (fig. 14); at other times they are either hemihexactinose (figs. 12, 13, 16) or quite hex- actinose (fig. 15). The terminal discs in a rosette of the kind may be in appearance similar to those in the thick-rayed discohexaster, except in being smaller and more weakly developed (figs. 12 and 13) ; or they may occasionally consist of 2-4, nearly transverse or outwardly diverging, slender claws at the ends of tapering terminals, in which case the rosette dçserveç tQ be çallçd au 96 ART. 1. — I. IJIMA : IIEXACTINELLIDA, III. onychaster (figs. 14, 16). The said variation in the development of terminal discs occurs irrespective of the rosette beiug normally developed, hemihexactinoso or hexactinose. In certain cases I have found the terminal claws branched (fig. lö). SYMPAGELLA O. Schm. SynipageUa, O. Schmidt, '70, p. 15. — F. E. Schulze, '86; '87; p. 119; '97, p. 528; '99, p. 32; '00 a. Aulascus, F. E. Schulze, '86; '87, p. 118; 97, p. 527. Caulophacids of cup-like body, showing tendency to form small colonies by budding. Parenchymalia as in Caidophacus. Dermalia, either hexactiiiic or pentactinic pinnies. Gastralia, hexactinic pinnies. Besides discohexasters, strobiloplumicome always present. The new species to be presently described under the name of SympageUa anomala niight, apparently with equal propriety be referred to the same genus as Aulascus johnstoni F. E. Sch. In fact, it appears to me that the genus Aulascus is scarcely sufficiently differentiated in characters to justify its separation from SynipageUa. Hence I have placed it under the synonymy of the older name O. Schiniidt's in spite of F. E. Schulze's ('97) opinion to the contrary. It seems to me fairly assumable SYMPAGELLA AXOMALA. 97 that the fragmentary specimen, on which tlie single species of Aidascus {A. johnstoni) is based, originally formed a part of an inclividnal shaped somewhat like Sympagclla anomala n. sp. The small rough " discohexactins " mentioned by F. E. Schulze as occnrring in the parenchyma of Aidascus are evidentl}^ nothing else than hexaetinose discohexasters. To them, as also to the presence of the pentactinic hypogastralia or of a proximal ray to the dermalia in that sponge, I can attach no more than specific importance. The following contains all the more important differential points in the characters of, and will serve as a key to, the three species contained in the genns as defined by me. «. — Body ellipsoidal rind witli a single osculnm ; borne on ends of branches of the I'amified stalk, Dermalia pentactinic pinnies; gastralia hexactinic witli very slender pinular ray. Without hypogastral pentactins S. mix. O. Scum. (N. Atlantic, Mediterranean) '»• — Body irregularly sacciform; incompletely divided into persons, each with an osculuni. Dermalia and gastralia hexactinic ; both with similarly shaped pinular ray. o'. — Predominant di^coliexaster with rough -surfaced terminals, each ending in a small, slightly arched, transverse disc which marginally runs out into recurved prongs. With hypogastral pentactins S. johnstoni (F. E. Sen.). (Prince Edwards Is.) b^. — Predominant discohexaster with nearly smootli terminals, each ending in a whorl of 2-6, very fine straight, forwardly and outwardly diverging branchlets (not recurved). Without hyp igastral pentactins ...5". anomala n. sp. (Saga mi Sea). SYMPAGELLA ANOMALA. X. sp. Plate YIII. A score or more of specimens of this new species have been at my disposal for study. They were all collected by Kuma from depths of 430-572 m. (235-313 fms.) in the Sagami Sea. 98 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. The more exact localities are : Maye-no-Yodomi ; Okinose, both Inside and Outside ; Homba ; Gokeba. The specimens are usually attached to the dead skeletons of other Hexactinellids, such as Chonelasma calyx, Perlphragella elisce, Hexactinella ventilahrum, etc. In one case (PI. VIII., fig. 1) about half a dozen small individuals of the species were seated on a large irregularly branching mass of a dead Hexactinellid skeleton, which on examination proved to be the stalk of an old individual, or individuals, of the same species as the living specimens borne on it. The entire configuration of the stalked sponge is quite irregular ; hence the specific name I have given to it. The sponge-body proper may in general be said to be saccular or cup- like and more or less laterally compressed. The wall is moderately thick ; the oscular edge is simple and sharp, but not thin. In size the body may be nearly as large as one's fist. It usually bears, in indefinite positions, bud-like or tubular evagiuations, each bavins; an osculum at the end and leadins; into the common gastral cavity. In other words, the body appears to be composed of variously sized and incompletely divided persons. The number of these in each case is never numerous, being limited to four or five at the most. The stalk is firm, branching and anastomosing ; it is at times moderately long, at other times rather short. To illustrate with concrete examples, I have shown in fig. 2, PI. VIII., a sjDecimen measuring 168 mm. in total height in which the stalk is about as long as the compressed pouch-like body. The latter has a wall about o mm. thick in the thickest part. It seems to have originally borne no less than five buds or secondarily formed 2:>ersons, of which two remain intact while SYMPAGELLA ANOMALA. 99 the rest have been torn off. The stalk shows an open perforation ill tlie upper part ; lower clown it splits into two branches, each of which is inserted in a piece of dead G/ionelasma. Fig. 3, PL VIII., represents one of the largest specimen I have seen. It is 185 mm. high. The main body consists of two large persons joined together at base. One of them is irregularly tubular ; the other is funnel-like, expanding above, and distinctly laterally compressed. Maximum thickness of wall, 9 mm. Four short and comparatively slender stalks proceed from the common base but unite below into one before forming a knob for attach- ment. On one side, one of the stalks sends out, obliquely downwards, a free ending branch which probably at one time w^as likewise fixed to the substratum. The dead, irregularly branching stalk before mentioned and seen in fig. 1, PL VIII., is at places 1-5 mm. or more thick. Probably it represents fused stalks of several individuals. Of far greater interest are the small 3'oung specimens which are seen growing on it. These are of the size of a walnut or smaller and represent various stages of the change in form undergone by the sponge during growth. In an early stage the entire body is simply elongate-ovoid, being attached by the narrower end while in the opposite end opens a round simple osculum. One specimen in this stage of development measured only 14 mm. in length and 62 mm. in breadth in the broadest part ; breadth of stalk- like base 2ï mm ; diameter of osculum 2 mm.; thickness of wall 2 mm. and under. This simple original form of the body is soon lost with the formation of buds. These are at first mere thicken- ings of the sponge-\vall. They gradually take the form of a papilla-like protuberance and sooner or later a new" osculum opens at the rounded apex. The gastral cavity of the bud in an early 100 AßT. 1. 1. IJIMA : HEXACTINELLIDA, III. stage of its formation either widely communicates with that of the mother or is apparently separate from, and independent of, tliis. The latter condition how^ever passes ultimately into the foi'mer evidently by the widening of a connecting excurrent canal, as assumably the new gastral cavity itself likewise arose by the widening of an excurrent canal of the mother. In some young specimens I have found, apart from the terminally situated osculum, a roundish gap in the wall, this being in no degree specially elevated at the spot. The gap is sometimes surrounded by a thin iris-like membrane. I consider this to be a secondarily formed osculum, formed either pre- cociously at a place where an evagination of the wall may yet take place or as standing alone by itself for a person in suppression of the evaginating process. In the latter case, it would be exactly comparable to the openings which I have called parietal or secondary oscula in the Euplectellidœ. The originally simple stalk may possibly send out off-shoots and thus change into the ramified condition seem in old speci- mens ; but there are also other w^ays by which this may l)e brought about. Some of the little specimens above-mentioned are in contact with, and fixed to, the substratum at more than one point ; in other cases, individuals of apparently separate origin but growing side by side are fused together in the upper part. In both cases the result is much the same, and it is easy to imagine that with the growth of the sponge there should arise as many stalks as there are points of attachment to the substratum with variations in their arrangement according to circumstances. The even but gently undulating, external surface of the sponge-body is covered with the extremely fine and approximately quadrate-meshed dermal latticework (PL VIII., fig. 20), in which SYMPAGELLA ANOMAL A. 101 the piuular rays can be discerned under the hand-lc-ns as minute white spots. The meshes measure mostly 100-135 /-« in lengtli of sides. The hypodermal latticework presents meshes of trian- gular, trapezoidal, rectangular or irregular shape, with sides of 400-700/^- length. At places, especially in the lower part of the body, the thin liypodermal beams are more or less augmented in strength so as to appear like sinuous and intersecting fibrous bundles which are in no wav disting-uishable from those in the choanosoma. Through the dermal layer are visible the entrances into incurrents canals measuring 2 mm. and under in diameter. The gastral surface (see fig. 3) is not covered over by a continuous gastral layer, but leaves open all the apertures of excurrent canals. These apertures are round and ])it-like ; they may measure 2 mm. across, but there exist all sizes down to those that can be called minute pores. Any two of the larger apertures may be separated from each other by a space up to 5 mm. in width which in turn is occupied by much smaller apertures down to the smallest. Altogether the appearance of the gastral surface closely resembles that in Hyalonenia affine. The stalk is firm. Unless denuded and the deeper fibrous texture exposed, it is densely covered with dermalia forming a pure-white powdery crust, which can be easily ruljbed off. In- ternally it is almost solid, being at most traversed by a system of insignificant excurrent canals. The color of the sponge in the fresh state is, according to the statements of Kuma the collector, light pinkish on the stalk, fading above into colorlessness on the body. The small specimens shown in fig. 1 retained that coloration for some time even after thev were brought to me in the desiccated state. As to the histology, no point of particular interest has 102 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. attracted my attention. The cup-like or tliimble-like chambers measure 65-100 />« in diameter. In a certain specimen an unusually large quantity of yellow, refractive, fat-like spherules were observed hanging on the subdermal trabecuhie (see fig. 22). In all prob- ability they were products of the thesocytes. Spiculation. The pa}'enchymalia are predominantly slender diactins, ar- ranged in loose bundles or running singly in indefinite directions. They are generally under 1.5 mm. in length and 27/'- in breadth in the middle ; occasionally, somewhat longer than 2 mm. and 35 /^- broad. The center is commonly even-surfaced ; occasionally supplied with an annular swelling or with four knobs. The rays generally taper towards the end, which is conically pointed and subterminally rough-snrfaced. Included here and there among the parenchymalia are oxyhexactins of medium and small dimensions. In the character of rays these oxyhexactins are exactly comparable to the pent- actinic hypodermalia, except that they are generally somewhat less strong. Axial length, mostly under 700/^-; breadth of rays near the center, 15 /^- and under. The smaller parenchymal oxhexactins seem to pass over gradationally into the canalar oxyhexactins. Exceptienally the parenchymalia are tauactinic and staurac- tinic. These forms are especially met with directly beneath the gastral layer and also along the canalar (both incurrent and excurrent) surfaces. SYMPAGELLA AXOMALA, 103 The peiitactinie IiupodennaUa are of a moderntely large size. Parataiigentials as long as 600 z-^; the unpaired proximal ray somewhat longer than the paratangentials of the same spicule ; thickness of rays near the center, up to ?A !>■. All tlie rays, the unpaired ray not excepted, are straight and gradually taper towards the sharply pointed end, which is subterrainally nearly smooth or sparingly supplied with obsolete microtubercles. Pentactinic liypogastralia are, as a general matter, not present in this species. In this respect it is like S. nux (F. E. Sch., '97, p. 529) but unlike S. Johnston i. It is possible that their absence is in a measure compensated for by the occurrence, before mentioned, of parenchymal tauactins and stauractins in touch with the gastral layer. The dermalia (PI. VIII., figs. 4 and 5) are exclusively hexactinic pinnies. The stoutly developed pinular ray is club-like or spindle-like in shape. Starting fioni the base, it is smooth for l-\ the entire ray-length ; then it commences to thicken as the rhachis and to send out obliquely upwards and outwards a number of elongate-conical scale-like spines. The conically pointed apex of the rhachis projects more or less at the tip as the central conus. The development of the entire ray is subject to a not inconsiderable variation according to individuals, as may be seen by comparing figs. 4 and ö which are taken from two different specimens. In a certain specimen, the ray never exceeded 100 1^ in length and 35/^ in greatest breadth, while in another it often reached a length of 148/-« and a breadth of 40/^ with a greater number of lateral spines. — The five remaining rays are all much more slender and taper gradually towards the conically pointed end. A number of rather sparsely set conical microtubercles 104 ART. 1. 1. IJIMA : HEXACTINELLTDA, TIT. roughen tlie surftice for the outer | or f of their leugth. They are 75-100 !>■ long, the proximal ray being generally somewhat shorter than the paratangentials in the same spicule. Sections through the wall show that, where the dermalia are superposed upon the hypodermalia, the paratangentials of both are not in direct contact, but there intervenes a space which is nearly as wide as the length of the dermal proximal rays (see fig. 22). The gastrcdia (figs. 7 and 8) are likewise hexactinic pinnies. As compared with the dermalia in t'i3 same specimen, they show a general agreement in shape and arrangement but are distinguished by an appreciably weaker development of all the rays. This concerns not so much the length as the thickness of these. The pinular ray, besides being thinner, has spines somewdiat smaller in size and less in number. It does not exceed 33 /-« in the broadest part. The above gastral pinnies are replaced on the wall of excurrent canals l)y a less differentiated kind of hexactins, the oxyhexactinic eanalaria (fig. 9). These j)resent about the same axial length as, or are somewhat larger than, the gastralia ; but the rays are very much thinner. All the six rays are alike in appearance ; generally straight but sometimes bent ; and sparsely beset with conical microtubercles on the outer half or less of their length. The transition of the gastralia into the eanalaria at the edge of excurrent apertures is rather abrupt ; nevertheless, there are not totally wanting in this position certain intermediate forms — such as have a pinular ray in an incipient dfgree of differentiation — which may be considered as connecting links between the two. On the other hand, as before mentioned, the SYM PAGELLA ANOMALA. 105 canalaria seem also to pass hy gradations insensibly into the smaller parenchymal oxyhexactins, there being such forms as stand between the two in respect of both the situation and the characters of rays. — The canalaria occur at places in tolerable abundance, placed side by side in irregular disposition but always with one ray freely projecting into the canalar lumen. At other places they are found only in scattered distribution. The incurrent canals seem to be totally wanting in s[)ecialized canalaria of the kind. The hexasters of the species consist of the strobiloplumicome and a peculiar onychaster-like kind of discohexasters, leaving out of account all those which occur but inconstantly or which in their distribution are confined to the stalk. The ony chaster -like liexastcrs (figs. 12-15) occur in the choanosomo, not very abundantly but still in moderate frequency. They are slender-rayed and rather small, measuring 68-100/^ in diameter. Each short jDrincipal bears 2 or 3, sometimes 4, widely divergent, nearly straight or slightly bent terminals, which are thickest at base and thin out to a very fine caliber towards the end. They are obsoletely rough-surfaced or nearly smooth. Occasionally the rosette is hemihexactinose (fig. 13). Now the very terminal point of the terminals is without a trace of a disc- like expansion but bears a whorl of 3-6, sometimes only 2, short and exceedingly fine prongs or branchlets, not more than 4 /^ long. Unlike the claws in a true onychaster the terminal branch- lets are straio;ht — not recurved — and are directed sometimes nearlv transversely, but generally obliquely forwards and outwards, so as to form a strongly divergent fork or umbel (fig. 15). The branchlets easily fall off, and then, as also when they are over- 106 ART. 1. 1. IJIMA : HEXACTINELLTDA, III. looked linder an insufficient power of the microscope, the hexaster might readily be mistaken for an oxvhexaster. The strobiloplumicome (figs. 10 and II) is common, sometimes quite abundant, in the subgastral space. In the subdermal region it occurs only occasionally. Diameter, 34-64 /^ The smaller ones have much finer and more delicate-looking terminals, and probably represent a not fully developed state of the rosette. The knob at the end of principals is hemispherical, and from its convex surface arise the terminals in about four closely set whorls (fig. 11). As usual the terminals of the innermost whorl are the longestand measure about 27/^ in length; those of the outermost whorl, only about 10/^. The small central process at the distal end of the terminal-bearing knob is frequently difficult to see, ])ut seems to be generally present. Exceptionally was I impressed of its being really absent. Under favorable circumstances I have distinctly seen the axial filament go light through the knob into the distal process. As hexasters of inconstant occurrence I consider the peculiar kind of discohexaster shown in fig. 16. I have discovered it in a limited number in the body of a medium-sized specimen (Sei. Coll. Mus. No. 473, from Outside Okinose by the Iwado-line, 500 m.) together with the usual hexasters of the species. It has never been met with in any other specimen. In its general appearance and in the character of its terminal discs, the disco- hexaster in question is not unlike certain others (fig, 17) which occur in the stalk ; but the remarkable difierence consists in the fact that either all or some of the terminals split into disc-bearing branchlets [at a point near the outer end. Diameter, 42/-« and under. Terminals, 3-4 to each short principal, rough-surfaced. Branchlets, 2-5 to a terminal in an umbel-like tuft ; their length SYM PAGELLA ANOMALA. 107 is not iinifonn in the same hexaster but reaches up to 15/^. Terminal disc, small, either rudimentary or convex with five, distinct, marginal teeth. It now remains to consider the spiculation of the stalk, As usual it is only the parenchymalia of this part of the sponge that are synapticularly fused together. The thicker beams of the rigid framework thus formed are beset with small prickles on the surface. Superiorly in the stalk the ankylosis becomes confined to the axial portion and finally ceases to exist before the body proper is reached. In the upper part of the stalk the dermalia and the hypo- dermalia exhibit essentially the same characters as on the body ; only they are both more closely set together. Towards the basal end, the dermalia change their character very considerably (fig. 6). They are here small regular hexactins in which all the six rays, being beset all over with nearly vertically outstanding- prickles, are alike in appearance ; they can not therefore be called pillules. Besides the usual onychaster-like hexaster and the strobilo- plumicome, there occur with tolerable frequency in the stalk certain discohexasters which seem to be p)eculiar to it (figs. 17- 19). These vary considerably in size and general appearance. A large specimen (fig. 17) of them may have a diameter of 90 ,". Each short principal is supplied with 2 or 3, sometimes 4, rough- surfaced, diverging terminals, which perceptibly thicken towards the small, convex, terminal disc with 5, recurved, marginal teeth. Smaller specimens (figs. 18 and 19) of the same may be so small as to measure only 30 ,« in diameter. While diminishing in size, the terminals become more slender in caliber while the number l08 AET. 1. 1. IJIMA : HËXACÏINELLIDA, lit. of them (4-8) to each principal increases. The miinite terminal disc may be marginally toothed or sim2:)ly pinhead-like. Transi- tional forms of the above discohexasters leading into the onychaster- like hexasters have not been discovered. DIAGNOSES OF THE FAMILIES, GENERA AND SPECIES TREATED OF IN THIS CONTRIBUTION. By way of a summary and conclusion, I reproduce here the definitions I have given to the families and genera treated of in this Contribution, also giving in proper places short diagnoses of the species described. EUPLECTELLID^^. Lyssacine Hexasterophora of tubular, cup-like or massive body; sometimes stalked; either rooted by a tuft of basal spicules or firmly attached by compact base; generally possessing numerous separate oscula. Dermal skeleton composed of hexactinic dermalia, the proximal ray of which is as a rule much longer than any other in the same spicule; no hypodermal pen- tactins. Hexasters various. For a list of the genera referable to the family, see p. 20. Diagnoses of the families, genera and sI'Ecies. 109 The family I divide into two subfamilies, viz.: 1. Euplectellinœ. Euplectellidse rooted in the substratum by a tuft of basal spicules. 2. Corbitelllnce. Euplectellidse firmly attached to the sub- stratum by a compact base. To the latter belongs PLACOSOMA Nov. Gen. With one species. Placosonia jfff^'ndicffjmiff h- g-, n. sp. — Corbitellinœ with laterally compressed, massively developed, soft body and moderately long, firm stalk. On top, a comparatively small primary osculum leading into the shallow gastral cavity. One side (the front) of the body is covered with a regularly quadrate-meshed, dermal latticework which is supported on another latticework composed of hypodermal fibrous bundles ; the other side (the back) presents a more dense-looking surface in which open a large number of secondary oscula leading into the excurrent canal-system. Paren- chymalia, principally diactins and occasionally hexactins. In certain positions on the front the dermalia may have tho proximal ray so reduced in length as to be not longer than the short distal ray. Gastralia, hexactins, Hexasters in three varieties : the hexactinose discohexaster (30-60/^ dia.), smallest and most numerous; the spherical discohexaster (lGO-240/^ dia.), large and extremely beautiful, confined to the back side of the sponge; and the hexactinose codonhexaster (110-17(3« dia.) found under the gastral layer. Floricome and graphiocome, not present. 110 ART. 1. — I. IJIMA : HEXACTINELLIDA, III. LEUCOPSACID^: Ij. n. fam. Lyssacine Hexasteropliora of tllick-^Yalled, cup-like or ovoid body; sometimes stalked; firmly attached by base (? or rooted by basal spicules). Dermal skeleton composed as a rule of moderately large pentactins with the unpaired ray directed proximad ; hypo- dermalia not distinguishable. Hexasters represented mainly by discohexasters (no oxyhexaster). A key to the genera and species of this new family is found on p. 33. LEUCOPSACUS Ij. Ijcucopsacids with small, ovoid or spindle-like body, which may be stalked. Parenchymalia chiefly hexactins ; diactinic parenchymalia present, but \)]i\j a subordinate part. Gastralia represented by hexactins similar to those of the parenchyma. Discohexasters in part hexactinose and in part hexasterous. Leucopsacus oyfhodocus Ij. — Leucopsacus with the inferior end of body narrowed into a stalk. Parenchymal hexactins regular and straight-rayed ; forming an approximately regularly quadrate-meshed framework. Besides hexactinose discohexasters (110-168/^ axial length), there occur smaller hexasterous forms (60-75 /^ dia.) in which the terminals to each principal form a distinct bell-shaped tuft. Len((>iisf(cus scoUodocus Ij. — Leucopsacus without a long DIAGXOSES OF THE FAMILIES, GEXERA AND SPFCIFS. Ill stalk. Parenchymal liexactins with curved rays. The hexastcrous discohexaster (46-70/^- dia.), present besides the hexactinose form (100-180 /-« axial length), is spherical in shape. Of inconstant occurrence is the small and delicate tylfloricome. CHAUNOPLECTELLA Ij. Leucopsacids with moderately large, ovoid or vase-like and thick-walled body, attached by short stalk-like base. Parenchy- malia chiefly hexactins and diactins. Dermalia either all pent- actinic or with a variable number of rays, none of which however are distally outstanding. Gastralia represented by hexactins similar to those of the parenchyma. Discohexasters always hexasterous, the lai-ger ones with terminal prongs arranged in a whorl like anchor-arms ; with or without sigmatocome in addition. CJiaunojilectella cavernosa Ij. — Cliaunopleclella of moderate- ly large size ; sometimes laterally compressed ; with wide canals and subdermal sj^ace. In young and small specimens the dermalia and parenchymal ia are as in Levcopsacus ; but in the larger ones both consist of irregular spicules with a varying number of rays, though the dermalia never show a freely outstanding distal ray. Discohexasters of varying size and appearance ; when fully de- veloped, they may attain a diameter of 250-100/^, and then have the terminal whorl of reverted prongs shaped like a bell expanding towards the rim. Small delicate sigmatocome present in addition to the above. Çhaunoplectella sphiifera n. sp. — ÇhaunoplcdiUa of ovQid 112 ART. 1. 1. IJIMA : IIEXACTINELLTDA, ITT. bodv. Dermalia and parenchymalia as in Leucopsacus ; the former consisting of oxypentactins and the latter, mostly of oxyhexactins. Some, but not all, dermalia with bent spines along tlie outer side of paratangentials. Discohexasters of varying size and appearance, but not larger than 114/^ diameter. Sisimatocome not found. CAULOPHACID^^ Ij. n. fam. Lyssacine Hexasterophora of cup-like or mushroom- like body; always stalked and firmly attached at base; solitary or forming a small branched colony by bud- ding. Dermal skeleton composed of a layer of small hexactinic or pentactinic pinular dermalia and of large pentactinic hypodermalia. Hexasters represented main- ly by discohexasters, either alone or in association with strobiloplumicome. The family comprises two genera, Caulophacus and Sympagella. CAULOPHACUS F. E. ScH. Caulophacids of cup-like or fungiform body ; solitary (al- ways ?). Parenchymalia chiafly diactins but including also large regular hexactins. Dermalia, hexactinic pinnies ; on the stalk these may be replaced by pentactinic forms. Gastralia, either hexactinic or pentactinic pinnies. Discohexaster present in vary- ing forms. Without strobiloplumicome. A key to all the species, at present known, of this genus is given on p. 86, DIAGNOSES OF THE FAMILIES, GENERA AND SPECIES. llo Caulophacus Joti folium n. sp. — Caulopltacys with inverted- conical body, continued below into a long stalk. Dermalia and gastralia, hexactinie ; the former with a spindle-shaped, and the latter with a slender, elongate, pinular ray. On the lower part of the stalk the dermalia lose the proximal ray, while the distal ray assumes a club-like or balloon-like shape. Spherical disco- hexasters (60-115/^- dia.) with thick, ])arbed terminals ; sometimes hemihexactinose or hexactinose ; they lead by gradual transition into much thinner-rayed, onychaster-like forms. Without lopho- discohexaster. SYMPAGELLA O. ScHM. Caulophacids of cup-like body, showing tendency to form small colony by budding. Parenchymalia as in Caulophacus. Dermalia, either hexactinie or pentactinic pinnies. Gastralia, hexactinie pinnies. Besides discohexasters, strobiloplumicome always present. For a key to species referred to this genus, see p. 97. Sympagella anomala n. sp. — Sympagella with irregularly- shaped body, composed of a small number of incompletely divided persons in each of which an osculum opens at the upper end ; stalk multiple, or branched and anastomosing. Both dermalia and gastralia, hexactinie ; similarly developed, though the former is much stouter ; pinular ray, spindle-like. Discohexaster, mainly an onychaster-like form (up too 100,« dia.) in which the ter- minals are finely attenuated towards the end and bear at the point 2-6, short and exceedingly fine branchlets in forwardly and outwardly directed (not retroverted) arrangement. Strobilo- plumiçonie of the usual shape and structure. 114 ART. 1. — I. IJIMA : HEXACTINELLTDA, 111. ROSSELLID.^. Lyssacine liexasterophora of cup-like or sacciform body ; sometimes stalked ; generally firmly attached at base and exceptionally rooted by tufts of basal spicules. A few secondarily formed oscula may occur besides the main terminal osculum. Dermal skeleton composed of small dermalia with a variable number of rays and of large hypodermalia . The latter are generally pent- actins which often show a tendency to protrude out- wards in such a way that the paratangentials form a veil-like coverin o- over the dermal surface. The der- malia, when hexactinic, have the distal ray not pinular but much like the rest in appearance. The hexasters are various but mainly oxyhexasters and discohexasters, these generally occuring together; but sometimes one kind occurs to the exclusion of the other. Oxyhexas- ters are often hemihexactinose and hexactinose. Disc- ohexasters frequently modified into discoctasters. Though this family has not formed a subject for treatment in the present Contribution, the diagnosis is here given for the sake of comparison with those of other lyssacine hexasterophorous families. i'OSÏSCillPT I. SCHRAMMEN 's HEXACTINELLID SYSTEM. 115 Postscript I. At a time when tlie priiUiiig of this memoir was iiearlv finished I received A. Schrammen's paper "Zur Systematik der Kieselspoiigien " (Mitth. Roemer-Museum. No. ID. Hildesheim 1903). This contains an emendation of the Hexactiiiellidan system advanced by the same writer in "02 and to wliich I have had occasion to refer critically in the foot-note on pp. 23-25 of this Contribution. It is satisfactory to observe that Schrammen's new^ system shows a great approach to that which I have in my mind and of which I have made a brief exposition in the foot- note just mentioned, though it still differs from mine in an important point, as will presently l)e pointed out. Schrammen accepts F. E. Schulze's Amphidiscophora and Hexasterophora as the two suborders of the order Hexactinellida, while the little known pakeozoic forms formerly put together by him under a distinct suborder, the Stauractinophora, are placed in an appendix under the families incertœ scdis. This is, I think, quite in accordance with the present stage of our knowledge about the sponges in question. The Hexasterophora are divided by him into two tribes, the Hexactinosa and the Lychniscosa ; the former defined as " Hexasterophora, deren Stützgerüst aus Hexactinen besteht " and the latter as *' Hexasterophora, deren Stützgerüst aus Lychnisken besteht." It is decidedly an advance that the lychniscophorous forms (the Lychniscosa) are brought under the Hexasterophora, whicli they certainly are. The tribe Hexactinosa Schr. is made to comprise three subtribes, the Uncinataria, the Inermia and the Euplectellaria. Of these, the 116 ALT. 1. — I. IJIMA : HEXACTINELLIDA, lit. first t 111 the last exactly correspond in scope, each to each, to two of the three Hexastero}fhora tribes in my system which I have provisionally designated witli the fiist three letters of the alphabet (p. 25, /. c), namely, to the tribes C and A respectively. On the other hand, Schrammen's luermia differs from the tribe B of my system in that all the lychniscophorous forms (which, as above mentioned, are made by him into the Lychniscosa) are excluded from it ; whereas I place them under the tribe B, together with, but as representing a family or families distinct from, the Dactylocalycidae {Dactylocalyx, Jfyliusia, etc.). This is the point in which I stand in disagreement with Schuammen, — a disagreement which I am strongly inclined to regard as due to an over-estimation, on his part, of the lychnisc as a systematic character. The diagnoses of the Hexactinosa and the I^ychniscosa, as given by Schrammen and cited above, appear to me as wholly inadequate to characterize the grou])S. The Hexactinosa is stated to have the supporting framework composed of Jiexactlns. This may be said to hold good for the dictyonine forms of that grou]) but not for all : a mere reference to the parenchymalia in the Euplectellaria is sufficient. Be that as it may, the lychnisc, which should characterize the Lychniscosa as opposed to the Hexactinosa, is, fundamentally, likewise a hexactin or a part of a hexactin ; it is clearly a secondary structure or complication which has for its basis hexactins, such as compose an ordinary dictyonal skeleton (see Marspiall u. Mayer, Mittli. kgl. Zool. Mus. Dresden, 1877, p. 267 ; F. E. Schulze, Hexactinelliden des Kothen Meeres, 1900). Therefore, the supporting framework of the Lychniscosa is, in my way of thinking, as really composed of hexactins as that of certain Hexactinosa. The Ivchnisc as a POSTäCtllPT I. SCHRAMMEn'ö HEX ACTINELLID SYSTEM. Il7 systematic character can only be utilized to characterize a group standing subsidiary to another and more comprehensive grouj) in which the supporting skeletal framework consists of hexactins, — of dictyonally fused hexactins, I may add. Thus, the Hexactinosa and the Lychniscosa, as defined and placed in the system by Schrammen, seem to lose all ground for existence and may be entirely dispensed with. On the other hand, if it be justifiable, as I believe it is, to consider Aulocystls as the living representative of all the lychnisco- phorous Hexactinellids that have existed, there can be no tangible reason for not receiving these into (he group Inermia. F. E. Schulze (Chall. Rep.) placed Aulocystis in the same inermate family (Mœandrospongidœ) with Dactylocalyx, Mylmsia etc., and there was a time when it even passed under the name of Myliusia. While I do not follow F. E. Schulze's opinion as to the familv to which Aulocystis should belong, I can not but agree with him in regarding it as one of the Inermia. Granted this point, the three subtribes Schrammen 's may be called the tribes — exactly the same as those in my system — into which the suborder Hexas- terophora may be directly divided. It may here be noted that the division I have adopted of the Hexactinellida into the suborders, tribes and subtribes is in complete agreement with F. E. Schulze's idea of the Hexacti- ricUid phylogeny as ably enunciated by him in the end of the Challenger Eeport. If graphically represented, so flir as it goes would take essentially the same appearance as the genealogical tree given by that author /. c. page 495. Tokyo, April öth, 1903. 118 AKT. 1. — I. ijiîiA : heXactinellida, iii. Postscript II. Following Schramm en's paper above referred to, I have received, jiu-t in time to add this postscript, F. E. SniUEZE's latest Hexactinellid work in which Caulophacus ardlcm (A. Hansen) and Calycosoma gracile F. Fl Sch. nov. spec, are des- cribed in detail (Abb. kgl. preuss. Akad. Wiss. 1903). With Caulophacus arclicus, which comes from a depth of 1977 m. in the Northeast Atlantic, a sixth species is added to the five I have enumerated in this Contribution. As pointed out by F. E. Schulze, it shows in the spiculation an extensive agreement with, and therefore seems to be most closely related to, C. latus of the South Indiau Ocean. So far as can be made out with certainty from the descriptions and figures relating to both, the following seem to be the most important features that characterize C. arclicus as distinct from 0. latus : 1. The j^inular ray of the dermalia is narrower and i)ointed at the apex, being thus spindle-like instead of ovoid.'" 2. The slender pinular ray of the gastralia is shorter by one-half or more, measuring 250- 500/^ in lenoth as ao-ainst 1mm. or more of G. latus. 3. The occasional presence of hemihexactinose pachydiscohexasters in addition to hexactinose forms.f 4. The tufts of terminals to * In the Cliall. Report (p. 125) the dermal pinular ray in C. InHis is stated lo be usually 50 ij. long; but this scarcely accords with the size of its figure ns given /. c PI. XXIV., fig. 10, magnified 100 times. One is led to suspect a typographical error or errors in this connection ; but if it be that the scale of magnification^appended is correct, tlie dermalia of C. Ia!us mu.st be said to have the pinular ray strikingly larger than that of the same spicule in 'C. arclicus, — a fact which might conveniently be made use of as one of the dißerential points between the two species. t F. E. Schulze (/. c. p. 8-11) disapproves of applying the term hexaster to the quasi- hexactiu called by me tlie liexactinose hexaster. Grounds for my persisting to use tliis appella- tion antl tlie advantage to be derived therefrom, will be dealt with in another jiublication. POSTSCRIPT II. CAULOPHACUS ARCTICUS & SYMPAGELLA GRACILIS. 119 lophodiscohexasters are considerably broader, they being of a narrowly conical sliape in C. latus. The last point is considered by F. E. Schulze to constitnte the chief difference between the two species. I have gone into the comparison somewhat more critically than F. E. Schulze did, simjily in order to contrast them, as far as possible, with respect to the points utilized by me for distinguishing G. Intus in the key given on p. 86 of this Contribution. The diagnosis of the genus (p. 85) is in no way affected by the addition of the sixth species. As regards Calycosoma gracUe F. E. Sch., I regret, with (hie deference to the judgment of so high an authority as its describer, that I can not readily accept the generic denomination «iven to it. In studvinii; its characters as embodied in the de- scription, one will at once be struck with the close resemblance to my Sympagella anomnla in all points of the organization. In fact, I can scarcely discern in it any noteworthy difference from S. anomala beyond the facts that parenchymal oxyhexactins (up to 1.5 mm. axial length) and oxydiactins (3-Ö mm. long and GO- 100 /^ broad) are much more strongly developed and that there occur oxyhexasters in addition to the onychaster and the strobilo- plumicome. It is possible that the two forms in question may in the future be proved to be specifically identical and that the differences mentioned may be found to be simply matters of individuality ; but for the present they may on the ground of those differences be allowed to stand as two very closely related species. Be this as it may, it seems certain that they can not possibly belong to different genera. Eithe]- my Sympagella anomala is to be re-christened as Calycosoma anomalum, or Calycosoma gracile as Sympagella gracilis ; while a further possibility is that they may have to be put into a third ^enus, in which çasç 120 ART. 1. 1. IJIMA : Hi:XACTINRLLIDA, Hl. Äulascus (supposing this to be tenable as a genus distinct from Sympagella), as being the nearest to both, comes in question, since the ci'eation of a new genus for their reception is scarcely to be thought of. The question whether Ca^ycosoma should be amalgamated with either Aulascus or Sympagella, or with both together, seems to be uncalled for at present. I completely concur with F. E. Schi lzc (/, c, p. 20 ; vide also my Contrib. I., p. ob) in regarding the onychaster and the oxyhexaster as two hexaster-forms of comparatively trivial differ- ence. At the s^ime time it will be conceded by all that the onychaster is to be considered as a discohexaster with the terminal disc in the most rudimentary state of development, being in fact represented by a whorl of fine claw-like branches. Now in Aulascus {A. johnstoni) the hexasters in question are all disc- ohexasters, in which the terminal whorl of prongs is by no means strongly developed and many of which indeed present a resemblance to typical onychasters (F. E. Sen., '97, p. 527). On the other hand, Calycosoma (as represented by the single species C. validmn) has the corresponding hexaster represented solely by oxyhexasters ; it is of no moment that some of these are in the hexactinose foi'in. Now the new species C gracile shows the same hexaster partly in the form of oxyhexasters and partly in onychaster-like forms ; so that it may be said that in this respect it stands midway between Aulascus and Chlycosoma. It merges into the former through the onychaster-like hexaster and into the latter through the oxyhexaster. Eecourse must then be taken to some other differential indications than the hexaster in order to decide to which of the two genera the new species is more closely allied. So for as tlie spiculation goes, a point that can be utilized as such an index seems to be found in the dermalia and the gastralig. POSTSCRIPT II. CAULOPIIACUS ARCTICL'Sà SYMPAGELLA GRACILIS. 121 These agree almost completely with those of A. johnsloni in general appearance but especially in the character of the pinular ray ; whereas, they, as compared with the same of C. vaUdum, present (to use F. E. Schulze's words) "eine auffällige Differenz" in the development of the lateral spines. It is clear what this points to. Perhaps another not unimportant difference in the spiculation is the presence in C. validinn of prostal needles in tufts, which are totally wanting in the new species as well as in A. johnsloni. Further taking into our consideration the occurrence in the new sj^ecies of bud-like prominences on the wall which may lead to the formation of such incompletely individualized persons as are known in Atilascus, and of a distinct, branched stalk known to be common in the nearest allies of that genus, I can but think the evidence is decidedly in favor of considerinoj the new species to be nearer to A. johnstoni than to Calycosonia validimi. If the two genera are to be kept separate, it should rather be placed under Aulascus. But since I hold this genus as unitable with Sympa gella (p. 96), I should accept F. E. Schulze's new species into my system under the name of Sympagella gracilis (F. E. Sch.). With this change, the diagnosis of the genus Sympagella as given by me in this Contribution (pp. 96, 113) requires alteration only in so ftir as the oxyhexaster should not now be excluded from among the hexasters of the genus. The passage concerning these should be made to read " Besides discohcxasters, ivhich are sometimes accompanied ivith oxyhexasiers, plumicomcs are always vresent.^^ The family diagnosis given on pp. 84 and 112 may remain as it is. There is no denying the fact that Sympagella. gracilis brings the Caulophacids {Caidophaciis and Sympagella), whether as a 122 - AET. 1. 1. IJIMA : HEXACTINELLIDA, III. distinct family as I have made them or as a group taken over into the RossellidcT, into direct touch with the I^annginellinge, the contact point on tlie part of this Kossellid subfamily being- found in Calycosoma validum. And whether or not this genus and species should be joined to the rank of the Caulophacids, now appears to be largely a matter of individual opinion. Tokyo, April loth, 1903. LIST OF LITERATURE. 123 List of Literature referred to in this Contribution. Ijlma, I. '96. Notice of New Hexactiaellida from Sagami, Bay, IL — Zool. Anz., 1896, No. 504. ,, '98. The Genera and S[)ecies of Ro.sscllidte. — Annot. Zool. Jap., Vo]. II., Pars. If. „ '01. Studies on the Hexactiaellida. Contributi'in I. — Jour. Soi. Culh, Tokyo. Vol. 1,5. „ '02. Studies on the Hexactinellida. Contribution IL — Jour. Sei. Coll. Tokyo. Vol. 17. Schmidt, 0. '70. Grundzüge einer Spongienfauna des Atlantischen Ge- bietes. Schrammen A. '02. Neue Hexactinelliden aus d(T oberen Kreide. — Mittli. aus dem Roemer-Museum, Hildesheini. Nr. 1.5. Schulze, F. E. *86. Ueber den Bau and das System der HexactinelUden. ,, '87. The Challenger Report. Hexactinellida. „ '95. Hexactinelliden des Indischen Oceanes. IT. Die Hexas- terophora. — Abh. kgl. preuss. Akad. Wis?. Berlin, 189.5. ,, '97. Revision des Systems der Asconematiden und Ros- selliden. — Sitzber. kgl. preuss. Akad. Wiss. Berlin, 1897. ,. '99. Amerikanische Hexactinelliden. Jena, 1899. „ '00. Hexactinelliden des Indischen Oceanes. III. — Abh. kgl. preuss. Akad. Wiss. Berlin, 1900. ,, '00«. Mittelmeer-Hexactinelliden. — Denkschr. math.-natur- wiss. Cl. d. k. Akad. d. Wiss. Wien, Bd. LXIX. ,, '00^. Die Hexactinelliden. In ; Fauna Arctica, Bd. I. Tüi'SENT, E. 'Ol. Éponges nouvelles des Açores. (2'"° série) (1). — 3Iéin. Soc. Zool. de France. T. XIV. „ '01''<. Spongiaires. Résultats du Voyage du S. Y. Belgica en 1897-1899. (Expédition Antarctique Belge). 124 CONETi\TS. Contents. Placosoma paradiclyum, n. g., n. sp 2 Spiculation H Oliservalions on the Euplectellidœ generally 19 Leucopsacidae 29 Leucipscicus Ij o4 Leucnpsacua orthodocus Ij 154 Spiculation ,"6 Notes on the soft parts and the larva 40 Leucopsacus scoliodocus Ij 46 Spiculation 48 C'liavnopledella Ij 58 C'liaunopk'ctella cavanom Ij ,. ... 53 Spiculation 59 C'huunopteclella sjjiuifera, n. sp 71 Spiculation 72 Caulophacidae 78 Caulophacus F. E. ScH 85 Caidophacus lolifoltiun, n. sp 87 Spiculation 90 SijmpageUa O. ScHM 9(') Sympcujella anoniala, n. sp 97 Spiculation 108 Diagnoses of the families, genera and species treated of in this Contribution 108 List of literature referred to in this Contribution 122 Postscript 1 115 Postscript II 112 I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION HI. PLATE I. Placosoma paradictjum, y. g.; x. sp. Plate I. riacûsotna paradictijuin Ij. Pp. '2-19. Fig. ]. The type-specimen, seen from the iVont side. (Sei. Cull. Mus. Sp. No. .506). Fig. 2. Same, seen from the Lack side. Both ligures reduced to a size slightly smaller than half natural size. Ijiina, Hcxactiudlida. Cotitiihution III. Jour. Sei. Coll. Vol. XVIII, Art. 1, PI. I Placosoma paradictijum Ij. J]». 4 IMP. THE miO PWHTlItü » I. IJIWA. STUDIES ON THE HEXACTINELLIDA. CONFRIBUTION III. PLATE IL Placosoma paradictyum, x. g.; x. sp. Plate II. Placosoma 'paradictynm Ij. Pp. 2-19. Fig. 1, A dermalia fiMm the frontal lattice, with greatly elongated proximal ray. 1.50 x . Fig. 2. A dermalia from the frontal lattice, with short proximal ray. Lateral view. 1.50 x . Fig. 3. A gastralia from the gastral cavity inside the primary or main osciilum. 1,50 X . Fig. 4. A raediiim-sized hexactinose discohexaster from a hypodermal beam of the frontal lattice. 300 x . Figp. .5, G. Hemihexactinose discohexast( rs of occasional occurrence. From the back side of the sponge. 300 x . Fig. 7. A large hexactinose discohexaster, from the same side. 300 x . Fig, 8. A hexactinose codonliexaster from the main gastral surface. 300 x . Fig. 9. A normally developed discohexaster, occasionally fourni together with the large spherical form of fig. 11. From the back side. 300 x . Fig. 10. Another occasional form of discohexaster (codonhexaster) from the same side. 300 x . Fig. 11. A large, spherical discohexaster from the same side. 300 x. Fig. 12. A portion of the same, showing the central parts. 300 x. Fig. 13. A small portion of the frontal latti(H\ Above, the dermal lat- ticework. Below, the hypodermal beam. .50 x . Fig. 14. Dermal surface of the back. 50 x . Fig. 1.5. A small portion of septum from the choanosome. ;50 x . Fig. 16. Surface of the main gastral cavitv. 50 x. Jjinui. HexactineUida. Coutiihution 111. Jour. Sei. Coll. Vol. XVIII. Art. 1. PI. II. Placosoma paradictyum Ij. LITH & IMP. THE TOKIO PRIMTING Ki I. IJIIIA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION III. PLATE III. Leucopsacus ortliodocus Ij. Leiicopsacus scoliodociis Ij. This plate also contains fignres (1-13) taken from two small specimens of Staumcahjplus sp., not specifically determinal)le with certainty on acconnt of their yonnsj state. Both are from the Sagami Sea. References to these figures will be made in a future Contribution relating to the Rossellidfe. Figs. 1-6. Young Staurocalyptus sp. (Sei. Coll. Mas. Sp. No. 437). Fig. 1. Spicules of dermal suvfiice, seen from outside. 100 x . Fig. 2. S[)ictiles of gastral surface, seen from inside. 100 x . Figs. 3, 4. Common oxyhexasters, 300 x . Fig. b. Microdiscohexaster. 300 x . Fig. G. Portion of a discoctaster. 300 x . Figs. 7-13. Young Staurocalyptus sp. Fig. 7. Surface-view of dermal layer. 30 x . Fig. 8. Combination figure of spiculation of the wall. 30 x . Fig, 9. Entire specimen. 2 x . Fig. 10, 11. Deformed discoctasters. .300 x. Fig. 12, 13. Oxyhexasters. 300 x . Figp. 14-26. Leiœopsacns ortJiotïocus Ii. Pp. .34-46. Fig. 14. One of the type specimens (Sei. Coll. Mos. öp. No. 230). Nat. size. Fig. 15. Longitudinal section of wall stained ; above, the oscular edge. (Sei. Coll. Mus. Sp. No. 438). 3 x. Fig. 16. Hexactinose discohexaster of an average size. 300 x. From Sp. No. 230. Fig. 17. Hexasteroiis discohexiister, from the same specimen. 300 x . Fig. 18-20. Same, from Sp. No. 438, Fig. 20 represents an occasional form. 300 x . Fig. 21. Egg-like cell of doubtful origin and nature, found in S[).No.230. 300 x . Fig. 22. A moderately large-sized arcluxjocyte-congeries, drawn from a section. The spherules represent nuclei. 300 x . Fig. 23, Larva in an early development stage, yet without spicules. Seen in optical section. 300 x . Fig. 24, Older larva of alwut 70// diameter; in o[ttical section, but all the si)icules in a hemisphere, seen by dil'i'erent foci of the mi- croscope, arc drawn in. Histologic il elements in th.e central ])art a little too distinctly shown. 300 x , Fig. 2.5. Oldest larva observed ; all the spicules of the upper hemisphere drawn in. 300 .< . Fig. 26. Combination of sections to show spiculation of wall. Above, the dermal ; below, the gastral, surface. About .50 x . Figs. 27-37. Leiccopsacws scoliodocus Ij. Pp. 46-.52. Fig. 27. Type specimens. The largest in the middle is Sei. Coll. Mus, Sp. No. 233 ; the two others are from Sp. No. 23.5. Nat. size. Fig. 28. Stained longitudinal section of an entire s[)ecimen. Basidictyonal mass at base. 3 x . Fig, 29, Hexactinose discohe.Kaster, 300 x , Fig. 30. Portion of same, showing the extent of axial threads in the central node. Fig. 31. Delicate tylflorioome of inconstant occurrence, from Sp.No.434. 300 x . Fig. 32. Spherical discohexaster, from Sp. No. 233. 300 x . Fig. 33. Portion of same, showing the short principal. Fig. 34. Small form of spherical discohexaster, from Sp. No, 235. 300 x . Fig. 35. Unusually large discohexaster, occassionally met with. From Sp, No, 233. ,300 X. Fig. 36 Ptare form of discohexaster. From Sp. No. 235. 300 x . Fig. 37. danbination of sections to show spiculation of wall. Above, the dermal ; below, the gastral surface. About .50 x . Ijima, Hexactùiellida, Contribution III. 1 . ^ 3 Jour. Sei. Coll. Vol. XVIII. Art. 1, PI. III. ^ -^^ '^' ^^/ ^^^^ 22 2 -\^^^' ^"^ ->. ^ 12 / \ IS I . tV^ 29 r "^ 30 vL 23 24 2Ô 34 7 V : 33 37 1 — 6. Staurocalijptiis sp. 14 — 26. Leucopsacus orthudocus I.j. 7 — 13. Staurocalyptus sp. 2'}' — 37, LeiicopsacuR alwliodoats Tj IITH. & IMP THE TOKIO PRINTING CO 1, IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION III. PLATE IV. Chaunoplectella cavernosa I-: Plate IV. ClLCLunoplectella cavernosa Ij. Pp. 53-71. Fig. ]. Portion of tlie wall of a large individual ^ nat. sizr. AU the figures in tliis plate taken from this specialen (Soi. Coll. Mus. Sp. No. 443). Fig. 2. öigmatocoine. 300 x . Same as fig. 11 of the following Plate. Fig. 3. Two terminals from the same. About 1000 x. Fig. 4. Section of the wall. About nat. size. Above, the dermal surface. Arrows show the direction of water current. Fig. 5. Portions of spicules in the periphery, a, I), pentactinic and tniiactinic dermalia. c, d, two unequally rayed parenchymalia. About .50 x . Fig. 6. Parenchymal hexactins of different sizes (a, b, c). About bO x. Fig. 7. Dermal layer seen from outside, with soft dermal reticulum. .50 x . Fig. 8. Oxyhexactinic canalaria, witli a sigmatocome (fig. 2) at tiie end of the freely projecting ray. 100 x . See fig. 12 of the following Plate. Fig. 9. Lariie discohexaster with anchorate terminal umbels. 300 x . Jjinia, Hexactinellitia, Contribution III. Jour. Sei. Col I. Vol. XVIII. Art. I, PI. IV. 5 Chaitnoplectella cavernosa Ij. LITH, &. IMP THE TCrJO ^!?JTiNG CO. 1. I.IIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION III. PLATE V. Laniiginella pupa O. Scum. Cliauiioplectella cavernosa I J. Cliaunoplectella spinifera Ij. The figures (1-7) relating to Luniujindla pupa will be referred to in a future Contribution. Plate V. Figs. 1-7. LaniKjinella pupa 0. ^5chm. Figs. 1-3. Discohexasters from one and the same specimen (8ci. Coll. Mus. Sp. No. 43G). 300 x. Figs. 4-6. Same from anotlier specimen (Sp. No. 234). 300 x . Fig. 7, Portion of a stroLilo})lumicome. 1000 x . The short central boss on the terminal-hearing knob at the end of each ])rinci[)al ray has been omitted through oversight. Figs. 8-13. CliaimoplecteUa cavernosa Ij. Pp. 53-71. Figs. 8, 9. Young specimens (Sei. Coll. jMus. Sp. No. 407) of various sizes, attached to a dead Chonclasma. Nat. size. Fig. 10. A dried specimen complete (0. C. No. 4389). ^ nat. size. Fig. 11. Sigmatocome. 1000 x. Fig. 12. Septal wall in section. Above, tlie incurrent surface ; belijw, the cxcurrent surface. Above 25 x . Fig. 13. Basidictyonal framework. 70 x . Figs. 14-17. Chaunoplcctella spinifera Ij. Pp. 71-77. Figs. 14-17. Different forms of discohexasters found in the specimen (Sei. Coll. Mus. Sp. No. 435). 300 x. Ijima, Hexactinellida^ Cotitribntion III. 1 9. o \1 // ^/ 'f^ r/ Jour. Sei. Coll. Vol. XVIII. Art. I, PI. V. 65 4 '^/i L 15 16 ^ 12 U ? 1^ 13 vw^ X 1—7. Laniiginella pupa O.SCHM. S — 13. Chmuwplectclla cavernosa Ij. 14 — 17. Chaiinoplectella apinifera Ij. LITH. 4 IMP. THE TOKIO PRINTING CO. I. I.TIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION III. PLATE VI. Chaunoplectella spinifera Ij. Staurocalyptus japonicus Ij. Staurocalyptus tubulosus I J. Tlie figures relating to Stmroea'iipUs j.tpo:itcJ,s an;l tu'mlosu^ to he referred to in a future Contribution, Plate VI. Figs. 1-8. Chcnmoplectella spinifera Ij. Pp. 71-77. (All figures from Sei. Coll. Mos. Sp. No. 4.59). Fig. 1. The type-specimen. Natural size. Fig. 2. Oxypeutactiiiic dernialia with spine-bearing paratangential rays. 100 X. Fig. .3. Si)herical form of discohexaster from the peri[)hery of wall. 300 x . Fig. 4. Larger spherical form of discohexaster from deep parts. 300 x . Fig. 5. Discohexaster with terminals in six separate tufts. From deep parts. 300 x . Fig. G. Discohexaster of similar form ])!it with very slender terminals. From deep parts. 300 x . Fig. 7. Largest form of discohexaster. From deep parts. 300 x . Fig. 8. Combination-figure to show the arrangement of skeletal elements in the [)eriphery of the wall. Above, the dermal layer. 25 x . Figs. 9, 10. Staurocalijptus japonic us Ij. Fig. 9. Sei. Coll. Mus. Sp. No. 403, seen from oscular end. Natural size. Fig. 10. Same specimen in lateral view. Natural size. Figs. 11-17. Staurocalijptus tubulosus Ij. (All figures from Sei. Coll. Mus. Sp. No. 241). Fig. 1!. The type specimen. Natural size. Fig. 12. Dermal layer seen from outside. Several dcrmalia and two hypo- dermal pentactins. 100 x . Fig. 13. Portion of a jiaratangential ray of the prostal or hypodermal pentactin, to show the character of its surfice. 303 x . Figs. 14, 1.5. Stauractin;c and psntactinic dermalia. 300 x. Fig. IG. Discoctaster from dermal side. 303 x . Fig. 17. Oxyhexaster. 300 x . Jjitna, HexactindUda. Contribution m 4 Jour. Sei. Coll. Vol. XVIII, Art. I, PI. VI 1 — 8. ChajinopJectella spinifera If 9 — 10. Staurocalyptus japonicHS Ij LITH ilMP.THf TOKIO.PRINTINS CO 11—17. Staurocalyptus tubulû^us I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IIL PLATE VII. Caulophacus lotifolium Ij. Plate VIL Caulophacus lotifolium Ij. Pp. 87-96. Fig. 1. The tyj->e specimen, now preserved in the " Kgl. Naturalien Kabinet" in Stuttgart. About i natural size. Fig. 2. Gastral hexactinic pinule of average size. 300 x . Fig. 3. Dermal hexactinic pinule of average size. 300 x . Figs. 4, 5. Normally developed, thick-rayed discohexasters from subdermal or subgastral region. 300 x . Fio-. G. Hemihexactinose form of same. 300 x . Figs. 7, 8. Hexactinose form of same ; from deep parts. 300 x . Figs. 9-11. Pentactinic dermalia from the lower part of stalk. 300 x . Figs. 12-16. Slender-rayed discohexasters, which pass over by gradations into the forms of figs. 4-8. Figs. 14-16, onychaster-like. From peripheral region of the sponge. 300 x . Fig. 17. Central part of a hexactinose discohexaster, showing the extent of the axial cross. Observed in glycerine. 1000 x . Fig. 18. Part of a section through the sponge, with soft parts preserved. Above, dermal layer (ectosome). sd. s. subdermal space. Below, periphery of choaoosome. ex. c, excurrent canal 50 x . Fig. 19. Dermal skeleton, seen surftice on. Beneath the dermal latticework, the hypodermal lattice with larger meshes. 50 x . Fig. 20. Surface of the stalk in oblique view. Above, the crowded layer of balloon-shaped pentactinic dermalia. 50 x . Fig. 21. Part of a section, adjoining the gastral surface. Below, gastral layer. 50 x . Ijima, Hexaetinellida, Contrih . III. Jour. Sei. Colt. Vol. XVIII. Art. 1, PI. VII. .? 3. Caulophacus lotifolium Ij. flviied ly KoMu-Kmla Tok/o Jif- I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION HI. PLATK VIIL Sympagella anomala Ij. Plate VIII. SympagcUa anomaJa Ij. Fio'. 1. A niimLer of small and young specimens attached on an old dead stalk of the same species. J nat. size. (Sei. Coll. Mns. Sp. No. 455). Fio-. 2. A medimn-sized specimen, attached to a \^\ece oï dead Chonel asm a. I nat. size. (Sei. Coll. Mus. Sp. No. 367). Fig. 3. A large specimen, with a part of the wall removed to show the gastral surface. | nat. size. (Sei. Coll. Mus. Sp. No. 355). Figs. 4, 5. Dermalia from different specimens. 300 x . Fig. 6. Modified dermalia from the lower part of stalk. 300 x . Figs, 7, 8. Gastralia from different specimens. 300 x . Fig. 9. Canalaria from excurrent canal. 300 x . Fig. 10. Strobiloplumicome. 300 x . Fig. 11. Part of a strobiloplumicome, partially constructed to show the structure. 1000 x . Figs. 12-14. Onychaster-like hexasters. 300 x . Fio". 15. Outer end of a terminal from an onychaster-like hexaster. 1000 x . Fig. 16. Peculiar discohexaster, with branched terminals, — an inconstantly occurring form. 300 x . (From Sei. Coll. Mus. Sp. No. 473). Figs. 17-19. Discohexasters from stalk. 300 x . Fio-. 20. Dermal skeleton (dermalia and hypodermalia), seen surface on. About 40 X. Fig. 21. Gastral skeleton (gastralia and underlying parenchymalia), seen sur- face on. About 40 x . FIl'. 22. Peripheral part of a section through the sponge-wall. Above, the dermal layer ; below, the periphery of choanosome. About 40 x . Fi'^ 23. Part of a section through the wall. Below, the gastral layer. About 40 X . Ijivia, Hexactinellida, Coiürih.IU. Jour. Sei. Coll. Vol. XVIII. Art. I, PI. VIII. Sympagella anomala Ij. /hnUd iy fCodiUa /inu/a. Tckj/o. JofHUv. I JOURNAL OF THE COLLEGE OF SCIENCE, IMPERIAL UNIVERSITY, TOKYO, JAPAN, VOL. XVIII., ARTICLE 2. Cretaceous Cephalopoda from the Hokkaido. PART I. LytoceraSf Gaiidryceras and Tetraf/onites. By Hisakatsu Yabe, RigahisM. Geological Institute, Science College, Imperial University, Tokyo. With 7 plates. Introductory Remarks. In 1877, Benjamin Smith Lyman first announced the oc- curence of Cretaceous Ammonites in the Hokkaido, and his collec- tion was examined by Edmund Naumann who published a short account of the fossils without any detailed description.^^ A more detailed account appeared in the work of Prof. Matajirö 1) E. Naumann : Ueber das Vorkomnien der Kreideformation auf der Insel Jezo. Mitth. d. deutsch. Gesell, für Natur u. Völkerkunde Ostasiens, Ed. XXI. 18S0. p. 28. The following names of Ammonites are mentioned in this work: — Lytoceras Sacya Forbes. StoUczhda sp. Phylloceras n. sp. afF indra P'orbes. Ainsoceras (eiiuimlcatuin Forbes. „ cfr. subalpinum d'ORBiüNY. „ sp. „ Velledœ MiCHELfX. Anisoceias n. sp. Amallheus swjcda Forbes. Flychoceras gaultinum Pictet. Hcipl'iceras n. sp. „ n. sp. „ planvlatum So-\VERBY. ,, n. sp. „ Gardeni Baili". In a preliminary note in the same volume, D. Erai'NS mentions an Ammonite from Ui'akawa as identical with Siephanoceras coroiiatum Brtjg. 2 ART. 2. H. YABE : CRETACEOUS YoKOYAMA,'^ who described and figured Cretaceous fossils, not only of the Hokkaido, but also of other parts of Japan. This work was indeed the first which treated systematically of the Cretaceous fauna of the Hokkaido, but the nature and extent of the Cretaceous forma- tion on that island was not known until the appearance of the publica- tions of Prof. KoTORA JiMBO and his assistants who, for several years, were occupied in the geological survey of the whole island. Jimbö^^ 1) M. YoKOYÄMA: Versteinerungen aus der Japanischen Kreide. Palaeontograpliica. Bd. XXXVI. 1890. The following species of Ammonites are described in this work: — rhyJloeeras Vcll.erlœ Michelin. „ ezocnse Yokoyama. Ijijioceras Sacya FOBBES. „ sp. „ sp. Ptychoceras pseudogauUinum Yokoyama. AniaocercK suhqua/lraiuni YoKOYAMA. „ Haradanuni Yokoyama. „ subundulaluin YoKOYAiMA. Anisoceras cfr. rugaium I'oRBES. „ sp. Desmoceras surjata Forbes. „ sp. „ Gardeni Baily. ,, gaudama FoREES. Pachydiscu'^ ariyalurensis StoliczkA. „ Sutneri Yokoy'AMA. „ Naumanni Yokoyama. 2) K. JiMBo ; Beiträge zur Keiintnii-s der Fauna der Kreideformation von Hokkaido. Palœontologische Abhandlungen. Neue Folge, Bd. II. Heft 3. 1894. In this work, besides Phylloceras Velledœ Mich., P. ezoënse Yok., H) ami ilie general us:ige of Japanese geologists. 2) H. Yabe: Notes on s )nie Shark's Teeth from the Mesozoic Formation of Japan. Journ. Geol. Sue. Tokyo. Vol. IX. No. 110. l'JU2. 4 ART. 2. H. YABE : CRETACEOUS age. European authorities, such as A. de Lapparent/^ Ernst Koken'-^ and Franz Kossmat^' believed the existence of different horizons in our Cretaceous upon the palaeontological data, corres- ponding in age to the different subdivisions of the European upper Cretaceous. On the other hand, J.. Böhm^' and R. Michael^^ noticed tlie occurrence of fossils characteristic of the Senoniau, while Rudolf Zueer*"'^ expressed the opinion that the Cretaceous of the Hokkaido and Saghalien contain lower Cretaceous type of Ammonites. 1) a. de Lapparent : Traité de Géologie. Quatrième édition. III. Iö99. Pp. lo46, 1361 and 1394. 2) Ernst Koken: Die Vorwelt und ilire Entwickelungsgescliichte. 1893. P. 421. 3) Fkanz Kossmat : Ueber die EeLleutung der sü lindisclien Kreideforniation für die Beurtheilung der geograpliischen Verbältnisse wübrend der späteren Kreidezeit, Jalirb. k. k. geol. Reichsanstalt, Wien. Bd. XLIV. 1894. Heft. 3, p. 470. KossMAT also criticized the Ammonites of the Hokkaido in liis " Untersuchungen über die südindische Kreideformation." He considers Pliyllnceras ezo'ênse YoK. as a form very closely allied to P. Forbesianum d'Orb., while he considers Uesmoceras Damesi Jimbö as probably identical with I), .vigntd Forbes. He brings Lylncerw^ slriatum Jimbö, L. crassicoslcUum JlMBo and demepUcaturn Jijibû under Gaudryceras and L. glabrvni Jjmbô under Telragonites. One of the many Hamite'i described as H. sp. by Jimbö is regarded by him as identical with //. {Anisoca'us) largemlcatua Forbes, and Acanthoceras rotomagense Defr. var asialica JlMBo as identical with A. Neivholdi Kossmat. Anisuccra>i cfr. rugatus of Yokoyama, according to KoriSMAT, is distinguished by its finer, obliquely arranged ribs from that of Forbes. Pachydisctis Dunisoiiianus vStol. has been divided into the three species of Hulcodiscus >--payicostatus, P. Jimboi and P. Denisonlanus, by Kossmat and all of them are quite different from what Jimbö described as such. He also doubts whether Desmocenxs hhlkawal Jimbö is not a Ilolcodiscua and takes D. gaudama of Yokoyama for a coarsely ribbed variety of D. indopacifica Kossmat. •J. F. Whiteaves of the Geological Survey of Canada, pointed out the occurence of Pachydiscus Ilaradai Jimbö and Hainites ohdrkiua Jimbö in the Cretaceous of Vancouver Island. (J. F. Whiteaves : On some Fossils from the Nanaimo group of Vancouver Cretaceous. Trans. Royal Soc. Canada. Stction IV. 1895. Pp. 130 and 132). In one of liis letters to Jimbö in 1895, he expressed the opinion that Pesmoceras Ishikawai might possibly . be identical with Haploceras (afterward considered as an Holcodiscus) cumsheiuaense Whiteaves ffom the Queen Charlotte Islands. 4) JoH. Böhm: üeber Ammolntes pcdenvjlis v. Buch. Zeitsclirift d. deutsch, geol. Gesellschaft. Bd. L. Heft. 1. 1895. P. 200. 5) R. Michael: Ueber Kreidefossilien von der Insel Sachalin. Jahrb. d. k. preuss. geol. Landesanstalt für 1898. 1899. P. 1Ü4. 6) Rudolf Zuber : Geologie der Erdöl-Ablagerungen in den galizisclien Karpathen. I. Allgemeiner Theil. Heft 1. 1899. P. 48. CEPHALOPODA FROM THE HOKKAIDO. 0 The present writer went to the Hokkaido mainly for the purpose of settling the question of the relation of the Cretaceous Ammonite-bearing deposits and the coal-bearing series, and was there during four summer vacations from 1899 to 1902 inclusive, thus having ample opportunities to make a valuable collection of the Cretaceous fossils and also careful researches in the strati- graphical order of the rocks. He proposes the following subdivisions of the Cretaceous deposits of the Hokkaido, although the thickness of each division has not yet been ascertained. Beginning from below, we have : I. The lower Ammonite-beds Avitli Orbitolina-limesione. IL The Triffonia-a-dndstone. a. Lower Accuitkocer as- zone or Trlgonia longiloha-zowe. b. Theiis-zone. c. Pec tunc ulus-zouQ. HI. The ujiper Ammonite-beds. a. Upper Acantlioceras-zowe. b. Scaphits-heà^. c. Fachydiscus-heds. The upper part of the Cretaceous passes gradually into the coal-bearing series. The layers of the Cretaceous complex are everywhere perfectly conformable to one another and evidently represent a continuous sedimentation. At the base of this formation, there is a thick complex of shales and sandstones ; in it the former predominates over the latter and contains in some places lenticular masses of limestone with Orbitolina concava Lam.'* This complex is what the present wiiter calls the lower Ammonite-beds. Above ]) H. Yabe: 0/'6t, p. 202, PI. XXVII. fig. 1.— T. AV. Staxtox and Diller: The Shasta-Chieo Series. Bull. Ueol. Soc. America. Vol. V. 1894. P. 446. — T. W. Staxton : Contribution to the Cretaceous Puhxiontology of the Pacific Coast : The Fauna of tlie Knoxville Petls. Bull. U. S. Geol. Surv. No. 133. 1895. P. 75. PI. XIII. figs 9—11. 3) The flattening of the body-whorl, wliicli is at fii'st nearly round, does not seem to be wholly due to meelianical action. It is, therefore, to be considered as one of the distin- guisldng characters of the species. 22.0 cm. Ratio. 100. 10.0 „ 45. 5.0 „ 22. 7.0 „ 31. 12 ART. 2.— H. YABE : CRETACEOUS The inner volutions, which gradually broaden anteriorly, are nearly round in section, showing only a slight lateral compression. The body-chamber is longer than one half of the last volution ; it is high and narrow. Umbilicus moderate in size ; relatively narrower on the last volution than on the inner. The surface of the septate portion is apparently smooth while the-body chamber shows radial farrows which are 15 in number ; the furrows are shallow and not well defined, becoming gradually obsolete toward the umbilicus. The interspaces are flat, slightly elevated, broadest along the ventral side, measuring 1.5 — 2.5 cm. in breadth. These flat interspaces and furrows extend also over the ventral surface. The suture line which is partly preserved, shows a lanceolate siphonal saddle with serrated margin. The external saddle is inclined inward and asymmetrically bitid, while the first lateral lobe is broad and deep. The suture line and the appearance of the septate portion of this species remind us of an Ammonite of the group of Lyioceras fimbi'ialum Sow. sp., while the suddenly enlarged and laterally compressed body-whorl, together with its peculiar sculpture, distinguishes it from all the known allies. Locality : — A clift' somewhat above the lowest gorge of the Ikushumbets, Prov. Ishikari, where the species occurs together with Puzosia subcorbarlca m. Horizon : — Lower Ammonite- beds. CEPHALOPODA FROM THE HOKKAIDO. 13 GAUDRYCERAS Ghossouvre emend. Kossmat. We have one addition to make to the remarks of Kossmat on the genus Gaudryceras, and that is respecting its suture line. Putting aside the question whether Gaudryceras is a well defined genus or not, it has been divided into two well marked sections, to the first of which belong G, Ayassizianum Pictet^^ from the upper Gault of Switzerland and G. 3Iarut Stoliczka"^ from the Utatur group of South India. The former shows the antisiphonal lobe prolonged and reflected on the preceding septum as illustrated in detail by Quenstedt."^ ; Ijut the suture line of the latter has been only partly figured. The second group, or the group of G. Sacya, contains more than thirty species, of which there are only two, — G. vertebratum Kossmat^' and G. varayurense Kossmat'^ — whose lobes have been examined ; but as to the septal lobe in these two species, there exists neither a drawing nor a description. However, after careful examination of several Japanese species of Gaudry- ceras, the writer was able to find three species showing the septal lobe just as in G. Ayassizianum Pictet. These three are G. tenuiliratum, m., G. striatum Jimbo sp. var. pictum m. (PL IV. fig. 6 c) and G. limatum m., each representing a particular type in the group of G. sacya. The presence of the septal lobe in the above three and in G. Ayassizianum leads us to presume that it is a character common to all species of Gaudryceras although 1) J. F. Pictet et W. Roux: ^Mollusques fossiles des Grès Verts des environs de Genève. 1847-53. P. 47, PI. IV. figs. 3, 4. 2) Stoliczka: 1. c. p. 1Ü2, PI. LXXIX., lig. 1.— Kossmat: 1. f. p. 34 (130). 3) F. A. Quexsteut: llandljucli d. Petrefuktenkunde. III. Auflage. ;SS5. P. 579, Pi. XLV. fig. 11. 4) Kossmat: i. c. p. 2G, PI. IV., lig. 2. 5) Kossmat: 1. c. p. 3U, PI. 1., ligs. 4, 5. 14 ART. 2. — H. YABE : CRETACEOUS it is not usually visible unless the specimen is in a very favorable state of preservation. All the species of Gaudryceras seem to be very imperfectly known. This is partly due to the fact that although they are very common especially in the upper Cretaceous deposits of the Indo- Pacific region, the specimens usually belong to immature ani- mals, and consequently closely resemble one another so that the determination is not only very difficult, but often quite im- possible. Äloreover, the aspect of the shell of this genus is so different in its younger and older stages, that without a large series of specimens for conq^arison, the larger and smaller forms are often liable to be separated into distinct species. There are already several instances of confusion among the species of this geiius. The full specific character, as it seems to the writer, appears always in the mature stage or at least after the shell has attained the middle stage of its life. The differences existing among the young indviduals of different species are very slight and often apparently less marked than those between individuals in one and the same species, while the shell of the full grown forms ol each species is w^ell characterized and easily distinguishable from its allies. However, owâng to the rapid enlargement of whorls in adult specimens and the extreme thinness of the shell, it is not always easy to get specimens with a well preserved shell. The largest Japanese species is Gaudryceras tenuiliratum, one of the specimens measuring over 30 cm. in diameter and being entirely septate, but still showing the sharp ribs as usual. The following is what Fr. Schmidt^^ wrote about an Ammonite from the Cretaceous of Saghalien which he identified with G. Sacya. 1) Fk. Schmidt: Die Petrefakten der Kreklefonnation von der Insel Suehalin. 1S73. P. IG. CEPHALOPODA FROM THE HOKKAIDO. 15 "Bei meinem grössten Stück, das über einen Fuss im Durch- messer hielt, und von dem ich nur ein Bruchstück mitgebracht habe, ist der Steinkern der äussersten Windung ganz glatt, nur über and über mit ineinander gewirrter Lobenzeichnung bedeckt ; die erwähnten dichteren Bippen sieht man an nächst der inneren Windung des nämlichen Stücks." The smoot state of the outer volutions of the older indivi- duals, as above referred to, is very common also in the Japanese species of Pachydiscus, but has never yet been observed in the Gaudryceras, occurring very abundantly in the Hokkaido and sometimes also in Amakusa. The writer is now acquainted w^ith nine species and five varieties of Gaudryceras from the Hokkaido, all of which belong to the category known as 'the group of Gaudryceras Sacya.^ This group is again divisible into six subgroups, chiefly according to the character of the full grown shells. I. Subgroup of Gaudryceras tenuiliratum Yabe. This subgroup is characterized by the full-grown whorl having numerous ribs of two kinds, a number of narrow^er ribs being inserted between each pair of broader ribs. The latter on a closer examination are found to be made up of two or three finer ribs placed close together. The whorls grow" exceedingly rapidly in height and breadth. The type is G. tenuiliratum^^ from the upper Ammonite beds of the Hokkaido and is very variable in form. G. varayurense Kossriat"^ sp. probably belongs to this subgroup, and G. mite 1) Vide p. 19. 2) Kossmat: 1. c. p. 2G (122), PI. IV (XVIII), fig. 2 a, h, c; PI. Ill (XVII), fig. 9. 16 ART. 2, H. YABE : CRETACEOUS Hauer^^ sp. from the Senonian of Go>ssau and France is also referable to this subgroup. II. Subgroup of Gaudryceras crassicostatum Jimbö sp. Whorls compressed on the sides, with broad ribs and finer inter- stitial ones. Suture line in young specimens more complicated than in those of the first subgroup. The type is G. crassicostatum JiMBö"^ sp. The two other species from the Hokkaido — G. densepli- catum JiMBö^^ sp. and G. striatum Jimbö^' sp. — show a sculpture somewhat deviating from this type species. Ammo7iites glanegensis Redtenbacher^^ from the Gossau formation of the north-eastern Alps, being intermediate between G. tenuiliratum and G. den- seplicatum, forms a connecting link between this subgroup and the preceding. G. subtililinealum Kossmat sp. is also probably referable to this subgroup. III. Subgroup of Gaudryceras limatimi Yabe. The stride on the whorls in young specimens are so extremely fine that the surface appears to be almost smooth. The adult shells present ribs which are broad and elevated, and crow^led together without any interstitial stride as in the foregoing subgroup. Suture line very complicated, characterized by the bipartite second 1) Fr. v. Hauer: Neue Cephalopoden der Gosaiigebilde. Sitznngsb. Akad. Wiss, Wien. 1866. Bd. LUI. p. 7 (305), PI. IT., figs. .3-4. A. de Grossouvre : 1. c. p. 227, PI. .XXVI., fig. 4; PI. XXXIX. 2) Vide p. 29. 8) Vide p. 30. 4) Vide p. 31. 5) Anton Redtenbacher : Die Cephalopodenfaiina d. Gosaiischicliten in d. nordöst. Alpen. Abh. d. k. k. geol. Reichsanstalt. Bd. V. Heft 5. P. 119, PL XXVII., fig. 3. CEPHALOPODA FROM THE HOKKAIDO. 17 lateral lobe. Type: — G. limatum m}^ G. polilissimum Kossmat^^ from the upper Trieliinopoly group of South India is a species closely resembling this Japanese form. Lytoceras sp. of Schlüter'^^ from the Senonian of Germany may also be cited as an ally of the latter. IV. Subgroup of Gaudryceras derimanense Whiteaves. This embraces these Ammonites whose surface bears numerous simple, equal, sharp libs in the full grown state. It is closely related to the first subgroup, but is distinguished from it by the absence of periodic ribs. Type: — G. denmanense Whiteaves sp.,^^ once treated as L. JuTcesii? Sharpe^\ and considered by Kossmat'''^ as identical with G. Kayei Forbes sp. G. Jukesii is a species from the Senonian of England which was founded on a single fragment. Y. Subgroup of Gaudryceras Sacya Forbes sp. A full grown shell shows many, very shallow and narrow, transverse grooves separated by wide interspaces. The suture line resembles that of the subgroup of G. tenuiliratum. Type : — G. Sacya Forbes'' sp., the adult specimens of which were called by 1) Vide p. 34. 2) Kossmat: 1. c. p. 32 (128), PI. L (XV). figs. 7 a, b, c. 3) Schlüter: Cephalopoden d. obereu deiitscken Kreide. P. 16, PI. XLII. figs. 6, 7. 4) Whiteaves : On some Fossils from the Nanaimo group of the Vancouver Cretaceous. 1S95. P. 129, PL II. figs. 1-2. Note on a Supposed New Species of Lytoceras from the Cretaceous Rocks at Denman Island. Ottawa Naturalist. 1901. Vol. XV. No. 2, p. 31. 5) Whiteavks: Mesnzoic Fossils. Vol. I. Pt. 2, p. 119, PI. XIII. figs 3, 3 a, b. 6) Kossmat: I.e. p. 125. 7) Stoliczka: I.e. p. 154, PI. LXXV. figs. 5-7; PI. LXXVI. figs. 1-3.— White aves: 1. c. Vol. I. Pt. 1, p. 4;;, PI. II. figs. 2a,b,c, 3. Pt. 3, p. 203, PI. XXV.— Kossmat : 1. c. p. 23 (119). 18 ART. 2. H. Y ABE I CRETACEOUS Forbes Ammonites Budha. The single fragment of an Ammonite from the Hokkaido shows the character of the subgroup. VI. Subgroup of Gmidryceras varuna Forbes sp. This subgroup hokls a somewhat isolated position as compared with the above five, including G. varuna Forbes sp.,^^ G. odiense KossMAT sp.,"' G. anaspatnm Redtenbacher sp.'^^ and G. Kawanoi JiMBö sp.^' The whorls of these species are considerably compressed on the side, the involution being very deep and the umbilicus narrow from the beginning of the shell, while the surface is nearly smooth. To these differences is still to be added another concerning the course of the suture line. There are numerous auxiliary lobes slowly diminishing in size toward the umbilicus. Although it appears somewhat artificial to make such sub- divisions of Gaudryccras as above enumerated, yet in this way, not only geologically younger forms are easily separated from the older, but also the extreme confusion which has hitherto prevailed among the Ammonites of the whole group may also be avoided. 1) Kopsmat: I. c. p. 34 (130); p. 65 (Ifil), PI. II (XYI), figs. 4 a, h ; PI. Ill (XYII), fig. 8. 2) Kossmat: ]. c. p. 33 (129), PL IV (XVIIH, fig. 1 a, h, c ; PI. V (XIX), fig. 3. 3) Rebtenbacher : 1. o. p. 113, I'l. XXVI. fig. 1. 4) Vide p. 41. CEPHALOPODA FKOM THE HOKKAIDO. 19 SUBGROUP OF GAUDRYCERAS TENUILIRATUM. GAUDRYCERAS TENUILIRATUM m. PI. III., figs. 3 and 4. 1890. Lyioceras Sacya Yokoyama : Versteinerungen îuis der japanischen Kreide. P. 178, PI XVIII. figs. 12 a, b, 13 a, b. 1894. Lytoceras Sacya Jimbö : Beiträge zur Keiintniss der Fauna der Kreideformation von Hokkaido. P. 13 (159)— 18 (164). This species is the most common form in our Cretaceous, playing the rôle of the leading fossil of the upper Ammonite- beds. It has long been known under the name of * Lytoceras Sacga ' and the occurrence of the middle Cretaceous in Hokkaido has been based by several authorities on this determination. The writer took special care in studying this species, as it has been often much confounded with allied forms. Dimensions : — Diameter. Height of the last wliorl. Breadth of the hist whorl. AVidth of umbi- licus. (1) (2) (3) (^) (5) 30.0 cm. 100. •23.0 cm. 100. 4.2 cm. 100. 3.95 cm. 100. 3.50 cm. 16.0 „ 53. 13.0 „ 56. )) — 1.45 „ 36. 1.2),, 16.5 „ 53. 12.0 „ 52. 1.6 „ 38. 1.45 „ 36. 1.35 „ 5.5 „ 18. 4.8 „ 20. 1.8 „ 42. 1.60 „ 40. 1.50 „ 100. 34. 38. 42. 1) The largest specimen ever found; from a cliff opposite the mouth of the Panke-moyfibari, a branch of the Yûbarigawa. 2) A specimen from Urakawa. 3) A specimen from Urakawa (after Yokoyama). 20 ART. 2. — H. YABE : CRETACEOUS 4) A specimen from tlie Kikumezawa, a tributary of the Ikusliumbets. 5) A specimen from the Kikumezawa, a tributary of the Ikushumbets. The description of t]iis species by Yokoyama is very accurate, but it is based on young specimens and not on adult ones. The full-grown individual of G. tenuiliraturii must have attained a considerable size, for a specimen from a clifiP opposite the mouth of the Panke-moyCibari measures 30 cm. in diameter and is en- tirely septate. By removing the outer volutions, we can trace each stage of development back to the embr3'onal chamber. However, when specimens are very small, it is not always easy to determine whether they represent a young stage of the present species or of other allied forms. That the young and adult specimens belong to one and the same S]3ecies is shown by the abundant occurrence of many transitional forms. For the sake of convenience the writer describes this species in three different stages of develop- ment, although in fact there is no sharp boundary between any two of them. The young stage^^ : — Shell discoidal, composed of mau}^ slightly involute whorls. A specimen, 3. 5 cm. in diameter, is composed of six volutions. Whorls transversely reniform, broader than liioh and broadest near the umbilical edo-e. Umbilicus very wide, with a moderately steep wall. Surface with very numerous fine striœ and few periodic ribs ; stride unequal in length, more or less S-shaped and acute, much narrower than their interspaces. The longest striae embrace the entire surface of the whorl, while the shortest ones are only on its ventral side ; others are on the ventral as well as on the J) Yokoyama: 1. v. PL XV^III. lig. 1;]. CEPHALOPODA FROM THE HOKKAIDO. 21 lateral sides. All of these striœ are very fine, but their size varies much in each individual. The periodic ribs are round and parallel to the stride, five of them being on each volution, and correspond to faint grooves on the cast of the interior. These ribs are also covered with fine striœ. The suture line has one external, two lateral and three aux- iliary saddles, diminishing gradually in size toward the umbilical suture. The above description and that of Yokoyama clearly indicate that this Ammonite is closely related to G. muUiplexiim Koss.^^ { = G. Sacya Forbes var. multiplexum Stoliczka), with which Yokoyama indeed identified it, while Kossmat considered it to be G. Sacya. The former species is distinguished from this by the difiereut forms of the auxiliary saddles. The latter, when compared with this, possesses less numerous volutions, although it is often impossible to find this distinction, when we have not full grown forms before us. G. dcnmanense AVhiteaves'-^ sp. f]"om tlie Vancouver Cretaceous which is identified with G. Kayei Forbes by Kossmat, is also hardly distinguishable from the young forms of the present species. Whiteaves, after a direct comparison of the sj)ecimens of his G. denmaneiise with those of G. tenuiliraium which had been sent to him for examination, informed the writer that he saw at pi-esent, no reason why the latter should be distinguished from the former. G. alamedense J. P. Smith''^ is also certainly related to G. tenuiUratani, but according to the above author to whom 1) Kossmat: 1. c. p. 25 (121), PI. I (XV), figs. G a, b, c. 2) Whiteaves: On Some Fossils from the Nanaimo group of the Vancouver Cretaceous 1895. P. 129, PI. n. figs. 1-2. 3) Smith : Tiie Development of Lyloccra^ and riijlloccras. Proc. Calif. Acad. Sei. Series. lU. Vol. 1. Ko. 4. 1S98. P. 130. 22 ART. 2. H. YABE : CRETACEOUS some specimens of the Japanese species were also sent for com- parison, the Californian form is said never to attain snch a lai'ge size, and to be less involute. This is the more remarkable, as the specimens sent to him were all of a young stage. Now, as shown by the descriptions of the varieties of this species, the invo- lution of whorls is also not always constant, therefore it seems to the writer that the distinction based on this character is not of much value. G. voragurense Kossmat^^ from the Trichinopoly group of South India is also one of the nearest allies, but it possesses a more compressed shell and higher whorls. The middle stage'': — Whorls rather rapidly growing in height and breadth ; the height and breadth being at first nearly equal, while later the height becomes greater than the breadth. Umbili- cal wall very steeply inclined. Surface with striœ and ribs, the shortest strite disappearing imperceptibly as the shell grows, while the remaining ones, on the contrary, become thicker and more elevated and separated by wider intervals than on the whorls of the yonng stage. Ribs very numerous, with seven or eight striie between, somewhat elevated, round, composed of two or three strite which are very close to- gether and run into one on the lateral side. Involution also somewhat deeper, aljout one half of the preceding volution being- covered. Suture line as in the preceding stage with some slight compli- cations. G. mite Hauer''^ and G. varagurense Kossmat in form closely resemble this stage. The first species which is from the 1) Kossmat: 1. c. p. 26 (122), PI. IV (XYJII), figs. 2a, b, c; PI. Ill (XVII), fig. 9. 2) Yokoyama: 1. c. PI. XVIII. figs. 12 a, b. .".) Haueu: 1. c. p. 7 (o05), PL II. figs. 3-4. A. de Geos^ouvee: 1. c. p. 227, PI. XXVI. fig. 4 ; PI. XXXIX. CEPHALOPODA FROM THE HOKKAIDO. 23 Gossan bed of the Alps, on comparison, has less involute whorls (about i) and the ribs are less conspicuous. G. varagurense which is allied to (r. ienuiUraium in the young stage of its shell, is a species based on fragments from the upper Trichinopoly group of Varagur, South India. It shows only one periodic rib in a space corresponding to the last quarter of the last volution. The mature stage : — Whorls enlarging very rapidly ; rounded and at the same time broadest at the umbilical edge, the breadth from this point gradually diminishing toward the ventral side. Involution more than i. Umbilicus narrow, very deep, surrounded by a perpendicular wall. Surfoce with very narrow, sharp, prominent, simple, trans- verse ribs whose greater part entirely surrounds the whorls. Between the ribs, sometimes, shorter ones of the same character are inserted. Interspaces between the ribs flat, measuring 2 mm. on the ventral side. Periodic ribs composed of a partial union of two ordinary ribs, with a regular interspace of about 2-3 cm. along the ventral side, 22 of which are on the last volution of the largest well preserved specimen. The suture line is of the same type as in the preceding stage, only it is very much complicated, the accompanying figure having been taken from a large fragment found on the Bannosawa which is, no doubt, considerably larger than that which the writer has already referred to at the beginning. The full specific character is shown after the Ammonite has attained its adult stage. The full grown shell is quite different from others in form, while that of the young or middle stage is very liable to be confounded with allied species. This may be partly due to the fact that the adult shells of the allied forms are mostly imperfectly known. 24 ART. 2. H. YABE : CRETACEOUS G. denmanense Whiteaves^^ sliows phases of growth similar to those of G. tenuiliratum. Its adult stage is provided with a single kind of ribs which are somewhat thicker and separated by wider intervals than in Japanese species. G. glanegense Redtenbacher^^ from the Gossan formation of the Alps also more or less closely resembles the Japanese species, showing surface marking intermediate between it and G. dense- plicatum Jimbö sp. Localities : — Abundant wherever the Pachydiscus-and the Scaphites-heds occur, but especially frequent in the Fachy discus beds. Found in Sorachigori and Yübarigöri in the Province of Ishikari ; Yûfutsugun in the Province of Iburi ; along the Opiraushibets, and the Abeshinai, a tributary of the Teshiogawa, in the Province Teshio ; Urakawa, in the Province of Hidaka ; Sôyagori, in the Province of Kitami. Outside of the Hokkaido, this species has been also found in the Cretaceous of Amakusa.^^ I. Variety with a Wider Umbilicus. Var. ORNATA m. PI. IIP, figs. 2 a, b. Compare : — 1865. Ammonites Kayei Stoliczka: 1. c. p. 15, PI. LXXVII. fig. 1. 1871. Ammonites Kayei Griesbach : Geology of Natal. Quart. Journ. Geol. Soc. London. Vol. XXVII. p. 63. 1) Whiteaves: On Some Fossils from the Nanaimo Group of the Vancouver Creta- ceous, p. 129, PI. II. figs. 1-2. 2) Redtenbacher : 1. c. p. 119, PI. XXVIL figs. 3 a, b. 3) H. Yabe: Note on Three Upper Cretaceous Ammonites from Japan. P. 10, <% CEPHALOPODA FEOM THE HOKKAIDO. 25 1895. Lytocei^as Kayei Steinmann : Die Cephalopoden d. Quiriquina Schichten. Neues Jahrb. f. Mineral., Geol. und Palaeont., Beilageband X. p. 86, PI. Y. fig. 5 a, b. 1895. Lytoceras [Gaudryceras) Kayei Kossmat : 1. c. p. 162, PI. II (XVI), fig. 5 a, b ; PI. III (XVII), fig. 2 a, b. Dimensions : — (1)r..o' (5) Ratio. 0^) Diameter. 5.S5 cm. Height of the last I 1.85 „ whorl. I Breadth of the last 1.80 „ whorl. Width of umbili- cus. 100 h2 4.00 cm. 100. 2.25 cm. 1.20 „ 31 1.35 47 1.80 „ 30. 0.G5 „ 34. ! 0.75 Ratio. 100. 28. 45. 1.20 53. (4) 2.10 cm. O.GO „ 0.70 „ 1.10 „ Ratio. 100. 28. (5) 2.?,0 0.70 1.20 „ 100. 1.2f>cm 0.30 „ 0.50 „ 0.70 „ (^^Ratio. 100. 24. 40. 56. 1). A specimen of G. Kayei from the Valudayur beds of Pondicherry (after Kossmat). 2). A specimen of var. ornata from the Pachydiscus-hediB of the Kikumezawa (PI. III. fig. 2a,l)). 3). A specimen of G. Kayei from South India (after Forbes). 4). A specimen of G. Kayei from the Quiriquina bed of Chili (after Steinmann). 5). A specimen of var. ornata from the upper Ammonite beds of the Ikushumbets. 6). A specimen of G. Kayei from the Valudayur beds of Pondicherry (after Kossmat). The specimen, Xo. 5, consists of seven volutions. The umbilicus is very wide, being broader than one half of the whole diameter of the shell, slightly covering the ventral portion of the preceding volutions. Aperture subquadrate, broader than high ; in older stages, becoming higher and laterally compressed. Surface 26 ART. 2. — H. YABE I CRETACEOUS with exceedingly fine striœ and five periodic elevations. The suture is similar to that of the normal form, the head of the first auxiliary saddle touching the umbilical suture which is followed by two or three small denticulations. This variety is distinguished by a very slow growth of whorls and an umbilicus wider than the normal form. The writer thought at first that this form was a distinct species, and probably identical with G. Kayei Forbes sp. from the Senonian deposits of S. India, ^^ Natal,-^ Chili'"'^ and Tunis,'*^ whose European ally is G. planorhiforme Böhm sp.;^^ but the presence of many transitional forms between this and the normal specimen of G. tenuiliratum convinced him that it is impossible to treat them as different species. The above mentioned characters being common to this variety and G. Kayei, it is doubtful whether the greater part of the forms described as G. Kayei are not a mere variety of G, tenuiliratum. It is quite impossible to decide whether an Ammonite described under the name of G. 'planorhiforme resembles Senonian or Ceno- manian forms when we have only young specimens. The writer is also in doubt whether the large individuals figured in PI. XXXIV. fig. 4, 5. and PI. XXXV. fig. 7, by A. de Grossouvre really belong to the adult stage of G. planorhiforme. An Ammonite from the Vancouver Cretaceous first described 1) Stoliczka: 1. c. Kossmat: 1. c. 2) Griesbach : 1. c. 3) Steinmann: 1. c. 4) L. Pekvinquière : 8ur nn faciès paiticularier de Sénoinen de Tunisie. 1898. (by référât). 5) J. Böhm: Palœontograpliica. Bd. XXXVIII. p. 49, PI. I. fig. 12— V. Uhliq : Bemerkungen z. Gliederung karpath. Bildungen. Jahrb. d. k. k. geol. Reichsanstalt. Wien, 1894. P. 216, fig. I.-Geossodvre : 1. c. p. 231, PI. XXVII. fig. 2. CEPHALOPODA PROM THE HOKKAIDO. 27 as L. Juhesi Sharpe ? and afterward under the name of L. denmanense sp. nov. and considered to be identical with G. Kayei by KossMAT, is believed by the writer to be a distinct species, he in this respect agreeing with the latest view of Whiteaves. Localities: — The Kikumezawa, a tributary of the Ikushumbets, and the upper course of the latter ; one from each locality. Horizon : — Pachydiscus-heds. II. Varieties with a Narrower Umbilicus and Coarser Ribs. Var. intermedia, m. PL III., figs. 1 a, b. 1895. Lytoceras Sacya Jimbo : 1. c. p. 34 (180), PI. VI. fig. 1. 1895. Lytoceras denseplicatum var. Jimbo : 1. c. p. oG (182). Dimentions : — (1) K.,i.,. Diameter. ■ 3.1 cm. Height of the last whorl. : 1.3 „ l^reacUh of the last whurl. ■ 1.3 „ Width of umliilicus. , 1.05 ,, 100. 41. 41. ("/ Katiii. 3.40 cm. 1.45 „ 1.50 „ 1.10 „ 100. 42. 44. 32. (•^) llati 4.1 cm. 1.8 „ 1.5 „ 1.3 „ 100. 43. 31). 30. (^) Ka, 11.7 cm. 5.7 „ 5.3 „ 3.0 „ 100. 48. 45. 25. 1) A specimen from the Yfibarigawa, Prov. Ishikari. 2) The original of Jimbô's PI. VI. fig. 1 ; loc. Yübarigawa. 3) A specimen from the Penke-moyübari. 4) A specimen from the Sanushibe, Prov. Iburi. This variety is distinguished from the normal form by the following characters : — When yoiuig, the whoi-ls grow more rapidly in height ; the umbilicus is narrower and somewhat deeper, a little more than 28 ART. 2. H. YABE : CRETACEOUS one third of the inner volution being visible ; the aperture is nearly round ; the strite on the surface are coarser. The suture line iô the same in both. This variety differs from G. madras'patanimib^ Blanford sp. from the lower Utatur group of South India in its coarser striie ; but in form they are remarkably alike. Lytoceras sp. (Yokoyama : 1. c. p. 80, PI. XIX. fig. 3 a, b.) may probably be referred to this variety, as it agrees quite well in form and sculpture ; periodic ribs are very indistinct in this and the next variety, being almost invisible in some specimens, although more distinct in exceptionally well preserved ones. A specimen from the Hokkaido in the Imperial Museum of Natural History in Berlin was considered by Jimbö, to be a variety of G. denseplicaium. But the examination of a plaster- cast kindly sent to the writer by O. Jjeckel, the director, shows that it really belongs to lliis variety of G. tenuiliratum. Localities : — The YCibarigawa, above the mouth of the Panke-moyübari ; the Ikushumbets ; the Sanushibe, a tributary of the Popets. Five specimens were examined. Horizon : — Upper Ammonite-beds. Exclusively in the Fachydiscus-hed^ ? Var. INFREQÜENS m. PI. IV., fig. 3 a, b. 1894. Lytoceras Saeya Jimbö (par.) : 1. c. p. 34 (180). Dimensions : — Diameter. 4.30 cm. 100.' Height of the last whorl. 1.7Ö „ 40. 1) SïOLiczKA: I.e. p. 151, PI. LXXV. fig. 2.— Kossmat : I.e. p. 32 (128). CEPHALOPODA FllOM THE HOKKAIDO. 29 Breadth of the last whorh 1.90 cm. 44. Width of umbiUcus. 1.40 „ 32. Shell broader than in other varieties ; whorls broaden slowly toward the mouth, broader than high. Umbilicus wide, surrounded by a vertical wall. Surface with striœ of the same type as in the preceding variety but considerably coarser and more prominent. Localities : — On the Opiraushibets, about 35 miles from its mouth, Prov. Teshio ; Makka-ushipe, a branch of the Sanushibe, Prov. Iburi ; the Kikumezawa, a tributary of the Ikushumbets Prov. Ishikari. Only one specimen from each locality was ex- amined. Horizon : — Upper Ammonite -beds. SUBGROUP OF GAUDRYCERAS CRASSICOSTATUM. GAUDRYCERAS CRASSICOSTATUM Jimbö sp. PI. IV., fig. 4. 1894. Lytoceras crassicostatum Jimbö : 1. c. p. 36, PL VI. figs. 7, 7 a. This species is exceedingly rare. It was first founded on a somewhat water- worn but yet full grown specimen from Cape Soya in the Province of Kitami. The writer has drawn up the following diagnosis from an examination of a specimen which belongs to a young stage. 30 ART. 2. — H. YABE : CR] ETACEOUS Dimensions : — Diameter. 4.5 cm. Ratio. 100. Height of the last whorL 3. '5 „ 33. Breadth of the last whorl. 1.2 „ 26. Width of umbilicus. 2.1 „ 46. Shell discoidal, compressed ; composed of about 7 volutions. Umbilicus very wide, about twice as broad as the height of the whorl, becoming gradually narrower as the shell grows. The inner volutions slightly inflated, merely touching one another, but the last volution is higher than broad, embracing one half of the preceding one. The surface sculjDture is similar to that of G. tenuillratuvis but the striœ become 2;raduallv finer toward the mouth where the shell appears almost smooth. The whorls of the later stages possess many thick periodic ribs as shown in fig. 7, PI. VI. in the work quoted above. This species, being characterized by the peculiar sculpture of the adult specimen, and the very loosely coiled volutions of the young individual, is easily distinguished from others. Locality : Cape Soya, Prov. Kitami. Horizon : Upper Ammonite-beds. GAUDRYCERAS DENSEPLICATUM Jimbö sp. 1894. Lyiocerru denseplicatum JiiMBö : 1. c. p. 36 (182), PI. III. fig. 1. This beautiful but rare form is represented by a single specimen in the writer's collection, this and the type figured by Jimbö being the only ones found up to the present time. The writer's specimen is less complete than Jimbô's, having only a part of CEPHALOPODA FROM THE HOKKAIDO. 31 the last volution which is marked by the characteristic sculpture of the adult stage. When judged from the sculpture, the Japan- ese form comes near to Gaudryceras glanegense Redtenbacher'^ from the Gossan formation of the Alps which, with a somewhat similar form and sculpture, shows a position intermediate between this species and G. tcnuiliratum,. It is not yet possible to ascertain the exact relation between the present form and G. striatum Jimbö sp. described below, because the imperfect state of preservation of the former does not allow the examination of its inner volutions which alone would settle its relation to the latter. Localities : — The type specimen of Jimbö is from the Ekimo- moanoro, a tributary of the Yûbarigawa, while the other is from a cliflP of the Yûbarigawa, close to its junction with the Penkemo- yübari. Horizon : — Pachydiscus-hQà^. GAUDRYCERAS STRIATUM Jimbö sp. PL lY., fig. 5. 1894. Lytoceras striatum Jimbö : 1. c. p. (181), PI. YI. fig. 6 a, b. The type specimen, which has been described and figured in the above paper, is a young individual of 4. 7 cm. in diameter from the Abeshinai-rubeshibe, a tributary of the Teshiogaw'a. Another less imperfect but somewhat larger specimen from the same place has been lately found in the collection of Jjmbö which necessitates the revision of the diagnosis given by him. 1) Redtenbacher : I.e. p. 119, PI. XXVII. fig. 3. 32 ART. 2. H. Y A BE : CRETACEOUS Dimensions : — ■Ratio . Diameter. 7.2 cm. 100. Height of the last whorl. 3.0 „ 41. Breadth of the last whorl. 2.8 „ 38. Height of the W'liorl last but one. 1.2 ,, 16. Breadth of the whorl last but one. 1.4 ,, 19. Width of umbilicus. 2.4 „ 33. Inner volutions broader than high ; wdien 4.0 cm. in diameter, they are as high as broad, after which they rapidly in- crease in height, so that a cross-section near the mouth of the larger individual shows a subquadrate form, higher than broad, being broadest near the umbilical edge, but narrowing toward the rounded ventral side. A change also takes place in the surface sculpture, just as in G. ienuiliratum and G. crassicostatum. The surface is at first covered with numerous very fine striœ and about five periodic ribs, corresponding to the feeble furrows on the cast. The striœ and the ribs are slightly S-shaped as is usual in shells of this genus. These periodic ribs appear more frequently when the shell has attained a diameter of about 5 cm. The figured specimen shows 7 of them on the last quarter of the last whorl. The suture line of our specimen when compared wâth that shown in fig. 6 b, PI. VI. of Jimö's w^ork, is î-dv more finely toothed. The relation of this species to G. denseplicatimi Jimbô sp. has been already alluded to. It does not appear, at first sight, to be different from the subgroup of G. temdliratum, and it is commonly not easy to distinguish the former from the finely sculptured variety of the latter. The main points in which they differ lie in the fact that the suture line in G. sfriatum is more finely divided than in G, tenuilirahmi, and the thicker ribs and CEPHALOPODA FPO>r THF. TTOKKATPO. Ô6 tlu'ir interspace!^ on tlie former are nnitbi'inely covered witli very fine striiTC which do not heconie gradually coarser and less in iHnnl)er in the older volutions, as we have seen in G. tenuiliratitm. On account of tlie aliove characters I have provisionally referred G. striatum to the sul)giou]) of G. rrassicostatum, altliough its aflfinity to this subgroup is not yory close, as is also the case witli G. (Jeni^eplirafimi. Locality: — Tlie Al)eshiii;n-rn]»('shi])e, a trihutary of tlie Teshio- gawa, Prov. Tesliio. AVe possess only two s]^eciniens — one of which is the original of Juinö's illustration and the other is represented in figure 5 in PI. IV. Horizon : — Pachydiscus-hed^. Var. PICTA m. PI. IV., fig. 0 a, b, c. The third specimen of G. striatiuii is a somewhat modified form ; the following table exhibits the dimensions measured on the external as well as on the internal volutions of the same specimen. Dimensions : — Diameter. Height 8 ]'.reailth of the last wliorl. •"•- „ .SO. » — AVidtli of unihilicus. 2.7 „ 32. 2.G „ .s;i CEPHALOPODA FROM THE HOKKAIDO. oO 1) A specimen from tlie Ikiisliumbets. (PI. IV., figs. 3 a, Ij). 2) A specimen from the Yübarigawa. (PL IV., fig. 2). Shell discoidal, somewhat compressed, composed of numerous ^vho^]s ; whorls oblong, becoming gradually higher as the shell grows, the broadest part lying near the umbilical edge. Umljili- cus wide, shallow, exposing nearly three-fifths of the inner volu- tions ; Umbilical wall perpendicular, suri'ounded by a rounded edge. The suture of the shell except on the last volution is smooth, being provided only with very fine stria? of growth. The striic begin at the umbilical suture, bend forward at its edge and extend into the lateral side of the whorl with a slight flexure, passing straight over the ventral side. When the shell becomes larger than about 8-9 cm. in diameter, the thick transverse ril)s gradually begin to appear, encircling the whorl parallel to the striœ and rapidly becoming broader and more elevated toward the periphery. These ribs are broad, measuring 0.6 cm. in average breadth with interspaces of about 0.4-0.8 cm. in breadth on the siphonal line. However they become broader and more closelv set together toward the mouth. Suture line very complicated ; the external saddles as well as two lateral ones and the first auxiliary are bipartite, while the remaining three auxiliaries are simple. The lobes are narrow, the two lateral ones being bipartite. The auxiliary saddles and lobes hang obliquely on the umbilical wall. The septal lobe has been seen in a well preserved specimen. There is only one ally of the present form, and that is Gaudryceras politisslmum Kossmat sj-».^^ from the upper Trichino- poly group of Varagur, South India. This Indian species resem- bles in many respects the Japanese. It is known (jnly by a specimen at the middle stage of growth, which is distinguished 1) KossiMAt: I. c. p. 32 [Via), PI. I. (XVj, tigs. 7 ;i, b, c. 36 AKT. 2. H. YABE : CKETACEOUS by a wider umbilicus and relatively low, mucli compressed wliurls. Still tliere is no doubt that there exists an intimate relation between these two forms. The Japanese species is very variable in its form ; some (PI. v., fig. 2) show a wider umbilicus and lower whorls, strongly reminding us of the form of G. iJolUissiinum, while others are characterized by a narrower umbilicus, higher whorls and the earlier appearance of broad ribs. The latter is now con- sidered merely as a yariety (var. obscur a). Fig. 3 a, b. in PI. YI. is a form intermediate between this and the normal type. Localities and Horizon : — The uj)per course of the Ikushum- bets, above the mouth of the right branch, the Yoshiashizawa, and along the latter itself; found in the u])per Ammonite-ljeds. Along the Opiraushibets, Prov. Teshio, in the /Scaphites-hiida, GAUDRYCERAS YOKOYAMAI m. PI. VP, figs, la, b, 2 a, b; PI. YIP, fig. G'. There are only a few specimens belonging to this species, one of which is a cast of an interior of a moderate size, showing the suture line ; the others are smaller, but mostly provided Avith the shell. P)imension : — iJianieter. 1.5 cm. 100. Height ol" the last whorl. 2.0 „ 44. 4.2 „ 4S. Breadth of the last whurl. 1.8 ,, 40. 40 „ 4C, Width oC mubilioiis. 1.35 „ 30. 2.1 „ 24. 1) A young individual from the Yilbarigawa. 2} The largest specimen from tlie Ikushumbets (') Kan,,. S.G cm. 100. CEPHALOPOUA FE03I THE HOKKAIUo. 37 Shell discuklal, thick; whorls rüuiicl in the early stage; subquaclrate in the adult, higher than broad, and Ijroadest near the umbilical edge. Ventral side uniforniely rounded. Umbilicus moderate in size, showing about three fifths of the circumference of each volution and surrounded by a perpendicular wall. Surface ornamented with very fine striœ, and also very feeble and narrow farrows. The striœ are so fine that we can hardly detect them without the aid of a magnifier ; they rise at the umbilical suture, bend abruptly forward at the umbilical edge and then traverse the sides without any further flexion. The faint furrows on the body cluiml)er of the largest specimen are narrow, parallel to the stria*, and arranged at intervals of al)out half a centimeter from one anotlier. The sutui'e line is very complicated, but shows on the whole the character of the group of G. Sacya Forbes. There are six or seven saddles on each side ; the external, the two lateral and the first auxiliary saddles are symmetrically bifid ; the lii'st lateral lol)e is the deepest, slightly surpassing the external lobe in depth and also symmetrically Ijipartite. Auxiliary saddles and lobes hang obliquely toward the uml)ilical suture. The fine strite of growth and numerous feeble furrows of the present form make it seem at first sight to be allied to Pseudo- l^hylliles indra Stoliczka' sp., to which its general aspect also closely conforms. The most characteristic feature of Pseudo- lihylliics, however, lies in the suture line (jf the young shell, which shows tripartite saddles with plylloid termination, while in individuals of an advanced stage both branches of the external saddle become unequal in size. The Indian form is, therefore, not only difierent from the Japanese, but also from all other 1) Stoliczka: 1. c p. 112, PL LVIII, fig. 2.— Kossmat: 1. c. p. 41 (J37), PI. II. (XVJ.), figs. üa,b, 9a,b; PI. III. (XVII.), figs. 6, 7a,b; PI. IV. (XVIII.), fig. 3. 38 ART. 2. H. YAlîE : CRETACEOUS species of Gaiidryceras. On this account, it was placed l)y Naumann and Whiteaves in tlie genus Phylloceras,^^ while KossMAT treated it as Pseudopli y lilies^'' a new subgenus of Lyio- ceras ; but the writer is at present not in a position to advance any opinion on the matter. In common with the preceding S2^ecies and (/. Rouvillei Grossouvre,"^ Gaudryceras Yokoyamai has fine striœ traversing the ventral side, bending neither forward nor backward, but running always straight. The small specimens of G. Yokoyamai and G. limatwii are liable to be confounded with one another, but those of the former possess higher whorls and a narrower umbilicus than those of the latter. G. Rouvillei from the Santon ien of Southern France is too small for a closer comparison, but it seems to be distinguished from the young of the Japanese species by its deeper involution. Localities and Horizon: — In pebbles from the Yubarigaw^a and the Ikushumbets. Found also in the Fachydiscuü-hedn along the Kikumezawa, a left tributary of the Ikushumbets. GAUDRYCERAS YAMASHITAI m. PI. IV., fig. 7. Only two immature specimens of this species Avere found, the laro'er of which measures : — Diameter. 8.8 cm. 100. Height of the last whorl. 4.4 „ 50. 1) Whiteaves: On some Foss. Nanaimo group Vancouver Cret. p. 129. 2) Kossmat: 1. c. p. 41 (137). ;i) Guus^ulvuk: 1. e. !>. 228, PI. XXXVII. figs. 7 and 10, CEPHALOPODA FPvOM THE HOKKAIDO, 39 ]*>!•< 'adtli of tlio last wlioi-1 3..'] em. 37. Width of uiubilicLis 2.2 „ 2ö. Shell cliscoidal, compressed; when yoinig, whorls nearly rounart of the Ix^ly chamlier of an adult specimen, no- thing can be said about its general shape and suture line. But the surface with narrow flexnous furrows is quite similar to that of G. Sacya figured by Sïoliczka^- and Whiteaves."^ The space l)etween two successive furrows measures above 1.2 cm., while the fui'i-ows themselves measure al)Out 0.3 cm. in breadth on an avei'age on the ventral side ; ]»esides there are nimierous fine stria' covering tlie entire surface of the shell. Tlie specimen is too imperfect for exact determination, l)ut its resemblance to G. Sacya is so great that there is hardly any doubt of its representing a form closely related to, if not actually identical with that species. Locality ; — A specimen, found in the collection of Jimeo, is in a marly nodule from Ishuikarushihara, on the Penkeohoshikep, a l)rauch of the Teshiogawa. Horizon : — leiden own. 1) Sioliczka: I.e. p. 154, PI. LXXV., fio-. 7. 2) WiliTEAVEs: Mesnzoic Fossils. Vol. T. I'l. ;'.. P. 20."., P]. X.W. CEriTALOPODA FROAf THE ITOKICATDO. 41 SUBGROUP OF GAUDRYCERAS VARUNA. GAUDRYCERAS KAWANOI JimeH s^p. 1894. Democerai^ Kawanoi JiMr.r.: 1. e. ]i. 28, PI. T. (XVII.), figs. 7 n, 1). This species is re]-) resented l)y two specimens, one of wliicli helongs to tlie type described by JimbH. Tlic other slightly deviates from this type, Init is considered by the writer as l)elonging to the same species. Dimensions : — (1) Rutin. (2) Rnlin Diameter. 2..") cm. 100. 2.2 em. 100. Hoiglit of llio M>.t \vl»orl. 1.0 „ 4). 0.9 „ 40. I'.roadth of tlic last wliorl. o.s „ 32. 0.75 „ 34. "Width of ninlii icns. 0.8 „ 32. 0.7.5,, 34. 1) The type specimen of Dewwcrra.^ Kaivanoi Jimbö from Tkandai near Urakawa, Prov. Hidaka. 2) The second specimen from the Bannosawa, a tributary of the Ikushnmbets near the Poronai coal-mines, Prov. Ishikari. The diagnosis of tliis species given by Jiinibo is as follows: — "Schale flach-scheil)enförmig. Umgänge sclinell an llülie zunehmend, die grösste Dicke des I'mgangcs in der Mitte. J )ie Zahl der Umgänge ist ca. 7. Nabel weit. Die Oberfläche ist von äiis.-^erst feinen Streifen bedeckt und zeigt 3 schwache Einschniii'- ungen auf der letzten AVinduns;. Die Streifen und Einschnürunoen bilden auf der Aussenseite leichte, vorwärts gerichtete Buchten." In this diagnosis, the last part needs some alteration. The periodic furrows and numerous fine striic extend obliquely forward from the umbilical suture to the ventral side, without the slightest 42 ART. 2. — IT. YABE : CRETACEOUS trace of a forward pi-oloiigalioii. Tlio eross-^oction of tlio whorl is oblong, being broadest in the middle of the lateral side. The umbilicus is of a moderate size, about 5 of the inner volution being exposed ; the umbilical wall gently slopes toward the um- bilical suture and to the lateral side. The suture line shows many lobes on each side, which are finely and deeply incised aiid diminish in size toward the umbilicus. The external lobe is as deep as the first lateral. The external and the lateral saddles as well as the first lateral lobe are bipartite. Judging from the general aspect of the shell, and especially from the suture line, the present species seems to belong to a particular subgroup of Gavdryceras, which contains only thiee species from foreign Cretaceous, viz., (t. odicm^c Koss. of the lower T^tatur group of Houth Indi;!. (1. vdruna Forbes of the VaUulayur bed of the same region and G. anaxpatum Redtenbacher sp. of the Gossan bed of north-east Alps. Our form is distinguished from G. odiense hj its broader umbilicus and its lower whorls, from G. varuna and G . anaf^patum by the different form of the whorls. G. varuna is a species also found in the Quiriquina bed of Chili."*^ Localities : — Chashikots in Ikandai near Urakawa, Prov. Hidaka, and the Bannosawa near the Poronai coal-mines, Prov. Ishikari. Horizon : — Pachyd isru.^-hoih. 1) Kossmat: 1. c. p. P.-'î (129), PI. IV. (XVIir.i, fi-. 1:., b,c; V\. Ilf. (XVTL), fig. 8. 2) Kossm.vt: 1. c. p. U (130); p. Go (101), PL II. (XVD, fio-. 4 a, h ; PL III (XVII), fis- S. 3) Redtenbacher: I.e. p. ll:î, PL XXVI., f;"'. 1. 4) Stkixmaxn: \.c. p. 'IS, PI. V., fi,us. 2 a, I.. CEPHALOrODA FR03I THE HOKKAIDO. TETRAGONITES Koosmaï. Of the species of Tetragoniles described in the following liney, three luive been already described liy Jim bo. One of these — Lytoceras glahrum JimbH — is considered l)y Kossmat" to belong to his newly established subgenus Tetrcujonltes. He states that nothing definite can Ije said about the other two — viz., L. crassum. JiMBô and L. sphaerojwtum Jimbö. However, after a careful examination of the types and other specimens the writer has come to the conclusion, that these two should also be brought under the same genus. TETRAGONITES GLABRUS Jijibô sp. PL VIT., figs. 2 and o. 1894. Lytoceras glabnim Jimbö : 1. c. p. 34 (180), PI. VI. (XXIL), figs. 2,2 a. 1895. Lytoceras {Tetr agonîtes) glahrum Kossmat : I.e. p. 133. Dimensions : — Dian: Ileig BretK Widt 1) A sj^ecimen from the Sanushibe, Prov. Iburi.(Pl. V., fig. 2). 2) A specimen from the Opiraushibets, Prov. Teshio. This and the next species possess very inflated whorls in the later stages of growth, and comparatively a very thin test. There- Diameter. Ileiglit of the last whorl. Breadth of the last whorl. Widtli of umbilicus (1) Initio. (2) 3.5 era. 100. G.5 cm. 1.7 „ 48. 3.2 „ 1.9 „ 54. 3.6 „ 0.8 „ L4„ 100. 49. 21. 1) Kossmat: I.e. \>. 3/ (133). 41 AKT. 2. — H. Y A BE I CRETACEOUS fore, Llic outer volution in adult specimens is generally not well ])reserved, wliieli makes exaet measurement very diilieult. These two species are closely allied to each other, and if we compare their dimensions, it is quite evident that there is no niaterial difference between them. The coincidence is not only in dimensions, but also in nearly all other characters if we except the periodic ribs. There is a group of Ammonites in which the shell is smooth until it has attained about 7 cm. in diameter, after which a periodic arrest of growth begins to appear, while another gr(»u]) shows periodic ribs from the very beginning; still it not rarely happens that we can not determine which of these two characters is shown by a specimen not in a good state of preservation. However, in spite of the imperfect nature of the tvpe specimens, we can easily see that Jijjbö's L. .yj/iaeronolufiL is of the first category, while his L. glahruiii is of the second. It is to be added that the writer is, at i)resent, not quite sure whether the absence of periodic I'ibs on the inner volutions is suflicient for specific distinction. Tctragonitc^ cpigoniis Kossmat,^^ a closely related species, is stated to be rather variable in the appearance of the ribs, some young individuals showing them typically, while the larger ones are smooth. Although this statement leads us to assume that the ribs are not of nuich im- portance in the distinction of species, yet the ])resent case must be considered as somewhat different from the above one. As above stated, in T. sphacrouolus the periodic ribs appear after the shell has attained a certain size, while in T. glabrus they are already present when the shell is young. Forms intermediate between these two have not yet been found. Therefore the writer 1) Kos^i.Vi: 1. 1-. 11. 3'J (i;;5), ri. HI. (XVll), llg. '1,5. CEPHALOrODA FJIOM TJIK HOKKAIIX». 45 is for tilt; timo being conipellcd to treat the two as specilically di.stuict, UH has akeady been done l)y Jimbö. It is necessary to make a fe\v remarks on the inteiiial suture luie of T. glahriis whieh was not figured by Jimbö. It sliows two saddles and the corresponding kjbes. Tlie antisi[)honal and the lateral lobes are narrow and deep ; the two saddles are also slender and of equal height, the inner ending bifid and the outer simple. The above characters of the internal IoIjcs and saddles, toge- ther with those of the external part of the suture line, are exactly what we lind in the type of T'elragoniics and especially in T. epi(jonn8 which also shows two internal saddles. Localities : — The upper course of the Ikushumbets, and the Yrd)arigawa, Ixjth in the Province of Ishikari; the 8aushi-sanushibe, a tributary of the Popets in the Prov. Iburi ; the Opiraushibets and the Ijranches of its u})per c(jurse, iu the Prov. Teshio. JNIany specimens fiom all the localities meutioned. Horizon : — Upper xVmmonite-ljcds. TETRAGONITES SPHAERONOTUS Jjmüö sp. PI. YIL, iigs. I a, 1). 1894. Lytoccras sphaeronotuui J im no : 1. c. p. oo (181), PI. VI. (XXII.), ligs. 3, 3a. Dimensions : — Dianit'ter. lloiglil of tlie lawl wliorl. Breadtli of tlie last wliorl. ^VillLll of iiiuliilieii--. (1) [;:ili(i. (-') , 14.5 cm. 100. G.Ociii. 7.0 „ 48. '"•"* )> 7.2 „ 49. 3.8 „ 3.0 „ 'JO. 1.4 „ loo. 47. 54. 'JO. 4.0 MIT. J. II. YABE : CllETACEOUS 1) A specimen from the Opiniusliil)ets, Prov. Tesliio. 2) \ spécimen from the Ynbarigawa, Prov. Ishikari. ►Shell thick, discoidal, con) posed of many whorls ; whorls some- what trapezoidal being laterally flattened, ventrally inflated and slightly broader than high. Umbilicus narrow, deep, with a perpendicular wall and ])luntly angular edge. Involution about i. Surface with verv fine but distinct stride of irrowth, without any trace of periodic ribs. The older specimens show whorls more rapidly increasing in height and Ijreadth, and ])rovided with peri- odic ribs. These ribs are slightly elevated and rounded, correspond- ing to shallow furrows on the cast of the interior. They are wddely separated, there being only two or three in one volution. The fine strite as well as the periodic ribs which are also covered with striie, rise at the umbilical suture, l)end obliquely forward on the lateral sides and then backward on the ventral side. Besides there are a few longitudinal lines on the external half of the lateral sides which are visil)le only on the last half of a volution of the largest specimen in the writer's collection. The periodic ribs begin to ap})ear when the shell attains 7 cm. in diameter. The length of the body chandjer of Teiragoniics is not yet known. The largest specimen, mentioned above, shows two thirds of the last volution not septate, but as the anterior end of the whorl is broken off, the length of the body chandjer nuist l^e more then 5 of a volution. Suture line as in the preceding species. There are many well })reserved individuals of this species in the writer's collection, one of which is shown in fig. 1, PI. VII. The al;)ove diagnosis was written after a study of them and the two specimens in Jindjo's description essentially agree with them. CEPHALOPODA FPtOM THE HOKKATDo. 47 Besides tlu'so sjx'ciincns, tlici'c are also a soiiicwliat. modified form distinguished l)y a slightly wider iimbilieus and compressed whorls. For the present, this is considered only as a variety. Localities : — The Yubarigawa ; the Ikiishnmbets and its branches, the Yoshiashizawa and the Kikumezawa ; between the Motomari and the Chietomanai, Söyagöri, Prov. Kitanii ; Shiintsuru- hara, on the left l)ank of the Teshiogawa ; the Opiraushibets and its tributaries, the Penke-kenebets and the Panke-kenebets, in the Province of Teshio. The figured specimen was found in a pebble from the Yübarigawa. Horizon : — Upper Ammonite-beds; abundant in the SWtphites- beds. TETR AGONITES CRASSUS JiMRr. sp. 189o. Lytoceras crassum, Jimp.ô : 1. c. p. oö (181), PI. YI. (XXIT.), fig. 5 a, b. AVhorls nearly round, slightly flattened on the sides and inflated ventrally, l)roader than high. Umliilicus moderate in size, deep, surrounded by a perpendieukr wall. Involution about il. 8urfece unknown, the only specimen being a cast of an interior. The suture line is shown in Jimbö's work. The siphonal saddle, however, is high and trigonal, and not so low or flat at the apex as is shown in his figure. The external as well as the first lateral lobe is asymmetrically bipartite ; besides, there are four auxiliary saddles and corresponding lobes ; the first auxiliary saddle is just at the umbilical edge whence the others incline obliquelly downward to the nmlulical suture. JiMBo took the shape of the whorl to be the character distingiushing the present speci(\s fi-om the precerliüg. He do- 48 ART. 2. IT. YAEE : CRETACEOUS scribes T. crassus as " Windung seitlich niclit abgeplattet " and 7\ sphaeronofns as having " Seitenflächen etwas abgeplattet."'^ The jii-esent species being based on a single young specimen represented by an internal cast, the writer is not in a position to draw a distinct line of demarcation between it and T. spliaeronotus or T. (jlabrw^. However a slight difterence in the form of the whorls and the somewhat wider umbilicus of the present species, as compared with the above two, seem to indicate that it belongs to a distinct species. Locality: — In a pebble from the Pombets, Prov. Ishikari. Horizon : — Unknown. TETRAGONITES POPETENSIS m. PI. VTI., iigs. 4. a,b, and (1. Dimensions :■ Diameter. ITeiglit (ii" till' last whorl. Breadth of tlie la^t wlmrl. Wndtli of niubilion.s. (J) ,:.,i,J (2) 0.4 em. 1.4.') „ 1.50 „ 1.0, „ 100. 2.0')em. 42. 0.1)1» „ 44. no. 0.90 „ 0.55 „ 100. 4.1. 2fi. 1) A specimen from the Shi-sanushil)e. (PI. VII. figs., 4 a, b). 2) A specimen from the Penke-opushikep, a tributary of the Teshiogawa. (PI. VII., fig. C>.) Shell discoidal, compi'essed, the smaller specimens measured above, l)eing composed of five volutions. Umln'licns moderate in size; involution about n. Whorls sulxpiadralc in section, inflated ventrally, and the umbilical wall perpendicular. Surface smooth, with exceedingly fine lines of growth. Periodic arrest of growth 1] .Jnrnri: 1. e. f'EPITALOrODA FROM TTTE TTOKKAIDi». 49 marked only l)y fool)le farrows on tlic cast ; al)Out six of tlicm wore counted in tlic smaller specimens. These strire or lines and furrows are of exactly the same nature as those of the ahove three sjiecies, hut project more strongly forward. The suture line shows four saddles on the external side, a saddle and one or two denticles following them on the umbilical wall, and two saddles on the internal side. In the suture line and the general aspect of the shell, the only species which approaches this is T. epigonus Kossmat. The suture lines of both species can hardly be distinguished from each other. However, the Japanese form is rather widely umbilicated and its Avhorls are lower. Localities : — The Makka-ushipe, a ti-ibutary of the Saniishibe, Prov. Ihuri ; the Fenke-opushikep, a tributary of the Teshiogawa, Prov. Tesliio. One specimen from each locality. From the Shi- sanushibe, also a tributary of the Sanushibe, we have three specimens, one of which is figured. Horizon : — Pachycl iscus-heds ; the horizon at the last men- tioned locality has not yet been ascertained. TETRAGONITES cfr. EPIGONÜS Koss^fat. PI. VIL, fig. 3. Compare : — 189-5. Lijioceras epigonumKoBSM at : 1. c. p. 39 (135), PI. HI. {XVIL), figs. 4, a, b, c, and 5, a, b. 1865. Amiiionit(^i^ Tim.otheanu^ p.p. i^TOLiczrcA : I.e. p. LIH, PI. LXXIIL, %. 5. 50 ART. 2. IT. YAP.E : tRRTACEOUR imensions : — Diameter. 3.2 cm. Uatio. 100. Height of the last whorl. 1.4 „ 43. Breadth of the hist wliorl. 1.^)5,, 48. Width of umbilicus. 0.90 „ 28. Geueral shape as in T. spliaeronotus. Umbilicus moderate in size, surrounded by a perpendicular wall. Involution about i. Whorls rather slowly growing, subquadrate in cross-section, some- what broader than high and broadest near the umbilical margin. Surface smooth, but with periodic arrest of growth, indicated by inconspicuous round ribs of which four have lieen counted on the last half of the outer volution. The character of the ribs is the same as in T. glahrus and T. sphaeronotus being distinguished only by a somewhat more oblique course. Suture as in the preceding s])ecies. The Japanese specimens agree quite well with those described by KossMAT under the name of T. epigonus from the upper Trichinopoly group of Andur and Varagur, S. India. The l)ody Avhorl of the larger Indian specimen figured by Kossmat measures 2.6 cm. high and 2.7 cm. broad, and its umbilicus 1.5 cm. in width, the diameter of the shell itself being 5.9 cm. Among the Japanese specimens there is none which attains such a large size. But they are comparable to, and agree in every particular, with the smaller ones of T. epigonus also figured by Koss:\rAT. The periodic arrest of growth is sometimes entirely absent and sometines present on the young shells of T. epigonus, so that this seems to be of no. special importance in the distinction of the species. In the specimens in the writer's collection, the periodic i'il)S are shown with moderate distinctness and their number agrees with those of the larger specimen figured by Kossmat. CEPHALOPODA FROM THE HOKKAIDO. 51 The reason wliy tlie writer hesitates to identify the Japanese form witli T. epigonus lies in the fact that his specimens are all in a young stage, and that there is none which clearly shows the suture line. Distinguished from T. popetensk by the higher and broader whorls and the narrower umbilicus ; from T. sphaeronoius and T. glabrus hy the lower whorl and the wider umbilicus. T. crciësus, on the other hand, is distinguished by its broader whorls. Localities : — The Penke-kenebets and the Kenekawen-opirau- shibets, both branches of the Opiraushibets, and the Abeshinai, a tributary of the Teshiogawa. Five specimens have been examined. Horizon : — fScaphites-heds. Foreign Localities of T. eplgonun : — U[)per Trichinopoly group of S. India and h'enonian of Tunis." ■r^-^^r^r -1) L.-'T'ervinqui EKE: Sur un ■ Facies particulier du yénuuieu tlu Tunisie. 1898. 52 ART. -H. YABE : CKETACEOUÖ Bibliography. F. M. Anüei;son : J. Böhm : J. Böhm : M. Boule : D. Bkauns : A. FuiTScii und U. W. M. Gabiî: L. C. CluiESBACii : A. DE Grossouake A. DE GUOSSOUVKE F. VON Hauer : A, Hyatt : K. JjMiîu : E. Koken : F. KossMAT : F. KOSHÎIAT hJoine Cretaceüus Beds of Fiuuge Hiver Valley, Oregon. Jonrn. Geol. Vol. IIJ. 1895 Die Kreidebildungen des Fürbergs und Sulzbergs bei tSiegsdorf in Ober-Bayern. Paheontographica. Vol. XXXVIII. 1891. Ueber Ammonites pedernalis v. Buch. Zeitscbr. d. deut. geol. Gesel, Bd. L. Heft. I. 1895. Note sur de nouveaux fossile secondaires Madagascar. Bull, du Mus. d'liist. nat. Paris. 1899. Vorläufige Notiz über Vorkoniuisse der Juraformation in Japan. Mittb. der deutscb. Gesell, für Natur — u. Völkerkunde Ostasiens. 1880. ScHEOENi'.ACH : Cepbalopodeu der br)limiscben Kreide. 1872. Pahoontology of California. Vol. I. 18G4. Vol. IL 18G9. Geology of Natal. C^uart. Journ. Geol. Soc. London. A^ol. XXVIL 1871. llecbercbes sur la Craie su[)(''rieuro. Mémoires pour servir à l'explication de la Carte géulogi(|ue détaillée de la France. 1893. Sur la Genre Neoptij dûtes. Bull. Soc. Geol. France, Serie III, Tome 24. Neue Cepbalopoden der Gosaugebilde. Sitzl). d. k. Akad. d. wiss. matb.-nat. Classe. 1866. Zittel's Text Book of Paleontology, translated by Ch. P». Eastman. Cepltalopoda. 1899. Beitnlge zur Keuntniss der Fauna der Kreidc;- formation von Hokkaido. Pal. Abb. Ijd. VI (Neue Folge, Bd. II). Heft 3. 1894. Die Vurwelt und ihre Entwickelungsgeschicbte. 1893. Die Bedeutung der Südindischen Kreideformation fiu' die Beurtbeilung der geograpbyscben Verbälnisse wjüirend der späteren Kreidezeit. Jahrb. d. k. k. geol. Keichsanstalt. Bd. XLIV. Ileif 3. 1894. Untersuchungen über die Südindische Kreidefoima- CEPHALOPODA FKOM THE HOKKAIDO. ÖO F. G. KOSSMAT : KuAussE : DE LaITAKENT E. Naumann : M. Neumayr : A. d'Oubigny ; A. d'Oubigny : J. F. PiCTET et \V. Fl!. Aue;. Qui'N^i'iii'T A. lÎEDTENBACIJEli : Cl. Schlüter : Fl!. Schmidt : \). SiiAurE ; J. Slmionescu : J. Slmionescu : J. r. Smith : tion. Beiträge z. Palasontologie u. Geologie Öster- reich-Ungarns IL d. Orients. Bd. IX. 1895. The Cretaceous ])ei)Osits of Pondicherry. Record Geol. Surv. India. Vol. XXX., pt. 2. 1897. Ueber Tertiäre, Cretaceische und ältere Ablagerungen aus West Borneo. Samm. d. Geol. Mus. z. Leiden. Series I. Bd. .3. Traité de Géologie. 4 Édition. III. 1899. Ueber Kreidefossilien von der Insel Sachalin. Jaiirb. d. k. preuss. geol. Landesanstalt für 1898. Ueber das Vorkommen der Kteideformation auf der Insel Jezo. Mitth. der deutsch. Gesell, für X^atur-u. Völuerkund Ostasiens. 1880. Die Ammoniten der Kreide und Systematik d. Aramonitiden. Zeitschr. d. deutsch, geol. Gesell. Bd. XXVII. 1875. Paléontologie Française. Terrains Crétacés. Tome L 1840-1841. Prodrome de Paléontologie. 1850. lîoux: Mollusques fossiles des Grés Verts des environs de Genève. 1847-53. : Petrefaktenkunde Deutschland. Abth. I. IhI. I. Cephalopoden. 1846-49. Die Cephalopodenfauna der (^nisauschichten in den norddeutschen Alpen. Abh. d. k. k. geol. Reichsanstalt. Bd. V. 1875. Cei)halopoden der oberen deutschen Kreide. Pahe- ontographica. Bd. XXL, 1871. Bd. XXIV., 1874. Die Petrefakteu der Kreideformation von der Insel Saclialin. Mém. de l'Acadamie Impériale des Sciences de St. Petersburg. Serie VII. Tome XIX. No. 3. 1873. Mollusca of the Chalk. Pal;eontograi)hical Soc. London Vol. VII— IX. 1853—1855. Ueber eine Unter-Cenoman Fauna aus den karpatlien Piumaniens. Verh. d. k. k. geol. Keichsanstalt. 1897 Fauna- Cretacica Superiora de la Ünnüs (Transilvania). Academia Puniana. No. IV. 1899. The Develo[)nient of Lytoceras and Phyllogeny. Proc. Calif. Acad. Sei. 3rd Series. Geology. Vol. I. No. 4. 54 ART. L'. — H. YABE : CRETACEOUS F. ÖTOLICZKA : V. Uhlig : V. UiiLiG : J. F. WlIlTEAVKS T. \V. »Stanton : The Fauna of tlie Sliasta-and Cliico-fürmatiun.s. Bull. Geol. Soc. America. Vol. IV. 1893. T. W. Stanton and J. S. üiller : The Shasta-Chico Series. Bull. Geol. Sog. America. Vol. II. 18.94. T. W. Stanton : Contribution to the Cretaceous Paleontology of the Pacific Coast : The Faima of the Knoxville Beds, Bull. U. S. Geol. Surv. No. 1.33. 1895. G. Steinmann, W. Deecke and VV. Müricke : Das Alter und die Fauna der Quiriquina Schichten in Chile. Neues Jahrh. Min. Geol. und Pakeont. Beilage Band X. 1895. The Fossil Cephalopoda of the Cretaceous Rocks of Southern India. Palœontologia Indica. 186.3. Bemerkungen zur Gliederung karpathischer Bild- ungen. Jahrb d. k. k. geol. Reichsanstalt. Bd. XLIV. Heft. 2. 1894. Abwehrende Bemerkungen zu R. Zuber's " Strati- graphie der karpathischen Formationen." Verh. d. k. k. geol. Reichsanstalt. 1900. On some Fossils irom the Nanaimo Group of the Vancouver Cretaceous. Trans. Roy. Soc. Canada. 2nd Series. Vol. I. Sect. IV. 1895. J. F. Whiteaves : Mesozoic Fossils. Vol. I. Part 1-4. 187G-1900. J. F. Wiin'KAVES : Note on a Sujiposed New Species of Lytoceras from the Cretaceous Rocks at Denman island. Ottawa Naturalist, Vol. XV. No. 2. 1901. H. Yabe : Orbitoli na-IÀmestone in the Hokkaido (in .Japanese). Journ. Geol. Soc. Tokyo. Vol. VII.' No. 91. H. Yabe : Note on Three Upper Cretaceous Ammonites from Japan outside of the Hokkaido. .Journ. Geol. Soc. Tokyo. Vol. VIII— IX. 1901-1902. H. Yace : Note on Some Shark's Teeth from the Mesozoic of Japan. Journ. Geol. Soc. Tokyo. Vol. IX. No. 110. 1902. YoivOYAjfA : Versteinerungen aus der japanischen Kreide. Pake- on tograi.hica. Bd. XXXVI. 1890. Zuber: Geologie der Erdöl-Ablagerungen in den galizischen Karpathen. I. Allgemeiner Theil. Heft. I. Strati- graphie der karpathischen Formation. 1899. M. R. CEPITAT.OPODA FltOM TFTE ITOKKAIDO. 55 Contents. IntrofUictory Remarks I/ijtorei-as, Gaudnjcems and Telrcif/onilcs, in iienenil.. Jjylncera.i Lytoreras ezoëme m ,, imperialem Gaudryceras Gaudri,C2ras tenuiUruluin ni „ „ var. ornata m. „ „ var. intermedia m. „ „ var. infrequens in. „ crassicosfatum JiMBÖ sp. „ denseplicafum Jimbo sp. ... „ striatum Jimbo sp „ „ var. pi(ta ni. ,, limatiim m „ „ var. obgrura m. „ Yoloyamai m „ Yamashitai uj „ sp „ Kuivanoi Jimbo s]i Tetragonites Tetra fjonitcs ylabrus JiMBÜ sp „ sphaei-onnlus Jimbü sp „ crassus JlMBü sp ,, popefensis m „ cfr. epifjonus Kossmat sp. BiMiography Tack. 9 9 U 19 24 27 2?^ 29 30 31 33 34 3G 3G 38 40 41 43 43 45 47 4S 49 H. YABE. CRETACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE I. Lytoceras. Plate I. Lytoreras ezo'ènse m. Pp. 9-11. Fig. 1. Side view of the tyjie, preserved in the 8ci. Coll. Mus. From lower Ammonite-beds of the Pompets, Prov. Ishikari. Piodnced to £ nat. size. Jour. Sei. Coll. Vol. XVIII. Art. 2. PI. I. J. [nhizaki del. ir. Y ABK. CRETACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE II Lytoceras. Plate IL Lijtoccras imperiale m. Pj). 11-12. Fig. 1. Side view of the type, ]ireserved in tlie 8ci. Coll. Mus. From lower Aiiimoile-beds of the Ilcii.shiuiiLetH; I'rov. Isliikari. licdiiccd to | II at. size. Jour. Sei. Coll. Vol. XV I II. Art. 2. PI. II. J. Ishizaki del 11. YABE. CRETACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE III Gaudryceras. Plate III. Gaudrijccras tcvuHirotaia m. Pp. 10-20. (Ail Ihc specimens in llie »^ci. Cuil. JMus). Y'vs.. 1. var. intermedia, a, side view; b, front view of a specinien i'roni upper Ammonite-beds of the Sannshibe, Prov. Ibnri. Nat, size. Fi^-. 2. var. ornata. a, side view ; h, front view of a specimen from r((chi/(liscu.s-\)i'{h of tlie Kiknnie/awa, Pntv. Isjiikari. Nat. size. VW. 3. A part of the suture line of a large s})ecimeu uf normal ty[)e. From Fachydiscus-hcd^ of the Bannosawa, Prov. Ishikari. Nat. size. — s., siphonal line. Y'vj^. 4. Suture line of a s}>ecimen. 3..") cm. in (ham., of normal type. From Pachydiscus-heda of the iSanushibe, Prov. Iburi. — s., sjphutial line ; ura., umbilical margin ; u-\, umbilical suture. 2 i x . Jour. Sei. Coll. Vol. XVIII. Art. 2. PI. III. K 4-%- '^^■ Fig. 3. A â Fig. ^.b. Fig. 2.a. \. J. Ixhiziki del. Fig. La. Fig l.h. H. YABF. CREFACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE IV. Lytoceras and Gaudryceras. Plate IV. Li/ioceras imperiale m. P. 12. Fig, 1. Sntnre Une, drawn from the specimen figured in PI. TL Nat. size. — 9., siphonal line ; nm., ninl)ilical margin ; ns'., nmLilical suture. Gcmdryceras limatum m. P. 35. Fig. 2. Suture line, drawn from the specimen figured in fig, 2, PI. V. Nat. size. Lettering as in fig. 1. Gaudryceras ienuil irai urn m. Pp. 28-29. Fig. :'. var. infrequens. a, side view; b, front view of a specimen preserved in the Sei. Coll. Mus. From ScapJdtes-lmh of the Opiraushihets, Prov. Iburi. Nat. size. (landrycerns crassicosfaf/rin Jimbo s}). l^p. 29-30. Fig. 4. Side view of a specimen in the Sei. Coll. Mus. From ujjper Ammonite-heds of Cape Soya, Prov. Kitanii. Nat. size. Gaudryceras striatum Jimbo sp. Pp. 31-34. Fig. 5. Side view of a specimen in the Sei, Coll, Mus. From Puchydiscus- heds of the Abeshinai-rubeshibe, Prov. Teshio. Nat. size. Fig. G. var. pieta. a, side view ; b, outline of a transverse section ; c, surface of a septum, showing the septal lobe, of a specimen in the Sei. Coll. Mus. From FacJ/yrïisc/is-heih of the LTipets-rubeshibe, Prov. Teshio. Nat. size. Gaudryceras Yamashifai m. Pp. 38-39. Fig. 7. Side view of the type in the Sei. C(dl. Mus. From upper Ammonite- beds of the Saiuishibe, Prov. Iburi. Nat, size. Jour. Sei. Coll. Vol. XVIII. Art. 2. PI. IV '*^, Fig. 1. ^...tf^fyTff^^ï^- ïïS- F ig. o.b. Fig. 2. Fig. S.a. ,-<^' Fig. O.a. Fig. O.h. Fig. 7. y^f'^ Fig. O.e. Fig 4. F.,-. ..• ■J . Uh'izaki del. H. YABE. CRETACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE V. Lytoceras and Gaudryceras. Plate V. LijInciTdFi f'zoi'iifie m. P. 10. Fig. 1. Snture line, di'iiwn iVdin tlic sixcinicti fii^nrcd in PI. T. Nat, size Gaudryceras limatnm m. P. 3fî. Fig. 2. 8ido view of a specimen, preserved in the Sei. Coll. Mus. From upper Ammonite-lieds of the Yfiharigawn, Prov. Ishikari. Nat. size. Jour. Sei. Col!. Vol. XV HI. Art. 2. PI V. Fig. 1. M^ Fig. ;?. / ■%-. ^m IshJziU del. H. Y A BE. C.^iTACEOaS CEPHALOPODA FROM THE HOKKAIDÜ. PLATE VI, Gaudrvceras. Plate VI. Gaudryceras Yol:oyamai m. Pp. 36-38. Fio-. ]. a and ]), side and front views of the type in the Sei. CdII. Mus. From njiper Aninionite-lieds of tho Yilharigawa, Prov. Tsliilcari. Xat. size. Fig. 2. n, side view ; h, front view, of a younger specimen in tlie Rci. Coll. Mus. From upper Amnion ite-l)eds of the Yfibarigawa, Prov. Ishi- kari. Nat. size. Gaudryceras limatnm m. Pp. 34-3(). Fig. 3. a, side view, somewhat restored ; b, front view, of the tyi)e in tlie Sei. Coll. ]Miis. From u]iper Ainmointe-beds of the Yoshiasliizawa, a tributary of tlie Ikushumlx'ts, Prov. Ishilcari. Nat. size. Jour. Sei. Coll. Vol. XVIII. Art. 2. PI. VI. Fig. la. Fig. 3.b. Fig. 2.h. f " ■" Fig. 2.a. Fig lb. Fig. S.a. J. hhlzaki del. II. Y ABK. CRETACEOUS CEPHALOPODA FROM THE HOKKAIDO. PLATE VII. Gaudiyceras and Tetragonites. Plate VII. TeiTCKjonites sphaeronotus Jjmbô sp. Pp. 45-47. Fig. ]. a, Tartaric acid. I) Alkiilis Atiuuonia. Yellow. ami Caustic jwtash. )) Suits. Caustic soda. ,, Lead acetate. Yellow precipitate. Sodium carbonate. Slightly yellow. Potassium bichromate. Reddish brown precipitate (not all). Ferric chloride Black (deep green) precipitate. Ferric chloride + potassium ler- Deep blue precipitate. rocyanide. Ferrous sulphate. No reuiarkable reaction. Silver nitrate. Brown precipitate. IMillon'.s reagent. Browinsh yellow. Iodine in potassium iodide. No reaction. Quinine sulphate. „ Potassium cyanide. Yellow. Flenuuins's solution. No reaction. Chloral hydrate. ,, X Naphtol + sulphuric Acid. (Mesophyll violet). Alum. Y^cllowish orange. Antipyrine. No. reaction. Ca Heine. " From these cheniieal reactions protaiitliocyaii and yellow antboeyan may prol)ably be considered as some allied compounds of tannin or modified ])licnol compounds, as Pfeffer^' remarks. Indeed I have often detected a substance like protantbocyan in the epithelial cells of white Hower petals of. many other plants, 1) Pfetlér.— PH:mzen Physiologie. Ed. I. 1807, p. VÔÔ. AXTITOCYAN IX TTTE PETALOID CALYX. / but this substance never develops later on into any coloured anthocyan. Xow tbe question arises wliy in certcun flowers protanthocyan precedes a coloured antbocyan wbile in some otbers it does not. Tbe answer is to be sougbt, probably, in tlie bereditary disposition of eacb plnnt. Tbat tbe colourless protantbocyan becomes yellow by tbe action of alkalis, as stated in preceding table, is wortb noticing, for it proves tbe presence of a tannin substance, as Wiesner^^ observed. Tbe cbange of tbe colour of yellow anthocyan, wben well developed, to a ligbt red by tbe action of acids, is also of some interest. jNIoreover, tbe existence of intercellular air-spaces and the exceedingly abundant presence of tbe crystals of ciileium oxalate at tbis stase is wortb reraarkins:. The Second Phase. Duration of observation : July 1 — July 20. External form. — Eacb wbite calyx-leaf of tbe opened flowei's now begins to colour, by tbe appearance of red spots from tbe distal end of its upper side. Tbe colour proceeds along tbe tnargin centripetally over tbe surface (Fig. .^.4). In general, tbis propagation of colour occurs earlier on tbe larger calyx-leaf tbnn on tbe smaller one, tbougb at the end tbe wbole u]3per surface becomes coloured, except tbe borders of tbe veins. For tbis development of colour it is necessary tbat tbe calyx sbould be exposed directly to tbe sunlight, the overlapped part of the upper side of a calyx-leaf and also the wbole lower side being not at all tinged till the end of this phase. 1) Wiesner, J. — Einige Beobachtungen über Gerb- nn'l Fni-bstoTe der Blnnienbliitter. (Bot. Zeitg, 1862, p. 389). 8 ART. T. TCHTMUKA : OX THE FORMATION OF Microscopical structure. — It is noticeable that the epidermal cells of the upper side become much projected upwards, assuming the shape of a cap, whilst those of the lower side remain unchanged. The glossy-white appearance of the calyx-leaf which is observed at first is chiefly due to the abundant presence of air in the intercellular space and partly also to the reflection of light by the cap-shaped projection of the epidermal cells. Later, many epidermal cells containing red anthocyan, besides those containing the yellow variety, are found on the upper side, whilst on the lower side coloured anthocyan has not yet made its appearance. The microscopical study of the calyx-leaf of red Hortense certainly reminds one of that of the flower-petals of several other plants. The cnticnlar radiating striations of the upper side become, finally, very distinct and the paralled striations of the lower side, commonest in epidermal cells of all plants, also l)ecome visible. The air-spaces and the crystals of calcium oxalate show a tendency to gradually decrease and are mostly concentrated toward the hypodermal layers. The chemical reactions of red anthocvan are shown in Tal)le TL Table TL Een gents. Reactions of red anthocyan. Acids. Hydrocliloric acid. No reaction. Nitric acid. Yellow. Sulpluiric aeiil. Yellowish hrown. Acetic acid. No reaction. Phosphoric acid. „ Salicylic acid. )) Tartaric acid. V Aî^HOCYAN IN THE PETALOID CALYX. Alkalis Ammonia. Yellowish green. and Caustic potash. Green. Salts. Caustic soda. » Lead acetate. Green precipitate. Sodium carbonate. Yellow (contracts into" globul mass). ir Potassium carbonate. tJreen ( „ „ „ . Potassium bichromate. Deep blue precipitate. Ferric chloride. Black (deep blue) precipitate. Ferric cliloride + potassium fer- Deep blue precipitate. rocyanide. Ferrous sulphate. No reaction. Silver nitrate. Brown precipitate. Millon's reagent. Brownish yellow. Iodine in potassium iodide. No reaction (contracts into globular mass). a Quinine sulphate. No reaction. Potassium cyanide. Violet yellow. Flemming's solution. Violet, dark blue, brown. Cliloral hydrate. Dissolves out. X Naphtol + sulphuric acid. Brown. Alum. No reaction. Antipyrine. J) Caffeine. )) Looking over the table it will be seen, in general, that acids do not produce any marked change in the colour of the red anthocyan but that alkalis change it into green. The presence of sugar in the mesophyll layer remains doubtful. Chlorophyll- grains have not yet nppeared. The Third Phase. Duration of observation : July 20 — August 1. External form. — The majority of the flowers of red Hortense now tend to nutate around their stalks, though some of them are not yet wholly reddened on the upper side of the calyx. Each calyx-leaf becomes turned upside down. This process which is 10 ART. :î. T. ICHIMURA : ON THE FORMATION OF physiologically produced by the epinastic growth of the flower stalk may from a biological point of view be described as a phenomenon of " gamotropism."^^ Being exposed directly to the sunlight, the lower side of the calyx, now in an nppermost position, turns red beginning at the base and ending at the periphery in a way just opposite to that observed in the case of the upper side. The red colour on the lower side is not so bright as it is on the upper side, but is rather dark except at the veinal surface (Fig. 7.). Besides, the darkness of the red colour is in- creased with the concentration of the spot and with the appearance of green chlorophyllgrains. In the meantime, all parts of the calyx-leaf on both sides, hitherto whitish, become noticeably green (Fig. 7,8.). Microscopical structure. — On the npper side the number of red coloured epidermal cells and the intensity of the red colour are both gradually increased. Some slightly blue cells are often met with in the hypodermal layer. On the other hand, the red cells on the lower side are mainly observal)le in its hypodermal layer consisting of idioblastic irregular cells, Init partly also in its epidermal layer, the cells of which mostly remain colourless. Such colouration of the hypodermal cells may be characteristic of the calyx-leaf of Japanese Hortense, for it is not usually found in the flower-petals or petaloid calices of other plants, which contain anthocyan. The multipolar horns of the idioblastic cells of the entire mesophyll layer gradually become prominent and, at the same time, the layer appears more spongy, the intercellular spaces being more conspicuous. In the normally healthy state (i.e. when not diseased or injured) the red idioblasts are confined to the hypodermal layer and never go any deeper. This condition 1) Ludwig, Fr.- Lehrbuch der Biologie der Pflanzen. 1895, ? 74. ANTHOCYAN IN THE PEÏALOID CALYX. 11 differs greatly from that of the autumnal red colouration of leaves. Crystals of calcium oxalate decrease still more than in the preceding ])hase. Some raphides are rarely found in the middle part of the mesophyll layer. Chlorophyll-grains begin to appear, at first, in both hypo- dermal cells and later extend toward the middle of the mesophyll. The formation of starch, accompanying the chlorophyll-grains, is soon afterwards observed. That chlorophyll-grains appear later than red anthocyan has been noticed by former investigators^^ in many cases, and this fact is no doubt to be explained from a teleological point of view. The chemical reactions of dee[) red and blue anthocyans are Liiven in Table III. Table III. ^\., Ecagcnts. Eeactions of deep red anthocyan. Eeactions of l)lue anthocyan. Acids. Hydrochloric. No reaction. Eed. Nitric. ij )) Sulphuric. !) Eed, brown, yellow. Acetic. )) Eed. Phosiihoric. „ (contracts into a globular mass). >) Salicylic. » » Tartaric. )) I) Alkdis. Aiiiuionia. Deep green (blackish). Green. and Caustic jiotash. Green (then yellow). Green (soon afterward Salts. yeUow). 1) Pick, Bedeutung des rothen Farbstoffs bei Phanerogamen. (Bot. Cent. Blatt. Bd. 16. p. 2S], 1883). Kerner, Plkmzen Leben. 2. AuH. Bd. I. p. -170, Stahl, fAnnales du Jardin Butaniipic de Buitenzorg. Vul. XIII. p. 137-216, ISUti). "*■■"■ ' 12 AllT. d. T. ICHlMUllA ! ON THE FORMATION: OF Lead acetate. Sodium carbonate. Potassium carbon- ate. Potassium bichro- mate. Ferric chloride. Ferric cliloride + potassium fer- rocyanide. Ferrous sulphate. Silver nitrate. Millou's reagent. Iodine in potas- sium iodide. Quinine sulphate. Potassium cyanide. Flemming's solu- tion. Chloral hydrate, a Naphtol + sul- phuric acid. Alum. Antipyrine. Cafleine. Deep green prc»ipilate. Dark green (contracts into a globular mass). Black (deep green) (con- tracts into a globular mass). No reaction. Black precipitate. Deep blue ]>rccipitale. No reaction. ;) (Violet, brown, dark). Reddish brown precipitate, (later dark brown). No reaction (contracts inlo a gloljular mass). Violet, green, yellow, (granules dissolve later). CIrccn, yulldw (ccmlraets into a globular ma^s). Violet, them black, brown- ish yellow. No reaction. Dissolves. No reaction (slightly violet). No reaction. ( Jreen (precipitate ?) Greenish yellow. Slightly red. Momentarily red, then black. Momentarily red, then black and blue pre- cipitate. No reaction. » (slightly brown). Momentarily red and then reddish brown. No reaction (slightly black). Violet, green, yellow. (granules dissolve later). Sligiit green, yellow. Momentarily red, then slightly black, brownish yellow. No reaction. Slightly red (dissolves). No reaction. Here tue presence of «ngar mostly around the libro-vaseular luiiulle of the calyx-leaf cau be denioustrated. The Fourth or Last Phase. Duratiou of ol)servatiou : August 1 — Septeudjer 1. External form, — Flowers standing erect cau no longer be fo\ind. The cT^yx-leaves nutate and are soon reddened on the ANTHOCYAN IX THE PETALOID CALYX. 13 lower side. A very quick and remarkable develoj)raeiit of red aiitliocyaii on some injured ]mrts of the calyx-leaf is particularly noticeable, and is most apparent on the uj^per greenish side. (Fig. 1).). This phenomenon is in entire agreement with the observations made by Linsbauer, Ludwig and Molisch. Now when the red colour of the lower side attains its maximum, it appears as a somewhat dark red, whilst on the veinal surface it is a rather distinctly crimson red. This dark red colouration is chiefly due to the mixing of the epidermal deep red, and the hypodermal blue, anthocyan. (Fig. 10.). In the course of the process of the degeneration of the calyx-leaf the interveinal mesophyll i)arts become dark brownish, then gradually dry up with the decrease of the sap, and finally die out. After the removal of the destroyed mesophyll parts there still remains the brownish skeletal vein-net of the calyx- leaf. This dead calyx-net is not deciduous till the spring of the next year. Microscopical structure. — Both red and blue anthocvans now l)ecome nuich more concentrated in all the cell-sap. Nevertheless, the colouring matter is still to be found in some cases in its various younger stages — colourless, yellowish and of a pale red. In this phase it is wortli noticing that violet or bluish crystals and refractive globules appear in the outer layers of the calyx-leaf. (Fig. 12.). ]\Iany of the violet crystals appear granular and measure about 5 /^- on the average. Some of them have the form of a needle shaped prism, 7-9 p- long. They are found in the hypo- dermal or epidermal cells on both sides of the calyx-leaf. Their chemical reactions completely agree with those of the blue or violet cell-sap, and hence these crystals must be regarded as 14 ART. 3. T. ICH IM Uli A : ON THE FOKMATION OF those of antliocyan. (Fig. 14). They dissolve in hydrochloric, acetic, and other acids to form a red solution. With sulphuric acid, however, no sooner are they dissolved, than they reajopear as red brownish amorphous granules. Potash dissolves them into a pale green solution, whilst chloral hydrate dissolves them without any special change of colour. They are, no doubt, identical with Zimmermann's "pigment secretion" and Krœmer's^^ "violet chromatophore," which are generally considered as somo tannin compounds of antliocyan. Hitherto the crystal form of antliocyan in a living cell has been but little known. Besides the two writers above named, Husemanir^ had enumerated the plants in which " j^igi^i^^^t corpuscles " (Farbkörper) occur. Stras- burger's " l^lue star," consisting of short needles of crystallized antliocyan, is found in many cells of the blue coloured calyx-leaf of the Delphinium- flower. It is identical with Zimmermann's "pigment secretion." A peculiar body, the so-called refractive globule, is found constantly in this last phase within each epidermal cell of the calyx-leaf of the Japanese Hortense. (Fig. 13.). It a.ppears, at first, as numerous line drops around a nucleus, which gradually unite into one large refractive globule, measuring 4-5 1'-. It is stained by alkannin, coloured a light brown by sulphuric acid, and is insoluble in alcohol, hydrochloric and acetic acids. This body closely resembles Krœmer's " strong refractive globules " found in the coffee-berry, and only differs from it in being almost insoluble in chloral hydrate and alcohol (5: 2). Aqueous ammonia ]:»roduces no reaction with it. Probably it may be a certain 1) Krœmcr, K. — Ueber das angebliclie Vorkoiuinen vmi violrtten Cliromatoplioren. (lîot. Cent, iiiatt. Nr. 41. T.d LXXXIV. 1900). -) Hiisciiiann, Th.— Die rilanzeubil.()lié. löS".', Hd I, ]>. 'lîM. ANTHOCYAN TN THE PETALOID CALYX. 15 protein substance com1)ine. 2) Th. Lestibondois. Étude sur I'anatomie et la physiologie des végétaux. Lille 1840. 3) Ph. van Tieqhem. Memoir sur la racine. Paris 1872. 4) G. DuTAiLLY. Kecherches anatomiques et orgauogéniqucs sur les Cucurbit acées. Congrès de Montpier. 1879. 2 ART. 4. — A. YASUDA : ON THE Cucurbitacea?, including that of the leaf-traces and of the bundles in the leaves themselves. A further investigation was conducted by Lotar/^ who compared the structures of the stems, hpyocotyls, leaves, tendrils, roots, and seeds of many genera in this family, such as Gitrullus, Lagenaria, Luffa, Goccinea, Cucumis, Oucurbita, Cyclanthera, Sicyos, Momordica, Abohra, Rhyncliocavpa, Bryonia, Ecballium, and Thladiantha. In his essay he described in great detail the distribution of the fibro-vascular bundles, with accompany- ing diagrammatic figures. Fischee,^^ examined the arrangement of the sieve-tubes in the stems of several genera, and classified them under four heads : (a) vascular-bundle sieve-tubes, (b) ecto- cyclic sieve-tubes, (c) entocyclic sieve-tubes, and (d) commissural sieve-tubes. He also referred to the mode of distribution of the sieve-tube system, and described six distinct types : (i.) Alsomitra type, (ii.) Luffa type, (iii.) Bryonia type, (iv.) Cyclanthera type, (v.) Lagenaria type, and (vi.) Cucurbita type. Penzig"^^ observed in the epidermal cells on the under surface of the leaves of 3Iomordica echinata and M. Charantia, globular cystoliths which, he found, oc- curred in groups, radiating from a common centre. In regard to tendrils Warming,^^ Dutailly,'^' Cogniaux,*^^ Müller,^^ and 1) Henri-Aimé Lotar. Essai sur I'anatomie comparée des organes végétatifs et des téguments séminaux des Cucurbitacées. Lille 1881. 2) A. Fischer. Untersuchungen über das Siebröhrensystem der Cucurbitaceen. Bot. Centralbl. 1885. Ed. XXI. No. 4. 3) O. Penzig. Zur Verbreitung der Cystolitlien im Pflanzenreich. Bot. centralbl. 1881. Bd. VII L No. 13. 4) E, Warming. Verzweigungsverhilltnisse der Phanerogamen. König, deutsch. Acad. der Wissenschaft. 1877. 5) G. DUTAILLY. Recherches organogéniques snr les formations axillaires ches les Cucurbitacées. Association français pour l'avancement des sciences. Congrès de Harure. 1877. C) A. CoGNiAUx:. Cncurl)itaceen in De Candolle's Monographiie Phanerogamarura. 1881. Vol. III. 7) E. G. O. Müller. Untersnchnngen über die Ranken der Cucurbitaceen. Cohn's Beiträge zur Biologie der Pflanzen. 188G. P.d. IV. lieft 2. COMPAEATIVE ANATOMY OF THE CÜCÜEBITACE^. 3 others came to the conchision, that the main axis of the tendril corresponds to a branch, which carries several leaves. The testa has been investigated by Bischoff/* Stkandmaek,-^ Höhnel,"^ Fickel/^ GoDFKm,^^ Hartwich/^ and Haez.'^ According to HöHNEL the testa consists of ten layers of cells, of which the first layer arises from the epidermis of the carpel ; the second, the swollen epidermis, is made up of prismatic cells with thickened- ridges on their walls, and together with the following three layers is derived from the outer integument of the ovule ; the third consists of irregular cells, abounding in intercellular spaces ; the fourth, of large stone- cells; the fifth, of reticulate cells with many intercellular spaces ; the sixth, of compressed cells derived from the inner integument of the ovule ; the seventh originates from the epidermis of the luicellus ; the eighth, from the perisperm ; the ninth and the tenth, from the endosperm. So far the chief results of investigations hitherto made con- cerning the anatomy of the Cucurbitacese ; for my part, I have tried to examine and compare the internal structures of various organs of those Cucurbitacese which are found wild or are cultivated in Japan. The number of our known genera belonging to this family is fifteen : namely, Adiiiosiemma, Melotkria Muhia,^^ 1) G. W. BisCHOFF. Handbuch der botanischen Terminologie und Systemkunde. Nürnberg 1833. Bd. I. 2) J. E. Strandmakk. Bau der Samenschale. Jcsï's Bot. Jahresber. 1874. Bd. II. 3) F. V. HöHNEL. Morj)hologische Untersuchungen über die Samenschale der Cucur- bitaceen. Wien 187(). 4) J. F. FiCKEL. Ueber die Anatomie und Entwicklungsgeschichte der Samenschalen einiger Cucurbitaceen. Bot. Ztg., 1876. Bd. XXXIV. Nr. 47-50. 5) J. GoDFRiN. Etude histologique sur les téguments séminaux des Angiospermes. Paris 1880. 6) C. IIartavich. Ueber die Samenschale der Coloquinthe. Arcliiv. der Pharm. 1882. Bd. CCXX. 7) C. D. ÎIarz. Landwirtliscliaftliche Samenkunde. 1885. Bd. II. S) H. KuKonvA. A List of Phanerogams collected in the Southern Part of Okinawa, an island of the Loochoo Chain. Botanical Magazine, Tokyo. 1900. Vol. XIV. No. 163. p. 123. 4 AßT. 4. A. YASUDA : ON THE Zehneria^'' Schizojjeiwn, Momordica, Lvjf'a, CUrullus, Cucumis, Bryonopsis^^'' Benincosa, Lagenarla, Trichosanthes, Oiicurbiia, and Gymnostemina. Of these I have examined twelve genera, leaving for a future occasion the study of llukia, Zehneria, and Bryonopsis, which are at present unfortunately inaccessible to me. I have studied the anatomical characters of the stems, hy- pocotyls, . blades, petioles, cotyledons, tendrils, roots, fruits, and seeds of the following sixteen species : Aciinosiemma racemosum Maxim., Melothria japonica Maxim., Schizopepon hryoniœfoUus Maxhn., var. japonicus Cogn., 3Iomordica cliarantia L., Luffa cylindrica (L.) lloem., L. acutangula (L.) Roxb., Citrullus vulgaris Schrad., Cucumis sativus L., C. Melo L., Benincasa cerifera Savi., Lagenarla vulga,ris Ser., Trichosanthes cucumeroides Maxim., T. japonica Hegel., T. muUllGha Miq., Ciieurhiia Fepo L., and Gymnostemma cissoides Fr. et Sav. Whenever it has appeared necessary, old as well as young specimens have been examined and compared. In the present paper I shall treat the subject in nine chapters, to be followed with a summary of the chief results of my investigation. Here I wish to express my hearty thanks to Prof. ])r. J. Mastumuka of the Imperial University, of Tokyo, under whose care my investigations have been conducted. My thanks are due also to Mr. Y. Tanaka, Dr. K. Miyabe, Mr. T. Making and Mr. T. Yoshinaga, all of whom have kindly supplied me with valuable materials for study. 1) 11. KuKonvA. loc. cif. p. 12;î. 2) Ditto, p. 12;î. COMPAKATIVE ANATOMY OF THE CUCUllßlTACE.E. CHAPTER I. THE STEM. Coiitoni'. In tlio Caeurbitacejt} the contour of a transverse section througli the stem differs considerably in different species and varies even in one and the same individual according to its age. When young the outline of the stem is pentagonal, but becomes more or less roundish as the stems grows older. The stems of Benincasa cerlfera and Lagenarla vulgaris are especially remark- able for a secondarily produced process in the furrows. Generally speaking, young stems are solid, but in advanced age some stems acquire a wide lumen in the centre, and the formation of a narrow compressed cavity is common in the very old stems of many species. In Benincasa cerlfem (PI. I. fig. 1-4), Lagenaria vulgaris, and Cucurbita Pepo, when the stems are still young, a remarkable roundish hollow is already formed in the centre by a gradual splitting of the central tissue. The full-grown stems of Luffa cylimlrica and L. acutangula as well as the old stems of 3Iomordica charantia and Actinosiemma racemosum have sharply ridged angles. ]\Iicroscopical examination shows that the ridges of Luffa cylindrica (PL II. Fig. 35) and L. acutangula consist only of outgrowths of the collenchyma, while those of Momordica charantia (PI. II. Fig. 36) and Actinostemma racemosum, in spite of their resemblance to the former in external appearance, are formed Ijy newly developed secondary tibro- vascular bundles, which have originated from the 6 ART. 1. A. YASUDA : ON THE secondary meristem outside of the outer ring^^ of the vascular bundles. Stoniata. The stomata on the stems generally lie in the same plane with the adjoining epidermal cells, but they are sometimes elevated with the epidermis above the surface of the latter, appearing as if they were supported by a short thick hair. GiLCurbila Pepo, Benincasa cerifera, Lagenaria vulgaris (PI. II. Fio-. 34), Luffa cylindrica, and Trichosanthes cucwmeroides give good examples of the elevated stomata. The average numbers of stomata on a square millimetre of the stems of the Cucurbitacese in question are as follows : Trichosanthes cucuiiieroides 23 T. japonica 23 T. multlloha Gitrullus vulgaris Cucurbita Pepo 21 Actinostemma racemosuvi 20 Lagenaria vulgaris 18 Cacuiiiis Melo 18 Momordica charanlla 17 Liùffa cylindrica 16 L. acutangula 16 Benincasa cerifera. 15 Cucumis sativus 14 iWelothria japonica 12 Schizopepon hryoïiiœfolius var. japonicus 8 Gymnostemma cissoides 7 22 9') 1) The fibro-vascular bundles in tlie stems of the Cucurbitacece are arianged in two circles, outer and inner, each of which contains five bundles. COMPARATIVE ANATOMY OF THE CUCUKBITACEiE. 7 Tn'cJwmes. The trichomes on the stems are multiceUular in every case, and may be divided into four classes : (i.) sharp- pointed conical trichomes, (ii.) blunt-ended conical trichomes, (iii.) short-stalked glandular trichomes, and (iv.) long-stalked glandular trichomes. The first kind is found in Melothria japonica (PI. I. Fig. 5), Citrullus vulgaris^ Cucumis sativus, C. Melo, Benincasa ceriferay Lagenaria vulgaris, and Cucurbita Pepo ; the second, in Actinostemma racemosum, Schizopepon hryoniœfolius, var. japonicus (PL I. Fig. G), Momordica eharantia, Luffa cylindrica, L. acut- angula, Lagenaria vulgaris, Tt'ichosanthes cucumeroides, T. japonica, T. multiloha, and Gymnostemma cissoides ; the third, which has an oval head composed of several cells, in all species without exception; the fourth, which has an oval head consisting of one or more cells, in Momordica charantia, Liffa cylindrica, L. acutangula, Citrullus vulgaris, Cucumis sativus (PL I. Fig. 8), C. 3Ielo, Benincasa cerifera, Lagenaria vulgaris, Trichosantlies cucumeroides, T. japonica, T. multiloha, and Cucurbita Pepo. Among these the trichome of Benincasa cerifera is especially characterized by having two processes at the apex (PL I. Fig. 9) ; that of Trichosantlies cucumeroides, by having a single enlarged terminal cell (PL I. Fig. 10), and that of Cucurbtia Pepo, by having the head consisting of two cells, one placed above the other (PL I. Fig. 11). Cuticle. The cuticle on the stems is mostly thin and smooth. As exceptions, however, are to be noted Actinostemma racemosum and Gymnostemma cissoides, the former of which has the cuticle striated at the angled portions, while the latter has it so on all surfaces. Collenchyma. The collenchyma is developed at the angles 8 AET. 4. A. YASUDA : ON THE of the stems, and often in the fnrrows between the angles. For example, Actinostemma racemoswriy 3Ielothria japonica, Momordica charantia, Lagenavia mil gar is, Trichosanihes curwneroides, T. japonica, T. multiloba, and Gymnostcmma cissoides have the eol- lenchyma developed only at the angles ; while Luffa cylindrical L. acutangula, Citrullus vulgaris, Cucumis saiivus, C. Melo, Benincasa cerifera, and Cucurbita Pepo have it at the angles as well as in the fnrrows. The number of groups of collenchymatons cells developed in the furrows varies from two to five. Cucumis saiivus and C. Melo have only traces of this tissue in each furrow ; Luffa cylindrica and L. acutangula have two masses of it ; Citrulhis vulgaris and Benincasa cerifera, three masses ; and Cucurbita Pepo, five masses. The layers of the collenchymatous cells are much thicker at the angles than in the furrows. Their maximum number (sixteen) is found in the angled portions of the stem of Ljiffa cylindrica, while their minimum number (one) is found in the furrowed portions of the stem of Cucumis Melo. Selevenchym a . In young stems the sclerenchyma forms a ring,^^ but when the stems become old the ring breaks up into several sections, so that the latter are placed outward of each fibro-vascular bundle. In some old stems the sclerench3ana is also formed inside of the inner j^hloëm. In other stems a secondarily formed sclerenchyma is developed within the primary one. The secondary sclerenchymatous cells are short and thick- walled, manifesting many pore-canals ; while the primary ones are long and fibre-like. 1) E. G. O. Müller. Cucurbitaceen in Englejî und Prantl's die natürlichen Pflanzen- familien. 1894. Th. IV, Abt. 5. p. .3. COMPARATIVE ANATOMY OF THE CUCURBITACE^. 9 The old stems of some species have only one group of sclerenchymatous cells outside of each fibro- vascular bundle, as in Aciinoslemrna racemosuin, MelotJwia japonica, Schizopepon hryoniarfolius, var. japo7iicus, Citrullus vulgaris, Lagenaria vulgaris, Tricliosanthes cucumeroides, T. japonica, T. multiloba, and Gym- nostemma cissoides; some have two sclerenchyma masses tangentially arranged, which have been formed by a rupture of the original continuous ring, as in 3Iomordica Oharaniia, Luffa cylindrica, L. acidangula, Cktcumis sativus, C. Melo, Benincasa cerifera, and Gucurhita Pej)o. There is a remarkable development of secondary sclerenchyma in the very old stems of Actinostemma racemosum, 3Ielothria japonica, Momordica Charantla, Luffa cylindrica, L. acutangula, Citrullus vulgaris, Tricliosanthes cucumeroides, T. japonica, T. multiloba, and Gymnostemma cissoides. In 3Iomordica Charantla and Actinostemma racemosum it is especially noticeable, that the sclerenchyma at the angled portions has a double ar- rangement, one mass lying just outside a primary fibro-vascular bundle of the outer ring, and the other lying outside a newly formed secondary bundle'^ which is placed externally to the primary one. In the old stems of Trichosanthes cucumeroides, T. japonica, and T. multiloba the secondary sclerenchymatous cells or stone- cells are developed in great masses and form an incomplete ring around the stem. Fihro-vasculav BurifUes, The fibro-vascular bundles in the stems are arranged in two rings, the outer and the inner. Each ring contains five bundles, the members of one ring being situated alternately to those of the other. The fibro-vascular bundles of the outer ring are smaller and nearly equally-developed, but those 1) See below, 10 ART. 4. A. YASUDA I ON THE of the inner ring often gi'ow irregnlai'ly, and in some cases one or two of them are wanting. The old stems of llomordica Charantia and Actinostemma racemo^um present a cliaraeteristie arrangement of the fibro- vascular bundles. We may o])serve doable bundles overlapping one another at the angled portions; i.e., the secondary bundles grow outside of the primary ones from a secondary meristem, so that the angled portions are raised into keels. (a) Phloem. The fibro-vascular bundles are of a bicollateral type, so tliey have two kinds of phloem, tlie peripheral and the axial, of which the former is always better-developed than the latter. Sieve-tubes occur not only in the phloem but also outside of it. They are, according to Fischer,^^ chxssified under four heads : (i.) vaseular-l)undle sieve-tul)es, winch are found in the phloem; (ii.) ectocyclic sieve-tubes, between the epidermis and the sclerenchymatous ring; (iii.) entocyclic sieve-tul)es, within scleren- chymatous ring ; (iv.) commissural^sieve-tul)es, serving to connect the other kinds of sieve-tubes with one another. In reference to the mode of distribution of these four kinds of sieve-tubes we may distinguish three types in the Cucurbitaceœ : the first type has the vascular-bundle and entocyclic sieve-tubes, but no ectocyclic or commissural ones ; the second has the vascular- bundle, entocyclic and commissural sieve-tubes, but lacks the ectocyclic tubes ; the third has the vascular-bundle, ectocyclic, entocyclic, and commissural sieve-tubes. To the first type belong Luffa eylindrica and L. acutancjula ; to the second, Actinostemma racemosum, Melothria ja,ponica, Schizopepon hryoniœfolius, var, japonîcus, 3Iomordica Charantia, CitruUus vulgaris, Oiieumis sativus, C. 3felo, Benincasa, cerifera, Lageyiaria vulgaris, Trichosanthes 1) A, Fischer. Inc. dl. p. 104. COMPAKATIVE ANATOMY OF THE CUCUKBlTACEiE. 11 cucmneroides, T. japonica, T. multiloba, and Gymnostemma cis- soides ; to the tliird, Cucurbita Pepo. The size of the sieve-tubes is so large that it serves as a distinctive character to the family. The following numbers denote the diameters of the large sieve-tubes in the stems of each species; Luffa acutangula 0.0(S8 nmi. Cucurbita Pepo 0.088 „ Luffa cylimlrica 0.087 „ Trichosanthes japonica 0.085 ,, Benincasa cerifera, 0.080 ,, Lagenavia mdgaris 0.075 ,, Movbordica Gharantia 0.075 „ Trichosanthea multiloba 0.070 ,, T. c lie lime ro id es 0.070 ,, Gitrullus vidgaris 0.065 ,, Cucumis 31elo 0.005 ,, C. sativus 0.065 ,, Hchizopepon bryoïiiœfolius, var. japonicas 0.054 „ Actinodemma racemosum 0.045 ,, Gymnostemma cissoides 0.040 ,, 3Ielothria japonica 0.040 ,, (b) Xylem. Among the characteristic features of the xylem vessels the following are to be noticed. They are very well- developed and become especially noticeable as the stems advance in age, when they can be easily perceived even with the naked eye. As a rule thyloses^' are developed in old vessels. By com- paring the diameters of the large vessels in each species the following average measurements were obtained : 1] H. A. LoTAK. loc. cit. p. 30. 12 ART. 4. — A. YASUDA : ON THE Luffa acutangula 0.50 mm Lagenaria vulgaris 0.50 Cucurbita Fepo 0.48 Benincasa cerifera 0.48 Luffa cylindrica 0.46 Trichosanihes jajwnica 0.45 Moitiordica Charantia 0.4o Tricliosantlics cuctwteroidcs 0.32 Citrullus vulgaris O.ol Tricliosanihes multiloba 0.30 Cucumis Melo 0.25 (7. sativus 0.24 Schizopeprm bryoniccfolius, var. japotiicus 0.23 Gymnostemma cissoides 0.22 Actinostemma racemosuni 0.17 Melothria japo7iica 0.14 Periderm. In the very old stem.s of many species periderm is often formed, l>ut in none is it so well developed as in Tricliosanthcs cucumcroidcs, T. juponica, and T. iiiuUiloha. Tlie periderm originates from phellogen, wlncli lias been formed on the outside of the sclerenehvmatous sheath. Starch-grains and Crtjstals. Although some si)eeies contain reserve starcli-grains in the full-grown stems, yet most species first accumulate them in the old stems, where they are reserved chiefly in the medullary rays. The stems ef Gymnostemma cis- soides and Actinostemma racemosum contain reserve starch-grains from a comparatively early age. COMPARATIVE ANATOMY OF THE CUCUKBITACE^. 13 The diameters of the large starch-grains stored in the stems of many species are as follows : Gymnosiemma cissoides 0.040 mm. Blelothria japonica 0.025 ,, Trichosanihes mull'iloha 0.017 „ T. cucumeroides 0.017 „ CuGurbila Fepo O.Olö „ Luffa cylimlrica 0.015 „ L. acuta )i(jula 0.015 „ TricJiosanthes japonica 0.01 o ,, Aciinostemma racemosum 0.013 ,, CllruUus vulgaris 0.012 „ CacumU Melo 0.012 ,, 0. sativiis 0.012 ,, The stems of Schizojjepon hryouiœfolius, var. japoiiicus, Mo- mordica Charantia, Benineasa cerifera, and Lageiiarla vulgaris contain no reserve starch-grains that are appreciable. Crystals of calcium oxalate are found in the old stems of Momordica chai'aniia, Trichosanthes cucumeroides, Luffa cyUndrica, L. acuiangula, Lagenaria vulgaris, Benineasa cerifera, Cucumis sativus, 0. Melo, Schizopepon hryoniœfolius, var. japo7iicus, etc. Of these 3Ioviordica Cliarantia has an abundance of the large crystals displaying various forms belonging to the quadratic system. Hhizonies. Rhizomes occur in Melothria japonica and Gym- nosiemma cissoides. In both cases, they have scales at each node which, when completely developed, are three in number, and are to be anatomically distinguished as a shoot, a leaf, and a tendril : for, the first has ten fibro- vascular bundles; the second, five bundles with a semi-circular contour ; and the third, three or four bundles 14 ART. 1. — A. YASUDA : ON THE with a circular contour ; — all wliicli structures characterize the above organs of Melothria japonica aiicl Gyinnostenima cissoides as such respectively.^^ The rhizomes serve as a reservoir of food-materials, so that though other parts of the plants perish with the approach of winter they preserve their life till the next spring, when the young plants grow from the rudimentary scales destined to be the shoots ; thus Jleloihria japonica (PL II. Fig. 37) and Gymnostemma cissoides lead a perennial life by means of their subterranean stems. Generally speaking, the anatomical structures of the rhizomes follow the type of the terrestrial stems, but in particulars they differ widely from the latter. The collencliyma is reduced to a trace, or can be no longer distinguished as such. Jn Jlcloihria japonica a very snudl group of the delicate sclerenchymatous cells is found only on the outside of each fibi'o- vascular bundle of the outer ring, while in Gijmnosicmina cissoides the degeneration is not so great as in Melothria japonica ; the sclerenchymatous cells composed of "2-5 layers, extending outside of each libro-vascular bundle of the outer as well as of the inner rings. The libro-vascular bundles in the rhizomes show great de- generation. The sieve-tubes and the vessels are much smaller than those of the terresti'ial stems. The following are the diameters of the large sieve-tubes and vessels in the rhizomes of Meloihria japonica and Gymnostemma cissoides : 8ifvc-luli(_'. Vessel. Melothria japonica 0.032 mm. 0.07 mm. Gymnostemma cissoides 0.031 ,, 0.12 ,, ; from which we see that the degeneration is nuicli greater in Meloth'ia japonica than in Gymnostemma cissoides. 1) Compare p. 9. p. 20. and p. 33. COMPARATIVE ANATOMY OF THE CÜCURBITACE^. 15 The pnrencliy matous cells in the rhizomes are remarkahly large as compared with those of the ordinary stems on account of tlieir reserved starch-grains. On the epidermis of the rhizomes stomata and trichomes are also sometimes found. The large starch-grains found in the rhizomes are of tlie following sizes : BianiPtor. Gymnostemma cissoides 0.060 mm. 3IeIothria japonica 0.034 „ Those of the former plant (PI. II. Fig. 38) are the largest among the starch-grains contained in any organs of the Cucur- bitaceae. CHAPTER II. THE HYPOCOTYL. The germinating plantlets of the Cucurhitaceœ have long terrestrial hypocotyls or very short subterranean hypocotyls. Aclinostemma racemosum, Melothria japonica, ^Schizopepon bryo- niœfolius, var. japonicus, Jfomordica Charantia, Lniffa cylindrica, L. acutangula, Citrullus vulgaris, Cucumis sativus, C. 3Ielo, Benincasa cerifera, Lagenaria vulgaris, and Oucurbita Pepo, have the former kind ; while Trichosanthes cucumeroides, T. japonica, T. multiloba, and Gymnostemma cissoides, have the latter. Contour. As seen in a transverse section the contour of the 16 ART. 4. — A. YASÜDA : ON THE hypocotyl may be (a) roundish, (b) roundish with a central hollow, (c) square, and (d) square with a central hollow. The first is found in Actinostemnia racemosum, Melothr'ia japonica, Jfomordira Charantia, Luffa cylindrica, L. acutrmgula, Citrullus vvlgaruy Trxcliomnihe?, cucumeroule^, T. japonica, T. multiloba, and Gymnostemma cissoides ; the second, in Benincasa ccrifera, Lagenaria vulgaris, and Cucurhita Pep>o ; the tliird, in Cucumis sativus ; and the fourth, in Chicmim 31e] o. Stomata. Tlie average lunnbers of tlie stomata on the surface of the hypocotyl are as follows. They are entirely wanting in the species having the under-ground liypocotyls. Luffa cylindrica 36 per sq. mm L. acidangula 32 Citrullus vulgaris 26 Cucurbita Peijo 24 Cucumis Älelo 20 , Benincasa cerifera 19 Lagenaria vulgaris 14 Cucumis sativun 8 Actinostemma racemosum. 7 Melothria japonica 6 Mo7nordica char an fia 0 Trichosanthes cueumeroides 0 T. japonica 0 T. midtiloha 0 Oymnostemma. cissoides 0 Trichonies. Some species bear many tricliomes on the hypo- cotyl, while others have only a small num1)er of tliem or none COMPARATIVE ANATOMY OF THE CUCU RBITACE.E. 17 at all. Tlie hypocotyls of Melothria jajwnica, Cucumis sativus^ C. Jlelo, Benincasa cerifera, Lagenaria vulgaris, and Cucurbita Fcpo, are hairy ; those of Actinostemma racemosum, Momordica Cliaranlia, Luffa cylindrica, L. acutangida, and Ciirullus vulgaris, are nearly smooth ; those of Trichosardlies cucumeroides, T. jnponica, T. nndtdoha and Gyninoslcmma cissoides, are entirely naked. The trielionies on the hypocotyl may he divided into the same fonr classes as those on the stem. ('ofl( iifhfjniff. In all species the eollenchyma is developed so as tu entirely snrrouiid the hypocotyl without interruption. Sclerenelif/ma. The sclerenchyma of the hypocotyl presents a eharaeteri.stie arrangement: it forms a small mass on the outside of each fil )ro- vascular bundle, Fihro-rasculfir Btiiidles. The lunnber of the fibro-vascular l)undles in the hypocotyl is generally smaller than that in the stem. The typical nund)er is six, a median and two angular bundles^^ being symmetrically situated opposite each cotyledon. Exceptions are however found in Ciirullus vulgaris and Cucurbita Pepo, the former of which has twelve bundles, while the latter has ten. ]) H. A. LoTAR. he. cit. p. 14. 18 ART. 4. — A. YASUDA : ON THE GHAPTEE III. THE BLADE. TJpidernifs, AVhcn viewed from tlie surface the outlines of tlie epidermal cells of the blades are straight or wavy : the former is the case on the upper surface and the latter on the lower, except in Schizopepon bryoniccfoHus, var. japonicus, 3Iomordica Charantia, Lagenaria vulgaris, Trichosanthes japonica, T. multiloba and Gymnodemma rissoides, where the walls of the epidermal cells on the upper surface are also wavy. The size of epidermal cells of the blades is greatest in Gymnoüemma cissoides, the average diameter reaching 0.O7 mm., while in those of the remaining species it is 0.0.3-0.04 mm. The epidermal cells on the upper surfiice of the blade of Tricliomnthes cucumeroides are very characteristic ; they are raised into conical papillae pointing outwards (PL III. Fig. 39). In certain places on the under surfice of the blade of CunirbUa Pepo the epidermal cells are two or three-layered, and these places consequently assume a somewhat etiolated appearance (PI. III. Fig. 42). The epidermis on the under-side of the blade of Jlojnordica Charantia is characterized l)y having many enlarged cells, each f which contains a globular cystolith" (PI. ill. Fig. 40). JIo//wrdica Cliarnidia is the only species a.mong the plants we have studied, which contains cystoliths. They are fixed to a lateral w;dl ; iKjt, as we see in Fica.^, Jlorus and Zelkowa, to the outer wall. The epidermal cells which contain the cystoliths are () 1; U. Pexzio. 'Joe. cil. p. 39;".. COMPARATIVE ANATOMY OF THE CUCURBITACE^. 19 united into p;ronps, radiating from a common centre, a.s the surface view of the epidermis will clearly show. Each group of these enlarged epidermal cells consists of two or more cells with the cvs- toliths arranged head to head, as many as eleven of these having been counted. The diameter of a large cystolith is O.Oü.") mm. Sfomafa. The number of stomata on tlie upjier and lower surfaces of the blade differs remarkal)ly in each species. The following are the average numbers of stomata found on the U2)])er surface : Cucumis saiivus a jMo Gucurbiia Pepo Benincasa ccrifera CltruUus vulgaris Lagenaria vulgains Lujfa cylitich'ica L. acutangula Trlchosanthes cucumeroides Actinodeniiiia racemosum Meloihria japon'wa Momordica Ch aran tia Trlchosanthes japon lea T. muUlloha. Sehizopepon bryonuefoUvs, var japon lens (ri/m ri osieînma clsso ides As can be seen from the above table Jfoinordlca Charanlln 160 per sq. nmr 156 J5 146 5) 112 5» 100 »> 84 J> 76 >> 72 >J 66 »> 51) J' 9 >» ot >» ot > J ot J> 0 V 0 • » t Stomata rarely occur. 20 ART. 1. — A. YASUDA : OX THE rarely has stomata on the upper surface of the blade ; Trichosan- ihe>< japonica and T. multiloba have them ; still more rarely, in the neighbourhood of the veins ; and Schizopcpon bri/oniœf alius, var. japonicus and Gymnostemma cissoides have none at all. In all species the under-surface of the blades has greater number of stomata. The following show the average numbers of them on the under-surface : Cucurbita Pepo Benincam cerlfera Cucumis JL'lo C. sali vus Trichosanthes cucumeroides Lage n a) • in v ulg avis Citr u II u s vu lija ris Trieb osanthes japoiiica T. multiloba Luffa cylindriea L. acutangula 3Iomordica Charantia Schizopepon hryoni07iicus 200 ,, Melothria japonica 172 ,, Actinosteïiinm ractmosuin 1()S ,, Gymnosteimna cissoides \)1 „ Trirhoiius. The trichomes on the blades follow the types of those on the stems. Their shapes are (]uite the same on the up})er nnd lower surfaces. Their number on the lower surface is generally greater than that on the upper surface, except in Jlelothria japonica, /Schizopepon bryoniœfolius, var. japonicus, Jlomordica 347 per sq. nun 344 831 324 310 2DS 297 288 27() 221 220 211 COMPARATIVE ANATOMY OF THE CUCURBITACE^E. 21 Charantia, Trichosanihes japonica, T. mitUiloha aiul Gymnoslemma rUsoides, in wliieh tlie trichomes on the iiiulcr-suifacc are borne only uj)on the veins. Actinostemma racemosum lias them only upon the veins on both the upper and lower surfaces. The following- are the average numbers of the trichomes on the upper surface of the blades, omitting those species in which the trichomes grow only upon the veins : Lagenaria ndyaris Oiicurbita Pepo Trichosanihes cucumeroifhs Liijf'a cylimh'ica Jlcloihrla japonica Oiicuinis sativus Benincasa cerifera Cacumu Jlelo Luffa acutangula Trichonan the s japon ica T. multiloba Schizopepon hryoniœfolius, rar japonicus Citru litis vulgaris Jlovi.ordica C 'Ii a i •« n t la Ggjnnostcin ma cissoides The average numbers of the trichomes on the under-surface are as follows, again omitting the species in which they are borne only upon the veins : Cucurbita Pepo 80 per s(p mm. Lagenaria vulgaris 52 ,, Trichosanihes cucumeroides 52 ,, Beyiincasa cerifera 28 „ 36 per sq. mm. 1^0 m 13 12 12 12 10 9 2 2 2 2 18 in Ö sq. mm. 4 J? 22 ART. 4. A. YASUDA I ON THE Cucumis Jlelo 19 per sq. mm. Luffa cylindrlca 16 „ L. acuiangula 10 ,, Gucmnis saiivus 9 ,, CUridhcs vulgaris 4 ,, Some of the tricliomes are soft, while othei-s are stilf, sihcify- iiig at the base centrifugally over the epidermal surface.^^ The trichomes on the upper surface of the blade of Jleloihria japonica, and on both surfaces of the blades of Luffa cylindrica, L. acuian- gula, Ciirullus vulgaris, Cucumis saiivus and C. Jlelo, are good examples of the latter kind. Thickness of Blades. The thickness of blades differs some- wliat in each species. The average thickness of full-grown leaves is as follows : Liffa acutangula 0.27 mm. L. cylindrica 0.25 ,, Cucumis Jlelo 0.21 ,, Benincasa cerifera 0.20 ,, Cucumis saiivus 0.20 ,, Jlomordica, Cliarantia 0.18 ,, Ciirullus vulgaris 0.18 ,, Trichosaiithes cucumcroidcs 0.18 ,, T. japonica 0.18 ,, J\ multiloba 0.17 „ Lagenaria vulgaris 0.16 ,, Gymnosiemma cissoides 0.16 ,, Aclinosiemma racemosum 0.14 ,, 1) A. DE Eary. Cutiiparative Aniitoiny of llic rhaiieroj;aiiis and Ferns. Oxford 1884 p. K'îî. COMPARATIVE ANATOMY OF THE ('UCURP.ITACE^.. 23 Melothria japonica 0.1 tin m . Cucurbita Pepo 0.13 ,, Schizopepon hryonkefolius, var. japonicus 0.12 „ PallisfuJe rareuehf/mf(. The thickness of tlie pallisade parenchyma is not uniform in all individuals. The results of microscopical measurements of it are given in the following table: Luffa acutangida 0 120 mm. X. cylindrica 0.095 ,, Gucurhila Pepo 0.085 ,, Cucumis Melo 0.075 ,, Benincasa cerifcra 0.070 ,, Cucumis sativus 0.070 ,, Lagenaria vulgaris 0.063 ,, Melothria japonica. 0.0G2 ,, Jlomordica Charantia 0.0G2 ,, Ciirullus vulgaris 0.060 „ Trichosanthes cucumcroides 0.055 ,, Ggmuostemma cissoides 0.055 ,, TricJiosanthes japonica 0.052 ,, T. mvUiloha 0.050 „ Actinostemma racemosum 0.050 ,, Schizopepon bryonicefolius, var. japonicus 0.050 ,, The pallisade cells mostly occupy three- or four-tenths of the thickness of the blades, and sometimes even exceed six-tenths in Cucurbita Pepo, where the pallisade cells are at times arranged in double rows. The following will show these j'elations more clearly : 24 ART. 1. — A. yasuda: ox the Ckcurhita Pepo 0.085/0.13=0.65 Luffa acuta7igula 0.120/0.27=0.44 Äleloihria japon ica 0. 002/0, 14=0.44 Shizopepoîi hryoiiv if alius, var. japon icux 0.050/0. 12 = 0. 42 Lagenaria vulgaris 0.063/0.1 G = 0.o9 Luffa rglimhirn 0.095/0.25 = 0.38 Cucumis Jfc/o 0.075/0.21=0.36 Actinosiemma racemosuui 0.050/0.14=0.36 Cucumis sativus 0.070/0.20 - 0.35 Benincasa cerifcra 0.070/0.20 = 0.35 Momordica Charaniia 0.062/0.18 =0.34 (rymnosleuumi cissoixJcs 0.55/0.10 = 0.34 Citrullus Tu/garis 0.60/0.18 = 0.33 Trichosanlhes cucumcroidcs 0.55/0.18 =0.31 T. mvltiloha 0. 50/0. 1 7 = 0. 29 T. japonica 0.52/0.18 = 0.29 A« to tlie brendtli of the pallisiide cells .^ome have a large diameter, as in Aclinosteut/ua racemosuui and Ggmnosicuima cissoides, in wliieli tlie breadth is neaily equal to the length (PI. Til. Fig. 41), while others have a comparative] y small diameter, those of Cucurbita Pepo, Lujfa cylindricu, and L. aeutaiigula being remark- al)ly long and narrow. SjmKffi/ Pitrenrhfntui. Tlie spongy parenchyma consists of 2-6-layered cells. The spongy cells of Ggmnostemma cissoides are characterized by being only 2-3-layered and by being far larger than those of other species. Sfarch-ffraiiis. The chloroplasts in the mesophyll of Actino- COMPARATIVE ANATOMY OF THE CUCURBIT ACE Jî. 25 slemma racemosum Ilelothria japonica and Gymnoüemma cissoides contain large spindle-shawled starcli-grains as a product of assimila- tion (PI. III. Fig. 41). 3Iid-rihs. The median line of the upper siirflice of the mid- ribs is always raised into a ridge, ^Yhere the collenchyma is well-developed, and of the species of the Cucurbitaceœ under consideration, the one having the highest and sharpest rido-e is Gymnostemma cissoides. The fibro-vascular bundles of the mid-ribs near tlie pi'oximal portion are arranged in six ways : the first type has a sinole bundle situated at the centre of the mid-rib (PI. III. Fio-. 43); the second has two bundles one above the other (PL III. Fi«-. 44) ; the third has three bundles, a larger one at the centre, tlie other smaller two on each side and above (PL III. Fig. 45) ; the fourth also has three bundles, but they are arranged in a straight line drawn from above downwards, the uppermost l)undle being the smallest, while the lowest is the largest (PL III. Fig. 46) ; the fifth has four bundles, and is distinguished from the third type by having one more small bundle at the upper part of the central one (PL III. Fig. 47) ; the sixth has seven bundles, the largest bundle being undermost, and the other smaller six on each side and above (PL III. Fig. 48). Actlnostemma racemosunn and Schizopepon hryoniœfolius, var. japonicus belong to the first ; Melothria japonica, to the second ; Gymmosteinma cissoides, to the third ; Cucumis saiivus, C. Melo, Benincasa cevifera, Lagenaria vulgaris, to the fourth ; 3Iomordica Cfiarantia, Luffa cylindrica, L. acutangula, Trichosanthes cucumeroides, T. japonica and T. mulliloha, to the fifth ; CHrulltis rulgaris and Oucurbita Pepo, to the sixth, 26 ART. 4. — A. YASUDA : ON THE In tracing the fibro- vascular bundlles of the mid-ribs from base to apex we find that the first type retains the single bundle to the last ; the second loses the smaller upper bundle on the way, and is reduced to a single bundle ; the third is also reduced to one bundle near the apex, the associated bundles on each side being out of sight ; the fourth loses the uppermost bundle first, then the middle one, the lowest bundle remaining to the end ; in the fifth, the lateral bundles first disappear and then the upper- most l)undle, the lowest one being the survivor; in the sixth the lateral bundles disappear from above downwards till the lowest bundle remains isolated, Collenchynia. The position of the collenchyma developed in the mid-ribs varies somewhat according to the above six types : in the first, second, third, fourth, and fifth it is developed only at the upper and lower nmrgins, but in the sixth, liesides the preceding places, collenchyma is also developed on both sides of the mid-ribs. GHAPTEB IV. THE PETIOLE. Contour. The contour of the petioles, as seen in a transverse section, is not of a constant shape in tlie proximal, middle, and distal portions even in one and the same species. Generally COMPARATIVE ANATOMY OF THE CUCURB1TACEJ<:. 27 speaking, a transverse section cut through the distal portion shows a deep groove on the upper surface, while that of the proximal portion appears roundish, the groove becoming extremely shallow. The middle portion has an intermediate character. Besides, though many of the petioles are solid some of them have a large central hollow throughout, this being the case in Benincasa cerifera, Cucurbita Fepo (PL I. Fig. 15-17), and Lujfa acutangula. Stoinafa. Stomata are not found on the epidermis, where the collenchyma is developed ; they are present only in the intercollenchymatous portions. The following are the average nund^ers of the stomata on the petioles of eacli species : (Jucurbita Pepo 12 per sq. mm Ti'ichosan thes multiloba 12 3> T. cucumeroidcs 11 >» Momordica Charantia 11 M Äctinostenwia racemosum 11 >) Cucumis, Melo 11 >> Citrullus vulgaris 10 >J Lagenaria vulgaris 10 )> Benincasa cerifera 8 JJ Cucumis sativus 8 J> Trick osanth es japon ica 7 >J Luffa cylindrica 7 » L. acutangula 7 i> Gymnostemma cissoides 6 >» Jlelothria japonica 4 J> Schizopepon bryoniwfolius, var. japonicus 3 i> 28 AKT. 4. A. YASUDA : ON THE Ti'iclionics. The petioles may bear several kinds of trichomes or tliey may be nearly smooth. Momordica Cliarantia, Luffa cylindrica, L. acutangula, CUridlus vulgaris, Cucumis sativus, C. 3ïelo, Benincasa cerifera, Lagenaria vulgaris, Trichosanihes cucumeroides and Ciwurhiia Pepo, belong to the first group ; Aciinosiemma racemosum, Jleloihria japonica, Scliizopcpon hnjo- niœfolius, var. japonicus, Trichosanihes japojiica, T. mulliloba and Oymnostemma cissoidcs, to the second. Collenchy^na. In the middle portion of the jietioles the collenchyma is developed outside of each fibro-vascular bundle. The number of the collenchyma groups varies from five to thirteen, the extremes being represented respectively by Gymno- sLemma cissoidcs and Cucurbita Pepo. In the distal as well as the proximal portions of the petioles several isolated collenchyma groups are united with one another, and in an extreme case all the groups of the collenchyma are entirely fused together. Ht'lereitc/i!fnn(. On the outside of each fibro-vascular bundle several-layered sclerenchymatous cells are formed, which are particularly well-developed in Lujfa cylindrica and L. acutangida. Fibvo-vascular J iu miles. The fil)ro-Viiscular bundles of the petioles have a circular arrangement. They are always in pairs laterally, while the undermost bundle is unpaired. Sometimes there appears a small bundle in the fundamental tissue under the groove. The undermost unpaired l)undle is the largest of all and towards the upper portion of the petioles the jiaired bundles gradually decrease in size. The uppermost pair, when the upper surface of the petioles is deeply grooved, enters into COMPARATIVE ANAT03IY OF THE CÜCUKBITACE^. 29 tlie ridges raised on both sides of the groove (PL I. Fig. 12-18). The number of the fil)ro-vascuhir bundles varies from five to thirteen. Aclinostemma racemo8iun, Jleloihria japoniea and. Gym- nostemma cissoides have five bundles ; Schizopepon bryonicefolius, var. japonicus, seven ; Jlomordica Cliaraiitia,^^ Cihmllus vulgarisy^^ Cucumis saiivus, C. JTelo,^^ Lagenaria vulgaris, Trichosanthes cuGumeroidcs, T. japoniea and T. multiloba, nine; Luffa cylin- drica, L. acutangula and. Benincasa cerifcra, eleven ; Cucurbila Pepo, thirteen. CHAPTEE V. THE COTYLEDON. The cotyledons are above the ground or underneath it. Those of TrichosaniJics cucurneroidcs, T, japoniea, T. vndliloba, and Gymnoslcmni.a cissoides, are entirely buried in the soil. The anatomical structure of the cotyledons differs from that of the foliage-leaves. Epidermis. When observed in a surface view the contour of the epidermal cells of the cotyledons is either straight or wavy as in the case of the foliage-leaves. On the upper surface 1) H. A. LoTAK. loc. clt. p. 103. 2) Ditto, p. 95. o) Ditto, p. 96. 30 ART. 4. — A. YASUDA : ON THE of the cotyledons of Momord'wa Charani'ia, Luffa cylindrica, L. acidaiifjula, Citrullus vulgaris, Cucuîuis saltvus, C Melo, Tricho- santhes cucunieroides, T. japonica, T. muUiloba, Cucurbita Pepo, and Gymnodemma cissoides, as well as on the under surface of those of Blomordica Charantia, TriGhosanthcs cucitmeroides, T. japonica, T. muliilolm and Gymnodemma cissoides, the walls of the epidermal cells are straight ; while they are wavy on the upper surface of those of Actinosiemma racemosum, 3Ielothria japonica, Benincasa cerifera and Lagenaria vulgaris, as well as on the under surface of those of Actinostemnia racemosum, Melothria japonica, Luffa cylindrica, L. aculangula, Citrullus vulgaris, Cucumis sativus, C. 3Iclo, Benincasa cerifera, Ljagenaria vulgaris and Cucurbita Pepo. Stomafif. In nearly all of the species, as far as we have seen, stomata are found on both the upper and under surfaces of the cotyledons. Exceptions are however to be found in some species, where they are either almost exclusively on the upper side or only on the under. For example, in 3louiordica Cliarantia, Trichosanthes japonica and 2\ multiloba, they are cliiefly on the upper, very rarely on the lower surface, while in Gymnostemvia. cissoides they are exclusively on the lower. The numbers of the stomata found on the upper surface of the cotyledons of each species average as follows to a square millimetre : Citrullus vulgaris 270 Luffa cylindrica 268 L. acutangula 260 Cucurbita Pepo 224 Lagenaria vulgaris 164 COMPARATIVE ANATOMY OF THE CUCURBITACE^. 31 Cucumis Meto 142 Benincasa cerifera 125 Aclinoste^nma racem osum 115 Cucumis sativus 93 llelothria japonica 43 Mœnordica CJi arantia 40 Trichosan Ikes c ucumer aides 28 T. japonica 10 T. multUoba 4 Gymnoslemma cissoides 0 Tliose on the under surface : Läiffa cylindrica 3G8 per sq. mm. L. acutangula 3G0 5> Cucumis 3Ielo 314 >> C. sativus 270 >? Cucurbita Fepo 253 JJ Citrullus vulgaris 228 >J Benincasa cerifera 177 11 Lagenaria vulgaris 153 J) Melothria japonica 135 )> Actinostemma racemosum 118 J> Gymnostemma cissoides 15 >) Trichosan thes cucumeroides 11 5> Ilomordica Charantia ot »> Trichosanthes japonica ot » T. multiloba Ot J> Trichontes. Some cot3dedons are furnished with trichomes on both the upper and the lower surfaces ; others only on the t Stomal a rarely occur. 32 AET. 4. — A. YASUDA : ON THE upper, and yet others have none on either surface. Thus in Luffa cylindrica, L. acutangula, Cucumis sativus, C. Melo, Beniiicasa cerifera, Lagenaria vulgaris and Cucurbita Pepo, they are present on both surfaces ; in Melothria japonica and Citrullus vulgaris, only on the upper ; in Aethiostemma raceynomm, Mo- moî'dica Charantia, Trichosanthes cucumeroicles, T. japonica, T. muliiloha and Gymnostemma cissoides, they are entirel}^ absent. Pallisade JPat'eiichf/ma. The pallisade pareneliyma of the cotyledons is generally many-layered ; but there are some excep- tional cases, in which the pallisade parenchyma can be scarcely recognised. The commonest type of the pallisade parenchyma consists of two* or three-layered cells : e.g. Actinostemma racemosum Melothria japonica, Luffa cylindrica, L. acutangida, Citrullus vulgaris, Cucumis sativus, C. Melo, Benincasa cerifera, Lagenaria vulgaris, and Cucurbita Pepo. In 3Iomordica Charantia, Tricho- santhes cuGumeroides, T. japonica, T. multiloba and Gymnostemma cissoides the pallisade cells are indistinct. Spongy Parenchyma. The layers of the spongy parenchyma- cells of the cotyledons are much more numerous than those of the foliage-leaves. A very peculiar structure of the spongy parenchyma appears in the cotyledons of Actinostemma racemosum where it pre- sents a reticulate appearance, the cells radiating from a fibre- vascular bundle and thus leaving very large intercellular spaces among them, a character which reminds us of the inner structures of water- plants (PI. IV. Fig. 49). The spongy parenchyma- cells of the cotyledons of Momordica Charantia, Trichosanthes cucumeroides, T. japonica, T. multiloba and Gymnostemma cissoides, are not well difierentiated, and possess no marked intercellular spaces. COMPARATIVE ANATOMY OF THE CUCURBITACE^. 33 CHAPTEE VI. THE TENDRIL. Contour, The contour of a transverse section through the proximal portion of a tendril is nearly circular. In the more remote portions it takes a more or less polygonal shape, and at the distal portion it becomes flattened. Especially in Luffa acutangula and Benincasa cerifera the surface of the main axis of the tendrils is elevated into noticeahle longitudinal ridges. In Benincasa cerifera (PI. IV. Fig. 50), Cucvrhata Pe^^o, Luffa aciUangida and Schizopeiwn hryoniœfolius, var. japonicus a central hollow is formed in the main axis of the tendrils. Stoniata. The average numbers of the stomata on the tendrils are not widely different in each species, as may be seen from the following table: Triclwsanthes cucumeroides T. japonica T. multiloba Gymnostemma cissoides Actinostenima racemoswn Melothria japonica Luffa cylindrica L. aeiitangida Benincasa cerifera Momordica Charaiitia Oitrullus 'vulgaris 24 per sq. mm. 24 24 '24 20 20 20 20 20 16 16 34 ART. 4. A. YASUDA : ON THE Oucumis saiivus 16 per sq. mm a Melo 16 )> Lagenaria vulgaris 16 j> Cucurbita Pepo 16 J) Schizopejwn bryonicefolius, var. japonicus 16 >j Collenchytiia. At the proximal portion of the tendrils we may distinguish three varieties of coUenchyma according as it is developed : (i.) on all sides of the tendrils, (ii.) on the outside of each fibro-vascular bundle, and (iii.) outside of each flbro- vascular bundle as well as of each medullary ray. The first variety is found in 3Ieloihria japonica, Oucumis saiivus, 0. Melo, Triohosanihes cucumeroides, T. japonica, T. multiloha, Gymnostemma cissoides ; the second, in Actlnostemma racemosum, Schizopepon hrgoniœfoVras, var. japonicus, 3Io7nordica Charantia^ Luffa cyUndrica, L. acutangula, CUrullus vulgaris and Lagenaria vulgaris ; the third, in Benincasa cerifera and Cucurbita Pepo. In the distal portion, the collenchyma masses on the under side of the tendrils fuse into one another, Sclerenchyma, At the proximal portion of the tendrils the sclerenchyraa forms a continuous ring.'^ At the distal portion, however, it is developed only on the ventral side of the tendrils.^^ In Benincasa cerifera and Lagenaria vulgaris the sclerenchyma masses occur separately outside of each fibro-vascular bundle near the base of the tendrils, and those masses which are on the ventral side of the tendrils at the distal portion towards their apex, are united. 1) H. A. LoTAR. loc. cit. p. 139. 2) Ditto, p. 138. COMPARATIVE ANATOMY OF THE CUCURBIT ACE^:. 3o Fibro-vasculav BiituUes. At the proximal portion of the tendrils the number of the fibro-vascular bundles varies from four to ten, gradually decreasing towards the distal portion. The maximum number of the bundles occurs in Luffa cylindrica, while the minimum is found in Actinostemma racemosum and Gymnostemma cissoides. Luffa acutangula, Gltrullus vulgaris, Benin- casa cerifera, Momordica charantia, Cucurbita Pepo, Cucumis 3Ielo, C. sativus, Lagenaria vulgaris, Tricliosanthes multiloha, T. cucume- roides, T. japonica, Schizopepon bryoniœfoUus, var. japonicus and Melothria japonica have numbers intermediate between the ex- tremes in a descending order. CPIAPTEB VII. THE ROOT. Besides ordinary roots, tubers are met with in some species. They are, however, limited to one genus only, Trichosanthes which thus leads a perennial life, just as Melothria japmnica and Gymno- stemma cissoides preserve their vitality through the winter by means of rhizomes. I shall now first take up the investigation of the anatomical structure of the ordinary roots, and then that of the tubers. Contoui'. The contour of the root is not diversified like 36 ART. 4. A. YASUDA : ON THE that of tlie stem. It is mostly roundish, and in some cases it is more or less triangular or square, according to the number of fibro-vasculer bundles present. Perklet'rn. Although the formation of periderm is common in old roots yet none have it so well-developed as that observed in the roots of Tr'ichosniilhcs cucwneroldes, T. japonica, and T. muUiloba. Sclerenchyma. When the roots are young, sclerenchyma is developed on the outside of each fibro-vascular bundle, but when they become old it breaks up into several parts, often form- ing two or three tangential parallel rows. When still further advanced in age numerous masses of sclerenchyma are found scattered in the cortex outside of each bundle as well as of each medullary ray. The degree of development of the sclerenchyma varies according to the species : some have well-developed and numerous sclerenchyma groups, while in others they are delicate and reduced. In Melothr'ia japonica, Irichosanthes cucumeroides^ and Gymnosteinma clssokles the w^ell-developed sclerenchyma forms a nearly continuous ring, but in Schizopepon bryoniœfoUus, var. japonlcuSy Benincasa cerifera, and Cucurbiia Fcpo it is greatly reduced. At the angled portions of old roots of Momordica CJiarant'ia the sclerenchyma shows a charateristic double arrange- ment like that seen in the old stem, a mass of it being developed outside of each of the double fibro-vascular bundles heaped one upon the other. Fibro-vascular BumUes. Most of the fibro-vascular bundles COMPARATIVE ANATOMY OF THE CUCURBITACEiE. 37 in the main roots are tetrarcli.'' In the lateral roots, however, di-, tri- or polyareh bundles are found especially in the adventitious roots : thus, for examj)le, the pentarch occur in those of Actino- stemma racemosum, Jlelothria japonica, Lvffa cylindrica, L. acutangula and Benincasa cerifera ; the hexarch, in those of Trichosanthes cucumeroides, T. japonica, T. multiloba, and Gymno- stemma cissoides ; the octarch, in those of Cucurbita Pepo.'^ In old roots of Momordica Cliarantia the fibro-vascular bundles are arranged in double rows at the angled portions as in the old stem, one small secondary bundle, or occasionally two, being newly formed outside of the primary one. (a) Phloem. The sieve-tubes found in the roots are of smaller sizes than those of the stems. The diameters of the large sieve- tubes under consideration are shown in the following table : Lagenaria vulgaris 0.075 mm. Beniticasa cerifera 0.070 ,, Momordica Charantia 00.70 ,, Luffa cylindrica 0.068 ,, L. acutangula 0.0G7 ,, Cucurbita Pepo 0.067 ,, Trichosanthes japonica 0.060 „ T. multiloba 0.0Ö8 ,, Citrullus vulgaris 0.057 ,, Trichosanthes cucumeroides 0.055 ,, Cucumis sativus 0.055 ,, a 3Ielo 0.054 „ Schizopepon bryoniœfoliuSy var. japonicus 0.0-10 ,, 1) 11. A. LoïAR. loc. cU. p. 155. 2) Ditto, p. 170. 38 ART. 4. A. YASUDA : ON THE Äctinostemma racemomm 0.040 mm. Melothria japonica 0.030 ,, Gymnostemma cissoides 0.025 ,, (b) Xylem. The xylem in the roots is well-developed and occupies the greater part of them. The sizes of the vessels in the roots are somewhat smaller than those of the stems. Ben'uicasa cerifera iJiuiuelir of Die large vessel. 0.50 mm. Luffa cylindrica 0.44 L. acutangida 0.43 Lagenarm vidgaris 0.43 Cucurbita Fepo 0.35 3Iomo) 'd ica C/ia ra n I ia 0.33 OitruUus vulgaris 0.29 Trichosanthes )ii ultiloha, 0.28 T. cucumeroides 0.25 T. japonica 0.24 Cucumis 3Ielo 0.22 C. sativus 0.20 Schizopepon bryoniic/oUus, rar. japonicus 0.17 Äctinostemma racemos am 0.15 Gymnostemma c isso ides 0.14 Jlelothria japonica 0.09 Thyloses^^ within the vessels occur most commonly in old roots as they do in tlie old stems. Inter cellular Spaces. The intercellular spaces in the cortex of young roots are numerous and large. Those found in the roots of Äctinostemma racemosum are exceptionally large, a trans- 1) H. A. LoTAB. loc. cil. p. 132. COMPARATIVE ANATOMY OF THE CUCURBITACE^. 39 verse section presenting an appearance like that of a water-plant (PL IV. Fig. do). This peculiar character of the root of Adino- stemma racemosum is not to be wondered at, when we consider that this plant always grows in the neighbourhood of a stream or in moist places, it having evidently acquired the character of a water- plant in order to adapt itself to its surroundings. Starch'fjrains mid Crystals, Reserve starch-grains of appreciable size are contained in old roots of the following species : Diaiiic'tor of tlie large starch-gruin.-i. Gymnostemma cissoides 0.023 mm. Trichosanthes multlloha 0.020 ,, T. japonica 0.020 „ T. cuGumeroides 0.018 ,, 3Ielothria japonica 0.017 ,, Luffa cyllndrica O.OIG „ L. acutangula 0.014 ,, Actinostemma racemosum 0.010 „ Citimllus vulgaris 0.009 „ Cucurbita Pepo 0.008 „ Crystals of calcium oxalate are met with in the old roots of Momordica Cliarantia, Luffa cylindrica, L. acutangula, Lagenaria vulgaris, Benincasa ceriferaj Cucumis sativus, C. Melo, &c., those of Momordica Cliarantia being the most remarkable in size and quantity. Tubers. Tubers are found only in Trichosanthes cucumeroides, T. japonica and T. multiloba. They occur in simple or grouped forms and attain an enormous size. A transverse section of a tuber shows a great degeneration of fibro-vascular bundles, the paren- chyma itself being much increased so as to serve as a reserve tissue. 40 AKT. 4. — A. YASÜDA I ON THE The periderm is well-developed and forms a thick layer around the tuhers. The sclerenchjnna of the tubers is best developed in Tricho- santhes cucumeroides, where it forms an almost continuous ring. In Trichosanthes mnltlloba the development of the sclerenchyma is somewhat checked and in Trichosanthes japonica it is very much retarded. The size of the sclerenchymatous cells and the thickness of their walls are much greater in the first two species than in the last. For the sake of a comparison I give the follow- ing numbers obtained from the measurements of the large scleren- chymatous cells in the tubers of the above three species : nhimeter. ïliickness of the wall. Trichosanthes multiloba 0.080 mm. 0.020 mm. T cucumeroides 0.050 ,, 0.020 „ T. japo7iica 0.030 „ 0.008 „ The fibro-vascular bundles are most reduced in Trichosanthes cucumeroides, where many of the vessels are separated from each other. The following are the diameters of the large sieve-tubes in each species : Trichosanthes multiloba 0.070 mm. T. japonica 0.050 „ T. cucumeroides 0.045 „ Diameters of the large vessels : Trichosanthes multiloba 0.42 mm. T. japonica 0.35 ,, T. c2ic2imeroides 0.30 ,, The starch-grains contained in the tubers attain their maximum size in Trichosanthes multiloba, those of Trichosanthes japonica coming next, and then those of Trichosanthes cucumeroides. In point of size the starch-grains from the tubers of Trichosanthes multiloba, however, stand second among those contained in the fOMPAUATIVE ANATOMY OF THE CUCURBITACE/E. 41 various organs of the Cucurbit ace?e, the first place 1)eing held hj those from the rhizome of Gymnosteiiuiia clsso'ides.^- The diameters of the large starch-grains from the tubers of Trichosanthes are as follows : Trieliosrfnthes m ^lUUohn 0.042 mm T. japon tea 0.027 „ T. citcinneroides 0.018 „ CHAPTER YIII. THE FRUIT. The surface of tlie fruit of the Cucurbitaceœ is mostly smooth, except in Actinostemma vacemosum, Momordicn Chamntia and Cucumis satiüus, where it is raised into tubercles. The nature of the tubercles varies, for an examination of their anatomical structure shows that tlios3 of Actinostemma racemosum (PI. lY. Fig. f58) and Jlomonlica Cliarantia are only parenchymatous outgrowths, while those of Cucumis saiivus are the large pro- tuberant bases of trichomes, on which the latter have formerlv rested (PI. IV. Fig. Ö7). Epidermis. The epidermal cells of the fruit have several distinct characters in the different species. To begin with, the shapes of the epidermal cells may be (a) radially flattened, ())) 1) See p. 15. 42 ART. 4. A. YASUDA : ON THE cubical, or (c) radially elongated. To the first belong* Aciinostemma racemosum, Schizopepon bryonîœfoUiis, var. japonicus, Jlomordica Qiarantia and Gymnostemma cissoides ; to the second, MeloÜiria japonica, Luffa cylindrica, L. acutangula, Citrullus vulgaris, Lagenaria vulgaris, Trichosanthes cucumeroides, T. japonica and T. multiloba ; to the third, Cucumis sativus, C. JTelo, Benincasa ccr'tfera 'and Cucurhita Pepo, the epidermal cells of Cucumis sativus beins: the lon^'est of all the radial diameter exceeding four times the tangential (PJ. V. Fig. 60) ; those of Benincasa cerifera are also characteristic, many of the cells having a tangential septum (PI. V. Fig. 61). A superficial view of the epidermal cells shows some variation n size, their diameters measuring 0.006-0.055 mm. Actinostemina racemosum, j\[omordica Charantia, Gymnostemma cissoides and Citrullus vulgaris, for example, being of the maximum size, while Cucurhita F-f issue ivifhiii the JEj>fdevniis. The structure of the pericarp-tissue Avitliin the epidermis is not uniform in all species. Some fruits have a hardened ring in the outer tissue of the pericarp (PL V. Fig. 61), while in others this is lacking (PL V. Fig. 60). In Aclinoslemma raceniosum, Blelothria ja]}onica, Luffa cyUndrica, L. acutangula, G'drullus vulgaris, Benincasa cerifera, Lagenaria vulgaris, Trichosanihes cucumeroides, T. japo- nica, T. multiloba and Gyumoslemnia cissoides the hardened ring is developed, that of Aclinoste/inna racemoswii and Melothria japonica being incomplete and interrupted in man}' places (PL IV. Fig. 58) ; in ScMzopcpon, hrgoniœfolius, var. japonicns, Mo/nor- dica Charantia, Gacumis sativus, C. Mclo and (Jacurhita Fepo the mechanical sheath is entirely wanting. The sclerenchvmatous cells constitutinii the hardened rin«; may be thin-walled with numerous pits, as in AciinosU'inuia racciiiosuni, Melothria japonica and Benincasa cerifera, or may be thick-walled with distinct })Ore-canals, as in Liifa cyUndrica, L. acntangala, (Jitrullus ralgaris, Lagenaria vulgaris, Tricliosanthes cucwneroides, T. japonica, T. iitultiloha and Gy)nnoslemma cissoides. The hypodermal cells lying outside of the hardened ring are rather smaller than those of the deeper fundamental tissue, and when no hardened ring is present the parenchymatous cells of the outer part of the pericarp under the epidermis act as such. In Lagenaria vulgaris and Benincasa cerifera the hypodermal 1) A DE BakY. lac. (il. \). 8(1. . COMPARATIVE ANATOMY OF THE CUCUFtBITACEJ!:. 4-5 cells are remarkably small, but in Actinostemma racemosum and Momordica Cliarantia they are exceptionally large. The number of the layers of the hypodermal cells outside of the hardened ring should also be noticed. While usually they are many-layered in Jlelothria japonica and Gijmnoüemma cissoidcs they are reduced to the minimum ; that is, in the former tlie hypodermal cells are one- or two-layered, l)ut in tlie latter only one-layered. Ffhro'Vasculnr Bun files. As a rule the fibro- vascular bundles in the fruit are weakly developed, and most of them have no sclerenchymatous sheath. Exceptions are found, however, in the fruit of Luffa cylindrlca and L. acutancjuln. The fibres con- stituting the net-work of the X^(^«-fruit are really the fil^ro- vascular bundles surrounded by the thick layers of the scleren- chymatous cells, which give a roundish shape to the fibres, and make them extremely elastic, the circumscribed bundles themselves remaining merely rudimentary (PI. V. Fig. 59). A peculiar arrangement of the fibro-vascular bundles is observed in the tubercles on the surface of the fruit of Momordica Gharantia. A bundle enters into the tubercle perpendicular to the surface of the fruit, briinching out near the top of the tubercle (PI. IV. Fig. 5(j). The sieve-tubes show a characteristic distribution in the fruit- tissue. Besides those found in the phloem, there occur isolated sieve-tubes in the tissue of the pericarp within the epidermis (PI. V. Fig. 61). In the species which have the hardened ring such as Benincasa cerifera, Trlcliom.nilies cucunieroldes, T. japonicri and T. midtiloha, they occur in the hypoderma on the outside of the ring, and in Hchr^opeiioii hnjonhrfollus, v(tr. j^, Trichosanihex cucumcr aides, T. jap oui ca and T. iimltiloha. In Melolhria japonica, Luffa cylindrica, L. acutangida, Cucumis saiiviis, C. Jlelo, Cucurhiia Pepo and Gyiiuioslemiita cissoides this layer occupies a comparatively small portion of the testa. That of Cœurbiéa Pepo^' and Gymiiostemina cissoides is composed of thin-walled cells, furnished with reticulate thicken- ings. In some cases the second layer passes over gradationally to the third layer. The stone-cells of the third layer of the testa are especially large, and polyhedral or roundish. They are characterized by the considerable thickness of their walls, and by the presence of a greatly reduced cavity as well as numerous branched pore- canals. This \i\yQY is made up of one or more rows of cells, which sometimes have small intercellular spaces. In Luffa cylindricrr' and X. ac alang ulci'^ it consists of rather large prismatic stone-cells (200 !'■ long and 37 !'■ wide), w'hich are placed radially parallel ; while in Gymnosteinma cissoides it is made up of large reticulate cells. The parenchymatous cells of the fourth layer are thin-walled 1) H. A. LoTAR. loc. cit. p. 213. 2) Ditto, p. 221. 3) A. Yasuda. On t!ie Anatomy of L^iffa ar>ila)ir/>ila (L.) Rnxli. R(it;inical M:iQ;nzine, Tokyo. 180G. vol. x. No. 108. p. 37. ^50 A ET. 4. A. Y A RUT) A : ON THE and pitted or reticulate. They are mostly irregular-shaped and often stellate, with many remarkable intercellular sj^aces. Tlie reticulate cells occur in Actinostemma raceniosum and Cucurbita Pepo," those of the former being especially large and noticeable (PL V. Fig. 63). The tangentially compressed cells of the fifth layer are also thin-walled. They are gradually transferred over from the cells of the fourth layer. The remnant of the nucellus coming under the fifth layer is a strongly refractive tissue, which is made up of the epidermis of the nucellus and several rows of compressed cells. Then follows the remnant of the endodermis, under which are laid the cotyledons (PI. V. Fig. 62-66). SUMMARY. 1. The old stems of Luffa cylmdrica, L. aciUangulay Momordicn Charaniia and Actinostemma Tacemosw)i are very characteristic. They have a sharp keel along the angled portions. Microscopical examination shows that the ridges of Luffa cylindrica and L. acutangula consist only of outgrowths of the collenchyma, while those of Momordica Gharant'ui and Actinostemma racemosum are formed by newly developed secondary fibro-vascular bundles. 2. Stomata on the stems of the Cucurbitacese may be 1) J. F. FicKEL loe. cit. p. 757. and H. A. Lotar. loc. cil. p. 214. COMPARATIVE ANATOMY OF THE CUCUIIBITACE.E. 51 sometimes elevated with the epidermis ubove the surface of the hitter, as if they were supported by short thick hairs. o. Trichomes on tlie stems are of four kinds : (i.) sharp- pointed conical trichomes, (ii.) blunt-ended conical trichomes, (iii.) short-stalked glandular trichomes and (iv.) long-stalked glandular trichomes. The glandular trichome of Trichosanthes cuanneroides is especially characterized by having a single enlarged terminal cell ; that of Benincasa cerifera, by having two processes at the apex ; and that of Cacurbila Pepo Ijy having the head consisting of two cells one overlying the other. 4. There are four types of the distribution of the sieve-tubes in the stems : (a) vascular-bundle sieve-tubes, (b) ectocyclic sieve- tubes, (c) cntocyclic sieve-tubes, and (d) commissural sieve-tubes. Luffa cylindrica and L. acuUn^guh have those of the first and third types ; Äci'niodemma racemosum, 31elothri<( jdponica, Schizo- l^epon bryonkefoUus, var. japonicus, Momordica Charantia, Oitrullus vulgaris, Cucumis sativus, C. 3Ielo, Benincasa cerifera, Lagenar'ut vulgaris, Trichosanthes cuct/meroides, T. japonica, T. muUiloba and Gytnnostemma cissoides, those of the first, third and fourth ; and Cucurbita Pepo, all four types. Ö. Khizomes occur in Melothria japonica and Gymnostemnia cissoides. They are long and thick, having at several nodes three scales, which are anatomically distinguished as a shoot, a leaf and a tendril. They are full of starch-grains, those of Gymnostemma cissoides being the largest (0.06 mm. in diameter) contained in any of the organs of the Cucurbitaceie. 6. In the rhizomes the collenchyma, sclerenchyma and fibro-vascular bundles are considerably reduced. Their degenera- tion is much greater in Melothria japonica. than in Gymnostemma. cissoides. OU AUX. J. — A. YASUDA : ON Ï1ÎE 7. The number of libro-vascular buiullcs in tliu hypocotyls is generally six, except in Citrullas vulgaris and Cacurhlla Pcjto, the former of which has twelve, and the latter ten. 8. The epidermal cells on the upper surface of the blade of Trlclioianthes cucumeroldes are raised into conical papilke pointing outwards. 9. The epidermis of the blades may be many-layered as in Cucurh'iia Pepo, a character which is limited to this species. 3Iomordic(i Chardiii'ia contains globular cystoliths (O.OGomm. in diameter) in the greatly enlarged epidermal cells on the lower surface of the blade. They are united into groups. 10. Stomata on the npper surface of the blades are rarely found in ^romordlca Charantia, Trichosanthcs japonica and T. muHilobf, while they are entirely wanting in Schizopepo7i hnjo- nurfoVins, var. japonicus and Gymiiostemma cUso'ules. 11. The pallisade parenchyma is usually confined to thrce- or four-tenths of the thickness of the blades, but sometimes exceeds six-tenths in Cucurbita Pepo, where the pallisade cells may be arranged in double rows. 12. The fil.)ro-vascular bundles of the mid-ribs Jiear tlie ])roxinial portion of the blades are arranged in six ways : the first type has a single bundle situated at the centre of the mid- ribs [Aclinostemma raceniosam, Schizopepon hryoïiiœfoUus, var. ja2Joniciis) ; the second has two bundles overlapping each other {3Iclothria jajoonica) ; the third has three bundles, a larger one at the centre, the other smaller two on each side and above [Gymnostemma clssoides) ; the fourth also has three bundles, but they are arranged perpendicularly {Cacumls sali vus, ('. Melo, Benincast in ('iiawhiUi Fcpo, Be/iuicasa cerifera and Luffa acuta nguhi, in which they are hollow. The number of fibro-vascular bundles varies from five {Actinosteinma raccmosuiii, MelotJirla japonica and Gyiiino- steinma cissoides) to thirteen {Cucurbita Pepo). 14. Stomata on the cotyledons are in many cases present on both the upper and lower surfaces. Bloiitord'ica Cltaratiiia, Tr'i('JiOS(i)dhes japonica, T. niultUoba and Gymnosiciinna cissoides form an exception to this rule, the former three having the stomata on the lower surface, very rarely on the upper ; while the last has them exclusively on the under side. 15. The pallisade parenchyma of the cotyledons is commonly several-layered, while the spongy parenchyma is many-layered. The mesophyll of the cotyledons of Jlouiordica C/iaraniia, Trirhosanthes cuniuieroidcs, T. japonica, T. inuUiloIxi und Gyiiinosteiniia( cissoides is, however, scarcely diiferentiated into the pallisade and spongy tissues. 16. The spongy parenchyma of the cotyledons of Aclino- steiniiia racemosuin shows a reticulate appearance, the cells radiating from a fibro-vascuhir bundle and thus leaving very large inter- cellular spaces, so that it reminds one of the leaf-tissue of water-plants. 17. Unlike the common solid types, the stems of Cucurbita, Fcpo, Benincasa cerij'era, Layenaria vulyarls, the hypocotyls of Cucurbita Fepo, Beninousa cerifera, Layenaria vulgaris, Cucuiais 54 ART. J. — A. YASUDA : ON THE Melo, and the iiiaiii axis of the tendrils of Cifcuröita Fepo, Ben'inrasa cerifera, Luffa aculanguh, Scliizopepon bryoniœfolius, vor. japonicits are equally furnished with a central hollow. 18. The number of stomata found on the tendrils is nearly the same in all species. The number of fdjro-vascular bundles varies from four {Ädhiostenuim raceinosu/ii, GyiimosieHuna cissokles) to ten {Luffa vyUndriva). 19. In the old roots of Moinordica Cliarantia the fibro- vascular bundles have a double arrangement at the angled portions, as in the old stems. 20. The young roots of Actinosieiuma raceutosu/u have remarkably wide intercellular spaces, the cause of which is to be found in the habitus of the plant. 21. Tubers are confined to Trivhosanthea cucuuieroldeSf T. japonica and T. niuH'doba. The largest starch-grains contained in the roots are those of the tuber of Trichosanthes uiuUiloba (0.042 mm. in diameter). 22. The epidermal cells of the fruit may be radially flattened, cubical, or radially elongated. Cucumis salivus is characterized by having the epidermal cells four times longer radially than tangentially ; and Benlncam cerifera, by having a tangential septum in many of them. 23. In the pericarp-tissue a sclerenchymatous ring, either complete or incomplete, is generally formed. In Scliizopepon bryonid'folius, var. japoriicus, Cucuinh saiivus, 0. Melo and Cucurblla Fepo, however, it is entirely absent. 24. The surface of the fruit of Actinosieinina raceinosum, Mo/nordica Gkarantht and Cucumis sali vus is raised into tubercles. The origin of the tuljercles varies ; for those of Actinostemma niceiiiusi/iit and Mouiordica Ckaranlia are anatomically proved to be C'OAr PA NATIVE ANATO^FY OF THE rrCUrvlUTAf'R.T:. t>,) parenchymatous outgrowths, while those of Ci/nonis salhus are tlie large protuberant hnses of trichomes, which have become detached. 20. Sieve-tubes have a characteristic distribution in the fruits. Besides those found in the phloem there are isolated sieve-tubes in the tissue of the pericarp. These extra-phloem sieve-tubes occur in the hypoderma outside of the hardened ring. And when no hardened ring is present they are found in the outer part of the pericarp-tissue. 26. Into the tubercles on the surface of the fruit of 3Iomordira Chamnt'ia a fibro-vascular bundle enters, and runs perpendicularly to the surface branching out near the top of the tubercle. Nothing of this kind is observed in other genera. 27. The anatomical structure of the fibro-vascular bundles in the fruit of Luffa cylindrica and L. acutangula deserves notice because of the fact that the well-developed sclerenchyma surrounds the remaining weakly-developed portion of the bundle, the whole forming a spongy mass. 28. We may distinguish three kinds of epidermal cells in the seeds : a) flattened or cubical cells, b) prismatic cells, and c) prismatic cells with thickened-ridges on their walls. To the first class belong Actinoslemma raeemos^on, SchirMpepon hryonkv- f oil us, vnr. japonic}! i< and GymnostemnH( ris^oules ; to the second, J/omordica Charanlia, Trichosanlhes curu/iierotdes, T. japonica and T. muIiUoha ; and to the third, Jlelothria japonica, Laffa cylindrica, L. acniangula, CUrullus vulgaris, Cucamis sativus, C. Melo, Benincasa cerifera, Lagenaria vulgaris and Cucurbita Pepo. The thickened-ridges of Benincasa cerifera, Lagenaria vulgaris and Cucurbita Pepo ramify dendritically at the apex, while those of 3Ielothria japonica. Cucumis sativus, C. Melo are pointed towards the apex and widen towards the base. m ART. 4. — A. YARUDA : ON THE Contents. Introduotory and Historical Cliapter T. Tlie Stem Chapter 11. The Ilypocotyl Chapter iir. Tiie Blade Cliapter IV. Tiie Petiole ... . Chapter V. The Cotyledon ... Cliapter YI. The Tendril ... Chapter Y II. Tlie Root Chapter VlII. Tlie h'rnit Cliapter IX. Tlie Seed Siiiiiiu!ir\ 15 IS 2G 29 .S3 n') 41 47 A. YASri>A. ON THE COMPARATIVE ANATOMY OF THE CÜCURBITACE.«, WILD AND CULTIVATED, IN JAPAN, PLATE I. Plate I. Fig. 1-4. Diagrammatic representation of transverse sections through the stem of Benincasa ceri/cra in various stages. Fig. 1. A young stem ; Fio-. 2. A oTowins; stem ; Fig. 3. A full-o-rown stem ; Fig. 4. An old stem, ri, fibro-vascnlar bundle of the outer ring; h, libro-vascular Imndle of the inner ring; c, central hollow. Fig. 5. Sharp-pointed conical trichome of Melotliria japonica. Zeiss, 2xD. Fig. (), Blunt-ended conical trichome (»f ScJit'y.opepoi) hriioniœfoJivf<, vor. japonicus. Zeiss, 2 x D. F'ig. 7. Short-stalked glandular trichome of Cucumw sativus. Zeiss, 2 x D. Fig. 8. Long-stalked glandular trichome of Cucumis sativus. Zeiss, 2 x D. Fig. 9. The same of Benincasa cerifera. Zeiss, 2 x D. Fig. 10. „ ,, „ Trichosanthes cucumeroides. Zeiss, 2 x D. Fig. 11. ,, „ „ Cucurbita Pepo. Zeiss, 2x1). Fig. 12-14. Diagrammatic representation of transverse sections through various portions of the petiole of Bloiiiordica Charontia. Fig, 12, The distal portion ; Fig. 13. The middle i)ortion ; Fig. 14. The proximal portion. /'', fibro-vascular bundle ; /'. tibro-vascular bundle entering the ridges, which are raised on both sides of the groove. Fig. lö-lT. üiagrammatic rei)resentation of transverse sections through various portions of a petiole of Cacurhita Fepo. Fig. 15. The distal portion ; Fig. IG. The middle portion ; Fig. 17. The proximal portion. /, tibro-vascular bundle ; h, central hollow. Fig. 18. Scheme of the median transverse section through the testa of Actinostemma racemosurn . The same of Melotliria japonica. „ „ „ Schizopejjon hri/oniœfoli ?/■'', vor. japoiiicus. ,, ,, „ ßlomordica Chnranfia. ,, „ „ Luffa cylindricn. ,, ,, „ L. axutangida. „ ,, ,, Citrtilhis vidijarift. ,, ,, „ C?icurnis sntivus. „ „ C. Mdo. „ „ ,, Benincasa cerifera. ,, ,, ,, Lagenaria vulgaris. ,, ,, „ TrichosanfJies cucumeroides. ,, ,, „ T. japonica. ,, „ ,, T. midtiloha „ „ ,, Cucurbita Fepo. ,, „ ,, Gymnostemma cissoides, Fig. 19. Fig. 20. Fig. 21. Fig. 22. Fig. 2,3. Fig. 24. Fig. 25. Fig. 26. Fig. 27. Fig. 28. Fig. 29. Fig. 30, Fig. 31. Fig. 32. Fig. 33. A. Yasuda, Cucurbitaceae. Jour. Sei. Coll. Vol. XV HI., Art. 4, PL /. ûM I ■ i ^ ' 30 32 \m. 1 IUP TME -TOO pwimic 00. A. YASl'l>A. ON THE COMPARATIVE ANATOMY OF THE CUCURBITACEi;, WILD AND CULTIVATED, IN JAPAN. PLATE IL Plate II. Fig. 34. Elevated stoma from the stem of Lagenaria vulgaris, g, guard- cells ; e, epidermal cells ; a, res})iratory cavity. Zeiss, 4 x D. Fig. 35. Transverse section through an angled portion of an old «tem of Luffa cyUndrica, showing the sliar[) ridge consisting of collenchyma. e, epidermis ; cl, collenchyma ; ct, cortex ; sc, sclerenchyma ; p, peripheral phloem ; p' , axial phloem ; x, xylem ; m , medullary ray. Zeiss, 2xB. Fig. 3(J, Transverse section through a keeled portion of an old stem ot Moniordica Charantia, showing the shar[) ridge consisting of a newly formed secondary fibro-vascular bundle, e, epidermis ; cl, collen- chyma ; /, newly formed secondary fibro-vascular bundle ; p, peripheral phloem ; p, axial phloem ; x, xylem of the primary bundle ; sc, sclerenchyma of the primary bundle ; sc', sclerenchyma of the secondary bundle; rn, medullary ray. Zeiss, 1 x B, Fig. 37. Rhizome of 3'Ieîothria Japonica gathered in May. At each node three scales are developed, which are destined respectively to be a shoot (.s), a leaf {I), and a tendril {I). Natural size. Fig. 38. Starch-grains from the rhizome of Gymnostemma cissoides. Zeiss, 4xL). A. lusuda, CaciU'bitaccat. Jour. Sei. Coll. Vol. XVIII., Art. 4, PI. II. un. i IMP. THE HMO mKTIlS A. YASLI»A. ON THE COMPARATIVE ANATOMY OF THE CUCURBlTACEiE, WILD AND CULTIVATED, IN JAPAN. PLATE III Plate III. Fig. oU. Transverse sectiuii ui' the l)]cide of Trlchouantltcs cucumeroidts. e, e})iderinal cell on the iiuder surface raised into conical papillae outwards ; st, stoma ; jxi, pallisade jiarencliynia ; .sp, s[)ougy parenchyma ; yj fibro-vasciilar bundle. Zeiss, 4xD. Fig. 40. Transverse section of the blade of MomonUca Charantia. c, cj'sto- liths ; e, epidermis ; st, stoma ; -jm, pallisade parenchyma ; sp, spongy parenchyma. Zeiss, 4xD. , Fig. 41. Transverse section of the blade of (Jijmaosteimna cissoldes. ch, chlorojtlasts containing large spindle shaped starch-grains; e, e})i- dermis ; d, stoma ; pa, pallisade parenchyma ; sp, spongy paren- chyma. Zeiss, 4 X D. Fig. 42. Transverse section of the blade of Cucurbita Fepo through the etiolated portion, c, epidermis ; e', many-layered epidermis ; pa, pallisade parenchyma ; sp, spongy parenchyma. Zeiss, 4 x D. Fig. 43-48. Diagrammatic representation of six types of the mid-ribs of the leaves in reference to the manner of development of tibro- vascular bundles. Cross sections. Fig. 43. Actinostemma raee- viosum ; Fig. 44. Jfelothria japonica ; Fig. 4.5. Gymnostemma cissoidcs; Fig. 4(). Cucumis sativ us ; Fig. 47. Blovwrdica Charantia; Fig. 48. Cucurbita Fepo. f, libro-vascular bundle. A. Yo.suda, Cucurhitaceae. Jour. Sei. Coll. Vol. XVIII., Art. 4, PI. III. sp z^^-S -3^/ 48 um t MP. 1)C TCNO pnwnc ca A. YASri>A. ON THE COMPARATIVE ANATOMY OF THE CÜCüRßlTACEi:, WILD AND CULTIVATED, IN JAPAN. PLATE IV. Plate IV. F\^. 49, Transverse section of a cotyledon of Actinostemma racemosum. e. epidermis ; st, stoma ; pa, pallisade;parench3'ma :^sp, spongy paren- chyma ; /, fibro-vasciilar bundle; /, intercellular si>a('e. Zeiss, 2x1'. Fig. 50-54. Diagrams of transverse sections of a tendril of Bem'ncafta cerifera through various portions. Fig. 50. The proximal ])ortion ; Fig. 51. A fartlier'])ortion ; Fig. 52 and Fig, 53. Still farther portions; Fig. 54. The distal portion. /; fihro- vascular bundle; h, central hollow. Fig, :x). Transverse section of a yonng root of Acthiostemma raccmosvm. e, epidermis; /, intercellular space; ew, endodermis ; ^>r, pericambium ; X, tetrarch bundle. Zeiss, 4 x B. Fig. 56. Scheme of a longitudinal section through' a tubercle on|_the surface of the fruit of Momordica Charanlia. f, fibro- vascular bundle ; t, tubercle. Fig. 57. Trichome from the surface of a young fruit of Ctccitmis sativum, sujiported on a protuberant parenchymatous base, a, trichome ; e, epidermis ; h, protuberant parenchymatous base. Zeiss, 2 x 13. Fi"'. 58. Lonoitudinal section throu2;li a tubercle on the surfacejîof the fruit of Jctinostemina racemosiim. t, tubercle composed of parenchyma ; e, epidermis ; p, parenchyma ; sc, a portion of sclerenchyma form- ing an incomplete ring around the pericar[) ; _/", reduced fibro-vascular bundle. Zeiss, ?> x B. A. Yasuda, Cuciirhitaceai\ Jour. Sei. Coll. Vol. XVI H., Art. 4, PI. IV. UTH, 4 MP- THE TO«! prontCCO A. YASUDA. ON THE COMPARATIVE ANATOMY OF THE CGCÜRBITACEJl, WILD AND CULTIVATED, IN JAPAN. PLATE V. Plate V. Fif'. 59. Transverse section of a fibro-vascular bundle in the fruit-tissue of Luffa cylindrica. sc, well-developed sclerenchymatous sheath ; /, enclosed rudimentary bundle ; j9, parenchymatous cell. Zeiss, 4 x B. Fig. 60. Transverse section through the pericarp of Cucumis sativus. e, prismatic epidermal cell ; st, stoma ; sb, extra-phloem sieve-tube ; p, parenchymatous cell. Zeiss, 4 x B. Fi'T. 61. Transverse section through the pericarp of Benincasa cerifera. id, wax ; e, epidermal cell having a tangential septum ; h, trichome ; at, stoma ; sb, extra-phloem sieve-tube ; sc, sclerenchymatous ring ; p, parenchyma ; /, fibro-vascular bundle ; h, crystal. Zeiss, 4 x B. Fi(f. 62. Transverse section through the testa of I/uffa acutangula. 1, first layer : prismatic epidermal cells with thickened-ridges ; 2, second layer : small thick-walled })itted cells ; 3, third layer : one row of characteristic large [irismatic stone-cells ; 4, fourth layer : thin-walled stellate pitted cells ; 5, fifth layer : thin-walled com- pressed cells ; 6, sixth layer : the remnant of the nucellus ; 7, seventh layer : the remnant of the endosperm, r, thickening-ridge. Zeiss, 2xD. FifT 63. Transverse section through the testa of Acthiosteiivuia racemosuvi. 1, first layer : llattened epidermal cells ; 2, second layer : thick- walled reticulate and pitted cells ; 3, third layer : several-layered stone-cells ; 4, fourth layer : remarkably large reticulate cells ; 5, 6, and 7, fifth, sixth, and seventh layers, like those in fig. 62. Fi(T. 64. Transverse section through the testa of Melotliria japonica. 1, first layer : prismatic epidermal cells with pointed, thickened-ridges ; 2, second layer: thick-walled pitted cells; 3, third layer: stone- cells ; 4, fourth layer : slightly pitted thin- walled cells ; 5, 6, and 7, fifth, sixth, and seventh layers, like those in fig. 63. r, thickened-ridge. Zeiss, 2 x D. Fio-. 65. Tangential section through the epidermis of the testa of 3Ielothria japonica. e, epidermal cell ; r, thickened-ridges. Zeiss, 2 x D. Fi"'. ^^. Transverse section through the testa of Schizopepon bryonice/oUus, var. japonicus. 1, first layer : cubical epidermal cells with the thickened lateral walls ; 2, second layer : thick-walled pitted cells ; 3, third layer : large stone-cells ; 4, fourth layer : stellate thin- walled pitted cells; 5, 6, and 7, fifth, sixth, and seventh layers, like those in fig. 64. Zeiss, 2 x D. A. Yasuda, Cucurbitaccac. Jour. Sei. Coli. Voi. XV Hi., Art. 4, Pi. V. um, 4 WP THE TOIO PRimc CO JOURNAL OP THE SCIENCE OOLLEaE, IMPERIAL UNIVERSITY, TOKYO, JAPAN. VOL. XVIII., ARTICLE 5. Untersuchungen über die atmosphärischen Pilzkeime. (I. Mittheilung.) von K. Salto, Rigakushi. mit Tafeln I.-V. I. Einleitung. Von Herrn Professor Dr. Miyoshi war ich während des verflossenen Jahres mit einer Arbeit betrant, deren Eesultate ich in vorliegender Abhandhmg niedergelegt habe. Das Thema der Arbeit war eine möglichst vollständige Erforschung der Schim- melpilz-, Bakterien- nnd Hefenkeime der Luft an verschiedenen Ortlichkeiten Tokios, oder, in anderen Worten : Untersuchungen über die Luftkeime der Stadt Tokio sowohl in quantitativer als auch in qualitativer Hinsicht. Bevor ich auf die Beschreibung der Versuche eingehe, sei es mir erlaubt, kurz über die ein- schlägige Litteratur zu berichten. Seit den bahnbrechenden Forschungen Pasteurs weiss man, dass die mikrobiologische Zusammensetzung der Luft in hygie- nischer und gährungstechnischer Beziehung von einer grossen Bedeutung ist; deshalb sind diesbezügliche LTntersuchungen vielfach 2 ART. 5. — K. SAITO : UNTEESUCHÜNGEN UEBER unternommen worden. Von den vielen Lnftanalysen, die von verschiedenen Seiten ausgeführt wurden, sind diejenigen von MiQUEL, welche in der Station zu Montsouris bei Paris viele Jahre hindurch vom mikrobiologischen Standpunkte aus durch- geführt wurden, in erster Linie zu nennen. In seiner, im Jahre 1883 erschienenen klassischen Arbeit'' legte Mtquel auf Grund vieljähriger Untersuchungen fest, dass sowohl die Zahl als auch die Arten der in der Luft suspendierten kleinen Körperchen (Mineralien, Pollenkörner, Algen, Pilze, Bakterien u. s. w.) nach den Jahreszeiten variieren, so dass z. B. die Keimzahl der Schim- melpilze in warmen und feuchten Jahreszeiten am grössten, in kalten und trockenen hingegen am geringsten ist. Auch die Arbeit von Hansen^' zeigte, dass die Keimzahl und der Artencharakter der Mikroorganismen der Luft je nach dem Ort und der Zeit verschieden und von manichfachen obwal- tenden Faktoren abhängig sind. Nach ihm waren die Schim- melpilze stets am häufigsten zu treffen, danach kamen in der Beihenfolge Bakterien und schliesslich Hefen. Zu erwähnen ist noch, dass die Luft am Meere nach Fischer''^ IMoreu und Miquel^' und Minervini^' einen relativ geringeren Keimgehalt aufwies ; gleiches gilt auch für höhere Luftschichten der Berge, wo nach Giacosa'^' die entwickelungsfähigen Keime 1) Les organismes vivant de l'atmosphère. 1883. 2) Recherches sur les organismes qui, a différentes époques de l'année, se trouvent dans l'air a Carlsberg et aux alentours, etc. (Ref. im Bot. Centralbl. 1882. III. p. 7.) 3) FiscHEK, B., Bakteriologische Untersuchungen auf einer Reise nach Westindien. Zeitschrift f. Hygenie, 1886, p. 421. 4) MoEEU und MiquEL, Gehalt der Seeluft an Mikroorganismen. Cheni. Centralbl., 1886, No. 26. 5) MiNERViNi, R., Einige bakteriologische Untersuchungen über Luft und Wasser in- mitten des Nord-Atlantischen Oceans. (Zeitschr. für Hygenie und Infectionskr, XXXV, p. 166.) 6) GlACOSA, p., Sopra i germi centenuti ell'aiia a grandi altezze. (Ref. in JT.st, Bot. Jahresb., 1883, Bd. I, p. 308.) DIE ATMOSPHAEKISCHEN PILZKEIME. 6 der Schizomyceten mit der Höhe an Zahl abnehmen, während die Schimmelpilze dabei noch recht zahlreich zum Vorschein konamen. Was die Luft in Städten, Schulen und Hospitälern anbelangt, so berichtet Miquel^^ dass die Luft in den Städten weniger rein ist als die der Umgebung, und dieselbe im Hospitale vielmal reicher an Bakterien ist als die im Garten. Miflet^^ untersuchte hauptsächlich die Bakterienkeime in der Luft von Kloakenräumen und Arbetszimmern seines Instituts, in dem Operationszimmer der chirurgischen Klinik, dem Sektionszimmer des pathologischen In- stituts u. s. w. und fand an diesen Orten verschiedene Arten der Bakterien ; Schimmelpilze jedoch erwähnt er nur nebensächlich. Selander^^ erhielt aus der Luft der Festung Vaxholm 192 Schimmelpilzkolonien, wounter do^A zu Pénicillium glaucwn ge- hörten und die übrigen Botrytis cinerea, Oidiiuii lactis und eine andere nicht bestimmte Art waren. Gleichartige Untersuchungen wurden neuerdings von Cacace^^ in der Schule zu Capua ge- macht, und verschiedene Mikroben und Schimmelpilze aufgefunden. In den obenerwähnten Arbeiten findet man eine ausführliche statistische Untersuchung über die Schimmel j)ilzkei me in der Luft nicht ; zwar erwähnt Hais'Sex^^ das häufige Vorkommen von Cladosporium herhariim, Dematium pullans, danach Botrytis, llucor und Oidiuiii. Da aber die Zahl und Art der Luftkeime, hier 1) MlQüEL, 1. C. 2) MiFLET, Untersuchungen über die in der Luft suspendierten Bakterien. CoHx's Beiträge, Bd III, p. 119. 3) Selander, N. E., Luftuntersuchungen bei der Festung Vaxholm. Sv. vet. Ak. Bih. Bd. 13, 1886. No. 9. (Just's Bot. Jahresb., 1886, Bd XVI, T, p. 230.) 4) Cacace, E., Die Bakterien der Scliule. Bakteriologische Untersuchungen ausgeführt an dem Staube der Normalschule zu Capua. (Centralbl. f. Bakteriologie, I Abtiieilung, Bd. XXX, 1901, p. 653-657.) 5) Citirt in Jükuexsex, iMikroorganismen in der Gährungsindustrie. Vierte Auflage. p. 50. 4 ART. ö. — K. SAITO : UNTERSUCHUNGEN ÜEBER speciell Scliimmelpilzkeime, nach Zeit und Ort variieren, so ist wenigstens an gewissen Orten eine ein ganzes Jahr hindurch geführte statistische Untersuchung erforderlich, um nähere Kennt- niss über die Luftkeime zu gewinnen. Ausser dem theoretischen Interesse haben die statistischen Untersuchungen praktische Bedeutung bezüglich ihrer Berück- sichtigung in der Gährtechnologie, Hygienic, Pathologie u. s. w. Von diesem Standpunkte aus übernahm ich die Arbeit und führte von Anfang Mai 1901 bis Ende Mai 1902, also ein ganzes Jahr^^ hindurch, Untersuchungen aus, deren Ergebnisse ich in folgenden Kapiteln zusammenstelle. Es sei mir gestattet meinem hochehrten Lehrer Herrn Prof. Dr. Miyoshi an dieser Stelle für seine vielseitigen Anregungen und Unterstützungen meinen verbind- lichsten Dank auszusprechen. Die Fragen, welche durch meine Untersuchungen gelöst werden sollten, waren folgende : 1) Wie viele Keime von Schimmelpilzen sind in der Luft vorhanden, und wie variieren dieselben nach den Jahreszeiten ? 2) Welche Arten sind in der I^uft vorhanden, und in welcher Weise variieren diese nach Ort und Zeit ? Während meine Arbeit im Gange war, erschien ein Aufsatz von Shibuya'', welcher die Luftkeime an verschiedenen Orten Tokios untersucht und zahlreiche Bakterien- und Schimmelpilz- kolonien gefanden hatte. Seine Versuche haben sich aber nicht auf ein ganzes Jahr erstreckt, sondern wurden nur im Zeiträume von Oktober 1900 bis December desselben Jahres ausgeführt. 1) Während ich mit der Arbeit beschäftigt war, zwang mich leider ein Unfall dieselbe willirend des Augusts 1901 einzustellen. Dieser Monat wird also in der Arbeit nicht erwähnt werden. 2) Shibuya, S., Ueber das Vorkommen der Mikroorganismen in der Luft. Miltheilungcn d. medic. Gesellschaft zu Tokio, Bd. XVI, 3 Heft (Japanisch). DIE ATMOSPHAERISCHEN PILZKEIME. 5 Ueber die qualitative sowie quantitative Analyse der Scliimmelpilz- flora der Luft sind wir durch seine Angaben nicht näher unter- richtet. In der vorliegenden Untersuchung beschäftigen wir uns nur mit Schimmelpilzen, während Bakterien und Hefen in einer bald darauf folgenden Arbeit behandelt werden sollen. IL Methodisches. Obschon bei der Bestimmung der in der Luft suspendierten kleineu Körperchen mannigfaltige Methoden seitens früherer Forscher empfohlen worden sind^\ diente mir für meine sämmt- lichen Versuche die Aussetzung einiger PETEi'schen Schalen, die mit Nährgelatine beschickt waren. -^ Der ancjewandte Nährboden für die Entwickeluna- der auf- gefangenen Keime war die Soyagelatine, welche neuerdings von Prof. MiYOsnr^ als ein vortrefflicher Nährboden für Schimmelpilze in Anwendung gebracht wairde."^ Da andere Luftkeime als Schimmelpilze ausserhalb des Zweckes vorliegender Untersuchung 1) MiQUEL, ]. c; Petri, E.. J., Eine neue Methode, Bakterien und Pilzsporen ia der Luft naclizuweisen und zu zählen. Zeitschr. f. Hyglenie, Vol, 3, 1887, p. 1-145; Feankland, New méthode for determining the numher of niikroorganisms in air. Proc. Roy. Soc. XLI. Hesse: Ueber quantitative Bestimmung der in der Luft enthaltenen Mikroorganismen. Mittheilung d. königl. Gesundheitsamtes, Bd. II, 1884. u. s. \y. Da der Einwurf gegen die MiQUEL'schen Aufsamnielungsmethode erhoben werden konnte, ob die aufgefangenen Keime noch entwickelungsfähig sind, oder ob sie nicht schon ihre Keimfähigkeit völlig verloren haben, so sind von mir die entwickelungäfilhigen Keim; allein auf der Nährgelatine gerechnet worden. 2) Aus keinem anderem Grunde als dem der Bequemlichkeit benutzte ich die Schalenaus- setzungsmethode, trotz der Empfehlung Koavat^ewsky's( Die Methode der quantitativen Bestim- nuing niederer Organismen in der Luft, 18S5. Ref. in Bot. Jahresbericht, 18S8, Bd. XVI, I, p. 230), dass das ruhige Niedersinkenlassen der Bakterien aus der Luft bei vergleichenden Bestimmungen dem Aspirieren vorzuziehen sei. o) Anwsndung japanischer Soya und deren Gemisch auf Pilzkultur, Bot. Mag. Tokio, Ed. IX, p. 403. 6 ART. 5. — K. SAITO : UNTERSUCHUNGEN UEBER lagen, so wählte ich einen Nährboden, welcher für die Entwicke- lung der Letzteren passt, aber gleichzeitig für Bakterien ungünstig ist ; zu dem Ende wandte ich die Soyagelatine mit gutem Erfolge an. Der von mir verwandte Nährboden enthielt die folgenden Koncentrationsverhältnisse der Mischsubstanzen : — Soya (im Handel) 5 ccm. Koncent. Zwiebeldekokt 10 ,, Kohrzucker 5 gram. Leitungswasser 85 ccm. Gelatine* 7-15 Yo. Nach der Aussetzung in der Luft wurden die Schalen in Zimmertemperatur (ca. 18°C. im Mittel) gebracht, und bei kälteren Jahreszeiten ins Treibhaus (16-21°C.) versetzt. Die Inkubations- zeit dauerte ca. eine Woche. Nachdem die auf der Gelatine entwickelten Schimmelpilzkolonien gezählt waren, wurde die gesammte Anzahl in den Schalen auf eine bestimmte . Flächen- grösse und Aussetzungsdauer (60 qcm. und 10 Minuten)^^ um- gerechnet, um Vergleiche zli erleichtern. Diejenigen Schimmelpilze, w^elche auf dem oben erwähnten Nährgelatineboden nicht zum Fruktificieren kamen, wurden in andere Nährböden, wie Fleischpeptonagar, Brot u. s. w. wieder eingeimpft ; bei einigen Arten fand nach der Transplantation eine reichliche Bildung des Fruktifikationsorgans statt, bei anderen aber nicht. Bei den statistischen Untersachungen habe ich mich bemüht, sowohl die Luft von gleichen Lokalitäten zu untersuchen, als auch gleichzeitig die Luft von verschiedenartigen Lokalitäten mit 1) Diese Flächengiösse und Aussetzungsdauer Avälilte ich aus Bequemlichkeilzwecken für die Berechnuug. Die von mir tutgewandle I'ETKl'sclie Schale hält ca. 6 qcm. i;u Mittel. DIE ATMOSPHAERISCHEN PILZKEIME. / einander zu vergleichen. Folgende waren die Orte, deren Luft untersuclit wurde : — A) Der hiesige botanische Garten. B) Die Strasse zu Kanda. C) Die Seefläche. D) Der Operationssaal der chirurgischen Klinik. E) Der Kloakenraum des botanischen Instituts. F) Das Vorlesungszimmer einer Mittelschule zu Kanda. III. Specieller Teil. A. Garten. 1901. Mai. No. Datum Temp, in Wind Wetter U 2 5 iH S _ « Uli c Oi 'S 'T3_C — O 4* ^ > — o 1 8* p.m. 10 10.5 Schwa'jh Regen. Trüb, 752.3 5 10 41 82 3 3 p.m 13 17.8 j> etwas Regen. 758.0 1) !I 68 136 4 5 p.m. 15 23.0 Ziemlich stark (Süd Klar. 7G5.8 " )) 95 190 5 4 p.m. 21 20.0 11 Trüb. 755.4 )> )' 60 120 Morgens klar. Tags 6 2i- p.m. 24 22.0 Still Klar. 748.7 11 )) 91 182 vorher Regen. Letzte Nacht 8 9U.ra.31 22.0 760.0 11 )' 62 124 Regen und Gewitter. ]\Ieteorologische Daten dieses Monates : Barometer -Stand 758.9 1) Die relative Zahl giebt die Bereclnning auf 60 qcm. und 10 Minuten an. ART. 5. K. SAITO : UNTERSUCHUNGEN UEBER Temperatur Humidität Regenmenge f Schnelligkeit Wind l Richtuno- Monatsmittel der Pilzkeirae Maximaler Keimgehalt Minimaler Keimgrehalt 16°C 78.0 149.3 3.9 SSO 139.0'^ 190.0 82.0 Bei den sämmtlichen Versuchen in diesem Monate wurden Cladosporium herbarum, Pénicillium glaiicum, Epicoccum piir- purascens, Verticillmm glaucum, Botrytis cinerea, Aspergillus glaucus, Mucor racemosus, Macrosporiwn cladosporioides, Monilia sp. und Pyhnidenbildner gefunden, und zwar in folgenden Verhält- nissen'^:— ^VS:!"^/- 1 3 4 r ) 6 i S '^^$^ A.Z. R.Z. A.Z. R.Z. A.Z. E.Z. A.Z. R.Z. A.Z. RZ. A.Z. R.Z. Cladosporium h erbarum . 21 42 27 54 38 76 30 00 49 98 30 72 Pénicillium glaucum ... 1 2 5 10 15 30 o 6 2 4 1 2 Epicoccum purpurascem. 1 2 4 8 15 30 18 36 12 24 11 22 5 10 8 5 10 10 1 1 2 2 Verticilllum (j'aucum ... Aspergillus f/'a vcus 1 2 2 4 Mucor racemcmis 2 4 3Incrospor!wn clado- .spo/'i otc/e.s 29 .-i8 15 30 14 28 8 10 2 4 Mouilia sp () 12 G 12 8 16 2 4 2 4 3 6 1 2 1) Diese und andere Yergleichszahlen beziehen sich auf die zelinte Kolumne (Relative Zihl). Ebenso weiterhin. 2) A. Z. = Absolute Zahl (gefundene Zahl) R. Z. = Relative Zahl (umgerechnete Zahl) DIE ATMOSPHAEEISCHEN PILZKEIME. 9 Juni. No. Datum Temp, in Wind Wetter Barometer- Stand Zeitdauer des Aussotzens der Schalen in IMinnlen c '^ 9 Cd 9 SJp.m. 1 22.0 Ziemlich stark. Klar. Trüb, 763.4 5 10 182 364 11 4 p.m. 4 13.0 Schwach. etwas Regen. 760.2 )) 7 33 94 12 4j p.m. 5 22.0 » Trüb. 761.8 )) 10 39 78 Beginn 13 lip.m. 8 24.5 )) Klar. 751.4 3 )> 105 350 der Kegen- 15 11 a.m. 14 23.0 Ziemlich stark. i> 755.3 „ 9 101 374 periüde. 16 oh p.m. 14 21.5 7) " 753.6 !) )) 72 267 Vormit- 17 3 p.m. 15 21.0 Still. Trüb. 749.3 )) 10 33 110 tags starker 19 3J p.m. 20 26.0 Ziemlich stark. Klar. 754.0 )) )) 71 237 Hegen. Tags 20 Mittag 23 22.5 Still. Trüb. 754.2 )) ij 54 180 vorher Kegen. Letzte 22 21 p.m. 27 26.S Ziemlich stark. )' 757.8 '» i> 74 247 Nacht starker Regen. Meteorologische Daten dieses Monates : — Barometer-Stand Temperatur Humidität Regenmenge f Schnelligkeit Wind l Richtunt^ Monatsmittel der Pilzkeirae Maximaler Keimgehalt Minimaler Keimgehalt Bei den sämmtlichen Versuchen in diesem Monate wurden Cladosporimih herbarum, Pénicillium glaucum, Epicoccum inir- 2mrascens, Botrytis cinerea, Verticillium glaucum, Aspergillus glaucus, 756.1 20.6°C 82.0 172.1 3.4 SSO 230.0 374.0 78.0 10 ART. 5. — K. SAITO : UNTERSUCHUNGEN UEBER ^4. giganto-sulphureus nov. sp/\ Jllucor racemosus, 3Iacrosporium cladosporioides, Pestalozzia sp. und Pyknidenhilclner gefunden, und zwar in folgenden Verhältnissen : — 9 11 12 13 15 16 17 19 20 22 S3 s; p4 s: < s: < s: < SS < n; < ^ si S3 < f4 si < 22 s; X. 73 s; < 21 s: -^ 70 C'ladosporlum herharum . . . 91 182 18 51 15 30 32 107 26 96 26 96 15 50 42 140 Pénicillium glaunim 10 20 2 6 4 8 6 20 6 22 2 7 0 7 5 17 S 10 2 7 Epicoccunpurpurasceriii... 73 146 8 23 8 16 34 113 52 193 15 56 9 30 6 20 6 20 10 53 8 16 3 1 9 o 5 10 10 1 33 5 19 4 15 1 3 12 40 2 7 10 Verticillium glaucuui 1 o 1 3 4 13 A. glganto-mlphareus f) 7 2 4 4 8 1 16 3 53 12 45 24 89 1 1 3 3 7 2;; 3 10 12 40 Macrosporium cltdosp. ... Pyknideubildner •J 10 9 7 Pestalozzia sp 2 7 Juli. f. i ,. ^ "w ^ .Z^ «— in r- o fcD Temp. "o ;=: ^ X __ O ■^ o '^ o > __ iNo. Datura Ill 0° Wind Wetter il "5 -2 o 24 2 p.m. 4 26.0 Schwach Trüb. 759.4 o 8 64 2G7 26 3 p.m. 8 18.5 Still Trüb, Ke-ell. 757.1 )> 6 38 211 28 2ip.m. 15 18.5 )) Starker Kegen. 754.4 )) •' 45 250 30 1 1 p.m. 19 20.5 Schwach Trüb. 756.2 >1 " 77 428 Meteorologische Daten dieses Monates : — 1) Ueber die Beäclnelhung dieser und anderer neuer Arten vergleiche den Anhang des Abschnitts V. DIE ATMOSPHAERISCHEN PILZKEIME. 11 Barometer-Stand 75^.4 Temperatur 22.1°C Humidität 88.0 Eegenmeuge 229.0 ( Schnelligkeit 2.8 Wind \ [ Richtung SSO Monatsmittel der Pilzkeime 289.0 Maximaler Keimgehalt 428.0 Minimaler Keimgehalt 211.0 Bei den Versuchen wurden Cladosporium herharum, Pénicil- lium glaucum, Epicoccum purpurascens, Botrytis cinerea, Asper- gillus (jlaucus, Ä. nididans, A. caesiellus nov. sp., Sepedonium chrysospermum, Botryosiporium pidchrum, Mucor racemosus, Macro- sporium cladosporioides, Catenidaria fuliginea nov. sp., Pestalozzia sp. und Heterobotrys sp. gefunden, und zwar in folgenden Ver- hältnissen : — ^S=ï-^ 24 26 28 3U A.Z. K.Z. A.Z. K.Z. A.Z. K.Z. A.Z. K.Z. Cladoi^m'ium heibaruni 27 113 17 94 17 94 29 3 17 10 56 Epicoccum purpurascerift 9 38 7 39 1 () 3 ]3 8 44 Aspergillus glaucus 4 17 1 6 2 11 A. nidulans 1 6 Ä. caesiellus 1 6 Sepedonium chrysospermutn ... 3 17 IBctryospoi-iuiii pndch-um 1 6 Mucor racemosus 3 13 Macrosporium chtdosp riuides. a 13 3 17 2 11 4 2 17 8 Pestalozzis sp , o 17 1 12 ART. o. K. SAITO ! UNTERSUCHUNGEN UEBER September. No. 31 Datum Î S ^ r; Temp. t, jS H S ^ ? in Wind Wetter S rt t— " ' — "3 rt "^ C° eitda usset Mil N^ II N^ H p.m. ]8 1 p.m. 21 2 p.m. 26 24.0 Still Klar. 763.1 5 10 102 204 2;i5 Schwach Starker Regen. 751.8 » » 69 138 22.0 Still Trüb. 759.1 )) 6 75 250 pq Meteorologische Daten dieses Monates : — AVind Barometer-Stand Temperatur Humidität Regenmenge {Schnelligkeit Richtung Monatsmittel der Pilzkeime Maximaler Keimgehalt Minimaler Keimgehalt Folgende Arten waren vorhanden : — Cladosporimn herbarum, Pénicillium glaucum, Epicoccum pur- purascens, Botrytis cinerea, Verticillium glaucum, Aspergillus glaucus. Asp. ßavus, Epicoccum nigrum, Mucor Jlucedo, 3Iacro- sporium cladosporioides, 3Ionilia sp>., Oospora I, Catenularia fuliginea, Pestalozzia sp. und Heterohotrys sp. Die folgende Tabelle zeigt den Aufschluss ihres Vorkommens, 758.5 22.3°C 80.0 130.8 3.2 NNW 197.0 250.0 138.0 DIE ATMOSPHAERISCHEN PILZKEIME. 13 Cludosporium herbarum Pénicillium glaucam Epicoccuvi pw-purascens Boh-ytis cinerea Epieo''cuvi n igru ni Verticillium fjlaucum Aspergillus glaucus A. flavus Mucor Macedo Jfasrosporium cladoRporioides ... Monilia sp Pestalozzia $p Calemdaria fidiginea Oospora I. Ileterohotrys sp 81 A.Z. 34 9 19 1 1 4 1 32 83 R.Z. A.Z. 68 18 38 12 2 4 13 2 2 R.Z. A.Z. 26 4 4 16 33 R.Z. 110 27 13 10 Oktober. 33 ■— Temp. > tX) .No. Datum in Wind Wetter - B i ^ a; = _ rt — "3 n "Z ^ C° £ JOi i « 5-^ '^ "3 — ^ 'ön: 0 •- -J^ 5302 n;:z; rl a '"' t5-< 1— 1 ^ 35 11 a.m. 2 19.2 Schwach Klar. Etwas Regen. 765.1 5 9 103 229 Taos 36 Mittng 6 21.0 Still 7G0.9 )j 10 350 700 Torlier schwül. 37 5 p.m. 10 24.0 )j Klar. 760.4 !) " 129 258 Schwül, 39 3 p.m. 14 )) Schwach j> 755.2 >> )! 165 330 bisweilen starker Regen. 40 2^ p.m. 16 21.0 )) Trüb. 762.4 )> >> 36 72 41 2 p.m. 18 10.5 Stark Starker Regen. 766.6 )) » 94 188 43 4 p.m. 22 17.5 Still Klar. 7.72.3 6 33 110 44 2 p.m. 28 18.0 Ziemlicli stark Trüb. 759.9 ' .. 10 64 12s Morgen starker Resen. 14 ART. 5. K. SAITO : UNTERSUCHUNGEN UEBER Meteorologische Daten dieses Monates : — Barometer- Stand 762.5 Temperatur 16.8°C Humidität 83.0 Keo'enmenoje 311.1 {Sclinelligkeit 3.5 Eichtung NNW Monatsmittel der Pilzkeime 252.0 jMaximaler Keimgelialt 700.0 Älinimaler Keimgelialt 72.0 Bei den Versuchen in diesem JMonate wurden CladosporiuTn herbarum, Fe?iicilliuiu. glaucum, Botrytis cinerea, VerticilUuîn glauciim, Epicoccum purpurascens, Fusarium roseum, Aspergillus glaucus, A. ßavus, A. nidulans, Oedocephalium crystallinum, Jlucor racemosus, Rhizopus nigricans, Catenularia fuliginea, Gle- nospora s/>., 3Iacrosporium cladosporioides, Monilia sp., Iletero- hotrys sp., Pestalozzia s]j., Cylindrocephalium sp. und Pylcniden- bildner constatiert. Ueber ihr Vorkommen in den Versuchsreihen giebt folgende Tabelle Aufschluss :^- DIE ATMOSPHAERISCHEN PILZKEIME. 15 4 Z'^^oh. -»,^^-' Cladosporium herbarum PenicUUura (jloucum Epicoccum purpiirascens — Botrytis cinerea VerticiUium glaucum ,. Fusarium roseum Aspergillus (jlaiwus A. flnms , A. nidulans Oedocephaliurn crystalliiiun .. Mucor racemosus Shizopus nigricans Glenospora sp Macrosporium cladosporioides Ca'env.laria fuliginea Monilia sp Pestcdozzia sp Cy'iadrocephaliu/n sp Heterohotrys sp. Pyknidenbildner ort N 47104 4 226 13 5 2 1 6 4Ö2 2( 10 4 2 12 37 < h4 73146 71 14 4 39 86 10 40 N N3 26 41 174 87 61 4 2 43 44 s: :n fi 10 3S 23 7 15 35 301 1 2 6 14 1 2 20 12 40 2 10 16 ART. 5. — K. SAITO : UNTEESÜCHUNGEN UEBER November. No. 46 Datum Temp, in C° Wind Wetter Zeitdauer des Aussetzens der Schalen in Minuten ^, p.m. 4 15.0 Siill Klar. 771.3 5 10 29 58 Letzte 48 11 a.m. 9 17.0 Schwacli )) 766.2 )) 47 94 Naclit Regen. 52 11 p.m. 16 11.0 Fast still Trüb. 759.7 8 34 85 )) 5?, 3* p.m. 20 ir,.o Stark (West) Klar. 754.1 10 89 178 55 " p.m. 25 10.0 Sehr stark (Nord) )) 755.8 )) 215 430 Morgens Frost, 57 P. p.m. 29 >) Schwach )) 756.5 )) 26 52 Meteorologische Daten dieses Monates : — Barometer-Sta nd Temperatur Humid ität Re2;enmen2:e Wind I Schnelligkeit Richtung Monatsmittel der Pilzkeime Maximaler Keimgehalt Minimaler Keimgehalt 762.5 10.2°C 74.0 63.6 3.2 NNW loO.O 480.0 52.0 Die gefundenen Pilze sind folgende : — Cladosporium lierharum, Pénicillium glaucum, Ejncoccum purpurasce7is, Aspergillus glaucus, A. 7iidulans, Fusarimii rosewn, 3Iacrosp)orium cladosporioides, Cephalothecium i'oseum, Catenularia fidiginea, Oedocephalium crystallinum, Pestalozzi a sp., Acremonium DIE ATMOSPHAERISCHEN PTLZKEIME. 17 aUernatum, illonilia sp., Heterobotrys sp., Dematium pullans und Pyhiîdenbikhier, und zwar in folgendem Verhältnisse : — "^%^ 46 4S 52 53 55 57 A.Z. R.Z. A.Z. R.Z. A.Z. R.Z. A.Z. R.Z. A.Z. R.Z. A.Z. R.Z. Ciadof^orinm herharnm. 7 14 8 IG 11 28 m 72 109 218 9 18. Pénicillium glaucum ... 2 4 3 6 2 5 2 4 1 2 4 8 Epicoccum purpurascernf. 2 4 1 3 2 4 1 2 As-pergillus glaucm^ 1 2 • A. nidulans 1 1 2 2 3 8 4 8 Fusarium rosenvi Macro^porium clado- sporioide^ p, 6 3 6 3 8 4 8 2 4 Cepha lo'Jicc ium rose un} . 1 2 Catenularia fidiginea ... o 6 Oedocephalium crystal- linum - 1 o Pestalozzia sp 1 1 2 2 1 o 4 8 1 1 2 2 Monilia sp Demniium pullans 3 G Acremonium altemalum. 1 2 ^ Heterobotrys sp 10 20 5 10 8 20 12 24 1 2 2 4 1 2 1 2 1 2 18 ART. ö. — K. SAITO : UNTERSUCHUNGEN TEE ER December. ^Meteorologische Daten dieses Monates : — Barometer- Stand 702.2 Tempeiatur 4.5°C Humidität 65.0 Kegenmenge 1 10.8 f Schnelligkeit 3.4 Wind "" l Richtung NNW Monatsmittel der Pilzkeime 165.0 Maximaler Keimgehalt 484.0 ^Minimaler Keimgehalt 50.0 Bei den Versuchen wurden die folgenden Pilze gefunden : Ciadosporium herbarum^ Pénicillium glmicum, Epicoccum inir- 2nt.rascens, Mucor racemosus, Rhizopus nifjricans, Aspergillus glaucus, DIE ATMOSPHAEßISCHEN PILZKEIME. 19 Asp. nidulans, Fusarnan roseian, Veréicillmvi tomeiilosulum, Macros- porluni cladosporioides, Oate/iularia fuliguua, Pestalozzi a sp., Uonilia sp., Dematium pidlans, Heterobotrys sp. und Ptjkniden- hildner, luul zwar iii foloeudeii Verliältiiisseu : — •ä^^^"-«, 58 59 GO (52 64 '"■==^^*e. U 66 Cladusporiuni herharum. Pénicillium ylnicum ... Epicoccum purpurus- Asperrjilluii glaucuts ., A. nidulans Mucor racemosus Rhizopus nigricanif Fusarium roseun VerticilUum tomen'.o- sulurn Macrosporiu m da d -i- sporioidei Peslalozzia sp Cdtenidaria fidhjinea.. Monilia sp Dematium pulluns Pijkiiidenbildncv Hcterobotrijs sp 9 .'10 R.Z. A.Z ' R.Z. 1 A.Z. 151 1 16 R.Z. A.Z. ' R.Z. A.Z. R.Z. A.Z. R.Z. 82 4 11 16 ! 15 ! 30 14 7 14 7 i 14 20 ART. 5.— K. SAITO : UNTEESUCHUNGEN UEBER 1902. Januar. 1 ^ ^ Temp. er tie ns de n in ten dl ä >^ so No. Datum in Wind Wetter S i r- CP 30 60 74 21- p.m. 29 12.0 Still Klar. 765.7 >> » 13 26 Meteorologisclie Daten dieses Monates: — Barometer-Stand Temperatur Humidität Regenmenge (Schnelligkeit Richtung 761.7 2.4°C 59.0 33.4 3.8 NNW 95.0 217.0 26.0 Monatsmittel der Pilzkeime Maximaler Keimgehalt Minimaler Keimgehalt In der Luft fanden sich Gladosporlam lierharum. Pénicillium glaucum, P. leucocephalum, Pénicillium sp., Epicoccam piirpiirasccns, Mucor racemosus, Hyalopus ochraceus, Pestalozzia sp., Ovularia sp., Macrosporium cladosporioides, Ccdenularia fuUginea, Pyhniden- hildner, Heterohotrys sp. und Monilia sp., und zwar in folgenden Verhältnissen : — DIE ATMOSPHAERISCHEX PILZKEIME. 21 "-%=. 67 68 70 ! 71 74 A.Z. R.Z. A.Z. KZ. A.Z. R.Z. A.Z. R.Z. A.Z. R.Z. Clados-poriiim herbanim. Pénicillium gluucum ... 11 7 1 1 1 7 23 o 3 o 23 16 4 1 1 1 2 1 2 5 32 8 2 2 2 4 2 4 10 2 8 1 1 1 4 16 2 2 2 1 5 22 2 10 44 4 1 2 '1 1 8 2 4 2 2 P. sp Epicoccum purpurascens. Tlyahpus ochraceus Pestalozzia sp Ooularia sp Macrosporium chch- Catenidaria fuUginca ... Monilia sp Hclerubulri/s sjj. , Februar. No. Datum Temp, in C° Wind Wetter Barometer- Stand Zeitdauer des Aussetzeiis der Schalen in Minuten II "ri "S N«2 -l'i si GO ai > ^ Ol CS] P3 75 IJp.ra. 5 6.5 Schwach Klar. 7<'3.6 5 8 150 Tags 77 2ip.m. 12 G.O » !) 7G9.5 " 10 42 84 vorher etwas Schnee. 79 3 p.m. 19 6.5 Stark » 763.0 )) >} 89 178 81 2 p.m. 24 1.5.0 Stark (Nordwest) Trüb. 7()C.l )) » 44 88 22 ART. 5. — K. SAITO : UNTERSUCHUNGEN UEBEE Meteorologische Daten dieses Monates Barometer-Stand Temperatur Humidität Regenmenge f Sclinelliffkeit Wind ' ^ 7G4.6 3.8°C 61.0 20.9 3.8 NNW 125.0 178.0 84.0 l Richtung Monatsmittel der Pilzkeime Maximaler Keimgehalt Minimaler Keimgehalt In diesem Monate wurden Cladosporium herbarum, Pénicillium glaucum, Ejncoccuiii purpurascens, Aspergillus glaucus, Asp. nidulans, llucor racemosus, Blacrosporium cladosporioides, Catenu- laria fuliginea, Pestalozzia sp., Heier oholryn sp., Monilia sp., und Fyknidenbildner gefunden, und zwàv in folgenden Verhält- nissen : — Cladosporium herbarum Pénicillium glaunan Epicoccurii purpuraacens Asper (j Hl US glaucus Ä. nidulans Mucor racemosus Macrosporium cladospurloldes. Catenularla fnllglnea Pestalozzia sp Mouilla sp Pyknldcnbildner Hetcroboinjs sp 77 AZ. R.Z. A.Z. E.Z. 30 20 18 ]ö 20 18 2 4 A.Z. 74 ü IS 11 • )•) 1 o r- / 14 44 81 E.Z. A.Z. 12 4 1 R.Z. 24 8 -} DIE ATMOSPHAERISCHEN PILZKEIME. 23 März. ^J - r; 3 Temp. «'S ^ c cl "^ O > _ to Xo. Datum in C° Wind Wetter Zeitdan Aussetze Schale Minn |S ^1 S S3 3 p.m. 5 9.0 Schwach Ziemlich Trüb. 7G0.8 5 8 32 80 Fortge- setzt klares 85 2 p.m. 12 14.0 stark (Süd) Klar. 7C5.2 )) 10 46 92 Wetter. Tags zuvor und am 87 2 p.m. 18 21.0 Scliwach )) 761.9 » )) 27 54 Tage selbst Regen und Schnee. Ziemlich Trüf), 88 -Ih p.m. 24 7.5 stark etwas 769.7 )) )) 17 34 (Nord) Kegen. Meteorologische Daten dieses Monates Barometer-Stand Temperatur Humidität Regenmenge I Schnelligkeit 762.1 8.4°C 71.0 90.4 4.3 N 65.0 92.0 34.0 Bei den Versuchen wurden in diesem Monate Cladosporium herbarum, Pénicillium glaucum, Epicocoum p2irj)urascens, Asper- gillus glaiicus, Botrytis cinerea, Vertieillium glaucum, 3Iacro- sporium cladosporioides, Catenularia fidiginea, Sporotrichium laxum, Pestalozzia sp., Pyhnidenbildner, Verticicladium sp., Toinda sp., Ilonilia sp. und Heterobotrys sp. gefunden. Die folgende Tabelle zeigt das Verhältniss ihres Vorkommens. Wind ^ l Bichtung Monatsmittel der Pilzkeime Maximaler Keimgehalt Minimaler Keimgehalt 24 AET. 5. — K. SAITO : UNTERSUCHUNGEN UEBER ^^Si '^f, ■''P/;, ^Jl '^t}> 'J}^ f}ie ''^ ey. Oadosporium herbarum , Penicillmm (jlmicum Epicoccum purpurascens Bofii/iis cinerea Verticillium glaueum AspergUhis glaucus Macrosporium cladosp Calenularia fuliginea Sporotrichium laxnm Monilia sp Pesfalozzia sp Pyhnidenhildner Verticicladium sp Toi'vla sp Heterohotrys sp 8.1 A.Z. R.Z 85 A.Z. 9 4 ] 2 E.Z. 87 A.Z. K.Z 2 4 90 88 A.Z. K.Z. April. No. 90 94 90 9S 99 101 Datiir 2 p.m. 1 IJp.m. 8 11 p.m. 18 ] p.m. 19 1 p.m. 2.3 4^- )>.ni. .10 Temp. 14.0 21.0 15..5 19..5 15.0 21.0 Wiiul Schwach Ziemlich stark Still Schwacli Siill Wetter Barometer- Stand Tiiib. 770.5 Klar. 75.5.7 Trüb. 76.5.2 Klar. 748.9 )) 762.0 " 7,55.5 ^ «3 )) 62 207 DIE ATMOSPHAEKISCHEN PILZKEIME. 27 Meteorologische Daten dieses Monates : — Barometer-Stand 759.3 Temperatur 16.3°C Humidität 78.0 Regenmenge 222.7 f Schnelligkeit 4.1 Wind ^ _. , ricir\ l Richtung bbU Monatsmittel der Pilzkeime 213.0 Maximaler Keimgehalt 333.0 Minimaler Keimgehalt 127.0 In diesem Monate wurden gefunden: — Cladosporium herbarum, Pénicillium glaucum, Epicoccum purpurascens, Botrytis cinei^ea, Verticillium glaucum, Aspergillus glaucus, A. nidulans, Macro- sporium cladosporioides, Catenularia fuliginea, Dendryphium sp., Acremoniurii alier natum, 3Ionilia sp., Dematium pullans, Pestalozzia sp., Oospora II und ///, und zwar in folgenden Verhältnissen : — 28 AßT. 5. — K. SAITO : UNTERSUCHUNGEN UEBEE '^'^%>,, 102 104 106 10 7 A.Z. KZ. A.Z. R.Z. A.Z. K.Z. A.Z. K.Z. Cladosporiitm herbarum 14 47 2 7 12 40 17 57 renicill'ium glaucum 6 20 5 17 4 i;; Epicoccum purpurascens !•) 7 o 10 VJ o;] 1 1 ;5 1 o 12 40 1 1 1 •> 1 2 7 8 27 8 27 Macrosporium cladosporioides. 5 17 1 1 o / 28 9;î 4 i;] 2 7 6 20 1 1 1 ;> 2 8 27 Äcrernonium alternalum 1 ;î Es ist aus der Reihe der monatlich ausgeführten Versuche leicht erkennbar, dass die Schiinmelpilzkeime in der Gartenluft nach den Monaten mannigfaltig abwechseln. Dieses scheint sehr verschiedene Ursachen zu haben ; es wird dessen noch später in den folgenden Abschnitten Erwähnung gethan. Hervorgehoben sei an dieser Stelle nur, dass der maximale Keimgehalt der Schim- melpilze im Juli und der minimale im März liegen. Die im ganzen Jahre in der Gartenluft gefundenen Schiai- melpilze waren folgende : — Cladosporium herbarum, Pénicillium (/laucum, epicoccum pu7'purascens, Botrytis cinerea, Verticillium DIE ATMOSPHAERISCHEN PILZKEIME. 29 glaucum, Verticillium tomentosulum, Mucor racemosus, 31. llucedo, Rkizopus nigricans, Aspergillus glaucus, A. nidulans, A. caesiellus, A. giganto-sulphureiis, A. flavus, 3Iac7'osporiuni cladosporioides, Sepedoniiim chrysospermum, Botj'yosporium pulchrum, Catenidaria fuliglnea, Epicoccum nigrum, Fusarium roseum, Oedocephalimii crystallinum, Glenospora sp., Gylindrocephalium sp., Cephalothecium roseum, Acremonium alternatum, Hyalopus ochraceus, Pénicillium leucocephalum. Pénicillium sp., Sporotrichium laxum, Ovularia sp., Heterohotrys sp., Monilia sp., Dematium pullans, Dendryphium sp., Torula sp., Oospora I, //und ///, Vei'ticicladium sp., Pestalozzia sp. unci Piihnidenbildiier. Von denselben waren Cladosporium herbarum.. Pénicillium glaucum, Epicoccum puipurascens und Jiacrosjjorium cladosporioides die in der Gartenluft stets häufig gefundenen Arten. Botrytis cinerea und Verticillium glaucum zeigten sich nur in wärmeren Zeiten, während einige Andere wie Fusarium roseum und Hetero- hotrys sp. mehr in den kalten Monaten vorherrschten. B. HïKASSE. In Strassenluft führte ich allmonatlich auch einige Versuche aus, um die Zahlen der Schimmelpilzkeime mit denen im Garten zu vergleichen. Hier sind sie in nachfolgender Tabelle A mit nebenbei stehenden meteorologischen Daten kurz zusammengestellt ; in Tabelle B zeigt sich das Vorkommenverhältniss der daraus gefundenen Arten. 30 AKT. 5. — K. SAIÏO : UNTERSUCHUNGEN UEBER Tabelle A. 1) <1> - c ^ C3 Fetnp. 3 t- 2 ^ 53 -73 » 56 933 961 65 Si a.m. 25 4.9 )> Regen. 764.1 )> )) 45 750 ) 80 133,3 76 jNIittag 6 ;u Schwach. 76;!.7 1 78 1300 78 2^ p.m. 13 0.7 Still. Schnee. 759.7 52 867 700 80 21 p.m. 19 6.7 Trüb. 76;'..0 )) 20 33>3 82 2Ä p.m. 27 12.4 Schwach. Klar. 768.8 )) )' 18 300 84 2.V p.m. 6 10.7 1 Klar, etwa;- staubig. 764.0 1 )» 27 450 ;; 86 J p.m. 15 I 13.7 Etwas Klar. 757.8 Î) )i 28 467 389 89 ill a.m. 29 10.3 staub- fülirend. 1 " 768.2 I) )) 15 250 95 10.", a.m. lOl 9.8 Still. Trüb. 746.7 ] 23 383 4 100 ]J a.m. 29 14.5 Etwas Regen. 762.0 32 53>3, 458 103 2J p.m. 8 16.8 Scliwach. Starker Regen. 752.8 1 1 }) 80 1333 1 Von 2850 105 ; 21 p.m. 15 16.6 ! Still. li-2 p.m. Starker Regen. 766.2 » j) 262 4367 DIE ATMOSPHAERISCHEN PILZKEIME. 31 pq o I— I Ca O t>- ^~' — — , ;t o t^ t^ I 1— 1 ZH O CO 00 e? lO 1-1 1— ( za 1—1 CO CO ^ iC S* (M rH r— 1 r— 1 et ZY ZY o o za o o za I— 1 ZY ZY o o O O O o -N •za o o o o I— ( C3 o 1— ( •za (M ■^ ZY -^ •ZY O rt rt I-H '^^ ■-^ o o o ^ o o o 00 za o o O CO o 2 1—1 1-1 (M za 00 CC' rH I-H CT ZY O CO ^ 1— ( 1-1 (N zv o o O o o O O ^ za o o CO oo o o et i-^ o o 1—1 O 1—1 •za '^ 1—1 et ZY CO 00 et i-i ■* '"' '"' ZY o o o c o o o o o o •za CO a: ce CO za CT> (N lO M "*! to CO Ci 00 'ZY o lO ce Tï^ ZY , o o o o o o o o (M za CO o (M •za c* ZY o (M (N ZY >o za za LO ^ ^ C-1 ZY ^ s ZY o o O o o o cc za o o r-H lO O o o et o o ira za eo (M lO l-H ZV I— 1 lO lO t~ et CO lO ZY ^ o O o 1—1 za o o CO (M et o lO o ô3 za 1—1 CO S tc 1-1 «o t^ •zh: iS ZY ZH ZY O - î2 1— ( I— I 1— 1 1—1 ce ce CO 1^ CO O O ZH CO CO 05 1-1 rt rH Cq 1— ( '*< -* CO ZV 07/ ^ / J/i r i : i : i 1 i S i « t § s. s g 1 1 1 J J i : S : 1 1 S ^ is 2 .S •g •§ s .s O s s 's 1 ^ S "g iî 11 'S o g ^ s c ~ "S o -S ~. '* S 2 i a, ô ^ "S § DTE AT3rOSPHAEEISCHEN PILZKEIME. 33 Es ergab sich, class die Scliimmelpilzkeime in Strassen- und Gartenluft bezüglich ihrer Anzahl bedeutend von einander ab- weichen, worauf wir in den folgenden Abschnitten noch einmal zurückkommen werden. Hier sei nur erwähnt, dass in der Stras- senluft Pénicillium glmicum und Catenularia fuliginea häufig und besonders in Juli 1901 am meisten zu finden waren ; nächst ihnen kamen Aspergillus glaucus und A. nidulans. Epicoccum pur- purascens und 3Tacrosporium cladosporioides, die in der Grartenluft stets reichlich gefunden wurden, erschienen in dieser Versuchsreihe nur selten ; merkwürdig ist auch das gänzliche Fehlen der Botrytis ciîierea. C. See. Während eines Aufenthaltes im April 1902 in der biologischen Station zu Misaki führte ich am 5 des Monates Versuche aus, um die Anzahl der Schimmelpilzkeime über der Seefläche, dem Strande und in dem Stationslaboratorium mit einander zu ver- gleichen. Hierbei erhielt ich folgende Kesultate : CJ V _ U r- s c fcJO Temp. 33 p =-^ ^5 > -H E :^o. Ort. in Wind "Wetter ^ o O 3 ^- 2 rtü ^ C° P II Ol ^ s Seefläclie2engl. Nord vSchwach 91 Meilen entfernt vom Strande 14 Klar. 5 6 9 30 Tags vorlier stark windig. 92 Strand )) » )) I) >> 58 193 93 Laboratorinra )) )> )> " )) 48 160 34 AKT. 5. — K. SAITO : UNTERSUCHUNGEN UEBER Auffallend bei diesen Versuchen ist, dass die Seefläclie sehr arm an Schimmelpilzkeimen ist gegenüber der Keimzahl der Strand- und Laboratoriumsluft. Die Anzahl der gefundenen Pilze ist hier in Tabellenform gegeben : ^"^ï*.^ 91 92 <è?. A.Z. R.Z. A.Z. R Z. i A.Z. i R.Z. Cladosporium herbarum 4 1 i:i 10 o 21 2 ;io 1 70 7 100 Ab 2 150 Epirocrum pirpurasrens ... Denuitiiini ptdlnns o 1 Chaelocladium Johnsii An dem reichlichen Vorkommen von Cladosporimn. Jierharum erkennt man die Zimmerluft, wo Pénicillium glaucum gänzlich fehlt ; dieser Pilz lässt sich auch in der Strandluft nicht finden, umgekehrt ist das Verhältniss bei Ejncoccum purpurascens. Zu erwähnen ist noch, dass Chaetocladium Johnsii aus der Seeflächenluft gefangen wurde. D. Operationssaal der chirurgischen Klinik. Einmaliofer Versuch um lU a.m. des 13. Novembers 1901. Lufttemperatur im Zimmer 17°C, äussere Temperatur 11°C. Zim- merluft sehr feucht. AVind still und Wetter trüb. Aussetzung 15 Minuten. Zahl der Schalen 6. Zahl der Kolonien insgesammt 47, darunter Cladosporimn herbarum 22, Pénicillium glaucum 22, Epicoccum purpurascens 1, 3Iucor racemosus 1, Heterobotrys sp. 1. DIE ATMOSPHAEEISCHEN PILZKEIME. 35 Vergieiclit man die Zahlen und Arten der Pilzkolonien dieses Versuches mit den in Garten und Strasse im November angestellten Versuchen, so findet man keine nennenswerthe Verschiedenheit zwischen ihnen. E. Kloake. Ein einmaliger Versuch wurde um 11 2 a.m. des 10. Novembers 1901 im Kloakenraume des Botanischen Instituts vorgenommen. Lufttemperatur l7.ö°C. Wind still und Wetter klar. Aussetzung 10 ^linuten. Zahl der Schalen 6. Zahl der Pilzkolonien ins- gesammt 40, darunter Clados2}orium herbarum 17, Pénicillium glaucum 1, E'picoccum purpurascens 3, 3Iacrosporium cladosporioides 2 und Heterobolrys sp. 3. Eine besondere Eigenthümlichkeit der Pilzkeimverbreitung in der Kloakenluft sieht man hier nicht. F. VOKLESUNGSZIMMEE EINER MITTELSCHULE IN KaNDA. Zweimaliger Versuch am 4 und 11 Juni 1901. Die meteoro- logischen Daten und die Kolonienzahl von Schimmelpilzen waren folgende : No. Datum Temp. in ^Vind Wetter ■r. --^ O . P. 5) erfolgt die Keimung der Conidien von Pénicillium f/laumm sclion bei 1.5 — '2.0°C. 2) Zopf, Die Pilze, p. 217. DIE ATMOSPHAEEISCHEX PILZKEIME. 4o welcher dem Mycelhyphen von Cladosporium herbanim eine holie Resistenz gegen Austrockung zuschreibt. Unter den anderen Schimmelpilzen ist der purpurrothe Epicoc- cum purpiirascens nicht zu übersehen. Er tritt auch sehr häufig bei den Luftanalysen im Garten auf, aber nicht immer in der Stras- senluft. Allerdings entwickelt er sich auf Soyagelatine stet« steril ; auf zuckerhaltigen Fleischpeptonagar wieder eingeimpft beginnt er allmählich bräunlichgelbe, rauhwarzige areolierte Sporen zu bilden. In meinen Luftanalysen begegnete ich einem Schimmelpilze sehr häufig, der einmal von Lindner^^ als der ,,Schokolade- farbene Pilz" vorläufig genannt w^urde. Er wird besonders riesig in Zahl von der Strassenluft isoliert. In morphologischer Hinsicht der Couidienträger und Conidienbildungsweise bin ich seneiût, ihn in die Gattung; Catenularia einzuschliesen, deren Beschreibung im Anhange dieses Abschnitts gegeben ist. Einige Pilze treffen wir stets in einer gewissen Jahresperiode, in anderen Zeiten dagegen nicht. Als ein Beispiel nenne ich Botrytis cinerea. Trotz des niederen Teraperaturminimums der Sporenkeimuug'-^ ist der Pilz in AVintermonaten nie in der Luft zu finden. Die im Sommer so zahlreich gewesenen Keime veschwanden schon im November vollständig, um nach mehr- monatlicher Abwesenheit im Winter im März wieder zum Vorschein zu kommen. Ob das Fehlen im Winter dem Mangel an zu seiner Entwickelung günstigen Pflanzentheilen zuzuschreiben ist, oder an der schwachen Widerstandsfähigkeit der jungen Keim- schläuche gegen Kälte liegt, muss dahin gestellt bleiben. Doch zeigt uns die Thatsache, dass Botrytis cinerea das ganze Jahr 1) Betriebbkontrolle der Giiliniugsgewerbe. Dritte Auflage, p. olö. 2) Kach HoFFMAXN (Jahrb. f. wiss. Bot. Bd. 11, 1Ö60, p. 267) ist die luinimale Temperatur der Sporenkeimuiig 1.6°C. 44 AET. 5. — K. SAITO I UNTERSUCHUNGEN UEBER hindurch nie in der Strassenhift gefunden war, die beschränkten Lebensbedingungen des Pilzes. Bezüglich des Vorkommnissverhältnisses steht Verticillium glaucum nahe der vorigen Species. Ihr Keim fehlt in Gartenluft im November, December, Januar und Februar ; nur selten kam der Pilz bei Strassenluft im April und Mai 1902 vor. Zu anderen Heeren von Lebewesen stehen einige in ganz umgekehrtem Verhältnisse, d. h. sie finden sich in kälterer Zeit reichlich ; z. B. Fusariaiii roseuni und Heterohotrys sp. Wie man aus der nachstehenden Tabelle sieht, beschränkten sich diese Pilze darauf, in der kälteren Periode hervorzutreten, und zwar kam der Erstere nirgends zum Vorschein als im Oktober, November und December, auch nicht in Strassenluft. Nur in dem Fehlen des entwickelungsbegünstigenden Nährmaterials er- blicken wir den Grund des begrenzten Vorkommens dieses Pilzes, jedoch nicht in Temperaturbedingungen, weil sie in höherer Temperatur wohl kultiviert werden können. Von den Pyrenomyceten waren Muco?' racemosus und Rhizopus nigricayis die in unseren Versuchen häufigst vorkommenden Arten. Von diesen tritt M. racemosiis in Gartenluft öfters auf, Rhizopus nigricans in der Strassenluft verhältnissraässig reichlich. M. Mucedo, der auf Pferdemist vegetiert, wurde bei unseren Versuchen nur selten und Chaetocladium Johnsil, ein Schmarotzer auf dem Ilucor, nur einmal in der Seeflächenluft gefunden. Eine andere Gruppe der Schimmelpilze, Äspei^gillus, kommt gleichfalls bei unseren Versuche nicht selten vor. Neben einer Reihe älterer, gut bekannter Arten habe ich noch zwei bisher unbeschriebene Formen isoliert. Eine Uebersicht der Formen ergiebt nach Wehmer^^ folgendes Schema. 1) Eot. CVnlralbl., IV\. LXXX, 1«9D, p. 449. DIE ATMOSPHAEEISCHEN PILZKEIME. 45 I. Weisse species. Aspzrfjillm albus Willi. II. Gelb-braiuie Species. 1) Mit einfachen Sterignaen. Asp. giffanto-sulphureus nov. sp. III. Grüne bis grüngelbe oder gelblicbgrüne Species. 1) Grossporige Arten, Makrosporen, nur unverzweigte Sterigmen. Asp. gkiTicus Link. Asp. flams Bref. 2) Kleinsporige Arten, Mikrosporen, Sterigmen einfach oder verzweigt. a) Mit zwergigen Conidienträgern. a) Einfache Sterigmen. Aap. caesiellus uov. sp. ß) Verzweigte Sterigmen. Asp. nidulans Eid. Unter diesen Arten traten am Läufigsten Asp. glaucus mid Äsp. nidulans auf, insbesondere in der Kegenperiode ; die erstere ist neben Pénicillium glaiicum ein mehr oder minder heftiger Feind des Haushalts. Hier dürfte noch hinzufügen sein, in welcher Weise die anderen Arten in der Luft auftraten. Artenname Versuchsreihe Asp. flavus. 33, 36, 37, 43 (Sep.-Okt. 1901) Garten. Asp. giganto-sulphur eus. lo (Juni 1901) Garten. Asp. albus. 82 (Februar 1902) Strasse. Asp. caesiellus. 28 (Juli 1901) Garten. Bei Garten- und Strassenluft nicht selten anzutreffen waren noch Macrosporium cladosporioides, Jlonilia sp. und ein andere mit rothbräunlichen Pykniden (vorläufig der ,,Pyknidenbildner" genannt). Sie kamen aber sehr unregelmässig in Zahl vor. Am Schlüsse dieses Abschnittes sind die relativen Monatsmittel der Schimmelpilzkeime in Garten- und Strassenluft, i.e. die Zahlen auf 60 qcm. Fläche bei 10 Minuten Aussetzung berechnet, angegeben. 46 ART. 5. — K. SAITO : UNTEESUCHUKGEX UEBER O C5 O . -* ■ o i^-ç 1 o ■^ — ^ l^ >3i zc LO rH Ci lO s -fi C/J I' O C^l O S^ C^l lO rf) CO '2> ■-H S^ '-' ■> 1—1 LO M Gi OJ Ci Cv| •<*i Oi ^ m g^ "' '^ et < d (M r^ O (M 1-1 I-H 00 -* 1-1 T-, rt I-H I-H rf t- LO OO 'tc t^ rH CO Oi CO CO CO •— Cß l^ LO ^H 1—1 CO s d I— ( o -rff 1—1 I— 1 rt 1^4 rt LO CT (M 1-H 1-1 OO r ,— ( ■* Tj< OS t^ rr 1— ( yj (M J rH OA (M d t-; O CO LO T-( rH rH i2 rH r^ l^ Oi O !M 1- J — ^- I^ o ai l^ T—^ CO w' L\ A I— t d i^ ÏÏ + " " ^ + "■ + 05 ■— 1 CO o er. , — C-. ^ CO CO CO œ o lO O lO- cc CC 00 (N CO CO C<1 O 1.- J^ CO (N 1—1 o d ^ lO «o c^ CO CC c CO LO o o c- O O ^ + ^ ^ ^ + S I^ Ci I— 1 ^^ o I-H (M GO 1-5 r-l CO " 05 OJ - s d l~ Ci et CO r-i (M (M rH '"' w- + â^ / / .J / / -^ r / 1 S e = 5 o 0/ / / o '^ / ^ 5 5 ^ 1 <;. r-^ :î ? .c> a '- / / 'î' ^/ / •'? '^/ / S "§ 1 "c "i '^ -1 ^ o . < ç =^ s Î s ^ ^ £ C 1 î K .Î 'S 1 "^ ^ "^ -^ pq pr; 5 1 1 1- o ^=1 DIE ATMOSPHAERISCHEN PILZKEIME. 47 + + + + + à .-^ «j S s -s- Q- Ch îQ O f^ ;^ C5 O s a -o S -2 ~ ~ b ~ T. « l."« ^ ^ . L^ ^ JJ ^O o •I i ^ ri o .5, !^ Ci? 55 Q' 4S ART. 5. — K. SAITO I UNTERSUCHUNGEN UEBER Anhang. Beschreibung der neuen Arten. 1) Aspergillus giganto-su^jhurevs. Der Pilz bildet eine flockige, anfangs weisse, bald hellgelb^^ erscheinende Decke auf dem Nährsubstrate. Bei älteren Kulturen geht die Färbung in Schmutzig-bräunlichgelb über. Die statt- lichen Conidienträger sind ansehnlich, über ein Millimeter hoch, meist einfach, seltener verzweigt ; die älteren Träger sind mit vielen Querwänden versehen. Die Wand des Conidienträgers ist glatt und farblos, meist derb ; das Ende des Trägers quillt zu einer kolbenförmigen grossen Blase auf. Die stets einfachen, dicken Sterigmen sind lang (über i des Blasen durchmessers oder fast gleich demselben, „langstrahlig") und bedecken den Blasen- gipfel allein. Die Conidien sind überall kugelig, glatt oder warzig, gross und vakuolenhaltig. Die vegetativen Hyphen sind ebenfalls dick und wuchern manchmal durch die Conidienträger hindurch. Grössenverhältnisse : Conidienträger ± 1 mm. Stiel desselben 6-16 /< dick. Wanddicke 2 i". Blase 28-32 /^ im Durchmesser (kolbig). Sterigmen 24-28 /^ lang und 7 /^ dick. Conidien 8-12 /^ im Durchmesser. Vegetative Hyphen Ca. 4 /^ dick. Fruchtkörper unbekannt. Selten. Gasblasenentbindung auf den Kulturen in Würz und Kojiextrakt- lösung, gleichzeitig mit schwacher Alkohol bildung (lodoformprobe). 1) Die Farbenzeichnung ist nach Saccardo's Chromotaxia angegeben. DIE AT3I0SPH A ERISCHEN PILZKEIME. 49 Affinität : — Dieser Pilz steht in vielen Beziehungen ^i. Ostimivs ^Vehmer nahe, jedoch unterscheidet er sich vor allem in der Farbe der Pilzdecke, da diese bei unserem Pilze von Hellgelb zu Schmutzig- bräunlichgelb übergeht, während sie bei ^4. Osfianvs durchaus beständig bleibt. Ferner ist die Blase bei A. Osltanifs kugelig und scharf gegen den Trägerstiel abgesetzt, während sie bei A. giganto-svlplmreus eine kolbenförmig ist. Der Conidienträger von A. giganto-suljjhureus ist zuweilen verzweigt, dagegen fehlt bei A. Osiianns eine solche Verzweigung vollständig. Von anderen morphologischen Besonderheiten sind die Sterigmen und Conidien zu erwähnen. Bei A. Osiiantis ist die Blase von allseitig ausstrahlenden Sterigmen besetzt, dagegen bei A. giganto-mlphiireus bedecken sie nur den Blasengipfel. Die Conidien von A. Oélianns sind kugelig bis oval und kleingestaltet, während die von A. gu/anto~sulphureuii kugelig und grossporig sind. 2) Aspergillvs caesiellns. Diese Art zeichnet sich durch zwerghaften Wuchs ihres Conidienträgers aus. Die Farbe der Decke ist ein etwas in's Graue übergehendes Grün, welches auch lange Zeit seine Nuance behält. Der Conidienträger ist sehr kurz, seine Wand stets glatt und sehr dünn. Der Stiel geht allmählich in eine kolbige Blase über, von deren Gipfel wenige, stets einfache Sterigmen (über l des Blasendurchmessers messend) ausstrahlen. Die Letzteren schnüren lange Ketten von meist glatten, gelegentlich auch warzigen ellipsoidisch-ovalen Conidien ab (kleinsporig). r)0 ART. "'. — K. SATTO : UNTERRÜCHÜNGEX TJERER Grössen verliiiltnisse : — Conidienträgor l()0-20() /^ Stiel desselben 4-6 !'■ dick. Wanddicke 0,-3 //. (dünnwandig). BLase 12-14 /' im Dnrchmessei- (kolbig). Sterigmen 12 x l] //.. Conidien 4x7 !'■ im Durchmesser. Vegetative Hyphen 2-4 /< dick. Frnchtkörper unbekannt, wärmeliebende Art (in o()°C' gut wach- send). Selten. Affinität :— Der zwerü'ia; "wachsende Pilz ist in vielen Bezichunii-en A. miniiirus Wehmer nahestehend (Persistenz der Deckfjirbe, Ton der Conidientarbe, einfache Sterigmen, zwerghafte Conidienträger, höhere optimale Wachsthumstemperatur u. s. w.). In anderen Punkten weicht A. minimns von unserem Pilze ab. Beim Ersteren geht der derliwandige Stiel fast unvermittelt in eine stets kugelige Blase über, beim Letzteren dagegen ist eine allmählich ü]:)er- sehende Blase vorhanden. Ferner übertrifft die Blase von A. inininius die Sterigmenlänge erheblich, während bei A. caesiellus die Länge beider Organe fast gleich ist. Die Sterigmen, welche bei A. minimus von der ganzen Blasen Oberfläche ausstrahlen, bedecken bei .f. cae.'iieUus stets nur den Blasengipfel. Beide Pilze sind auch in (Irösse und Gestalt der Conidien verschieden, indem bei A, minimus die Conidien meist oval und 2 ,« gross, bei A. caesiellm dieselben überall ellipsoidisch-oval und 4x7/^ gross sind. DIE ATHOSPHAERISCHElSr PILZKEFME. ol 3) Catenulan'a fiiUginea. Dieser Pilz, der einmal vun Lindxer als der ,,Schokolade- farbene Pilz" vorläufig genannt wurde, ist durch seine selirot- korngrosse, schokoladenfarbige Kolonieenlagerung auf dem Nähr- boden ausoezeichnet. Die Kolonie ist von kleinen Conidien staubiir bedeckt ; die letzteren stammen von kettenförmigen Abschnürungen des Endes des Conidienträgers und zerfallen leicht in einzelne Gliederzellen. Die Conidien sind ursprünglich kurz cylindrisch, später aber runden sie sich ab und werden 2-e3 !'■ gross. Beim Keimen quellen sie enorm auf. Die kriechenden ca. 3 ,« dicken vegetativen Hyphen sind anfangs farblos und in ihrer Contour gleichmässig, aber mit tlem ^Vltei' wird etrit1't, so Ijestätigten sich hauptsächlich die Angaben Miquels. Meine Versuche ergaben : — Die Gartenlut't enthielt in verschiedenen Perioden eine verschiedene Anzahl von Schimmelpilzkeimen. In den warmen und feuchten Jahreszeiten, ]) ßi^lier sind C soiijilex, C. aim uiul C. cchinuUi ber^clirieben. 52 ART. 5. K. SAIÏO : UNTERSUCHUNGEN UEBER d. li. besonders im Juli, sind die Pilzkeime am zahlreichsten, während sie in kalten und trockenen Zeiten dagegen geringer an Zahl sind, und im März den minimalen Gehalt erreichen. Dieselben Verhältnisse wurden auch in der Strassenluft constatiert, aber es zeigte sich hierbei die Luft im allgemeinen weniger rein als im Garten ; ebenso wich bezüglich der Schimmel- pilzarten die Gartenluft von der Strassenluft etwas ab. 3) Bei gleichen meteorologischen Verhältnissen sind die Monatsmittel der Schimmelpilzkeime von dei" Regenmenge abhängig. 4) Die Luft trägt bei stiirkem Winde eine reichliche Anzahl von Keimen in sich. ö) Gleich nach starkem Regen- und Schneefall war die Luft keimärmei' an Schimmelpilzen, 0) Beinahe keimfreie Luft findet sich über dem Meere, während am Strande die Luft noch viele Keime enthielt. Es ist ohne weiteres klar, dass ein lAiftraum liber der Seefiäche, wo die Schinnnelpilzkeime ganz fehlen, in nicht allzuweiter Entferiumg vom Strande 2;efunden werden dürfte. 7) Obgleich die Zahl unserer Versuche über den Keimgehalt der Laboratorium-, Hospital- und Kloakenluft noch recht gering ist, so ergiebt sich doch, dass eine l)esondere Eigenthümlichkeit der Schimmelpilzkeime dieser Luft nicht existiert. 8) Die bei sämmtlichen Versuchen am häufigsten gefundenen Schimmelpilze waren ('iadospori/n// licrharuni , PenlcUrunii (jlaKctnii und Kpicocciiia purpurascens, danach Äi>pcrgillus (jlaueus, A. iiiduhinH, ( 'aieiiiilarni fulig'incu, Mncor racemosu^, Ilhhopua nigricans, j\Laürosj)or'niiii clf/do-^porioidex, MoiiUhi .yj. und Fyhniden- hildner. U) Es ist als ein bezeichnender Charakterzug für die Ver- DIE ATMOSPHAERISCHEN PILZKEIME. Oo breituug der Schimmelpilze iiï der Luft anzuseilen, da.ss Bolrytls cinerea und VerticilUiun glaucum nur in den wärmeren Perioden in der Garteuluft vorkamen, dagegen Heterobotrys sp. und Fusarium roseum besonders in kälteren Zeiten vorherrschten. 10) Selten wurden dagegen die folgenden Schimmelpilze aus der Luft isoliert : — Mucor Jlueedo. Chaetocladium JoJuisii, Pénicillium leueocephaluiii, Pénicillium sp., Epicoccum nignini, Allernaria tenuis, Asperffillus ßavua, A. fjiganto-sulpliurens, A. albus, A. caesiellus, Cephalotlieciurn roseum, Oedocephalium cry- stalUnum, Sepedonium chrysospermu)!), Glenospora sp., VerliciUium tomentosulum, Sporotrichium laxum, Botryosporium pidchrum, Hyalopus ochraceus, Dematium puUans, Torula sp., Oospora I, II und III, Slysanus microsporus, Dendryphtum ptenicilliafc//), Dendryphium sp., Cylindrocephalinm sp., Acremoiiium alternaium, Ovularia sp., Pesialozzia sp. und Verticicladium sp. 54 ART. 5. — K. SAITO : UNTERSUCHUNGEN UEBER Anhang, Systematische Anordnung- der im Texte erwähnten Schimmelpilze. I. Pliycumycetes. Gattung 1, Mucor. M. Micedo L. (Fig. 9) J/. race m onus Fres. (F^g- '') Gattung 2, Rhizopus. IL nitjricans. (^^g- ^) Gattung o, Chaetocladium. ('h. Joh/isii F'res. (Fig. 27) IL Ascomycetes. Gattung 1, Aspergillus. ^i. f/lai'ci/s Link (^'"ig- 1^*) J. liidiilant^ Eidam (l^iS- H) .1. /^^n^ö- Bref. (Fig. 13) Ä. albus AVilh. (Fig. 15) A. glganto-sulphurt'its n. sp. (Fig. 12) A. caeslellits n. s}>. [F^t^- 14) Gattinig 2, Pénicillium. P. fjlaucini/ Link. (Fig. 2) P. leiLcocepJtalinn Rabenh. (Fig. 35) P. sp. (Fig. 32) (^attung 3, Botrytis. IJ. chierea Fers. (Fig. 3) III. F\nigi imperfecti. ( 1 ) Sphaelopsidales . DIE ATMOSPHAERISCHEN PILZKEO[E. Öü Gattung 1, ? Ein Pi/knidenhUdner. (Fig. 20) (2) ]Melanconiales. Gattung 1, Pestalozzia. P. sp. (Fig. 18) (3) Hypliomycetes. (Ti'uppe 1, Mucedinaceae. Gattung 1, Oospora. 0. I. (Fig. 28) 0. II. (Fig. 37) 0. III. (Fig. 44) Gattung 2, Monilia. 31. sp. (Fig. 23) Gattung 3, Oedocephalium. Oe. crystallinum Ces. (Fig. 29) Gattung 4, Cylindrocephalium. Cy. sp. (Fig. 31) Gattung 5, Botrvosporium. B. pulchriim Corda. (Fig. 40) Gattung 6, Hyalopus. H. ochraceus Corda. (Fig. 19) Gattung 7, Acremonium. A. aJternatum Link. (Fig. 36) Gattung 8, Ovularia. 0. sp. (Fig. 46) Gattung 9, Sporotrichium. S. laxinn Nees. (Fig. 38) Gattung 10, Sepedonium. S. cJwysospermuw (Bull) Fr. (Fig. 25) Gattung 11, Verti ci Ilium. /)6 ART. 5. — K. SAITO : TJNTERSUCHUNGET^ UEBER V. glcmcum Bon. (^'^Z- ^*^ V. tomentosuluni fS])eg. (Fi."'- 42) Gattung 12, Cephalothecium. 6'. roseum Corda. (Fig. 26) Gruppe 2, Dematiaceae. Gattung 1, Toriila. T: sp. (Fig. 17) Gatting 2, HeterobotrY!^=. H. sp. (Fig. 22) Gattung 8, Glenospora. 6^. sp. (Fig. 30) Gattung 4, Catenularia. 61 fidiginea n. sp. (Fig. 4) Gattung 5, Dematium. D. pullans de Bary et Low. (Fig. 39) Gattung 6, Verticicladium. V. sp. (Fig. 43) Gattung 7, Cladosporium. C. herbarwii (Pers) T^ink. (Fig. 1) Gattung 8, Macrosporium. 31. cladosporioides Desm. (Fig. 21) Gattung 9, Alternaria. A. tenuis Nees. (Fig. 45) Gattung 10, Dendryphium. I), penicilliatum (Corda) Fr. (Fig. 24) D. sp. (Fig. 41) Gruppe 3, Btilbaceae. Gattung 1, Stysanus. S. microsporiis Sace. (Fig. 34) Gruppe 4, Tuberculariaceae. DIE AT^rOSniAERTSCTTEX PTLZKEi:\rE. 57 Gattung 1, Fusarium. F. roseum Link. (f'^^-'- 1^) Gattung 2, Epicoccum. IJ. jmrjnirasceiis Ehrciib. (I'^ir- '"*) J^. nigrwih Link. (Fig. 33) December 1902. Botanisches Institut Kaiserl. Universität zu Tokio. ■rs-^ hS Ar.T. 5. — K. SAITO : TINTEnSUCIIUNGEN UEEEK Inhalt. Sk[t. I. Einleitung 1 ir. Metliodisches •' in. Speeieller Teil 7 A. Garten " B. Strasse '-^^ C. See ■'>'■> I). Operationssaal der chirurgischen Klinik •!•! K. Kloake :'••'> F. Vorlesnngszimmer einer Mittelschule in Kanda ">■> IV. Allgemeines über die periodisciien Variationen der Keimzahlen '■'>('> V. Einiges über die Zusammensetzung der Lnftkeime *l-î VI. Zusammenfassung •'! KUREIGENDA. P. 44 Zeile ]S ist zu lesen Phycomyceton statt Tyrenunivcelen. K. SAITO. UNTERSUCHUNGEN ÜEBER DIE ATMOSPHAERISCHEN PILZKEIME. TAFEL I. Tafel I. Tafel I î^eigt eine graphisclie Dar.stellimg des monatlichen Keimgelialts der ycliiramelpilze. Dabei stehen die Teniperatnr-, Hninidität- nnd Regen- niengeknrven. Jour. Sei. Coll. Vol. XVIII., Art. 5. PI. I. n i \ i \ i >^ I. Gartenluft. 11. Strasseuluft. III. Temperatur. IV. Humklität. V. Kegenmenge. / X- / \ /' \ / \ / \ \ \ \ \ \ \ \ \ \ \ / / / "~"-~-^/ / / / / / .-^ \ \ / \ / 0 ol K. SA HO. UNTERSUCHUNGEN UEBER DIE ATMOSPHAERISCHEN PILZKEIME. TAFEL IL Tafel II. Fig. 1. Chulosporium lierbaruin. a einf'aclie, /' verzweigte Coiiidien träger ( X 400). Fig. 2. Fenic'dliuni (jlaucum. a Conidieiiträger ( x 400), h Coiiidieii ( X 560). Fig. 3. Jioft'i/tis ci7terca. « (Jouidieutniger mit Con idieii, Z> Conidieiiträger, c, d Ap[)re);Sorieii, c Conidien. o, h, c, d (x400), e ( x .360). Fig. 4, Catenularia faliijinea. a Kolonienforni, b etwas vergrüssert (^•cliemati.scli), c ]Iyj)]ieii mit Conidien. d Keimung der Conidien. (i, d (x .360). Fig. .3. Epicoccum purpurasveus. a Coiiidieiilager, b Conidieiiträger, c Hyphen kultiviert in Knop'.sclu'r NiUirlö^ung mit Jvolir/Aicker. ■ a ( X 12.3), 0, c ( X 400). Fig. 6. rerticiniuii/ glaucuvi. a Conidientiiiger, h Ende des Conidien- trägeis, c Conidien, d Gemmen in untergetauchten Hyphen. a, d ( X 400), b, c {x .360). Fig. 7. Jfticor raceiiiosus. a Natürhehe Crosse, b Conidienträger, c »Sporangium, d, Sporen, c Gemnie.n, _/'Hetes[)rossung, r/ Zygosporeii. b ( X 125) c ( X 400), d, f ( X .360), e, Conidien ( x 560). Fig. 17. Torida sp. a Conidienketten (x400), h Conidien ( x 560). Fig. 18. Pestalozzia sp. Conidien allein ( x 400). Fig. 19. Hyalopus ochracens. a Conidienträger (x 400), h Conidien ( X 560). Fig. 20. Pijhnidenhildner. a Natürliche Gn'lsse, h reife Pyknid, welche nach Anfeuchtung mit Wasser ihre Conidienmassen entleert, c das Anfangsstadium der Knotenbildung an einem Fadenstücke. h, c (x400). Jour. Sei Coll. Vol. KVUl., Art. 5. PL Ut. UTH i IMP. T>*E TOWO PftlfOING CC K. SAITO. UNTERSUCHUNGEN UEBER DIE ATMOSPHAERISCHEN PILZKEIME. TAFEL IV. Tafel IV. Fig. 21. Macrosporîum cladof^porioides. Conidienträger mit Couidicn ( X 400). Fig. 22. Heterohotrys sp. a, h Mikroconiclienträger mit Conidien, <• j\I;dvVo- conidien in Ketten, d Keimung der Mikroconidien e Keimnng der Makroconidien. Alles Fig. ( x 560). Fig. 23. Monilia sp. ( x 400). Fig. 24. DendryjMiim pemcüliatttm. Conidienträger mit Conidien ( x 400). Fig. 25. Hepedonium chrysospevnm.m.. Conidienträger mit (Jonidien ( x 400). Fig. 26. Cephalotliecitiiii roseum. Conidienträger mit Conidien ( x 400). Fig. 27. (^haeiocladmm Jolmsit. a Conidienträger, b derselbe vergrössert, c Conidien. a ( x 125), h ( x 400), r ( x 560). Fig. 28. Oospora I. a Hyphen mit Conidien, J) Conidien. a, h ( x 400). Eig. 29. Oedocephalinm crystallimim. a Conidienträger, h derselbe im jungen Stadium, a, h (x 400). Fig. 30. Glenospora sp. a Conidienträger ( x 400), h Conidien ( x 560). Fig. 31. Cylindrocephalmm sp. Conidienträger mit Conidien (x400). Jour. Sei. Coll. Vol. XYIIi.^rt5,_PI. IV. £• Sailo . 3r. YOKOYAMA ON .SOME Professor Jimbo in his " Report of the Geological Researches of Eastern Rikuzen " (M S) mentions six horizons or beds as constituting the Jurassic of that region, which counted from below are as follows : I. Cjjrena bed of 3Iaf/enosn and N'h'anohainn, a dark clayslato filled with Cyrena. II. KaraJi'uwa Clay slate, with Gastropods, Lamellibranchs and Crinoids. III. Irlnosawa Clayslate with EHheria-WVe shells. lY. Trigonia Bed, consisting of a hard, light-grey, medium- grained sandstone filled with shells of Trigonia V-costata Lycett. V. Sandy Clayslate, with Arieliles and Belemnites. VI. Clayslate, with Ammonites. Of these six beds, he considei-s the first three as the lower, and the other three as the upper part of the formation separated l)y a line of unconformability. The subdivisions subsequently proposed by Iki are similar to the above, although not quite the same. The fossils described bolow are those obtained in the beds I., IV., v., and VI. They are the following : From Bed I. 1. Cyrena hnivlata n. sp. 2. Cyrena ohlonga n. sp. 13. Trigonia hosourensis n. sp. 4. Perna rikuzenica n. sp. 5. Gervillia trigona n. sp. G. Baiera f sp. From Bed lY. 1. Trigonia V-costata Lycett. JUKASSIC FOSSILS FR03I JUKUZEN. 2. Belemnopsis s[). 3. Belemnojysis sp. 4. Belemnites sp. 1. Amiiwnlieti sp. 9 From Bed V. Animoniiea sp. 3. Beleiiinites 8 p. From Bed Y I. 1. Harpoceruii Ikianinn n, sp. 2. Schlothehnia Jimboi n. sp. 3. Lytoceras ef. Uneatii/in Schloth. Jiidgiiig from tliese fossils, there is not tlie least doubt about Bed VI being of a Liassic age. It is of course not possible at 23resent to determine exactly to ^vllat part of the Lias it Ijelongs. But as Schlotheimia is a genus hitherto confined to the Lower Lias and Lytoceras lineatum is a species only found in the Middle Lias, so it is most probable that the bed represents the lower half of the Lias, and it is also not altogether impossible that it is again divisible into two parts, the lower or Schlotheimia-hQ'àvmg bed and the upper or Lytoc&ras-hQ'AYiiv^ bed. Bed Y is also probably Liassic, as the two Ammonites con- tained in it look very nnich like those of Arietinœ. By Professor JiMBO they have even been taken for Arlel'Ues. If such be really the case, then the bed must be assigned to the Upper Lias. Bed I Y is a very important one, as it contains Trigonia V-codata Lycett, a species hitherto found onh^ in the Inferior Oolite of England. ^Vnd as a Belemnojms which occurs together with it has never been found outside of the Dogger, so we may fairly assume that the bed represents the lower half of the Doggei-, roughly corresponding to the Inferior Oolite of England. 4 ART. 6. — M. YOKOYAMA ON SOME The fossils found in Bed I are all new. But the resemblance of Perna to Perna rugosa of the Dogger and the occurrence of a costated Triyonia which has its greatest development in the same formation and of the genus Cyrena which is abundant in the Middle Jurassic plant-bearing series of other parts of Japan tend to show that the bed is still Middle Jurassic, possibly be- longing to its upper part. Beds IT. and III., although their j^ahieontological characters are not yet known, must be considered as belonging also to the Dogger, as they lie between I. and IV. From what has been said above, it is quite clear that the strata are inverted. Bed VI. which Professor Jimbo considered as uppermost being the lowest, and Bed I which was taken by him as the lowest being uppermost. According to my opinion, the so-called line of unconformability betwceJi III. and IV. may possibly be a line of fault. DESCRIPTIONS OF THE SPECIES. A. Fossils from Bed VI. 1. SCHLOTHEIMIA JIMBOI xv. sp. PI. I, Fig. ('.. The I'orm which I designate under the above name is a discoidal shell somewhat distorted 1)y pressure. Its diameter measures about o4 nnn, while that of the umbilicus is about G mm, so that the shell may be said to be tolerably widely um- bilicated. The lateral sides of the whorls are only slightly convex and ornamented with somewhat ilexuous ribs which number o8-40 JURASSIC FOSSILS FROM KIKUZEN. 5 on the last whorl. Most of these ribs rise at the eds-e of the steep, almost perpendicular umbilicus, but there are some which commence either at a short distance from the edge or, as is sometimes the case, near the middle of the lateral side. The longest of these ribs are at first directed slightly forward and then curve somewhat backward, and then again forward on the external side, where they are interrupted by a deep narrow groove, an unmistakable sign of the genus Schlotheiriiia. Our only specimen being not in a good state of preservation does not permit its strict comparison with European species. But that it is not Schlothewiia angulata to which it presents some resemblance is clearly shown by the different course of the ribs. The genus Schlotheimia is confined in Europe to the lower Lias. Locality : — Bandy Clavslate of Hosoura. 2. HARPOCERAS IKIANUM n. sp. PI. L, Fig. 5. The shell fiattened and furnished with a distinct keel. The whorls are tolerably convex and about S involute. The umbilicus is small being only <> mm. in a shell Avhose diameter measures 20 mm. The umbilical wall is steep. The lateral sides of the whorls bear distinct ribs which are about 40 in number on the last volution. Of these 40 ribs, only about one half rise at the umbilical edge, the others being formed by their bifurcation at about * the distance from the umbilicus. But now and then between the bifurcating ribs there is intercalated a simple rib which does not reach the umbilicus. The course of the ribs may 6 ART. 6. — M. YOKOYAMA OX SOME be called weakly falcate, as they curve a little backward after bifurcation and then bend strongly forward near the external keel. This form closely resembles the IMiddle Liassic species Harpoceras lytlieiiae Youxg, but it is at once distinguished from it by its bifurcating ribs, in which respect it is not unlike some of the many varieties of Harpoceras aalense Zieten. A single cast of an external surface of the shell was obtain- ed in a sandy clayslate of Hosoura, our ligure having been drawn from a plaster cast of it. 3. LYTOCERAS cf. LINEATUM Schloth. PL II. Fig. Ö. An external impression of a Lytoceras found in a clayslate of Aratohama, though water-worn and imperfect, shows characters so peculiar to the group of Lytocevaa fi)iibrlalum Sow. of the Lias, that I do not hesitate to bring it under that group, and indeed very near to Lytoceras Imeaiimi Schlotii. The last volution of the shell measuring about 70 mm. in diameter is provided with many fine ribs which at certain intervals arc coarser and decidedly fimbriate. The fimbriation of the ribs becomes more frequent as we approach the shell-mouth near which almost every rib is fimbriate. The non-fimbriate ribs arc very fine, thread-like, more or less straight, and in some cases showing a tendency to fimbriation. Besides those above mentioned, there is a single wall-like elevated rib near the middle of the last half of the last whorl. JURASSIC FOSSILS FROM EIKUZEN. 7 Although our specimeu lacks the outer portion of the whorl near tlie mouth, yet the characters given ahove agree so well with those of Lytoceras liîieatum that it must be at least closely allied to, if not quite identical with, this European species. Lytoceras lineatum Schloth. occurs in the ]Middle Lias of England and Germany. Our figure was drawn from a pi aster- cast. B. Fossils from Bed V. 1. AMMONITES s p. PI. IL Fig. 11. This is an external impression of a flat discoidal shell, about 4Ö mm, in diameter with a wide umbilicus of about 20 mm, and furnished with coarse, forwardly bent ribs about 33 in number in the last whorl. It closely resembles some of the forms of Arietinse. Locality : — Hosoura. The figure was drawn from a plaster- cast. 2. AMMONITES sp. PL II. Fig. 10. Another impression of a discoidal, widely umbilicated shell, about 35 mm. in diameter, which may be more or less closely allied to the preceding form, but its ribs (36 in number) are finer, more rigid and sharper. The figure was drawn from a plaster-cast. Locality : — Hosoura. 8 ART. 6. — M. YOKOYAMA ON HOMiù 3. BELEMNITES sp. PI. II. Fig. 12. A single specimen. It is provided witli a part of the pliragmocone and is lanceolate in shape, measuring about 37 mm. in length, and about 7 mm. in breadth near the upper end, reminding one of Belemnites acutus Miller of the Lias. However the preservation is such that a strict specific determination is at present impossible. Locality : — Hosoura, in a sandy clay slate. C. Fossils from Bed IV. 1. TRIGONIA V-COSTATA Lycett. PL II. Figs. 2, 3, 4. Lycett. A Monograph of the British Fossil Trigoniœ, p. m, pL XIII fig. 5, pi XVI fig- 1-4. A Trigonia lielonging to the group of Undulatseand present in numerous specimens coincides so well in its characters with the species above named from England that I have not the slightest doubt about the identity of the two forms. The most important character of the species lies in the formation of a V-shaped angle by the meeting of the subtuberculated anteal and posteal ribs near tlie carina. This angle which is particularly distinct near the ventral margin of the shell gradually passes into a curve towards the })eak. Most of our specimens are small, being about the size of those figured as smaller ones by Lycett. Fig. 4 shows the inner JURASSIC FOSSILS FROM RIKUZEN. 9 side of a right valve on which is seen a rounded ridge on the posterior margin just below the upper edge and parallel to it. The species occurs in England in the Inferior Oolite. Locality : — Hosoura and Niranohama, in a greyish sandstone. 2. BELEMNOPSIS sp. PI. II. Fig. 1. There are several specimens of a lanceolate Belemnite, one of which is about 35 mm. long and 8 mm. broad near its upper end. They all show a deep ventral groove running down to the apex. Therefore, although much water- worn, it is quite certain that they belong to the subgenus Belemnopsis. Locality : — Hosoura. 3. BELEMNOPSIS sp. PI. II. Fig. 6. A specimen of another form of Belemnopsis, the lower end of which is not preserved, is much larger than that of the pre- ceding one. It seems to be broadest a little below the alveole where it measures about 12 mm. in diameter, thence slisjhtlv tapering upward. Locality : — Hosoura. 4. BELEMNITES sp. PI. I. Fig. 7. This fragment of a large Belemnite more than 20 mm. in 10 ART. 0, M. YOKOYAMA ON SOME diameter has the lower end of the phragmocone intact. The transverse section of the rostrum is shghtly oblong. Locality : — Hosonra. D. Fossils from Bed I. 1. BAIERA? SP. PI. I. Fig. 2. This is a fan-shaped leaf wanting the upper end, l)ut l(wking very much like the basal portion of either a Baiera or a Phœnicopsis. Fine, longitudinal, simple veins constitute all that can be observed in the leaf. Locality : — Hosoura-. 2. CYRENA LUNULATA x. sp. PI. II. Fig. 9. Shell moderately thick, ovatelv triangular, slightly inequi- lateral, subrostrated both in front and behind, with curved ventral margin. Beaks prominent, touching, and directed forward. Surface with fine but unequal concentric strise. In front of the beak is an ovate lunula al)0ut twice as long as broad and bounded l)y a fine deep groove running from the beak to the anterior margin. Behind the l)eak there is a sharp keel running to the posterior margin and enclosing a lanceolate valley-like area. Proportions of length to height and thickness are as 10 to 8 to 4. The figured specimen measures about 25 mm. long, 20 mm. high and 11 mm. thick, and may l)e taken as representing the average size of the shell. Locality : — Ilosoura. Very numerous. JITKAS.SIC FOSSILH FROM KIKUZEN. 11 3. CYRENA ELLIPTICA x. «r. PL I. Fig. 4. Shell elliptical, In'uader than high, somewhat inequilateral, with anterior and posterior margin rounded and ventral margin curved. Beaks submedian, prominent, touching. Surface only with fine concentric striations. Pi'oportions of length, height and thickness, 10 : (S : 4. This form is a, little larger than the preceding one, and seems to have been com^jaratively thin shelled. There are specimens larger than the one figured. , Locality : — Hosouia. Earer than the foi'egoing species. 4. TRIGONIA HOSOURENSIS n. sp. PI. I. Fig. 3. A few specimens of a Trigonia belonging to the Group of Costatœ were obtained from the Cyrena slate of Hosoura which, though imperfect, show sufficient characters to be created into a, new species. The shell is ovately triangular, flattened and strongly inequi- lateral. The anterior side is very short with its border rounded. The posterior side is long and produced. Beaks prominent, and recurvate. The surface is ornamented with about 17 smooth concentric ril)s Avhich near the marginal carina descend with a veiy steep slope so that the curvature is stronger than in any of the known species of this group. Escutcheon narrow and. lanceolate. Area depressed, and smooth. The group of Trigonia costaia has occurretl hitherto only in the Oolite. 12 ART. 6. M. YOKOYAMA ON SOME o. GERVILLIA TRIGONA n. sr. PL n, Fig. 7, 8. The shell is triangular, inflated and smooth. The l)eaks are prominent, close to the anterior extremity of the straight hinge- line, in which there are four ligamental pits, three of which are just l)elow the l)eak, while the remaining one is situated a little behind it. They are separated from one another by nearly equal distances, although differing in shape the two middle ones l)eing longitudinally elongated, the 2)Osterior transversely elongated, and the anterior obliquely elongated, the upper end being away from the others. The hinge line is anteriorly eared and posteriorly produced forming a large wing pointed at the end. The posterior border is nearly vertical and slightly concave below the hinge- l)order, while the anterior border which is also somewhat ex- cavated in the middle recedes obliquely to a short rounded ventral border, so that the shape of the shell is decidedly triangular. The ])roportions of the various diniensions of the shell are best seen in a cast drawn in fig. 7, in which the length of the hinge l)order measures 28 mm., the height of the shell 25 mm. and the thickness of the left valve U) nun. The specimen shown in fig. 8 is more than double the size of the above, but it looks broader than it actually is, as it has l)een distorted by pressure. The posterioi- wing of this species reminds one of the genus Pteropeima occurring in the Lower Oolite of France and England. Locality : — Hosoura. Tolerably frequent. JURASSIC FOSSILS FROM RIKUZEN. 13 6. PERNA RIKUZENICA n. sp. PI. I. Fig. 1. Several specimens of a large quadrate Perna were obtained, which are however more or less broken. The figured specimen has both valves preserved and measures 8 cm. long, 6 cm. broad and 3 cm. thick. The shape of the shell, which is decidedly four-sided and has the anterior border concave resembles that of Perna rugosa Goldf. figured by Morris and Lycett in their Ponograph of the Mollusca of the Or eat Oolite {pi. XIL, fig. 1 ; pi. XIV., fig. 16). But on a closer examination our shell differs in several points from the European. In the first place our shell is a little more inflated ; in the second place the greatest inflation lies very close the anterior border, so that the surface descends abruptly to it, while towards the posterior border the slope is very gradual ; and in the third place the surface is not rugose, being marked only by fine concentric lines of growth. Localitv : — Hosoura. >I. YOKOYAMA. ON SOME JURASSIC FOSSILS FROM RIKUZEN. PLATE I. Plate I. Fig. 1. Pcrna rikuzenica n. «}). a lateral view of a right valve ; h front view of both valves. Fig. 2. Baiera ? s[). Fig. 3. TrUjonia hosourensis ii. sp. a Left valve ; h right valve, e showing area and escutcheon. Fig. 4. Cyrena eUiptica n. sp. a right valve ; h front view of the same. Fig. 5. Uarpoceras Ihianwn u. sp. ct lateral view ; h view of the external side. Fig. 6. ScMotheimia Jiinhol n. sp. (f lateral view ; b showing the groove of the external side which is narrowed by pressure. Fig. 7. Jielannites sp. a and h seen from opj)osite sides, h showing a part of the phragmocone. Jour. Sei. Coll. Vol. XVIII.. Art. 6, PL I. Flg. 6. ^1^ S . V L M. Yohnyama : On Some .Juruftsic Fossils from lîikuzm. 51. YOKOYAMA. ON SOME JURASSIC FOSSILS FROM RIKUZEN. PLATE IL Plate II. Fis;. 1. Belemnopsis sp. A water-worn specimen ; a showing the ventral groove and alveole ; h the same seen from the opposite side. Fig. 2, 3, 4. Trifjonia V-costaia Lycett. 2 left valve. 3 right valve. 4 inner side of a right valve. Fig. 5. Lytoccras cf. lineatum Schloth. Fig. G. Bdemnopsis sp. a showing the ventral groove and alveole ; J> the same seen from the opposite side. Fig. 7, 8. Gervilla ivigona n. sp. Fig. 1 , internal cast. Fig. f). Cijrea lunulata n. sp. a lateral view ; J> seen from front ; c showing lunula and area. Fig. 10. Ammonites sp. Fig. 11. Ammonites sp. Fig. 12. Bclemnites sp. Fiq. 1. Jour. Sei. Coll. Vol. XVIII., Art. 6, PI. II. Fig. 4. M. Yokoyama : On Sotne Jurassic Fossils from Rikuzen. JOURNAL OF THE COLLEGE OF SCIENCE, IMPERIAL UNIVERSITY. TOKYO, JAPAN. VOL. XVIII., ARTICLE 7. Studies on the Hexactinellida. CONTRIBUTION IV. (MosseUidce). By Isao Ijima, %•• Ph- D.. Rig. -Hah.. Professor of Zoology, Imperial T'niversitv, Tokyo. With 23 plates. The Rossellidfe I define as follows : Lyssaeine Hexasteropliora of cup-like or sacciform body; sometimes stalked; generally firmly attached at base and exceptionally rooted by basal processes or by tufts of basal spicules. Besides the main terminal osculum, a few secondarily formed oscula may occur. Ectosomal skeleton composed of small rough dermalia with a variable number of rays and of large liypo- dermalia. The latter are generally pentactins which often show a tendency to protrude outwards in such a way that the paratangentials form a veil-like covering- over the dermal surface. The dermalia, when hexactinic, have the distal ray not pinular but much like the rest in appearance. The gastralia are generally rough 2 ART. 7. — I. uniA : IIEXACTINELLIDA, IV. liexactins; without pentactins as hypogastralia. The hexasters are variouR but mainly oxyhexasters and dis- cohexasters, tliese sjenerally occuring together; but sometimes one kind occurs to the exclusion of the other. Oxyhexasters are often hemihexactinose and hexactinose. Diseohexasters may be modified into discoctasters. In the alwve, I have slisrhtlv modified the dia2:nosis of the same family as given in my Contrib, III., p. 114. In '97 F. E. Schulze distinguished three subfamilies under the Rossellida?, viz., the Eossellinte, the Lanuginellinœ and the Acanthascinœ. To these I added in 'g8 a fourth under the name Leucopsacinae ; but since I have later (Contrib. III.) given to this group the status of a distinct family, there remain F. E. Schulze's three suliflimilies al)ove-meiitioned to make up the Rossellid^e. As was pointed out by that writer, they may be distinguished from one another by the absence or j^i'esence of strobiloplumicomes or of discoctasters, thus : ". — Without discoctiister. rt'.— With strobilophimirome... A, Lanuginellinœ. h'. — Without jstrobilophimieoine B. RusmlUnœ. h. — Witli (liseootastor ; no stroliilophimioomo C. Aeanthascinœ. A. LANUGINELLIN.^. Pentactinic hypodermalia always present. Gast- ralia, hexactins. Parenchymalia consist of diactins and of large or medium-sized hexactins. Strobilo- plumieome always present among the hexasters, which A. LAXUGIXELLIN.F.. Ö for the rest consist of either discohexasters or oxy- hexasters, or of both; without the discoctaster. Uijf'erentuil Ketf to tltc Knotrn (Jcuerri und Species. "• — Finiily attaclied at base to solid substratum. a\ — Without tufts of prostal diactins. Dernniliu stauiactinic. Only diiL(jhexasltrs present besides strobiloplumicomes 1. LanugiiieUa pupa O. >Schm. (N. Atlantic; Polynesia; Sagami 8ea). b\ — Willi tufts of jjicistal diactins. Dermalia pentactinic and hexactinic. Only oxy- hexasters present besides strobiloplumicomes 2. Ccili/cosiima validuni F. E. ScH.* (SE. of Mass.). ''. — Rooted in loose bottom by tufts of pentactinic anchor-needles. e'. — Both oxyhexasters and discohexasters present besides plumicomes. Shaft of anchor- needles barbed in the distal part. Body-surface veiled '.'>. MeUomjmpha velala(W. Thoms.). (Strait of Giltraltar). (iK — Only oxyhexasters present besides plumicomes. Shaft of anchor-needles smooth. Body-surface not veiled, but witli tufis of long prostal needles Lophocalyx. a'. — Body cup-like. Üermalia stauractinic. Oxyhexasters normal, with terminals sliorter than principals 4. Lophocdlyx philippinensis (J. E. Gray). (Philippines; Polynesia). 6'. — Body irregularly shaped. Dermalia, convex stauractins and pentactins ; the latter having the unpaired ray directed distad. Oxyhexasters normal, heaii- hexactinose and hexactinose; terminals about as long as principals 5. Jjophomhjx i>pinosa F. E. Sen. (W. of Andaman Is.). Of the above-iueiitioned four genera and five species referable to the subfamily, Lanuginella pupa is the only form that has as yet been discovered in the Japanese seas. A description of it after niv own studies follows. LANUGINELLA PUPA (J. ?Schm. Contrib. III., PI. V., figs. 1—7 ; and this Contrib., PL I. (). Schmidt, 70, p. 13 ; PI. XL, iigs. 1,3. — S. Kent, •70, p. 247 ; PI. LXV., figs. 1-6.— F. E. Schulze, '86, " For the grounds of placing here this genus and species, originally described b}' F E. ScH. ('99^ :'^ ^"1 Asconematid, see my Contrib. III., pp. T.l-So. 4 ART. 7.— I. IJIMA : HEXACTINELLIDA, IV. p. 47.— F. E. Schulze, '87 (!)=•=, p. 180; PI. LUI., %s. 3-Ö.— F. E. Schulze., '97 ( ! ), p. ö48.— E. Topsent, '95, p. 213.— I. Ijima, '98, p. 44. The species has long been known from St. Jago, one of the Cape Verde Islands (O. Schm.), from the coasts of Spain and Portugal (S. Kent,) from the Strait of Gibraltar (924 m.; Tops.) and from oft* Little Ki Island (236 m.; " Chall."). It occurs also in the Sas-ami Sea, so that it seems to be a verv widely dis- tributed species. In the Sagami Sea, in all nearly a score or more specimens have thus far been obtained at Döketsba, Mochiyama, Inside and Outside Okinose and at a. spot off' the east coast of Vries Island. The depths from which they were obtained were 183-572 m. (100—313 fms.). The species is known to occur there along with Eaplectella inarshalli, Leucopsacxs oHhodocus, L. scoliodocus, Slauro- calyptus pleorhaphides, Grotcroinorplia mojeri, Hyalonema affine, Semperella domata, Farrea. sp., Ilexacluiella lorica, etc. Tlie species is always lirmly attached at base tu the sub- stratum. I have seen it growing on pebbles and shells as well as on dead remains uf a coral, of a Bryozoa, of a Lithistid, of a Farrea and very fieipiently on Hexact'inella loriea. In the Challenger Eeport (p. 131), F. E. Schulze mentioned tliat the sponge might sometimes be rooted in a soft bottom by means of long anchors ; this mention, so far as it goes, was probably founded upon observations, not on the present species, but on young Lophoealyx ph'dippinemis, specimens of wliich, as we are told by him, were contained in the same bottle together * In the list of literature given for each specie-^ treated of in tliis Contribution, tlie principal or the more important work or works are indicated by an exclamation mark in piarentlicsis. LANUGINELLA PUPA. Ö with the Lanuginella pupa examined by him. In fact, a case of one and the same Hexactinellid species being firmly fixed when growing on hard substratum but producing a root-tuft when living on a soft bottom has never as yet been shown to exist. The general shape of the sponge is ovoid or sjiherical, usually contracted below into a short stalk-like base. The bodv in laroer specimens is often laterally compressed to a perceptible degree. Its small size led Carter {'73 c/, p. 283 ; '73 b, p. 359) to suspect that the species was based on young specimens of a larger sponge. It seems the species never attains a large size, — a size larger than, say, a large acorn or a hazelnut. One of the largest specimens I have measured was ovoid in shape, measuring 19 mm. in height and 11 nun. by 14 mm. in breadth at the broadest part. It had at the upper end a roundish osculum, o mm. in diameter. AVall as thick as A2 mm. in the middle of the body. Another large specimen, likewise ovoid in shape, measured 22 mm. in length. The specimen shown in natural size in fig. 1, PI. I., measures Kj mm. in height and 13 mm. bv 9 mm. in «-reatest breadth. The osculum, 3 mm. in diameter. Thickness of wall, as much as 4 mm. — The specimens of figs. 2 and 3 measui-e respectively 17 mm. and 10 mm. in height. The osculum situated at or near the upper pole is always comparatively small ; it is roundish or oval and has a thin smooth edge, never supplied with marginal prostals. It leads into a gastral cavity which is either pit-like or but slightly expanded internally, on account of the considei-able thickness of the bodv- wall in the middle. — F. E. Schulze found small young specimens of 2-3 mm. diameter with the osculum still unopened. That negative condition may sometimes, but certainly does not ahvavs, 6 Airr. 7. — I. ijiMA : hexactinellida iv. persist even wlieii the sponge lias attained the size of a pea or of a bean. This is attested by two specimens of the size in- dicated, which I have found in my collection. In both the sponge- wall is simply thinned out, without being broken through, at the point where one would expect tlie formation of the oscular aperture. Though I have not examined them on sections, I believe that the wall there lacks the chamber-layer as represented i)y F. E. HcHULZE in his fig. 5, PI. LIII. {I. c), and further that, in the absence of a single large aperture, the discharge of water from the gastral cavity must have taken place through the intertrabecular gaps of that part of the wall. All the I'est of my specimens, including the smallest of G or 7 mm. height, show an open osculum. In some specimens I have seen the external surface covered partly or nearly all over with a veil, formed of small hypodermal pentactins which had protruded through the dermal layer. Ho, for instance, in two of the specimens figured on PI. I. All the smaller specimens are without the veil. Not that all the larger specimens are provided with it ; on the contrary, the two largest specimens, respectively 19 mm. and 22 nun. high, seem to show^ J 10 trace of it. Presumably it is formed only under certain circumstances after the sponge has reached maturity. In forming it, the hypodermal pentactins stand out isolatedly but usually at such intervals that their para tangential s are nearly or quite in touch with one another. Their shafts are exposed over the dermal surface to a length of 2 — 1 mm. The real derjnal surface is smooth. A\^hen dried it is some- what shin}', which is however by no means a peculiarity of the genus or of the species. The reflection of light takes place principally, if not solely, from the smooth Ijardened surface of LANUG INELLA PUPA. 7 tho dermal membrano. Seen under the hand-lens, the dehcate dermal lattice^York contains minute meshes whicli are more or less regularly rectangular in shape, though in certain individuals I have found tliem irregularly shaped throughout nearly the entire extent of the layer. With the larger-meshed hypodermal lattice, it is quite usual that the intersecting of the thin laths takes place in an irreo'ular manner ; only seldom are the meshes of a rectan- gular shape. Generally speaking, the ectosomal skeleton seems to lie in tolerably close apposition to the choanosomal mass, so that the subdermal space is but inconspicuously developed. The en- trances into incurrent canals, indistinctly visible from the outside, are small ; the larger of them are separated from one another by a comparatively wide interspace. Towards the stalk-like base of the sponge the hypodermal lattice becomes unnoticeable. It is here rejilaced by parenchymal bundles running in the main longitudinally and with these the ectosomal layer is apparently in direct contact. The gastral cavity, instead of being lined with a continuous gastral or endosomal layer, shows the apertures of excurrent canals freely open. Many of the apertures are much larger than those of the incurrent canals in the same specimen. They may measure 2mm. across in fully grown specimens. Spiculation.-' Tho parcnchyrnaUa consist of oxyhexactins and oxydiactins. In both these forms the rays run out to a sharp point and are either smooth all over or show a roughness near the end. * I am intiebtcd to I'roiessor F. E. ScnULZK for a slide-preparation of Lanuginella pupa, — presnma})iy of oiip from tiio Oliallonger enlloetiDii. Tt lins heen useful fir the piir[)o«e of comparison. 8 ART. 7. — T. IJIMA '. HEXACTINELLIDA, IV. The oxyliexactins are numerously represented in tlie paren- chyma and play an important part in the formation of the sup- porting skeleton. Though somewhat variahle in size, most of them are of moderately large dimensions. They may l:>e so large as to present an axial length of nearly 2mm., the thickness of rays reaching up to 30/^ near the central node. All the six rays are not always of the same length. As a general rule, the spicules lie with one of the axes directed radially and are met with in a layer, — at places in a few ii-regular layers, — in the thickness of the choanosome. The oxydiactins likewise occur in considerable numbers. They are here decidedly more numerous than in Leucopsacus, but not so numerous as in the 2;eneralitv of Kossellid members in which diactins form the predominant, if not the only, megascleric ele- ments of the parenchymalia. In the present species, the spicules in question are mostly thin and small, not exceeding 22/;! in thick- ness near the center, which may or may not be externally marked by an annular swelling. Some are seen to run in company with the radial rays of parenchymal oxyhexactins or of hypodermal pentactins ; but the majority seem to j)ursue a in ore or less paratangential ly directed course, either isolatedly or arranged in bundles, which are strongest in the stalk-like basal region of the sponge. Here a diactin may reach a length of 3 or 4mm. Close to the attachment-surface there occur a number of stout-rayed and ytrickly-surfaced hexactins of about 100 /'• axial length, forming a thin layer. The same spicules are occasionally pentactinic and even stauractinic. There can be no doul)t that we have here to do with the basidictyonalia. Some of the spicules are loosely disposed ; others, especially those in direct contact with the substratum, are joined together by synapticulœ in an irregular LANUGINELLA PUPA. 9 manner, formin«;' n eontinnons basal plate in wliieli the iiidividnal spicules ean without diiticulty be recognized in their proper forms. The Jiypoderiiial'ta closely resemble the parenchymal oxv- liexactins in character, except in being always devoid of a distal ray. The cruciate paratangentials, 1 mm. or over in axial length and up to 34 !'■ in thickness near the center, are usually in a slightl}" convex plane in conformity with the curvature of the body- surface. They may exhibit a sparing quantity of obsolete micro- tubercles near the pointed end (PL L, lig. ('») or may be quite smooth throughout. The unpaired proximal ray, more generally smooth all over, is straight and ma}' be nearly three times as long as the paratangential. As before mentioned, the hypodermalia in certain specimens are protruded through the dermal layer, their paratangentials thus forming a veil over the external surface. In one such veiled specimen, — namely, the one depicted in PL I., fig. 1, — I have found most of the prostalia to be rough-surfaced on the paratan- gentials (PL I., fig. 7) as well as on the shaft, but the latter only for a short distance from the spicular center. This shagreen- like roughness, which is known to exist also on the same spicules of several other Rossellids, is due to minute, fine, erect and closely set processes. On the paratangentials, the processes are most pronouncedly developed on the outer surface ; laterally they become obsolete, leaving the inner surface along the middle line nearly smooth. Xot that all the prostal pentactins are shagreened in the manner described, for some are quite devoid of this characteristic. On the other hand, among the hypodermalia, /. c, the prostalia before protrusion, there occasionally occur such 10 ART. 7. T. IJI]\[A : ITEXACTINELLIDA, TV. as show the same roughness of siirfnco. It then seems that this shagreen-like character, beginning to arise, whenever it occurs, while the spicule is yet hypodermally situated, does not constitute a constant peculiarity of all old hypodermalia. This opinion is also supported by the fact that the specimen referred to is the only one in which I have seen the rougli peutactins, while in all the rest I have failed to find tlie same sjiicules, whether as prostalia or as hypodermalia, characterized in tlie same way. The (lermnlia (PI. I., fig. 4) are rough stauractins, exception- ally and very rarely tauactins ; they are nearly flat or perceptibly convex on the outer side. The rays taper 1)ut little towards the end which is rounded. The prickles on the surface are erect and generally tolerably conspicuous, but are subject to a consider- able variation in this respect according to individuals, as are also the spicules in respect of their size. In most specimens the axial length averaged 220 /'- (the maximum being 280 f) in length and 7 /^ in breadth near the central node. In one specimen, however, I found the average axial length to reach up to 330 f- (the maximum Ix^ing 374 !') ; the rays tapering gradually towards the conically or obtusely pointed end and being beset with rather inconspicuous prickles. F. E. Schulze ('97) had given 160-200 />« as the size of the dermalia. — It is by no means rare to meet, here and there among the dermalia, with perfectly smooth, small and unusually thin-rayed oxystauractins, such as are shown in PL I., fig. "). They are evidently dermalia in an incomplete stage of development. The gadrcdia (PI. I., fig. 8) are regularly shaped oxyhexact- jns, measuring 220 — 330 /^ in axial length and Ih /" in average LANUGINELLA PUPA. 11 breadth of rays near the central node The surface is uneven on account of obsolete niicrotubercies. The freely projecting rav is in no way differently characterized from all the rest. The gastralia occur abundantly but do not form a continuous lattice- work. Oxyhexactins of precisely the same appearance also occur in scattered distribution along the inner surface of excurrent canals — especially of the larger excurrent canals — as the canalaria. (See f]g. 9). The kexasters consist of the discohexaster and the strobilo- plumicome. The discohexaster (PI. I., figs. 11 and 12 ; also Contrib. III., PI. v., figs. 1-6) occurs commonly throughout the entire sponge- wall. It shows considerable variations in both size and appear- ance in the same as well as in different specimens. In general it may l)e said to be spherical or approximately spherical, in that the terminals so diverge peripherally that all the terminal discs are situated neai-lv equidistant from one another. In general appearance it most closely resembles certain discohexasters of Chaunopleclella and of Leucopsacus scoliodocus. Tiie diameter ranges from 40 /< to \)() r- (according to F. E. Schulz f, 132-100 /'-). In some s])ecimens of the species, however, the largest discohexaster does not exceed 70 ,"- in diameter. Aside from certain exceptional cases, each short or very short principal bears 2-4 or .") (most commonly 3) terminals. The number may vary with (liH'erent principals in one and the same rosette. The ter- minals ai)pear to l)e rather strong — at any rate not quite thin except in the ease of the smallest discohexaster ; they are either of about the same thickness tliroui»;hout the entire leuiith or thicken slightly towards the outer end. Tlieir surface, when 12 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. seen under ii high magnifying power, is rough on account of microtubercles (PL I., fig. 12). In some cases, tlie roughness of the surface is more pronounced than in others, the microtubercles being then visible as minute reverted prickles or barbs. The convex terminal disc, when well developed as is generally the case, exhibits comparativelv strong recurvate marginal jirongs, Ü-G in number. Occasionally the ]irougs are minute or quite obsolete ; they may even be ajiparently wanting, in which case the terminals appear as if ending bhuitly, iu all probability rejiresenting a developmental stage |)revious to the formation of the terminal prongs. In rare instance I liave seen small and delicate oxyhexaster-like forms, which I consider to be likewise an early stage in the genesis of the discohexaster. One individual (Mus. No. 436) of the species from ( )utside Okinose, which I have studied with special tlioroughuess as regards the spiculation, requires to be here particularly mentioned. In it I have discovered, though certainly as rare al)normalities, cases of the discohexastei' in which some — not all— of the termi- nals are bent outwards and backwards, directly after their origin from the principal, in a semicii'cle or in a comma-like manner, apparently with no dcliiiitc rule as to the I'elative orientation of the [)lane of curvature. They arc not unlike the peculiarly twisted oxyhexasters figured by F. E. Schulze from Jkithydonis farts of the choanosome. In one specimen of the species, the rosettes in ijuestion measured in diameter 34-42 // ; in another, 45-50 i'- ; and in still another 50-7G !'■. 14 ART. 7.— I. IJIMA : HEXACÏIXELLIDA, IV. Structurally and in appearance the rosette is quite similar to the same of Synipagella (Contril). III., ]^. 106). The hemi- spherical knob, bearing the featlier-duster-like bunch of delicate terminals, usually shows at the center of the convex outer surface a small process, into which the axial filament is seen to extend itself. The process tlien is a direct continiuitioii of the principal and represents the outer end of the primary ray of a hexactin ; the terminal-bearing knob and the terminids with it are then to be considered as secoJidary structures that have developed along the course of a primary my, not at its outer end. This may l)e held as an indication that the strobiloplumicome is a hexaster swi _y;, having arisen independently of the ordinary discohexaster and oxyhexaster in which the terminals a])pear to be ah origine at the very ends of primary hexactin-rays. The central process aljove I'eferred to may be mure or less atrophied and may even disappear altogether. Thus, several instances in the present species have come under my observation in which the process was reduced to a mere acuminous })oint, the entire knob in shape presenting a resendjlance to an acorn. In some other exce])tional cases I have convinced myself, as I did also with some of the same rosettes in Syiiipagella anoniala, that the central process was totally wanting. Soft Parts. All that I have observed as regards the soft ])a]-ts of tliis sj)ecies may be referred to in l)rief. In iig. *•>, PI. 1., the soft parts are shown as seen in a section under a low power of the microscope. The shading given to the chaujl)ers is somewhat artificial. With respect to fig. 10, which is meant to illustrate the highly inagnified appearance of LANÜGINELLA PUPA. 15 the chamber-wall as well as that of areliseocytes and of trabeculie arising from the chamber-rim, I must state at once that it is to a great extent a failure, in part due to the unsuccessful lithograph- ing and in part to the highly unsatisfactory state of the origi- nal preparation.'-' The chambers are shallow and cup-like or long and thimble- like or tubular. Their diameter, 77-132 r- ; on an average 100 ,"• Length, up to 440 />«. The longe.^t are found at the blind ends of excurrent canals, in the periphery of the choanosome ; they may sometimes present a lobed or branched appearance. All the more deeply situated chambers are cup-like. In all my prepara- tions, the cham])er-wall appears at the best as a faintly stained reticulum with minute irregular meshes. The nuclei are not discernible under ordinary powers of the microscope ; Init by using an immersion system they can, under favorable circum- stances, be recognized as ill-defiued spots found at short intervals and measuring not more than 2 /< across. In staining capacity they differ scarcely at all from the substance of the reticulum. Flagella seem to be in no case preserved. The trabecuhp are developed in moderate, and in some individuals in very great, abundance, (^n the external sponge- surface they are frequently spread out in a film-like manner to form the perforated dermal membrane. The nuclei are minute but distinct, l)eing well stained as usual. Not seldom have I seen, hanging on the trabecuL'e, homogeneous fat-like spherules which stained well with both carmine and hœmatoxylin ; they were no doubt the product of the thesocytes. * Many of the plates, now issued in liiis Contribution, were prepared and printed several years ago in the early period of ray studies anfl therefore contain shortcomintr^ of which I am more conscions than ever. For them I hcg to ask a lenient judgment. 16 AKT. 7. 1. IJIMA : HEXACTINELLIDA, IV. The arcliGeocyte? are as usual deeply stained and on the whole somewhat larger than trabecular nuclei. They do not exceed ok p. in diameter. Though sometimes found isolated, they more commonly occur in loose or compact groups varying greatly in dimensions, always among, and on the outer side of, tlie cham- hers. Many of the groups arc of a quite conspicuously large size (PI. T.. fig. 9). B. ROSSELLINiE. Pentactinic hypodermalia generally present ; ex- ceptionally wanting. Gastralia, liexactins; exceptionally pentactins. Parenchymalia, chiefly diactins and includ- ing medium-sized or small liexactins, or exclusively diactins. Hexasters consist as a rule of oxyhexasters and discohexasters ; the latter often in more than one variety, but may be totally absent; lacking both stro- 1)1 loplumicome and discoctaster. The following is a list of all the genera and species which 1 consider to make up the subfamily as it stands at present. 1. Jio>iHfUa aniavvtlca Carter. { — Acantkamm gro^mlaria Y. E. 8eh.). (S. of Kerguelen Is.; SE. of Prince Edwards Is.; Possesion Is.; 2-j6-ü49 m.). 2. //. <>'. — Body saccular or vase-like, witliout stalk. Discoliexaster in two (large and small) varieties Aiilomccus Ij. (Nos. 22-24). b. — With pentactinic hypodermalia. r;>. — Hexaster consists of oxyhexaster only (without discohexasler). a*. — Paratangentials of hypodermal pentactin, spiny. All oxyhexasters hexactinose. Aphorme F. E. Scir. (No. 27). 6'. — Paratangentials of hypodermal pentactin, not spiny but smooth and rough at ends onlv. Oxyliexasters, some or all normal- EOSSELLIN^i 19 a*. — Trichaster present besides ordiuaiy oxyliexaster with larbed terminals. Trichasterina F' E. ScH. (No. 26). i*. — "Without trichaster. Oxyhexaster terminals rough or nearly smooth Bathydorus F. E. ScH. (Nos. 28-oo). d^. — Hexaster consists of oxyhexaster and discoliexaster. c'. — Deriualia, pentactins with the unpaired ray distally directed Asconema S. Kent. (No. 25). d^. — Dermalia variable; generally without distally directed ray, which, if present, as it sometimes is, invariably belongs to a hexaster Jand not to a pentactin. c'-\ — Body smaller than acorn-size; parenchymalia include a considerable ([uantity of relatively large hexactins VitroUida Is. (No. 0.) (Z^î.— Body much larger; parenchymalia, diaclins only or may include medium • sized or small hexactins. «*. — Gastralia not hexactins, but pentactins or other forms witlxout free proximal ray. Sponge M'itli distinct stalk Craieromorphu F. E. ScH. (Nos. 11-15), 6*.— Gastralia, hexactins. Sponge generally without, but sometimes with stalk. a*. — Pentactinic Jiypodermalia i)artly witli spiny and partly witli smooth paratangentials. Rooted in loose bottom by basal pro- cesses Schaudinnia F. E. ScH. (No. 10). i'-— All pentactinic hypodermalia with smooth paratangentials rough at ends'only. Firmly attached at base. o^. — Discohexaster in one small form of moderately uniform size Hyalasciis Ij. (Nos. 16-18). i". — Discohexaster in inore tlian one form and diflering in size ; at any rate, the smallest being of about half tiie diameter of the largest, or even much smaller. a'. — Discoliexaster distinguishable into three varieties, of which the largest has the terminals to each principal arranged in a narrow or a ) erianth-like tuft. Other varieties spherical Rossella Cart. (Nos. 1-4). h'. — Discohexastcrs all spherical, and without the largest form mentioned under o' Scyphidium F. E. ScH. (Nos. 5-8). Further treatment is given, in tliis contribution, to only those genera and species knoNvn to me from Japanese waters. 20 ART. ". 1. IJIMA : HEXACTINELLIDA, IV. SCYPHIDIUM F. E. Sch. Saccular or vasiform, tliick-walled, firmly attached by contracted base; showing a disposition to produce buds on the wall. Parenchymalia exclusively diactins. Pentactinic hypodermalia with paratropal or regularly cruciate, smooth or finely shagreen-like paratangentials, without spines. Dermalia, stauractins or pentactins, or both ; rough-surfaced. Clastralia, regular hexactins with rough rays; forming a continuous layer over ex- current canalar apertures. Oxyhexaster normal, hemi- hexactinosc or hexactinose. Discohexaster distinguish- able into two forms differing in size, but botli being s})herical in shape. These two forms may grade over into each other; at all events, the smallest discohexas- ter is only about half as large as the largest or even much smaller. The genus was first instituted by F. E. Schulze ('oo) to receive the species septenlricmale from the Arctic Ocean. Two forms occurrino; in the Saffiimi Sea and described bv me before- in brief under the names of RosseUa longispina ('96) and Vitrollula namiyei ('98) I now consider to be generically uuitable with Scyphidiuni septentrionale F. E. Sch. Further, the speci- men from Messier Channel (Southern Chile), which had been described by F. E. Schulze ('99, p. 43) as Bossella sp. without receiving a specific name, likewise seems to be referable to the same genus and to rejDresent a species especially closely related to my ticyphidiuDL namiyei. Under Rossella I put together the species (Nos. 1—1 of the SCYPHIDIUM. 21 list) in which the hexaster consists of oxyhexasters and of disc- ohexasters in three varieties differing in size and appearance (macrodiscohexaster, mesodiscohexaster and microdiscohexaster). Now, Scyphidium in the scope given it by me differs from that genus chiefly in the fact that the hexaster occurs in only two varieties, the macrodiscohexaster of Rossella being wanting here. However there can be no doubt of the especially close relationship existing between the two genera. As another very close ally of Scyphidium is certainly to be considered Vitrollula ferlilis Ij., which however lacks not only the macro- discohexaster but also the mesodiscohexaster seen in Eossella, the only discohexaster-form present being apparently the microdisco- hexaster. The chief differential characters of the species of Scyphidium may be seen from the following key. «. — Dermalia, stauiactins. a*. — Pentactinic hypodennalia with smooth, regularly cruciate paratangential s. I^arger discohexaster 80 \j-, and microdiscohexaster 40 ;a in dia. Oxyhexaster, about 100 \>. in dia. Sponge with stalks; without prostalia 'S'. septentrionale Y. K. ScH. (N. of Spitzbergen). 6'. — Pentactinic hypodermalia with either paratropal or regulary cruciate paratangen- tials, which are either smooth or shagreen-like. Larger discohexaster 90-1Ö0 |J., and microdiscohexaster 24 [i, in dia. Oxyhexaster, 100-122 \i. dia. Sponge witli contracted base ; with long and conspicuous diaetinic ijrostalia, in addition to which pentactinic prostalia may occasionally occur S^ longispina (Ij.). (Sagami Sea). 6. — Dermalia, pentactins and stauractins. Pentactinic hypodermalia with smooth regularly cruciate paratangentials. Sponge with thick stalk-like base. c'.- — Larger discohexaster of 100 ij. dia. leading down to microdiscohexaster of .'!5 {)■ dia. Oxyhexaster, 5o-7G \i. \\\ dia. Without prostalia >S'. namiyei (IJ-). (SHgami Sea)- t/i. — Larger discohexaster of 100 a dia. leading down to microdiscohexaster of 20 i'- dia. Oxyhexaster in two varieties; tlie smaller and thinner-rayed, 80-100 [J. ; and the larger, 120-150 u. in dia. Synapticular fusion among parenchymalia. Prostal- ia? S. sp. (Messier Canal in Chile. Vide F. E. ScJi. '99). 22 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. SCYPHIDIUM LONGISPINA (Ij.) Plate II. Rossella longispina, Ijima, '96, p. 253. Two specimens are now before me. The type-specimen, which is the larger and on which my original description was based, comes from Yodomi in the Sagami Sea (depth, abont 429 m.). The second specimen, obtained last year, is from Inside Okinose (about 572 m.). In general appearance l)0tli are not unlike certain Slaurocalyptus or Rhahdocalyptus with spiny prostal needles. The type-specimen (PI. II., fig. 1) represents a pear-shaped, laterally compressed, thick-walled sac narrowed below into a stalk-like base, where it is torn off. Length, 51 mm.; greatest breadth, 37 mm. Thickness of wall, in ^^l^ces G mm. The osculum at the top is oval, 14 mm. by 7.5 mm.; its edge, thin and simple. The external surface is uneven on account of low conical elevations, from the aj^ex of which strong diactinic pros- talia project in an obliquely upward direction, some to a length of 30 mm. or more. These prostal needles stand out sometimes singly, but more commonly two or more (up to about half a dozen) t02;ether in tufts. Thev mav moreover l)e associated with a few small and inconspicuous pentactinic prostalia, wdiich are far two sparse to form a veil. Between the apices of the eleva- tions the dermal surface is quite smooth. (PI. IL, fig. 10). Attached to the diactinic prostalia and apparently pierced right through by these, there are two small young individuals of the same species. Possibly they arose as buds, which, after separa- tion from the mother-body, were shifted outwards along the prostalia, similarly as in Lophoccdyx fhili'piymensis. One of the SCYPHIDIUM LONGISPINA. ZO young individuals is 15 mm. long ; it already shows a small oscular opening in its outer end. The other is much smaller, measuring only 0 mm. in length ; the osculum is still unopen. Both possess thin and short prostal needles of their own. — Besides the young, a number of Foraminifera, some of which are about as large as a pin-head, are borne on the prostal needles of the old specimen. The dermal skeleton is in close contact with the choanosome. The extremely delicate lacework formed by the dermalia exhibits meshes, more or less regularly quadrate, which are so small as to be scarcely discernible with the naked eye. The hypodermal latticework is formed of thin strands which intersect one another at various angles and thus produce irregularly shaped and rather small meshes. The laro-er incurrent canals visible throuo;h the dermal layer may be 2 mm. wide. The simple gastral cavity is lined throughout w^ith a smooth and continuous gastral layer. Under the lens this appears meshed like the dermal layer. Some apertures of excurrent canals situat- ed right under it may be as wide as 3 mm. The second smaller specimen (Mus. No. 507) from Inside Okinose, is of an oblong ovoid shape. The lower end is likewise torn off. Length, .37 mm.; greatest breadth 18 mm.; osculum, 8 mm. by 9 mm. The lower half of the sponge presents a smooth undulating surface. The upper half bears a number of long and strong prostal needles ; in this part the surfoce is much lacerated, to which fact may be due the fact that conical elevations are not observable in this specimen. On the prostals again are attached some Foraminifera, but no buds or young. The spicula- tion essentially agrees with that of the type-specimen. 24 ART. 7. T. TJIMA : HEXACTINELLTDA, IV. Spiculation. Tho following description refers to the large type-specimen shown in PL I., unless the other is specified. The imrenchymalia are exclusively diactins, of which some may be distinguished by larger dimensions as the principalia. The ])rincvpalia are oxydiactins with smooth or but faintly roughened ends ; generally bow-like and sometimes boomerang- like in shape. The length may reach 12 mm. or more and the thickness at the middle, 350 p-. The oxydiactinic prostalia are evidently nothing else than principalia, which in a certain position have grown excessively in length. A large prostal oxydiactin measured was 50 mm, long and 275 n thick. Smaller prostalia gradually grade over into the intraparietal principalia. Similarly, the latter intergrade with the finest (7 /^ thick) parenchymalia occurring as comitalia. These and in fact all the more slender parenchymalia are of a nearly uniform thickness thoughout their length and terminate with rough- surfaced, conically or obtusely pointed ends. Externally the spicular center is indicated not even by an annular swelling. The thin strands constituting the irregularly meshed hyi^oder- jiml latliceu'orh are made W]) for the greater part of diactins, exactly comparable in characters to those of similar dimensions in the parenchyma. One of them selected for the purpose of measuring was about 2 mm. long and 23 /^ broad in the middle. Though occasionally found running singly, they are usually more or less combined into bundles of varying strength. SCYPHIDIUM LONGISPINA. 25 Penlactinic hypodermaUa seem to l)e confined in their distri- bution to the upper part of the l)ody. Here they are in places not uucommon, especially on the conical hillocks. I have been unable to detect any regularity in the mode of their relative arrangement. They are never very large ; the largest I have picked out measured : length of paratangential ray, 1.5 mm.; that of the unpaired proximal ray, 5 mm ; thickness of rnys near the center, oo !>. But the majority are considerably smaller. The paratangential rays, which taper outwards to a sharp point, are either regularly cruciately or more or less paratropally dis- posed. The latter form arises as the result of some strong diac- tinic prostalia in immediate proximity excercising a pressure in a lateral direction upon the paratangentials, pushing these away from them and thus widely opening one of the four angles which otherwise would all be right-angled. In some of the spicules, the rays are all perfectly smooth ; in others, the paratangentials present a finely shagreen-like surface (PI. II., fig. 3), caused by the same minute and thickly set processes which I have met with on certain hypodermalia of Lanuginella pupa (p. 9). The proximal ray, as also that part of the diactinic prostalia which dips into the sponge-wall, is accomjjanied with comitalia of the usnal appearance. As before mentioned, isolated pentactinic hypodermalia may project outwards, generally in association w^ith the diactinic prostalia on or near the apex of the external hillocks (PI. IL, fig. 10). The pentactinic prostalia, unlike the other kind, are quite inconspicuous. As rare abnormalities of the liypodermalia under considera- tion, I may mention a case in which a distal sixth ray was present as a short rudiment, and another case in which two of 26 AET, 7. T. TJIMA : HEXACTINELLIDA, IV. the parataiigentials bore — one a single and the other three — prickles, suggestive of the same occuring in Rossella antarc- tiea. The dermalia (PI. II., fig. 2) are stauractins, nearly flat or perceptibly concave on the inner side. The microtubercles, which beset the rays all over, are distinctly developed. The rays slightly taper outwards, to end in a rounded manner. Axial length, 250- 360 // (300 /'• on an average) ; breadth of ray near the center, 7-12 // (10 !>■ on an average). In forming the quadrate-meshed latticework, two rays of the directly adjoining dermalia run, as usual, alongside of each other for the greater part of their length. The meshes generally measure 140-180 // in length of sides. — Exceptionally and quite rarely the dermalia are met with in the straight diactinic form. Of more frequent occurrence here and there are oxystauractinic forms with smooth and un- usually thin rays. They are evidently dermalia that have not yet attained full development. The dermalia of the second specimen from Inside Okinose require special mention in that pentactinic forms were not seldom found amongst them. The gastralia (PI. II., fig. 4) are hexactins much larger than the dermalia, but with similarly characterized rays. Axial length, generally 0.55-1.00 mm. (0.75 on an average) ; breadth of ray close to base, 15-22 n. The free proximal ray shows no point of special differentiation from the rest. Though the spicules are present in tolerable abundance, the paratangentials form but rarely meshes of approximately quadrate shape. The layer lies closely upon parenchymalia, of which none can however SCYPHIDIÜM LONGISPINA. 27 be distinguished as hypogastralia. There exist no pentactins under the layer. In tlie two large specimens in hand, both being torn oÔ' at base, the basidictyonalia can not be observed. I have therefore examined one of the two young individuals for those characteristically ankylosed sj)icules, and actually found them in the basal region of the little sponge growing on a prostal needle. This was at the place directly invested by a thin and small-meshed siliceous reticulum, the basal-plate of F. E. Schulze, over which was a layer of irregularly distributed hexactins, the basidictyonalia. These are small but thick-rayed, with uneven surface. They are in fusion not only with the basal-plate but also here and there with themselves, in a manner similar to the condition seen in fig. 12, PI. XXI. The mode of their occurrence on a spicule of another individual reminds me of the small basidictyonal masses I have found in Slauroealypius fjlaher {cfr. Contrib. I., p. 18G, foot-note). The hexasters are of the following three types : 1. Oxyhexaster. This occurs in great abundance in the choanosome as well as in the subdermal space. Especially numer- ous is it near the gastral surface and in the gastral layer itself. It is not uncommon to see several attached to the proximal ray of the gastralia, as if they had been shifted on from below along it. Diametei-, 100-122 !>■. The principals are always exceedingly short, so that the bases of terminals are situated close to the somewhat swollen central node. The terminals are obsoletely rough-surfoced. As to their number and the degree of their development, there exists a certain difierence between the oxy- 28 AllT. 7. — I. IJIMA : HEXCTINELLIDA, IV. hexasters in the periphery and those in the deep part of tlie wall. The former, as for instance those situated in the subdermal space, are as a rule normally developed, showing generally o, but sometimes 2 or 4, terminals to each principal. The terminals are thin, measuring only about 2 /^ in thickness at base. The latter, as represented by those in the gastral layer, have perceptibly stronger terminals, about 3 !>- thick at base. More- over they are but seldom normally developed, exhibiting 2 ter- minals to each principal. The majority are hemihexactinose, there being also found not uncommonly quite hexactinose forms (PI. II., figs. Ö, G). Of the hemihexactinose forms, the total luimber of terminals to the entire rosette may vary between 7 and 11 (most frequently 1) or 10) indicating in each case the number of the principals wliicli remain l^iterminal and of those which had become unitermiual. In some cases one of the two terminals borne on a principal was found to be very much shorter than the other, these shorter terminals representing without doubt intermediate stages in the transition from a biter- minal principal into a uniterminal. In both the hemihexactinose and hexactinose forms, the uniterminal principal may show the well-known bending at the point of its junction with the single terminal remaining to it, the two parts being not straightened out as they quite frequently are. 2. JJiScohexaster. Much less numerous than the oxyhexasters are the discohexasters (PI. II., fig. 7), which occur subdermally and somewhat less sparingly in the choanosome. In a few cases they were found together with oxyhexasters on the free proximal ray of gastralia. Diameter, DO-loO /'. Each short principal, which is swollen knob-like at end, bears 4-6, long, slender, faintly rough-surfaced terminals diverging so as to give an a2)prox- SCYPHIDIUM LONGISPINA. 29 imately spherical shape to the entire rosette. The convex terminal disc consists of 6, small, recurved teeth (fig. 8). 3. Jlicrodiscohexasler. This (PL II., fig. 9) is clearly dis- tinguished from the above by its small size and by the more numerous and exceeding y fine terminals. It is found in sparse distribution both subdermalh^ and subgastrally. The shape is spherical with a diameter of only 23-25 /^. The terminal disc is so minute that it appears simply pinhead-like. Soft Parts. As the type-specimen was thrown into alcohol at the place of capture, I was able to make some observations on the soft parts, which may be worth recording in brief. All my sections are stained with borax-carmine. Figs. 10 and 11, PL IL, show the general appearance of the soft parts in section. T'he dermal membrane is represented for the most part by a sparse quantity of quite thin trabeculse running between the rays of derraalia. It is membranously developed only here and there in limited areas between closely situated parts of certain dermalia. The cup-like or thimble-like chambers measure 100 /-< in average diameter. Examined under very strong magnification (Zeiss' homog. immers. system), its wall or the membrana reticula- ris exhibits distinctly open meshes. The delicate beams consist of granular protoplasm which is scarcely stained by the coloring- reagent (PL IL, fig. 14). Each nodal point, as seen surface on, is occupied by a faintly colored and clear looking nucleus, con- taining a few granules in its interior. Its boundary can not be said to be well defined but is often sufficiently indicated by the cyto- oO ART. 7. — I. IJIMA : HEXACTIXELLIDA, IV. plasmic granules lying against it as well as by the slight but perceptible difference in the staining capacity of the plasma within and without it. Diameter of the nucleus, not more than 2h /i. Flagella, not observed. Archœocytes occur either isolatedly or in small flat groups on the outside of the chamber- wall. Size, 22-4 /^ ; exceptionally 5 or 6 //. Here as in Euplectella niarshaU'i, one is soon led to the conviction that he has before him small cells, — not nuclei, — unless an inadequate power of the microscope be used for the observation. At least the larger archseocytes are distinctly seen to be composed of deeply stained cytoplasm containing a still more deeply stained nucleus, which is indistinguishable from a trabecular nucleus both in size and aj^pearance. Sections of the large specimen are remarkably rich in pecu- liar bodies which I take to represent the thesocytes. We have here to do with clusters of fat-like globules found among the trabeculœ of both the subdermal and subgastral regions. They are shown, not quite successfully, in PL II., fig. 12. The globule, when of large size, may measure 20 !>■ across. Its substance is nearly homogeneous and weakly refractive, taking the stain toler- ably well. It may at times be of a conglomerate-like or of an irregularly granular appearance. Not infrequently a small nucleus is seen in direct contact with the surface ; the appearance then being that the bodies in (pestion are some unusually bulky cell contents or product which has pressed the nucleus of the turgid cell against the external limit. Together with the above there occasionally occur smaller and more weakly stained spheres with granular contents. They are shown in both figs. 12 antl 13, PL II., and will be easily recog- nized. In the case of these again I have frequently ascertained SCYPHIDIüM LONGISPINA. 31 that they contained a nucleus which was pressed against the wall. Cases also came under my observation, in which the spheres in question showed a close approach, in size as well as in the character of the contents, to the thesocytes which are filled u]) with the fat-like product. So that, I am inclined to think that they represent but an early stage in tlie develop- ment of thesocytes. In sections of one (the larger) of the two young individuals found attached to the prostalia of the specimen, I find the theso- cytes by no means so copiously present. On the other hand, T see in them an al)undance of peculiar bodies, the like of which I have not met with in the large specimen and the nature of which remains perfectly dark to me. The bodies consist of numerous, thin, deeply stained filaments, which reach up to 5Ô ,« in length and are arranged either radially so as to present an irregularly star-like appearance or in brush-like bunches diverg- ing from a point (PI. II., fig., 15). The filaments are sometimes nearly straight and needle-like ; sometimes gently lient or wavy. The general appearance reminds me of the groups of stearin and margarin cr3^stals, which A. Letellier* obtained from the alco- holic extract of the organ of Bojanus. But the stainability of the filaments at once excludes the idea of their beinof crvstals. Possibly we have here to do with something which is certainly not identical with, but is allied to, the groups of rod-like bodies I have met with in Acanthascus cactus and Euplectella marshalU (Contrilx I., p. 180) ; at any rate, they all seem to be bodies foreign to the sponges. * A. Letellier. Fonction urinaire chez les ]\IolliJSf|nes acéphale^;. — Arcli. Zool. Expér. T. V. 2me Serie, p. 48 ; PI. I., figs. 6, 7. 32 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. SCYPHIDIUM NAMIYEI (Ij.) Plate VI., %s. 9-17. Vitrollnla namiyei, Ijima, '98, p. 48. Two complete specimens and some fragments of this species were obtained by Kuma in one haul of the long-line at Outside Okinose by the Iwado-line, fi'om a depth of about 4o7 m. (250 faths.). Both complete specimens (PI. VI., %. 9) present a laterally compressed, irregularly shaped, moderately thick-walled, pouch- like body of a moderate size. Above, there exists in both a secondarily formed osculum l)esides the primary. Below, the somewhat contracted base had evidently been firmly fixed to the substi'atum by an irregular attachment surface or rather by a number of such. The larger specimen (PI. VI., fig. 9a) is Qiß mm. high. A cross-section of the laterally compressed body at its widest j)art would present an oblong outline, Ö4 mm. long and about half as broad. The wall is about 5 mm. thick in the middle of the body. The upper end is occupied by a large, irregularly oval, sharp-edged osculum. To one side of this and well apart from it on the sagittal edge of the l)ody, there is a papilla-like bud, the rounded apex of which shows a secondary osculum of 2 mm. diameter. This leads into a small gastral cavity about 5 mm. deep, which does not stand in open communication with the principal gastral cavity. The specimen also shows on one side an oval-shaped gap in the wall, leading into the main gastral cavity. The smaller sj)ecimen (fig. 9 b) is 5Q mm. high. It is widest at the upper end, measuring 41 mm. in the longest direction SCYPHIDIUM NAMIYEI. 33 whereas, from side to side in the middle of tlie compressed body, it measures not more than 20 mm. across. Two separate oscula of different sizes, lying side by side, occupy the upper edge of the body. There is a clear indication of the two having been formed from an originally single and narrow oscnlum by the coming together and fusion of the oscular lips at one part, in a way similar to the process which takes place at the bifurcat- ing point of a Farrea-tnhe. The smaller of the oscula thus formed occupies the end of a short tube which appears like a branch from the sponge-body. In both specimens, there are present along the sharp oscular edge some prostal needles, which are fine, short and quite incon- spicuous, besides being irregularly distributed and isolated. Simi- lar prostals also occur in quite small numbers in the upper part of the lateral wall. They are in all cases diactins projecting to the extent of 3 or 4 mm. at most. — In places there are to be seen some isolated hypodermal pentactins apparently more or less protruded beyond the dermal level. Whether tliis represents a normal state or is the result of rough handling, I am at loss to decide. On the whole the external surface may be said to be smooth. To the naked eye it appears as if loosely frosted. Not until ob- served under the hand-lens can the dermal latticework and the hypodermal supports be seen with distinctness. The dermal layer is in very close contact "svith the choanosome. — The subder- mally situated entrances into the incurrent canals are small, probably never more than 1^ mm. in width. The gastral surface is likewise smooth. It is continuously covered throughout l)y a most delicate gastral layer, supported on fine sinuous hypogastral strands. — The excurrent canalar apertures, 34 ART. 7. — T. TJIMA : HEXACTIXELLIDA, IV. indistinctly visible through the gastral layer, may be 2 mm. wide. In tlie wet state the sponge-wall is rather firm Imt can he easily torn off. When dried it is of a light, delicate and friable texture. Spiculation. The j)arencliym(dia consist of diactins only, — oxydiactins as a rule. Tliese comprise all sizes from the large, elongate-spindle shaped or bow-like principalia down to the shorter and very much thinner comitalia. They are somewhat closely felted to- gether in forming the parenchymal mass. (See PI. YL, fig. 17). The principalia, which occur rather numerously, may attain a leno-th of 10 mm. and a thickness of 120 n. in the middle. They are smooth all over, even near the sharply pointed ends. The smaller parenchyraalia, leading down to comitalia of only 10 // thickness, often show an indication of an annular swelling at the center and are sparingly beset with ol solete microtubercles at the ends, which taper to a point instead of being slightly swollen as usual. — Certain small diactins on the outer and inner surfaces of the choanosome are formed into thin strands and go into the support of the dermal and gastral layers, more especially of the latter. Hypodermal oxypentaetUis are abundantly present. The rays are all smooth throughout and always regularly cruciate in dis- position. They are never very large, the paratangential axial length not exceeding 3 mm. The unpaired proximal ray is usually more than three times as long as the paratangential in the same spicule. Breadth of rays at base, as much as 55 IK — No pentactins occur as hypogastralia. SCYPHIDIUM NAMIYEI. 35 Irregularly meshed basidictyonalia of the usual appearance are found in a thin connected layer over the places of basal at- tachment. The beams, whicli may be 20 u. thick, are uneven- surfaced with scattered spiny processes. The dennalia (PI. VI., %. 10) are spinose stauractins and pentactins of considerable strength. As to the relative quantitv of these, the former somcNvhat predominate over the latter in some places, and vice versa in certain other places. The rays are 90-165 n long (as measured from the center), all those in one and the same spicule being nearly equally long. Thev are thick (up to 20 ij. at base) and taper gradually outwards to the conically or bluntly pointed end. The microspines are well developed. The paratangentials are in a plane scarcely or Ijut slightly concave on the inner side. In the case of pentactinic dermalia, the unpaired ray is always directed proximad. The gastraHa (PI. VL, fig. 11) are hexactins and pentact- ins ; exceptionally stauractins. The hexactins seem to be numer- ically the predominant form, though in some places the pentact- ins (with the unpaired ray directed distad) are found in about as great abundance. In the dimensions of rays and in the nature of their microspines, the gastralia are quite like the dermalia. The proximal free ray presents no features of special development. The oxyhexasters (PI. YL, fig. 12) occur in very great abundance throughout the choanosome. There is no appreciable difference in appearance between those situated in the subdermal and in the subgastral region. They are of a rather small size, measuring 53-70 u. in diameter. Nearly all are normally devel- 36 AKT. T. 1. IJIMA : HEXACTINELLIDA, IV. opeclj each short principal bearing 2-4 terminals. Only occasion- ally do hemihexactinose forms occur, especially in the middle of the choanosome. The terminals are only moderately strong ; their surface is rough. The minute processes causing this rough- ness, when examined under a strong i)0\ver of the microscope, are seen to l)e distinctly inwardly directed (lig. lo). The discohexasiers (PI. VI., figs. 15 and IG), present like- wise everywhere in the choanosome, are less abundant than the above, though in several parts of the subdermal and subgastral regions they are found lying numerously together side by side. They are mostly of al)0ut the same size as the oxyhexasters or of a larger size, reaching up to 100 //. in diameter. Others are so small as to measure only 35 /y. in diameter ; discohexasters of such small size may be taken as representing the microdiscohex- aster of the species. However, it is important to mention that the larger discohexaster (fig. 16) and the microdiscohexaster (fig. 15) in the present species are much the same in general a2:>pear- ance and are besides gradationally linked together l)y intermedi- ate sizes. So tbat, it is also not improper to say that discohex- asters are present in only one form, which is quite variable as to size (35-100 //. dia.), the smallest being less than half the size of the largest as measured by the diameter. Much the same relation seems to ol)tain among the corresponding rosettes (25- 100 // dia.) of Scyphidiniib sp. { = Rossella sp. F. E. Sen. '99, \). 43) described by F. E. Schulze from the coast of Chile ; where- as, in S. septeiitrionale and S. longispina the microdiscohexaster is clearly distinguished from the larger discohexaster-form not only l)y its smaller size but also by having the terminals in far greater abundance. — In the present species it may be said in VIÏKOLLULA Ö7 general that the discohexasters, irrespective of size, have o-o (or occasionally more), slender and faintly rough-surfaced terminals to eacli very short principal. All the terminals so radiate from the ends of principals that a spherical shape is given to the entire spicule. The arched terminal disc is composed of o-S, distinctly developed, recurved prongs. VITROLLULA Ij. In '98 I included two species under this genus ; but since I now consider that one of them had better be referred to Scyphidium and have described it above as S. namiyei, there remains only V. fertilis to represent the genus. A generic diagnosis may therefore be dispensed with. The genus and species is a small-bodied sponge, which would be difficult to distinguish from a Lanuginella or a Leucopsacus without a microscopic examination of the spiculation. This close- ly resembles that of Scyphidium, of Crateromorpha and of Hyalas- cus. It differs from that of the first mentioned genus mainly in that the parenchymalia iuclude hexactins and in that discohexasters occur in a single form which apparently corresponds to the microdiscohexaster. In these respects the agreement with certain Crateromorpha may be said to be almost complete, but the con- spicuous difference with respect to the size and shape of the body may be regarded as sufficient to Avarrant the generic distinction. The difference from Hyalascus consists again in the much small- er size and further in the presence of hexactinic parenchymalia, in the oxyhexasters being pertinently normally developed, &c. 38 AKT. 7. — I. IJIMA : HEXACÏINELLIDA, IV. VITROLLULA FERTILIS Ij. Plate III. VUrolhda ferlUis, Ijima, 'g8. Four spécimens in all of this genus and species have been at my disposal for study. I regnrd it worth while to mention the following particulars about each of them. Two specimens, forming Sc. Coll. Mus. Sp. No. 228, were found attached in the dried state to a LophoheliaAike coral, ob- tained by KuMA April 2nd, 1894, at Okinosé from a depth estimated at about 429 m. (235 fnis.). Both are of about the same shape, l)eing spindle-like, slightly bent and broader at the oscular than at the somewhat contracted basal end. One of them is shown in PL III, fig, 1. It is 15 mm. long and 6 mm. broad in the middle, where the body is approximately circular in cross-section and the wall measures li-l] nun. in thickness. The small, thin-edged osculum at the superior end has a diame- ter of 1Ï mm. The other individual of the lot is slightly larger, being 16 nnn. long and 7 nnn. broad in the broadest part. The third sjiecimen (Sei. Coll. Mus. Sp. No. 231), shown in PI. Ill, fig. 2, was obtained by myself July 23rd, 1894, at a s])Ot about 4 kilometers oft' the village of Inatori on the eastern coast of the Province of Izu. The dei)tli was somewhere between 380 m. and 414 m. (180-228 fms.) ; the bottom consisted of sand, pebbles and shells. There it occurred together with Farrea, Aphrocalltstes, Hyaloneiiia ajjiiie var. and Seinperella stomata, specimens of all of these having been secured at the same time. The body of the specimen in question is laterally compressed ; it is 12 nnn. high and 8ï mm. bv 5 nnn. broad. ^Vall, 2 mm. VITROLLULA FERTILIS. 39 thick in the middle of the body. The circular osculum at the upper end measures 3 mm. in diameter. The contracted and hiterally compressed base is attached to a horny worm-tube. The fourth specimen (Sei. Coll. Mus. Sp. No. 433) was ob- tained by KuMA in November, 1895, from an unknown depth at Inside Okinose. In shaj^e it is ovoid and slightly laterally com- pressed. It is torn oif at the narrower end. Length, 14 mm.; breadth, 7 h mm. Wall, about 2 mm. thick. The oval osculum at the broader end measures 2h mm. I)v le mm. in diameter. All the specimens agree in having a smooth external surface. Through the dermal layer, which is in close contact with the choanosome, are seen the variously sized, ])ut generally small, apertures to incurrent canals ; they rarely exceed f mm. in width. The gastral surface presents a somewhat honeycombed appearance owing to the fact that excurrent canals open freely into the gastral cavity, the apertures being not covered over by a continuous gastral layer (PI. III., figs. 7 and 8). Some of these apertures may be li mm. wide. The gastral cavity is deep. The body-wall gradually thins out towards the thin sim^ile oscular edge. The texture of the sponge is delicate, soft and light. The basal end, for a greater or less extent, is firm, which is due to the basidictyonal mass being developed to a not inconsiderable thickness. Spiculation. The parenchymalia consist of slender diactins and more or less regular hexactins. The former are present in tolerable abundance, running 40 ART. 7. — I. IJIMA : HEXACTINELLTDA, IV. either isolatedly or in weak bundles. They are all tliin and filamentous, probably never attaining a thickness of more than 12 ij.. The majority, if not all, of them have the spicular center externally indicated by an annular s\yelling or by four cruciately disposed knobs. The rays are smooth except at the roughened ends, which sometimes terminate conically and at other times taper out to a point. Scarcely any of the diactins can l)e dis- tinguished as principalia. The latter are l)y no means uncommon, though they are more plentiful in some individuals than in others. They some- times appear to be comjmratively strong. Thus, a large parenchy- mal hexactin may measure 1.6 mm. in axial length, the rays being 30 // thick at base. However, the majority of the hexact- ins are considerably smaller and weaker. The rays gradually taper outwards and are smooth all over except near the conically pointed end. As a general rule the hexactins are so disposed that one of the axes is radially directed, without however show- ing much regularity in the arrangement. Sometimes their rays are seen to pursue a solitary course ; moi-e frequently are they joined with the diactins in small numbers to constitute the par- enchymal strands. As in so many other lyssacine Hexactinellids, the basal end of the sponge exhibits a typical basklictyonal mass (PI. III., fig. 22). This consists of a rigid, irregularly meshed reticulum of comparatively thick l)eamä, the surface of which is beset through- out with small tubercles. These beams, for the most part, may without difficulty l^e recognized as the rays of unusually stout hexactins, which are ankylosed either directly ray to ray or by means of synapticular formations. The synapticulse on the surface yiTROLLÜLA FERTILTS 41 of the mass in contact with the substratum form a tliin close- meshed limiting layer (see the lower part of fig. 22). The hypodenncdia (PI. ITT., fig. o) are moderately large pentactins ; their rays closely resemhle in character those of parenchymal hexactins. The regularly cruciate paratangentials may measure 1 mm. in axial length and 27 // in breadth near the spicular center. The proximal fifth ray is always longer than the paratangential in the same spicule and at times is nearly twice as long. Tn the smaller hypodermalia the size of the paratangential cross approaches or may even nearly coin- cide with that of the larger stauractinic dermalia, so that in certain cases it is scarcely possible to decide whether a pentactin is to be considered as a dermalia or a hypodermalia. This may perhaps be regarded as an indication of the low degree of dif- ferentiation of the species from the Leucopsacidae. Seen on sur- face-view preparations, the paratangential crosses are situated for the most part without any regularity of mutual arrangement, though at places they may show an attempt, so to speak, at the formation of a quadrate-meshed latticework. The hypodermalia are never observed protruded as prostalia ; nor are they ever found with shagreen-like surface. The dennaHa are stauractius, the plane of which is usually slightly convex on the outer side. The rays are relatively long and slender and gradually taper towards the conically or obtusely pointed end ; the surface is roughened, generally all over, on account of quite obsolete and insignificant microtubercles which are scattered over it at rather wide intervals (PI. III., fig. 3). The axial length fl^uctuates generally between 360 !'■ and 680 /z ; the 42 ART. 7. T. IJIMA : HEXACTINELLIDA, lY. bread tli of rays at base between 7 // and 12 I'-. On the mem- branons oscular margin the size may decrease to 264 fj. axial length. Exceptionally slender-rayed dermalia, found occasionally by the side of stouter ones, represent without doubt a develop- mental stage preceding the attainment of definitive dimensions. ' The dermalia are found irregularly scattered in the dermal membrane. They can not be said to l)e numerous; in many places they occur in no greater, if not in somewhat smaller, num- bers than the hypodermalia. Altogether, it may be said that the latter with their paratangentials are about as much concerned in the support of the dermal membrane as are the dermalia them- selves. (See PI. III., fig. 10. In this figure, the cruciate spicules di^awn in blue represent partly the dermalia and partly the paratangential crosses of the hypodermalia. Through the deeply stained dermal membrane, perforated by roundish gaps or pores, is seen the most peripheral part of the choanosome). The gastralia are represented l\v both hexactins and pentact- ins, the latter having the unpaired ray directed distad. I can not definitely state which of the two forms j^i'edominates, though in some places I have found several of the former form placed together side by side. In any case the gastralia are on the whole rather sparsely present, l)eing situated in isolated positions. The rays are similar in appearance to those of the dermalia ; their length as msasured from the spiciilar center is 165-176 ft. The hexasters are of the following two kinds : Common but not alnnidant are tlie oxyhe:(-aslers (PI. III., fig. o) in the choanosome as well as in the subdermal space. They are characterized by the possession of rather numerous VITKOLLULA FERTILIS. 43 terminals, of which generally 4-7 are borne on each short princi- pal. They are slender — at any rate not strong — and are distinct- ly rough-surfaced. Though bent at the base, they are straight for the rest of their length and so diverge from one another as to give a spherical shape to the entire spicule. This measures 114-140 !>■ ; on an average 120 /v. in diameter. Cases of a principal Ijearing less than three terminals probably never occur. Certain it is that hemihexactinose and hexactinose oxyliexasters a]"e both entirely foreign to the species. The only kind of discohexasters present is, as before indi- cated, comparable to the microdiscohexaster of certain other Rossellids. (PI. III., fig. 6). They measure only 26-30 r- in diameter. The convex disc at the outer end of each tolerably strong principal bears a bunch of numerous and exceedingly fine divergent terminals, which end each in a minute terminal knol). The shape of the entire rosette is spherical. In a certain speci- men the discohexasters in question were met with only occasion- ally ; in another they were quite common, especially near the dermal and gastral surfaces, where they seemed to be somewhat more numerous than the oxyhexasters. Soft Parts and Larvae. A glance at PI. III., figs. 8-11, will show at once that the general arrangement of the soft parts is in essential agreement with what we know of other Hexactinellids. The dermal mendjrane (fig. 10) is perforated l)y numerous pores of various sizes ; its tissue separating these from one an- other is at times quite thin and lilamentous, wdiile at other times it is flat and film-like. 44 AEÏ. 7. 1. IJIMA : HEXACTINELLIDA, IV. The trabecule are sparsely developed, without doid^t on account of the narrowness of the space that they occupy. Chambers of the usual cup-like or thiml)le-like shape are 80-150 jjL wide. In a few instances it seemed to me that a chamber freely communicated with a neighboring one through the end which should normally l)e closed and rounded. Their wall or the reticular membrane exhi]:)its minute open meshes not more than 11 n. wide (PI. III., tig. 11). Under a moderately strong power of the microscope, the nodes of the reticulum appear as swollen points somewhat more deeply coloretl than the delicate Ijeams. Seen under the immersion-system the choanocyte nucleus (about 2 //. dia.) is discernible not so much by itself as by the fact that the spot is relatively clear of the surrounding protoj^lasmic granules. It is scarcely stained by the borax-carmine or the haematoxylin. In some of my preparations the flagella is occasionally observable, though by no means in a complete state. The collar is unrecognizable. Close to the thin oscular margin the ehaniber-laycr is re- presented simply by the reticular membrane disposed in a conti- nuous undulating manner instead of being formed into distinct chambers (upper part of fig. 8). Superiorly, it finally ceases to exist, its disappearance taking place insensibly in that its reticu- lum gradually passes over into the wider-meshed cobwel) of the ordinary trabecuUe (fig. 11). Well-stained arclueocytes, either isolated or grouped together in varving numl)ei's, occur in abundance on the outer side of the wall of the chambers, exactly as I have described in Eupledella iiiarshaW (Contril). I., p. 1G5) and in Leucopmcus orlliodocus (Contrib. III., p. 41). PI. III., fig. 11, shows two such groups of quite insignificant size. The larger of the archœocyte-congeries VITKOLLUIA FEßTILIS. 45 are very conspicuous on stained preparations on account of their compactly packed elements being very deeply colored (seen in PL III., figs. 8-10 as deep blackisli spots). Though they may take a somewhat irregular shape in accordance with the circum- stances of the space occupied by them, the normal shape of the congeries after attaining a certain size seems to be spherical (figs. 13-15). A rather small congeries of 40 /^ diameter (flg. 13) is already evenly delimited on the external surface, though evident- ly it is still without a special enveloping membrane or epithel- ium. It may grow to doul)le or more than double that diame- ter without showing a morphological change, except of course in the numerical increase of the closely crowded cells. The sharj)- ly defined surface pregses against the incurrent side of the walls of the chambers, right in the midst of which the body is situat- ed. The extent of the chamber-wall surface in contact with this is such that inceptively several little groups of archseocy tes might have taken origin on it ; hence it is exceedingly probable that the growth of an arclueocyte-congeries takes place not only by multiplication of its elements but also by fusion of originally separate groups. The archgeocytes, taken singly, are only 2-4 l^ large. PL III., Fig. 12, which shows a small group of them as seen in a borax-carmine pi'eparation, is not a good representation in that it fails to indicate the nuclear outline in each cell-l)ody. A reviewed examination of the ^^réparations, long after the plate had been printed, Ijrought me to the conviction that here, as I be- lieve in Hexactinellidan archœocytes generally, there exists a greater or less quantity of cytoplasm around the nucleus, — in other words, that we have here to do with small entire cells and not with free nuclei {cfr. Contrib. L, pp. 158, 171). The 46 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. cytoplasm always takes up stains well, a fact which renders the limit of the inclosed nucleus indistinct. The use of the immer- sion-system for the examination will make the matter clear, un- less the cells are overstained, as is generally the case when hsematoxylin is used. The nucleus contains from one to several dark granules. It measures approximately I2 /', which is also al)Out the size of the trabecular nuclei. TJie Larva. — In the study of the present species I have been al)le to obtain a somewhat more definite knowled'=' fact it seems assumable that the single bodies here taken for the cells are not merely nnclei by themselves. The cells, as seen in both cross (fig. 21) and tangential (fig. 19) sections, are separat- ed from one another by narrow clear spaces. Internally against the inner mass the covering layer is sharply delimited. On the external snrface, so far as the pre- sence of the layer can l)e demonstrated, there is a coating of qnite clear appearance, showing a sparse granulation which here and there assmnes the form of a vertical striation. I think there is no doubt whatever that this coating represents the fla- gella, which have deteriorated as the result of the application of reagents. The same was observed likewise in the larva of Len- copsacus orthodocus (Contrib. III., p. 43) ; therefore the Hexacti- nellida seem to offer no exception to the rule that the sponge- larva is provided with an external layer of flagellated cells. The inner mass, as seen in toto preparations under a mode- rately high power of the microscope (PI. III., fig. 20), ajipears, leaving aside the spicules, simply as a dense assemblage of well- stained corpuscles, much like a large congeries of the archseo- cytes before spoken of. Closer observations on serial sections (subjected to after-staining with hematoxylin or hsematein-alum or with either of these in combination with eosin, &c.) reveal that to a certain degree histological differentiation already exists among the elements of the mass (PI. Ill,, fig. 21). Peripherally and right under the external epithelium, the corpuscular elements, in respect to which it is difficult to deter- VIÏROLLULA FEr.TILTS. 49 mine whether wo have to do simply witli iiucloi or with cells in- closing each a nucleus, are very small and comjiactly j^acked together. I am ratliei- inclined to view them, as in the case of certain more internally situated elements (arch^eocytes) of similar appearance l)ut of a somewhat larger size, in the light of com- plete cells. The spicules, to which I shall soon return, run for the most part in the peripheral layer just referred to. .Vt the two poles of the larval l)ody, which are apparently destitute of the external epithelium, cellular elements are not recognizalde ; whatever sparse quantity of the ]ii'otO]ilasm exists over and l)e- tween the bundled spicular rays simply appears finely granular. The more internal and l^y far the principal part of the inner mass is composed of at least two kinds of cells, viz., those which go to form a reticular kind of tissue and those which retain a more or less spherical shape (fig. 21). The former most likely correspond to the so-called "dermal cells", and the latter to the arclneocytes, known in the inner cellular mass of non- hexactinellidan sponge-larvcT. The reticular tissue is most plainly visible in the anterior part of the body, in front of the region where this is broadest (see fig. 21). There it presents a small and open meshed, col)- web-like appearance, consisting of irregularly branching and anastomosing filaments, which are well stained and tolerably sharply defined in contour. The small corpuscles occasionally contained therein, I take for the nuclei. What nature to ascribe to the fluid, which, though imperceptible, undoubtedly fills up the spaces of the meshes, is difficult to directly determine. On the otber hand, I am stroughy inclined to assume that we have in the reticulum an inceptional trabecular system, which, in my opinion (Contrib. I., p. 104), represents at once both the 50 ART. 7. — T. TJIMA : HEXAC'TINELLIDA, IV. connective-tissue cells and the pinacocytes of the ÙMonaxonia and the Triaxonia. The two kinds of cells just mentioned are l)oth the outcome of the larval " dermal cells ", which in the Hexacti- nellida seem very early to take the form of trabeculee. If I am right in the above assumption concerniuü; Ihc morphological nature of the larval reticulum, the spaces in it or the meshes are sim2:>le interstitial lacunœ, which later (after the immigration of the external flagellated cells inwards to forui tlie c]iam])ers) should come into free comnmnication with the exter- nal world. From tliis standpoint it is exceedingly questional)le if the fluid contained in them is to he regarded in the light of a connective-tissue mesoglœa, Avhicli, moreover, is something apparently totally nndeveloped in the adults (Contrih. I., p. 101). It seems more likely that the fluid is simply imbil)ed water, — an assumption which suggests itself as being by no means improb- able. The reticulum can be traced, from the anterior region before ]-eferred to, backwards into that lying posterior to the broadest ])nrt of the larval body. In fact it may be said that the reticu- lum pervades almost the entire inner mass. Only, in the more extensive posterior region just indicated, forming about four-fifths of the entire mass, it is not quite plainly visible, this being probablv due to the crowded co-existence here of small and approximatelv spherical cells. INIoreover, the meshes here are on the whole consideraltly wider than in the small anterior region which lacks the said cells. It may not be improper to consider that iu a measure their greater width stands in relation to the presence therein of the cells in question. These cells, which pro1)ably deserve to be called sim])ly the archœocvtes, measure onlv 2-2 ^ ,"-. As in the case of the same VIÏEOLLULA FEllTILIS. 51 cells in tlie mother sponge, tlie cytoplusui is deeply staiucd and thus makes indistinct the outline of the nucleus, which, except- ing the few chromatic granules contained in it, does not surpass Init rather falls behind the cytoplasm in staining capacity. Fur- ther like them, the larval cells in question either adhere to the reticulum (the trabeculœ) or lie heaped together in the meshes (the trabecular spaces). They are by no means uniformly packed ill, but are for the most part irregularly and apparently rather loosely arranged, so that there exist between them vacant gaps? evidently parts of the lacunar system traversed Ijy the reticular filaments. The gaps may be of quite an insignificant extent, though at other times they may measure 14 // across. While many of them are irregular in outline, others present a more or less roundish section and then may be bordered, either partly or nearly all arouiul, l)y an epithelium-like row of the cells. Such an appearance gives one the impression that he has before him follicle-like structures ; and at first I even thought of the pos- sibility of their representing the Anlagen of the chambers. How- ever, after more concentrated observations, I have had to throw off this illusion, 1)ecause : firstly, several closely situated follicle- like spaces, though a})parently distinct at first sight, could often be demonstrated to be parts of a continuous lacunar space ; secondly, they are at places found to be partitioned from one another l)y a single row of the cells, which should not be the case if each were a follicle having a wall of its own ; and third- ly, it not infrequently happens that the inner space is distinctly traversed by the filaments of the reticulum. After all, I Ijelieve the cellular mass is simply honeycombed, as it were, by a system of interstitial lacunie, which are the same as the meshes of the reticu- lum. In other words it iiiav be said that the cells are, at least 52 ART. 7. 1. IJIMA : IIEXACTINELLIDA, IV. in ])art, arranged in reticular tracts in the same way as those in the oklest larva I have seen of Leucopsacus ovthodocus (Contrib. III., p. 46 ; PL III., fig. 25), the cellular tracts occupying the same space as the trabecular reticulum with wliicb they are joined together. As to the })robable origin of the cliaml)ers (hiring the meta- morphosis, I can here do no more tlian refer tlie reader to the considerations I have laid down on p. 1(>J of my Contribu- tion I. iSTow as to the s[)icnles of the larva. As before stated, they lie mainly in the pei'iphery of the inner mass, — close under, but ]iot in direct contact with, the external e])ithelium. As in the larva of Leucopsacus orlhodocus, they are all stauractins {cfi\ pp. 44-46, Contrib. III.). They are always so oriented in relation to the form of the larva that we may sj)eak of the transverse and the longitudinal axes (PI. III., fig. 20). The former is al- ways the shorter, and the two rays in it are usually of about e(pral length. Of those forming the longitudinal axis, one is as a rule much longer than the other. The four I'avs are in a ])lane more or less concave on the inner side, in conformity with the curvature of tbe Ijody-surfiice. In the manner of distribu- tion of the spicrdes, a strict regularity is Jiot observable beyond the fact that the centers lie well separate from one another, the result being that the rays run singly without coming together into bundles. They ai'c of such a length that intersection is of l're(|uent occuiTcnce ; there is thus brought about a latticework, the meshes of which may l)e said on the whole to be rectangular ))ut not regular in shape. Towards each pole of the larva, a munber of longitudinally running rays converge and there come N'lTKoLLrLA FEKTILItf. ÖO together into a 1>iuich, llieiice to project tlieir pointed emlri for a short distance out of the soft tissue. Finally, some remarks on the develo])menl. In all appearance the larva originates from the arclueocyte- congeries of the mother individual. In this respect T have nothing to add to what I have said in Contrib. I., ])p. 18 7-11 X), and in Contrib. ill., pp. 42, 43.=^= PI. III., figs. 13-15, will give a fairly good idea of how arclueocyte-congeries of various sizes appear in sections. In them the external epithelial layer of the larvffi is still undeveloped. This comes into formation when a congeries has attained a diameter of 90-100 //, by which time it is invariably spherical in form. I must say that there can not be discovered in my series any stages which show the exact manner of the formation of the external epithelium. Xeverthe- less, I think it will not fall wide of the mark to assume that the peripheralmost cells in a congeries, which has grown to the proper size, take the epithelial arrangement and thus difterentiate themselves as a layer from the inner cellular mass. Possiljly this takes place synchronously with the development of flagella by the said cells. At any rate, the remnants of flagella in the foi'm of a clear, granidai' or striated-like crust are observable as soon as the external e})itheliuni has established itself as such. Whether the epithelium is at first formed alike all over the spherical embryo, must be left undecided. The ap})earance of * 111 Sonic 111' the sectiuns of VilroUnhi ferllll" I liavf mot with a few cases of veritahle ej;ss whicli were undergoing llie cleavage process. They measured ahout TOO \i. and were of a dark appearance owing to llie abundant presence of deuto[)hismic granules wliicli com- pletelv hid the nuclei. However, I liave made myself sure of tlie fact that they did not lielong t(j the sponge, hut to a small Crustacea which lived in the sponge-wall, — a fact which could at once be foretold from the very appearance of the vitellus. 54 AKT. ~—I. IJI3IA : HEXCÏINELLIDA, IV. the inner cellular mass at tliis stage difter^ in no way from that of a simi^le archseocyte-congeries. The spicules appear shortly after, in embryos of 110-130/^ diameter, in the same shape and j^osition as I have described for the same embryonal stage oï Leucopsacus orthodocii^ (Contrib. III., p. 44). (Bee PI. III., figs. 10 & 17). They are all minute oxystauract- ins and are situated in the ])eriphery of the inner mass and at a short distance from its limit against the external epithelium. They are distributed, widely apart from one another, in a single Jayer running parallel to the external contour of the spherical I.Mxly. The plane of the foui- rays in each spicule coincides with that of the layer and is therefore slightly concave on the inner side. In the earliest developmental stage of the oxystauractius (fig. 17) that I have seen, the axial length mersured 1~) u.. The central node was flat and disc-like and was j'clatively large in comparisoii with the small spiiie-like rays. I did not succeed in bringing the axial iilaments into view, nor could any of the cells directly adjoining the spicules l)e distinguished from the rest as scleroblasts. It is most unsatisfactory that the spicules could not 1)0 observed at the very beginning of their develop- ment. For such minute observations the methods I have used seem to have been inadequate. »Some time after the appearance of the spicules, the eml)ryo- nal ])ody l)egins to elongate and assumes for a time an ovoid shape. 1*1. III., fig. 18, shows an endjryo in this stage of development. It is ajjparently in longitudinal section, but I am not in a position to slate exactly the direction in which it had been cut, since it was ibund in the wall of the mother sponge which had been sectioned without any knowledge of its presence. As it appears on the section in question, the flagellated layer invests the body on CRATEROMORPHA. fjh the sides as well as at both ends, — in fact, all aronnd. That this represents the trne condition can be asserted only nnder the assnmption that the section really passed throngh the Uxo poles, which is however not certain. I regret that the point could be determined neither on other sections of the same embryo nor on any other of the material in hand. — The inner mass now shows an advance in that the reticulum, before described from fully developed larvae, is distinctly observable in it. The open-meshed reticulum is for the most part situated at the periphery in one moiety of the body, which moiety is then probably to be regarded as the anterior. The cellular elements lie denselv crowded in the central as well as in the posterior parts of the mass. There exists a distinct indication that the reticular and the cellular tracts penetrate to a certain extent into each other. In the latter tract there occur a few irregular slit-like gaps. The oxystanractinic spicules are still quite small. Intermediate stages leading over the one just described to the fully developed larva were not discovered ; Init it will not be difficult to imagine the changes l>y which the form and organi- zation of the latter is reached. CRATEROMORPHA J. E. CIray. r^ip-like or bowl-like, firmly attached by distinct stalk; large or moderately large. Excurrcnt canalai' apertures on gastral surface probably always freely open. Stalk generally not tubular but traversed by a system of anastomosing (excurrent) canals. Parenchy- 5Q ART. T. T. TJIMA : IIEXACTINELLTDA, IV. malin diactins, in addition to which medium-sized hex- actins may sometimes occur. Pentactinic hypodermalia with regularly cruciate paratangentials, always present. Dermalia, rough pentactins or stauractins or both. Gastralia, generally similar pentactins and sometimes stauractins. Oxyhexaster as a rule normally devel- oped. ])iscohexaster in one form, which is usually small (microdiscohexaster with diameter under r)0 ,"), Init may be of a considerably larger size (80-120 /i dia.). Til the Challenger Report F. E. Schulze instituted a genus Avlochone as distinct from Crateromorpha. In '97 (]). '139) the same investigator included the former under the latter, thus joining the two genera into one, on account of the far reaching agreement in spiculation. However, I think the distinction be- tween the two genera mentioned may be kept u]i in view of the fact that, while ( -rater omorpha possesses pentactinic hypodermalia, Aulochone is altogether devoid of these, — a sort of difference analogous to that which separates Aecmthascus from either RhahdoGalyptiis or Stmirocalyptiis. Under Aulochone as a distinct genus may be placed not only F. F. Sohulze's original A. cylindrica (from the Kermadec Is.) and A. lilt 11)11 (from the IVIeangis Is.) but also the South Afi-i- can species recently described by R. Ivtekpatrick ('02) under the name of Grateromorpha lankesieri. To the genus Craterornorpha I refer the species and varieties eml)odied in the followinir CRATEROMORPHA MEYERI. 57 Differential Key to the Specie.^. "• — Dormnlin, cxclu.sively or predominantly pcntactins. Itiscohoxasfer, spherical, up to a')Oiit 50 a in diameter, a».— Dermalia intermixed witli some stanractinic forms. Hypodermal ])entactins not conspicuonsly thick-rayed (not over 100 a in hreadtli of rays at basel a-.— Sponge-body smo.ith n\\ the outside, tlie entire sponççe being exquisitely wine- glass-like C. meyrrl .J. PI Gbay. (Philippines; Saganii Sea). />".— Sponge-body with rounded tnberole-like prominences on the outside C meyeri inberosa Tj. (Sagami Sea, Snruga Bay\ '^''^•— ^^ouse-body with numerous wrinkle-like ridges and irregular prominences on the outside C. meyeri rugosa Ij. (Sagami Sea). i\— Dermalia exclusively pentactins. Hypodermal pentactins have rays of striking thickness (000 ;j. or more at base) C. pacliyaclina Ij. (OH' Shikokn, Japan). l> — Dermalia, exclusively or i)redominantly stanractins. '•'•—Discohexaster rather large (80-120 a dia.) and strong-rayed ; spherical. No hexactins among the parenchymalia C. tumida F. E. Sen. (Banda Is). f/^— Discohexaster small (up to -lO u. dia) and delicate. d^— Sponge-body witli tortuous exterior, the depressions leading into intercommuni- cating intercanals. Xo hexactins among the parenchymalia. Discohexasters all spherical C. cornujata Ij. (Sagami Sea). «■• — Sponge-body with smooth exterior; without intercanals. Medium-sized hexactins present among the parenchymalia. :Most discohexasters with the terminals formed into six separate bunches (not si)herieal) C. Ihierfdderi F. E. ,^'cu. (Link- Ki Is.). CRATEROMORPHA MEYERI J. E. Gray. Plate IV., figs. 1-8 and 12. CrateromorpJta meyeri. H. J. Carter, '72, ]). 112, — J. E. Gray, '72, p. i:^G.— H. J. Carter, jyi.—R. J. Carter, '73^, p. 861.— H. J. Carter, '75, p. 199.— W. Marshall, '76, p. 12'"). — F. E. Schulze, '86, p. 52 (reprint).— F. E. Schulze, '87, p. 161, pi. Lxi (!).— F. E. Schulze, '97, p. Ô40.— I. Ijima, '98, p. 48, 58 ART. 7. T. IJIMA : IIEXACTINELLIDA, lY. Hyalonema anoinahim. J. S. I^owerp.ank, '77, p. 401 {vule F. E. Schulze '87, p. 188). Besicles tlie typical Grateromorpha meyeri I recognize two varieties or subspecies of it, viz., C. nieyeri iuhevom and C\ u}p.y- eri rugo>^5 m. in the neighborhood of Okinose. Total height, 114 mm. Stalk, 30 mm. long. Greatest diameter of body, 67 mm. Osculum, 63 mm. in t'KATEROMOnrHA MEYERI. 59 diameter. Surface smooth all over. The one point specially worthy of note was the relative shortness of the stalk bearing the elongate sacciform body. The second specimen (Sei. Coll. Mus. No. 364) is the one shown in PI. IV., fig. 1. Locality, Outside Okinose by the Iwado- line ; 429 m. (235 fms.). Total height 115 mm., of which about 55 mm. belong to the stalk. This is of uneven contour, measur- ing 9-13 mm. across. In the lower two-thirds of its length it is quite hard and close-textured owing to spicular ankylosis, while the upper portion presents a longitudinally fibrous appear- ance. The lower end is thickened into an irregular basal en- largement, by means of which the sponge is fixed to the firm, finely grained, tufaceous substratum. The bulging, cup-like body somewhat closes aljove but soon Hares out at the simple-edged oscular rim. It is irregularly roundisli in cross-section, with a diameter of about 50 nun. at its middle. The osculum is 35-40 mm. in diameter. The sponge-wall is thin and membranous at the rim but thiekens below, attaining a thickness of 10-12 nnn. near the insertion of the stalk. The smooth external surface, when seen under the lens or even with the naked eye, shows the delicate and exceedingly fine-meshed dermal layer, which is supported by a much coarser hypodermal network composed of straight, but often interrupted, streaks of tolerably uniform fineness and showing small angular meshes, usually not exceeding half a millimeter in length of sides (PI. IV., fig. 6). The hypodermal beams are finer than those in either 61 ///. iuherosa or C. m. nigosa, and do not form continuous strands of such length, which fact has its ground in a certain difference in the spicular elements composing them. 00 AKT. 7. — I. IJIMA : JlEXACTIx\ELLIJ)A, lY. ])Ut of this later. — The apertures of iiicurrent caual»-!, visible lliroiigh the dermal layer, are small, not exceeding 2 mm. in diameter, even in the middle of the bodv where the lanrefet oecu r. On the gastral surface the apertures of excurrent canals open freely. Near the oscular margin they are all small; lower down, larger ones add themselves to the small, and in the lower half of the cavity the largest may measure 4-5 mm. across. Centrally at the very bottom there exists a space occupied by a few small apertures only ; peripherally it runs out into five or six, septa-like, radial ridges, the interspaces between which arc taken up by closely crowded excui'rent apertui'cs. The compact stalk is traversed by a system of anastomosing excurrent canals as in most Ci'afe}'oiiioi'i>ha (not simply tuljular as in ('. thiei'f elder i). Externally it lacks the dermal layer which must have fallen away. The sponge as preserved in alcohol is colorless. So also the formalin specimen belonging to Mr. Owstox. It is pure while after desiccation. Kuma states that the Sei. Coll. specimen, which was obtained l)y him, was in the fresh state " yellow like the yolk of a hen's egg." JJoth specimens examined by me contained a large colony or colonies of Si/llis ramosa MTntosh. This remarkalde Annelid seems to be seated mainly in the excurrent canal-system, stretch- ing out in part into the gastral cavity. When the sponge is dried it may still be recognized, as its colored body adheres to the white sponge-tissue. The S])ecimens I have taken from the C meyerl in possession of the Sei. Coll. were studied In' Prof. A. Oka (Ueber die Knospungsweise der SyllU ra/nosa. — Zool. Mag., Tokyo, Vol. Vll. [1895], p. 117). CKATEROMORrilA MEYElll. 61 Spiculation. The spiculation wa« studied i)riiK'i|)ally <>ii tlie 8ei. Cull. specimen (No. 364). I am also greatly indel)ted to Professor. ^^ E. Schulze for a gift of slide-preparations made from Celui specimens, wliicli have Ijeen invaluable for the purpose of com- parison. As in external form, so also in spiculation I «ibserve an essential and far-reaching agreement between the Japanese and the Pliili])pine specimens. The parencJiyMaJia are mainly diactins. among which iK'xacl- inic forms are occasionally intermixed. The diactinic })arenchymalia are as usual of varied dimen- sions, ranging from filamentous comitalia \\\) to princi})alia ut* ô mm. or more in length and 80 //. in thickness in the middle. The larger diactins are met with more especially in the deeper ])arts of the body, close to the canalar and the gastral surfaces. They are bow-like or elongate spindle-shaped ; smooth through- out (not roughened at ends), -without central swelling or knobs and gradually running out to the pointed ends. This refeis more especially to the larger diactins of the body proper ; those i.)ï the stalk are generally roughened at the ends, which are often rounded instead of being pointed. The thinner diactins, which occur either isola tetl or as comitals, are smooth at the center or show there at most a weak annular swelling ; their ends are always roughened and either swollen and rounded or attenuated to a point. The isolated oxyhexactinic parenchymalia (some shown in IM. lY., fig. 7) are of a medium or moderately large size. Tluy may approach the dimensions of a hypodermal })entactin but are generally considerably smaller. The rays are smooth and straight 62 AKT. 7. 1. IJIMA : IIEXACTINELLIDA, IV. or iieurly «traiglit. F. E. Schulze does not niciition these liexactins in liis descriptions but lias not omitted tliem in his figure (Chall. Ive})., PI. LXI., fig. o) ; indeed I observe their ])resence in the preparations from Celju specimens. Such par- enchymalia are known to exist not only in (_[ meyeri, l)ut also in the su1)species iuhcrom of the same as well as iu C. thierfel- (leri. In the remaining members of the genus they seem to have disap})eared altogether, leaving the 2)arenehymalia composed ex- clusively of diactins, so fir at least as those of the sponge-l)ody are concerned. In the upper one-third of the stalk the jnirenchymalia seem to consist only of longitudinally disposed diactins which are dense- ly grou2)ed together but free. Synapticular connections between them commence to occur at about the beginning of the lower two-thirds of the stalk. At the same time there begin to a|)pear, among the diactinic parenchymalia of the region, small hexacti- nic — and occasionally pentactinic — elements (PI. IV., fig. 8), the j'ays of which are comparatively short and thick, have rounck'd ends and show inconspicous microtubercles on the surface either all over or near the ends only. The spicules in question are at Jii'st free but soon l)ecome fused to one another as w^ell as to the diactinic parenchymalia of the region in irregular orientation. Consequently, in about the middle of the stalk the skeleton is already entirely represented by a dense and stony siliceous frame- work, except for isolated oxyhexasters lying loose in the meshes. The small hexactins just mentioned I regard as homologous with those which I have called the bdsiJiclijoJialiff in other lyssacine Hexactinellids. Tliey have l)een mentioned and well figured by Y. E. Schulze (Chall. Eep., PI. LXL, figs. Ö, 6 cV: 8). CRATEROMORPHA MEYERT. 63 Tho hypodermnlia are mainly oxypontactins of a comparative- ly large size and with latlier strong rays. The paratangentials may he 1.") mm. long and the unpaired proximal ray, 2.5 mm. The rays at l)ase may attain a thickness of 100 //. 'J he pointed ends of rays usually show no roughness of surface. There occa- sionally occur exceptionally small and thin-rayed oxypentactins, situated somewhat deeper than those of more normal size ; they ]irobal)ly represent early stages in the development of the hypo- dermalia. The meshwork formed l)y the paratangential crosses is irregular (PI. IV., fig. 6). In addition to the pentactinic hypodermalia there are ob- served at intervals slender diactins, which, running either solitar- ily or in small bundles, are in direct contact with the dermal layer and so serve as its support. They are thus to be regarded likewise as hypodermalia ; however, it must be said that as such they play only a subordinate rôle in comparison with the pent- actinic forms. In the Sei. Coll. specimen the diactinic hyj^oder- malia are quite few and far l)etween ; in the larger specimen Ijelonging to Mr. OwsTOX, they are somewhat more numerous. It may be that as the sponge increases in size, their number is more or less augmented by transference from the ranks of the parenchymalia. It will later lie seen that in large specimens of both the subsjiecies iuberosa and rf/gosa, the numerical ratio of pentactinic to diactinic hypodermalia is reversed, the latter greatW predominating over the forjner in numl)ers. J>at it is of course impossible to predict that the same will ultimately take place also in the typical species after a great advance in growth. There exist in this species no spicules, which can be speci- fied as the hypogastralia. G4 ART. 7. T. IJIMA ! HEXACTINELLTDA, IV. The (Jeriiialia nvo rongli pentaetins ; occasionally stanractiDs (TM. TV., figs, !2 & 3). The rays are on an average 130 ti long as measnred from the center and 7.6 //. thick. They scarcely taper outwards at all or do so in T)nt a slight degree. The ends are rounded. Not infrequently the pentactinic form, in which the un]mired ray is always directed proximad, shows an indica- tion of the sixth distal ray in the form of a knob. The para- tangential cross is usually — hut not always — more or less convex on the outside, which is due more to the rays concerned being not (juite at right angles to the radial axis, rathei- than to the cnivalure of the rays themselves. Seen surf ice on, the delicate (Iciinal latticework (PI. IV., fig. <>) presents irregular meshes, though in places these show a tendency to assume a regular quadrate arrangement. Here and there occur unusually small and slender-rayed dermalia — in all probability not fully developed — in which the rays are but obsoletely rough and run out to fine points. On the thin oscular margin the dermalia present are all stauractins. These and some thin diactinic ]iarenchymalia seem to be nearly all the spicules that enter into the support of that part. The gaslralia are pentactins, but sometimes stauractins and rarely even diactins. The rays are characterized similarly to those of the dermalia ; only they are frequently of a nuich great- ei- length, while the microtubercles of the surface occur somewhat more sj)ai'sely. Without forming a continuous layer the spicules in (piestion are irregularly distributed over the choanosonial sur- face facing the gastral cavity. CKATEROMOKPHA MEYERT. 66 The .«anie kind of spicules as the gastralia just mentioned, are present on the surface of the excurrent canals, Tliese may tlien 1)0 called the canalaria. They are most frequently met with in the widened proximal region of the canals, directly ad- joining the apertures into the gastral cavity. It may safely lie concluded that the same kind of lining spicules, whatever he their names, extends from the gastral cavity into the excurrent canals, Followins; the latter distad towards and into their nar- rower lu-anehes, the canalaria become more and more scarce until they cease altogether to exist. To be explicit, the canalaria are, mostly at any rate, rough pentactins with the unpaired ray directed distad. So far as those in the body proper are concerned, I have not seen hexactinic forms amongst them, luit I believe that some might possibly have been discovered, had a more extensive search been made. On the other hand, in preparations of tissues from the upper part of the stalk, I find a considerable quantity of regular hexactins occurring together with pentactins. All these spicules are quite like the gastralia in appearance and without doul)t represent the canalaria lining the excurrent passages in the region mentioned. For Philippine specimens F, E. Schulze ('87, p. 16') ; '97) has given rough oxyhexactins as the canalaria gene- rally. T find this fully borne out by the preparations of a Oebu specimen at my disposal. Nevertheless, the difference here indi- cated as regards canalaria I regard as probably inconstant and therefore as being not of systematic importance. Of the hexasters, tlie oxyhexmier (PL IV., fig. o) occurs abundantly in the clioanosome, ])iametei', 90-120 //. Each very sliort princi])al bears usually 2-3, sometimes 4 and even 0, (jG art. 7- — T. 1JT.>[A : IIEXACTINELLTDA, IV. straiglit or sliglitl\^ l)ent, obRoletely rongli or nearly smooth, divergent terminals. There are in some oxyhexasters perceptibly more slender than in others. The small (lisrohexaster (PI. VT., fig. 4) may be said to be spherical in shape ; the terminals emanating from each principal do not form a distinctly separate tnft, as seems to have been the case in the specimens stndied l)y F. E. Schulze, Diameter, 4'")-50 r,.. As shown in the figure referred to, the minute terminal discs ol'ten appear as if they were situated in the periphery at un- c(|ual distances from the central })oint. T tliiidv this is due, not so much to actual differences in the length of the terminal rays as to the various directions in which these are viewed. The said discohexasters are scattered in moderate abundance in the subgastral region as well as along the surface of the excurrent canals. They are somewhat more common in the latter region than in the former. F. F. Schulze ('97) found the rosette in (|uestion generally shifted out to the free I'ay-tij) of oxyhexacti- nie canalaria. Similar instances were observed also by me, in Avhicli a rosette hung on the free end of a stray parenehymalia- ray that projected into the canalar lumen. CRATEROMOEPHA MEYERI TUBEROSA Tj PL IV., Hg. 1>; PI. v., figs. 12 ct 13. CJ. meyeri vai'. luhero.^a. Ijima, '98, ]i. 49. Of the specimens wdiich I refer to a subspecies of (J. meyeri under the above trinomial designation, several (no less than CKATEEOMORrUA .MEYEItl ÏUBEROSA. 07 fourteen) liuvc come nuclei- my observation. The localities in the ^:>agami 8ea, so far as known to me, are : Outside Okinose (about 429 m.), whence came most of the specimens; Homl)a (about 572 m.) ; and a spot a few miles E. of Habu, Vries Islaml. Xurth of the latitude of Ok i nose in the sea mentioned, no specimen as yet seems to have been o])tained. AVhereas in the Suruga Bay, on a rich Jfelacrinus ground near Enoura and of only al)out 80 fathoms depth, the " Albatross " (Stat. 3719 ; May 11, 19()0) trawled u[) a badly macerated specimen which at the time seemed to me to ))elong to the present subspecies. A specimen from Honiba (PL V., fig. 12) was growing'- un a loose stone covered all over with remains of Brvozoa, worm- tubes, &c. Others from Outside Okinose are attached to Ijlack lava, to masses of volcanic de- tritus or to some shells (in one case to a Balanus and in an- other to a ]3rachiopod). Sylli.^ ramosa seems to be a verv fre- quent, if not a constant, com- panion of the subspecies ; at least I have been able to deter- mine the presence of that com- mensal Annelid in all speci- mens (seven in number) from Outside Okinose. This lot of Okinose speci- mens is of further interest in that it comprises a graduated series of ditterently sized in- dividuals of the subspecies, the 'IVxl-figure 1. CnUeromuipha meyeri lubcrosa Ij. 5 natural size. 68 ART. 7. — I. JJIMA : IlEXACïIXELLIDxV, JV. extremes representing the smallest and the largest I have us yet seen. To mention a few specimens in particular. The smallest just referred to is the one shown in the accompanying text figure 1, A (S. C. M. No. 478). It is only 42 mm. high, the stalk being nearly as long as the body 2^i"0i)er. The latter is 18 mm. Inroad. The presence of tubercular prominences on this small specimen as also on such gradationally larger ones as mea- sure 57 mm,, 73 mm., 8-") mm., etc. in height, shows that their their formation takes place very early in the life of the indivi- dual. Text-tig. 1, L\ represents a modei'ately huge specimen with the typically characteristic shape of the subspecies. (From Out- side Okinose, S. C. i\I. No. 482). Total height, 115 mm. Greatest breadth of body, Go nun. PI. v., fig. 12, shows in half natural size the single speci- men I have from Homba (S. C. M. No. 444). Total height, lo9 mm. Greatest breadth of body, 100 nun. The irregukir protul)erances of the Itody are lO-oO mm. or ir.ore in height. In places the surfaces of directly adjoining protuberances have come into contact and have fused together leaving an arch-like passage between them. The stalk is of about the thickness of one's thumb. Fig. lo of the same plate depicts the largest s])eciinen (S. CA M. No. 445, fi-om Outside Okiuose l)y the Iwado-line) tliat has come under my observation ; it is cut open lengthwise, so as to shovv' the gastral surf ice and the system of excuri'ent canals traversing the stalk. Total height, 210 mm. Thickness of wall, 7-10 nun. in the lower })art, without taking the outbulging into consideration. CEATEROMOEriTA MEYEllI TÜBEROSA. 69 Summarily speaking, the general ajipeavance of the sponge is essentially like that of the typical sj^ecies, except in the fact already indicated that the external surface of the body proper is irregularly and conspicuously uneven on accovuit of large and small rounded protuberances. These are usually only slightly (tr not at all developed along the rim e. In all the larger specimens, the hypodcj-mal strands as seen -with the naked eye or under the lens are on the whole some- what coarser (u[) to 90 //. in thickness) and therefore more dis- tinctly visible than in the individuals I have seen belougiug to the tyjùcal Sjiecies. Moreover, they extend continuously to considerable lengths without beconnng Ijroken at short intervals in course. J>y branching and by intersecting v,ith one another they form a line nieshwork with small angular meshes. Micros- copic examination shows that these hypodermal strands consist mainly of line diactins in fascicular arrangement. Pentactinic hvpodermalia in condjination with the strands are comparatively sparse (cfr. ]). 03), though they occur abundantly on the stalk. The wide difference in character Ijetween the hypodermal lattice- work as described aljove and that known to me from the typical species will be apparent by comparing figs. 11 and 6 in PI. TV. Howevei', the difference is a])parently one which becomes pro- nounced oidy afier the subspecies under consideration has grown to a certain large size. Thus, in the smaller specimens — say, in those not over 100 mm. or so in height — the diactins and the CRATEEOMORPHA MEYERI RUGOSA. 71 hexactins are oitlior nearly equally represented or the latter predominate over the former in numerical proportion, in which cases the hypotlermal structure as regards its composition is much the same as in the typical species. In the smallest specimen at my disposal (text-fig. 1, A), the hypodermal latticework is scarcely developed as such and can not he distinguished from the choanosomal feltwork. The canalaria, which I have seen in l)oth incurrent and ex- curront canals, arc rough pentactins with or without a rudiment of a sixth ray ; occasionally they are stauractins or hexactins, though in some individuals the latter form may be said to he even al)undant. Of the hexasters, the microdiscohexaster (PI. IV., fig. 9) is met with in moderate al)undance in 1)oth the subdermal and sub- gastral regions. The terminals are perceptibly finer than in the typical species as exemplified by specimens from both Celni and the Sagami Sea, but this may l)e a varialde character. The dia- meter ranimes from oS /^- to '>() /i. CRATEROMORPHA MEYERI RUGOSA Ij. PL lY., figs. 10, 11 ; PI. v., figs. 14, irK C vieyerl var, rugom. Iji.afa, 'g8, p. 40. This subs2)ecies is established on the strength of five speci- mens. One of these belonged to ^Ir. Alax Owston (O. C*. No. G':)77) ; the rest are all in the Sei. Coll. Museum. The known localities are Outside Okinose by the Iwado-line and Homba (al)out 572 m.), each of which localities has thus far yielded two 72 AE,T. 7. — T. TJIMA : HEXACTINELLIDA, IV. specimens. The grounds may be said to he the same as those tVom which came both C mei/eri and C. meyeri tuberom. As- snmably the three forms thrive under different physical condi- tions of the bottom, and the possibility can certainly not be excluded that they represent what Bidder has recently called " the metamps." Subspecies rvgom is shaped much like tvherosa but is char- acterized by the fact that the external surface of the body pro- per, except close to the thin oscular margin, is extremely uneven on account of numerous wrinkle-like ridges and other irregular ])rominences. The sponge may grow to a respectable size, measur- ing 320 mm. in height, as attested by one specimen in the Sei. Coll. (JNlus. No. Ö03, from Okinose). In that specimen the stalk length is about equal to only one-fourth of the total height ; it broadens above to an unusual extent, so as to assume an inverted cDnical sluipe. The body proper is unfortunately much shrunk and jiartly destroyed. Very well preserved are the two S2)ecimens shown in PI. V., figs. 14 and lo, and hence they may be taken as models for description. Though they are about the smallest I have had, the height measures nearly 240 mm. in l)oth. The stalk is nearly as long as the body, near the small and irregular attach- ment disc is about as thick as one's finger and gradually thick- ens above towards the junction with the body. It is throughout compact-looking, being partly covered by a dense coating of der- mal s])icules which easily f;ill off, and partly firmly felt-like on account of the exposed parenchymal fibers that run in the main longitudinally. Tengthwise it is more or less prominently ril)bed in the upper part, the ribs passing above into the superficial irregularities of the l)ody proper. CRATEROMORPHA MEYER! RUGOSA. 73 This stalk expands somewhat abruptly at its upper end ; it is more or less distinctly compressed laterally. In the specimen of fig. IT), the major transverse axis of the body measures 172 mm., the minor falling short of it by nearly 50 mm. The wall, which is thin along the oscular edge, shows a considerable thickness below. The osculum is large and wide, being of an oblong shape though quite irregular in outline. The irregulari- ties of the external surface, which form the most conspicuous feature of the subspecies, are apparently due : firstly, to the thicken- ing out of the wall into protuberances similar to those of tubero- sa, these being generally most prominent in the lower part of the l^ody ; and secondly, to the tact that the general surface is thrown into low and sharp-edged wrinkles, such as arise on cer- tain soft substances when they become parched. In the sponges before us there can be no doubt whatever that the rugosity is something natural to them and not a postmortem feature. Except in the above character, the texture and general ap- pearance of the sponge are in essential agreement with typi- cal C. meyeri but especially with the subs2)ecies ivherom. I may specially mention that as regards the appearance of hypodermal strands and of the gastral surface (see fig. lö), what I have said under tuberosa is equally applicable here. The agreement extends to the spiculation also. But there exists one, probably not unimportant point of difference in the fact that in no specimen of rugosa have I found hexactins among the parenchymalia, these consisting exclusively of diactins. This negative result was reached in spite of a special search made in a number of preparations. Exactly as in tyberosa, the hypodermalia in the body pro- per are mainly diactins ; pentactins occur only here and there 74 AET. 7- 1. IJIMA : HEXACTINELLTDA, lA . amongst them (PI. IV., fig. 11). Tlie}" form relatively long and continuous strands of varying strength. Some of the strands, especially those running along or forming the edge of the more prominent wrinkle-like ridges, m;iy be 300 // or more in breadth. On the stalk, pentactinic hypodermalia are present in abundance ; numbers of them adhere to the finger on being touched. The canalaria are rough pentactins with or without the knob-like rudiment of a sixth ray. Eegular hexactins as canal- aria have not been met with. On other points in the spiculation special remarks may be entirely dispensed with, as they would be but a repetition of what I have already said under typical C. meyeri. CRATEROMORPHA PACHYACTINA Is. PI. IV., fig. 13. Crater omorpha jxichyact'nia. Ijdia, '98, ]). 49. This species is based on a single and, unfortunately, much injured specimen (Sei. Coll. ]\lus. No. 395) which is stated to be from the Tosa Sea, off the island of Shikoku. The specimen was found included in the exhibits of the marine j^roducts of Küchi-Ken (Prov. Tosa) in the Fourth Industrial Exhibition held at Kyoto 189Ö ; it was purchased by the natural history dealer " Mimatsu " of Tokyo and subsequently acquired by the Science College. I at first referred it to GraleromorpJia meyeri, but a closer examination of the structure rev^ealed a number of points which seem to be sufficiently characteristic to found a dis- tinct species on. CRATEROMORPHA PACHYACTINA. 75 The specimen consists of a stalk and a large fragment of the body projier. The former is about 100 mm. long and 18 mm. by 26 mm. thick in the upper portion, but narrower below and just above the swelling at the extreme base. In its general ap- pearance, in the mode of transition into the body proper, in the canal-system traversing the interior and in the ankylosis of principal spicules in the lower portion, the stalk corresponds ex- actly to that of C. meyerl. One thing which attracted my atten- tion from the outset was the fact that on touching it with the fingers it readily gave off' sharply pointed and disproportionately strong-rayed spicules (hypodermnl pentactins), on which account it was necessary to use extreme caution in handling it. The fragment left of the body-wall is in a mutilated condi- tion but still s'lfiiciently w^ell preserved for determining the more important features of the sponge. It is easy to conceive that the specimen, when entire, had approximately the size and general appearance of the two specimens of C. meijeri rugosa figured in PI. V., figs. 14 and 1.3. Thé external surface is ex- tremelv uneven on account of irregular elevations of varying height. The wall is thick, measuring not less than 15 mm. in thickness near its junction with the stalk. While the gastral surface is perforated with numerous large apertures of excurrent canals and looks much like that of G. nieyeri, the outer side of the wall presents a remarkably com- pact and densely felted appearance, apparently due to an ex- cessive development of the parenchymalia as well as to the fact that the dermal layer is closely adherent to the parenchymal mass. The subdermal space is scarcely perceptible, while incur- rent apertures and canals, so far as can be recognized with the naked e3^e, are narrow and widely separated from one another. 76 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. The largest incurreut apertures, seen here and there in scattered distribution, do not exceed 2 mm. in diameter. Spiculation. The pai^enchymalia seem to consist exclusively of diactins. Not a single parenchymal hexactin could be discovered although a special search was made for them. A number of the diactins may be called princijmlia. These are straight or bow-shaped spicules of varying strength ; smooth all over and sharply pointed at ends. In the body proper they may measure ü mm. in length and 275 /t in thickness in the middle ; in the stalk they are generally longer but more slender, reaching up to 15 mm. in length and 100 /j. in thickness. The larger diactins are found in especial abundance near the external surface in both the body and the stalk. In the latter they mostly run in longitudinal directions. But by fc\r the greater part of the parenchymalia is made up of very much finer diactins (accessoria) generally not over o mm. in length with a breadth of only 4-15 //. The ends of these are somewliat swollen, rounded and rough-surfaced. These filamentous diiictins occur in part as comitalia to the stronger spicules ; for the rest they stand alone by themselves and may be developed in such exceedingly great numbers as to form a tissue of very fine soft texture. Such a tissue exists even in the stalk but is confined to the inner portion of its upper part. When freed of any such coarser spicules as may be contained in it, which can be done without much difficulty l)y feeling for them, the tissue can be l)alled like wool or cotton by rolling it between the fingers. CRATEROMORPHA PACHYACTIXA. 77 The ankylosis of certain spicules in the lower ])art of the stalk occurs in much the same manner as in (J. meyeri. Per- haps it may Ije regarded as a point of slight difference that the obsolete microtubercles on the beams of the basal framework are comparatively sparsely |»resent in an irregular distribution. Btrongly developed as are the j^arencliymal piincipalia, a far more striking feature is olfered by the unusually thick-rayed hypodermal o.vypentaotins (PL IV., fig. lo). These occur in abundance on both the body and the stalk. Handling the sponge without due care is liable to lead to the irritating result of find- ing them impertinently sticking to the skin by their sharp points. While some of the pentactins — evidently those not yet fully developed — have indeed comparatively slender rays, most of them have rays so thick that they may be said to l)e nearly of an elongate conical shape. Wath a length of 24 mm. (as measured from the spicular center), the rays may be 330 // thick close to their base. They taper gradually towards the sharply pointed ends and are smooth throughout. All the rays in one and the same pentactin are of nearly equal lengtli. The plane of the paratangentials is usually convex on the outside. The pentactins in situ can be discerned witli the naked eye and picked up one by one by means of a pincette. In PI. IV., fig. 13, a few der- malia and some fine parenchymalia (comitalia) are drawn by the side of a hypodermal pentactin in order to show at once the striking difference in bulk. No other spicular forms than the above pentactins can be distinguished as hypodermalia. The dermal layer lies in most places in direct apposition to the parenchymal mass, and there- /8 ART. 7. 1. TJIMA : HEXACTINELLIDA, IV. fore it is scarcely possible to discriminate any one of the under- lying diactins as being hypoderinal and not parenchymal. The dermalia, so far as I have seen, are all small rough pentactins, exactly comparable to the same of G. mej/eri. They do not form a distinct dermal la ce work, being closely adherent to the tissues below. The gastraUa, which are likewise lOugh pentactins, also show no noteworthy point of diaerence from the same ot C. mei/eri. The same may further be said of the hexasters of this species, so that I need not enter into a description of them be- yond making the following cursory remarks. The oxyhexaster is abundantly present in all parts of the parenchyma. Diameter, 80-100 /a Number of terminals to each principal, 2-4. The microdlscoliexüstcr is common in the parenchyma gene- rally. Diameter, oS-50 //.. The figure of this rosette given in PI. IV., fig. 9, from (J. itieiferi tuberosa, may pass equally well as that of one from the present species. CRATEROMORPHA. CORRUGATA Ij. PI. VI., figs. 1-8. Craleromorplia corrugata. I.iima, '98, p. 49. I consider thi& as a very well characterized species, which, unless the specimen to be identified is too imperfectly preserved, can be easilv recognized. CRATEROMOEPHA CORRUGATA. 79 About fifteen specimens in all have passed tlirough my hands. ISTo doubt they all came from the Sagami Sea ; a more exact statement of locality can l)e made only in the cases of five specimens from Outside Okinosé by the Iwado-line (200-oOü fms.) and of one from a spot in Döketsba (100 fms.). The latter speci- men, together with Euplectella m'irs//al/i, Mttacrinus rotundus, etc., was obtained by Professor Mitsukuri in one of his excursions on the " Gülden Hind." At first sight the species may appear not unlike C. meyeri tuherosa or rugosa (PI. VI., fig. 6). The general shape of the body is that of a bowl of irregular contour, broadest at the base, the central portion of which passes below into the stalk. The periphery of the body-base may somewhat overhang the insertion of the stalk. The external surface of the body proper is very peculiarly characterized. Though it looks much the same as in C. meyeri along the thin smooth oscular border, the greater part of it pre- sents a nuich folded or corrugated appearance. The rounded and quite irregular folds causing this appearance may at once be dis- tinguished frDUi tlic simple j'rotuberances of G. meyeri tuberosa or from the wrinkled irreguhirities of C. meyeri rugosa. Between the folds are furrow like or 2:)it-like depressions; many of tiiese are shallow and plainly show the cul-dc-sac bottom, while others, esj)ecially the pit- like ones, are frequently deep and may even be so deep and canal-like that their course can be followed only by introducing probes or by cutting open the wall. And among such deeply penetrating pits or canals there invariably exist some that internally freely intercommunicate with others of the kind. The canals may divide in their course and the branches may by anastomosis from a tunnel-system that opens externally by more 80 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. than two openings. Thus the canals in question are strictly comparable in character to the intercanals of the Ascons, and indeed the species bears some external resemblance to certain of those Calcarea. Needless to say the above perforating passages, for which the name intercanal may be borrowed, are throughout lined with the dermal layer as is of course the entire surface externally exposed {vide PI. YI., fig. 7, in which the dermal surface is colored yellow). The external depressions are on the whole deepest, and the intercanals most frequently developed, in the lower part of the bodv. In that re2;ion the thickness of tlie bodv-wall — consider- ing this to be represented by the space between the general surfaces of the exterior and of the central gastral cavity — must be said to be considerable, being 90 mm. or more in very large specimens. But, if we restrict the application of the term " body-wall " to that layer of the sponge-tissue which is bounded externally by the dermal layer and internally by the excurrent surface, irrespective of whether the latter belongs to the gastral cavity or to the canals opeining into it, the wall-thickness is nowhere very great, being usually 2-3 mm. and probably never more than 5 mm. Apart from the external irregularities above described, the dermal surface is smooth. The dermal layer is of an extremely delicate kind. The hypodermal lattice-work is made up of beams which are so fine as to be barely discernible with the naked eye. The closely set incurrent apertures, visible through the dermal layer, are small, measuring not more than 2 mm, iu diameter. The gastral surface (PI. VI., fig. 7) presents an appearance essentially similar to that of C. meijeri. Above and near the oscular margin, there occur only small roundish excurrent aper- CRATEROMORPHA CORRUGATA, 81 tares, mostlv under 2 mm. in diameter. Lower down, larger apertures likewise of roundish or oval shape are added to them until the largest occurring in the deepest part of the cavity may- measure 1'") mm. or more across. Centrally at the bottom there is an irregular space with comparatiyely small apertures, as we haye seen also in C. weyeri. The larger apertures usually do not lie very close together but are separated from one another by a rather wide interspace occupied l)y small apertures only. While the smaller apertures show a sharp angular edge, the larger ones are without any edge at all, the transition of the central gastral cavity into the excurrent passages being gradual and imperceptible. The latter are sometimes shallow and niche- like, at other times much deeper and pit-like or canal-like. The appearance of their wall is essentially that of an extension of the gastral surface. The larger deeply penetrating excurrent canals, as can be determined by cutting them open, often but not always intercommunicate with one another. The anastomosis reminds one strongly of the gastrocanals or the tubar system of the Ascons. It is plainly associated, both genetically and in topo- graphical relation, with the intercanal system of the exterior. The stalk is nearly as long as, and sometimes perceptibly longer than, the body proper. In general appearance it is scarcely distinguishable from that of C. meyeri. It is firm throughout, the lower part being quite hard and compact. The lower end may expand into an attachment disc. Superiorly it gradually broadens, acquiring a densely and longitudinally fibrous appearance and an obtusely polygonal shape in cross-section. On severing the sponge-body, the upper end of the stalk is found to he divided into a few l)ranches ; so, at any rate, after the sponge has attained a large size. It is solely by this peculiarity 82 ART. 7. T. IJIMA : HEXACTINELLTDA, IV. that I identify the completely macerated stalk shown in PI. VI., fig. Ô, as that of the present species. The liranched state is to be accounted for by the intercanal or intercanals that go right through the sponge at the junction of the l)ody with the stalk. The figure just cited will give a good idea of the manner in which the excurrent canals traverse the stalk. I may now add some remarks concerning certain individual specimens. The largest specimen I have seen (O. C No. 40B4) was 020 mm. high, the body measuring fully 250 mm. in greatest breadth. The typically shaped specimen (O. C. No. 108), shown in PL VI., fig. 6, measures 247 mm. in height and 190 mm. in greatest breadth. Htalk, about 90 mm. long. Osculum, 120- 140 mm. in diameter. Gastial cavity not deeper than 100 mm. Greatest thickness of body- wall (as measured between two points in the gastral and the general external surface), 90 mm. or more. Another exquisitely preserved specimen — that depicted in PL VI., fig. 7, in a longitudinal section (8. C. M. No. 365, from (lutside Okinose)— is smaller. Total height IIT) mm. Breadth at the osculum, oO mm. Greatest l)readth, 74 mm. Stalk, ahout (>0 mm. long and ^-14 mm. })r()ad. Gastral cavity, 40 mm. deep. An individual with an uiuisually widely expanded calyx had a lireadth of 300 mm. and a height of 2öO mm. Of special interest are the two smallest specimens I have had at my disposal (S. C ^l. Nos, 484 & 485, shown in the appended text-figure 2). The smaller of the two (Ä) is about CRATEROMORPHA CORRÜGATA. 83 ïext-fiffure 2. 50 mm, high ; size of the oscuhim with out-tiaring rim, 26 mm. hy 11) mm.; and breadth of the body at base, 2o mm. l)y 18 mm. The other specimen [B] is 80 mm. high ; body, o4 mm. long, and 27- 37 mm. broad ; size of the oblong osculum, 34 mm. by 16 mm. In both the lateral compression of the body is distinct, a fact which I have not specially noticed in all the larger specimens. Both the small specimens show on the der- plia comujata ij., iVoio Outside okiiiosé. niai snrfacc of the bodv a numbci The arrows indicatu tlie intercumiaiinicut- Two small -pecimens of Craln-'naor- cornicfata Ij., iVoio Outside Okiiic arrows indicate tlie intercuimaiiniu ing external depressions, f nat. size. of dimple-like dcprCSsionS. SomC not all — of the deeper and pit-like depressions intercommunicate with one another, representing an early stage in the formation of intercanals. From their general appearance it is to be con- cluded that the intercanals result from the breaking through of adjacent external depressions, which become deeper as the sponge advances in growth, and apparently not by the fusion of tubular outbulgings from the sponge-wall, although the two processes, as both actually occur among the Ascons, are to be considered as modifications of one and the same process leading to the same result. In the two little specimens under consideration, the junc- tion of the body with the stalk is simple exactly as in C. ineyeri ; the piercing through ol' that part seems to take place at a later stage of the growth of the sponge. 84 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. Spiculation. The jtarenchymal'm do not contain hexactins but consist ex- clusively of diactins, as in C. meyeri rugosa. The diactins are for the most part thin and small, terminating with i-ough swollen ends; generally under l.ö mm. in length and 10 jx in thickness. Occasionally there occur larger diactins which may be called the principalia. These may attain a length of 3 mm. and a breadth of 5b II at the middle. They are bow-like or ahnot?t straight and differ from the smaller parenchymalia in tapering towards both ends. On account of the general smallness and fineness of the parenchymalia (PI. VI., fig. 8) the consistency of the sponge - body is soft and delicate, as ascertained on spirit specimens. The firm stalk, on the other hand, contains parenchymal diactins wdiich may be 10 mm. or more long with a thickness of nearly 80 ij.. In its lower portion is observed the usual synapticular coalescence between the principal supporting spicules. The beams of the rigid framework are nearly smooth all over, the microtubercles being present at places only in a sparse num- ber, much as I have seen them in 0. pachyactina. The hypodermalia are mainly pentactins, which are supple- mented by occasional diactins. The pentactins are small to medium sized, the rays measuring up to 500 //. in length and 83 [J. \n Ijreadth at base. The unpaired proximal ray is some- what longer than any of the paratangentials in the same spicule. Each ray gradually tapers towards the roughened, usually coni- cally pointed end. The above pentactins exist rather copiously^ the spicular centers l)eing separated -from one another by an interval which is a])proximately equal to the length of the para- CRATEROMORPHA CORRTJGATA. 8o tangential rays. With these rays they form a fine hypodermal network, the small meshes of which are irregularly angular or are often incompletely enclosed. — The hypodermal diactins are fine and ditter in no way from the smaller parenchymalia. In one specimen of the sponge they were found in tolerable fre- quence ; in others they were rather rare. The dermalia (PI. VI., figs. 1 & 2) are rough stauractins and pentactins, the former predominating. Length of ray as measured from the spicular center, 85-138 a. In certain speci- mens I have frequently seen the stauractinic forms in possession of the rudiment of the fifth (proximal) ray in the form of a boss (fig. 2). The four paratangentials of a spicule are in a plane which is but slightly convex on the outside and is often nearly perfectly fiat. The roughness of the ray surface is, as a general matter, less pronounced than in C. meyeri. It often diminishes towards the base of the rays where it is altogethor lost (fig. 1). Along the thin oscular margin the dermalia are found to be represented now and then by tauactins and even liy diactins. The latter seem to intergrade with the parenchymalia of the region by forms of intermediate size and character. The yastralia are quite like the dermalia. There occur both stauractins and pentactins amongst them, Itut their number must be said to be sparse, being found in scattered distribution. A considerable area of the gastral surface may sometimes be searched in vain for gastralia. No special canalaria have been observed. The Itexasiera of the species closely agree with those of most other Crateromorpha. They are : 86 ART. T.^-I. IJIMA : HEXAGTINELLIDA, IT. Firstly, the oxj^liexasters (PI. VI., fig. o) which are of quite common occiin-ence. Dinmeter, 80-114 p.. Ver}^ rarely I have met with oxyhexasters in which (jne or two of the principals — never all the six (hexactinose) — bore each a single terminal which was bent at base in the well-known manner. The rule is that the six principals bear each 2 or 3, sometimes 4, diverging terminals. Except at base, these are nearly straight ; otherwise they are slightly wavy. Their surface is oljsoletely rough or nearly smooth. I have noticed that the principals are, generally at least, perceptibly longer than those in the corresponding rosette of C. meyeri. But such finer points in the character of the rosette are probably subject to considerable individual varia- tions. And secondly, the minute discohexasters (PI. VI., fig. 4) which have been met with in some numbers — by no means abundantly — in the subderinal s})ace. In no other region of the body have they been discovered. Diameter, 40-30 /i. The entire shape is quite spherical, all the terminal discs being uniformly distributed on the surface. Under a high power of the micro- scope the minute terminal disc is seen to l)e supplied with six, and sometimes more, marginal teeth. The numerous fine termi- nals arise from all over the convex surface of the disc at the end of each principal, as is usually the case with the so-called mi- crodiscohexaster of the Ivossellidse. HYALASCUS. 87 HYALASCUS TjrMA. Vase-like, firmly attached by contracted base; large. Gastral surface lined with distinct endosomal layer covering over the excnrrent canalar apertures. Parenchymalia, diactins only. Hypodermalia, pentact- ins supplemented with some diactins. Dermalia, gener- ally rough pentactins; occasionally hexactins. Gas- tralia, similar hexactins. Oxyhexaster represented by hemiliexactinose and hexnctinose forms. Discohexaster in one small form with very fine terminals. The genus was originally instituted by me for the reception of a single species which I briefly described in 'g6 under the name of TL sac/amlensis. F. E. Schulze ('97, p. 520), in Ins revision of the Asconématidtç, placed this genus and species under that familv. For the orounds that have led me to take up the genus under the Rossellidiie, the reader is referred to my Contribution HI. ('03, pp. 78-82). In '98 I referred to the same genus a second species, If. glganteiis. And now T feel the necessity of establishing a third and new s]iecies, H. similis. The genus seems most nearly related to ScypJiiiliuiii, Vifrol- lula and Crater omoripha. Its distinction from tliese as well as from certain other allied genera may l)e gleaneut rarely in a perfectly intact state. HYALASCUS SIMILIS no v. sp. Just in time to admit of the insertion of this description, ]Mr. Alan Owston has shown me, with his usual courtesy, a l)eautiful and excellently preserved specimen (O. C. No. 7803) acquired by him not long ago from the coast of the Province of Tötömi. It at first appeared to me to l)e a second specimen of Hyalasciis sagamiensis, but a close study of the spiculation has led me to think otherwise and I propose to call it H. similis n. sj-». The exact circumstances of the capture of the specimen are not known. The shape of the specimen is shown in the accompanying text-figure 3 in one-fifth natural size. The saccu- lar and rather thin-walled body measures 380mm. in total height. The main part of it is distinctly ïcxt-fiouie 3. laterally com- Hyalaxcui^ aimilis n. sp. in i; natural size. Two views of tlie -, a j. j.r, same specimen seen from different directions. prCSSeu. A.t thC HYALASCÜS SIMILIS. 97 ventricose middle the breadth measures sagittally 230 mm. and transversely löO mm. The body contracts towards both ends, but less so above than l)elow. The oscular region, which termi- nates in a thin simple edge, is outflaring to a greater or less degree in different places. The osculum measures 135 mm. by loO mm. in diameter, the lesser and the greater diameter in relation to those of the laterally compressed main part of the body being exactly reversed in orientation. The basal end with irregular longitudinal ribs and furrows measures about 70 mm. across. The surface of attachment to the firm substratum is in several irregularly shaped patches. The wall is only al)Out 5 mm. thick in the middle of the sponge. The gastral cavity extends at the bottom into the stalk-like basal region. As regards the general appearance of both the external and internal surfaces, the texture and the canals, what I have said under H. sagamiensis (pp. 89 & 90) is equally applicable here. The similarity further extends into the spiculation so that this again need not be described in detail except in regard to one point which constitutes the chief, if not the only, distinctive character of the present species. The point in question concerns the discohexaster (text- figure 4). This occurs not uncommonly in scattered distribution throughout the parenchyma, though by no means so abundantly as the oxyhexasters. Being of very inconspicuous appearance, a careful ex- amination of the preparations is necessary in order to find one. Moreover, the exceedingly fine terminals easily break off, as that it is Text-fi^ire 4. „,,,., , usual to find the discohexaster in a more or less ilie discohexaster of i/. «w//«. 500 X. çiamaged condition and not infrequently repre- 98 ART. 7. — T. IJIMA : HEXACTINELLIDA, IV. seiited by only the central parts. The size is small, measuring only 46-50 p. in diameter. The general form can not be said to be spherical, since the terminals to each principal form a dis- tinctly separate tuft shaped like the perianth of a lily. The six principals arising from the small and spherical central node are short ; the measurement from end to end of two principals in one axis is scarcely 10 //.. Their outer ends do not appear to be expanded. The thin terminals (20 /^ long), of which there are 10, 12 or more in a tuft at the end of each principal, are of the most delicate description. They thicken just perceptibly to- wards the outer end which terminates in a minute pinhead-like disc. The entire ray, with the outwardly expanding tuft of ter- minals, is in appearance not unlike that of the octaster of certain Acanthascin?e. The above discohexaster as compared with that of H. saganii- ensis presents marked differences. (Compare text-figure 4 [magn. 500 times] with PI. VIL, fig. 5 [magn. 300 times]). It is con- siderably smaller (46-50 /i dia. against 80-90 // dia.), and the fact that ^ the much more numerous terminals are arranged in distinct tufts gives to the spicule a very characteristic appearance. I think the différences indicated are of sufficient import to justify the specific separation of IL similis from the foregoing species.* " Too late to admit of introducins; change?; in tlie text I iind that the discoliexaster liere described is not the only kind but that there is to be ascribed to the species another which I had entirely overlooked. Having occasion to re-examine the preparations, I have come across a discohexaster lying near tlie gastral surface, which closely resembles thai of H. sa'jamienÙH (PI. VII., fig. 5). Spherical in shape, with diameter of 76 [;.. Terminals, :! or 4 to each short and minute principal; exceedingly fine and very slightly thickened to- wards the outer end which terminates in a minute disc. Having once seen it, I have suc- ceeded after a prolonged search on several preparations in discovering a few more of tlie same kind. It mu-t be said that this discohexaster is of very rare occurrence; possibly it is on the verge of disajipearance. But then it seems undeniable that we have in it a dis- cohexaster form which is common to both H. sagamiensis and //. similis. Was not the smaller form (text-fig. 4), considered in the text to be peculiar to the latter species, over- HYALASCUS SIMILIS. 99 Of other points in the spiculation I may put down the fol- lowing notes, though these are in the main nothing but repeti- tions of what I have already stated under //. sagamiensis. Farenchymalia, slender diactins of varying length (up to 25 mm. or more) and thickness (up tp about 175 /^) ; no hexac- tins. Medium-sized diactins under the gastral layer often with cruciately disj)Osed knobs at the spicular center. Hypodermalia, slender-rayed oxypentactins with bent para- tangentials which may be 1 mm. long. Dermalia, pentactins, sometimes hexactins ; prickly all over. Length of ray, 80-114 u. Oastralia, hexactins in which the free proximal ray may attain a length of 285 ix; other rays 120-165 p. long. Eays prickly but smooth at base. Oxyhexaster, hemihexactinose and often hexactinose. Diame- ter, 90-125 II. All the spicules here remarked upon, let it be repeated, essentially agree with those of H. sagamiensis in the manner of their arrangement as well as in details of character. Finally, I regard it worth while to mention certain oxyhex- asters which seemed to be in a state intermediate, so to say, between the hemihexactinose and the hexactinose forms. Such looked by me in the specimen of the former ? If it was, H. simili-i at once loses its specific status and slionld be combined with H. sagamiensis. However, careful re-examinations of the preparations of tlie latter, partly newly made from a piece in my possession, have en- tirely failed to reveal a second discohexaster form in addition to the one attributed to the species in the text. So the matter stands thus: H. sagamiensis has a single discohexaster form ; H. similis possesses the same in quite a limited number and, in addition to it, a sec nd smaller form which occurs in moderate abundance. Whether with more specimens in hand the difference indicated can be maintained as a specific distinction, can not be foretold. Since anyway it seems impossible to base generic separation on that difiërence, it becomes necessary to make a slight emendation in the generic diagnosis given on p. 87, the last sentence on which sliould read " Discohexasters m one or two small forms witli very tine terminals," instead of " Discohexasters in one small form", &c. 100 ART. 7.— I. IJIMA : HEXCTINELLIDA, IV. Text- figure 5. All oxyliexaster in tlie last stage of becoming liexaetiiwse. 440 X . cases were incidentally met with more than once in the present species. The accompanying text- figure 5 sliow^s one of them. Five of the rays are simple, heing either straight or bent at base, exactly like those in hexastinose forms of the oxyhexaster ; in them the principals are strictly uniterminal. The sixth of the principals, which are all of quite an obsolete length, bears, Ijesides a normally developed terminal, another of spurious size. Were this as much developed as its fellow on the same principal, we should have a normally hemihexactinose form in which a single principal is biterminal and all the other five are uniterminal. On the other hand, if it should altogether dis- appear, as it apparently is on the verge of doing, the result w^ould be a hexactinose form, the exact like of which may not be difiicult to find among the oxyhexasters of the species. In my experience certain other Rossellids have also yielded similar cases of oxyhexasters being in the last stage of transition into the hexactinose statu. HYALASCUS GIGANTEUS Li. PL VIII., figs. 3-16. Hyalaacus (j'lganteus, Ijima, *g8, p. 'jO, This species is described on the basis of a large fragment which originally belonged to Mr. Alan Owston (O. C. Ko. HYALASCUS SIMILIS. lOl 4063) and is now to be seen in the British Museum. It came from the Sagami Sea, the more exact locality being unknown. The thick plate-like fragment is nearly as large as the blade of a tennis racket. Greatest length, 480 mm. Greatest breadth, 255 mm. Thickness, 45 mm. It is torn all around, had evidently been flattened out during desiccation and does not warrant in- ference as to the shape and size of the original sponge except that it must have formed part of the wall of a very large and presumably vase-like individual. The piece, being preserved in the dry state, is light and of a cavernous appearance on account of broad incurrent and excurrent canals. The spongy septa between the two canal systems are rather thick. The incurrent canals, which in places may be 15 mm. wide, seem to freely anastomose with their branches, thus forming a continuous system extending throughout the whole specimen, while the excurrent canals only occasionally intercommunicate with the branches and more usually remain separate from one another. The dermal surface (PL VIII., fig. 3) is much macerated, so that its exact nature is difficult to determine. Here and there the apertures of incurrent canals appear as oval or roundish openings of not over 10 mm. diameter. Not that they are all freely open, Init a large number of them are seen to be covered over with an uneven and irregularly cobweb-like layer formed of spicular bundles which intersect, unite and branch in quite an indefinite manner. The thicker bundles may show in places a thickness of over 1 mm. Between the canalar apertures the said layer is indistinguishable from the tissue of the parenchymal septa. It is not to be doubted that tlie layer constitutes the hypodermal framework. The dermal layer proper remains in small patches only in a few places. Even if it were extensively 102 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. preserved, I should think it would be rather inconspicuous be- cause it closely overlies the choauosoinal surface in parts and the cobweb-like hypodermal framework in other parts. Some deep and chasm-like slits occur on the external surface, their wall presenting a granular appearance. This is due to broken ends of the parenchymalia which there appear to be even- ly nipped off. The slits were evidently made by Ophiurons which took their abode in the sponge, as is apparent from such as still contain that animal. The gastral surface (PI. YIII., fig. 4) is in a much better state of preservation. The excurrent canalar apertures, of very various sizes under 18 mm. diameter, are covered over witli an endosomal latticew^ork composed mainly of moderately strong and compact strands which may be called hypogastralia. The meshes are angular but irregular in shape, mostly measuring 1-3 mm. in length of sides. They are all open, which I believe is the natural state. At any rate, the gastralia proper are only found either distributed singly on the beams, or several together on the nodes, of the hypogastral latticework. Remarkable is the fact that under certain circumstances the above endosomal latticework, instead of extending merely in a plane layer, may be developed more or less in the third dimen- sion also so as to form a trabecular system of some thickness. Thus, I have seen some cases of the excurrent canalar apertures being incompletely closed, so to say, by a spongy partition. The larger the excurrent aperture, the deeper is the pit-like canal it leads into. Usually about half-way through the sponge- wall, the larger canals begin to divide uj) into branches. And these branches, unless they happen to be small, are seen to begin with apertures which are spanned by essentially the same lattice- HYALASCÜS SIMILIS. 103 work as that ünardins: tlie entrance into the main canal on the gastral surface. In other Avorcls, the lining layer of the gastral cavity is duplicated on the wall of the main passage of excurrent canals. But this duplication does not occur in all the smaller canals. It must be said that in respect of the wall- structure raacro- scopically considered, the sponge shows a rather wide deviation from both //. sagamiensis and //. sinnlis. Nevertheless, there exist several points of close resemblance in spiculation, a fact which mainly weighed with me in referring the species to the same genus. The species is probably very nearly related to Crateromorpha ; but I was deterred from placing it under that genus solely by the presence of hexactins among the dermalia and of a covering latticework to excurrent canalar apertures and by the fact that the gastralia are all hexactins. Spiculation. The parenchymalia are exclusively slender diactins of widelv varying sizes. The largest, the principalia, may reach 23 mm. in length and 175 !'- in breadth at the middle, while the finest are of the ordinary dimensions of a comitalia only about 7 // thick. They are in general irregularly bent or wavy, either gradually tapering out to a point at both ends or terminating with conical or rounded tips. Subterminally the surfece seems to 1)e always rough on account of microtubercles that are sometimes wart-like and sometimes spine-like. As a rule there exists not the slightest trace of an external swelling marking the spicular center. Both the hypodermal and hypogaüral beams (PI. VIII., 104 ART. 7. — I. IJIMA : HEXACTINELLTDA, IV. figs. 5, 6), giving support to the ectosome and the endosome respectively, consist in the main of bundles of diactins, either loose or compact, which are comparable to the smaller parenchy- malia. Among themselves the hypodermal and hypogastral diact- ins are of various sizes, with thickness reaching up to 30 //. The shorter ones amongst them usually show a gentle annular swelling around the spicular center. Hypogastrall}^ there not infrequently occur diactins as short as, or at any rate not much longer than, the axial length of the gastral hexactins, like which spicules such short diactins may have microtubercles sprinkled nearly all over the surface. Their appearance is such that they might have arisen directly from the gastralia by loss of the four rays in two axes. Hypodermal pentactins were discovered only in a limited num- ber. There can scarcely be a doubt as to their being clearly dis- tinguishable from the dermalia, though under certain circumstances they seem to closely approach these both in size and appearance. In them the unpaired proximal ray is the longest of all the five rays ; it may be more than twice as long as the paratangential. The latter, in one case measured, was 700 ii long and in another case, considerably over that ; but it may sometimes be as short as 300 IX. In short the size is subject to considerable variation. The paratangentials are rough nearly throughout their entire length, the roughness being quite similar to that of the dermalia, while the proximal ray is generally smooth except towards the end. The dermalia are predominantly pentactins (PL VIII., fig. 3) with a sixth ray represented by a small hillock-like promi- nence on the distal side. In these the unpaired proximal ray is HYALASCüS GIGANTEUS. 105 usually somewhat shorter than the paratangentials. Not infre- quently the proximal ray is also reduced to a knob-like rudi- ment, so that the spicule takes the form of a stauractin (PL VIII., fig. 9). On the other hand, tlie distal knob of the pentactins is sometimes prolonged in varying degrees, leading to the regular hexactinic form which is in fact occasionally met with. In all cases the rays are rather slender, tapering but slightly outwards and terminating with rounded or subconical tips ; the roughness of surface, caused by rather unprominent microtubercles, is most pronounced in the outer half of the rays and becomes gradually weaker and inconspicuous towards the base. In size the dermalia show a somewhat wide range of variation. As measured on the paratangentials, the ray-length (half-axis) measures 120-200 // with a breadth of about 7 h // at base. The plane of the four paratangentials in a pentactinic dermalia is flat or just percep- tibly arched. In some parts of the sponge surface, I have seen the dermalia form a regularly quadrate-meshed lacework in which the meshes measured on an average 110 f-f- in length of sides. In other places they are evidently quite irregular in the mutual relation of their paratangential rays. The gastralia (PI. VI IL, fig. 7) are similar but on the whole much larger hexactins. The length of rays, as measured from the spicular center, is mostly 165-385 /^. In the largest of the spicules, the breadth of rays at base may reach 20 ij.. The six rays are often unequally long, but I could not deduce the rule that the free proximal ray is the longest. It is only occasionally that a number of the gastralia are found so grouped as to form approximately quadrate meshes with their paratangentials. The distribution is for the most part quite irregular. As before men- 106 ART. 7. — I. IJIMA : HEXACTINELLTDA, IV. tioned, a continuous autogastral layer covering np the meshes of the hypogastral framework seems not to exist. Perhaps worth noting is the fact that the paratangentials are frequently observed running among or under, instead of over, the diactinic elements of the hypogastral beams. Hexactins similar to the gastralia occur in some number as canalaria on the l)eams of the lattice-like layer lining the lumen of the larger excurrent canals. Of the hexasters, oxyhexasters (PL VIIT., figs. 12-16) occur in abundance in the choanosome. Diameter or axial length, 80- 125 //. They are mostly hexactinose and less frequently herai- hexactinose. In the latter case, one to three of the six, extremely short or nearly entirely atrophied principals bear each two widely diverging terminals. In fact, three seems to be the utmost num- ber of principals that may be biterminal in an oxyhexaster. At any rate, normally developed oxyhexasters, i. e., those in wdiich all the six principals bear more than one terminal each, were on no occasion met with. The uniterminal rays, i. e., those consist- ing of a principal continued into a single terminal, w^hether be- longing to a hexactinose or to a hemihexactinose oxyhexaster, are either bent at base or are straight or nearly straight from the origin at the central node. PL VIII., fig. 16, represents a rare case of a hemihexactinose oxyhexaster w^ith peculiarly bent rays. Now^ and then there are observed cases of a principal bearing a terminal of normal length and in addition the short rudiment of a second. PL VIII., fig. 14, is an example of such cases ; and text-figure 5, on p. 100, may well pass for one taken from the present sj^ecies. Occasionally a principal together with its terminal seems to be totally suppressed in development ; only AULOSACCÜS. 107 in this way can be explained such oxyhexaster forms as are shown in PI. YIII., figs. 12 and 13, or those I have seen in which less than six (e. g., only four) rays in all emanated from the central node. In all the oxyhexasters the terminals are rough-surfaced. The roughness is frequently seen to be caused by minute retro- verted tubercles. The precise extent of the short axial filaments forming the central cross can be clearly observed if one goes through the necessary steps of preparation. It needs simply to be stated that PI. XIV., figs. 24 and 25, may be said to represent exactly the state of things in the central part of oxyhexasters of the present species. The discohexasters (PI. VIII. , figs. 10 & 11) are common in both the choauosome and the gastral layer. At j^laces they are much more numerous than the oxyhexasters. They occur in one small and delicate form of spherical shape, measuring only 30-38 /^- in diameter. Each short principal bears sometimes about 10, and sometimes only about 6, very slender terminals ending in a comparatively large disc with about half a dozen, slender, marginal teeth. AULOSACCÜS IjiMA. Aulosaccus, Ijima, 'g6. Calycosaccus, F. E. Schulze, 'gg. Vase-like, thick-walled, firmly attached at base; moderately large. Gastral surface lined with a conti- 108 AKT. 7. — I. IJIMA : HEXACTINELLIDA, IV. iiuous endosomal layer. Parenchymalia, of diactins only. Without pentactinic liypodermalia. Dermalia, various according to species. Gastralia, rough hexact- ius. Oxyhexasters show a greater or less tendency to occur in hemiliexactinose and hexactinose forms. Disco- hexasters in two spherical or approximately spheri- cal forms: macrodiscohexaster and microdiscohexaster. The former with numerous terminals and usually strik- ingly large in size. The above diagnosis is drawn up regarding the genus, as it now stands, as made up of the three following species : 1. A. schidzei I J. li. A. ijiviai {¥. E. yicn.) = Calycosaccms ijimai F. E. Sch. 3. A. mitsukurll Ij. The first and the last mentioned species, which were years ago described by me in brief, will be treated of in full in this Contribution. The second mentioned species is one which was described by F. E. Schulze ('99), who made of it a special genus, Calyeosac- CU8. He was certainly fully cognizant of the close similarity of his genus and species to my Aulosaccus schulzei, so much so that, as he clearly implies (/. c, p. 100), he would not have hesitated to associate the two forms in the same genus, had it not been for a difference in the character of their dermalia. That difference was the one on which he based the distinction between the fami- lies Asconematidse and Rossellid?e. His species had to be placed under the Asconematidie, while Aulosaccus was to be considered a Rossellid. From such a position, it of course followed that the species must receive a generic designation of its own. Now, in AULOSACCUS. 109 mv Cüiitribution III., pp. 78-83, I have endeavored to show that the Asconematidse had better be dissolved and that a number of its genera, Calycosaccm for one, should be taken over by the Rossellidœ. If I am right in so doing, Calycosaccus as a name for a Eossellid genus seems to lose its claim for existence, as was indeed anticipated by F. E. Schulze himself. Nothing more than a specific value can be attached to the difference between G. i/jimai and Ä. schulzei. It is with a much less degree of certainty that I refer A. mitsukurii to the srenus. It differs decidedlv from both A. schulzei and A. ijiniai in the possession of strong prostat needles and in tlie comparatively small size of its macrodiscohexaster. That these differences sufficiently warrant generic distinction seems to me to be doubtful, so that the species may best be left for the present in the genus to which it was originally assigned. Aulosaccus shows greatest afiinity to Scyphidium and Rossella in that it possesses like these the two kinds of discohexasters, but differs from both in having no pentactinic hypodermalia. This negative character is also shared bv Aulochom under the Rossellime, but that genus lacks the macrodiscohexaster. If A. mitsukurii were only provided with pentactinic hypodermalia, I should have no hesitation in referring it to Scyphidium and plac- ing it l)y the side of S. longispina, which species it most closely resembles in the rest of its characters. The three species of the genus may be distinguished as follows : a. — Without coiiuli :ind needle-like prustalia. a'. — Dermalia, hexactins or predomiuanlly hexacliiis. Macrodiscohexaster may measure 400 !J. in diameter; its princijjals separate or represented by six heuîis[)lierical bosses AiilosacciitS ijimai (F. E. ScH.). 110 ART. 7. — 1. ijima: hexactin lleida, b^. — Dermalia, pentactins or predominantly pentaclins. Macrodiscohexaster iiiny measure nearly 1 mm. in diameter; its principals fused into a spherical mass Anlof^fiiru.'i Hchulzd I J. ''• — With conuli, from the apex of which ]iroject needle-like prostalia. e'. — Dermalia, ])redominantly staiiiactins. Macrodiscohexaster not over I'JO \>. m diame- ter; its principals separate Aalosaccux mil--; somehow concealed from view, perhaps in that they are too sparse and widely scattered to be easily discovered. At the same time I ACANTHASCIN^.. 135 liold it not impossible that their development may under cir- cumstances be quite sujipressed, in which case the limiting layer alone would stand for the basidictyonal plate. Finally with respect to the hexasters, it may be said that there are three forms occurring together in a. species ; viz. oxy- hexasters, discoctasters and microdiscohexasters. The oxyhexasters are the most al)undant of all. Strong and wide-spread is the tendency shown by them to assume hemi- hexactinose and quite hexactinose forms. In some species the hemihexactinose form predominates; in some others, the hexactinose. In 11. tenuis (F. E. Sch.), as before alluded to, all the oxy- hexasters present appear to be hexactinose. Basing our description on normally developed oxyhexasters, the principals are always very short, — often so exceedingly short as to be called vestigial. The number of terminals most frequently borne by a principal is two f' but it may sometimes l)e three and rarely four. The terminals are nearly smooth or more frequently rough. The roughness may be developed on their l)asal parts into retroverted prickles or barbs. In several sjiecies the oxyhexasters situated in the periphery of the sponge-wall, but particularly in the subdermal * For Ihjse oxyhexasters generally in which tlie principals appear bifurcated in that tliey are provided eacli with two terminals, it has been given by F. E. Schulze ('97 a) as an ajjproxiniate rule that the plane of bifurcation of a principal stands at right angles with that of another belonging to the same axis, and tliat the six separate bifurcation planes in one oxyhexaster of tlie kind correspond to tlie so-c.illed secondary planes of synnnotry in the isometric crystal system, the primary or principal planes of symmetry being given by the principals forming the three axes. So that the princijials and the terminal fork-; should represent all the nine possible planes of symmetry distinguishable in a regular cry.-tal (thre? principal planes determined by the axes and six secondary planes determined by diagonally opposite edges of a cube). I liave not specially gone into the t?sting of the truth rf the above statement; but so far as concerns the two bifurcation planes at the ends of any one axis, my experience makes me hesitate to lay it down as a general rule that tliey are relatively vertically oriented to each other, for the angle referred to seemed to me to be much too variable and indefinite, as observed in a large number of cases in various Hexactinellid species. 136 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. region, differ from those more deeply situated in having some- what longer principals and slightly more slender terminals. Moreover, tlie former generally show a greater total number of terminals, being usually normally developed, although hemihexact- inose and hexatinose forms may be common among the latter. (Compare in PI. XXI., figs. 4 and 5 with figs. 6-8 ; in PL XXII., figs. 7 and 8 with figs. 14 and 15). It must however be borne in mind that between the peripheral and the more deeply located oxyhexasters there is always a gradual transition within one and the same sponge. Of the three oxyhexaster-forms occurring together in a species, it is usually the hexactinose amongst which are found individual oxyhexasters with greatest diameter or axial length, — a fact which I have noticed also in some Rossellinae. It appears as if the reduction in the number of terminals to the minimum, i.e., to one to each principal, in a measure fivors the growth of the spicule in general size. In a number of the species I have specially gone into the observation of the axial cross in hexactinose oxyhexasters. With a little trouble I have found it in all cases an easy matter to demonstrate what I have repeatedly emphasized as to the extent of the axial filaments in the central parts of that oxyhexaster variety. For the rest I may let fig. 2.">, PL XIV., speak for itself. By the side of that figure is another (fig. 24) showing, for the sake of comparison, the extent of the axial cross in the central part of a normal oxyhexaster in which each principal is supplied with two terminals. It scarcely needs to l:»e reiterated that uniterminal principals, irrespective of their occurrence in a hemihexactinose or in a hexactinose oxyhexaster, are joined to their single terminal mostly ACANTHASCIN^. 10/ in a straight line in sncli a manner that the point of junction of the two parls is externally not in the least indicated ; but the terminal may sometimes be l)ent at base exactly as in a hexaster ray with two divei'gent terminals, save that one of these terminals is entirely -wanting, in which case it may be said to be well marked off from the principal bearing it. The discoctasters are spicules peculiar to the Acanthascinae ; in fact their presence constitutes the only reliable criterion by Avhich a Kossellid can be determined as a member of that subfamily. Hence a misgiving might be entertained that should they chance to l)e simply overlooked or not properly identified, or if there existed a species which had lost them only secondarily, the sponge in question would likely be taken up under the Hossellinse, which are apparently ab origine without these charac- teristic spicules. The discoctasters were first reco2;nized l3v F. E. Schulze ('93) to l)e strongly modified discohexasters in which the six ])rincipals have entirely or almost entirely atrophied while the terminals have undergone a new arrangement into eight secondary principals and terminal tufts at points of the central node cor- responding to the eight corners of a cube. It was pointed out l)y the above-mentioned writer that there may exist on the central node and in the center of the space surrounded by every four secondary principals a hump-like prominence representing the outer end of a primary princij^al, and further that this protuberance may run out at base into four radial ridges (see PL XL, fig. 20), marking the course along which the original terminals were laid down in order to combine into the eight secondary j^rincipals. These are frequently not quite cylindi'ical being longitudinally ribbed or at any rate somewhat angulate in cross-section, indicating lo,S ART. 7. — I. TJIMA : TTEXACTINELLTDA , IV. their formation by a coalescence of parts rnnning lengthwise. If more support is needed to establish the correctness of F. E. Schulze's enunciation as to the nature of the discoctaster, a note- worthy fact may be adduced with respect to the typical triaxial cross of filaments inclosed in the central node, which I believe I was the first to demonstrate in the spicule under consideration ('97). As I have repeatedly had occasion to remark, the cross becomes plainly visible if the spicule be examined in glycerine or in any otlier medium of a refractive power similar to that of the siliceous substance, and its six points are seen to be always directed towards the middle of the space surrounded by every set of four secondary principals. That space is either simple-surfaced and somewhat concave or shows in the middle the hump-like protuberance before alluded to. In the latter case, each arm of the axial cross extends directly into the protuberance towards which it is directed (see PI. XI., fig. 20). Of some interest is the not infrequent occurrence of malformed discoctasters in which there exist such a primary t 'rminal or terminals as stand out from the central node, having apparently been left fi-ee without fusing with any of the secondary principals (PI. XIIL, fig. -3; PI. XVI., fig. 10; PI. XVIIL, fig. T) ; etc.). The cases may be said to be in a measure suggestive of the condition of the spicule before the primary terminals were brought together and joined into the eight secondary tufts. Noteworthy also seems the presence of minute vacuole-like spaces in the secondary princi})als and sometimes also in the central ]iode (PL XIL, figs. 2.",, 27 ; PI XV., fig. 1) ; etc.). In the former as well as in the ridges running out from the six pro- tuberances on the central node, the little spaces are usually found arranged in a row or rows that run longitudinally. There can scarcely be a doubt as to their being vacancies between the ACANTHASCUS. 189 primary termmals, left unfilled by the siliceous deposit Avliieli solders these togetlier {cfr. p. 1 IG). The free terminals at the end of secondary principals are always rougli, tliougii often obsQletely so. The minute terminal discs are either simple like pinheads or else toothed at the margin. The general size of discoctasters and the proportion and shape of their parts are of great systematic value and may he utilized in distinguishing the different species. It frequently happens that in one and the same species the deeply situated discoctas- ters are considerably larger than those in the periphery of the wall. The Wiicrodiscoliexaslers are minute and extremely delicate rosettes of the usual structure. In general shape they seem to be uniformly spherical, measuring under 40 !'■ in diameter. In some species the diameter is only about 15/<. Not that they have been o1)served in all Acanthascine species, for several species have been found to be without them, though it is difficult to decide in all cases if that negative result was not due simply to oversight or to the individuality of the sponge. ACANTHASCUS F. E. »Sen. llypudermalia, diactins only; at any rate, no pent- actins amongst them ; hence, never veiled. 140 APvT. 7. — I. IJIMA: HEXACTINELLIDA, IV. This genus, as originally established by F, E. Schulze in the Challenger Report, included, besides A. cactus, two other species that were called A. grossularia and A. dubius ; neither of these however possesses discoctasters and hence can not be placed under the Acanthascinte. The former has since been declared by the same writer ('97, p. ô 37) to represent merely a young specimen of liossella antarctica C^vrtek, Avhile the latter was recognized to be a species which should properly be called Rossella dubia (F. E. Sch.). Subse(_j[uently in '98 (p. 55) I described A. alani from the Sagami Sea and in '99 (p. 45) F. E. Schulze put forth his A. plate'i from the Californian coast (E. of St. Diego, 572 m.). Accordingly, tlie genus as it now stands, comprises three species, the difierential characters of which may be gleaned from the following key : «. — Dermalia, predominantly stauractins, occasiüiuil]y peiitacliiis ; gastralia, iDc^lly peutaciiu.*, sometimes stauractins. OxyhexasttT, yo-l.)2 ;j. dia. DiKcot'tasttT, 10()-300 ]i. dia. Microdiscohexaster, lo-^Ö;;. dia \. cadiix F. E. ScH. b. — Dermalia, predominantly pcntactins, iisunlly witli a Unoli-like rudiment of the distal sixth ray ; gastralia, liexactins. a'. — Oxyhexaster, 1Ü0-1.">U ;j. dia. L>iscüctaster, llîO-JÔO [j. dia. Mierodiscohexaster, at most 20 [J. dia A. platei F. E. Sch. 6'. — Oxyhexaster, 144-19(1 a dia. Diseoctaster, loti-2"J0 a dia. Microdiscohexaster, 30- 35 u dia A. aani Ij. ACANTHASCUS CACTUS F. E. Sch. ri. XL, ligs. 10-22 and PJ. XII. Acaidhascas cactus. F. E. Schulze, '86, p. 49. '87 (!), p. 148; PI. LVII., figs. 1-7. '97, p. -351.— Ijima, '97, p. 48. ACANTHASCUS CACTUS. 141 The species was first described by F. E. Schulze from a single small specimen which was obtained l)y Döderlein probably from a shop in Enoshima. Some years later, as the fishermen became aware of the demand by naturalists, specimens 1)egan to be brought to us in such numbers that soon we had to decline to purchase them unless unusually large in size or beautiful in appearance. Evidently the species is one of the most abundant and most widely distributed Hexactinellids of the Sagami Sea. It is known to the fishermen along the coast under the name of " Wata " (cotton) or " Wataboshi " (cotton-hood). Nearly half a hundred specimens now lie before me, all obtained from depths between 220 and 572 m. and from a bottom of volcanic origin in the Sagami Sea. The more exact localities known to me are : Yodomi, Mochiyama, Inside Okinose, Outside Okinose and Hondta. I myself have collected several fragments at a spot about 2 kilo- meters oft' the village of Senzu on the northern coast of Vrits Island, from a depth of 220 fathoms and a l)ottom of a Ijlack basalt-like rock. Mr. T. Tsuchida, «Assistant in the ^NJisaki Marine Laboratory, obtained some fragments on the north of Okinosé from a depth of 235 fathoms. In 1900, the U. S. Fish Commission S. S. " Albatross " trawled up several fragments, together with llhahdocahjptus mollis, at her Station 3697 (about 4 kilometers off the mouth of the Sakawa river in Sagami Province; 265-120 faths.; black volcanic mud). While most specimens grow on a tufaceous substratum, I have seen some attached to the dead skeletons of other Hexactinellids [Hexavlindla ventilahrum, Farrea occa). On the following page aj-e given sketches of a few selected specimens in order to give an idea of the general appearance of the sponge. The shape is in general like that of a Avide-mouthed 142 ART. 7. — I. IJIMA : IIEXACTIXELLIDA, IV. Text-figure G. Acanthascus cactus. All about \ natural size. A. — S. ( '. M. Xo. 428. Two persons in connection with each (ither at l)asc. The larger person : 140 mm. high; 70 mm. hy 4.']nnn. broad at middle; osculum about 41mm. dia.; Mall Ô mna. thick at middle. E.— Ö. C. M. jS'o. 42U. 200mm. high; osculum 210 )nm. by 148 mm.; wall 7 mm. thick at middle, 10 mm. lower down. C.-S. C. M. No. 4;;(l. ;;i.")mn:. liigh; lUOmui. by 86 mm. broad at middU'; osculum 140 l)y llônnn.; wall 8 nnn. thick at michlle and 10mm. near base. 1».-S. ('. :M. No. 4;!]. 470 nnn. high; 140mm. broad at middle; osculum 178 mm.; wall S mm. thick at middle. vasc, l)roa(-k'8t at or near the top and narrowed towards the base. Young ;ind small individuals generally show the wall somewhat closed in at the oscular margin. This, in the larger specimens, is usually directed straight upwards ACANTHASCUS CACTUS. 143 or otlierwise somewhat outwards. Sometimes the body expands superiorly in such a manner that it presents a funnel-like shape. The sponge stands either erect Irom the basal attachment or is hent more or less in the basal part. Laterally it is frequently more or less distinctly compressed, especially in the basal region ; but the upper part is about as often simply roundish or irregularly roundish in cross-section. The wide and deep gastral cavity extends close to the infer- ior end, so that the blind wall presenting the irregularly shaped surface of basal attachment is not thicker than the adjoining 2;)art of the lateral wall. This is on the whole only moderately thick, gradually thinning above towards the thin and simple-edged oscular margin. While many specimens represent pure and simj)le individuals, each being in jDOSsession of a single large osculum at the superior end, others show complication in that they consist of a main j)erson bearing one, two or sometimes more, generally much smaller daughter persons secondarily produced by budding. The bud is at first a blind-sac-like outbubino; of the wall, which eventuallv opens an osculum at the top. The oscular rim is for some time closed inwards but may flare out later. The bud-formation seems to be confined to the basal region, or at any rate to the lower half of the mother body. Text-figure 6, A, represents a case in which two persons, one somewhat smaller than the other, arise side by side from a common basal part. The specimen shown in C\ same figure, Ijears three buds : two on the convex sagittal edge of the bent l)asal region, and one on the lateral side. Several cases have been encountered of specimens similarly bent at base and bearing on the convex side a single bud directed in the opposite direction from the main part of the mother sponge, the 144 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. general shape being tlien not unlike that of the Rhahdocalyptus shown in PL XYIT. The Imds are sometimes represented merely bv secondary oseula which are iinaccomjmnied by any outbulging of the surrounding parts and thus appear like simple gaps in the wall. Thus, the specimen marked B in text-figure 6 is in pos- session of more than one small sucli osculum in addition to the large primär}^ one at the superior end. The species grows to very considerable dimensions. The largest complete specimen before me (S. C. M, No. 432) is 432 mm. high, and consists of two large persons and of a l)OSS-like cœcnm of the size of one's fist, all these parts arising from a common basal portion. There is also in the Science College a larger fragment with an osculum measuring 240 mm. in diameter. On the other hand, the smallest individual I have seen was only al)out the size of a walnut. A striking feature of the sponge consists in the sharply apexed conical elevations of the external surface. These occur at various but leather wide intervals. They are usually most numerous, largest and tallest near the middle or in the lower half of the body. In large specimens they may be as high as 25 mm., but are much shorter in the smaller specimens. They grow more and more sparse superiorly towards the osculum, becoming at the same time smaller and more inconspicuous, until they no longer exist close to the oscular margin. The cones l)ear on the apex moderately strong, needle-like prostalia lateralia, which project either singly or in tufts of a few together to various lengths, up to about 20 mm. They are directed on the whole radially, though frequently inclined obliquely one way or the other. When the prostalia are well preserved, as is especially the case in the smaller specimens, the resemblance ACANTHASCUS CACTUS. 145 of the sponge to a cactus is so remarkable that one is at once impressed with the appropriateness of F. E. Schulze's selection of the specific name. No doubt all the cones arise in connection with the prostalia, but it is by no means uncommon that we meet with such as show no trace of the latter, which evidently have been lost. Cones without prostalia are especially frequent in the larger and therefore older individuals ; in fact in these it is rare that we find any of the prostal needles at all. The loss is undoubtedly partly due to a mechanical breaking off at the apex of the cones due to some external cause, as is proven by the proximal part of the needles still remaining in the cones, the outer end being broken off. At the same time I am inclined to assume that a large part of the loss is to be accounted for by the fact that the spicules in their entirety are cast off by some natural physiological process of the sponge. Only in this way can be explained the entire absence of even the proximal remnant within the cones in so manv cases. Probablv the lono-er and therefore older prostals are the first to be thrown out, and in sujDport of this theory stands the fact that those still remaining on the larger specimens are not so long nor so strong as some to be seen on the smaller individuals. If I am right in the above assumptions, it follows that the period of the greatest develop- ment of the prostalia is when the sponge is comparatively small and young, or at any rate before its full size is attained — a fact which is not without parallel in other Acanthascinae {S. glaber, R. victor, li. capillatus). Near the oscular margin there may occur some fine prostalia which project either singly or in small tufts without having cones at bases. They are however never so numerous as to form anything like a fringe to the osculum. 140 ART. 7. — I. IJIMA : HEXACÏINELLTDA, IT. Figs. 16 and 17, PI. XI., show in natural size the appearance of the endosomal and the ectosomal layer respectively. A close familiarity with the characters of either will be sufficient to enable one to recognize the species even if the specimen to be determined 1)6 a small fragment, provided of course one or the other of the layers is well preserved. In the ectosome (PI. XIL, fig. 34) the extremely delicate dermal layer exhibits minute quadrate meshes, just discernible as such with tlie naked eye. JNIuch more distinct is the hypodermal latticework (see PI. XL, fig. 17), the meshes of which are triangular, trapezoidal or polygonal in shape and measure rarely over 2 ram., but far more frequently less than 1 mm., in the length of their sides. Towards the ajDcx of the cones the hypodermal beams converge more or less from all sides, making the meshes between them narrower Init more elongate than in other joarts. The beams may measure up to 130 /^ in breadth. They are therefore not very thick, but still are thick enough to be distinctly perceived with the unaided eye. The points of intersection often appear somewhat thickened in a node-like manner, owing to dense concentration there of the soft parts. The subdermal space in the intervals between the incurrent canalar apertures may be as wide as 1 mm. It is traversed by numerous strands, the pillars, con- necting the ectosome to the choanosome. The apertures of the incurrent canals are of various sizes. In the larger specimens, the largest canals may reach 4 mm. in diameter but are in general much smaller. The gastral surface is on the whole tolerably smooth, though it may sometimes exhibit a wrinkled appearance. The canalar apertures on this side of the wall are on the whole larger, but necessarily less in number, than those on the external side of the ACANTHASCÜS CACTUS. 147 same specimen {cfr. PI. XI., figs. 16 and 17), as is tlie rule with all Hexactinellids of similar shape and structure. All the ex- current canalar apertures, in fact the entire gastral surface, is covered over by a well diiferentiated endosome (PI. XI., fig. ig; PI. XII., fig. 35). In this the gastralia are however never so abundantly present as to form a continuous layer of latticework by themselves ; so that, the endosomal meshes, which present themselves as such to the naked eye, may be said to be freely open, except for some fine and insignificant trabeculic of the gastral membrane that may occasionally extend across them. The said meshes are of about the same size and shape as those of the hypodermal framework ; only they are generally more rounded at the corners, while the beams inclosing them are perceptibly thicker. These beams, so far as the skeletal parts are concerned, consist in the main of the strands of hypogastralia, along which there occur the gastralia proper in scattered distribution together with numerous hexasters. Certain hypogastral strands are some- what conspicuously thicker than the rest and can be traced continuously for a considerable distance, pursuing an irregular course and frequently intersecting others of similar strength. The sponge in the living state is colorless or whitish. The texture mav be said to be rather firm. Spiculation. As to the spiculation I have l)ut little of importance to add to what is alreadv known throii2;h the investis^ations of F. E. Schulze. To that gentleman I am indebted for the gift of a slide-preparation made from the type-specimen described in the 148 AKT. 7. 1. IJIMA : HEXACÏINELLIDA, IV. Challenger Keport, wliicli preparation was of great service to me in insuring a correct identification of specimens in the early stage of mv studies. Principal parenchymalia, elongate spindle-like, sometimes bow-like. Length, up to lomm.; thickness at middle, up to 300 ft. Thier pointed ends smooth, but sometimes rough as in all the smaller parenchymalia. There exists a gradational series of vari- ously sized parenchymalia, from the principalia down to comitalia of only 6/^ thickness. (Fig. 36, PL XII., is intended to illustrate the different modes of termination of the parenchymal diactins. Unfortunately the representation of the subterminal roughness in the thinner spicules has turned out to be a failure). The 2^^'ostalia are diactins of great length, which may measure up to 25 mm. or more. They are probably to be looked at simply in the light of excessivel}^ elongated parenchymal principalia. Like these, the part imbedded within the sponge- wall is accompanied by filamentous comitalia as well as by other loosely and radially arranged parenchymalia whose outer ends are directed towards the apex of the cones. The hypodennal and liypogadrcd strands consist of diactins, and of diactins only, which aie arranged in loose or compact bundles of variable strength (PI. XIL, figs. 34 and 35). The elements look much like parenchymalia of similar dimensions. They are gerierally not longer than 3 mm. nor broader than 38 !'■ in the middle. Towards both ends they gradually taper, the very tip being either simply acuminate or conically pointed. iSubterminally, the surface is roughened by microtubercles which ACANTHASCÜS CACTUS. 149 are however never numerously present. Usually the spicular center is externally provided with a gentle annular swelling ; more rarely, with two opposite protuberances. When the sponge is torn from the rock to which it was attached, the basidictyonal plate is left adhering to the latter. The plate can easily be separated out by boiling in acid. It is found to be in places quite thin, being represented by an uneven and small-meshed siliceous reticulum (the limiting layer of the basidictyonal plate, see p. 134), the beams of which may at intervals show^ spicular axial filaments in the form of a plane cross (as in PI. XXII., fig. 17). At other places there exists, besides the same limitin s; or attachment laver and in direct union with it, a somewhat wider-meshed sponge-work (PL XII., fig. 37) which increases the thickness of the basidictyonal plate to half a millimeter or more. The relalively thick beams of the sponge- work are rough-surfaced all over on account of the presence of numerous conical microtubercles. The manner of their arrange- ment is apparently irregular, but by exact observations it is not difficult to make out that the foundation to the structure is given by certain hexactins, the basidictyonalia, which are directly as well as synapticularly fused together. Isolated instances also occur of diactins or of parts of diactins — apparently belonging to the parenchymalia — being incorporated in the beams. Observed in glycerine, such a diactin is seen to be of smooth-contour and distinctly traversed by the axial canal but entirely enveloped by an irregular and synapticula-forming secondary deposit of siliceous matter. But by far the greater part of the parenchymalia in the region have their lower end simply loosely inserted into the interspaces of the basidictyonal framework. 150 ART. 7. — T. IJIMA I HËXACTINELLIDA, IV. Dermalia predominantly stauractins, rough all over, nearly plane or sliglitly arched so as to be convex on the outer side (PI. XII., fig. 23). Axial length, 200-350 /A Eays perceptibly tapering towards the rounded tip or nearly uniformly thick ; sometimes slightly swollen at the ends. Thickness at their middle, Sh y- on an average. The quadrate meshes formed by mutual apposition of the rays of separate derraalia measure on an average 165 !'■ in length of sides. Occasionally among the dermalia there occur peutactinic forms in which the proximally directed, unjwired ray is somewhat shorter than, or nearly equally as long as, tlie paratangentials. They scarcely ever occur on the cones. Under exceptional circumstances a stauractin may have one of the rays so shortened that it approaches a tauactin in shape. A few cases of unusually small, smooth and slender-rayed oxystauractins and oxypentactins that I have seen I hold to be dermalia in an in- complete state of development. Gastralia rough pentactins, in which the unpaired ray is directed distad (PL XII., fig. 24) ; occasionally stauractins. They are (juite like the dermaliii, though perhaps on the whole slightly smaller. As before mentioned they are never so numerous as to form a continnous gastral lacework, occurring, as they do, in widely scattered distribution on the hypogastral strands (see PI. XII., fig. 35). Oxyhexasters, 00-152// in diameter (120/^ on an average). Those occurring in abundance in the choanosome as well as in the endosome (PL XII., figs. 29-32) have mostly exceedingly short principals which in many cases may be said to be almost ACANTHASCUS CACTUS. 151 reduced to nothing. In them the terminals are comparatively strong, reaching up to nearly 5 /^ in thickness at base ; they are obsoletely rough all over. The roughness is especially pronounced at their base and here it may often be distinctly observed, under a very high power of the microscope, that it is caused by reverted microtubercles or minute barbs. Each principal may bear two, seldom three or four, divergent terminals, thus giving in all 12 or more than 12 terminal points to the entire oxyhexaster. More usuall}'' the oxyhexasters seem to be hemihexactinose and not seldom quite hexactinose (figs. 30 and 31). In the former case, the principals are biterminal if not uniterminal. Thus, forms with 11, 10, 9, 8, 7, or only 6 terminal points are not uncommon. Of the oxyhexasters situated in the subdermal space, many (not all) are considerably different in appearance from those described above and which are more deeply situated in the wall. In them the principals are appreciably longer and more slender, while the terminals are simply more slender ; moreover, there occur as a rule two or three terminals to each of the six principals. One specimen of such peripheral oxyhexasters is to be seen in PI. XL, fig. 18, on the right hand side. A com- parison of that oxyhexaster with the others shown in PI. XII. will at once make apparent the differences above indicated. In all the oxyhexaster varieties the central cross can easily be brought into view and the extent of its arms exactly deter- mined, if proper steps be taken in preparing. PL XII., fig. 33 represents a case of what seems to be an abnormally developed oxyhexaster. It was met with but once in the dermal membrane of a certain specimen and may possibly have been only an extrinsic object. Central node irregularly 152 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. shaped, but still with indications of the six principals. Total number of terminals, 20 at least ; some of them bifurcated at the end as shown in the figure. DisGoctasters (PI. XII., figs. 2ö and 27) various in size, measuring 106-260/'- in diameter. Secondary principals J — è as long as the free part of the terminals. Number of terminals in a tuft, 3-7 ; most usually 4 or 5. The discoctasters in the periphery of the wall are the smallest (106-137/^ dia.), while those in the deepest parts may be twice as large (generally 200- 260 /^ in diameter). In other respects than in size there also exist some noteworthy différences between the two. The smaller peripheral discoctasters (fig. 25) usually show distinct hump-like prominences on the central node, in the middle of the spaces surrounded by every set of four secondary principals. The four ridges radiating from the prominence to- w^ards and into each principal are also generally but not always distinct. The principals as seen in a lateral view are frequently uneven in contour ; a more exact observation may show that they are longitudinally ribbed throughoui the length, indicating their formation by a coalescence of parts running lengthwise. PI. XL, fig. 20, representing the central part of a discoctaster taken from the subdermal region, demonstrates the points mentioned above, besides showing the disposition of the axial cross in the central node and the series of small vacuole-like spaces contained in the ridges that go to form the secondary principals. The terminals are obsoletely rough-surfaced ; their disc at the end is minute, appearing simply like a pinhead. In the larger, deeply situated discoctasters (fig. 27), the central node may be so large as to measure 20 /f- across. On it ACANTHASCUS CACTUS. 153 the hump-like prominences are seldom seen, these being usually either merely indicated or not at all present. The space where they should occur may even be concave. It seems that the prominences, as also the ridges arising from them, disappear as the central node becomes larger with the growth of the entire spicule ; possibly they become, so to say, covered up by the siliceous deposit added to the node. The secondary principals, which may measure 7* /^ across in the middle, are obsoletely rough-surfaced. The terminals are similarly rough. Near their outer end the roughness grows somewhat coarser and is here seen to be caused by retroverted microtubercles (PL XIL, fig. 28). In direct proximity to the terminal disc, which marginally runs out into 7 or 8 small teeth, there exists a very short tract devoid of the microtubercles. Not rarely malformed discoctasters are met with, in w^hich some primary terminals have apparently failed to unite in the proper way with any of the secondary principals but remain more or less free, appearing like supernumerary appendages to an ordinary discoctaster. As a case of such malformation is to be considered the spicule figured by F. E. Schulze in the Challenger Eeport, PI. LVIL, fig. 4. Of special interest are the isolated cases of true discohexasters I have come across in certain specimens and which I consider to represent the primitive form whence the discoctaster was derived. In general size and in the appearance of the terminals they are exactly compara1)le to the smaller discoctaster of the periphery, but the essential difierence consists in the fact that the principals are six in number, these being short, thick and knob-like and arranged in the usual disposition. Four or five terminals arise divergingly from each principal. The discohexaster, 154 ART. 7. T. IJIMA : IIEXACTINELLTDA, IV. which 18 much too rare and inconstant to be counted amons; the regular hexasters of the species, may be said to be in appearance not unlike the macrodiscohexaster of Aulosaccus ijimai F. E. ScH.; it is apparantly in the state of nearest approach to the (liscoctaster, into which it may l)ecome converted by comparatively short steps of change. Micro(l(!^cohexasiers (PL XII., fig. 26) of only 1-3-25/-'- diameter and of the usual appearance are common in both the dermal and gastral membranes as well as in the subdermal and subgastral trabecul^e. In some individuals they are much more abundant than in others. In the choanosome they do not seem to occur. In stained pi'eparations a nucleus — in all probability representing a silicoblast — is almost always seen in each of the angles formed by the principals, as shown in PL XL, fig. 18. The interspaces between the exceedingly delicate terminals are filled up by a protoplasmic matrix, so that the entire rosette first attracts one's attention as a small, faintly colored mass of spherical shape. Soft Parts. 8ome fragments fixed and ])reserved by myself at different times at the place of capture have been utilized-for making the following observations on the soft parts. The chambers are so numerous and densely crowded together that it is difficult to exactly determine the shape and extent of each (see PL XL, fig. 22). However there can be no doubt of their being as usual cup-like or thimble-like in shape. The ACANTHASCUS CACTUS. 155 diameter varies from 90/^ to 190/^. The chamber wall I have seen as a reticular membrane witli 02:>en meshes (PL XI., fig. 19). The choanocyte nuclei, though scarcely differentially stained as they lie in the nodes of the reticulum-forming granular protoplasm, can at places be distinctly recognized as svich. They are of about the same size as the trabecular nuclei. None of the preparations are such as to allow anything to be said about the flagellum or the collar. On the outside of, and in close ap])Osition to, the chamber wall are variously sized groups of well-stained archœocyies reach- ing up to 3i !>■ in diameter. A small group of the same is seen in PL XL, fig. 19, on the left. The archaïocytes lie close together, generally arranged in a single layer Hat upon the chamber wall or between the walls of directly adjacent chambers. The trabecidce, both external and internal, are as usual cobweb-like. Their nuclei, measuring about 1 !'■ in diameter, are well-stained and contain generallv more than one chromatic granule. Along the lumen of the larger canals, both incurrent and excurrent, the trabecuUe are at places extensively exjmnded into a thin lining film, so that they might be spoken of as constituting here a canalar membrane. The dermal membrane (see PL XL, fig. 18, left) is film-like j^erforated by large and small "pores" of roundish or oval shape. The parts between closely adjoining "pores" may be reduced to mere threads indistinguishable from the trabecula3. Not only does the essential agreement in histological nature but also the fact that the same thesocytes and microdiscohexasters occur in both 156 ART. T. 1. IJIMA : HEXACTINELLIDA, IV. the dermal membrane and the trabeculse, strengthen me in the belief that the two parts just mentioned are genetically and fundamentally one and the same structure, whatever differences apparently exist between them being due to the circumstances of their respective situations (see Contril). I., pp. 122, 147). The gastral membrane is of much the same appearance as the dermal. The thesocytes of the species are of a characteristic appearance. They may occur in the dermal and gastral membranes as well as on trabecuhe in all parts of the sponge-wall, but seem to be most abundant in the subdermal region. They are show^n in numbers in PI. XL, fig. 18, in wdiich they appear as morula-like masses consisting of numerous small spherules. The quantities in which they occur are subject to variation according to individuals ; but they seem to be present all the year round, since I have observed them in specimens preserved in the months of April, July, August and December. The thesocytes are spherical, ovoid or somewhat irregular in shape and may be more or less flattened when situated on or against a membranously expanded trabecula. They measure 8- 20 /^ across. A fine cell-limit and the nucleus are only excep- tionally distinguishable ; the former can be perceived only when the spherules contained in it are not developed in too excessive numbers, while the latter is generally concealed among the spherules. The spherules, which measure 1.7-2.5/^ are refractive and homogeneous, appearing yellowish in the unstained state. Their reactions towards different stains have already been noticed on p. 178 (foot-note) of my Contribution I. and therefore need not be ACANTHASCüS CACTUS. 157 mentioned here. They are generally tolerably uniform in size within a tliesocyte. In certain cases, however, I have found them quite unequal in size, the larger ones appearing to have arisen from the combination of several smaller ones. Further in certain individuals I have not infrequently met with thesocytes in which the spherules were apparently in the process of disintegrating into irregular granules ; or in other cases, of dissolving into a diffuse state. The diiferences in appearance may be partially due to the drastic influence of the preserving reagents ; nevertheless I believe that they may in general be taken as representing the changes which the spherules, as a nutritive substance held in re- serve, undergo by a natural physiological process before they are consumed. And it would be but natural if we should find that some thesocytes are quite or nearly quite devoid of the fat-like contents as the result of consumption. As such thesocyte relics I consider certain pale-looking cells which are now and then found wherever thesocytes might be expected to occur. A few cells of the kind in question are to be seen in the left upper part of fig. 18, PI. XL They are of about the size of ordinary thesocytes filled up with spherules. Probably they are of a more or less collapsed shape. The faintly colored and finely granular cytoplasm incloses a distinct .nucleus, while its external limit is well-defined and is sometimes distinctly provided with an envelop- ing membrane. The cells can scarcely be viewed in the light of early thesocyte stages before the formation of the spherules ; for, there exist other cells which alone can be viewed as inceptional thesocytes, viz., those very much smaller than full-sized thesocytes and which already contain the spherules though yet in quite a limited number. I may here add that in some instances the fat-like spherules 158 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. presented the appearance of dispersing after having been set free from the thesocyte wliich originally contained them alL Qnite isolated spherules, evidently the same as those of thesocytes, are occasionally found sticking to tlie trabecuhe. The peculiar rosette-like bodies depicted in PI. XT., fig. 21, were met with in profusion in a prejDaration of a specimen taken in April. Reference to them has already been made on p. 31 of this Contribution, under Scyphidium longisjjina. The body consists of a varying number of well-stained spindle-like or rod- like pieces in radial arrangement. These measure up to 4 /^ in length. The figure was included in the plate under a suspicion that the bodies might represent stages in the spermatogenesis, but now I think they are something, extrinsic or otherwise, at any rate quite foreign to the sponge. ACANTHASCUS ALANI Ij. PI. X., figs. 16-23. Acanthascas alani. Ijima, '98, p. -vj. The species is based on a single specimen (PI. X., fig. 16) which originally belonged to Mr. Alan Ow^ston (O. C. 4097) but is now in the collection of the British Museum. It is a dried specimen greatly macerated on the exterior. In shape it is ovoid, goblet-like and slightly laterally compressed. The lower end contracts into a short, stalk-like and irregular ACANTHASCUS ALANT. 159 base. Total height, 190 mm. Greatest breadth of body, 133 mm.; lesser breadth in the same region, 104 mm. Stalk-like base, about 60 mm. in greatest breadth. Superiorly the body-wall closes in towards the thin oscular margin which is apparently simple-edged. The irregularly roundish osculum is about 60 mm. in diameter. The deep gastral cavit}'- extends into the stalk-like base. The body-wall in the lower part is as thick as 35 mm. Notwithstanding the lacerated condition of the external surface, it can be ascertained that this was by no means smooth but must have shown a number of irregular hillock -like or ridge- like elevations, judging from what appear to be their rests. The more prominent of the elevations may have been 10 mm. high, as measured from the bottoms of adjacent depressions. Possibly their summits were originally provided with diactinic prostalia as in other sj)ecies of the genus, but no such spicules have been found preserved in the specimen. Fortunately there has remained the ectosome in some places, though in small patches (PI. X., fig. 23). It is uneven, being- much creased. The minutely and more or less regularly quadrate meshed dermal lacework is exceedingly delicate ; it is supported l)elow by fine intersecting hypodermal strands, mostly under 0.1 mm. in thickness, which pursue a sinuous course. Over the interaperturai spaces, as also over the external prominence, the ectosome is generally in close contact with the choanosome, so that it can not be very clearly distinguished. Where it has fallen off, the sponge-surfîice appears somewhat crisp-hairy on account of the dishevelled fibers of the parenchymal strands. The stalk- like base presents the usual fibrous texture. The incurrent canalar apertures measure mostly under 5mm. 160 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. in diameter, though a few may be as large as 0 mm. Those of an approximately similar size lie separated from one another by a space nearly equally as wide as, or sometimes much wider than, their diameter. On the gastral side, the endosome is not developed in a continuous layer, so that the excurrent canalar apertures all open freely into the gastral cavity. The apertures are of various sizes, some being as wide as 15 mm.; their edge is either sharp and distinct or but slightly indicated. The interapertural space shows an irregular interlacing of fine fibers. All the principal spicules are very fine, which fact accounts for the soft and delicate texture of the sjoonge. The septa between the two systems of canals are thin ; consequently, the sponge is light, and cavernous in appearance. Its general aspect is not unlike Chaunoplectella cavernosa or Aulosaccus scJmlzel. The species is apparently more closely allied to A. ^j/a/e-z! than to A. cactus. From the former it is distinguishable by a number of points, of which I may mention the relatively wider canals, the freely open excurrent apertures, the smaller discoctaster and the larger oxyhexaster and microdiscohexaster. Spiculation. Parenchymal diactins small, slender ; generally less than 1 mm. in length and at most 12 r thick. In the stalk- like base, they may be over 1mm. long and 30/^ thick but can scarcely be said to be coarse. Center, usually without external swelling or bosses. Nearly uniformly thick throughout except at the ends which are more or less swollen and rough ; extreme tip rounded. Oxyhex- ACANTHASCUS ALANI. IGl actins met with here and there in preparations of the septa are either canalaria or gastralia. Hypodermalia quite like the above. DerriiaUa (Ph X., fig. 16), ahiiost always pentactins ; nsnally with a gentle prominence representing the distal sixth ray. Rays, slightly attenuated outwards but with rounded tip ; microtubercles, not numerous, nor strongly developed, so that the roughness of surface is not pronounced. Axial length, generally 250-350 ,«. Breadth of rays at base, 9-11 !'-. Gastralia (PL X., fig. 18), hexactins ; irregularly scattered. Rays, like those of dermalia but with less roughness of surface. Axial length, 280-460 /^. Hexactins similar to the gastralia occur along the excurrent canals as canalaria. In the incurrent canals pentactins resembling dermalia were occasionally found ; they are probably to be considered likewise as canalaria. Oxyhexaster, present in abundance. Diameter, 144-190,«; on an average 160/^. From the characters of its rays, two varieties can be distinguished ; both seem to occur together promiscuously. In the one variety (PI. X., fig. 20) the center is swollen to a globular node, while the short principals are rounded in a knob-like manner. Slender rough terminals, generally 3 or 4, arise from each principal. They seem to be very liable to be broken off near the base, the fragments being found in abundance in the soft parts. It seems that this is the more abundant of the two oxyhexaster varieties. In the other variety (PI. X., fig. 21) the terminals are considerably stronger while the principals are much less distinctly 162 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. indicated, being in fact quite abortive. Number of terminals to each principal, generally 2 ; seldom 3 and sometimes only 1 to some of the principals. Insignificant microtubercles are sometimes seen on the surface of terminals, at the base of which they are reverted barb-like. Discoctaster (PL X., fig. 19), common in all parts ; varying in size. Diameter, 136-220/'-. Central node, always with the six boss-like prominences. Secondary principal, about as long as, or longer than, the terminal tuft. The latter consists of G-8, fine terminals (drawn too thick in the figure) ; narrow at base and expanding more or less distally. Terminal disc, small and pinhead-like. Microdiscohexaster (PI. X., fig. 22), very sparsely present, having been found in canalar septa only in a few instances after a careful search ; spherical, with 30-35 !'■ diameter. Central node, swollen to spherical shape. Principals moderately long, with numerous fine terminals. STAUROCALYPTUS It. Ijima, '97, p. 53. Hypodermalia include pentactins in which the paratangentials are smooth or rough, but never armed with spines. Generally veiled, sometimes not. STAUEOCALYPTÜS. 16^ The genus, as I coDsider it at present constituted, comprises tlie eleven species embodied in the following- Differential Key to Species. 'f.— Dermalia nearly exclusively, or at any rate predominantly, pentactins. a'. — Discoctasters (both t^iibdermal and subgastral) of less than 200 h- dia.; rarely up to 213 i>. dia. a^. — Thin-walled. Gastral surface with a continuous endo^onml layer covering over the exclurent canalar apertures. (AVithout prostalia. Discoctasters 80- 100 [J- dia.; with characteristically short and cylindrical tufts of terminals) ,S'. fasciculatus F. E. Sch. (Coast of California, 690mm.). 6'. — Tolerably thick-walled. Gastral surface without a continuous endosomal layer, the excurrent canalar apertures opening freely. a^. — "Without prostalia (?). "With peculiar pit-like subdermal cavities whence arise narrow iucurrent canals. Excurrent canalar apertures wide (up to S mm. dia.). (Discoctasters 128-180 jj- dia.) .S'. rœperl (F. E. Sen). (Coast of Patagonia, 7.31m.). b^. — AVith botli pentactinic and diactinic prostalia. Subdermal cavity in- conspicuous. All canals narrow, even extremely narrow ; not more than 2 mm. wide, a*. — Gastral surface smooth. Among the gastralia pentactins more numerous than hexactins. Rays of dermalia and gastralia smooth basally. (Discoctasters 150-200 [i dia.) S. solidus F. E. Sch... (Coast of California, 48G-1254m.). b*. — Gastral surface hairy on account of short projecting ends of fine diactins. Gastralia hexactins. Eays of dermalia and gastralia entirely rough. A considerable number of stauractins among the dermalia. (Discoctasters 130-213 |j. dia.) S. lubuloms n. sp. (Sagami Sea). /,i._Pnbgastral discoctasters larger than 220 [J-, reaching up to nearly 300 [j- or over in diameter ; subdermal discoctasters considerably smaller. (Tolerably thick-walled ; with pentactinic and 'generally also diactinic prostalia ; canals comparatively narrow). c'^.— Smallest discohe:j;asters not smaller than 200 \i- dia. (m fact 228-320 \j. dia.). Gastral surface smooth. Prostal pentactins small; their paratangentials up to 2.2 mm. in lengtli. Principal parenchymal oxydiactins under 8 mm. in length and 41 \i. in thickness. Gastral hexactins 80-100 [x in length of raj^s ,S'. doidlncjl (Lambe). (Pacißc coast of IST. America, 63-512 m.). f/-.— Smallest di^:cohexasters (subdermally situated) much smaller than 200 H- dia. Gastral surface with numerous projecting diactins. Prostal pentactins moderately large; their paratangentials reaching up to 4mm. or considerably more in length. Principal parenchymal oxydiactins much coarser than in c" (may be 80-600 iJ. thick). Gastral hexactins on the whole as large or decidedly larger than in c*. 164 AllT. 7. 1. IJIMA : HEXACTINELLIDA, IV. c^.- — Gastral surface hairy on account of short projecting ends of fine diactins. Excurrent canalar apertures mostly freely open ; in some places covered over by an irregularly coarse-meshed endosomal layer. Oxyhexasters largely hexactinose S. affinis n. sp. (Sagami Sea). cP. — Gastral surface conspicuously hispid on account of projecting and rather coarse diactins. All excurrent canalar apertures covered over l)y an irregularly coarse-meshed endosomal layer. Oxyhexasters, rarely hexactinose. »S'. entacanthus n. sp. (Saganii Sea) 6. — Dermalia nearly exclusively, or at any rate predominantly, stauractins. c'. — With small and inconspicuous prostal pentactins. (Dermalia uniformly micro- tubercled on all sides). Gatralia hexactins, forming a continuous lacework in the endosome. Discoctasters 114-128 ij- dia S. microchetus Ij. (Sagami Sea). (iK — Without prostal pentactins, the dermal surface being smooth, (though in e"^ isolated and slender prostal diactins may sometimes occur). c^. — Dermalia decidedly spiny on the external side but obsoletely microtubercled on the inside of rays. Gastralia hexactins in which the free proximal rays are over 450 |j. in length; forming a continuous gastral lacework over excurrent canalar apertures. Discoctasters very large, 500-660 a in diameter S. glaber Ij. (Sagami Sea). /-. — Dermalia slightly rough, (jastralia pentactins and stauractins, with rays up to 100 [J. length ; not forming a continuous layer over excurrent canalar apertures. Discoctasters 110-210 ;j. diameter & heteradinus Ij. (Sagami Sea) f.— Dermalia straight diactins. (Gastralia represented likewise by diactins. Disccctaster.-* 140-200 (A dia. Sponge veiled and with long diactinic prostalia S. pleorhaphidcs Ij. (Sagami Sea). Ill the present Contribution I propose to treat in full nine species wliicli I have studied. S. fasciculaim and S. solid us are two American species of which my knowledge has been solely derived from the describer's work (F. E. ScH., '99). I beg here to offer a few remarks with regard to both. /S. fasciculalus is undoubtedly a well differentiated species. Its peculiarly characterized discoctaster (only 80-100/^ large and with 4-6 terminals, about 12/^ long, forming a cylindrical tuft not broader than tiie principal of over 20,« length) should alone be sufficient to distinguish the species from all the rest in the genus. STAUKOCALYPTUS. 165 "With respect to S. soUd^is it seems that F. E. Schulze (/. c, p. 52) regards it to be distinct from the closely similar S. dowlingi mainly on the strength of two peculiarities, viz., that the general shape is invariably barrel-like, straight and broad- based instead of being cup-like, outbulged on one side and narrowed towards the base ; and that the discoctasters are much smaller and more slender-rayed. It occurs to me that perhaps some other points might well be added to the specific difference, to which view I am led from certain facts contained in F. E. Schulze's own description and figures. On p. 105 (/. c), in a short diagnosis of IS. solidus, is a statement to the effect that as gastralia there occur a greater number of pentactins than hex- actins ; whereas, in iS. dowlingi the same spicules are known to be hexactins and only occasionally pentactins. Again, on his Plate X. (/. c), F. E. Schulze gives figures of the dermalia and gastralia from aS'. solidus; both the spicules mentioned are shown to have rays which are basally quite smooth, instead of being rough all over as in S. dowlingi. The differential characters above indicated, if shown to be constant, would certainly be of no small importance in distinguishing the two sj)ecies. Concerning the three new species descril)ed for the first time in this Contribution (y/,r,, S. iubulosus, S. ajfinis and S. entacan- thiis), it may here be mentioned that they are all very nearly related to S. doivlmgi and that some of the specimens on which they are based were at first referred by me to that species, though with some hesitation. The characters of the specimens referred to were therefore taken into account in drawing up the diagnosis of S. dowlingi in my " Revision of Hexactinellids with discoctasters " (Ijima, '97, p. 53). However, a renewed examina- 166 AET. 7. 1. IJIMA : HEXACTINELLIDA, IV. tion of the Jajoanese materials and a comparison with typical S. dowli7igi, the characters of which have since become more precisely known to me, have led me to think that I was mistaken in the identification. S, dowlingl is, for the present at least, to be eliminated from the list of JajDanese Rossellids ; on the other hand, I think it fairly justifiable, under the circumstances, to add to the list the three new species mentioned above. But I. can not * Table sliowing some points in tbe characters of those Siaurocalypius species (or speci- Name of species. Size of speci- mens described. Dernuil side. Gastral side. JJiactinic pnjstalia. Pentactinio prtistalia. Incurrent canalar apertures. Projecting needles. Excurrent canalar apertures. .. 140- 240 [i.. 130- 152 p.. Very nu- merous. 120- 200 a. 280- 400 IJ.. 20 IX. 25 nun. 520 [J.. 40 ram. 180 -J.. 5-6 mm. 100- 170 [J.. 140- 175 [X. 115- 160 .X. Very nu- merous. 160- 183 IX. 350- 380 IX. 19 IJ.. 9 mm. 250 !J.. 2-4 mm. 130- 200 iJi. 95- 130 IJ.. 100- 132 -x. Not pre- sent. 155- 220 !x. 220- 286 tj.. Not found. 1.5 mm. 80 p.- 4-6 mm. 130- 200 iJ.. 140- 250 p.. 120- 186 (X. Xumer- ous. 143- 176 IX. 262-352, rarely 428 IJ.. 19 !J.. 168 ART. 7. 1. IJIMA : HEXACTINELLTDA, IV. a table showing for each species or specimen certain noteworthy points from its characters). STAUROCALYPTUS RŒPERI (F. E. Sch.). PL XIV., figs. 26-32. Rhabdocalyptm Rœperi. F. E. Scpiulze, '86, p. 51. '87 (!), p. 1Ö8; PI. LXV. '97, p. 553. Slaurocalyptus rœperi, I. Ijima, '97, p. 55. '98, p. 53, The following account of this species is here introduced mainly to record certain points which go in a measure to sup- plement the excellent description we have already received from F. E. Schulze, the original describer, and which seem to be of importance for a sharp distinction between this and certain other species closely resembling it. To the liberality of the investigator jnst mentioned I am indebted for a gift of two slide-preparations made from the type- specimen, and on which my own observations have been solely made- The species was obtained by the " Challenger " to the south of Puerta Buono in Patagonia (depth, 731 m.). It should exhibit a medium-sized and moderately thick-walled, vase-like or cup-like body, firmly attached to the substratum by a short stalk-like base, No prostalia of any kind were found, possibly as the result of abrasion. Through the even ectosome were seen elongated, angular or spindle-shaped and pit-like^subdermal spaces, whence arose rather Z2 arrow incurrent canals. On the gastral side the STArROCALYPTUS RŒPERT. 169 large (up to 8 mm. dia.l, sharply contoured, excurrent apertures were freely open, i.e., not covered by endosome. The above seem to constitute tlie more important macroscopic features of tlie species. As to the .^piculalioii, all the skeletal elements are remarkable for their slenderness. The strongest ixirenchymal diactins seem not to exceed 35 !'■ m lireadth. The ^''ai'enehymalia in general have sparsely micro- tuberculated ends, the very tips being pointed. The smaller or the more slender of them show kno1)S cruciately disposed around the spicular center. The oxypentactinic hypodermalia have slender paratangentials under 2 mm. in length and Vo f- in thickness at base; these are smooth except at the rough outer ends and are not quite straight, but rather wavy. — Besides the oxypentactins, strands of diactins also serve to support the ectosome. These hypodermal diactins are of various lengths ; they are all cruciately tubercled in the middle. While those of çrreater leno-th are indistinsfuishaljle in appearance from a parenchymal diactiu, the shorter ones may equal the dermalia in axial length and may have rays exactly similarly characterized as in the latter ; so that, such short diact- ins, especially when in isolated positions, might as well be classed under the dermalia as under the hypodermalia. The statement therefore seems justifiable that the dermalia pass over l)y gradation to the parenchymalia through the intermediation of hypodermal diactins. The derma/la (PL XIV., figs. 30 and 31) are predominantly pentactins, not infrequently stauractins and rarely monactins, not 170 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. to mention the occasional diactins above referred to. In all the different forms it is quite usual that the suppressed ray or rays are indicated by vestigial knobs. Such rays as are well developed are 100-165 !>■ long and (> /< or less broad at base ; the micro- tubercles on their surface are sparse and weakly developed ; their tips nre conically pointed. Not only are the rays considerably more slender tlian in the corresponding spicules of all other closely I'elated species of the genns, but also the roughness of surface is far less prononnced, as has already been pointed out by F. E. Schulze. The (jadralia (PI. XIV,, fig. 32) are oxyhexactins with similarly characterized rays, which measure 100-220 1'- in length and not over 5 /< in breadth at base. Over the interspace between the freely open excurrent canalar apertures, the gastralia are present in abundance, forming a nearly regularly quadrate-meshed latticework. In this, the laths are formed frequently by two or more gastral paratangentials combined in a loose strand. The oxyhexasters (PI. XIV., figs. 27, 28) are small and very slender-rayed. Diameter, 88-130 /<. Those situated in the periphery of the wall are on the whole somewhat smaller than others in deeper situations. The short, often exceedingly short, principals bear each 2-3, nearly straight or slightly wavy and obsoletely rough terminals. Less frequently there is only a single terminal to one or more principals in an otherwise normal oxy- hexaster (hemihexactinose). In the preparations at my disposal I have not discovered any regularly hexactinose oxyhexasters, but I presume that the spicules spoken of by F. E. Schulze as " small, weakly developed oxyhexacts " (Call. Rep., PI. LXV., STAUnOCALYPÏUS KŒPERI. 171 fig. 6) and wliieli should occur isolated arc nothing else than such. Their rays are said to be wavy and to have a rough surface (F. E. Sch., '97). It frequently happens that in biterminal rays of the oxy- hexasters a third terminal is represented by a minute spine which seems not to have attracted F. E. Schulze's attention. In PL XIV., fig. 28, I have figured a case in which every principal bears a spurious terminal besides two well developed ones. Special examinations were made to convince myself of the fact that the minute spine was not an axial elongation of the principal to which it belonged. In one instance I observed a principal bear- ing two terminals which were quite normally disposed except that one of them was very much shorter and more rudimentary-looking than the other. There can be no doubt whatever that the cases in question represent stages preliminary to the reduction of the number of the terminals. The dm-ocladers (PI. XIV., fig. 26) measure 128-180/^ in diameter. Those situated subdermally seem to show no appreciable difierence in any respect from others lying near the gastral surface. The slender secondary principal, not over 4 ,« broad, is nearly as long, in some cases only about h as long, as the free terminals belong- ing to it. The terminals number 2-5, usually 3 or 4, in a tuft which broadens Ijut slightly in the distal direction. They are nearly straight ; the surface is obsoletely rough when seen under a very high power of the microsco])e. The minute terminal disc is always shaped like a pinhcad, and not marginally toothed. The delicate microdiscohexasters (PL XIV., fig. 29) are found in tolerable abundance in the gastral region. IJiameter, 22-24 fj. 17.2. ART. 7.— I. IJIMA: IIEXACTINELLIDA, IV. (32 (i, according to F. E. Sch.). In shape tliey are spherical, the six tufts of terminals in the same spicule being not widely separated from one another. The terniinals are not very nuuiei'ous; they are exceedingly fine and about twice as long as the principal. The slimness of the rays in all the spicules, the sparsely microtuberculate character of the dermalia and gastralia as well as the small size of the discoctaster seem to make up the more conspicuously distinctive features in the spiculation of this species. STAUROCALYPTUS DOWLINGI (L. l\. Lamüe). R]iab(locaIyptu6 dotdhuji. Lamjje, '93 (!), p. o7 ; PL III., figs. 2, 2a-2h. — F. E. Schulze, '97, p. 554. Siauroccdypias (JowUnfji. F. E. Scjiulze, '99 (!), p. 47 ; PI. IX., ligs. 1-6. — {Kol IjiMA '97, p. 5o ; nor Ijima '98, p. 53). The following account of this species is given more for the purpose of furnishing basis with which to compare certain Japanese forms, than to give the results of my own observations on the sample taken from the Canadian type-specimen and which was kindly sent to me at my request by ISlv. L. M. Lambe of Ottawa. As mentioned before (pp. 165-167) I now consider, contrary to my earlier assumption (Ij. '97), that this species is not represented in the Japanese waters so far as is known at present. The localities then to be assigned to the species are : Strait of Georgia (Canada ; al)Out 60 m.), near St. Ivosa Island (California; 221m.). and near Lenard Rock (Aleutian Islands ; 512 m.). STAUROCALYPÏUS DOWLINGI. 173 According tu the descriptions of L. M. Lambe and F. E. Schulze, the species should show a tolerably thick-walled broadly sacciform body, somewhat outbulged on one side and Jiarruwed at base. It mav attain a lariie size, as attested l)v a fraoment of the wall 300 mm. long and lö mm. thick, obtained by the "Albatross " on the Californian coast. From all over the dermal surface there stand out both diactinic and pentactinic jirostalia. The gastral sur- face is smooth but may show in places the ends of some parenchymal oxydiactins projecting beyond it. The subdermal space is evidently of a quite insignificant width. The apertures of both the in- curj'cnt and excurrent canals seem to be small as a rule. Lambe gives 11 mm. as an average diameter of the larger and il mm. as that of the smaller incurrent apertures of a specimen 100 mm. high. The same specimen showed on the internal surface evenly distributed excurrent apertures of about ^ mm. diameter. A continuous endosomal layer is apparently not present over the excurrent apertures, which thus seem to open free into the gastral cavity. The single imperfect specimen, — 220 mm. high and 120 mm. broad, — obtained by the " Albatross " near the Aleutian Islands and referred to the present species by F. E. Schulze, requires s])ecial mention in so far as it has, in contrast to the specimens from Canada and California, very wide canals and subdermal spaces, on which account the upper part of the wall is said to have presented an a])pearance almost of a lamellar structure. It seems to me that this structural deviation apparently extends somewhat bevond the urdinarv ranoe of variabilitv to be exiDCcted I « O t J. ill different individuals of a Kossellid species. With respect to spiculatioa the more impurtant points, knuwn 174 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. to me from the descriptions of Lambe and Schulze as well as froai my own observations, may be summed up as follows : rarenchyntalia, oxydiactins under 8 mm. length and 41 /^ breadth. The small diactins with 2 or 4 knobs at the middle and of 0.28-1.37 mm. length, mentioned by Lambe as occurring in the dermal and gastral layers and seen also by Schulze, are apparently either those that enter into the composition of hypoder- mal beams together with the paratangentials of oxypentactinic hypodermalia or what might be called the hypogastralia. Prodal oxydlaciliis, as long as GUmm. with a thickness of 300 /A Hypodermal oxxjpcnlact'mn, mostly with orthotroi)al (/.), somewhat as in Lanuginella pupa and Scyphidium longùpina (this Contrih., PI. L, fig. 7 and PL IL, fig. o). They stand nearly vertically or slightly inclined outwards, giving a fine and densely hirsute, rather than " granular," appearance to the surface beset by them. This appearance is acquired by the paratangentials before the pentactins are protruded as prostalia, apparently as the last step in the development of hypodermal (/. e., prospectively prostal) pentactins. This is clearly indicated by those groups of hypodermal pentactins (Contrib. III., PI. VI,, fig. 12) in which the most superficially situated pentactin-head consists of rough paratangentials while another or others following in deeper levels are smooth and successively more slender rayed. A small numl)er of diactins associate with the paratangentials of hypodermal pentactins in forming the support to the dermal lacework. The dermalia (Contrib. III., PI. VL, figs. 14 and 15) are predominantly rough pentactins but intermixed with a goodly number of stauractins. The former frequently, but not always, show a knob-like relic of the atrophied sixth and distal ray. Karely and exceptionally I have encountered hexactins among STAUROCALYPTUS TUßÜLOSÜS. 179 the derinalia. In all these the rays taper slightly towards the rounded or obtusely pointed ends. Their length as measured from the spicular center, 84-190//.; on the average about 14.0 n. Breadth of rays near base, 7-12 ^. I have noticed that the stauractins attain on the average a considerably larger size than the pentactins (see Contrib. III., PL YI., fig. 12). (For the sake of comparison I may mention that in /V. doivlingi the dermaliii are known to be <»;enerallv pentactins and occasionally diactins. Length of rays, lGO-180/^; according to Lambe, 160/^- on the average). The (/aslralla are all hexactins with rays similar in appearance to those of the dermalia. Length of I'ays, 130-200/^ (in >S. dowUng'i, 80-100//). Oxyhexaders (Contrib. III., PI. VI., fig. 17) occur in abun- dance. They are mostly normally developed ; less frequently are they hemihexactinose. Quite hexactinose forms are very rare. The central node is often swollen to a spherical shape. The principals are as usual short — often exceedingly short. The dia- meter is in general 75-115;^ (in S. dowlirtgl 100-120 /i). The oxyhexasters in the subdermal region mostly measure less than 100 y- in diameter. In them the terminals are slender, slightly rough and 2 or o in number to each principal. Those situated more deeply in the wall are on the whole somewhat larger, many also having perceptibly stronger terminals, which are then nearly smooth-surfaced. From this species again I am in a position to record a case of an oxyhexaster in which five of the principals bore each a set of two, nearly equally developed terminals, but the sixth principal l80 ART. 7. — ï. IJIMA : HEiACÎlNELLIUA, IV. rail out into a single normal terminal besicles showing near its base the short and spurious rudiment of a second terminal. By one short step of reduction, the oxyhexaster might change itself into a hemihexastinose form with one uniterminal princijml. The discodasters may in general be said to have a diameter of 130-213/^; the principal takes up nearly one half or some- what less of the entire ray-length. Terminals straight, nearly smooth, 2-5 in a tuft which broadens gradually and slightly outwards. The sinaller discoctasters show six tuljercles on the central node, l)ut the larger ones do not. Those in the subdermal region are smaller than others in the subgastral ; the former measuring 130-170/^, and the latter 17'j-21o/i, in diameter. The specimen spicule shown in Contrib. III., PI. VI., fig. IG, is from the subdermal space. (In -S'. dowl'myi the discuctaster diameter is known to be 228-320 /', manifestly never falling below 200 n). Spherical iiiicrodiscoJiexnders (jf 1Ü//. diameter and of the usual appearance are found very sparsely distributed in tlie endosomal layer. I consider that they recpiire no S2)ecial de- scription. STAUROCALYPTUS AFFINIS nov. sp. PI. XIII. and PL XIV., hgs. 14, 10, 17, 22-25. Staarocalypius Vowlituji in part. Ijima, '97, p. 53 ; 'gS, p. 53. This new species is based on two specimens in the Science College Museum, which were at first — errorneously, I now^ think SÎAUKOCALYl^TUS AFFINIS. 181 — taken by me for S. dowUngi. This would suggest its close resemblance to that species and also to S. tuhulosus, a fact which can not be gainsaid. The two specimens differ in certain points of outward appearance but show an essential or almost complete agreement in spiculation. Such differences as exist between them are manifestly referable to difference in individual age and partly also to the different state of their preservation ; so that, I have scarcely a doubt as to their specific identity. I propose to call them S. affinis. Each of the two specimens serves greatly to supplement the knowledge to be derived from the other. They will be separately described as regards their macroscopic characters. The one I take up first (S. C. M. No. 194, from which all the figures in PI. XIII. are taken), was purchased of a Misaki fisherman in 1891. The locality as put down on the label is Okinose ; depth not stated. The specimen consists of large and small dried fragments which must originally have formed the wall of a tubular or vase-like form, assumably about 300 mm. in heiglit and not less than lOOmm. in diameter. The maximum thickness of the wall is about 9 mm. Owing to imperfect desiccation the texture is soft and loose, the tissues easily falling off in crundjs and shreds. The external surface is much abraded. No lateral prostalia, which I presume Avere once present, are preserved. However, the delicate ectosome still remains here and there in small patches. It adheres closely to the choanosomal surface and consists of the usual dermal lacework supported by thin and irregularly intei'seet- ing hypodermal strands (see PI. XIII., fig. 10). The dernial meshes, of approximately quadrate shape, are on an average about 165/^ wide. 182 AßT. 7. — I. IJIMA : HEXACTIXELLIÛA, IV. The external clioanosomal surface (PL XIII., fig. 1) is tolerably even and presents a coarsely fibrous appearance. It exhibits variously sized, roundish, incurrent canalar apertures, tlie laro-est of which mav measure 02 mm. in diameter. The larger apertures are separated from one another by an interspace nearly as wide as themselves or even considerably wider. They lead into deep pit-like canals. The gastral surlace (PI. XIII. , fig. 2) is on the whole ex- cellently preserved. On it there open luunerous, closely set, roundish and sharp-edged excurrent apertures measuring not more than 2 mm. in diameter. Most of them open freely and directly into the gastral cavity without a covering endosome, similarly as in several species of the genus (aS'. rœperi, dowlingi, tubulosus, solidus). In many cases the apertures are provided with an iris- like rim. It is, however, a remarkable fact that a covering endo- somal latticework is not entirely undeveloped. Such a structure is iu fact to be fouiid here and there in small irregular areas, some of which are to be seen iti PI. XIII., fig. 2. It consists of irregularly reticulate beams, mainly composed of diactinic hypo- gastralia and inclosing small angulate meshes. These meshes are open, not filled in by a continuous gastral lacework ; they can easily be distinguished from the underlying canalar a};ertures proper l)y their position, small size and irregular shape. The above endo- somal layer, so far as it is distinctly difiërentiated, greatly resembles in appearance that of Acanthascus cactus (PI. XL, fig. 10). Xo diactins are found that project their ends through and beyond the gastral surface except such occasional cases as may be regarded as due to unnatural displacement. The loose feltwork of the parenchyma contains a quantity STAIIROCALYFTLTS AFFINIS. ISo of coarse strands, running more or less parallel to the surfaces but otherwise in all directions (PL XTIL, fig. 1). Especially coarse is its appearance on the external side of the wall. Picked up ])j means of a pincette the strands easily come off in strips often as long as 40 or üO mm. At the torn edge of the wall the fibers bristle out in a conspicuous mannei-. The second specimen (S. C. ]\r. No. 400, shown in PI. XIV., fig. 14) was obtained by Ktma in the Uraga Channel (Nago Hill-e-])aibusa Spit and about 4 kilom. off the latter). It resembles a tube which laterally is slightly and irregularly compressed. It is 103 mm. long, 24-27 mm. l)road in the middle and considerably broadened at the lower end. At one corner of the latter there is a small stalk-like base for attachment. At the opposite corner the wall is outbulged and thinned out at the top, apparently in the first stage of the breaking through of a secondary osculum. It is to be noticed that both this outbuloins; and the attachment base lie in the sagittal plane of the body, as is the rule with so many other Acanthascinic with similar parts. The upper truncated end of the body is taken up by an oval-shaped osculum. The wnll is about G mm. thick in the inferior part, gradually 1)ecom- ins; thinner towards the oscular eds-e. From all over the external surface, except near the basal stalk, there arise both oxydiactinic and oxypentactinic prostalia lateralia. The prostat diactins are not ver}' luunerous ; they are long needles that spring out isolatedly to a length of 30 mm. or less, directed obliquely outwards and upwards. In the upper part of the body they are finer and much shorter than those situated lower down. Along the oscular edge they project straight upwards, the exposed portion not exceeding 4 mm. in length. — The large prostat 184 ART. 7. — T. TJIMA : HEXACTINELLIDA, IV. pentactins, either isolated or in loose tufts, arise from the top of small papilla-like prominences of the dermal surface, situated at intervals of 1-2? mm. The pentactins, wherever they have escaped the influence of abrasion, are so numerous and crowded together that they form a gauzy layer of 3 or 4 mm. thickness over the dermal surface. The latter is tolerably smooth except for the above-mentioned papillae. Close beneath the thin ectosomal layer are seen the roundish apertures of tlie incurrent canals, which, in the lower part of tlie sponge, may measure up to Ih mm. in diameter. The smallness of the canals in comparison with those of the first specimen seems to be sufficiently accounted for by the much smaller size of the individual. The characters of the gastral surface are essentially the same as in the first specimen. Namely, while most excurrent canalar apertures are freely open, certain others are covered over liy an ii-regularly meshed endosomal latticework. A number of small oxydiactinic parenchymalia are seen to project their fine ends beyond the gastral surface to the extent of about 1 mm. or more. Taken all in all, the general appearance of the species is closely similar to that of both ä dowli7igi and S. tubulosus, but especially of the latter. With our present knowledge it would be unsafe to pick out this or that macroscopic character in the above description of the specimens as in any degree distinguish- ing the species from either of the two just mentioned. It may further be stated that the spiculation, so far as the kinds of spicules entering into it are concerned, is to a great extent, if not essentially, the same in all the three species (to which I may add S. entacantliKS n. sp. as a fourth). Nevertheless, there exist STAUROCALYPTUS AFFINIS. 185 certain points of difference shown by several of their spicules, that seem to warrant the separation of S. affinis as a distinct species. Spiculation. The following description applies to the first specimen (S. C. M. Xo. 194 ; PI. XIII.), unless the other specimen (S. G. M. No. 400 ; PI. XIV., fig. 14) is explicitly referred to. The principal ijarenchymaUa are oxydiactins of unusually large dimensions, often 35mm. or more in length and GOO/-« in thickness in the middle. (A portion of one is seen in PI. XIII., fig. 12). They occur in great abundance, especially in the periphery of the choanosome. They are entirely smooth, tapering gradually towards both finely attenuated and sharply pointed ends. They are slightly wavy or else are bent either in a bow-like or in a boomerang-like manner. Diactins of gradationally intermedi- ate shape and dimensions lead over the principalia to filamentous comitalia of only 12 /^ or less in thickness. In the second and smaller specimen the principalia attain a length of 25 mm. and a breadth of 520/^. — It may be remarked that in no other Staurocalyptus species are the parench^^mal elements known to develop such length and coarseness. — The needle-like prostal oxydiactins which were found only in the second specimen may be 40 mm. long and 180 ,« thick. The diactins which join in the formation of hypoäermal strands are mostly short, having a length of 1-2 mm. and a breadth of 14-33 /i at the middle, where they may show a slight annular swelling. The ends are rough, somewhat tapering and obtusely or conically pointed, — not swollen as in the much longer 18G ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. parenchymal coraitaliii. The diactins run sometimes singly but more usually a few are combined into thin bundles. The Jiypogastral diactins, likewise running singly or in loose strands (see PI. XIII , fig. 11), are similar to the hypodermal. The oxypentacilnic hypodermalia (to be seen in PI. XIII., fig. 12) are large spicules with smooth, nearly straight or slightly bent rays. The paratangentials are in most cases more or less distinctly paratropal, but sometimes regularly cruciate. TJieir length meas- ures 7-12 mm. and their thickness at base, about 100/^. The unpaired proximal ray is longer ; it is accompanied l)y numerous thin comitalia (about 9/^- thick) of the usual character. In the second specimen I find the hypodermal (and also the prostal) oxypentactins somewhat smaller, the paratangentials meas- uring mostly between 5 and 6 mm. in length. Nevertheless it may be said that the spicule iu question is on the whole con- siderably larger iu the present species than in either S. dowlinyi or S. tubulosus. Decidedly larger does it seem to be tlian in the former species, in which the paratangentials are said not to exceed 2.2 mm. in length. — The paratangentials of those pentactins already protruded as prostalia remain smooth in most cases. Only occasional- ly have I met with such prostal pentactins as have slightly rough or granular paratangentials. The roughness is here due to the pre- sence of rounded and iuconspicuous microtubercles somewhat like those in S. doidingi but unlike the same in S. tubulosus. The dermalia (PI. XIII., fig. 3) are nearly exclusively pentactins ; rarely and exceptionally, stauractins. The paratan- gentials are neaily straight or very slightly arched. All the rays are rough throughout ; they generally taper perceptibly outwards STAUROCALYPTUS AFFINIS. 187 to terminate -with rounded tips. Their leugth as measured from the spicular center is 130-200//; on an average, 160/^; breadth close to base, 11-15 fi. In the second specimen the length was 100-170 /i or on an average 130 /<; the breadth, 11-10 p.. — The dermal paratangentials form a tolerably regularly (juadrate-meshed dermal lacework (PI. XIII., fig. 10). The gastralia (PI. XIII. , fig. 4) are hexactins with similar but often longer rays than those of the derraalia. Length of rays, 140-240//.; in the smaller specimen, 140-175//. The free proximal ray is not distinguished from the others by a greater length. The gastralia are present in far too sparse a number to form a continuous lacework by themselves (PI. XIII., fig. 11). They are found in irregular distribution in the spaces between the excurrent canalar apertures as wx41 as on the endosomal reticular beams, where these are developed. The oxyhexasters (PL XIII., figs. 7 and 9 ; PI. XIV., figs. 16 and 17) measure 130-152// in diameter (in the second specimen, 115-160//). They are exceedingly numerous both in the choano- some and the endosomal layer. While in the former the hexact- inose and the hemihexactinose forms predominate, in the latter region the normally developed form is by far the most abundant. The normal oxyhexasters, occurring in the endosome or in its proximity (PI. XIII., fig. 9; PI. XIV,, fig. 16), show two or sometimes three, generally smooth, nearly straight and widely divergent terminals to an excessively short principal. The central node is often spherically swollen, but not sharply dermarcated from the principals. In the subdermal space, oxyhexasters are altogether sparse and those that do occur there are as a rule again normally 188 AMT. ".— I. IJliMA: IIEXACTINELLIDA, IV. sha]3ecl. They are on the whole slightly smaller and more slender- rayed than in those in or near the endosome ; besides, the terminals are always more or less distinctly rough (PL XIV., fig. 17). The roughness is in many cases plainly attributable to minute and inward- ly directed barbs, especially distinct in the basal parts of terminals. Hemihexactinose forms of the oxyhexaster most commonly show one or two uniterminal principals, the rest of the principals being each in possession of two terminals as is usually the case. The terminals are nearly straight and more or less rough-surfaced. Hexactinose oxyhexasters (PL XIII. , fig. 7) have likewise rough rays which are however frequently not quite straight but somewhat wavy. They are quite numerous in the choanosome, a fact which may perhaps be regarded as constituting one of the peculiarities of the species. Occasionally I have met with interesting exceptional forms of the oxyhexaster, in which the hexactinose character is still further modified into the pentactinose and even the stauractinose by a complete suj^pression of the terminals on one or more of the principals. The principals, thus deprived of their terminals, remain in their position as smoothly rounded off prominences which leave no room to doubt that the terminals have not been lost by mechanical breakage Ijut were undeveloped from the first. The above, evidently abnormal, forms were not noticed in the second specimen. In this, on the other hand, I have not infre- quently met with the small rudimentary-looking spicules shown in PL XIV., fig. 22. These contain a small axial cross of filaments and there can be no doubt about their being oxyhexasters with terminals but little developed, though it is difficult to decide whether we have to do with early developmental stages or with abnormal appearances. SÏAUROCALYPTUS AFFINIS. 189 The discodaêters occur in tlie sabdermal space as well as near the gastral surface. They are not very abundant, though in places several are found together. Right in the choauosome they must be very rare, if indeed they are ever present. Two sizes are distinguishable according to their position. The smaller disc- octaster form (PL XIII., lig. ö), measuring 120-200/^ (in the second specimen, 160-183 ij) in diameter, occurs only in the subdermal space ; while the larger (PL XIII., fig. 6) of 280- 400/^ (in the second specimen, 350-380 /j) is confined to the proximity of the gastral surface. The range of variability in the size of discoctasters is here decidedly wider than in either -'S'. dowlingi or S. tubulosus. As to general shape the discoctasters in general closely agree with those of several other nearly allied Staurocalyptus species and also with those of Acanihascus cactus. The principals take up i-'s of the entire ray-length, are unevenly contoured, thicken somewhat in the outward direction and carry a slightly expanding tuft of 3-7, straight or nearly straight terminals. The central node, especially that of the smaller discoctasters, frequently shows the six well-known tubercles. In the larger form, the distal portion of the terminals is beset with minute but distinct barbs (fig. G) and the terminal disc exhibits a number of marginal teeth exactly as I have figured them from Acanihascus cactus (PL XI r., fig. 28). In the smaller form, the terminals are simply rough and the terminal discs pinhead-like. The discoctaster as well as the oxyhexaster, — of the latter, both the normal and the hexactinose forms, — I have subjected to special examinations in order to determine the position and extent of the central cross of axial filaments in the central node. The results are embodied in PL XIV., figs. 23-20, which will speak 190 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. sufficiently for themselves, so that comments upon them may here be dispensed with. Microdiscohexaders (Fl. XIII., fig. 8) of spherical shape and 19-20 /^- diameter are not uncommon in or near the endosomal layer. They were occasionally observed in the ectosome also. From a spherical node arise comparatively thick principals, which in length are about I the radius of the rosette and each of which carries at the outer end a small disc, usually provided with a central tubercular prominence on the external side. The terminals are exceeding fine and difficult to count, but probably not more than 10 to each principal. Finally, the basidictyonal plate, found preserved in the second specimen, is rather thick and consists of an irregular reticulum of beams of variable thickness (up to 30 /^), inclosing mesh-like spaces up to 100 n in widtli. The beams are beset with small prickles as w^ell as with larger conical protuberances. Here and there can be observed hexradiate axial crosses in the nodal points. Also some rough and thick-rayed hexactins are seen in the first process of being amalgamated with the basidictyonal beams. PL XY., fig. 12, taken from S. glaher, equally well illustrates the condition of the basidictyonalia in the present species. The chief })oints in the spiculation of the present species, requiring special attention in relation to such other species as are most likely to be confounded with it, seem to be : the great length and coarseness of the principal parenchymalia ; the large size of prostal pentactins, in which the paratangentials are in STAUEOCALYPTÜS ENTACANTHUS. 191 most cases smooth but occasionally slightly granular ; the compara- tively large size attained by gastral hexactins ; the great abun- dance of hexactinose oxyhexasters ; and the wide range (120 /<- 400 ijt) of variation in the size of discoctasters as well as their large maximum size. (For a comparative talkie of the main distinctive features of this species and of certain others, see the foot-note on pp. 166-167). STAUROCALYPTUS ENTACANTHUS noy. sp. PL XIV., figs. 15, 18. Staui'ocalyptus Doiclingi in part. Ijima, '97, p. 00 ; '98, p. 53. The specimen which I now make the type of this new species is another that was at first placed by me under S. dowlingi. Its characters were therefore taken into account in drawing up the diagnosis of that species given by me in '97. The said specimen (PL XIY., fig. 15. S. C. M. No. 242) comes from the north of Onigase and from a depth of between 429 and 572 meters. It is a large piece of wall torn from the superior end of an individual, which when entire must have had a sac-like or vase-like shape with a diameter of nearly a foot. It exhibits a part of the natural oscular edge, which is sharply angular but not thin. The marginal zone, is divided into a number of broad triangular flaps, folded outwards and backwards and even partially rolled up, so that in places the oscular edge is in touch wdth the dermal surface of the lateral wall. Between 192 ART. 7. 1. IJIMA : HEXACTINELLTDA, IV. each two of the flaps there remains in the oscular edge an unreflected point ; consequently there arises at each such position an ear-like projection, which is all the more prominent since the edge of reflection of the flaps on its either side is depressed so as to present a concavity directed upwards. There are three such ears in the fragment on hand, measuring 105-145 mm. from tip to tip of two adjacent ears. In the entire state of the individual there must have existed at least seven ears and as many reflected flaps around the superior opening of the gastral cavity. The w^all is 13 mm. tliick at a point ahout 100 mm. below the tip of the ears and must liave been thicker towards the sponge-base. The state of preservation of the specimen which has been dried, is as good as could be desired. The dermal surface is tolerably smooth, distinctly showing the delicate, regularly quadrate- meshed dermal lacework and the irregularly meshed hypodermal network. On close observation, the thin fillers composing the latter are seen to be so arranged as to present numerous radial figures, the centers of which are 2i-ß mm. apart from one another. Pentactinic prostalia of medium size are found only on parts of the external surface tliat are protected from abrading influences by the folding of the marginal flaps. They are irregularly dis- tributed, usually solitary and project to the length of at most 3 mm. Except a few isolated and quite insignificant needle-like spicules springing out on the sharp oscular edge, no diactinic prostalia are present. Presumably, however, the species in a young stage of growth is provided with a greater quantity of both pentactinic and diactinic prostalia. The larger apertures of incurrent canals, vaguely visible STAUROCALYPTÜS ENTACANTHUS. 193 through the dermal layer, may be 3 mm. wide, and that in a position not more than 100 mm. distant from the uppermost end of the specimen. In the interspaces between the apertures the ectosomal layer is in tolerably close contact with the choanosomal surface. The gastral surfoce, so far as this extends in the specimen, is entirely lined with a well-developed endosomal latticework, which is for the most part very distinctly set apart from the underlying choanosome and covers up all the excurrent canalar apertures (up to 3 mm. dia.), in a measure concealing these from view. Its appearance is not unlike that of the same structure in aS'. affinis, in which species, however, it is developed in l^ut small irregular patches and not all over the gastral surface. The endosomal beams, the main suj^port of which is afforded by strands of hypogastral diactins, are of a moderate strength and inclose small, irregularly shaped meshes (mostly i-f mm. wide), which are open and not filled in by a continuous lacework of gastralia, unlike those in S. fasciculatus but quite like the same of Acanthascvs cactus (see PI. XL, fig. 16). Tliis endosomal latticework, proba1;)ly developed over the entire gastral surface and leaving none of the excurrent canalar apertures quite freely open, may be regarded as one of the distinctive features of the present species as contrasted with many closely related forms in the same genus. Another noteworthy, and I think characteristic, point in the features of the gastral surface is the fact tliat numerous, moderately strong oxydiaetins project freely ])eyond the surface, mostly to a length of 4 or 5 mm. There is no I'ule as to the direction of the projecting needles, which occur over the entire gastral surface so far as this is represented in tlie fragment, but 194 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. somewhat more abundantly in the region of tlie reflected marginal flaps than lower down. They give to the gastral surface a peculiarly hispid or spiny appearance, an idea of which may be obtained from the appearance presented ])y the upper border of fig. 15, PI. XIV. Similarly disposed needles are known to me to occur on the gastral surface of S. tubulosus and jS. affinis ; but in both these species their trichodal ends project to so short an extent that their eflect upon the surface is to render it at most delicately hairy. S. dowllngi also seems to be not altogether devoid of needles protruding from the gastral surface ; but here, F. E. Schulze ('gg, p. 49) was led to regard them as something of occasional and rather accidental occurrence ; while L. M. Lambe ('gs), in his description of that species, says that the gastral surface is smooth, and makes no mention of projecting needles. Now, in the present species, the needles in question are spicules, coarse rather than fine and of a considerable length, and moreover so abundant that they can not fail at once to attract one's attention ; and in view of the well-preserved state of the tissues, I see no reason for not assuming that we have here to do with a normal character, probably constant to the species. Here I may interpolate an account of a specimen which I provisionally refer to S. entaeantJms ])ut with a query. AVhile agreeing as regards many characters with the above-described type- specimen, it shows several points of difference, especially in regard to the spiculation, — differences, whicli, on further study with more material, may possil)ly turn (nit to be of more than individual value. The specimen in (|uestion {^'. 0. M. No. 403) was obtained at Inside Okinose by the Iwado-lino from a depth of 618 m. STAUKOCALYPTUS ENTACANTHUS. 195 Two views of it Iiave already been published iii my Coutributiou III., PL VI., figs. 9 and 10. Appended to tliem will be found the name of S. japou'waa, an appellation which I beg herewith to withdraw. The specimen is the upper portion of a thick-walled and probably tubular or cylindrical individual. It is about 62 mm. long by 42 mm. diameter ; the wall is 9 mm. thick in the thickest part. While the gastral cavity is nearly cylindrical throughout, the external surface curves in at the top forming a rather thick oscular edge. The osculum is irregularly roundish with a diameter of 22-24 mm. I presume that pentactinic prostalia in tufts, were originally present but have been lost by breakage, for there exist stumps of what a])pear to be their shafts arising in association with small papilla-like prominences of the dermal surface, which is on the whole tolerably smooth. The said prominences occur on the average at intervals of about 3 mm. Of diactinic prostalia some remnants seem to be represented by a few, fine and isolatedly projecting needles. The apertures of the larger incurrent canals reach up to nearly 2 mm. in diameter ; those of an approximately similar size are separated from one another by an interspace usually wider than their own diameter. Over this interspace the thin and delicate ectosomal layer is closely adherent to the choanosome. The endosomal layer, closely similar to that of the type- specimen {i.e., consisting of a small and irregularly meshed lattice- work, in which the meshes are not filled in by a dermal laceW'Ork but remain open), is well differentiated over the entire gastral surface. Accordingly, no excurrent canals can be said to ojien freely and directly into the gastral cavity. Not only in this respect does there exist an agreement wdtli the type-specimen 196 ART. 7. 1. IJIMA ; HEXACTINELLIDA, IV. but also ill the strikingly and deusoly hispid character of the gastral surface (see Coutrib., III., PI. VI., fig. IJ). This is caused by nuineious [)rojectiiig ueedles of a moderate strength, occuring from the oscular edge downwards over the entire gastral surface of the specimen. They are firmly implanted in the sponge-wall, their inner ends protruding 5 mm. or more, sometimes vertically but more usually inclined in indefinite directions. It is entirely beyond question that the spiny character of the internal surface is not accidental but natural. The spicules of this specimen will be specially dealt with after 1 shall have described those of the type-specimen in the following paragraph. Spiculation. First, that of the tyi)e specimen (PI. XIV.,, fig. 15). The principal parenchymalia are oxydiactins which may attain a length of 9 mm. and a breadth of 250 /-« in the middle. They are relatively thick, nearly straight but sometimes bow-like or boomerang-like spicules ; entirely smooth and not very much attenuated towards both sharply pointed ends. Small and finer parenchymalia are of the usual characters. The needles projecting from the gastral side are oxydiactins similar in appearance to the principal parenchymalia. They may be 9 mm. long and 90 /^- thick. A varying number of fine coinital diactins are found accompanying them. Hypodermal oxypentactim are mostly found seveial together in loose groups, in which the smaller and therefore younger ones are always more deeply situated than those of older formation. STAUEOCALYPTUS ENTACANTHÜS. 197 When fully grown, the paratangentiaLs, wliich are either paratropal or regularly cruciate, measure 4 mm. in length and 65/^- in thickness at l)ase. They may acc^uire a rough surface, due to rounded microtubercles, while still remaining in the hypodernial situation. But this apparently does not take place in all cases, for among the oxypentactins already })rotruded as prostalia I have found some with rough and others with perfectly smooth para- tangentials. Diactins also enter into the composition of the hypodermal strands though not in large numbers. These hypodermal diactius are somewdiat variable in size and characters. While the larger of them are indistinguishable from ordinary parenchymalia, others are small, tubercled around the center and rough all over instead of being so at the ends only. The latter kind might easily be taken for derinalia, were it not for their association with unmis- takeable hypodermal elements. The Jiypogastralia are diactins similar to those of the hypo- dermal strands. The dermalia and gaslralia may be said to be essentially like those of S. doivUngiy tubulosus and affinis. On the pentactinic dermalia, which are by far the predomi- nant form, I have frequently observed a low prominence in the place of the aborted distal ray. Length of rays, 180-200/'. Earely stauractins and still more rarely diactins are met with among the dermalia. With respect to the hcxactinic gastralia, it may l)e mentioned that these are on the whole somewhat smaller than the same spicules in either S. tubulosus or affinis, agreeing closely in this respect with S. dowliugi. Length of rays, üö-loO /^. 198 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. The oxyhexasters measure 100-1 o 2 /v. in diaiueter, tliiiy com- ing next in size to the same rosette in ^S. dowlingi and ti/balosus. A goodly number oF them are hemihexactinose, but I have found none in quite hexactinose form. This negative character may perhaps be regarded as one of the j^eculiarities of the species, or at any rate of the particular specimen now being described. The terminals are rather strong (often 4 fJ- thick at base), — stronger than I have found them to be in other closely similar species of the genus. Moreover, they are always rough, the roughness being frequently develo[)ed into short barbs on the basal parts. Of the discociasters, those in the subdermal space measure 155-220/^ in diameter ; those in or near the endosome, 220-286/^. It may be pointed out that the lower limit (155 //) in this range of variation falls considerably below that in S. doivlingi (228 /i) but coincides in an approximate way with the same in S. iubulosus and ajjinis. In all of the discoctasters the principal is somewhat shorter than, or sometimes nearly as long as the terminal tuft belonging to it. The general appearance of the spicule agrees well with that of the same in S. ajjinis. In no case, however, have I noticed the six prominences on the sides of the central node. Jlicrodiscohexaslers I have failed to discover anywhere in the specimen ; but, in view of the ease with which they might be overlooked, I am not fully prepared to assert their total obsence. Now let me describe some points in the spiculation of the smaller specimen (No. 40; > ; Contrib., IT!., PI. VI., ligs. 9 and STAUEOCALYPTUS ENTACANTHUS. 199 10), provisionally referred by nie with some degree of hesitation to S. entacmülms. Here the principal parenchymal oxydiactins attain a maximnm size of 15 mm. length by 80/^ breadth. Thus they may be longer and decidedl}^ more slender than in the type-specimen. Another appreciable point of difference consists in the more gradual manner of tapering towards both ends which are slender and more or less rousfh-surfaced. The oxydiactins protruding from the gastral surface may be 20 mm. lono; and 250/^ broad in the middle. Thev are much longer and thicker than the corresponding spicules in the type- specimen. This fjict seems noteworthy especially when the smaller size of the specimen under question is considered. Among the derraalia I have found, though exceedingly rarely, regularly developed hexactinic forms. In many of the dermal pentactins the distal i-ay is represented l)y a mere knob. The different kinds of hexasters are quite similar to those of S. affinis. Oxyhexasters are normal, hemihexactinose or hexactinose. The last mentioned form is numerously represented, and constitutes another notable point of difference from the type- specimen. Diameter, generally 120-186/^; the hexactinose form may be larger, sometimes attaining 242 /^ in axial length. Peripheral discoctasters, 143-176/^; those deeply situated, 262- 352 !^, rarely up to 428 n in diameter. The upper limit in the size of discoctasters seems to extend very considerably over and beyond that determined for the type specimen. Microdiscohexasters of spherical shape and 11)// diameter are occasionally found in the endosome. 200 ART. 7. 1. IJIiMA : HEXACTINELLTDA, IV. Soft Parts. Care was taken by Kuma, the collector, to preserve in strong spirit small pieces from both the specimens described above ; so that I have been able to make some studies on their soft parts. This however has yielded no special additions to the knowledge we already possessed. I therefore restrict myself to recording only a few points. The thiml)le-like and sometimes cup-like chambers were observable with especial distinctness in the specimen (No. 403) refei-rcd tentatively to S. eniacanthns. Here they have a dinmeter of 110-187/^ (on an average, 155//). The usual reticular structure of the chamber-wall is in places very beautifully shown (PI. XIV., figs. 20 and 21). The meshes, the open nature of which can not be doubted, measure about 5/^- in length of sides. Observed under the immersion system, every nodal point is seen to be occupied by a vesicular, not specially well-stained nucleus, 1.7-3.4/^ in diameter and containing some chromatic granules. The smaller nuclei present a more compact appearance than those of a larger size. In optic sections of the wall, the nuclei have a distinctly oval outline. In both specimens the trabecuhe are frequently spread out in the form of a membrane especially on the surfaces which come in contact with the external world, — not only on the dermal and irastral sides but also alon«; the lumen of the larwr canals. The *D O Ö finely granular substance of trabecuhe may exhibit in the film-like parts somewhat deepei'-stained and often irregularly liranched streaks, which in some places may run more or less parallel to the edge of the film-like plate (PI. XIV., figs. 18 and 19). The first thouo-ht on observinir them would likelv be that we have STAUROCALYPTUS ENT ACANTHUS. 201 here to do with a fibrillar structure. But by close observntions I have come to the belief that the streaks do not represent fibrils actually developed as such but are merely wrinkles or irregularities in the thickness of the fihn. One strong reason in support of this view is the fact that the streaks are frequently quite ill-defined as to their contour and may, at one place or another, pass over gradually and insensibly into the general substance of the film. They are iu part probably the effect of shrinkage or contraction, such as might be caused by the action of reagents or by protoplasmic activity in the living state ; the rest are to all appearance simply the terminations of those filamentous trabeculœ which have just reached the film -like portion in order to join it. In certain places it seemed to me that the spaces between the trabecular nuclei {2-'2h /^ large) were unusually wide. The thesocytes are of a somewhat different appearance in the two specimens. In the type specimen of S. entacanthus, they are conglomerate- like in that the contents consist of a group of variously sized spherules (PI. XIV., fig. 18). These are usually deeply stained but not always to the same intensity. Their substance is nearly homogeneous or shows a faint granulation. In many cases the contents appear to be in the process of breaking up and becoming resorbed. The nucleus evidently lies concealed among the spherules. In the other specimen {S. entacanthus f) the thesocytes have moderately well-stained protoplasma-like contents which are rather coarsely granular (PI. XIY., fig. 19), instead of being composed of spherules in conglomeration. The nucleus is distinctly visible ÎIS a dark spot. The cells ai'o roundish or oval in shape with a diameter of 20 /^ or under. A delicate enveloping membrane can often be distinctly made out. I have found the cells in abundance 202 AßT. 7. T. IJIMA : HEXACTINELLTDA, lY. on the dermal and subdermal trabecule as well as on those aronnd the Inmen of the larger incurrent canals, bnt none in any part of the trabecular system of the excurrent side. STAUROCALYPTUS MICROCHETUS ^ It. PI VIII., figs. 17-20. iSiavrocalypius microchetvs. I.tima, '98, p. 58. The single type-specimen of this species (S. C, M. No. 450) is from Outside Okinose l)y the Iwado-line (about 572 m. depth). It is attached to a piece of rust-colored, easily breakable tuff. The body represents a rather thin-walled, laterally compressed tube, gently and slight!}^ outbulged between the two ends (see PL VIII., fig. 17). It is attached at the lower end by a tubercular base and also by an accessory base a short distance above the first. Total height of the specimen, 95 mm. Breadth at the middle, 37 mm. l)y 23 mm. Thickness of wall in that region, 3 mm. The simple-edged, wavy oscular i im is directed straight upwards. A veil covers the entire external surfoce. It is by no means conspicuous, because the pentactinic spicules composing it are very small and project only about 1 mm. or less beyond the smooth dermal surface. Here and there a few, short and very fine needles are seen standijig out from the extei'n;d sui-face in the manner of diactinic prostalia, which probably they really are. * iJiv-poi:, and ö/etO!: canal. STAUROCALYPTÜS MICROCllETUS. 203 The wall is tolenibly firiii on account of the closely interwoven state of the parenchymal npicules as well as of the small size of the canals. These, both incurreiit and excurient, are never so large as to measure 1 nnn. across. Their ai)ertures lie close together and are indistinctly visible through the tissues (ectosomal or endosomal as the case may be) immediately above them. The gastral surface, like the dermal, appears smooth and uniformly compact, ^ome parenchymal needles are seen to project into the gastral cavity but not in considerable numbers. I may say that the essential microscopic characters of the sponge are quite unlike those of any other /Skcurocalypius known to me, so that they alone should in my opinion suffice to prevent the species being confounded with any other in the genus. Spiculation. Some jiarenchymal diactliis, evidently the principalia, are comparatively very lai-ge. They may be as long as 24 mm. with a thickness of 143 /^- in the middle. Such large diactins are invariably without a central swelling; they gradually taper towards both ends wdiich are nearly smooth or only slightly rough. The smaller parenchymalia and the comitalia are of the usual charac- ters. The hypodermal pcnlactuis (seen in IM. VIII., fig. 24) are isolated, not in groups. In relation to this fact the four paratan- gentials always are regularly cruciately, and not paratropally, disposed. They are nearly straight or only «lightly bent, meas- uring up to 11mm. in length and 32// in breadth at base. The 204 AKT. 7. J. IJIMA : HEXACTJNELLIDA, IV. unpaired [)r()xiüial ray may l)e twice a« long as the paraiaugeiitials ill the same .spicule. In most cases all the rays ure smooth except for ;i s[)arse number oi' obsolete niicrotul)ercles near their conically ])ointed or rouiiJed ti[)s. Howevei', some of the spicules — without doubt those that are old and ready to be protruded as })rostalia — show paratangentials which are entirely rough, this being due to a thick covering of minute processes similar to those I have observed on the corresponding spicules of Scyphidium longlspina (PL II., fig. o). On the other hand, among those pentactins already protruded fi'om the dermal surface as prostalia, I find some with paratangentials roughened in the manner just mentioned but others exhibit no trace whatever of such roughness. Thus, it seems that this roughness may possibly never appear on some of the pentactins, though under certain circumstances it develops while the spicules are still hypodermally situated. The derinalia (PL VIII. , fig. 18) are generally stauractins ; exceptionally pentactins. The former are nearly flat or just perceptibly arched. The latter have the unpaired ray directed ])roximad. The rays are strongly prickly all over, gradually tapering from the base towards the pointed end. Length of rays, 70-100// (on an average 8Ö/'-). Their average thickness at base, 0 /'. The arrangement of the spicules in their relation to one another is rather irregular (see PL VIII., fig. 24), though here and there an approacli towards forming (j[uadrate meshes is observable. The fjadralla (PL VI IL, fig. 19) ai-e hexaetins with rays characterized similarly as in the dermalia. All the rays in the spicule are generally about equally long, though sometimes the STAUROCALYPTLS MICROCHETüS. 205 free proximal ray is found to be soinewliat longer tlian any other of the six. Length of rays, 120-143/-«; average tliiekness at base, 10 p-. The S})icules form a eontinuous lacework whieli is for the most part regularly (juadrate-meshed (PI. VIII., fig. 20). It lies close over the parenchymalia, from which scarcely any diactins can be distinguished as hypogastralia. The oxyhexasters may be distinguished as of two kinds, though these seem to iutergrade into each other by transitional forms. They may be said to differ in respect both of size and general appearance. One kind (PI. VIII., fig. 21) is met with, not abundantly but in moderate frequency, only in the proximity of the ectosome. It is comparatively small and of a delicate appearance. Diameter, 68-100 /-«. From each short and slender principal there arise o- o, thin and widely divergent terminals. The other kind (PI. VIIL, fig. 20) occurs in the choanosome as well as in the endosome and is far commoner. Besides l)eing larger, the terminals are somewhat stronger and the piincipals usually so exceedingly short that they may be called abortive. Diameter, 106-1 o6 /^-. The central node is frequently swollen to a globular shape. The number of terminals to a principal is usually 2, but may rarely be 3 or sometimes only 1. Hemihex- actinose forms are of occasional occurrence. Hexactinose forms were not found ; if thi-y occur at all, they must be very rare. The terminals are nearly smooth or obsolctely rough. The (lisrodasters (PI. VIIL, iig. 23) are very abundant subgastrally and a}iparently also in the endosome itself. I have not found them in any other position. Diameter, 11 1-128 />«. 206 ART. 7. 1. IJIMA : HEXACTINELLDIA, IV. The central node \a provided with the six tabercles which are moderately [)roiiniieiit, I'hc necondary principal measures in length between h and i that of the entire ray. 'J'lie number of terminals to each principal may be |)ut down at 5-10, but is more usually 8-10. The terminals appear to be rather strong ; at any rate they cannot exactly be called fine. The tuft they form is about 10 fji broad at base and may expand to thrice that breadth at the outer end. Each terminal gently bends slightly outwards in its course towards the thinly attenuated en(], which is furnished with a pinhead-like terminal disc. The microdiscoJicxaslcrs (PI. VII I., iig. 22) seem to be very sparse ; in fact they were met with only in a few isolated cases, in what })art of the wall I can not definitely state'. Diameter 23-2(3 /^. The terminals are exceedingly fine and not very numer- ous. They give a spherical shape to the entire spicule. The })rincipals may be simply knob-like as shown in the figure ; but this is not always the case ; for, in some specimens of the spicule I have seen each shoi't principal furni«lied with the usual ter- minals-bearing discs at their outer ends. Finally, the hciiiidiGlijomd i)hite. This I have isolated in small fragments from the attachment surface. In these I have made out that the single spicules, which by amalgamation go to constitute the plate, are mainly stout stauractins and occasionally pentactins. The rays in these spicules may be as thick as lo /^, their surface showing microtubercles in moderate numbers. The fact that these spicules are not hexactins as is the rule with dic- tyobasalia in general, probabl}^ has some connexion with the thinness of the plate in the present case. STAUROCALYPTUS GL AB ER. 207 STAUROCALYPTUS GLABER Ij. PI. XV. Sfriurocali/pli/s (jhtha-. Tjima, '97, p. ö7. This species seems to be not altogether uncommon in tlie Sagami Sea. More than a dozen specimens, representing varions stages of growth, have been examined by me. The exact localities they came from are : jMaye-no-Yodomi, in depth between 501 and 572 m. ( = 274 and 317 fms.) ; Inside Okinose by the Sengenzuka- line, about 500m. ( = 274 fms.); Outside Okinose by the Iwado-line, about 480m. ( = 262 fms.); Homba, between 501 and 572m. ( = 274 and 317 fms.) ; etc. In some specimens the sample of the bottom attached to the base is of a tufaccous nature, but the majority are attached to dead remains of other Hexactinellids, such as PeripJiragella elisa\ Clionelasma calyx, Hexactinella lorica, Farrea sp. and Hijalonema sp. On that most remarkable Chonelasma calyx which I have mentioned on p. 25 and figured on p. 31 of my Contiib. I., a small specimen of the present species was found in association with Chaunoplectella cavernosa and Ilhahdocalyptus capillatus. To mention some of tlie representative specimens in particular. The largest specimen I have seen was of a laterally compressed, vase-like shape, measnring 250 mm. in height, OG mm. by 95 mm. broad; at about tlie middk^ and 17 mm. in thickness of wall in the same part. The specimen was in a l)ad state of preservation. In Pi. XV., figs. 1 and 2, are shown in half natural size 208 AET. T.— I. IJIMA : HEXACTINELLIDA, IV. two exquisite specimens in the possession of the Sei. Coll. Museum. The individual of fig. 1 (Mus. No. 244) represents a thick- walled vase, slightly laterally compressed, rounded above and narrowed towards the base by which it is attached to the side of a dead Periphragella elisœ. Entire height, 109 mm.; breadth in the broadest part, 47 mm. by 64 mm.; thickness of wall in the same part, 14 mm. Somewhat on one side of the upper end is situated the roundish osculum, 27 mm. in diameter. A sparse number of inconspicuous prostal diactins, protruding to a length of 14 mm, or less, is present, especially in the region adjoining the oscular margin. That of fig. 2 is pouch-like and distinctly laterally compressed. Height, 130 mm. Breadth near the middle, 80 mm. in one direc- tion and 54 mm. in another. Thickness of wall at about the middle, 14 mm. The osculum is of an oblong shape, being elongate in the sagittal line. The basal attachment is on one side of the inferior end. The sponge is therefore bent in the basal region, the bending being as usual in the sagittal plane of the laterally compressed body. Nearly opposite to the basal attachment there exist two small secondary oscula, situated close to each other. No prostalia of whatever sort are to be seen on the specimen. Small, and therefore evidently young, specimens of the species are characterized by the possession of fine, but conspicuous, prostal needles. The smallest specimen I have had was of a nearly globular shape, measuring oiîly G mm. in lieight and with a roundish osculum of 2 mm. in dia. It showed a small numlier of fine ])rostal needles, apparently in th(' first stage of protruding out- wards. Much laro-er are the two vouns individuals shown in PI. XV., STAIIROCALYPTUS GL AB ER. 209 fig. 3. They are both of an ovoid shape, growing side by side on a dead Clionelasma calyx. One is 38 mm. and the other 32 ram. hio'h. The hitter shows an oscnhnn of G mm. dia. Fine and long prostal needles, springing ont isolatedly and directed ontwards and upwards, are present in moderate numbers. They occur more especially in the upper half of the body. Some of them may be 14 mm. long. From the constant absence of pentactinic prostalia and from the peculiarly cavernous appearance of the wall, I find it easy to distinguish the young of this species from those of certain other discoctasterophorous species, such as Rhab- docalyptus victor (PI. XIX., fig. 17) and B. capillatus (PI. XXII., fig. 3-5). One more specimen (text-fig. 7) requires special mention as being highly remarkable in more tlian one respect. The sponge- body (64 mm. long and 27 mm. by 37 mm. broad in the middle) presents the appearance of run- ning out basally into a root- tu ft consisting of a number of strong needle-like spicules, instead of being firmly fixed to a hard substratum as in all other specimens of the species. F. E. Schulze (Chall. Kep., p. 41) conceived the idea that with difiei'ing conditions of the sea-bottom, it was possible for one and the same species to produce Text-fig. 7. a root-tuft in one case and not in another. This s. giaher growing on assumptlou of F. E. Schulze's I at first thought basal spicules of Hya- lonema sp. (S. M. No. had becu actually realized in the case under Ma'; isofaToat^de Consideration ; but this proved to be simply an Okinosé). illusion, for by microscopic observations it could clearly be demonstrated that the tufted spicules at the basal end of the specimen do not belong to that sponge and are nothing 210 ART. 7. T. TJIMA : IIEXACTINELLTDA, IV. else than old and much lacerated, anchoring needles of a Hyalo- nema species, — presumably, of //. affine. The ends of the needles are invariably broken ; some of them show the characteristic spines on the surface ; and their surface, so for as it lies imbedded in the sponge, is covered with a thin siliceous reticulum, the same as the so-called basal-plate which is known to develop on the attachment surfece of so many Hexactiuellids firmly fixed to the substratum. It is then beyond dispute that the S. glaber in question had attached itself to, and was growing on, a tuft of Hyalonema needles. Another very interesting fact in relation to the specimen under consideration is tlie presence in large numbers of certain, peculiar, small bodies lodged among the tissues of the choanosome. To these bodies I have already had occasion to refer in Contrib. I. (p. 186, foot-note) of this series of Studies. To the naked eye they appear as whitish spots of various sizes under 0.6 mm. diameter. Except in the ectosome and the endosome, they occur throughout the sponge in tolerably uniform distribution. Under the microscope the body is found to be a reticular mass of no definite shape ; it consists of an irregular rigid framework of microtuberculate siliceous beams (PL XV., fig. 13). The mass is always traversed quite through by a few parenchymal diactins of the sponge containing it. It is further easy to make out that the aforesaid beams are formed by the solderins; t02;ether of small but comparatively thick-rayed hexactins that exhibit no regularity as to their relative disposition. The hexactins (under 100 ,« in length, and 10 n in thickness, of rays) are in characters quite similar to those that go to comj)Ose the basidictyonal framework of the species (see PI. XV., fig. 12). In fact, I have no hesita- tion in regarding the above described reticular bodies to be the STAUROCALYPTÜS GL ABER. 211 basidictyonalia belonging to quite young individuals of the species : it appears that a brood of the young had temporarily or otherwise settled themselves on, and had each formed its own basidictyonal framework around, the parenchymalia of the old specimen in question. It is impossible to do more than speculate as to the mode of origin of that brood of the young. But one point appears to me to be almost certain, viz., that the reticular bodies treated of are the same as those described by F. E. Schulze ('99, p. 64 ; PI. XIV, figs. 2-6) from the l)uds borne on the prostalia lateralia of Rhabdocalyptus mirabilis. Seen in this light and from my point of view concerning the phylogenetic relation between the " Lyssacina " and the " Dictyonalia " (Contrib. III., pp. 23- 25, foot-note), it seems to me readily explicable, if, in the de- velopment of a firmly seated lyssaciue Hexactinellid, certain sup- porting spicules early underment ankylosis and thus acquired a dictyonine character. But to return to the general characters of the species. Summarily speaking, w^e have here to do with thick-walled, sac- cular or vase-like, firmly seated forms, which, after attaining a certain size, are laterally compressed to a greater or less degree. Superiorly the wall somewhat closes in towards the sharp, but not thin, oscular edge. The young are ovoid in shape and are furnished with conspicuous, but slender and isolated prostal needles, which, after the full growth of individuals, seem to become entirely or almost entirely lost. Pentactinic prostalia do not occur in any stage of the life. The dermal surface is moderately smooth. The delicate ectosomal skeleton consists of a minutely quadrate-meshed dermal lacew^ork and of supporting hypodermal strands. The latter are 212 ART. 7. 1. IJIMA : HEXACTIA'ELLIDA, IV. of various thicknesses but on the whole are thin. They run in somewhat curved courses and intersect with their fellows at various angles, thus forming meshes of very irregular shapes. Frequently a number of the strands are seen to radiate, though in an irregular manner, from the point where a pillar or pillars from the choanosme join the ectosome. The endosome appears much like the ectosome. It lies closely applied to the internal choanosomal surface. The gastralia form a continuous lacework over the meshes bounded l)y the hypo- 2;astral strands which are on the whole somewhat thicker than the hypodermal. Characteristic of all the specimens are the rather cavernous appearance, and the somewhat loose and light-looking texture, of the wall. The former character is due to the spacious development of the subdermal cavity and to the comparatively large caliljer of the canals proceeding from it. The subdermal cavity is of a width known to me in no other Acanthascine species. In large specimens it may in some places be nearly 10 mm. wide. Conical or irregu- larly ridge-like projections — the pillars — ^^join the choanosome to the ectosome. The spaces between such adjacent pillars vary in ex- tent and may each contain sometimes a single and sometimes several large incurrent canalar apertures. These are round or oval, measur- ing up to about 7 mm. dia. in large individuals, and may be separated from one another by an interspace of 4 mm. or more. The incurrent canals are not deep and pit-like, but rather shallow and funnel-shaped. This is probably in some way related to the wide development of the subdermal space, which not in- fre(][ueutly passes over into the canals without any ^perceptible demarcation. The apertures of branch canals opening into the primary canals are generally plainly visible from the outside. STAUEOCALYPTÜS GLABEE. 213 Excurrent canalar apertures ou the gastral side are ou the whole much smaller and situated more closely together than the iucurreut. The canals they lead into are deep and pit-like, unlike those of the iucurreut system. The soft paris (see PL XV., fig. 11) were studied hy me but without results of any special interest. The limiting trabe- culœ of both the dermal and the gastral surfaces are to a great extent membranously developed. The free proximal ra3'S of the gastralia heave up the gastral membrane in a tent-like manner. External trabecule are most numerously present Avhere the choano- some most closely approaches the ectosome. A large part of the wide subdermal spaces is quite free of them. Chambers, of the usual shape and appearance; 100-165 // in diameter. Arclueocytes and thesocytes much as I shall later describe from Rhabdocalyplaa capillatus (PL XXIII.) . Spiculation. PL XV., fig. 11, will serve to give a general idea of the spiculation. The parcncJiymal princlpalia are elongate spindle-shai)ed oxy- diactions with rough-surfaced ends and aie more or less bent in a bow-like manner. They may attain a length of 13 mm. and a thickness of 190 n in the middle. The smaller ^^'ii'enchymalia and the comitalia are of the usual characters. The diactinic proslalia, present on the younger specimens, may be 25 mm. long or longer. In thickness they may measure 214 ART. 7. — I. ijima: HEXACTINELLIDA, IV. 90 [i, but are usually thinner. The major part of the length is protruded free, while the part rooted in the wall is accompanied by fine comitalia. The hypodermal'ia, which go to form the hypodermal strand.*, are pentactins and diactins, in both of which the rays are al- ways smooth except at the roughened ends. The extreme tip of the rays is either rounded or conically pointed. The hypodermal pentactins are of a large size. The curved or nearly straight paratangentials may be 8 mm. long with a breadth of 80 // at base. The unpaired proximal ray, which is always straight, is much longer and is invested with fine comital diactins in the usual manner. This ray, together with the roots of the diactinic prostalia that may ])e present, forms a part of the so-called pillars. Seen on the sponge surface, the centers of hypodermal pentactins are situated either isolatedly or a few together in loose groups. Their loose arrangement sufficiently accounts for the fact that the four paratangentials in a spicule generally have a regularly cruciate disposition, each being free to take its natural direction during development (cfr. p. 131). Only occasionally have I met with such hypodermal pentactins as have paratropal paratangentials. In certain very small specimens I ascertained that the hypo- dermalia consisted of pentactins alone. Some of these showed paratangentials with microtubercles sparsely distributed throughout the length, and seemed to intergrade with the larger dermalia. Whereas, in all the larger specimens the paratangentials of hypodermal pentactins are usually accompanied by a greater or less number of slender diactins, which are scarcely distinguish- able from parenchymalia of similar dimensions. In some places STAUROCALYPTUS GL ABER. 215 the strands consist of tlie diactins only, and in still other places these are seen to rnn singly, not being combined into bundles. The hijpogaHral strands are made up of similar diactins, and of diactins onl3^ The dermaUa (PL XV., figs. 4-G) are almost exclusively stauractins ; rarely pentactins and still more rarely tauactins or diactins. The stauractins are very slightly convex on the out- side. On some of them the atrophied proximal ray, and oc- casionally also the distal, may 1)0 represented by a boss-like pro- tuberance. Length of rays, 100-165 I' ; breadth at base, 6-9? n. The rays may be said to be entirely rough; and it is a remark- able fact that the microtubercles on their outer surface are developed into more or less prominent, conical and vertically erect spines. Those on the lateral sides of the rays are much less strongly developed, while those on the inner side may be said to be obsolete. The length attained by the spines on the outer surface differs somewhat in different individuals, but at all events their unusually strong development constitutes one of the characteristics of the species. Even in a small specimen of only 6 mm. height, the spines in the position indicated are quite prominent, though much thinner than in the larger specimens. The gaslraUa (Ph XV., fig. 7) are oxyhexactins of a moder- ately large size. All the six rays in one spicule may sometimes be of nearly equal length, but more frequently the free proximal ray is the longest and the distal the shortest. Length of proximal rays, 450-560 ji ; of ])aratangential rays, 330-352 /^ ; and of distal rays, 190-262 /x Thickness at base, about 10/^ on an average. The tapering rays belonging to one and the same spicule may 216 ART. 7. 1. IJIMA : HEXACTINELLTDA, IV. all be nearly equally rough on account of the presence of mi- crotubercles, but the usual condition is that the prolonged proximal ray is, in comparison with the others, much more pronouncedly rough, the microtubercles on it being developed into distinct, conical and vertically projecting prickles. The gastral hexactins are generally so arranged as to form with their paratangentials a continuous quadrate-meshed lacework, in which the sides of the meshes, exactly like those in the dermal lacework, are formed of the two apposed rays belonging to each of every two adjacent gastralia. Occasionally two gastralia lie very close together, making the four laths proceeding from their position three rays strong instead of two. In the smallest specimen I have had (G mm. high), the gastralia were quite sparsely i^resent in scattered arrangement. No special canalaria were found in the species. The oxyhexasters (PI. XV., fig. 10) measure 98-114 // in diameter. They all seem to be normally developed, neither hemihexactinose nor hexactinose forms havins; been noticed. Fur- ther a distinction between those in the periphery of the wall and others more deeply situated can scarcely be drawn. The prin- cipals are usually short but distinct ; sometimes they are quite obsolete. Two to four — commonly three — terminals are borne by a principal. They are generally thin and nearly straight, show- ing a slight roughness at base when seen under a high power of the microscope. In some of the rosettes the terminals may be nearly twice as thick as in others with which they promis- cuously intermingle. The above oxyhexasters are not found in the ectosorne, but occur abundantly in the choanosome, especially in the deeper STAUROCALYPTUS GLABER. 217 parts. Very frequently they are seen shifted out to a point midway towards, or right on the tip of, the free proximal rays of gastral hexactins (fig. 11). The discoctaslers (PL XV., fig. 9) are common in the deeper parts of the wall, probably never occurring in the subdermal space. They may attain a size larger than is known to me from any other Acanthascinje. Diameter, generally 500-660 /^, but sometimes reaching down to 300 /^- or even less. In the very young specimen of only 6 mm. height, the discotasters, of which only two or three in all were found, measured 240-352 /^. The central node is of a plain appearance. The secondary principal gently thickens outwards, forms about one-third of the entire ray-length and splits at the outer end into five or six slightly rough, but sometimes nearly smooth, straight and slightly divergent terminals, which together give rise to a tuft of an elongate conical shape. The terminal disc is without marginal teeth, being simply like a 2:)inhead in shape. llicrodiscohexaslcrs (PI. XV., fig. 8) of spherical shape and 15-22 fi diameter occur extremely rarely in the older specimens, but are not uncommon in the younger. They are found on or near to both the dermal and gastral surfaces. About ten termi- nals of extreme fineness, each having a minute terminal swelling, occur to the flat or externally convex disc at the outer end of each moderately long principal. Finally — the basidictyonal plate (PI. XV., fig. 12) is of quite a similar appearance as in other species of the subfamily. In the larger specimens it is of a considerable thickness and consists of both directly and synapticularly connected, thick-rayed and mi- 218 ART. 7. T. IJIMA : HEX ACTINELLIDA, IV. crotubei'cnlate liexactins, amongst whicli there occasionally occur pentactinic and even stauractinic forms. Its snrfaee in direct contact with the substratnm is lined with a thin and particnlarlj close-meshed reticular layer, the limiting basal-plate. — Of the small basidictyonal mass shown in fig. 13, I have already spoken on p. 210. STAUROCALYPTUS HETERACTINUS Li. PI. XI., figs. 1-10. Staurocalyplus heteradinvs. Ijima, '97, \). 50. This sjiecies was described liy me in '97 from a single specimen, A second specimen has not been obtained. The type-specimen (PI. XI.,' fig. 1 ; S. C. M. No. 409) comes from a depth between 501 and 572 m. It is of the size of a bean and represents a strongly laterally compressed pouch, measuring 21 mm. long and 10^ mm. broad in one direction and 6 mm. broad in another. On one side of the upper rounded end is the small oval-shaped oscuhim with its thin simple edge. The op- posite end shows two processes with torn off terminations. Both these processes probably served to fix the sponge to the substratum. Thickness of wall at the middle of the body, about 21 mm. Possibly the specimen is a young individual. The external surface is tolerably smooth, being without pros- tals of any kind. Examined under the hand-lens, there are seen to proceed upwards ft'om the basal processes obliquely running and intersecting strands of rather coarse fibers which build up STAUKOCALYPTÜS IIETERACTINÜS. 219 the parenchyma. Close over this is an ill-defined hypodermal network with irregular meshes of various sizes. The dermal layer is only partially preserved, and is very indistinct. The incurrent canalar apertures are represented by small gai)s between the parenchymal strands and quite occasionally by pit-like but «hallow depressions. The gastral surface exhibits a nnmljer of freely open ex- current apertures, which may reach up to | mm. in diameter. Some of the excurrent canals are pit-like but never deep in relation to the thinness of the sponge-wall. Altogether, the appearance of the wall is much like that of the oscular region of a larger Acanthascine species. Spiculation. The parenchymalia include an abundance of comparatively strong princijnilia of elongate spindle-shaped or bow-shaped oxy- diactins, which may attain a length of 12 mm. and a breadth of 220 fJ- at the middle. In them the ends are smooth. The smaller parenchymalia present no features worthy of special mention. The hypodermalia (PI. XL, fig. 2) are moderately large and sti-ong pentactins, occurring commonly but in irregular distribu- tion. The paratangentials, which are regularly cruciate and never paratropal, usnally measure under I mm. in length l)ut are some- times longer. The unpaired proximal ray is always much longer than the paratangentials. The rays may be 30 /^ thick at base. The surface near the conically pointed tip is either quite smooth or sparsely beset with microtubercles. Occasionally the hypoder- 220 AKT. 7. 1. IJIMA : UEXACÏIMÙLLDIA, IV. malia are represented by paratangentially situated slauraetiiis ; more rarely by tauactins. Certain diactins seem also to lend themselves to the formation of hypodermal strands. The dermalia (Bl. XL, figs. 3 and 4) are predominantly stauractins, occasionally pentactiiis and tauactins and rarely diactins. The rays are slightly rough all over and have rounded tips, which are occasionally somewhat swollen. The size varies considerably, some of the spicules being in axial length twice or even three times as long as the smallest. Thus, a small dermalia measured had rays of only 90 /^ length and 0 n thickness at base, while a large one may have rays 270 n long and 13 n thick. The largest nearly approaches the size of the smallest hypoder- malia, but the latter seem to be always distinguished from tha former by the fact that the rays are rough at the ends only. Exceptionally thin-rayed and smooth oxystauractins, found here and there in the dermal layer, I take to be dermalia that are still in an incomplete state of development. The sei)arate dermalia are irregidarly disposed as regards the mutual orientation of their paratangentials. The gaüralia (PI. XL, fig. 5) are pentactins and stauractins, the former being present in somewhat greater relative abundance. Rarely are they tauactins or diactins. The rays are quite like those of the dermalia but are generally smaller. Length of rays, 55-100 /A Average thickness at base, 62 //. In the pentactins the unpaired ray is directed distad ; sometimes the proximal ray is represented by a boss. The gastralia are nowhere numerously present. STAUROCALYPTUS HETERACTINUS. 221 The oxyhexasters (PI. XI., figs. 7 and 8) liave a diameter of 106-114 It-. They seem to be never hemihexactinose or hexactin- ose. As regards appearance they may be divided into two cate- gories, which seem however to intergradate. '1 he one category (fig. 7) is composed of those mostly lying in the periphery of the sponge-wall, and in them each principal of a perceptible length bears 2-4 (usually 3), slender and obsoletely rough or nearly smooth terminals. To the other category (fig. 8) belong the great majority of the oxyhexasters abundantly present near the "-astral snrface and occasionally a few of those fonnd in the periphery. In these the principals are extremely short and bear each 2 (sometimes 3), strong and distinctly rough terminals. The roughness may be developed into retroverted microspines or barbs on the basal parts of the terminals. The discodasters (PI. XL, fig. 6) vary in diameter from 110 />« to 200 lu They are tolerably common, especially near the gastral surface. Those in the peripheiy of the wall are on the whole smaller than others more deeply situated. In the former (of which PI. XL, fig. 15, in reality taken from an undetermined Staurocalyiyius, might well pass for a representative) the secondary principals are as slender as 4 /^ or less. In the latter (fig. 6) the same may be fully 6 y thick. Generally speaking, the central node is either plain or is supplied with the six boss-like promi- nences. The secondary principals make up about two-fifths of the entire ray-length. The terminals, 2-7 in number to a prin- cipal, are straight and form a tuft which expands generally but little outwards. The terminal discs appear simply like piidieads. The microdiscohexasters (PI. XL, fig. 9) are of the usual 222 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. appearance. Diameter, 16-19 jj.. They occur in scattered dis- tribution both near the gastral surface and in the peripheral parts. In some places in the dermal membrane I iiiul them to be quite common. STAUROCALYPTUS PLEORHAPHIDES Ij. PI. XVI. SlaiU'ocalypias pleorlLapJLldcs. Ijima, '97, p. öÖ. Three specimens of this species have come under my ob- servation. Rougldy speaking, tliey all represent thick-walled, elongate, pear-shaped sacs, lirmly attached by the narrowed lower end and provided with prominent 2>i'0stal needles and pentactins. In external aj^peanmce the species bears a close similarity to Scyphidium longispiyia. The first specimen (PI. XVI., fig. 2 ; S. C. M. No. 220) is from a depth of between 42Ü and 572 m. in Honiba. Height of body, 43 mm. Greatest breadth, 23 mm. Greatest thickness of wall, 6 mm. The roundish osculum at the upper end, 7 mm. in diameter ; leading into a deep cylindrical gastral cavity. Base, stalk-like ; 6 mm. thick. The second specimen (PI. XVI. fig. 2 ; S. C. M. No. 415) comes from a depth of about 300 fms. (say, ôôO m.) in Inside Okinose by the Sengenzuka-line. It is attached on the external side of a Hexaciinclla lorica. Body, 5Ö mm. long ; somewhat laterally compressed, measuring 3-3 mm. by 26 mm. in breadth at the middle. Basal end, lU-12mm. broad. The irregularly el- STAUROCALYPTUS PLEOEHAPHIDES. 223 liptieal oscukim on one side of the npper end, 18 mm. b}^ 7 mm. across. Thickness of wall in the middle, 8 mm. The third specimen again is attached to a large Hexadlnella lorica (S. C. Isl. No. 448), together with Lanuginella pupa, Leu- copsacus scoliodocus, etc. Locality, Outside Okinose by the Iwado- line. Total height, 40 mm. Greatest breadth, 30 mm., the body being nearly circular in cross-section. Thickness of wall in the lower part, about 10 mm. The roundish osculum, 13 mm. in diameter. The following are the details of the macroscopic characters of all the three specimens taken together. The oscular margin is thin and simple-edged. By far the greater part of the external surface is uneven on account of the presence of numerous small conuli, lying at distances of 3-10 mm. from one another. From each such conulus there spring a num- ber of prostat spicules arranged in a loose divergent bunch. The prostat bunch generally comprises both diactins and pentactins, but sometimes consists of the latter alone. The diactinic prostals are generally strong needles of various lengths. They may project to a length of 25 mm. or more, be- ing directed on the whole obliquely outw^ards and upwards, though there may exist some that proceed straight outwards or even somewhat downwards. Those in the broadest part of the body are the longest. The needles give an altogether spiny appearance to the sponge. Seen under the hand-lens, some of them may present a dirty brownish color due to a thin incrustation of some foreio;n substance. Adhering to them are seen here and there some animal remains, among which Foraminifera shells are the more common. The pentactinic prostals are of a modeiately Inrge size. They 224 ART. 7. — I. uni A : hexactinelltda, iv. form over the dermal surface a gossamer-like covering, which in places is about 5 mm. thick. They generally protrude in groups of two or more but may sometimes stand out singly. The dermalia, together wdth the hypodermal spicular rays, form quite an irregularly meshed latticework. The meshes are never quadrate but triangular, trapezoidal or indefinite in shape. This irregularity has its explanation in the fact that the dermalia are predominantly straight diactins, wdiich may lie in all possible directions in the plane of the dermal layer (PL XVI., fig. 15). The subdermal space is inconspicuously developed. The variously sized apertures of incurrent canals are largest in the broadest region of the body. Here they occasionally measure 3 or 4 mm. across. Apertures of a similar size are separated from one another by a space as wide as, or wider than, their own width. The gastral surface presents a loosely felt-like appearance, not being covered with a well differentiated endosomal layer (PI. XVI., fig. 1). On it the excurrent canals, up to o mm. in width, open with free apertures. These canals are mostly not deep but are seen soon to divide up into branches. Spiculation. The larger parenchymalia, which may be called the princi- palia, are elongate spindle-shaped oxydiactins, either nearly straight or gently bent in a bow-like manner. They may attain dimen- sions of 7 nun. length and 80 /^ breadth at the middle. The ends are usually rough in varying degrees ; in other cases they are smooth. Under certain circumstances, one end of the diactins STAUROCALYPTUS PLEORHAPHIDES. 225 may be rounded while the other is as usual acutely pointed, as e.g. in some of those diaetins that just reach the dermal surface with one of their tips, in which case it is the distal end which may be rounded off (PL XVI., fig, 14). The smaller parenchymalia, including the fine comitalia, are of the usual appearance. They seem to grade down in dimensions uninterruptedly to the small diaetins which I shall describe further on as the gastralia. The j:)ro.?/a/ diaetins, which are in fact to be regarded as enormously developed parenchymal principalia, are of various sizes. A small one may measure only 10 mm. in length, while the largest measured was 40 mm. long and 253/'- thick. The needle- like spicules are straight or nearly straight, tapering perceptibly towards both ends. The outer end is usually found to have been broken off; the inner is either acutely or bluntly pointed, the subterminal surface being smooth or sparsely beset with micro- tubercles. The entire exposed surface of the prostat needles, — at least, of the larger of these, — is minutely and densely rough, in exactly the same way as the paratangentials of certain prostal pentactins soon to be described. The roughness extends a short distance into the parts rooted in the sponge-w^all, gradually fading into a perfectly smooth surface. The parts in the wall, as also the same parts of the shafts of hypodermal and prostal pentact- ins, are generally accompanied by some comital parenchymalia. The hypodermal pentactins are somewhat variable in size. The paratangentials, which are generally not quite straight but rather wavy, may be 5 mm. long or longer. The straight shaft or the unpaired proximal ray is always much longer than the 226 AKT. 7. — I. IJIMA : HEXACTINELLIDA, IV. paratangential in the same spicule. The pentactins situated near the oscular margin and already externally protruded as prostalia, I have found to be unusually small, measuring not more than half a millimeter in the length of paratangentials, which in these small pentactins are generally arranged in the form of a regular cross to each. This cruciate arrangement of the paratangentials also occurs, but only occasionally, in the larger pentactins lower down on the sponge. In by far the greater number of these the rays referred to are paratropal. The paratropism is carried out to varying degrees in different spicules. In some cases, one of the four angles formed by the rays is simj)ly obtuse while the rest are all acute though greater than 30° each ; in other cases the four rays form only three angles, each of about 30° or less. It is to be noted that whenever two or more paratropal pentactins form a close group, whether by themselves alone or in company with diactinic prostalia, the paratangentials of each pentactin are, as it were, pushed away from the center of that group (PL XVI., fig. 14). I shall have to return to this point under Rhabdocalyptus victor. All the rays of the pentactins are at first smooth except at the ends which are rough. So are they in most of the spicules in the hypodermal situation ; however, here and there among these are to be seen such as show the paratangential cross finely rough all over. The roughness is caused by minute, erect and sharply pointed processes, which, when seen from above, appear to be somewhat laterally compressed so as to present a shape elongated in the direction of tlie axis of the ray they beset (PL XYL, fig. 13). The microspines reminds me of those I have seen on the prostal pentactins of Scyphidium lo7igispina. To judge from what appear to represent developmental stages, the STAUKOCALYPTÜS PLEORHAPHIDES. 227 roughness first sets in as an extension of that at the ends of the paratangentials ; thence it proceeds to develop towards the spicular center and from this point proximad for a short distance on the shaft. Kot that the roughness develops on all the old pentactins before these are protruded as prostalia. Of those which already stand out and form the gossamer-like covering on the exterior, many indeed have the rays roughened in the manner described; but in certain others these are perfectly smooth exce})t subterminally. It is difficult to say if all the latter will eventually acquire the character of the former. Kunuing along with hypodermally situated paratangentials are occasionally seen some fine diactins, which thus help to sup- j^ort the dermal layer. The dermalia (PL XVI., figs. 3-5) are rough diactins. Exceptionally they may be stauractins or tauactins ; extremely rarely, orthodiactins and monactins. The diactins measure 264-520 a (on an average, 400 n) in total length and about 9? « in thickness at the middle. They are generally gently bent in a bow-like mannei", the concavity facing downw^ards. They taper slightly tow^ards both ends which are either rounded or conically but bluntly pointed. The center is usually without an external swelling ; only occasionally it is marked by a pair of opposite bosses and much less frequently by a single unilateral boss. A noteworthy feature of the present species consists in the fact that the (jastraUa are represented by straight diactins. I at first felt inclined to pass over the matter by simply assuming a total absence of gastralia for the species. Upon further studies, how- 228 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. ever, and having found the same diactins in like abundance over the entire gastral surface of the tlyee specimens examined, I have come to see no inappropriateness in calling them the gastralia. Moreover, a second Acanthascine species with diactinic gastralia is now known, viz., Rhabdocalypius phmiodigiiaius K. Kirk p. It should, however, be said that the gastralia in the present species seem to intergrade with the parench3analia by an uninterrupted series of transitional forms. Like the dermalia, the gastralia lie in quite irregular disposition on the gastral surface (PI. XVI., lig. 16). Many of the gastralia are much like the dermalia both in dimensions and general appearance. AVhat may perhaps be men- tioned as slight deviations shown by the former are the facts that the microtubercles on the surface are on the whole more sparse and more thinly scattered and that the middle of the spicules is frequently, but not always, marked by a gentle annular swelling. Many others of the gastral diactins are considerably larger (PI. XVL, fig. 6), ac(i(uiring characters approaching tliose of parenchy- mal diactins. AVith the increase in size, the microtuberculation becomes more and more sparse except at the ends, finally rendering the middle parts of the spicules perfectly smooth. In one specimen (No. 226), in wliich I have studied the spiculation most closely, I find the gastral skeleton in the prox- imity of the osculum scarcely at all distinguishable as to its elements from the dermal. That is to say, the gastralia in that region contain, besides an abundance of the rough diactins, occasional stauractinic and tauactinic forms, underlying which spicules are some hypogastral pentactins with cruciate paratan- gentials. Deeper down and in by far the greater part of the gastral surface, the hypogastral pentactins are not found, while STAUROCALYPTUÖ PLEORHAPHIDES. 229 as gastralia there occur only diactins though in somewhat diminished numbers. The oxyJiexasiers (PL XYI., lig. 7), occurring in moderate abundance in all parts, are partly normally developed and partly hemihexactinose. Hexactinose forms, with the terminals either straight throughout or bent at base, were met with in only a few cases. Diameter, 114/^ on an average. The terminals are rather strong, measuring about 2 u. across at base ; they are nearly straight or slightly bent and show a rough surface, due to microtubercles which are sometimes distinctly retroverted but never much pro- longed. The principals are extremely short and often almost obsolete. In most normal oxyhexasters all the principals bear two terminals each, so that the total number of terminal points is tw'elve. Sometimes some — not all — of the six principals may show" three terminals, in which cases the points in one spicule number more than twelve in all. Thus, not larely have I come across normal oxyhexasters with sixteen or seventeen terminal points. In the hemihexactinose forms, the uniterminal rays are either straight or bent at base, the rest of the rays being always biterminal. The peculiarly twisted and evidently malformed oxyhexaster, shown in PI. XVI., fig. 8, was observed l)ut once. It possesses six terminals in all, but these evidently belonged not to as many principals, but to the four or five remaining of the original six principals. The curvature of the rays is in different planes, the relative orientation of which could not be determined. The dUcociaders (PI. XVI., figs. 9-11) are present in abun- 230 ART. 7. 1. IJIMA : HEXACÏINELLDIA, IV. dance in the entire wall. They are slender-rayed and on the whole small thongh of varions sizes. The diameter is usually 140-200 I'- ; occasionally, only 100 /^. The central node is either plain or tubercled. The principals are about half as long as the terminals. Number of termiuals to a principal, 2-4 ; usually 3 ; probably never more than 4. They form a slightly diverging tuft and are nearly straight or perceptibly bent outwards. Under a high power of the microscope they appear to be rougli-surfaced. On the minute terminal discs the marginal serration is unrecog- nizable. Malformed discoctasters, in which one or more primary terminals stand free without fusing with any of the secondary principals, are of no infrequent occurrence. The microdiscohcxaster (PI. XVI., fig. 12) are of the usual appearance and 20 !'■ in diameter, and are found, mostly near the gastral surface, in small numbers and in scattered distribution. Finally, as to the basal plate. I have seen this in the specimen shown in PL XVI., fig. 2, which is attached to a Hexactinella lorica. The dictyonal skeletal parts of this sponge, at the place where the said specimen is fixed, are enveloped in a thin, small-meshed, siliceous reticuhun, evidently the limiting basal-plate of the latter. The beams of this plate are sparsely microtuberculate and look quite like those I have figured in PI. XXII., fig. 17, from Rhahdocalyptiis capillaius. Outside that plate and in the sponge under consideration there may possibly occur at places some basidictyonal hexactius, but these were not actually encountered. UNDETERMINED STAÜROCALYPTÜR. 231 UNDETERMINED STAUROCALYPTUS. Here I wish to mention three specimens («-r) which I have studied but prefer to leave unnamed, tliough placing them provisionally under Staurocalyptus. They are all very small and therefore very jDrobably 3'oung specimens in which the characters may not have been fully developed. Sfam'ocfdypttfs sr. o.. (Figures already published in Contrib., III., PL III., figs. l-G). This little specimen (S. C. M. No. 437) was found on a dead Hexaclinella lorica from Outside Okinose, together with Leu- copsacus scoliodocus, Lannginella pupa, etc. The body of a barrel- like shape is somewhat larger than a grain of rice. It shows at the lower end a rigid basidictyonal mass measuring about 2 mm. across. From that mass, as also from the body proper, there spring out some fine prostat needles of a considerable length, mostly directed obliquely upwards and outwards. The dermal surface is smooth. The parenchymalia are chiefly diactins under 14/-« in thickness. In most of them the spicular center is marked externally by an annular swelling or by two or four knobs arranged in the usual manner. It seemed to me that the subterminal roughness extended over a relatively greater area than usual, some of the shorter diactins showing sparsely distributed obsolete microtubercles nearly all over them. Highly remarkable is the fact that some of the parenchymalia are apparently hexactinic. The parenchyjnal 232 ART. 7.— I. IJIMA : IIEXACTINELLIDA, IV. oxyhexactins may have rays 95 t^- long and 8 n thick at base, the entire surface being thinly microtubercled. Their position and manner of occurrence scarcely warranted interpreting them as gastralia. The derraalia are both stauractins and pentactins in about equal numbers (Contrib. III., PL IIL, fig. 1). Size various, 43- 152/^ m length of rays and 4-9/^ in breadth at their base. The relatively strong and slightly tapering rays are entirely rough on account of sparse but distinct microtubercles. Tlie atrophied rays are sometimes represented hj an external and an internal knob in the stauractins, and by an external knob in the pentactins. The larger dermal pentactins approach in size the pentactinic hypodermalia in which the paratangential rays may reach 380 1^ in length and are smooth except near the ends. The gastralia are hexactins found in scattered distribution {I. c, PI. IIL, fig. 2). Rays as in dermalia ; length, 34-72 IK Oxyhexasters (/. c, PI. Ill,, figs. 3 and 4) are of common occurrence. They are normally developed, there being two or three, slender and rough-surfaced terminals to each principal, which is short. Diameter, 88-1 OG //. Discoctasters (/. c, PI. IIL, fig. G), a small number found ; small and si ender- ra yed ; measuring 130-144 /< in diameter. Microdiscohexasters (/. c, PI. IIL, fig. ö), very small and delicate; 15,« in diameter. They occur in abundance in the wall, l)ut especially in the periphery. The basidictyonal mass is of the usual structure, l^eing com- posed of fused hexactins with thick, short and sparsely micro- tubercled rays. The skeletal beams of the Hexactinella, at the spot giving attachment to the specimen, are entirely enveloped by a thin and small-meshed limiting plate. UNDETERMINED STAUROCALYPTUS. 233 The spiculation as described above seems to come nearest to, and indeed closely resembles, that of Staurocahjptus rœpcrl (F. E, ScH.). But, under the circumstances, I hesitate to make a definite specific determination. Staurocalypffts sp. ß. (Figures in Contrib. TIL, PL III., figs. 7-13). On another HexactineUa lorica from an unknown spot in the Sagarai Sea was found the small specimen here to be described, together with some other Hexactinellid species {Leucopsacus scoliodocus, Lanuginella pupa, Skiurocalyptus pleorhaphides). The specimen is shown in /. c, PL III., fig. 9. Barrel- shaped ; only al)0ut 4 2 mm. high ; with numerous fine prostal needles and a deep gastral cavity. A comparatively wide subder- mal cavity exists. The parenchymalia are all slender diactins, rough only near the sharply pointed ends and generally without a swelling around the spicular center. Prostal needles, 4 mm. or over in length and up to ö7 /^ in thickness. The dermalia are mainly oxystauractins and rarely oxypent- actins, in both of which the cruciate paratangentials are slightly convex on the distal side. Rays, slightly rough ; 100-154/^ long and 6-7è ix thick at base. (For the dermalia and hypodermalia, see /. c, PL III., fig. 7). The hypodermalia are oxypentactins with paratangentials up to 450 ix long and 27 /^ thick at base. The rays are usually smooth except at the roughened ends, but in some of the spicules I have found the paratangentials thickly beset all over with 234 AKT. 7. T. T.IIMA : HEXACTINELLTDA, IV. pointed microtiibercles in much tlie same manner as in Stauro- calyptus pleorhaphides. Special gastralia have not been found. Oxyhexasters (/. c, PL III., figs. 12 and 13), very abundant, generally normal and rarely hemihexactinose. Diameter, GS-IH/-«. Terminals, two (at most three) to a principal ; nearly straight ; obsoletely rough ; much thinner in the smaller than in the larger oxyhexaster. Discoctasters or rather their representatives were found in quite a limited number, — only two cases in all which were dis- covered after a careful search throughout the entire specimen. They are both shown in /. r., PI. III., figs. 10 and 11. The case of fig. 10 is without doubt that of a malformed discoctaster, — at any rate, one in which some of the primary terminals remain free without uniting into secondary principals, though certain others are united into such for a short distance and situated in proper positions on the central node. This central node is nearly cul^ical and shows the rounded bosses corresponding to the primary principals. Diameter, 122 /-«. The other case, shown in /. c., fig. 11, may almost be called a discohexaster, in which the primary pi'incipals are in part still distinct and partly fused together. The terminals are all free and radiate in all directions from the ir- regularly shaped central mass, showing as yet no trace of rearrange- ment into the eight bunches of a discoctaster. I regard the s[)icule, though possibly a case of deformity in itself, as representing an early stage in the transformation of a discohexaster into a discoctaster. Speaking on a priori grounds, discoctasters should have passed through a phylogenetic stage appearing much like the spicule now in question. Diameter, S5 p-. Microdiscohexasters, quite like those of aS'. sp. a [I. c, fig. 5). UNDETERMINED STAUKOCALYPTÜS. 235 They are common on or near the gastral surface. The terminals are so fine that it is difficult to observe them in Canada-balsam preparations. Basidictyonal beams as in Ä sp. «; nearly smooth, the scanty microtubercles present being quite obsolete. Staiivocalypt as sp. '/. (This Contrib., PI. XL, figs. 11-15). On the same Hexactinella lorica as that which bore the foregoing specimen ((^), was found another (PI. XL, fig. 11) of about the same size and appearance but differing somewhat in some points of the spiculation. Parenchymalia, as in S. sp. [-i. Prostal oxydiactins, under 34 IX in thickness. Dermalia (fig. 13), exclusively stauractins ; more or less convex on the outside and with slightly roughened rays 90-230/^- long and not thicker than 7? ix at base. Hypodermal oxypentactins, moderately large ; with compara- tively slender rays, which are smooth except at the roughened end. Length of paratangentials, up to 700 // ; thickness at base, under 111 i'-. The unpaired proximal ray is longer than, — fre- quently fully twice as long as — the paratangential in the same spicule. Xo gastralia, nor microdiscohaxasters, were discovered. Oxyhexasters in moderate abundance (fig. 14) ; normal and all of a uniform appearance. Two or three, sometimes four, slender, rough terminals to a principal wdiicli is very short. Diameter, 128-138 //. 236 AßT. 7. — I. IJIxMA : HEXACTINELLIDA, IV. Discoctasters (fig. 15), common. The six bosses present on the central node. Prhicipals slender, at most 4 /^ thick ; about i the length of the entire ray. Terminals fine, 3-6 in a gently expanding tuft, with the minute terminal disc sliaped like a pinhead. Diameter of the spicule, 1'28-160/a Basidictyonalia, as in S. sp. ß. In spiculation the specimen seems to resemble most closely a young aS''. glaher, but differs not inconsiderably in the qualifica- tion of the roughness of derraalia and in the size of oxyliexasters as well as of discoctasters, — differences which at least render doubtful the propriety of considering the specimen as of that species. I have come to this view after carefully comparing it with the smallest specimen (6 mm. high) of S. (jlaher in my possession. RHABDOCALYPTUS F. E. Sch. Hypodermalia include pentactins in which the paratangentials, when fully developed, bear series of strong hook-like or prong-like spines. Veiled, but in some cases the veil may be lost. Key to the known species. a Dcriualia, predominantly iientaclins, or pentactins and stunractins in nearly eijiial nuail ers. a^— Dibcoctaster not more tlian 100;). in diameter. a-.— Discoetaster wilii nearly straiglit terminals, wliicli diverge Imt slightly outwards in each tuft. Dermalia pentactins, occasionally stauractins R. damoni (Lambe). (Pacific coast of N. America). KHABDOCALYFÏUS. 237 b"^. — Discoctaster with terminals bent outwards in each tuft. Dermalia pentactins, occasionally hexactins. (Oxyhexaster witli spherical central node) R. tener F. E. Sen. (C. of California). 6\ — Discoctaster more than 150 [J- in diameter. c". — Oxyliexasters normal, hemihexactinose and hexactinose. a^. — Discoctaster, 150-200 u. in dia. Dermalia pentactins, less frequently staur- actins. Gastralia, hexactins and pentactins B. ai^pcr F. E. ScH. (C. of California). b^. — Discoctaster, 240-300 \i.. in dia. Dermalia, stauractins and pentactins. Gastralia oxyhexactins. (Oxyhexaster with splierical central node) R. 7ioduIosn.'i F. E. ScH. (C. of California). d''. — Oxyhexasters, all liexactinose. Dicoctaster about 200 [i. in diameter. Dermalia pentactins, stauractins and diactins. Gastralia hexactins with the proximal ray 600-800 [i or more in length R. tenuis (F. E. Sch). (C. of California). It- — Dermalia nearly all stauractins. (Discoctaster, 180-240 \>- in dia. Oxyhexaster, 180-280 \>. in dia.) R. victor Ij. (Sagami Seaj. «• — Dermalia nearly all or at least predominantly compass-needle-like diactins. c^ — Sponge firmly attached at base to solid substratum. Gastralia hexactins. c'-. — Discoctaster more than 130 \i. in dia.; re.'iching up to about 180 \3-. c*. — The radial axis of gastralia about as long as, or shorter than, tlie length of the longest dermalia. The free proximal ray of gastralia about as long as any other ray in the same spicule. a*. — Among the diactinic dermalia, stauractins and pentactins occur but rarely. Oxyhexaster terminals more or less distinctly barbed at base. R. mollis F. E. Sch. (Sagami Sea). b*. — Among the dermalia, diactins are most numerous, but stauractins and pentactins ai-e also abundant. Oxyhexaster terminals slightly rougli- ened, not barbed. (Paratangentials of hypodermalia chagreened besides being spined) R. aiistralis Tops. (Antarctic). rf^.— The radial axis of gastralia nearly twice as long as the longest dermalia, or even longer. The free proximal ray is considerably prolonged over any other in the same spicule. (Among the dermalia, other forms tlian diact- ins occur, but quite exceptionally. Oxyhexaster terminals smooth or ob- soletely rough). c*. — Oxyhexaster only occasionally hexactinose. Discoctaster with princi- pals 20-25 [J. in length ; terminal discs toothed on the external side only or with the teeth much more strongly developed on that side than on the inner. Microdiscohexaster, 22-30 (J- dia R. unguiculcdus n. sp. (Sagami Sea). rf*.— Oxyhexaster, mostly hexactinose. Discoctaster with principals only about S [J. iu length; terminal disc equally toothed all around. Micro- discohexaster, 32-40 [J. in dia R. niirabilin F. E. Sch. (S. of Alaska). /^. — Discoctaster small ; 82-106 [i. in dia.; terminal tuft distinctly flaring at tlie outer end R. c((pilluiLi.>i Ij. (Sagami Sea). (/'. — Sponge with tinger-likc Ijasal processes and rooted in loose bottom by means of anchor-needles, ((rastralia, compass-needle-like diactins like the dermalia. Discoct- aster of two kinds, large (130-160 p. dia.) and small (00 [j. dia.) R. plumodir/iUUua KlKKP. (S. Africa). 238 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. The four species occurring in the Sagami Sea, viz., R. victor, R. mollis, R. imguiculaius and R. cajnllakis, will now be dealt with in detail. RHABDOCALYPTUS VICTOR Ij. Pis. XVII., XVIII., and XIX. Rhabdocalyptus victor. Ijima, '97, p. 52. — Cii. Gkavier, '99, p. 421. Specimens of this species have passed through my hands in no small numbers. The localities in tlie Sagami Sea where they have been obtained, so far as is known to me, are : Homba (572-859 m.) and Outside Okinose (by both the Iwado and Sen- genzuka lines, 501-572 m.). In these localities the species seems to be moderately common. INIost of the specimens bore at the base samples of a tufaceous bottom ; others grew on corals, on dead Chonelasma or on Hexactinella lorica. On the following page (text-fig. 8) and in Plate XVII., I have depicted several of the better preserved specimens that I have seen, which are illustrative of the shapes assumed Ijy the species. In text-figure 8, in which all the figures are drawn to the scale of s natural size, A represents a specimen which belonged to jMr. OwsTON. It was 23 inches high. Oscular margin directed upwards. Lateral compression of body especially distinct at the base, which is bent in the sagittal plane. RHABDOCALYPTUS VICTOR. 239 B and B', two views of a similarly shaped but relatively broader specimen (S. C. No. 426) ; B', as viewed from the side of the lesser curvature of the body. Height, 550 mm. Diameters Text-figure 8. — Ehabdocilyptua victor. All in \ natural size. of the oval-shaped osculum, 176 mm. and 118 mm. At the middle of the body, the likewise oval cross-section presents 240 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. diameters of 212 mm. and 148 mm. The strongly compressed base measures only 37 ram. in thickness from side to side. Thickness of wall at the middle of the body, 9.5 mm ; farther l)elow, up to 12 mm. Basal end attached to the substratum at two places, between which is a free dermal surface still preserving a number of pentactinic prostalia. C, a specimen (S. C. No. 460) slightly bent at the base and with the oscular margin expanded outwards all around and bent backwards. Height, 570 mm. Body laterally compressed Init only to a slight degree. Pentactinic prostalia preserved in many part of the dermal surface. D and D' , two views of a specimen with l)ent bod_v (S. C No. 2()7). Height, 348 mm. The oval osculum, 170 mm. by 73 mm.; elongate in the sagittal direction ; with flaring oscular rim. In the middle of the body the breadths are 100 mm. sagittally and 77 mm. transversely. The base is nearly round in cross- section, measuring about 40 mm. across. A low couical outbulging is seen at about the boundary between the lower and the middle third of the body on the side of the greater curvature but to one side of the median line. This is undoubtedly the beginning of the formation of a daughter person ; l)ut a secondary osculum has not yet opened itself at the summit. Thickness of wall at about the middle of the body, 10 mm. E, a slightly bent specimen (S. C. No. 416) attached by tubercle-like basal processes to a branched coral. Below and on one side is a funnel-like daughter person. Height, 285 mm. Near the ujiper end the walls of the two sides have come into contact and have fused together. Above this place of fusion, the gastral space and the original main osculum are narrow and slit- like ; below it is the main part of the gastral cavity which RHABDOCALYPTUS VICTOR. 241 communicates with the exterior only by means of the (secondary) oscuhim of the daughter person. Essentially similar in shape is the magnificient specimen shown in PI. XVII., reduced to one-fourth of the natural size (S. C. M. No. 423 ; from Homha, about 859 m.). The circumstances of its capture I have already had occasion to relate in my Contribution I., pp. 24-25. Its total height is 880 mm. As in many other cases, the body is laterally compressed, the compres- sion being especially distinct in the stalk-like basal region which is bent towards one side. The major and minor diameters of the osculum are 400 mm. and 300 mm. respectively ; those of the body at about its middle, 270 mm. and 220 mm. The basal region measures only 50 mm. trasversely from side to side. The lower end of the body is continued, in the direction opposite to the main osculum at the upper end, into a laterally compressed outbulging, wdiich soon divides into two thin-walled tubes, situated one behind the other in the median sagittal plane and each terminating in a secondary osculum directed downwards. The irregular attachment surface of the bent base is (snpero-inferiorly) 180 mm. long ;ind (transversely) 85 mm. broad. Judging from its disposition in relation to the directions of the oscula present, it is highly probable that the sponge was growing on a per- pendicular surface. The wall is 14 mm. thick in the middle of the body ; lower down, it is as thick as 19 mm. The gastral cavity extends into the laterally compressed base in the form of a vertical slit-like space, giving to the wall at the cul-de-sac end a thickness of only about 11 mm. On one side of the lateral wall (not seen in the ligure) and at a short distance from the main osculum, there exists in the wall an irregularly shaped gap, apparently the result of a mechanical injury. The torn edge of 242 ART. 7. 1. IJIMA : HEXACTTNELLIDA, IV. « the gap had, so to say, healed and regenerated a natural looking edge which appears nuich like thnt of the oscuhim. Speaking of the larger specimens in general, the shape is exquisitely vase-like and in a measure laterally compressed. The sponge stands either erect or is hent at the base. In the latter case the bending takes place invariably in the median plane con- taining the major diameter of the laterally compressed body. The daughter person or persons, occasionally found on the specimens, seem to be restricted in their location to the convex side of the bent basal region. The moderately thick wall is gradually thinned out at the oscular margin, which is sometimes outflaring and sometimes not. The base can not be said to be solid, since the gastral cavity extends almost to the attachment surface. The edge of the extreme base, as seen from the exterior, is often tubercled and indented reminding one of a cat's paw. The thin oscular edge is of a finely granular or densely felt- like ajopearance. It may be perfectly simple but more usually is provided with a thin row of fine needle-like marginalia, not over 10 mm. in length. The intervals at which these stand out are usually irregular. The dermal surface, when in a good state of preservation, is on the whole tolerably smooth except for the small papilla-like prominences and the slightly raised hypodermal strands, both soon to be referred to again. The dermalia form an exceedingly delicate lace work in which the minute meshes are quadrate in shape and measure on an average 170 /^ in length of sides. The hypodermal strands appear as thin streaks which, running in all directions, frequently intersect one another (PI. XVIII., fig. 2). Upon closer attention it will be seen that they form a number of radial EHABDOCALYPTUS VICTOR. 243 patterns, — that they converge towards numerous central points, lying 3-7 mm. apart from one another. At each of these is situated the papilla-like prominence mentioned above. The radial pattern is caused by the paratangentials of a number of hypodermal pentactins which are arranged in a group (PL XVIII., fig. 16), their shafts and the accompanying comitalia combining together to form a loose, vertically dipping bundle. The axis of this bundle is occupied by the erstwhile comitalia to the shafts of prostal pentactins which were once present at the spot but are now lost. The outer ends of the comital spicules in question form a compact bundle and, projecting more or less beyond the dermal surface, give rise to the small papilla-like prominence at the center of each radial figure brought about by the hypodermal strands. In the inferior region of the body, the prominences are generally disintegrated and take tlie form of little parallel tuft of fine spicules, projecting to a length of 2 mm. or less (PI. XIX., fig. 23) ; whereas, in the superior parts they are usually to be seen as whitish, slightly elevated spots. The prostal pentactins, when present, invariably stand out on the papilla-like prominences, either singly or a few together in a tuft. However, prostal pentactins are in general Ijut sparsely present in the mature specimens : they seem to become readily lost as they are somehow shed ofi". Many individuals are quite or nearly destitute of the prostalia in question, showing at most only a few isolated representatives of them. But under certain circumstances they may be somewhat extensively preserved not only in such parts as seem to be protected in a way from abrading influences but also on the exposed lateral surfaces. The spiny armature of the prostal paratangentials can be easily recognized when seen under the hand-lense. Diactinic prostalia, apart from those which 244 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. sometimes fringe the oscular margin, were not observed on the lateral wall of the larger specimens. The subdermal space is narrow. The variously sized apertures of incurrent canals, visible through the ectosome, may be as large as 5 mm. in diameter (PI. XVIIL, fig. 2). The interapertural choanosomal surface, usually not wider tlian the width of the directly adjacent canalar apertures, shows an interweaving of fibers or strands which run irregularly but in tlie main in oblique directions. In the proximity of the basal attachment the parenchymal texture is dense and coarsely fibrous. On the inner side of the wall, the endosome show^s a con- tinuous gastral lacework (PI. XVIIL, hg. 3) the quadrate meshes of which are distinctly visible to the naked eye. The gastral layer is supported below by a w^ell differentiated system of hypo- gastral strands which are long and of various strengths under oOO n and which by intersecting one another enclose wide meshes of an irregularly angulate shape. Frequently the strands are seen to run over and across the apertures of the excurrent canals. Interaperturally the endosome is closely adherent to the choano- some. The excurrent canalar apertures are of about the same size as the incurrent on the external side, but are somewhat more closely set together. Both the incurrent and excurrent canals are pit-like ; when seen in sections of the body-wall, they are alternately arranged, their wall showing variously sized, oval or roundish entrances into the branches. Finally, a few words with regard to the small and young specimens of the species. The smallest I have seen was only 23 mm. high, thin- walled and with an osculuni of 3 mm. diameter KHABDOCALYPTÜS VICTOR. 245 at the upper eud. The next larger is the one depicted in PI. XIX., fig. 17 ; height of body, 37 mm.; wall, 5 mm. thick in the middle ; oscuhim, 7-9 mm. in diameter. All the young of a similar or somewhat larger size are ovoid or barrel-shaped, and, besides being covered with a gossamer-like layer of prostal pent- actins, show numerous, fine, needle-like (diactinic) prostalia which spring singly from all parts of the external surface. Thus the young are, in general appearance, scarcely distinguishable from those of R. capillatus (PI. XXII., figs. 3-5). It seems the diactinic prostalia lateralia are all lost during later stages of the growth, though in some individuals similar proslals muy arise in later life but in such cases they are restricted in their distribu- tion to the oscular rim (marginalia). Spiculation. The parenchymalia principalia are bow-like oxydiactins with tapering rays which subterminally are either smooth or sparsely microtubercled. In large specimens they may attain 28 mm. in length and 400 /^ in thickness at the middle ; such coarse spicules occurring especially abundantly near the external surface in the lower part of the body. The strength of the principalia of course varies wâth the size of specimens. Close to the basal attachment I have found, abundantly in one large specimen but not in the same proportion in others, straight or bent parenchymal diactins, 8-15 mm. long and 20 /^ thick on an average, in which the center is externally marked by an annular swelling while one or both of the ends are swollen, round-tipped and thickly beset with well-developed, conical micro- 246 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. tubercles. When one end only is thus swollen (PI. XVIIT., fig. 4), as is most frequently the case, the spicular center is situated very much nearer to that end than to the other ; the spicule is quite unequally rayed, the ray which is directed towards, and almost or quite touches, the basidictyonal plate, being always the shorter and having the rounded tip. Even in this extreme basal region synapticular fusion never takes place between the parenchym- al spicules, nor between these and the basidictyonalia. The smaller parenchymalia, including the comitalia, are of the usual ajipearance and require no special description. The marginalia, present on some s])ecimens, are needle-like oxydiactins which may be 10 mm. or more in length and 40 1>- thick in the middle. The similarly shaped prostalia lalGvalia, seen only on small and young specimens, may be 20 nnn. long and 90 1>. thick. Diactins further go to compose the long liypo- gaüral strands. These are combinations of long comitalia-like diactins, 7-35 // (generally about 10 ft) thick and mostly without an annular swelling in the middle. The hypodermalia are moderatey large oxypeutactius with paratropal paratangentials. They occur in close groups, generally of 4-8 each (PI. XVIIL, fig. 16), the manner of arrangement being typically that which I have described on p. 131. There are usually in each group one or two pentactins which have entirely smooth paratangentials ; they are always the smallest and the most slender-rayed — i. e., the youngest — of all in the group. The older pentactins being always situated at a higher level than the younger, the shafts of the former obstruct the paratangentials of the latter and prevent them from developing in a regularly cruciate RHARDOCALYPTUS VICTOR. 247 disposition ; hence, the paratropism. It may be said that in each group the oklest pentactins are the most centrall}" situated ; thus, after tlieir protrusion as prostalia hiteralia ; they stand out from the center of a hypodermal group ; and when cast off, they leave behind in tliat position a compact little l)uncli of the outer ends of fine needles that accompanied their lost shafts as coraitalia. After full development, the hypodermalia have the paratan- gentials armed from base to tip with strong and sharply pointed prongs, arranged in two series along the lateral sides of the rays. The prongs are situated at tolerably regular intervals, those of the two sides alternating with one another. In the l)asal parts of the rays, the strongest prongs may be 100 ," long ; there they all spring out nearly vertically but soon become bent in a claw-like manner, the bending taking place generally either backwards or forwards, and occasionally downwards away from the dermal surface. I do not remember ever to have seen the prongs bent upwards. Towards the tip of the rays and along with the gradual attenu- ation of these, the prongs grow continually smaller, and in the terminal parts they are simply thorn-like, being directed obliquely forwards. Apart from the above prongs, the surface of the paratangentials is perfectly smooth. The unpaired shaft-ray is never pronged ; it is entirely smooth except for a few microtubercles which may be present near its inner pointed end. The prostal pentactiiis, i.e., the hypodermalia after protrusion through the dermal layer, always show^ pronged paratangentials. The protrusion evidently takes place only after complete develop- ment of the armature, a fact which seems to hold true for all members of the genus. The paratangentials are, as measured on the prostalia, generally 5.5-6.5 mm. long and the shafts, 4.5-8 248 AET. 7. 1. IJIMA : HEXACTINELLIDA, IV. mm. Thickness of mys, not more than 65/^- at base. The shaft may in length exceed the paratangential in the same spicule by about half the length of the latter ; in other cases it is only about as long or even somewhat shorter. Compared with the same spicules of certain other species [e.f/., R. capillatus) in which they form a persistent veil, the shaft rays in the present species fall behind considerably in length, a fact which may have bearing on their tendency to ])e readily cast off. The dermalia (PI. XVIII., figs. 8-Ll) are j^redominantly rough stauractins. The center of these is generally plain but occasionally shows a gentle swelling on either the external or the internal side or on both sides. Rarely the dei-malia are pentactins in which the unpaired ray is directed proximad while the distal ray is represented by a vestige in the form of a mere swelling. I have found the pentactinic forms especially along the hypoder- raal beams. Still more rarely are the dermalia diactins and tauactins, lying with all their rays in the dermal plane. In the former the suppressed rays are indicated by four knobs around the center ; in the latter the atrophied para tangential usually leaves a knob-like relic, while the ladial rays may or may not be similarly represented. Length of dermalia rays, 114-194/^ (on an average 156/^). Their thickness at base, 9*-12e /^ (on an average 10/^). Sides of the quadrate dermal meshes, on an average 180 /^ The grtstralia (PI. XVIII. , fig. 18) are rough hexactins in which the free proximal ray is not specially characterized. Length of rays, 150-230/^ (on an average 180/^) ; breadth at base, 11- 15 /^ (on an average 12è /^). Sides of the quadrate gastral meshes RHABDOCALYPTÜS VICTOR. 249 average 200 /^ in length. Sometimes two gastralia may lie close together, in which case the directly adjoining gastral laths are three rays strong instead of two as usual. Unusually small and slender-rayed oxyhexactins occasionally present in the layer are without doubt gastralia which have not yet attained complete development. Taking part in the formation of gastral laths, diactins are not infrequently found, the rays of which are generally somewhat thinner but Ioniser than those of the liexactins. The center is marked either by an annular swelling or by four cruciately dis- posed knobs ; the surface may be nearly smooth but is more generally roughened by the presence of microtubercles in varying numbers. There can scarcely be any impropriety in classing some at least of the diactins under the gastralia ; at the same time they may be looked upon as spicules linking the gastralia proper to the hypogastralia. Oxyhexasters (PI. XYIIL, figs. 5, 7 and 15) of a large size, measuring 180-280 /^ in diameter, are abundant in the choanosome. Normal forms are rather scarce ; more frequently are the oxyhexasters hexactinose and most commonly, hemihex- actinose. The principals are extremely short or nearly obsolete, making the terminals almost appear to radiate directly from the common central node. The terminals are strong, measuring up to 4 /^ in thickness at base ; their surface is always rough, the roughness developing into small barbs in the basal parts. The uniterminal principal is either straight or bent at its junction with the terminal. In the former case the atrophied terminal may some- times be represented by a unilateral boss (fig. 7). In the normal oxvhexasters the number of terminals to each 250 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. and every principal seems to be generally two, giving twelve termi- nal points in all to the entire rosette. Quite rarely I have counted as many as fourteen terminal points, in which cases I presume one or two of the principals have had more than two terminals each. The hemihexactinose forms show 7-11 terminal points. In one specimen I have noticed the presence in the subdermal space of a number of oxyhexasters, wdiich, besides being normal, had appreciably thinner rays than others in deeper situations. Probably this differentiation of the peripheral oxyhexaster obtains in all individuals of the species, but is not ahvays manifest owing to the sparseness of this kind of oxyhexasters in the periphery. The discoctaslers (PI. XVIIL, figs. 6, 12) are of about the same size, and occur in nearly equal abundance, as the oxyhex- asters. Diameter, 180-240,«. Large and small discoctasters occur together promiscuously. The central node is either plain or is supplied with the six hump-like prominences, the latter form being especially common in the case of the smaller of the rosettes. Principals smooth ; about half as long as the terminal tuft ; 5- 7h /^ thick in the middle and somewhat thicker towards the outer end. Terminals slightly rough; gradually attenuating; nearly straight or just perceptibly bent after the manner of the latter S. Their number is 6-8, sometimes as few as 4, to a tuft, which expands gently towards the outer end. Terminal discs, minute and pinhead-like ; not dentate. 31icrodiscohexasters of the usual size and shape, most nearly resembling those of E. mollis (PI. XX., fig. 4) or of E. vngui- culatus (PI. XXI., fig. 9), were found exceedingly sparsely in RHABDOCALYPTÜS VICTOR. 251 one specimeu. In other specimens I have sought after them in vain, though certainly I can not be quite sure that they ^Yere really totally absent in all these cases. As to the hasidiclyonal plate of the species (PI. XVIII., fig. 14), I have always found the structure to be thin and extremely uneven. The irregularly contoured beams are sparsely micro- tubercled. The meshes are small and roundish, oval or irregular in shape. Soft Parts. A specimen of the size of an apple, killed and preserved in alcoliol, was utilized for the study of the soft parts. However, this led to no important results. Kot a little information as to the arrangement of the soft parts may be gathered from figs. 18- 23 in PI. XIX. and from the explanations appended thereto ; and I here limit myself to recording some of the points determined, which are as follows : Chambers, 120,« in average diameter. j\Ieshes in their walls open, generally 3-G /^ wide. Xuclei, about 2 /-« large in diameter ; rather distinct though stained in about the same degree as the reticulum-forming protoplasm ; containing one or more darkly stained granules. At the chamber-rim (fig. 19), as also along the termination of the chamber-layer close to the oscular edge (fig. 18), the finely reticular wall passes into film-like or filamentous trabeculfe without any sharp demarcation. In the thin oscular margin the chamber-layer assumes the character of a continuous, irregularly undulating, reticular membrane, instead of being dif- ferentiated into separate chambers (fig. 18). 252 ART. 7.— I. IJIMA : HEXACTINELLIDA, IV. Onseveral occasions I have distinctly observed the presence of a film-like memhrana reuniens (Contrib. I., p. 130) filling up the gap between the circular apopyles of three or four, directly adjoining chambers (fig. 20, /.), thus shutting off at the spot the iutercameral incurrent space from the excurrent lacunse of the internal trabecular layer. At other times, however, the gaps were clearly seen to be open (fig. 20, g.), so that I believe the membrana reuniens is not a thing of constant occurrence. Dermal and gastral membranes, rather extensively film-like (PL XVIII., fig. 16). Trabeculœ, thread-like ; rather sparse in the subdermal space as well as along the lumen of both incurrent and excurrent canals. Their nuclei, at most 2 ,« in diameter ; scattered at irregular intervals. Archœocytes, found in small flat groups on the wall of chambers ; not forming large congeries, which fact is probably due to the immature state of the specimen. Scattered in irregular distribution and suspended on the trabeculœ, both external and internal, are found numerous fat- like spheres of various sizes (shown in figs. 20-22 as black dots). Diameter, 72-20/^. Sometimes they seem to be inclosed in a thin envelope ; more often they are apparently quite naked. Their substance is either homogeneous or granular ; it is intensely stained by carmine or hsematoxylin, and is browned by osmic- acid but is not coloured blue by iodine. Neither alcohol nor ether dissolves it. I have therefore no doubt as to the bodies being thesocytal products. Frequently two spheres are seen closely apposed to each other, as if they had taken origin and were actually lying together within one and the same cell, — which is probably the real fact, though the nucleus can not be distin- o'uished. RHABDOCALYPTÜS MOLLIS. 253 Together with the above bodies or in places where one might expect to find them, there not nncommonly occur roundish, oval or irregularly contracted cells of pale appearance and each con- taining a darkly stained nucleus (fig. 20, c). They measure 7i- 12/^ or more across. An enveloping membrane is distinguishable; the contents are thinly granular. I take these cells for old thesocytes, which have lost the fat-like products by consumption. RHABDOCALYPTÜS MOLLIS F. E. Sch. PL XX. Rhabdocalyplus mollis. F. E. Schulze, '86, p. 51 ; '87 (!), p. 155, PI. LXIV ; '97, p. 552.— I. Ijima, '97, p. 50.— Ch. Gkaviek, '99, p. 421. The specimens I have seen of this species number not less than a dozen. Most of them belonged to Mr. Alan Owston. That all came from the Sagami Sea there can be no doubt, but only in the following three cases can the localities be more exactly stated: Homba (about 572 m. depth), where Kuma obtained two very small specimens together with R. capillalus ; off Odawara in the Province of Sagami (about 500 m.), at which locality I myself obtained a fragment attached, together with a number of other animals, to the carapace of a JIacrocheirus kcempferi ; and the " Albatross " Station 3697 (about 4 kilom. off the mouth of Sakaw^a River in Sagami, 265-120 fms., black volcanic mud), where a tubular fragment of the species was trawled up by the " Albatross " (1900) together with Acanthascus cactus. 254 AKT. 7. — I. IJIMA : HEXACTINELLIDA, IV. Nearly all of the specimens I have seen bore on the basal attachment samples of the bottom which consisted of a fine- grained, tufaceous clay. I may begin the description of tlie general characters with the two little individuals (S. C. M. No. 421) mentioned above as havinir been obtained at Homba. Both are of about the size of a pea and are torn off at the base. They are both ovoid in shape and show a small round osculum. They possess some fine prostat needles together with a few prostal pentactins and are thus indistinguishable in outward appearance from the young of R. caplllaim (PL XXIL, figs. 3, 4). The sj)ecies attains rather large dimensions. The individual shown in PL XX., fig. 1, wliich is one of the largest I have seen, measures 337 mm. in height and 6-8 mm. in thickness of the w'all at most parts ; that of fig. 2, same plate, 366 mm. in total height and 10 mm. in thickness of the wall in the inferior half of the body. In general all the larger specimens are of a saccular, funnel- like or vase-like shape, gradually contracted towards the knobby base and more or less distinctly compressed in a lateral direction. In the specimen of PL XX., fig. 1, which, so much of it as is preserved, retains the natural shape in a perfect state, the osculum at the upper end is roundish measuring approximately 82 mm. in diameter ; but a third of the height lower down, the body is sagittally 153 mm. and transversely 94 mm. broad; and still lower, the greater breadth is 82 mm. and the lesser, only 30 mm. The species seems to be in a high degree prone to form secondary oscula ; in fact all the large specimens I know of are in possession of one or more such in addition to the primary or main osculum. They may be represented by simple perforations RHABDOCALYPTUS MOLLIS. 255 Text-fig. 9.—Ehabdo- calypius vwllis F. E. Sch. (O. C. No. 104). Drawn in } natural size. Ol., main oscnlum or oscnlnm of the mother person. O2., oscuhnn of the first-formed daugliter person. O3., oscuhim of the second-formed daugh- ter person (injured). Text-fig. 10.— Ehabdo- calyphis mollis F. E. Sch. (O. C. No. 105). Drawn in J natural size. O1-4., oscnla num- bered in the order of their formation, as in the preceding figure. 256 ART. 7. r. IJIMA : HEXACTINELLIDA, IV. but are more generally situated at the end of tubular outbulgings or buds of the sponge-wall. These are at first csecum-like but after a time an osculum opens at the blind end of each ; they, as also the simple secondary oscula, should be considered to repre- sent daughter-persons formed by the mother-person, ill- defined and persistently continuous with the latter though they are. The bud may grow to conspicuous dimensions and thus may give to the sponge a very peculiar and characteristic shape. It seems to be invariably the rule that the simple secondary osculum or the tubular daughter-person, whichever be the case, arises at some point on one of the median edges— not on the broad lateral sides — of the laterally compressed mother-sponge, and that, when two or more secondary oscula or daughter-persons coexist, these are all situated in a row on the same body-edge. My material comprises a series of specimens varying in the number of daughter-persons from one up to five. The two specimens of PI. XX., figs. 1 and 2, show each a single daughter-person. In the one (fig. 1) the wall is somewhat bulged out sagitally in the upper half of one of its obtuse edges. At the lower end of this out-bulging is an artificial gap in the wall, produced without doubt by the severing off of a daughter- person which had grown there. Assumably the upper part of the same out-bulging is in an inceptional stage of giving rise to a second daughter-person. The other specimen (fig. 2), a strongly laterally compressed individual, shows a conspicuous tubular daughter-person, 183 mm. long and springing from the upper part of the mother-wall in an obliquely upward direction. In this case we may presume, from the circumstances of space, that a second daughter-person, were it ever to arise, would be formed below the one already present. In the third specimen I have RHABDOCALYPTUS MOLLIS. 257 seen with a single tubular daughter-person, this constituted a very conspicuous feature in that it was somewhat longer, though certainly less broad, than the mother-body. It arose from near the base, so that the formation of a future bud w^as to be anticipated most likely in a position above the daughter-person just referred to. That the second bud may be situated either above or below the first formed, is established by actual cases to be mentioned diiectly. In text-figure 9 I have represented a specimen with two secondary oscula in addition to a large primary osculum (0,). Of the former, the larger one {0.>) is situated at the end of a broad branch from the mother-body ; it is evidently the older. The other (OJ — the smaller and younger— is situated in the axil between the branch-like daughter-person and the mother-body ; or, it would be proper to say simply that in that position is situated the second formed daughter-person which seems to have suffered much mechanical injury. The general shape of the sponge has been suggestively described by Mr. Alan Owston as like " a hand with thumb extended." Another specimen of essentially the same shape existed in the collection of the gentleman just mentioned. In it the second daughter-person was well-preserved in the form of a tube, smaller than the first da ugh ter- person and situated between the origin of this and the oscular margin of the mother-person. Here a reference may be made to the two specimens which are mentioned by F. E. Schulze in the Challenger Report. Tlie smaller of them is said to have exhibited near the lower end two small roundish apertures. I consider it probable that these apertures were secondarily formed oscula or at any rate some sort of gaps indicative of daughter-persons. The other specimen. 258 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. which is figured, should have possessed on one side of the lower part a tube-like branch opening in a wide orifice at its extremity; directly below it a small cœcal protrusion or boss ; and just above the solid base a small round perforation in the body-wall. What- ever may be the nature of the last mentioned perforation, there can be no doubt about that of the tubular branch open at the end and of the ctecal boss. The latter w^ould after a time have developed an osculum at the blind end, like the older bud just above it. The specimen of text-figure 10 bears three daughter-persons. The lowest, of a long tubular and chimney-like appearance, is without doubt the first formed (O2). As the second formed I consider the uppermost, a great part of which had evidently been torn off and lost, leaving an irregular gap in the w^all (O3). Between the above two is the third tubular bud (O4) which is open at the end and is of a comparatively small size ; this I regard as the one last formed. Another specimen was essentially similar to the last mentioned except in the fact that it showed a distinct trace of the fourth bud which had been torn off. If it be justifiable, as in a great measure I think it is, to infer the order of successive formation from the relative caliber of the buds, then the lowest situated and by far the largest of them all is the oldest ; the one directly above it, the third ; next above comes the second ; while the uppermost is the fourth and last formed. A specimen with five well-preserved buds is distinctly bent in the sagittal plane. On the body-edge which formed the greater curvature the tubular buds are arranged in a row, after the manner of glove-fingers. Counting from below upwards, the first (the lowest) bud is very long, cbimney-like and open at the end ; this is probably the oldest daughter-person. The second resembles RHABDOCALYPTÜS MOLLIS. 259 a tube only about half as long as the first and still closed at tlie tip. The third is a long tube, broader but slightly shorter than the first and open at the end ; this I regard in point of age as next to the oldest daughter-person. The fourth is a relatively short blind tube like the second. Finally the fifth, i. e., the uppermost situated near to the margin of the main osculum, is a tube of about the same size as the last but open at the end ; it probably represents a daughter-person younger than the third but older than either the second or the fourth, both of which terminate iu a cul-de-sac indicative of their incomplete develop- ment and may have originated nearly simultaneously. Thus far concerning the peculiar external shape which the species may acquire as the result of budding out tubular daughter- persons. Now, the gastral cavity of the mother-person — it is scarcely necessary to say — is directly continuous with the like cavities of the daughter-persons. It is very deep and extends close to the basal attachment, so that a solid stalk can certainly not be attributed to the species. Externally, the circumference of the basal attachment may show irregular pad-like thickenings. The oscular margin, which is always thin, is generally simple- edged in the daughter-persons. That of the mother-person may, on the other hand, show a varying number of fine, short prostal needles, projecting upwards to a length not exceeding 5 mm. These partly springing from the very edge and partly from the dermal surface adjoining it, may form an inconspicuous and interrupted fringe to the osculum. That diactinic prostalia occur with comparatively greater prominence in the young stage of the sponge, may be concluded from tlie condition presented by the two very small individuals before mentioned. 260 AKT. 7. — I. IJIMA : HEXACTINELLIDA, IV. As in R. victor the prostal pentactiiis seem to fall off readily. Thus, in the specimen of PI. XX., fig. 1., in wliich the external surface is in an almost perfect state of preservation, the pent- actins, protruded to an extent of 3 or 4 mm., occur isolatedly and sparsely in certain parts only, while over the greater part of the sponge they are totally missing. The delicate dermal lacework is thrown into little creases, possibly as the result of desiccation. Observed under the lens the meshes appear to be irregtdar and indefinite in shape, which is due to the fact that the dermalia are diactins that cross, or are joined to, one another at various angles (PL XX., fig. 12). The thin and wavy hypodermal strands, in places so thin as even to closely resemble tlie dermalia in strength, form also irregularly angulate, but of course much larger, meshes. They are seen, sometimes with a fair degree of distinctness, to converge towards separate central points which are 3-6 mm. apart. This is on account of the pentactinic hypodermalia being arranged in groups, much as we have seen them in R. victor. In the lower parts of the sponge, especially near the base, the centers of the hypodermalia groups are usually well indicated by short small projecting tufts of fine needles which either cohere or are loose and brush-like. The same tufts we have likewise seen in R. victor. The prostal pentactins that may occur spring out in connexion with the tufts. In one specimen I have seen on the dermal surface several cicatrice-like spots which appeared white and densely textured. The same have sometimes been noticed in several other Acan- thascinse. It is more than probable that they arise by hyper- regeneration of tissues, both soft and scierie, at places where the sponge had suffered an injury. RHABDOCALYPTUS MOLLIS. 261 The internal surface of the sponge-wall is covered tlirougliout with a continuous gastral lacework, in which the meshes, for the most part regularly quadrate in shape, can be discerned with the naked eye (PL XX., fig. 13). Beneath the layer are seen hypogastral strands varying in thickness, running in indefinite directions and intersecting one another at various angles. The canals, both incurrent and excurrent, are deep and pit- like. In the larger specimens, the roundish canalar apertures of the incurrent system may be as large as 4 or 5 mm. in diameter, while those on the gastral surface measure at most 2 mm. The latter are thus on the wdiole smaller than the former but are situated more closely together (PI. XX., fig. 2). On cross-sections of the sponge-wall, the canals of the two systems do not alternate so regularly as in certain other species in which they are of ap- proximately the same caliber. Spiculation. With F. E. Schulze's excellent descriptions of the spicula- tion of this species and with two slide preparations from the type specimen which he gave me, it was an easy matter for me to recognize the species. The principal parenchymalia are bow-like oxydiactins which may be 20 mm. long and 90 /^ thick in the middle but are more generally smaller. The middle is not externally marked by swellings. The rays are gradually attenuated toward the end. The smaller parenchymalia, down to comitalia of only 4 /^ thickness, are of the usual description. A central nodal thickening 262 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. is by no means of such frequent occurrence as one might be led to suppose from F. E. Sciiulze's statement in the Challenger Report. The tips are acuminate, rounded, conical or mucronate ; occasionally swollen to a club-like or even a bulbous shape. Sub terminally rough in varying degrees ; the microtubercles, when strongly developed, are retroverted. The parenchymalia undergo synapticular fusion close to the basal attachment and in conjunction with the basidictyonalia. In certain specimens the fusion was limited to the central parts of the base ; in others it extended to the external surface. The basididyonal plate is at places represented by a very thin reticular layer quite agreeing in appearance with F. E. Schulze's pi. LXIV., fig. 3, in the Challenger Keport. At other places it is of a considerable thickness, being composed of basidictyonal hexactins ankylosed together in the usual manner. The hypodermal pentactins show essentially the same char- acters and arrangement as in R. victor. Paratangentials, not exceeding o mm. in length ; the shaft, longer, up to 10 mm. The former are more or less paratropal, but cases of their being quite or nearly regularly cruciate are not infrequent. This may be explained by the fact that the hypodermalia do not lie close together in their groups (PI. XX., fig. 12) and consequently their shafts run a certain distance apart from one another, thus depart- ing in a measure from the condition that causes the paratropism in a developing hypodermal pentactin. Not uncommonly four or five pentactin-heads are seen in a group, of which the most superficially situated {i. e., the oldest) one or two have the four paratangential rays provided with the spiny armature. The spines RHABDOCALYPTUS MOLLIS. 263 are somewhat less developed, with respect both to their size and number, than in R. victor. They are directed obliquely outwards, the stronger ones in the basal parts of the rays being curved generally likewise outwards but occasionally in some other direc- tion, in a claw-like manner. Though arranged as a rule in two irregular series on the lateral sides, there may occasionally exist a spine or spines on the external side of the ray, especially in the basal parts. The spines on the two sides may alternate with tolerable regularity, but this is subject to frequent interruptions. Those on the same side are situated at intervals which frequently may measure 160/^ or thereabout. In some cases — not all — of the spined hypodermal pentactins, I have found the general surface of the paratangentials to be finely shagreen-like on account of the presence of dense micro- tubercles, similar to those in certain Staiirocalyptus species. In this, as also in several other features of the spiculation, the present species stands in very close agreement wàth R. avslralis Tors. A number of slender diactins, indistinguishable from paren- chymalia of a similar strength, associate with the paratangentials in forming the hypodermal strands. A number of comitalia accompany the shafts. The dermalia (PI. XX., fig. 12) are generally rough and rod-like diactins, much like those given in PL XXL, fig. 2 (from R. unguiciilatus) . Earely and exceptionally are they orthodiactins, tauactins, stauractins or pentactins. According to F. E. Schulze monactins should occasionally occur also. The diactins, which are either straight or very slightly arched, are 265-485 /^ long as measured from tip to tip. Breadth 264 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. in or near the middle, 9-15/^. The spicular center is generally plain ; otherwise it is externally marked, sometimes by an annular swelling and sometimes by knobs which may occur either in a single pair or in two pairs. The rough-surfaced rays generally taper slightly tow^ards the tip which is rounded off or conically pointed. The gastralia (PI. XX., fig. 13) are hexactins with rays appearing exactly like those of the dermalia. Length of rays, 165-245/^ (about 200/-« on the average). Breadth at base, 11- 15/^. All the six rays in the same spicule may in general be said to be nearly equally long, though in some cases the distal ray has been found to be somewhat shorter, and the free proximal ray to be somewhat longer, tlinn the paratangential rays. The microtubercles may be slightly more pronounced on the proximal ray than on any other, but in any case the differentiation of that ray is never carried out to any considerable degree. The unusually small and slender-rayed gastralia, which are occasionally met with, are apparently those that have not yet attained full development. The oxyliexasters occur much more abundantly in the deeper parts of the wall than in the periphery, as has been pointed out by F. E. Schulze. Diameter, 102-160 i"; usually about 120 /i. The microtubercles on the basal parts of terminals show the tendency to develop into barbs in an unusual degree. One point concerning the spicules, which has not been mentioned by F. E. Schulze, is the fact that they occur in two slightly differing forms, the one chiefly in the subdermal space and the other in more deeply situated parts. It must however be said at once that the two varietal forms intersirade. RHABDOCALYPTUS MOLLIS. 265 The subdermal oxyhexasters (PL XX., fig. 3) are distinguished by being nearly always of the normal shape and by having very slender terminals. Each exceedingly short principal in a spicule carries 2-4 (usually 3), nearly straight, rough terminals. These seem to be very fragile ; the broken off terminals are usually found in numbers sticking to the dermal membrane. The rough- ness of surface becomes more pronounced towards the base of the terminals, where it may distinctly appear to be due to reverted microtuberclcs. Under certain circumstances these may be de- veloped even into long and conspicuous barbs, but such is by no means the case with the generality of the peripheral oxyhexasters. Karely is the variety in question hemihexactinose or hexactinose. Deep in the choanosome it is only occasionally met with pro- miscuousl}^ with the following. The second variety (PI. XX., figs. 0-11) is by far the more numerously represented, occurring in greatest abundance in the subgastral region as well as in the endosomal layer. In it the principals are reduced almost to nothing, so that the terminals appear to radiate directl}^ from the central node. The terminals are — in some individuals of the sponge slightly and in others decidedly — stronger than in the other variety. In some exceptional cases of the spicule I have found all the rays nearly^or quite smooth from tip to base ; but the general rule is that their basal parts show^ a varying number of barbs or unusually strongly developed retroverted microtuberclcs, while the distal parts are either obsoletely rough or quite smooth. Sometimes the basal barbs occur in quite a limited number ; it may even happen that in the same spicule some rays are provided with only a few of them while the rest are entirely smooth. As a general rule, however, the b;irbs are numerous, gradually increasing in length 26G ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. towards the very base of the ray to which they belong. As has been known from F. E. Schulze, the more basally situated barbs may sometimes be so long that those of adjacent rays almost unite. In a few instances I have seen some of the barbs of the different rays actually in fusion, thus producing around the central node a structure which bears a certain resemblance to a lychnisc. The total number of rays in an oxyhexaster of the second variety may be twelve or even thirteen or fourteen. These evi- dently represent cases of normal oxyhexasters, in which the six, nearly suppressed principals are either all biterminal or are partly l)iterminal and partly triterminal. On the other hand a large, if not the greater, number of the oxyhexasters in question are apparently hemihexactinose, leading down to the quite hexactinose forms which are also numerously represented. Both the reduced oxyhexasters just named are of the usual general shape and need not be specially described. But I should mention that again in this species I have encountered several instances of a principal bearing divergently a well developed terminal and the spurious rudiment of a second (PI. XX., fig. 10). The rudiment may occur as a small unilateral spine even in the cases in which the principal and the persistent terminal have straightened out as a simple ray, not bent at the base (PI. XX., fig. 7). F. E. Schulze has found in some numbers peculiar oxy- hexasters— reduced forms with six or less rays in all — in which all or some of the rays present are spirally twisted (Chall. Rep., PI. LXIV., figs. 10 and 11). A similar oxyhexaster was noticed by me in Staurocalyplus pleorhaphides also (p. 229, PI. XVI., fig. 8). As regards the present species, certain specimens indeed were found to possess the twisted oxyhexasters though sparingly, but in several others I have searched for them in vain. I am there- RHABDOCALYPTÜS MOLLIS. 267 fore disposed to consider them as of inconstant occurrence in the species. Possibly they are produced only under certain abnormal conditions. The discodasters (PL XX.,. fig. 5) are found scattered in the wall, but in especial abundance in the subdermal space as was noted by F. E. Schulze. I may here note that contrari- wise in R. capillaius the same spicule is most numerous in the subgastral region. What causes such a dissimilar distribution can certainly not be told. In diameter the spicules in question vary from 130,« to 175 /j- (on an average 140/^-). They are nearly as large as, or but slightly larger than, the average oxyhexasters of the species. The central node exhibits the six hillock-like prominences more or less distinctly. The principals (secondary) take up about one-third or less of the entire ray-length ; they are 5 n or more thick iu the middle. Number of terminals in a tuft, 5-9 ; rarely as few as 3 or even 2. The tuft is gently expanded distally, the terminals composing it being each slightly bent outwards or, as is sometimes the case, nearly straight. The surface of terminals is slightly rough ; under the immersion system the roughness may be seen to be caused by reverted microtubercles. The terminal discs are either simply pinhead-like or show seven or eight marginal teeth, the latter being the case in the larger discoctasters. Not infrequently are the discoctasters malformed in that some terminals remain free without being incorporated into any of the secondary principals. Noteworthy seem the few instances that came under my observation, in which the terminals stood out in eight tufts directly from the tubercled central node, without coming into fusion in the basal parts. 268 ART. 7. 1. IJIMA : HEXACTIXELLIDA, IV. The microdiscohexasters (PI. XX., fig. 4) are of the familiar appearance. Diameter, 22-27 /A F. E. Schulze found these spicules in especial abundance in the proximity of the dermal membrane. In the preparations that he sent to me I find them quite common in the endosome i.s well as in the subgastral region, once counting as many as a dozen in the same visual field of the microscope. Far less commonly and even quite sparsely were they represented in several specimens in which I have specially looked for them. In these they were to be met with only now and then, both subdermally and subgastrally. RHABDOCALYPTUS UNGUICULATUS nov. sp. PI. XXI. The species here described as new under the above name is unquestionably a very near relative of R. mollis F. E. Sch. and of R. mirabilis. From both these species it can scarcely be said to differ so far as the categorical forms of the spicular elements are concerned ; but in the details of characters I find, common to all the individuals referred to it, certain constant peculiarities which, I think, may be considered to be of specific value. I have had at my disposal for examination no less than eight specimens, all of which I place under the present species. The known locality is Okinose, both Inside and Outside ; depth, about 500 m. Three of the said specimens are small, young individuals, quite similar in external appearance to the young of R. victor KHABDOCALYPTUS ÜNGUICULATUS. 269 (PI. XIX., fig. 17) or of R. capiUalus (PI. XXII., figs. 3-5). Two of them grow together on a mass of coal-cuider in company with the hirge and beautiful specimen sliown in PI. XXI., which will soon be described in detail as the type of the species. A fifth specimen (S. C. ]\I. No. 473) shows a tubular shape, 58 mm. high. A sixth (S. C. INI. Xo. 402), 100 mm. high, is much macerated ; it is attached to a branch of Ceratoisis sp. together with a small R. victor. Finally two more specimens, which belonged to a collector, came under my inspection ; they were both tubular and somewhat fusiform, measuring respectively 115 mm. and 150 mm. in height. These two also grew on a piece of coal-cinder, erect and side by side. In the gossamer-like covering and in other points of the macroscopic characterization, they agreed w^ell with the type-specimen. In no specimen have I seen any indication of the formation of secondary oscula or daughter persons. The type-specimen referred to above (PI. XXL, fig. 1 ; S. C. M. Xo. 501, from Outside Okinose) is an exquisitely preserved individual of the shape of a thick-walled, somew4iat laterally compressed, tubular sac, gently swollen out on the sides. It is attached to the substratum by an irregular basal mass, situated on one side of the lower end and as usual in the sagittal plane of the body. Total length, 365 mm. Breadth in the middle (ex- clusive of the layer of prostals), 140 mm. sagittally and 120 mm. transversely. Thickness of wall in the same position, 21 mm.; it is much greater than in specimens of R. mollis of approximately the same size. The oval osculum at the upper truncated end, 72 mm. by 107 mm. in diameter. In the peristomal region there exist numerous, fine, needle- like oxydiactinic prostalia of variable length. They may project 270 ART. ".— I. IJIMA: HEXACTINELLIDA, IV. to the length of 14 mm. or more. They spring out partly isolated and in irregular distribution but more usually in tufts from the top of little tubercle-like prominences of the dermal surface. Within a centimeter's distance from the oscular edge the oxy- diactinic prostalia become entirely replaced by large oxypentact- inic ones, which cover the rest of the external surface down to the basal mass. All the pentactins arise in tufts from the summit of small tubercle-like or papilla-like prominences, which are always of small diameter but may sometimes be 2-3 mm. high. Thej are usually separated from one another by a space of 5-10 mm. In many cases there are only two or three pent- actins in a tuft ; in other cases there may be more than ten. The paratangential heads in a tuft do not all lie in the same level but usually rise to various heights from the dermal surface, representing different stages of the protrusion of the shafts from the latter. Thus, there arises the gossamer-like layer, in places 15 mm. thick, over the sponge exterior. The conditions of the prostalia are exactly the same as in E. capillatus {cfr. PI. XXTL, fig. 12, etc). On the basal mass there bristle out numerous hair-like prostals. These, on close observation, were found to be simply the shafts of such prostal pentactins whose distal parts had been broken off. The dermal surface is rather uneven. Under the lens it shows the dermal layer to be irregularly meshed, which is due to the diactinic dermalia. The comparatively fine hypodermal fibers exhibit a certain regularity of arrangement in so far as they appear to radiate from the base of each prostal tufts. In short, the structure of the ectosomal skeleton is essentially the same as is known to me from E. mollis and E. capillaius. RHABPOCALYPTUS UNGüICüLATUS. 271 The subdermal sj^ace presents itself as a rather conspicuous layer in sections of the wall. In places it is as much as 3 mm. across. It is traversed by the pillars which join the ectosome to the choanosome and consist mainly of the shafts of prostal and hypodermal pentactins and of their comitalia. The appearance of the wall in section partakes much of that which I have figured from R. capUlalus in PL XXII. , fig. 12. As to the roundish apertures of the incur rent canals a certain différence is noticeable with respect to their size and manner of distribution on the two lateral sides. On the side shown in PL XXL, fig. 1, some of them measure up to 7 mm. in diameter, and lie promiscuously with others which are at most about half, and often much less than half, as wide. On the other side the apertures are of a more uniform size, mostly measuring 3-5 mm. in width and situated somewhat more closely together. The gastral surface is covered over by a continuous gastral lacework, the regularly quadrate meshes of which are plainly visible to the naked eye. They are much larger than in R. mollis or R. capillalus, measuring i-h mm. in length of sides. In the interspace between the excurrent canalar apertures, the layer is in tolerably close contact with the choanosome; over the apertures it frequently heaves up in a vault-like manner. Kemarkable is the fact that even over the largest apertures there are seen no hypogastral strands in suj^port of the gastral layer. The excurrent canalar apertures in most parts of the gastral cavity are rather closely set, reaching up to 5 or 6 mm. in diameter. In the bottom of the cavity some of them are as wide as 8 mm. 272 ART. 7. — I. IJIMA : HEXACÏINELLIDA, IV. Spiculation. The following account is based on my observations on the type specimen, except when other specimens are specially referred to. The principal parcnchymcdia are large oxydiactins which may measure 30 mm. in length and 175 p- in breadth in the middle. They are generally more or less bent and gradually attenuated towards both ends, which are rough for a short distance even in the largest of the spicules. The accessory parenchymalia present no point that seems to require special mention. The prostal oxydiacti7is of the marginal zone may attain a length of 25 mm. and a breadth of 85 /z in the middle. They are smooth throughout, without any indication of the spiny character of the distal parts, such as has been observed in H. mirabilis. The prostal oxypentactins have paratangentials which are nearly always paratropal and are 8-11 mm. long and about 85/^ broad near the base. The spines on the paratangentials may be as high as 150 /a Basally on the rays they are rather irregularly disposed but more distally are arranged in two lateral rows. The smooth shaft-ray generally measures 15 mm. or over in length. On the basal mass of the sponge it may be as long as 40 mm. With respect to the arrangement of the same oxypentactins as hypocJe7'malia and of the younger spineless hypodermal oxy- pentactins, what I have said under R. victor may be considered to apply equally well to the present species. EHABDOCALYPTUS UNGÜICÜLATUS. 273 No cliactins seem to occur as hypodernialia, unless those tliat occasionally extend from the choanosome into the ectosome or unless those which will presently be mentioned as unusually elongate dermalia be regarded as such. The dermalia (PI. XXI., fig. 2) are nearly exclusively diactins and only occasionally stauractins or tauactins. The rays are rough all over and slightly tapering towards the rounded or obtusely conical end. The diactins are nearly straight or only perceptibly arched. In them the center is usually not marked by any swelling ; only occasionally it is indicated by tw^o opposite lateral tubercles, never by four such in ci'uciate disposition, — a fact which suggests the direct derivation of the diactins from stauractins and not immediately from either pentactins or hexactins. The diactinic dermalia generally measure 330-420 ,« (about 400 /^ on an average) in total length and 8-15 !'■ (about 10 <" on an average) in thickness near the middle. Rarely there occur, among the dermalia, di- actins, 800 /^ or nearly 1 mm. long. With such an increase in length of the dermalia, the roughness of the surface shows a tendency to become confined to the terminal parts, leaving the middle parts smooth. The gastralia (PI. XXL, fig. 3) are rough oxyhexactins of much greater axial length than the dermalia. They are con- siderably larger than the same spicules of R. mollis. The radial axis is 640-850/^ (on an average about 780/^) long. Length of the free proximal ray, 440-550 (j- ; that of the distal ray, 230- 300 /^ ; that of paratangentials, 275-330 /a Thickness of rays near base, about 15/^ on the average. The proximal ray, which is not only the longest but generally also the thickest of all in 274 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. the same spicule, shows no swelling in its course but gradually tapers like the rest of the rays, whereas in II. mirabilis the same ray should be thickened in the middle. The microtubercles on that ray are more numerous and somewhat more strongly developed than on all the other rays on which tliey appear rather sparsely. The oxyltexaslers, which occur abundantly in all parts, measure 130-100/^ in diameter. Those situated in the periphery of the wall are mostl}" normal, having usually two and occasionally three, slender (about 2 /^ thick at base) and obsoletely rough terminals to a principal, the latter being very short (PI. XXI., figs. 4 and 5). It seems that the terminals are very easily broken off at base, for it is seldom that one meets with an oxyhexaster of this variety in a perfectly intact state. In the deeper parts of the wall the oxyhexasters change in their character, though in general size there exists no appreciable difference (PL XXL, figs. 6-8). Thus, in the subgastral region as well as in the endosome they are scarcely ever normal but are nearly always hemihexactinose and occasionally regularly hexactinose. Moreover, the terminals are quite smooth throughout and are considerably stronger than in the peripheral oxyhexasters. — sometimes fully twice as thick at base as in the latter. The exceedingly short principals bear at the most two terminals. The discodasters (PI. XXL, figs. 10 and 11) resemble in shape and size those of E. mollis or R. mirabilis. They are found more abundantly in, as well as directly under, the ectosome, than in the deeper parts. Diameter, 175-190 /-«. The six bosses on the central node are sometimes distinct and sometimes not. RHABDOCALYPTUS UNGüICULATUS. 275 The principals are 20-25,« long as measured from the spicular center, — in any case they are much longer than in E. mirabilis ; they are of uneven contour and slightly swollen towards the outer end (about 10/^ tliick). The terminals number 7-12 to each principal and form a rather broad, lily- like tuft, expanded at the outer end to a breadth of about 60/a Each single terminal is nearly smooth, rather strong, thickens slightly near the distal end and terminates with a small disc of a characteristic shape. This disc distinctly shows some recurved marginal teeth, which are however not uniformly developed, but are longest and strongest on that side of the disc which is turned away from the axis of the perianth-like tuft of terminals. On the opposite side of the disc, I.e., the part of the disc circumference nearest the axis, the teeth are smaller and often obsolete or even totally suppressed (fig. 11). Thus, the disc approaches or fectually assumes e?i miniature the shape of the unguiculate terminal plate in an Euplectellid flori- come. The teeth, when present all around, number six, but there may be only the three or four on the external side of the disc. In width the disc may measure about 7/^, of which nearly 4/^ constitute the length of the longest tooth. 31icrodiscohexasters (PL XXL, fig. 9) of 22-30 />« diameter are sparsely distributed in the dermal membrane. They are of the usual description. And now a few words about the hasidictyonal plate (PI. XXL, fig. 12). This consists as usual of a thin limiting plate, inside of which are found some small but thick-rayed hexactins in fusion with it. The uneven and irregularly cril)rate limitino- plate shows roundish meshes, generally under 35 /-« in width ; its 276 AKT. 7. — I. IJIMA : HEXACTINELLIDA, IV. beams, 25 /^ or less in tliickness, are sparsely microtubercled and contain, here and there in the nodes, axial canals in the form of a plane cross. The basidictyonal hexactins are likewise sparsely microtubercled. There can be scarcely any doubt that these are the same as those which F. E. Schulze ('99, p. 63) discovered forming a rigid framework in the buds of R. mirabilis. Finally it should be stated that the spicules, as might be expected, are subject to certain variations according to the age or the individuality of the sponge. Thus, in a young specimen not larger than a nut and attached to the same piece of coal- cinder as the type specimen, I find several of the spicules considerably smaller than those mentioned above. They measure as follows : Paratangentials of prostal pentactins not over 5 mm. in length ; dermalia 275-330 /^ (on an average 320 /^) long and generally about 7? /^ thick in the middle ; proximal ray of gastralia 330-385 /^ long ; paratangentials of same 187-240/^ long; diameter of discoctasters 143-155 /^ ; principals of same about 19 /-« long ; etc. lu a certain specimen I have found the terminals of oxyhexasteis sparsely microtuljercled at the base, instead of being entirely smooth. RHABDOCALYPTUS CAPILLATUS Ij. Pis. XXir. and XXIII. lUiabcIocalyplud capiUatus. I. Ijima, '97, p. öl. Of this species some sixteen specimens of various sizes have thus far been examined bv me. The known localities are : EHABDOCALYPTUS CAPILLATÜS. 277 Maye-no-Yoclomi ; Mochiyama ; Inside Okinosé by the Ena-line ; Outside Okinosé by the Iwado-line ; and Homba ; — all in the Sagami Sea. Depth, between oOO and 572 m. (274 and 313 fms.). The samples of the bottom found attached to some of the speci- mens are tufaceous. In one case six small specimens were found seated on a dead Ckonelasma calyx from Maye-no-Yodomi. An- other dead Chonelasma calyx from Mochiyama bore three young individuals of the species together with two of Staurocalyptus glabcr. The smallest specimen I have had was globular in shape, measuring only about 3 ram. in diameter. It already possessed a small osculum, while from its surface there emanated, almost radially, a number of fine prostal diactins that stood out isolated. These were the only prostalia to be seen on it. The next larger specimen, shown in natural size in PL XXIL, fig. 3, is bcirrel-like ; it shows in addition to prostal diactins some small pentactinic prostalia. But these are as yet quite few in number. Young specimens (figs. 4 and 5), still larger but under 30 mm. or thereabout in height, may be said to retain the barrel- like shape, the cross-section of the body being of an approximately circular outline. Both the diactinic and the pentactinic prostalia have greatly increased in number. The former are more abun- dant in the upper parts of the body than in the lower, being most numerous in the marginal zone around the circular osculum. The latter are present all over the body, mostly arising in tufts from slight elevations of the dermal surface. The general appear- ance of young individuals closely resembles that of R. victor of about the same size, so that in their cases the microscopic examination of the spicules is quite indispensable to insure correct identification. 278 ART. 7. — I. IJIMA : HEXACTINELLIDA, IV. After attaining a moderately large size, the body is poncli- like, tubular or vase-like, and laterally compressed to a greater or less degree. The wall is moderately thick. The deep gastral cavity extends close to the basal attachment. PI. XXII., figs. 1 and 2, and text-figure 11 will serve to give a fair idea of the external appearance of what I consider to be quite or nearly mature specimens. FW. 1 shows a specimen — from Inside Okinose by the Ena- line — of a strongly laterally compressed pouch-like shape. Height, 97 mm. Breadth sagitally, 40 mm. at the attachment base and 77 mm. in the broadest part. Tlie osculum is slit-like and com- paratively small, occupying a position at one end of the upper body-edge, the other end of whicli forms a part of a rounded outbulo-ing of the wall. This outbulging shows a small perforation at the top besides a thinning-out of the wall at two places. It contained in the internal cavity a cluster of Cephalopod eggs in which the embryonal development was still in an early stage and which therefore could not have been there long enough to justify the assumption that their presence had acted as the cause of the outbulging. The specimen shown in fig. 2 is from Outside Okinose. It is one of the largest I have seen, measuring 210 mm. in height. The lower half of the body is distinctly compressed laterally, the breadth measuring 90 mm. sagittally and 50 mm. transversely. The upper parts are swollen, the osculum being roundish with a diameter of about 45 mm. Thickness of the wall in the middle of the body, about 14 mm. (exclusive of the gossamer-like layer of prostals). The basal attachment is at one corner of the lower end. In the angular corner opposite to this opens a small secondary osculum of an irregularly oblong shape. On one side of the speci- KHAEDOCALYPTÜS CAPILLATUS. 279 men there is a large cicatrice- like patch where the wall is very thin, probably the result of the healing of an injury received there. The annexed text- figure 11 represents another large specimen which belonged to Mr. Alan Owston (O. C. No. 102). The shape is that of a spindle-like vase, stand- ing erect and but little compressed laterally. Height, 210 mm. Breadth, 72-79 mm. in the middle. Thickness of wall in the broadest part, 7-16 mm. The irregularly shaped osculum at the narrowed upper end is 3»5 mm. long and about 20 mm. broad. The body likewise narrows below but some- what expands again at the very base. In all the larger sj)ecimens, the dis- tribution of diactinic prostalia is confined to the oscular edge and to a narrow zone along it. I presume that in this species, as in R. victo7' and Staurocalyptus glaher, the early formed diactinic prostalia of the young are cast off and thus become lost during growth, though new spicules of the same kind continue to be produced in the growing parts, /. e., in the oscular marghi. Here the spicules spring out either singly or in small tufts. In the latter case they generali}^ arise from the apex of small dermal elevations, the same as those which, lower down in the body, bear the tufts of pentactinic prostalia. The mar- ginalia project generally straight upwards to a length of 7-12 mm. and taken together may form a bristly wall around the oscular opening. The secondarily formed osculum is provided with none or at most wâth only a few of such marginalia. Texl-fig. 11. — Rhabdocalyp- tus capillahis. J natural size. 280 AKT. 7. 1. IJIMA : HEXACTINELLIDA, IV. AU the rest of the external siirftice is thickly covered with large or moderately large pentactinic prostalia, forming a persistent gossamer-like layer much as we have seen in R. unguiculatus. The layer, which may be 9 mm. thick, is sufficiently resistent to allow handling the specimens without touching the dermal surface. The pentactins arise invariably in tufts, often ten or even more in number to each, situated at intervals of 4-10 mm. from one another. The ectosome projects at the tuft-basis in the form of a narrow papilla-like prominence which may be 2 mm. high (PI. XXII., fig. 12) ; and this prominence always occupies a position central to each irregularly radiating group of hypodermal fibers ( = paratangentials of hypodermal pentactins ; same plate, fig. 16). In the prostal tufts the different paratangential heads usually lie at various distances from the dermal surface, according to the order in which their successive protrusions took place. In certain specimens I find in some places a film-like sub- stance stretching in streaks and patches between the prostalia. The presence of nuclei in it can not be demonstrated and tliere can be little dou])t that it represents something — probably mucous — foreign to the sponge. The dermal surface is rather uneven. As in all species with diactinic dermalia, the dermal meshes are observed under the lens to be irregularly shaped, — not quadrate. The layer is supported by the intersecting hypodermal fibers already referred to (fig. 16). There is a comparatively wide subdermal space, traversed by the shafts of hypodermalia and prostalia and also by the distal ends of certain parenchymalia. On sections of the dried body- wall it plainly presents itself to the unaided eye as a layer which may be 1-2 mm. wide. RHABDOCALYPTUS CAPILLATUS. 281 The roundish entrances into the incnrrent canals are on the wliole rather wide. In large specimens they may attain 4 or 5 mm. in diameter, those of nearly similar sizes being separated from one another by an interspace measuring in width nearly as much as their own diameter. The entire inner surface of the wall is covered by a gastral lace work with regularly quadrate meshes of 180-240 /^- (on the average 175 /-«) on a side. When viewed in an oblique direction under the hand-lens, the gastral layer presents a velvety appearance on account of the projecting proximal rays of the hexactinic gastralia. Directly beneath it are distinctly observable hypogastral strands of varying strength. They run in various directions and by intersecting form wide and irregularly shaped meshes. As usual they are frequently seen to stretch right across the excurrent canalar apertui'es. The stronger strands may push out the gastral surface in a ridge-like line. A somewhat different state of the gastral surfece obtains in very small young specimens. Until these have grown to a certain size, the gastralia are not present in numbers suffi- cient to form a continuous layer. Hence, there occur between them gaps by which the excurrent canals open directly and freely into the gastral cavity. The little specimen of PI. XXII., fig. o, still exhibits this condition of the gastral sur- face ; whereas, that of fig. 4 is already in possession of a con- tinuous gastral layer covering the apertures of all the excurrent canals. These, in large specimens, are smaller than the average incur rent apertures on the external side, generally measuring 2-3 mm. in diameter. They are closely packed together. The main canals, both incurrent and excurrent, are pit-like 282 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. and deep. On sections of tlie wall it is easy to make out that those of the two systems alternate (fig, 12). Spiculation. The principal 'pnrenchymalia are slender bow-like oxydiactins, reaching up to 24 mm. length and 132/^ in thickness at the middle. The gradually tapering rays are subtermina.lly more or less rough. Of the more slender parenchymalia I see no points worth special mention. The hypodermalia consist of large or moderately large oxy- pentactins with paratropal paratangentials which may reach 12 mm. in length and about do /^ in thickness at base. The shaft, up to 15 mm. in length, is always the longest of all the rays ; it is smooth except at the end which is somewhat rougli. The paratan- gentials are likewise smooth in the incompletely developed state of the spicule, but finally acquire the spiny armature characteristic of all Pvhahdocalyptus species. The stout conical spines occur here in a rather irregular distribution — at any rate, not strictly in two lateral rows — at intervals varying from 70 n to 200 //. Those situated on the basal parts of the rays may be 80/^ long, springing at first vertically and then becoming more or less bent in an in- definite direction. Towards the end of the rays they become shorter and more slender, being pointed obliquely away from the spicular center. Exactly as I have enunciated under R. victor (p. 243), the hypodermalia occur in close groups of two or more — frequently as many as eight or nine, — the paratangential heads in each group forming hypodermally an irregularly radiate system RHABDOCALYPTUS CAPILLATUS. 283 of fibers (PL XXII., fig, 16). In each such system, the most superficially situated head or heads, generally 1-3 in number and belonging to the oldest hypodermalia in the group, consist of rays which exhibit the spines ; all the more deeply lying and succes- sively younger heads are made up of smooth rays. From the center of the radiate hypodermal system arise the prostalia in a tuft, the base of which is enveloped in the papilla-like prominence of the dermal surface. A number of comitalia accompanying the shafts and those of certain more or less radially dis[)osed parenchy- malia extend their distal ends into this papilla-like base of the prostalia. The prostal pentactins, being nothing else than old hypodermal pentactins protruded through the dermal layer, need not be specially described. Diactins do not seem to associate with the hypodermal paratangentials in forming the support to the dermal layer. It may be worth mentioning here that in the smallest specimen I have had — measuring only 3 mm. across the body — the small hypodermal pentactins were all found to have regularly cruciate paratangentials. This may be explained by the f^ict that they were situated isolatedly and hence there were wanting in their close proximity shafts of older pentactins, the presence of which shafts should cause the paratropism of the younger pentactin-heads. None of the hypodermalia in this little specimen were protruded as prostalia, nor were any of them as yet in possession of the spines. The Jiypogaslnd strands consist of diactins, essentially the same as those which make up the parenchymal bundles. The dermalia (PI, XXII., fig. 6) are (juite predominantly 284 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. (liactiiis ; rough throughout, straight or slightly arched and some with the central annular swelling or knobs. Total length, 286- 462/^ (on an average 380/^); thickness near the middle, 72-lle /A Rarely the dermalia are stauractins or tauactins ; even pent- actins and hexactins were met with amongst them during my studies, though in quite a limited number of instances. I regard it important to mention that up to certain early stages in the post-larval growth of the sponge, the dermalia may be predominantly stauractins, not diactins as in all the larger individuals. Thus, in the little sj^ecimen of PI. XXII., fig. 3, as also in the still smaller one of only 3 mm. diameter, I find stauractins more numerously represented among the dermalia than diactins ; whereas, in the specimen of fig. 4, same plate, the two forms of dermalia occur already in the inverse proportion. This change in the predominant form of dermalia during the ontogeny might well be regarded as the repetition of that which had taken place during the phylogeny of the species. The ffastralia (PL XXII., fig. 13) are almost all liexactins with rough, tci})ering and pointed rays. Exceptionally forms witli a less number of rays and even diactins may occur as the gastralia. The hexactins have distal and para tangential rays 120-220// in length and 12-1 3 2 ,« in thickness at base ; the free proximal ray may be twice as long as any other ray in the same spicule and the microtubercles on it may be slightly more strongly developed than tliose on the others. However, there not infrequently occur such hexactins as have the proximal ray in no way specially distinguished from the others. With their paratangentials the hexactins form a continuous and regularly (|uadrate-meshed lace- work. The sides of the meshes, 132-242 /^ long, are composed EHABDOCALYPTUS CAPILLATUS. 285 usually of two rays, but occasionally of three, belonging to as many ditlerent spicules and running side by side. No special canalaria are found. The oxyheT: asters occur in abundance in all parts of the sponge-wall. In a few instances I have seen them even outside of the dermal layer, and frequently they are borne, together with ' some discoctasters, on the free proximal rays of gastralia. They are mostly normally developed, the terminals being in general slender, minutely rough and nearly straight or slightly wavy. Diameter, 106-136«. The oxyhexasters in the periphery of the wall are somewhat differently characterized from those in the deep parts. The former (PL XXIL, figs. 7 and 8; also PL XXIIL, fig. 19) are on the Avhole slightly larger ; their princii)als are of a perceptible length ; and there occur usually 3, sometimes 4, terminals to a principal. In the latter (figs. 14 and 15) the principals have almost disappeared ; the terminals are somewhat stronger but less in number, there being generally 2 or rarely 3 of them to a principal. Occasionally the deeply situated oxy- hexasters are hemihexactiuose and only rarely hexactinose ; more- over, their terminals may sometimes be provided at base with some distinctly barb-like microtubercles, — all of which seems never to be the case with the peripheral oxyhexasters. In a certain specimen I have found the terminals in some of the subgastral oxyhexasters supplied with basal barbs about as prominently developed as in R. mollis, but in any case this seems to be a character varying much according to individuals as well as within the same individual. In the smallest specimen I have had (3 nnn. in size), oxy- hexasters were already numerously present, but all these agreed in 286 ART. 7. 1. IJIMA : HEXACTINELLIDA, IV. appearance with those occurring in the peripheral parts of the larger specimens, though the terminals numbered usually two to each principal. The discoctasters (PL XXII., figs. 9 and 10) are met with in fair abundance in all parts of the wall, though much less numerously than the oxyhexasters. They occur most abundantly in the endosome, in which they may nearly equal the number of the meshes in the gastral lace work ; on tlie other hand, they are found but rarely, if at all, in the ectosome. Of all the Acan- thascinie known to me from the Sagami Sea, this species has the smallest discoctasters. Diameter, 7(3-110/^ (on an average 91/^). Thus, they are in general smaller than the oxyhexasters. The central node exhibits the six hump-like tubercles more or less prominently developed. The principal is about one-fourth, or more commonly somewhat less than one- fourth, of the length of the entire ray ; it is longitudinally ribbed and uneven in contour, being thicker in some parts than in others. The slender terminals number 6-12 to each tuft; they are arranged not in a circle but in a- solid bundle. The tuft widely expands distally since the tei-minals are bent outwards. A tuft measuring 7^ ,« across at base may be oO // wide at tlie outer end. The terminal discs are minute and })inhead-like. Sometimes they may show 7 or 8 recurved marginal teeth when examined under a very high power of the microscope. Now" and then there occur discoctasters in which some primary terminals stand out free from the central node, without combining with any of the eight secondary })rincipals. Both in size and shape the discoctasters of the species seem to most closely resemble the same of B. iener F. E. Sch. Their RHABDOCALYPTUS CAPILLATÜS. 287 appearance is so vorv characteristic that they constitute a highly important clifferentinl point in the identification of the present species. The microdlscohexasters (PL XX IT., fig. 11) are sparsely present. I have met witli them mostly in preparations of the ectosome, but they are not altogether wanting in those of the endosome. Diameter 20-25 /A The figure I have given will suffice to give a good idea of their appearance. In the smallest specimen frequently referred to, I have failed to discover a single microdiscohexaster, although the entire body was made into preparations and subjected to examination. The basidictyonal plate was found to be represented for the most part by the usual thin and small meshed limiting layer (PI. XXII., fig. 17). Here and there in the sparsely microtubercled beams are seen spicular axial canals in the form of plane crosses. At certain places in the layer the beams are developed, though irregularly, in all the three dimensions ; thus, bringing about in the parts a dictyonal framework of some massiveness. Probably hexactins, and possibly pentactins also, participate in the forma- tion of such parts. I have omitted to determine the disposition of the axial canals contained in those parts. Soft Parts. Unsatisfactory as is my knowledge concerning tlie histology of the species, I may put down some points from notes taken long ago. I beg to mention here that Plate XXIIL, illustrative 288 AKT. 7. — I. IJIMA : HEXACTTNELLIDA, IV. of the soft parts, had been prejiared and printed before I found betler matcrinl for the study in Euplectella rnarshalli (Contrib. I). An idea of the arrangement of the soft parts may be obtained from PL XXIII., figs. 22-24, comments upon which I consider unnecessary beyond the explanations attached to them. The chambers measure 88,"- in average diameter. In fjivorably situated parts of their small-meshed reticular wall, I have seen, though faintly, choanocyte-nuclei, generally one to each of the nodes, as shown in fig. 20. The three round or oval bodies in the lower part of the figure just referred to are without dou1)t archreocytes, though at the time of drawing them I was under the impression that they were merely nuclei of a special sort. Small groups of the same cells are also seen on the charaber-w\all in fig. 19. Occasionally they were observed formiug compact masses of a quite large but varying size. The trabecular cobweb is very extensively developed. Nuclei, 2-2': /' ill diameter ; usually with several nucleoli-like granules within. The trabecuhe, generally very thin and filament-like, are frequently expanded into films (fig. 19), especially on the dermal and gastral limiting surfaces (fig. IS). The film-like parts may appear as if containing irregularly branched and often ill-defined fibers or streaks of condensed protoplasm ; this is evidently due simply to irregularities in the thickness, or on the surfece, of the film as enunciated on pp. 220-201. The trabecuUie of the gastral surface are pushed out, so to say, by each and every proximal ray of gastralia, so as to form a little cone around the latter (fig. 24). The cone may well be compared to a tent, but it must not be accepted that it invariably exhibits externally a membranous limiting surface. More frequently it is RITABDOCALYPTÜS CAPILLATFS. 280 made up of a cobweb of filamentous trabeeula' iiiteriiallv as well as ill tlie most superficial situation, similar to the trabecular cones on the dermal surface of Euplectella marshalU (Contrib. I., p. 123). Sometimes the trabecular substance may extend as a thin film vertically between adjacent proxini;il rays of gastralia in a sail-like manner ; at other times I have seen it in the form of thin isolated threads running straight from one cone to another or directly between the terminal parts of the rays, each of which basally supported a cone. Such being the real circumstances, the term gastral membrane, unless used with certain explicit reserva- tions, would be apt to lead one into misconceptions. Thesocytes occur in aVuindance. I have seen them mostly in the dermal layer and on trabecuhie of the subdermal s})ace. Numbers of them may l)e recognized in figs. 18 and 19 as spherical or ovoid and often conglomerate-like ])odies which are stained as well as trabecular nuclei but are much larger. Several are shown much more highly magnified in fig. 21. The cells are of various sizes, from 4// up to 15/^ in diameter. The small nucleus is generally seen pressed against the cell-surface, where there probaldy always exists a delicate envelof)ing membrane. In the smaller of the cells the contents may sometimes appear to be simply finely granular or homogeneous ; otherwise they are seen to contain, or rather to consist of, one or more fat-like spheres of the same microchemical character as the thesocytal contents of other Hex- actinellids. Of the larger thesocytes the majority are represented by compact conglomerate-like groups of the same spheres, showing the appressed nucleus at some part of the periphery. The not very numerous spheres in one such cell are comparatively large and may give it a rosette-like appearance. Thesocytes of the above description I consider to be such as have attained full 290 ART. 7. T. TJIMA : IIEXACTINELLTDA, IV. (lovclopnioiit. Ceitciiii other thesocyte^ of tlie larger size sliow irregulurly and ratiier coarsely granular contents, the granulation often heilig considerably i-arified or even entirely ohliterated in the peripheral parts. At times the cell-outline is irregular, ap- parently as the result of shrinkage. Such thesocytes may be regarded as those in which the contents are undergoing, or have undergone, disintegration preparatory to Ijecoming consumed. In the present species I have observed a, number of oxy- hexasters which apparently were not yet fully developed, or had l)ut recently attained full size ; at all events they seemed to preserve undisturbed their original relation with the surrounding soft tissues. Reference has already l)een made in Contribution I., p. 109, to the developing oxyhexaster and the scleroblasts. In this connection I can add nothing of much importance to what I then said ; but attention may be called to the three oxyhexasters included in fig. 19. The smallest depicted is about the smallest and youngest I have met with ; the fact that the terminals appear to be stained throughout their length indicates the presence of a protoplasmic envelope. The two others may be considered to be of nearly or quite mature development. In all the three oxyhexasters the central parts are enveloped in a nucleated j^rotoplasmic mass running out in the periphery into filamentous trabecuhc. The mass itself may indistinctly show a cobweb-like structure and I am inclined to regard it as for the most part nothing else than the trabecular sul)stance. Accordingly, some of the nuclei contained in the mass are probably simply trabecular nuclei, and the rest, scleroblast-nuclei wliich should l)e present. It is difficult to dis- criminate between the two sorts of nuclei ; however, I believe that at least those Iving immediately around the central node of the DIAGNOSE« OF THE SPECIES. 291 spicules may safely be taken for the scleroblast-iinclei. The nuclei in (|uestion occur in this position constantly in all such spicules as are apparently still growing and also for some time after these have attained full dimensions. As viewed in an equatorial optical section of the sjiicule the said nuclei are seen to be four in number, one situated in each of the four angles formed by the short principals. In certain cases these particuhir nuclei were found to l)e somewhat larger than others lying close beside them. For the present I can defmitely say nothing more aljout the matter. DIAGNOSES OF THE SPECIES TREATED OF IN THIS CONTRIBUTION. Genus LANUGINELLA O. ScHM. L, jmi>n O. SCHM. — Small, spheiical or ovoid, lirndy at- tached at base ; with a small, smooth-edged osculum leading into a nearly tubular gastral cavity. External surface smooth or covered with a veil formed of small j^i'ostal pentactins. Aper- tures of excurrent canals not covered by gastral layer. Dermalia, rough stauractins. Gastralia, regular oxyhexactins. Hexasters consist of discohexaster and stroljiloplumicome. Discohexaster, varying in size but under iUO /'• dia.; spherical ; each short 292 ART. 7. — 1. IJIMA : HEXACTINELLIDA, IV. principal generally with 2-4, rougli, moderately strong terniinal^^; marginal prongs of terminal dise, clistinetly developed. k5trobilo- plumieome, 34-7(3 P- in dia. Genus SCYPHIDIUM F. E. Sch. (Diagnosis on p. 2(1). .S. lonfßis/tiie(f (I.j.).- -Saccular, moderately thick-walled, some- what like a pear in shape but smaller, wdtli an osculum at the upper rounded end ; firmly attached l)y the contracted base. External surface with low conuli, fi'om which there arise long and rather strong, diactinic prostalia. In association with these may occur a few small and inconspicuous pentactinic prostalia, with paratangentials which are smooth or shagreen-like and either regularly cruciate or paratropal. Dermalia, predominantly stauractius Avith s])inose rays. Gastralia, similarly rayed hex- actins of a larger size. Oxyhexaster, 88-104 f- dia., fretpiently hemihexactinose and hexactinose in the deeper part of the cho- aiiosome. Discohexaster, 90-130 /'■ dia., with slender terminals. Microdiscohexaster of the typical shape, 23-25 /^ dia. ,V. n(imit/- dia.; often hemihexactinose and not seldom hexactinose. Peripheral disc- DIAGNOSES OF THE SPECIES. 297 octaster, 106-137 !'- dia.; more deeply situated discoctaster, gen- erally 200-2G0 r- dia.; secondary principal in length shorter than, or about equal to, the terminals. Microdiscohexaster, 15-23 /nlia. A, alatn Ij. — Ovoid in shape ; thick- walled. (Prostal di- actins present ?). l!)ermalia, almost exclusively pentaetins. Gas- tralia, hexactins ; not forming a continuous lacework over excurrent canalar apertures. Oxyhexaster, 144-190 /^ dia.; generally normal, with 2-4 terminals to each principal ; seldom hemihexactinose. Discoctaster, 136-220 ,"■ dia.; secondary principal about equal to, or longer than, the terminals. INIicrodiscohexaster, 30-35 fjt dia. Genus STAUROCALYPTUS T.T. (Diagnosis on p. ICi'I;. S. rœpct'i (F. E. 8ch.). — Cup-like, with short stalk -like base. With peculiar pit-like subdermal cavities whence arise nan-ow incurrent canals. Excurrent canalar apertures, relatively very wide (up to 8 mm. dia.) ; all freely ojien. All sj^icules with remarkably slender rays. Pentactinic hypodermalia, small ; with paratangentials under 2 mm. in length. Principal parenchy- malia under 35 !'■ in thickness. Dermalia, slightly rough pent- aetins ; not infrequently stauractins. Gastralia, similar hexactins. Oxyhexaster, 88-130 // dia.; some hexactinose. Discoctaster, 128-180 // dia. Microdiscohexaster, 22-24 /^ or more in dia. S. fJotrlhtf/i (L. ]\[, Lambe). — Broadly sacciform, somewhat outbulged on one side and narrowed at base. Canalar apertures 298 ART. 7. — T. T.TIMA I TTEXACTTNELLTDA, TV. snmll ; those on the gastr;il Rurface freely open. Both diaetinic and pentactinic prostalia present over the entire 1>o(ly ; the latter are small with granular- siirfaeed paratangentials a1)out 2. '2 mm. long. Principal parench3mialia, fine (less tlian 41 !>■ in hreadth). Der- malia, rough pentactins; exceptionally diaetins. Gastralia, similar hnt smaller hexactins, with rays of 80-100 !>■ length. Oxyhexaster, 100-120 !>- dia.; both hemihexactinose and hexactinose forms numerous. Discoctaster, 228-820 !'■ dia. ^Nlicrodiscohexaster, 20 // (lia. ,S*. lulntlosHs T.T.— Similar to ,S'. (loivViiu/i in general char- acters and in spienlaiion. ]]ody, tubular. Gastral surface with freely open canalar apertures ; hairy on aeeonnt of the project- ing ends of fine parenehymalia. Paratangentials of pentaetinie prostalia, generally under 2 nun. in length ; rarely 4 mm.; acquiring a rough surface due to nearly vertical, fine and sharply pointed microspines. Principal parenehymalia may l)e 12 mm. long and 130// l)road. Dermalia, rough pentactins but frequently stauractins and rarely hexactins. Gastralia, similar hexactins, with rays 130-200 /< in length. Oxyhexaster, 7')-115 !'■ dia.; very rarely hexactinose. Discoctastei-, 130-213 !>■ dia. IMicrodiscohex- aster, 19 ;'- dia. ,V. ((jpitis Tj. — Similar to S. (Joinliiu/i in general characters and ill spiculation. Body, tuludar or vase-like, ('analar apertures on the gastral surface, some freely oi)en while others are covered with a small-meshed endosomal lattice, developed in irregular patches and in which the gastralia do not form a continuous lacework. Pentactinic prostalia, large ; with smooth or granular paratangentials o-12 mm. in length. Principal parenehymalia, DIAGNOSES OF THE SPECIES. 299 long and thick (.some over 2-3 mm. long and over oOO u. thick). Dermalia, rough pentactins; rarely stauractins. Gastralia, similar liexactins with i-avs 140-1 7'3 nnn. long or even longer. Oxv- hexaster, 11'')-1C)() ;/. dia.; hexactinose forms (juite numerous. Discoctaster, 120-10!) /' dia. ^Microdiscohexaster, 19-20 /^ dia. *S'. ritt((c(nit/tt(s 1j. — Similar to S. doivUngi in general char- acters and in spiculation. Body, vase-like (?). Gastral surface, entirely covered witli a small-meshed endosomal lattice in which the gastralia do not form a continuous lacework, but which extends over all the excurrent canalar apertures. This surface is beset with rather coarse projecting needles which render it conspicuously spiny. Diactinic prostalia not found on the lateral surface. Pentactinic prostalia with smooth or granular para- tangentials reaching up to 4 mm. or more in length. Principal parenchymalia may be 250 /^ thick. Dermalia, rough peutactins ; rarely stanractins and diactins. Gastralia, similar hexactins with rays 95-loO/'- or more in length. Oxyhexaster, 100-132/'- or more in dia. ; hexactinose forms not found (entirely wanting ?). Discoctaster, about loO !'■ up to nearly 300 /^ in dia. Microdisco- hexaster, not found (wanting?; if present, 19 /< dia.). -S. itiict'orhcfHs Jj. — Tubular, laterally compressed. Giuals very narrow. With inconspicuous veil, formed of pentactins in which the paratangentials are regularly cruciate and measure not more than 11 ra.m. in length. Dermalia generally stanractins, exceptionally pentactins ; rays beset with rather strongly developed microtubercles. Gastralia, similar hexactins. Oxyhexasters in the subdermal region, 68-100/'- dia.; those in deeper parts, 106- 136 [X dia., with stronger ]'ays ; occasionally hemihexactinose ; 300 AKT. 7. 1. IJIMA : HEXACTINELLIDA, IV. lit'xactiiiüHu luriiiH iiut fouiul. Discoctaster, 114-128/-«; found in subgastral region only. Microcliöcoliexaster, 2o-26 !'■ dia. ; rare. ;V. ffluber Ij. — Saccular or vase-like ; tliiek-walled ; some- what laterally compressed after attaining a certain size. Small young specimens with tine long prostal needles which seem how- ever to Ijccome lost with growth. Pentactinic prostalia, never present ; so that old specimens have a quite smooth external surface. The extensive subdermal space and large canals give a cavernous appearance to the delicatel}^ textured wall. Dermalia, nearly all stauractins with microtubercles most strongly developed on the outer side of the rays. Gastralia, moderately large oxyhexactins, generally with the free proximal ray longer and more rough than the others. Oxyhexasters, 08-114/-« in diameter ; all normally developed, having 2-4 terminals to a princi])al. Discoctasters may be of an unusually large size ; 240-060 !>■ in diameter. Micrcxliscohexasters, 1Ö-22 i>- diameter ; not unconnnon, but sometimes extremely rare. S. hrtei'futt lints fj. — Laterally compressed, pouch-like, with narrow canals. No prostalia of any kind. Dermalia, predomi- nantly stauractins, occasionally pentactins and tauactins, rarely diactins ; with slightly rough rays. Gastralia, similarly rayed but smaller pentactins and stauractins ; not forming a continuous lacework. ( )xyliexasters, always normally develo})ed ; 100-114/^ in diameter. Discoctasters, 1 1 ( )-2( )0 /'• in diameter. Microdisco- hexasters, lO-lO/-« in diameter ; common in certain parts. S. ithofhd indes Jj. — Thick-wjdled, pear-shaped, attachetl by the narrower end. With .strong ])rostal needles and a gossamer- DIAGNOSES OF THE SPECIES. oOl like layer of jn'ostal pentactins, sj)ringing out from the small conuli of the surface. Dermalia, nearly all rough diactius. Gas- tralia, similar diactius. Oxyhexasters mostly normal or hemihex- actiuose ; rarely hexactinose ; with rather strong, rough terminals ; 114 /^ in average diameter. Discoctasters, small ; 100-200 !'■ in diameter. JMicrodiscohexasters, 20 /^ in average diameter ; sparsely present. Genus RHABDOCALYPTUS F. E. ScH. % (Diagnosis on p. 2o6). B» victor Ij. — \^ase-like ; somewhat laterally compressed, especially at base. Veil and fine prostat needles present in the young stage but later are generally cast off. Dermalia, nearly all rough stauractins. Gastralia, similarly rayed hexactins of a nearly regular shape. Oxyhexasters, with rough terminals ; 180- 280 /^- in dia. ; hemihexactinose form, numerous ; hexactinose form, not uncommon. Discoctasters, 180-240 /'• in dia. ; terminals nearly straight. jNIicrodiscohexaster of the usual size and shape in sparse distribution. /»*. impllis F. E, ScH. — Vase-like or sac-like, laterally com- pressed ; with a strong tendency to form tuljular daughter per- sons along one of the sagittal body-edges. Veil generally cast oft'. Dermalia, nearly all rough diactius. Gastralia, similarly rayed hexactins of nearly regular shape. Oxyhexasters with ter- minals more or less distinctly barbed at base ; hemihexactinose and hexactinose forms, common ; 102-160 i'- in dia. Discoctasters, 130-176 !'■ in dia. ; terminals nearly straight or slightly bent 302 ART. T. — 1. IJIMA : HEXACTINELLIJDA, IV. oiitwarcl«. ]\Iicro(li8C()li(jxaHtery, present in variable niiiul)er8 ; 22- 27 /'- ill diel. K. tfH{/tflrtff((fKs L). — Vase-like ; entirely eovered with ;i thick irossaiuer-Kke veil. Dermalia, iiearlv all rouoh dicictin.s. Gastralia, hexactiiis ; similarly rough l)ut with iiiuch longer ray«, the free proximal ray being especially well develo])ed ; the radial axis, 640-850 y- long. Oxyhexasters, loO-lGO /'- in diameter ; not infrequently heniihexactinose and only occassional ly hexaetinose. DiscoL'tasters, 143-11)0 /< in diameter; principals, 20-2-5 /< long; terminal tufts ex[)ajided at the (juter end ; terminal discs with strongest marginal teeth on the side turned awav from the axis of the tuft, or toothed only on that side. ^Lierodiscohexasters, sparse ; 22-30 f^- in diameter. Ji. caplllatiis Ij. — Yuse-like or sac-like, more or less laterjdly compressed ; entirely covered with a thick gossamer-like veil. 4)erinalia, nearly all rough diactins. Gastralia, rough hexactins in which the free [)roximal ray may be twice as long as any other ray in the same spicule. Oxyhexasters, 10(3-130 !'■ in dia- meter ; occasionally heniihexactinose and rarely hexaetinose. Disc- octaster small, measuring 82-lOG !'■ in diameter ; terminal tuft distinctly outHaring at the outer end. MicroLliscohexasters, s[)arse ; 20-25 !'■ in diameter. LIST OF LITEr.ATT'KE. oOo List of Literature referred to in this Contribution. Kent, W. S. '70. On tlie '' Hexactinellida3," or hexradiate spiculed siliceous sponges taken in the 'Noma' Expedition, etc, — Monthl}' IMicrosc. Jour., 1\"., \)\^. 241-252 ; pi. LXI1I-LX\^ Schmidt, O. '70. dirnndzüge einer Spongien-Faniia «les Atlantischen Gebietes. Cauth-.r, it. J. '72. Descriptions of two new sponges from the Philippine Islands. — Ann. and Mag. Nat. Hist., ser. 4, vol. X., pp. 110-113. GuAY, J. E. '72, < 'n a new genus of hexradiate and other sponges discovered in the Fliilii)pine Islands by Dr. J. B. Meijer. — Ann. and Mag. Nat. Hist., ser. 4, vol. X., pp. 134-139. Cautku, ]\..]. '73'. l^escription of Laharia hchrispherico, etc. — Ann and Mag. Nat. Hist., ser. 4, vol. XT., pp. 27.')-28(). ,, '73/;. On the Hexactinollidiü and Litliistidte generally, etc. — Ann. and Mag. Nat. Hist., ser. 4, vol. XII. '75, Notes introductory to the study and classification of the Spongida. Pt. II. — Ann. and Mag. Nat. I [ist., ser. 4, vol. XVI.. p. 199. IMarshai.l, W. '76. Ideen über die Verwandtschaftverhältnisse (Ur Hexactinelliden. — Zeitschr. 1'. wis';. Zool.,vol. XXVHI. P)()WKRP,Axiv. .1. S. '77. Descriptions of five new species of sponges dis- covered by A. B. Bleyer on the Philippine Is- lands and New Guinea. — Proc. Zool. Soc. Loud., 1877, pp. 4Gl-4n4. Sc;riri,zK, F. E. '86. Ueber den Ban und das System der Hexactinelliden. '87. The Challenger Report. Hexactinellidn. Lambk Ij. ^I. '93. Sponges from the Pacific Coast of Canada. — Trans. Roy. Soc. Canada. Sect. IV., p[t. 37-38. Sciiui.ZR, F. E. '93, Ueber die Ableitung der Hexactinelliden Nadeln vom regulärem Hexactine. — Sitz.-ber. Akad. wiss. Berlin. 1893, pp. 991-997. TopsENT, E. '95. Campagnes du Yacht Princess Alice, etc. — Bull. Soc. Zool. de France, t. XX., pp. 213-21."). 304 LIST OF LlTERATUPvK, Ijima, I. '96. Notice of new Hexactiriellida from f^agami Bay. — Zool. Anz. lS9n, No. :"304. ,, '97- Revision of Hexactinellids with discoctasters, etc. — Annot. Zool. Jap. Vol. I., pj). 4.3-.j9. S('iii:r,ZR, F. E. '97. Kevision des Systèmes der Asconematiden und Ros- selliden. — l*>itz.-l)er. Al^ad. wiss. Berlin, XXVI., pp. .■)20-.").)8. ,, '97". Ueber einige Symmetrieverhältnisse hei Ilexacti- nelliden Nadeln. — VerhantU, Deutsche Zooj. Ge- sellsch. 1897, pp. .3.")-.37. TjiMA, T. '98. The genera and species of Rosselhdie. — Annot. Zooh .Tap. Vol. ir., pp. 4l-."')5. Uhavjkr, Cit. '99. Sur une collection d'Épmïges (Hexactinellides) du Japon.— Bull. Mus. d'hist. nat. Paris, t. V., No. 8, pp. 419-423. Sr'nni,zE, F. E. '99. Amerikanische Hexactinelliden. .Jena, 1899. ,, '00. Die Hexactinelliden. Fauna Arctica, vol. I., pp. 87-108. Ijima, I. '01. Studies on the Hexactinellida. Contribution, I. — .Tour. Sei. Coll., Tokyo. Vol. XV. KiRK PATRICK, V\. '01. Description of a new Hexactinellid sponge from S. Africa.— Ann, and Mag. Nat. Hist., ser. 7, vol. VII., p. 457. „ '02. Descriptions of S. i\frican Sponges. C. of Good Hope, Dept. of Agriculture, No. 4, Ijima. 1, '03. Studies on the Hexactinellida. .Contribution III. .Tour. Sei. Coll., Tokyo. Vol. XVIII. NOTICE. 305 N. B. A Correction with Regard to the Generic Name Placosoma applied by me to an Euplectellid. In ContrilMition III., I described a new Enplectellid genn.s and species nnder the name of Placofioma para(Jict//?riû. Now, Professor F. E. Schulze, in a letter, has very kind!}' pointed ont to nie the fact, which I liad entirely overlooked, that the generic name " IVacosoma " had been pro- occupied, having been employed bv Tschudi for a reptilian genus in 1847 (Arch. f. Natnrgesch., Bd. 13, I., p. ."^O). I therefore beg to withdraw that generic name as applied Iw me to the Enplectellid and to substitute for it " Bolosoma " (-/} ,9(^/,o: = cluinp), — a designation, for the suggestion of which I am likewise indebted to Professor Schulze. PIencofSch i;;;) Key to the species 140 A. cactua F. E. SuH 140 Spiculation 147 Soft parts 154 A. cdani Ij 158 Spiculation KjO Stanfocdlyptas Ij Dj^ Key to the species 16;; 6'. rœperi F. E. ScH KJS 'S', dowlingi (L. M. Lambe) 172 S, iubulums nov. sp 175 Spiculation 177 6'. aßiiis nov. sp ISO Spiculation 18-3 S. glaber Ij 207 Spiculation 21;» S. heteractinus Ij 21S Spiculation 211) .y. pleorliaphides Ij 222 Spiculation 22-1 I ' udeterni iuable Sl() 11. victor Ij 2;]8 Spiculation 245 Soft parts 251 R. mollis F. E. ScH 25;> Spiculation 261 R. unguieidatus uov. sp 2(i8 Spiculation zi J, R. capillalus Ij 27fi Spiculation 282 Soft parts 287 Diagnoses of the species treated of in tliis Contribution 21)1 List of literature referred to in tliis Contribution ;jo;', N.B. A correction with regard to the generic naiue " I'lacosoma " applied by uic to an EuplecLclUd []()b I. IJINA. STUDIES OW TEE HEXACTINELLIDA. CONTRIBUTION IV. PLATE I. Lanui>inella pu])a 0. Scum. Plate I. LaniKjindla pupa 0. t^ciiM. P. 3. Figs. 1-3. Tliree specimens in natnral size. Fig. 1, atiachecl tu a dead coral. Figs. 1 and 2, with veil over the surface. Fis. 4. Dermal ia in lateral view. 300 x . Fi«'. 5. Same in an early stage of development. 300 x . Fi". G. Part of a large hypodermal pentactin as seen from distal side. 150 X. Fi"-. 7. Part of a prostal pentactin forming tlie veil ; seen from distal side. 150 x . Fig. 8. Gastralia in lateral view. 300 x . Fig. 9. Section of sponge-wall, showing a wide and hranched excurrent canal, the chamber-layer with archteocyte-cougeries, etc. Above, the dermal side ; below, the gastral side. 50 x . Fig. 10. Parts of the walls of three chambers lying close together. Archj^eo- cytes and membrana reticularis very badly represented. Above, trabecnlfe arising from chamber-rim. 450 x . F'i-r. II. üiscohexaster. 300 x. Fig. 12. Outer end of discohexaster terminal. Highly magnitied. Fig. 13. Plumicome. 300 x . (Figs. 4-l;î were all taken iVoni the .speciiiifu shown in fig. 1 [Sei. Cull. Mus. Nu. 2*J9]. For more ligures relating to the s[ieeies, see Cuntrilnitiun ill., PI. v.). Ijima, Hexactinellida. Contribution IV. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. I. 8. 5. 10. o^o ^^ 6. 7. 11. 12. 13. Lamiginella pupa 0. ScHM. LITH, & IMP. THE TOKIC POINTING CO I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE II. Seypliidium longispina (Ij.)- Fio-. 2. Fig. 3. Fig. 4, Fig. 5, Plate II. Scyplddmin lonf/hpino (Lt.). P. 22. Fig. 1. Tlie typo specimen (Sei. Coli. Mus. No. 222); with two small young individuals attached to the prostalia of the large ,one. Natural size. All the other figures in this plate, except fig. 15, are taken from the largest individual. Dermalia in lateral view. 300 x . Part of a {)entactinic hyiiodernialia. ir)0 x . Gastralia. 300 x. Hexactinose oxyhexaster. From a deep part of the wall. 300 x . Fig. G. Hemihexactinose oxyhexaster. From a deej) part of the wall. 300 X . Fig. 7. Discohexaster. 300 x . Fig. S. Outer end of a discohexaster terminal. Ahout 1000 x . Fig. 1). Micr(uliscuhexaster. 300 x . Fig. 10. Section of wall ; part adjoining th(i outer siiriace. Above, dermal layer ; with a pentactinic }irostal and two strong diactinic jirostals cut off. About 25 X . Fig. 11. Section of wall; ])art adjoining the inner surlac(,'. P»e]ow, gastral layer. About 25 x . Fig. 12. Trabeculiü and thesocytes ; from a borax-carmine preparation. The nuclei appear as deeply stained dots. The thesocyto product is represented by the large, more or less globular, strongly stained bodies. Smaller, mcire weakly stained and granular s|)hert's re- present probably a stage in the development of tlie thcsocyte ])rodnct. 450 x . Fig. 13. Chamber-rim ; not well ]n-eserved. 'J'he larger deeply stained dots in groups are ]irobably arcli;oocytes ; the weakly stained spheres of granular a})|)earance are likely tliesocytes not yet fully grown. Stained with borax-carmine. 450 x . Fig. 14. Chamber-wall. Tlie deeply stained dots are pr<^bal)ly partly archîeocytes and partly trabecular nuclei. Choanocyt(^ nuclei are scarcely indicated in the figure. Strongly magnified. Fig. 15, Peculiar bodies of uncertain nature, consisting of well-stained threads arranged in diverging l)unches or in a radiid manner. Seen in one of the two young individuals attached to the jirostal needles, 300 x . Ijima, Hexactinellida. Contribiition IV. 3. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. II. 2. 4. 13. S V Scyphidium longispina(Ij.} LITH, & IMP THE TOKIO PRINTING CO I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE III Vitrollula fertilis Ij. Plate III. VifrolJiila JhiiUs Ij. P. oS. Fig. 1. One of 8. C. M. Sp. No, 228, found attached to a dead coral. From Okinopé. Nat. size. Fig. 2. S. 0. M. Sp. No. 231, attached to a worni-tul)e. From off Inatori. Nat. size. Fig. 3. Part of a moderate-sized stauractinic dermal ia. 300 x . Fig. 4. Part of the ]iaratangential cross of a hypodermal pentactin. 300 x . Fig. ;■). Oxyhexaster. 300 x . Fig. G. Microdiscohexaster, the only form of discohexaster present in thi^ species. 300 x . Fig. 7. Part of a stained transverse section tlu-ongh tho body of Sp. No. 231. Exciirrent canal opening freely into the gastral cavity. 3 x . Fig. 8. Part of a longitudinal section of the same specimen. Above, the oscular margin. About 2.5 x . Fig. 9. Surface view of the stained wall of S. C. M. Sp. No. 433 ; aliove, oscular margin. Seen from the gastral side. Al)out 50 x . Fig. 10. Same, seen from the dermal side. About 50 x . Fig. 11. Showing the gradual transition of memhrana refici(Iari-'< (above) into trabecule (below) near the oscular margin. From a hfema- toxylin preparation. 300 X. Fig. 12. A small group of archa='ocytes lying on the chamber-wall. From a borax-carmine [)reparation. 1000 x . Figs. 13-16. Stages in the development of a larva, from sections of Sp. No. 231. Stained with borax-carmine. 150 x. Fig. 17. Stauractinic spicules from the larval stage of fig. IG. The lower figure, in side view. 440 x . Fig. 18. Section of a developing larva, somewliat more advanced than that of fig. 16. Stained with h;\3matein-alum. 300 x . Fig. 1!). The outer layer of flagellated cells, seen surface on. From the same larva. 300 x . Fig. 20. A fully developed larva in optical section, but with tin? peripherally situated spicules of one side drawn in. 150 x . Fig. 21. Longitudinal section of a, fully developed larva, stained with h;ematein-alum, 300 x . Fig. 22. Pasidictyonalia, joined below to the small-meshed attachment ])late. 100 X. Ijima, Hexactinellida. Contribution IV. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. III. Vitrollula fertilis Ij. LITH. & IMP. THE TOKIO PRIhfTiNS CO. t. ULM A. STUDIES ON THE HLXACTINELLIDA. CONTRIBUTION IV. PLATE IV. Graieromorplia meyeri J. E. Gray C. meyeri tuberosa Ij. C. ineyeri riigosa Ij. C. jmcliyactina Ij. Plate IV. Fig.s. 1-8. Typical Crateromorpha nicycri J. K. Giîay. P. 57. Fig. 1. A «i)ccimeii ol' ty[)ical C. vieijcrL (tS. C. M. No. 364). Nat. size. Figs. 2-8 were all taken fruiu this f5[)eciiijeii. Figs. 2, 3. Dermalia. 300 x . Fiu'. 4. Microdiscoliexaster. 300 x . Fig. 5. Oxyhexaster. 300 x . Fig. 6. l)ermalia and hypodermalia in siUi ; seen l'n»ni unlside. A bunt 20 X. Fig. 7. Section of wall. About 20 x . Fig. 8. Basidictyonalia, from middle of stalk. 100 x . Fig. 9. C. meijeri tuberosa Ij. P. QQ. Fig. '.). 31ierndisi.(»liexaster. 30U x . (Tiiis ligure might [>ass for that of the curres[)onding s[(icule of Ü. meijerl riujosa). Figs. 10, IJ. C. ineijt'rl riKjosa Ij. P. 71. Fig. 10. Caualar surface of stalk, with canalaria and fused parenchymalia. .)0 X . Fi". 11. Dermalia and hvnodermalia in situ ; seen from outside. About 20 X . (This figure might pass for that of the corresponding skeletal parts of 6'. meyeri tuberosa). Fig. 12. U. ïïieijtrL Fig. 12, (Jn^ss-sectioii of stalk, showing the canals running through it. ■;| nat, size. Fig. 13. (J. puclujactina Ij. P. 74. Fig. 13. Tliick-r-i\cd hypodermal petitactin, together with a few dermalia and comitaha. (From S. C. M. No. 395). 20 x. Ijima, Hexactinelliâa. Contribution IV. 2. \ >^.\ 'M±^ 13. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. IV. 3. 10. _ /'^r-^^uiz^ S^xC  ii. 20» * >r N / T / j y b^> r- ^ .^ 1 — 5, 12. Crateromorpha vieyeri {J. E. Gray) Carter. 9. C. meyeri tiiberosa Ij. 10, 11. G. meyeri ru.gosa I.j. 18. C. pachyactina Ij. xa i iw? the 'okio printing a I. IJ19IA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE V. Crateroraorplia meyeri iuberosa Ij. Crateromorpliti meyeri rugosa Ij. Plate V. (AU ligiires in huit' natnrîil ,si/.e). Fi■^ 1*2. A sjieciuieii ol" Crateromorpha htvijcri tahenmi 11 I'mni Jldiiilni. 8. C. M. No. 444. Fig. 13. Another specimen of .sanie, from Outside Okinose. ?S. C. M. No. 445. A part of the wall and of the stalk removed. Fig. J4. A specimen of CraterojuorpJia inc/fcrl rngosa Ij. iVoin Outside Okinosé. S. C. M. No. 366. Fi'^ 15. Another s[)ecinien of same, ii\)m llomba. h). CM. No. 3GU. A })art of the wall and of the stalk nnnoved. Ijima, HexactinelUda. Contribution IV. Jour. Sei Coll. Vol. XV III. Art. 7. PI. V. 12, 13. Crateromorpha meyeri tiiherosa Ij. 14, 15. C. meyeri rugosa Tj. LITH, i. IMP. THE TOKIO PRINTING CO I. I.ÎIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE VI. Crateromorplia corrugata I J. Scyphidium namiyei (Ij.)- Plate VI. Crateromorphn rm'ritgafa Lt. P. 78. Figs. 1. t^' 2. Dermalia drawn in jiavt. oOO x. Fio-. ?t. Oxyhexaster. 300 x . Fig. 4. Part of a discoliexastcr. 1000 x . Fig. ~). Stalk and basai dise; the former eut o})en longitudinally to show the internal canals. Above, the stalk is branched at the jnnction with the body proper, which is lost, i nat. size. Fig. (J. An entire specimen (0. C. No. 108). ^VT "'"i^- '"^i^^- Fig. 7. The body of another complele specimen (S. C. M. No. 305) longitudinally bisected. The dermal surface, including that of intercanals, is colored yellow. Nat. size. Fig. 8. Spicnlation of the body-wall. Above, dermal surface ; below, gastral surface. Al)ont 20 x . Figs. 9-17. Scyphidium narniyei (Ij.). P. 32. Figs. 9, a & h. The two type-specimens (S. C. M. No. 302). Aliont I nat. size. Fig. 10. Pentactinic dermalia. 300 x . Fig. 11. Gastralia. 300 x. Fig. 12. Oxyhexaster. 300 x . Fig. 13. A terminal from the same. 1000 x . Fig. 14. Onter end of discohexast(M- terminal. 1000 x . Fig. 15. Microdiscohexaster. 300 x . Fig. 16. Discohexaster. 300 x . Fig. 17. Spiculation of the wall. Above, dermal lay^r : below, gastral laver. About 20 x . Iji?na, Hexactinellida. Contribution IV. ( Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. VI. "^xi 1 — S. Grater oTnorpha corrugata Ij. 9 — 17. Scypliidiiim namiyei Ij. jTirU^a- oy fwsnioa, KanJa. 7oÂyo, Joßan.. I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE VII. HvalascLis sao-amiensis I J. Plate VII. TTynlnf^rvf^ fiar/nmiens/-^ It. P. 88. V\'j;. 1. The tyiie-s]x>cimen, which i-^ ]trol);il)ly to lie tntiiid proRervpcl at Amherst Colleor,. (Pn,f. ]>,. K. Emkhkox), Mass., U. S. A. About 1 nat. tize. Fio's. 2, 3. Dermalia. 300 x . (Fig'. 3 is not gnod, tlu- rays Ix-iiig ihawii relatively too short or too thick). Fig. 4. Gastralia. 300 x . (This figure is again not good, the rays l)oing drawn relatively too sliort or too thick). Fig. .5. Discohexaster. 300 x . Fig. 6. Ilypoderinal oxypentactin, TOO x . Figs. 7-10. Oxvhexasters. 300 x. Fig. 7, a hexactinose form. Fig. 8, an exceptional degenerate I'orm in which the rays are reduced to only four in number. Figs. 9 and 10, hemihexactinose forms. Ijima, Hexactinelliäa. Contribiitio7i IV. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. VII. 5. 2. B yi ^. •Î .ü^ 7. 8. 10. Hyalascus sagamiensis I.j. LITH. & IWP THE TOKIO PRINTING ( I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION lY, PLATE VIII. Ilyalascus saganiiensis I J. II. giganteus Is. Staurocalyptus microchetus Ij. Aulosaccus scliulzei Ij. Plate VIII. Figp. 1-12. HyaJascHS scu/amiensis Ij. P. 88. Fig. 1. tSpiciiles on and near the dermal «iirface in obli(|U(^ view ; Sonie dernialia, hypodennalia and parenchymalia. .50 x . Fig. 2. Gastral layer (below) and some .spicules (|)arenchymalia and oxyliexasters) directly beneath it ; from a section. .3(J x . Figs. 3 — 16. Hijalasciis (jiganteus Ij. P. 100. Fig. .J. A small portion of the dermal surface, much abraded, of the type specimen [)reserved in ihe British Museum. Nat. size. Fig. 4. A small portion of the gastral surface of same. Nat. size. Fig. ."). Spicules on and near the dermal surface in oblique view : some dermalia, a hypodermal ]ientactin and some parenchymalia, .30 x . Fig. 6. Gastralia and hy[)()gastral beams in surface view. .00 x . Fig. 7. Gastral hexactin. 300 x. Fig. 8, 9. Dermalia. 300 x. Fig. 10. Disojhexaster. 300 x. Fig. 11. Portion of the same, more highly magnified. Figs. 1'2-16. Various forms of ox3diexaster. 300 x. Figs. 17-2.5. Staurocalyptas 'hticrochclus Ij. P. 202. Fig. 17. The ty[)e specimen (S. C. M. No. 4.50). Nat. size. Fig. 18. Dermalia. 300 x . Fig. 19. Gastralia. 300 x. Fig. 20. Oxyhexaster from a deep ])art of tlie wall. 300 x . Fig. 21. Oxyhexaster from periphery of the wall. 300 x . Fig. 22. Microdiscohexaster. 300 x . Fig. 23. Discoctaster. 300 x . Fig. 24. Dermal layer in obli(|U(' view ;uid underlying spicuU;s, some hypodermal [)entactins protruded as prostalia. .50 x , Fig. 2.5. Gastral layer in surface view ; some discoctasters and ox\liexasters in situ. .50 x . Figs. 2f)-28. AalosatvuH nclmhxi Ij. 1\ 110. Fig. 2(!. Dermal layer and liy[)odermal Ix'ams (ectosomal skeleton) in surface view. ,50 x . Fig. 27. Two gastralia (below) and adjoining spicules in situ. Amongst the latter, a macrodiscohexaster. .30 x . Fig. 28. Central spliere of macrodiscohexaster, examined in glycerine, showing the axial cross. 440 x . Ijima, Hexactinellida. Contribution IV. / Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. VIII. A9. /^ , :#:'6^-;r „>".*2**^-:^ -0. 7 — 2, Hyalascus sagamiensis Ij. 17 — 25. Staurocalyptus microchetus Ij. 3 — 16. H. giganteus Ij. 26 — 28. Aidosaccus schulzei Ij. I. IJI.UA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE IX. Aulosaccus scliulzei Ij. Plate IX. Aidosuccus schulzci Li. V. llU. Fig. l. Till- typo speciinen, now preserved pmbaljly Mt Amlierst College (l'rof. B. K. Emeuson), Mass., U. S. A. About i natural size. Figs. 2-7. Oxyhexasters. 300 x . Fig. '2 is hexactiiiose, the connuoncst foru). Fig. G is lieuiihexactiuose. All llie rest are degenerate forms, in which cue or more [u'ineipals, together with their ter- minals, have disappeared. Fig. S. Macrodiscohexaster. 1.50 x . Fig, !). Outer end of a terminal belonging to a macrodiscohextister. 4.")0 x. Fig. 10. ]\Iicrodiscohcxaster. 300 x . Fig. 11. Gastralia. 300 x. Fig. 12. iJermalia. Not (juite 300 x. Figs, i:], 14. Sco[)ul;e of extrinsic origin, fourni in the ty[ie specimen. 300 X. The figures were inadvertently introduced into the plate. Ijima, HexacfmelHda. Contrihution IV. Jour. Sei Coll. Vol. XVIII. Art. 7. PI. IX. Aulosaccus schuhet Ij. lITH i. IMP, THE TOKIO PRINTING CO I. um A. STUDIES ON THE HEXACTINELLIDÂ. CONTRIBUTION IV. PLATE X. Aulosaccus mitsukurii Ij. Acanthascus alani I-i. Plate X. Figs. 1-1)5. AuJosaccKS initsulmrU I.j. P. 117. Fii;'. I. The type-specimen in the Sei. Coll. ]\Tnseuni (No. 427), attached to Isis. ^ nat. size. (Figs, ^-l;") were all taken from this specimen). Figs. 2, 3. Macrodiscohexasters. 300 x . Figs. 4-7. Oxjhexasters. 300 x . Fig. 8. Microdiscohexaster. 300 x . Fig. 9. Dermalia. 300 x. Fig. 10. Gastralia. 300 x. Fig. 11. Showing the si^iculation of the wall. Almve, a eomilns with a tuft of prostalia. 20 x . Fig. 12. A small part of a section through the liody-wall, stained with hfematoxylin. Above, two dermalia and the dermal membrane. The latter connected with the choanosome by a pillar consisting of spicules and a dense cobweb of trabeculfe. Below, parts of two flagellated chambers. 100 x . Fig. 13. A gastralia with the free ray directed downwards, and the trabecular cobweb in connection with it. 100 x . Fig. 14. Chamber wall (membrana reticularis) in optical section. 1000 x . Fig. 1."). Same in surface-view. Below, a. group of strongly stained archreocytes. 1000 x . Figs, lG-23. Acanfhnscns alairi It. P. 158. Fig. iC). The type-specimen now belonging to the lîritish Museum. .] nat. size. Fig. 17. Dermalia. 300 x. Fig. 18. Gastralia. 300 x. Fig. 19. Discoctaster. 300 x. (The terminals are drawu a little too jliiek.) Figs. 20, 21. The .two varieties of oxyhexaster. 300 x. Fig. 22. Microdiscohexaster. 300 x . Fig. 23. Skeletal ])arts of the ectosome : dermal lacework and hy[iodermal strands. 50 x . Ijima, Hex'ictinellida. Contribittion IV. Jour. Sei. Coll. Vol. XVIII. Art. 7. PL X. 1 — 15. Äidosaccus mitsukurii Ij. 16 — 23. Acanthascus alani I.j. I ' JHfUeJ ij/ ÂoiAîÀt. Kuula, Tok^.Jaftan. I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XL Staurocalyptus lieteractiniis It. Staurocalyptus sp. Acanthascus cactus Y. E. Sen. Fig. 1 Fig. 2 Figs. 3 Fig. Ô Fig. G. Fig. 7 I<^ig. 8. Fig. Fig. 9. 10 Plate XI. Figs, 1-10. Strn/rncn?//j)f/tf< Jtcfernrfinvs T,T, P. 218. Till* type-speciiiien (S. C. M. N.\ 409). Nat. size, ilypodermal [leutactin. I.IO x . 4. A large and a small stanractinic (Imnalia. ^00 x . Gastral pentactin. 300 x . Discoctaster from the deeper part of tlie wall. .300 x . Oxyhexaster from the periphery. 300 x . Same from a deep [)art. 300 x . Microdiscolu^xaster. 'M)0 x . 10. Spiculation of the wall. Comhi nation fignre. Ahont 30 x . Figs. 11-15. Stanrocahjptus sp. P. 23.5. Fig. 11. The entire S})ecimen, attached to a dead I[exa,Mi'' I. IJI.MA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XII. Acaiitliascus cactus E. E. iScii. Plate XII. Äcantliascus cactus F. E. !Sch. P. 140. Fig. 23. StcUiractinic dermalia iu lateral view. 300 x . Fig. 24. Peiitactinic gastralia in lateral view. 300 x . Fig. 25. Perii)heral discoctastcr. 300 x . Fig. 26. IMicrodiscohexaster. 300 x . Fig. 27. üiscoctaster from siibgastral region. 300 x . Fig. 28. End of a terminal of a subga.stral discoctaster. Very highly magnified. Fig. 29. Oxyhexaster, in which each nearly atrophied princi})al seems to bear two terminals. 300 x . Fig. 30. Hemihexactinose oxyhexaster. 300 x . Fig. 31. Hexactinose oxyhexaster. 300 x. Fig. 32. Oxyhexaster with four terminals to each princii)al. A rare Ibrnj. 300 X . Fig. 33. Abnormal oxyhexaster, only once met with. 300 x . Fig. 34. Ectosome in surface-view, sliowing the dermal lacework, the hy[)odermal strands and the perforated dermal membrane. The larger black dots in the last represent deei»ly stained thesucytes. About 4.") X . Fig. 3."). End(^some in surface-view, showing the perforated gastral membrane and the various s[)icules found in and below it. J31ack dots as in last figure. About 4.3 x , Fig. 3(i. Ends of parenchymalia of various sizes. 100 x . Fig. 37. fSmall portion of basidictyonal framework. 100 x . Ijima, Hexactiaellida. Contribution IV. 23. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. XII. 2é. AcanthasGus cactus F. E. ScH. \K I. IJI9IA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XIII. Staurocalyptus afiQiiis ii. sp. Plate XIII. Staurocalz/iAus affinis u. s}). 1*. 180. Ail lignres from Sei. Coll. I\lus. Sp. No. lî)4. Fig. 1. Exteriuil surface of the wall. Dermal spicules wautiug in iJie greater part. Nat. size. Fig. 2. Uastral surface of the same. Canalar apertures mustly freely open hut some covered with small-meshecl endosomal lattice which is (lilferentiated in irregular jjatches. Nat. size. Fig. 3. Dermalia. ."jOO x . Fig. 4. Gastralia. 300 x . Fig. 5. Small discoctaster from subdermal space ; imperfectly develu})ed in so far as some terminals remain isolated from the secondary principals. 300 x . Part of a larger discoctaster from the subgastral space. 300 x . iiexactinose oxyhexaster from choanosome. 300 x . Microdiscohexaster. Fig. 9. Smooth-rayed oxyhexaster from eudosome. 300 x . Fig. 10. Small ])ortion of ectosome in surface view. JJermalia ami hypo- dermalia held together l)y desiccated soft tissue. (]0 x , Fig. 11. Small jiortion of endosome in surface view. GO x. Fig. 12. Peri[)heral ]K)rtion of a section through the wall. Above, der- mal surface. To the right, a group of paratropal hy[)oderiiial pentactins. In the center, tlie central jjortion of a huge parenchymal principalia, surrounded by comitalia. 30 x . Fig. 13. I'ortion adjoining the gastral surface; of a section thnaigh the wall, ßelow, the gastral surface. 30 X . Fig. 6 Fig. 7. Fig. 8. Ijima, Hexacfinellida. Contribution IV. Jour. Sei Coll. Vol. XVIII. Art. 7. PI. Kill. Staurocalyptus affinis Ij. LITH. & IMP THE TOKIO PRINTING CO \1./ I. um A. STUDIES ON THE HEXACTINELLIDl CONTRIBUTION IV. PLATE XIV. Stanrocalyptus afiPinis n. sp. S. entacantlius n. sp. S. rœperi (F. E. Sen.). Plate XIV. Sfanroraliipfnfi alfinif; n. sp. P. 180. Fis;. 14. Tlie smaller of the two type specimens of S. aßnis (Sc. Coll. Mus. No. 400). Aljout | nat. size. Fig. J C). Smooth-rayed oxyhexaster from gastral side. 300 x . (From Sp. No. 400). Fig. 17. Rough-rayed oxvhexaster from subdermal s])ace. 300 x. (From Sp. No. 400). Fig. -22. Two rudimentary oxyhexasters (earlv develo])iu(;ntaI stage?). 300 x . (From Sj). No. No." 400). Fig. 23. Central node of a discoctaster, showing the disposition of axial filaments. Seen in glycerine. 1200 x . (From Sp. No. 194). Fig. 24. Central node of an oxyhexaster, showing the axial cross of fila- ments. Seen in glycerine. 1200 x. (From Sp. No. 194). Fig. 2."3. Central node of a hexactinose oxyhexaster, showing ti>e axial cross of hlanients. Seen in glycerine. 1200 x . (From S)>. No. 194). StauroccdyptKS erdacroit/ivs n. sp. P. 191. Fig. l."3. The type specimen (Sc. Coll. Mus. No. 242), a fragment from the oscular margin of a large individual. Seen from the dermal side. Marginal lappets reflected outwards and backwards, thus showing the spiny gastral surface. About i nat. size. Fig. 18. Snhderrnal traheculfu, showing four thesocytes. Fixed with alcohol and stained with htematoxylin. 440 x . Sfaurocalypius entaccmthus ? Pp. 194 & 198. So. Coll. Mus. No. 403, which specimen is shown in these Studies, Contrib. III., PI. VI., figs. 9 and 10, mider the name of ,S'. japomcus. Fig. 19. Trabecular tissue, with four thesocytes, from along the lumen of an incurrent canal. Alcohol and hajmatoxylin. 440 x . Fig. 20. Chamber-wall, seen surface on. Well-stained archa^ocytes in small grou[)s. Alcohol and hajmatoxylin. 440 x . Fig. 21. Same in optical section. 440 x. StauroeaJypiKs rœperi (F. R. Sru.). 1^. KiS. All figures from F. E. Schulze's tyj)e-spe(;inien, obtained by th<' " Challenger." Fig. 20. Discoctaster. 300 x. Fig. 27. Oxyhexaster, from gastral side. 300 x . Fig. 28. Another oxyhexaster, in which each principal bears, besides two well- developed terminals, a minute rudiment of a third terminal. 300 x . Fig. 29. Microdiscohexaster from gastral surface. 300 x . Fig. 30. Pentactinic dermalia. 300 x . Fig. 31. Stauractinic dermalia. 300 x . Fig. 32. Castralia. 300 x. Ijima, Hexictinellida. Contribution IV. Jour. Sei Coll. Vol. XVIII. Art. 7. PI. XIV. 14, 16, 17, 22—25. Staurocalyptus aßiiis Ij. 15, 18. St. entacanthus Ij. 'l9—21. St. entacanthus ? 26—32. St. roeperi {F. E. ScH.) UTH. &. IMP. THE TOKIO PPJfJTlNG CO. I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XV. Staurocalyptus glaber Ij. Plate XV. Staitrocaïyptfis (jJaher It. P, 207. Fig. 1. A large specimen attached to a dead Periphragella elisœ (Sei. Coll. Mus. No. 244). 1 nat. size. Fig. 2. Another attaclied to a specimen of a tnfaceous bottom (Soi. Coll. Mus. No. 398). A nat. size. Fig. .3. Two young specimens attached to a dead Chonelasma calyx. Nat. size. Figs. 4-6. Dermalia in different views. 300 x . Fig. 7. Gîistralia. Below, the free proximal ray. 300 x . Fig. 8. Microdiscohexaster. 300 x . Fig. 9. Discoctaster. 300 x . Fig. 10. Oxyhexaster. 300 x. Fig. 11. Portion of a longitudinal section throngh the wall of a yonng specimen. Above, the oscular edge. 30 x . Fig. 12. Small portion of the basidictyonal plate. 100 x . Fig, 13. One of the small basidictyonal masses, fonnd in abundance in the parenchyma of Sei. Coll. Mus. No. 361, — j-rubably basidicty- onalia belonging to a yonng brood which had fixed themselves, temporarily or otherwise, to the parenchymalia of the mother- sponge. 50 X . Ijima, HexactineUiâa. Contribution IV. 4. Jour. Sei. Coll. Vol. XVIII. Art. 7. PI. XV. 7. Staurocalyptus glaher Ij. LITH. 4 IMP. THE TOKIO PPJNTING CO. I. IJI1IA. STUDIES ON THE HEXACTINELLID.^. CONTRIBUTION IV. PLATE XVI. Stauixcalvptus pleorliapbides Ij. Plate XVI. Staurocalyptii-s plcorlioplddcs. 1j. P. 222. Fi"-. J. S. C. M. No. 226. A part of the wall cut off and the ga^tral surface exposed. Nat. size. Fif, 2. Another entire specimen. S. C. M. No. 415. Nat. size. Figs. 3-.5. Derinalia. .300 x. Fif. 6. Gastral diactin of a moderately large size. 300 X . Fig. 7. Normally developed oxyhexaster. 300 x . Fjo-, 8. Malformed oxyhexaster, observed but once. 300 x . Fig. Î), 10. Discoctasters of the larger size. 300 x . Fi«»'. 11. Part of a very small discoctaster. 300 x. Fiy;. 12. Microdiscohexaster. 300 x . Fig. 13. Part of the paratangential ray of a prostal pentactin, to show the roughness of surface caused Ijy minute and pointed microtu- l)ercles. 300 x . Fig. 14. Spicules in and un a conuhis of the sponge-surface. Two i)rostal, paratropal oxypentactins and a part of a prostal oxydiactin. About 30 X. Fig. 1.5. Surface view of the ectosome, with dermalia which are mostly diactins. 60 X . Fi<''. 16. Surface view of the eudosome, with diactinic gastralia. 60 X . Ijima, Hexactinellida, Cmtribidion IV. Jour. Sei. Coll. Vol. XVIII., Art. 7, PI. XV L 4 5 Staurocalyptus pleorhaphides Ij. Lith rf- Imp Th^ Tokio frintimi Co. I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XVII, Eliabclocalyptus victor Ij. Plate XVII. Ilhubducaljptas victor Ij. P. 238. Fig. 1. Sei. Coll. Mus. No. 423. Slightly le.ss than -i nat. size. Total length 880 mm. ; diameter at middle 220-270 mm. Ijima, HexactinelUda, Covfributimi IV. Jour. Sei. Coll. Vol. XVIII., Art. 7, PI. XVII Rhabdocalyptus victor Ij. LITH 4 IMP, THE TDKIO PRINTING CO I. IJIHA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XVIII Ehabdocalyptus victor Ij. Plate XVIII. Rhahdocalyptm victor It. P. 238. Fig. '2. A jiortion of tlie dermal surface, magnified about two diameters. Dermal lacework, blurred. Hypodermal strands radiating from several central points. Fig. 3. A portion of the gastral surface, magnified about two diameters. Gastral lacework supported on hypogastral strands, covering the excurrent canalar apertures. Fig. 4. A large, unequally rayed, parenchymal diactin from the l)asal region. 1.^0 x . Fig. .3. A heraihexactinose oxyhexaster with nine terminal points. 300 x . Fig. G, Portion of a discoctaster, with a primary terminal which has remained free, unfused wilh the secondary princip;il running by its side. 300 x . Fig. 7. A hexactinose oxyhexaster. 300 x . Figs. 8-11. Different forms of dermalia, the stauractinic form of fig. 10 being by far the most common. 300 x . Fig. 12. A discoctaster, with ]ilain central node. 300 x . Fig. 13. A gastral hexactin. 300 x . Fig. 14. A small })ortion of the basidictyonal plate. 1,">0 x . Fig. If). A hemihexactinose oxyhexaster with sev'en terminal ))oints. 300 x . Fig. 16. A small portion of the ectosome, seen from outside ; including a group of hypodermal pentactins with the paratangentials in a radial disposition. In the center of the gronp, a dense projecting tuft of fine diactins, 30 x . Ijima, HexdctinelUda, Contrihution IV. Jour. Sei. Coll. Vol. XVIII.. Art. 7, PI. XVIII. % \ I ^ 10 .IT fi.t ' -« " f «? - v < If- ' ^ > » V 4K0 'V>'ß -. K 'i*»^ .21 23 n. ,/• "-*•■' «/owr Sc/. Coll. Vol. XVIII., Art. 7, PI. XIX. '■■ 9 Jf', \ 4 : % 4 ^ > > c n "-? 5 f ",. "» -"T, « ! "1; ' >." -"^ ;2;!^ Rhabdocalyptus victor Ij. LM ä' Imp Tht Toki« Printimf Co. I. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XX. Kliabdocalyptus mollis F. E. Scii. Plate XX. lÜtabdocalyptufi 'mollis F. E. tScH. P. 253. P'ig. I. A s})ecimeu (0. C. No. 103) well preserved, exce[>t in that a bud has been torn off. ^ natural size. Fig. 2. A specimen (8. C. M. No. 420) iVoni which a triangular piece of the wall had been cut away. ^ natural size. (Figs. 3-9 were all taken from this specimen). Fig. 3. A normal oxyhexaster from the subdermal sjjace. 300 x . Fig. 4, Microdiscohexaster. 300 x . Fig. 5. ]3iscoctaster. 300 x . Fig. 6. A hemihexactinose oxyhexaster with 10 terminals in all, from deep parts, 300 x . Fig. 7. A liexactinose oxyhexaster; two of the straight rays each showing at base the remnant of an atrophied terminal in the form of a unilateral spine. 300 x . Fig. 8. A quite liexactinose oxyhexaster. 300 x . FiiX. Î). Another liexactinose oxvhexaster with all the ravs crooked at base. 300 x . Fig. 10. A nearly hexactinose oxyhexaster ; one of the rays is simply bent at base, while another similarly l)ent ray shows the rudiment ol' a fellow terminal belonging to the same ])rincipal. From 0. C. No. 104. 300 X. Fig. LI. A hemihexactinose oxyhexaster with very strongly developed basal barbs to the rays. From 0. C. No. 104. 300 x . Fig. 12. »Spicules of ectosome seen in surface view. From ß. C. M. No. 420. 30 X . Fig. 13. Spicules of endosoiiie seen in surface view. From S. C. M. N(j. 420. 30 X . Ijima, Hexactinellida, Contribution IV. Jour. Sei. Coll. Vol. XVIII., Art. T^^l. XX. Rhabdocalyptus mollis F. E. ScH Litti A- Imp Tht Tokio Printimj Co. I. ijinA. STUDIES ON THE HEXACHINELLIDA. CONTRIBUTION IV. PLATE XXI. EhaMocalyptus unguiciilatus Ij. Plate XXI. EJiabdocalyptti.s anijaiculatus Ij. P. 268. Fig. 1. The tvpe-.specimeii (Sei. Coll. Rîus, No. 501). l iiatiinil size. AH the following figures of .s[)icnlc,s wore taken fronj thi« s[)ceinien. Fig. 2, Two dermal diactins. 300 x . Fig. 3. A gastral hexactin. 300 x . Figs. 4, .5. Thin-rayed oxyhexasters from Ihu i)eri|)liery of the sponge wall. 300 X. Figs. 6, 7. Stronger-rayed, smooth, hemihexactinose oxyhexasters from the gastral side of the sponge wall. 300 x . Fig. 8. Plexactinose oxyhexaster from the same region. 300 x . FiiT. !). Microdiscoliexaster from the dermal membrane. 300 x . o Fig. 10. A discoctaster. 300 x . Fi<ï. 11. Uno;uiculate terminal discs of discoctaster. One in lateral view ; two as seen from above. 070 x . Fig. 12. A small fragment from the basidictyonal plate. The small-meshed limiting layer and some basidictyonal hexactins in fusion with it as well as with one another. 100 x . Jjima, Hexactinellida, Gontribtition IV. Jour. Sei. Coll. Vol. XV III., Art. 7, PI. XXI. 2 ,.:i Bhahdocaly ptas unguiculatus I.j LITH 8.IHRTHET0KIO.PR1NTING CC T. IJIMA. STUDIES ON THE HEXACTINELLIDA. CONTRIBUTION IV. PLATE XXII. Eliabdocalyptus capillatus Li. Plate XXn. Uliahdocahjptus capillatus II. P. 276. Fig. 1. A specimen from Inside Okinosé l>v the Ena-line (S. C. M. No. 287). I nat. size. Fig. 2. Another from Outside Okinosé hy tlie Iwado-line (S. C. M. No. 397). J nat. size. Figs. 3, 4. Two very small specimens from a lot of six, all of which were found attached to a dead Chonelasma cahjx from Maye-no- Yodomi (8. C. M. No. 406). Nat. size. Fig. 5. A young specimen from a lot of three, found attached to a dead Chonelasma calyx from Mochiyania (S. C. M. No. 413). Nat. size. Fig. 6. Dermalia. 300 x. Figs. 7, 8. Oxyhexasters from sulxlermal s]iace. 300 x. Figs. 9, 10. Discoctasters. 300 x . Fig. 11. Microdiscohexaster. 300 x . Fig. 12. Wall of dried specimen in section. The arrows indicate incurrent and excurrent canals. 2 x . Fig. 13. Gastralia. 300 x . Figs. 14, 15. Oxyhexasters from subgastral space. 300 x . Fig. 16. Arrangement of the paratangential heads of hypodcrmal [»Mitactins in groups. About 4 x . Fiir. 17. Part of hasidictvonal i)late. 100 x. Ijima, HrxactineUida, Contributian IV Jour. Sei. Coll. Vol. XVIII., Art. 7, PI. XXII. IS u I BJiahdocalyptus capillatns fj Lith & Imp The Tokio I'livtinu Co. I. IJI9IA. STUDIES ON THE HEXACTINELUDA. CONTRIBUTION IV. PLATE XXIII. Rliabdocalyptus capillatus Ij. Plate XXIII. Rhabdocalyphis capillatus Ij. P. 27G. (AU figures from sections stained with borax-carmine). Fig. IS. Surface view of ectosome. 300 x . Fig. 19. A small part of the subdermal region. Below, blind end of a ciiamber and a small (yuimg) oxyhexaster. 300 x . Fig. 20. A small part of chamber wall. Below, three archreocytes. 1000 x . Fig. 21. Thesocytes of varied appearance. 1000 x . Fig. 22. Peripheral i>art of a section vertical to the surfaces. 30 x . Above, dermal surface ; s. s., subdermal space ; in., incurrent canal ; ex., excurrent canal. Fig. 23. Part of a paratangential section. 30 x . in., intercommunicating incurrent spaces ; ex., excurrent canal. Fig. 24. Gastral part of a section vertical to the surfaces. 30 x . Below, the gastral surface. Lettering; as in the above figure. Ijivia, HexactiiieUida, Cnntribution IV. Jour. Sei. Coll. Vol. XVIII., Art. 7, PI. XXIII. 18 20 21 ■1 t\ 23 22 Mk ex. f f*"^^.^ 7 ^ :. J_*^.,-^^ 2é ^ % 3 Rhabdocalyptus capillatus Ij. Litk exth. Fl. Hongk. p. 169 ; Clarke, Comp. Ind. p. 18 ; He.msl. in Voy. Chall. Bot. I, part o, p. 159 ; Hook. f. COMPOSURE FORMOSA X.E. 5 FJ. Brit. Ind. Ill, ]>. 235; Fokb. et Mkmsi.. In.l. Fl. Sin. T, p. 401; HknrYj List PI. Formosa, |>. öl. Centratlierum cliineiist.^ Less, in Linnii-a, I\ , p. o2i). Cijanoph pnhescens DC. Prodr. A", p. <)9. Conyz'i patida Willd. Sp. PI. Ill, p. 1!)11». Hab. Hüsoan, Shinshoa, leg". H. Kawaka.mi, anno 181»') ; Taipea, leu". T. Makixo, anno 189G ; Pachina, leg. T. XiiXAMi, anno 1897 ; ipse, ann(j 1900. DiSTRiB. ]\Ialaya peninsnla et ins. Philipina?. 3. V. cinerea Li>^- in Linniea, I\', i>. 291, et M, p. (iTo; DC. Prodr. \, p. 24 ; :\IiQ. Fl. Ind. Pat. II, p. 11 ; Bkxth. Fl. Ilongk. p. 169 ; HooKKK, Niger Fl. p. lo4 ; Gkisebach, Fl. Prit. West Ind. Isl. p. o5o ; Dex'JII. Fi. Austral. Ill, p. 459 ; Ceakke, Comp. Ind. p. 20 ; Oliver, Fl. Tropic. Africa, p. 275 ; Hook. f. Fl. Brit. In.l. Ill, p. 235 ; Hemsl. in Voy. Cliall. I. ])art 3, p. 159 ; Hielehraxd. Fl. Hawaiian Isl. p. 191 ; Fork, et Hemse. Ind. Fl. Sin. I, ]>. 4<»I ; Henry, List PI. Formos. p. 51 ; Diels, Fl. Centr. China, in Fxol. Bot. Jahrb. XXIX. (191)1) p. 608. Hab. Shintekn, leg. T. AIakexo, ann<;) 189() ; Taipea, leg. C, OwATART, anno 1897 ; Kuclui. Shintem-koe, leg. K. MiVAKE, anno 1899; Senton, ipse, anno 1900; Pacliina. leg. T. XllXAMl. DlSTUlB. Asia tropica, Africa et Australia. 2. Elephantopus Lixx. Conspectus Specie rum. Capitulorum glomeruli loiige pedunenlati, laxissimc corymbosi...!. E. saiher. Capitnlonim glomeruli secus ramos fioridos elongatös sessiles ...2. K. spiaittix. AKT. 8.-B. HAYATA. 1. E. SCaber J-inn. Sp. PJ. ed-i^ p. lolo ; Koxb. Fl. lud. p. 445; Li-:88. Synop. Comp. p. 149; DC. Prodr. Y, p. 8() • Hooker, Niger FJ. p. 429; Miq. Fl. Ind. Ixit. II, p. -Jl ; Wight, Ind. Ic. ri. Ind. or. t. 108G ; Benth. FJ. Hongk. p. 170, et Fl. Austral. Ill, p. 4(U ; Gi'JSEbach, Fl. Brit. West. Ind. Isl. p. o54 ; Clarke, Comp. Ind. }>. 28; Oliver, Fl. Tropic. Africa, p. 299 ; IIemsl. in Ijiol. Central- America, j^ot. II, p. 70 ; Hook. f. FL Brit. Ind. III, p. 242 ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 4(32; Henry, Li.st PI. Formos. p. 51. Hab. Ho-soan, leg. H. Kawaka.mi, anno 1895 ; Shintekii, Taipea, \eg. T. Marino, anno 1896; Kierung, ]3iyoritsi» \eg. K. Honda, anno 1897; Ku-cliö, Shintem-koé, leg,. K. ]\Iiyake ; Senton, ipse, anno 1900. JJisTRiB. Asia troj)ica, Africa, et Australia. 2. E. spicatus P. Juss.; Less. Synop. Gjinp. p. 149; DC. Prcjdr. V, p. 87; Grisebach, FI. Brit. West. Ind. Isl. p. 355; Hemsl. in Biol. Central- Americ. II. p. 7() ; Heniîy, List PL Formos. p. 51. Hab. Linga-riyao, Sui-cho-rü, Pak-kan-koe, leg. C. O'vVATARi, anno 1898; Daito, Paikokuwa, Koho, ]eg. K. MiYAKE, anno 1899. DiSTRiB. America tropica. 3. Adenostemma Forst. 1. A. viscosum Forst; Less. Synop. Comp. p. 15(j ; DC. Prodr. y, p. 111 ; Sieh, et Zucc. Fl. Jap. Fam. Xat. p. 181; IJENTH. Fl. Hongk. p. 171, et Fl. Austral. Ill, p. 4(i2 ; Seemann, FI. Vitiensis, p. 140; Franch. et Sav. Enum PI. Jap. I, p. 219; Cj.arkEj Comp. Ind. p. 28 ; Ojjver, Fl. Trot. I, jxirt o, p. 159 ; :\[iQ. FJ. Ind. Hat. ir, p. 23; Hillkbrand. FJ. Hawai. IsL p. 192 ; Maxim, in Engl. Jalirb. VI, p. 08 ; Hook. f. Fl. Brit. Ind. Ill, p. 242; FoKB. et Hkmsl. Ind. Fl. Sin. I, p. 403; Henry, List PI. Formos. ]). 51 ; DiELs, Fl. Centr. China, in i:ngl. Jahrb. XXIX, p. G08. Spilanthes tiuctorius LoiK. Fl. Cochinch. ed-2, p. 484. Icon. Jap. Somokudziisetz, vARKE, Comp. Ind. p. 30; Oliver, Fl. Tropic. Africa, III, p. 300 ; Hemsl. Voy. Chall. Bot. I, p. 40, et part 3, p. 159 ; et in Biol. Central.-Americ. II, p. 81 ; A. Gray, Synop. Fl. X. Americ. ed-2, I, part 2, p. 93 ; Hillebrand, Fl. Hawai. Isl. p. 193 ; Hook. f. Fl. Brit. Ind. Ill, p. 243; Forb. et Hemsl. Ind. Fl. Sin. I, p, 403 ; Henry, List PI. Forms, p. 51. Hab. Taipea, leg. T. Makixo, anno 1896, et C. Owatari, anno 1897. DiSTRiB. In calidis regionibus universaliter dispersa. 8 ART. 8.— B. HAYATA. 5. Eupatorium Li\n. Conspectus Spccieruni . Canlis erectus. Folia uou partita. Folia saepissime ovata. luvolneri l)rac-teie obtnsae. Folia basi rotundata 1. h'. Ilei'ie-iii. Folia basi angiistata 2. K. japauiciDn. Folia oblongo-lanceolata. luvolneri bractejTe acutîe..3. i'A fj'iiillei/dinnii.. Folia tri-partita i. K. Cltincnse rar. tn'/nntituni. Canlis scandens 5. /•,'. Trishiini. 1. E. Reevesii Wall.; DC, Pnxlr. V, p. 170; Hook, et Aux. Bot. Beech. Voy. p. '267; Iîenth. Fl. Hoii,ü-k. p. 17 '2 ; Clauke, Comp. Ind. p. 33 ; Hook. f. Fl. Brit. Ind. [11, p. 213; Foiin. et Hemsl. Ind. Fl. Sin. I, p. 405 ; Henry, List Fl. Forrnos. p, öl, Hab. Loco non indicato. DiSTRiB. China et Japonin, 2. E. japonicum Thunb. Fl. dap. p, 30S; DC. Prodr. A', p. 180 ; Sieb, et Zucc. Fl. Jap. Farn. Xat. p. I-Sl ; Miq. in Ann. Mus. Bot. Lugd.-Bat. 11, p. 107; Franch. et Sav. Enuni. Fl. Jap. I, p, 21i>; Franch. PI. David, p. 1()0 ; Maxlni. in Engl. Jahrb. \'l, p. (iS ; FoRB. et Hemsl. Ind. Fl. Sin. I, p. 403; Hexry, List PI. Formos, p. 51 ; Diels, Fl, Cent. China, in Engl. Jahrb. XXÏX, p. (iOS. Eupatorium. U'allichii DC. Prodr. V, p. 179; Iîexth. Fl. llongk. p. 172 ; Haxce in Jonrn. Linn. Soc. XIII, p. 100. Jap. Icon. Somokudzusetz, vol. 15, fol. 5(); Honzodsnt'u, vol. 11, fol. 14 recto. k\va-wi, herb., vol. 3, fol. 14, Hab, Daitocho: Daikako, leg. K. Mivake, anno 1891). DiSTRiB. Japonia, China, et montosis Indiie borealis region i bus. COMPOSITE FOEMOSAX^E. 9 3. E. Lindleyanum DC. Prodr. V, p. 180; Bentil, Fl. Hongk. p. 172, et Fl. Austral. Ill, p. 462; FoPxB. et Hemsl. Inot. Lugd.-i^at. II, p. 167 ; Forr. et Hemsl. Ind. Fl. Sin. I, p. 405. Hab. Taipen, Tamsui, Kierung, leg. T. Makixo, anno 1896; Kucbu, Shintemkoe, leg K. Miyake, anno 1899. 5. E. Tashiroi Hayata, sp. nov. cum tab. I. SuttVutices; caulis scandens, teres, laevis, ramosus, ramis gracilibus divaricatis. Folia opposita, petiolata, tenuin, glabrn, lanceolata, v. ovato-lanceolata, acuminata, basi obtusa v. rotundata, ■dentata, dentibus ascendentibus, acutis v. mucronatis, o-nervia, 4-S cm. longa, 2-3 cm. lata, petiolis brevibus (l-i^ mm. longis. Panicula) laxe "Corymbosa?, pedunculis o-5, remotis, oppositis, 2-4 cm. lorjgis, pedicellis 1-1 1 cm. longis, capitulum bis a:}quantibus, pubescentibus. Capitula 5-flora. 1 cm. longa. Involucrum anguste campanulatum, bracteis elongato-lanceolatis, 1-2-seriatis, siccis, scariosis, in- terioribus 5, 5 mm. longis, exterioribus 5-6, minoribus, additis o-4-bracteolis, minimis 1^ mm. longis. Keceptaculum minimum €onvexum, nudimi. Corolhc 4 mm. longa?, tubo tenui, limbo anguste campanuhito 5-dentato. Styli rami longissimi, 3 mm. longi corollam valde excedentes, superne compressiusculi. Achaenia glabra, nigra 5-costata, costis prominentibus sub-rugosis, 2^ mm. longa, apice truncata. Fappi setie l-seriatie co , rigidie, scabra', 3 nmi. longii?. Ab F'i. iresinoides Kth. diflert foliis glaberrimis. jO ART. 8.-B. HAYATA. Hab. Sliarioto, leg. T. Making, anno 1896 ; prope Taito, leo-. Y. Tashiro, anno 1897 ; Kuchu, leg. K. Miyake, anno 1899; in montibus Morrison 5000 ped. alt., ]eg. II. ToRii, anno 1899; Senton, ipse, anno 1900. 6. Mikania Wju.d. 1. M. SCandens Willd. Sp. PI. Ill, p. 1748, Less. Synopsi. Coni}). J). 157 ; DO. Trodr. \, p. 199 ; Griskbach, FJ. Brit. West Ind. Isl. p. 364; Cj.ahke, Comp. Ind. [). 84; Oliver, Fl. Tropic. Afric. Ill, p. 301 ; PIook. f. Fl. Brit. Ind. Ill, p. 244 ; Hemsl. in Biol. Oentral-Americ. IV, p. 104 ; A. Gray Synop. Fl. X. Americ. ed-2, I, part 2, p. 94 ; Furb. et Hemsl. Ind. Fl. Sin. I, p. 405 ; Hexuy, List PI. Formos. p. 59. Eiipatorium scandens LiNX. Sp. PI. ed-2, ]). 1171. Mikania vohihilisV^UAA). Sp. PL III, p. 1743; Miy. Fl. Ind. Bat. II, p. 28. Hab. Taichu-ken, leg Y. Tashiro, anno 1897 ; Botansha, les: K. Miyake. anno 1897. DiSTRiB. In Asia, Africa et America, late diit'usa. 7. Solidage l INN. 1. S. Virga-aurea Lixx.; Tiuxk. Fl. Jap. p. 317 ; Willd. Sp. PI. p. 2065 ; Less. Synop. Comp. p. 163 ; DC. Prodr. V, p. 338; Benth. Fl. Hongk. p. 179; Sowerby, F]nglish Bot. (1873) t. 1278 ; Franck, et Sav. Enum. PI. Jap. I, p. 228 ; Clarke, Comp. Ind. p. 35 ; Hook. f. Fl. Brit. Ind. Ill, p. 245 ; K. COMPOSITE FORMOSAX^. |q^ MiYABK, Fl. Kuile Isl. ill Mem. üo.st. Soc. Xat. lli.vt. I\^, (1890) p. 1^0 ; FuRB. et IIkmsl. Iiid. Fl. Sin I, p. 406 ; Hknky, List PL Forms. ]». 52 ; A. Gray, Synop. Fl. X. Americ. e.1-2, I, part 2, p. 148. aS'. cdtitoniensis et S. decurrens Lour. Fl. Cocliincli. '2-t'd. j). (512, DC. Pro.lr. \, pp. 341-2. Icon- Jap. Somokudsnsetz, vol. 17, fol. 40 ; HijndsodsufiL, vol. 14, fol. 15, recto. Hab. Tamsui, leg T. Making, anno 1896 ; et K. Xiixami anno 1899 ; in montibus Morrison 5000 pe-l. alt., leg, K. ToRii, anno 1899; Sa-bo-san-sho, ipse, anno 1900. DiSTEiB, Europea, Asia temperata, Jap«jniu er America boreal is. 8. Dichrocephala dc. 1. D. latifolia DC. Prodr. V, p. 372; Miq. Fl. Ind. Pat. [I, p, 37; Wight, Ic. PI. Ind. or. t. 1096; Benth. Fl. Hongk. p. 173; Fk. et Say. Eniim. PI. Jap. I, p. 229 ; Sonder, Fl. Capens. Ill, p. 115 ; Clarke, Comp. Ind. p. 36 ; Oliver, Fj. Tropic Afric. Ill, p. 303 ; Hook. f. Fl. Brit. Ind. ID, p. 245 ; Maxim, in Engl. Jahrb. VI, p. 68 ; FoRB. et Hemsl. Ind. Fl. Sin. 1, p. 306 ; Henry, List. PL Formos. p. 52; Diels, FL Central-Chin, in Engl. Jahrl). aXIX, p. 115. Hab. loc«3 non indicato. DiSTRiB. Asia tropica snbtropicave et Africa. 12 ART. 8.— B. HAYATA. 9. Grangea adaxs. 1. G. maderaspatana Poiu.; Less. Synop. Comp. p. '202 ■ DC. Prodr. V, ]). 373; Miq. FI. Ind. I^>at. II, 39; Bextîi Fl. Hongk. p. 185; Wight, Ic. PI. Ind. or., t. 1097; Clark. Comp. Ind. p. 37; Olivkij, Fl. Tropic. Afric. III, p. 304 ; Hook. f. Fl. l'.rit. Ind. III, p. 247; FoRB. et HvMsi.. Ind. Fl. Sin, I, p. 407 ; Hi:xry, List. PI. Form OS. p. 52. Hab. Hosoan, Sensoreisho, IL Kawakami, anno 1895 ; Tai-cliii-ken: Ho-ri-sha, leg. Y. Tashiro, anno 1897 ; ])0-riyo. Sin-ye-slia, leg. C. Owatari, anno 1898 ; ïai- chu-ken: Daibo-ho-soo, leg. H. Satake, anno, 1899. DrsTHiB. 'ri-o[)ica Africa et Asia. 10. Aster Li XX. CoHspeciiis Specici'fdii. Folia margiue non ciliata. Folia lineavia 1. J. altniciis. Folia ovata v. ovato-lanceolata. Pappus copiosus. Capitula longe pedunculata 2. A. trinerrius. Capitula feœ sessilia 3. A. luacharoides. Pappus 4. A. iiuliciis. Folia mavgine duplicato-ciliata 5. A. (^hUninii. 1. A. altaicUS Willd.; Xeks. Gen. et Sp. Aster, p. 22S ; Clarke, Comp. Ind. p. 4(5 ; LIuok. f. Fl. lîrit. Ind. HI, ]). 251 ; Franck. PI. David. \). 161 ; Foub. et Hemsl. Ind. Fl. Sin. I, p. 408. COMPOSIT.E FORMOSAN.E. 13 Caliweris altaica Xees, Gen. et Sp. A.ster. p. 228 ; DC. Piodr. Y. p. 258. Calimcns ciliata A. Gray, Bot. Jap. p. o94. Aster striatus Fr. et Say. Emud]. PI. Jap. I, p. '2'2'2. Hah. Manka, leg. T. Making anno 1896 ; Biyo-iin, leg. Y. Tashiro, anno 1897 ; Taito-cbo: Suibi, Periinir, Cliihon, leg. K. ]\Iiyake, anno 1899. DiSTRiR. in montibiis Altaii, ^Mand.sliuria et Japonia. 2. A. trinervius PtoxB. Fl. Ind. Ill, p. 438; Nees, Gen. er Sp. Ast. p. 17o ; Bexth. Fl. Hongk. p. 174 ; Fr. et Say. Enuni. PI. Jap. I, p. 222 ; Hook. f. Fl. Brit. InrsL. Ind. Fl. Sin. I. p. 414 ; Henuy, List PI. Formos. p. 52. Hab. Senton, ipse, anno 1900. DisTUiiL Phuita endemica. a. Erigeron linx. 1. E. linifolius Willd. Sp. PI. Ill, p. 1955; Benth. Fl. Hongk. ]). 17(î, et Fl. Austr. Ill, p. 496; Clarke, Comp. hid. p. 50; Hemsj.. \"oy. Chall. Jîot. I, part I, p. 42 ; A. Gray, Synop. Fl. X". Americ. ed-2, I, part 2, p. 220 ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 41S ; Hexrv, List PI. Formos. p. 52; Diels, Fl. Central-China, in Engl. P.ot. Jahrb. XXIX, p. 612. Coiiii:a aiiilii jiKi DC. Prodr. A . p. 381. COMPOSURE FORMOSAX^. I5 Hal). Tuipea, leg. T. Makixo, anno 1896. DisTRiB. Iridiges Mediterraneae regionls, nunc late colo- catas. 12. Conyza les«. Com^pectus Spccii'rwn. Folia oblongo-ovata 1. '_ '. Jdpnitici. Folia lanceolata pinnatilida 2. C ae;ii/jitica. 1. C. japonica Less.; DC. IVxIr. A', p. 382 ; Sieb, et Zicc. Fi. Jap. Fani. Xat. p. 18-t ; Fu. et Sav. Enum. PI. Jap. T, p. 229 ; Clarke, Comp. Ind. p. Go; Hook. f. Fl. l^rit. Ind. Ill, p. 258; Forb. et Hemsl. Ind. Fl. Sin. T, p. 419 ; Henry, List PI. Fornios. p. 52 ; DiELS, Fl. Centr.-Chin. in Engl. Jahrb. XXIX, p. 612. ComjKi vt'rnonicarfolia Wall. DC. Prodr. V. p. 382; Benth. Fl. Hongk. p. 176. Hah. Loco non indicato. DiSTRiii. Japonia, China centralis, India horeali.s. 2. C. aegyptica Arr. Hort. Kew. ed-2, X, p. 2S ; DC. Prodr. V, p. 382 ; P.ENTH. Fl. Austral. Ill, p. 497 ; Oliver Fl. Tropic. Afric. Ill, p. 314 ; FIanxe, in Journ. l^ot. (1878), p. 108 ; Ho(..k. f. Fl. Brit. Ind. Ill, p. 258 ; Forb. et Hemsl. Ind. Fh Sin. I, p. 419 ; Henry, List. PI. Foi-nios. p. 52. Hab. Sinteku, leg. T. Maklxo, anno 1896. Kotosh«^ leg. Y. Tashiro, anno 1897 ; Taichu-Ken : prope Tc-run- gai, leg. Y. Tashiro, anno 1897. DisTUii!. Tropica Asia, Africa, et AustraHa. ](3 ART. 8.— B. HAYATA. 13. Microglossa dc. 1. M. volubilis DC. I^-odr. Y. p. 320; MiQ. Fl. Tnd. Bat. II, p. 34 ; Claeke, Comp. Iiiot. lîeecli. \o\. p. 195. Hfib. Chuko, leg. K. Honda, anno 1S97. DiSTUiB. Japonia, China anstralis. 2. W. calendulacea Ia-.s<,. Synops. Comp. p. 222 ; DC. Prodr. V. p. 5P)9 ; MiQ. Fl. Ind. V>o\. H, p. 08 ; Wight, Ic. PI. Ind. or. t. 1107 ; Iîfxth. Fl. Hongk. p. 182 ; et Fl. Au-stral. Ill, p. 537 ; FuAX'CH. et Sav. Ennm. 1^1. Jap. I, 2o;) ; Clakke, Comp. Ind. ]>. 136 ; Maxim, in Engl. Jahrb. \\, p. 68 ; Fohb. et Hkmsl. Ind. Fl. Sin. I, p. 434 ; Hexry, List. PI. Formos. p. 54. Verhesina calendulacea Lixx; l^ori;. Fl. Cochinch. ed-2, p. 619 ; HooK. et Arx. Bot. Beech, p. 265. Ic. Jap. So-niokudzusetz vol. 17, fol. 3(). Hab. Sensoreisho, leg. H. Kawakami. anno 1895 ; Keirung leg. T. Makixo anno 1896; Biyo-zoku, leg. K. Honda, anno 1897 ; Pachina, leg. T. Xiixami, anno 1899 ; Pi'ope dai-to, Kwarenko, leg. K. Miyakk, anno 1899 ; Senton, ip-se, anno 1900. DiSTHiB. In tropica Asia late dispersa. 3. W. biflora Bknth. Fl. ll.^ngk. p. 183; et Fl. Austral. Ill, p. 539 ; Claiîke, Comp. Ind. p. 137; Olivek, Fl. Tropic. Afric. Ill, p. 376; Hook. f. Fl. lîrit. Ind. Ill, p. 306 ; Haxce, in Journ. Linn. Soc XIII, p. 108 ; Hemsl. Voy. Chall. Bot. I, part 3, pp. 160, et 241; FoBB. et He.msl. Ind. Fl. Sin. I, p. 433; Hexry, List PI. Formos. p. 54. Wolladonia hißnr,i DC. Prodr. V, p. 54(). .Miq. Fl. Ind. l>at. II, p. 70. Hab. Ho-soan, Shinshoa, leg. H. Ka\vaka.mi, anno 1896. Bivo-ritsu, leg. K. Hoxda, anno 1897, Shario-to, leg. 2D ART. 8 — B. HAYATA. T. ^[akixo, armo 1896 ; et K. ^Iiyake, anno 1899 ; Daito-cho : Biyoko-sha, leg. K. Miyake, anno 1899; Kieriing, ipse, anno 1900. DiHTRiR. in niaritimis regionibns India? et Malayee. 18. Spilanthes linn. 1. S. Acmella Lixx. ; Thlnb. FJ. Jap. p. 321 ; DC. Prodr. V, p. G2?> : MiQ. Fl. Ind. I'.at. II, ]\ 79 ; Clarke, Comp. Ind. p, 138 : Oliver, Fl. Tropic. Afric. Ill, p. 384 ; Ho.^k. f. Fl. l^rit. Ind, 111. p. :'07 ; HiNiiv, List ri. Formos. p. 54. llali. Taiclui-ken : Chigito, leg. Y. Tasiiiro, anno 1897; Daito: Raiko-kwa, leg. K. jMiyake, anno 1899. DiSTRiiî. üeneralliter in calidis reoionibu.«. 19. Bidens lixx. Conapecfuti Specicnini. Flos albus 1 . J>. pilosd, Flos tlavus '2. a. hi/iiiuuita. 1. B. pilosa Finn. Sj*. 1M. e : Pextii. Fl. Hongk. p. 183 ; et Fl. Austral. III. p. 543 ; Grisebach, Fl. West. Ind. Isl. p. 373 ; Ci,AUKE, Comp. Ind. ]>. 140 ; Oliver, Fl. tropic. Afric. III, p. 393 ; Hemsl. in Piol. Centr.- Americ. II. ]>. iOl ; Fkaxch. PI. David, p. 1G5 ; A. Ghay, Fl. Synop. 44. X. Americ. ed-2, I, part 2. ]). 297 ; Fonn. et Hemsl. Ind. Fl. Sin. I, ]>. 434 ; Hexrv, List PI. Formos. ].. 54 ; Diels, Fl. Centr. Chin, in Engl. Jahrb. XXIX, p. GIG. Bidens mil/ichii DC. Pr(j. 438 ; IIkxüy, List V\. Formos. p. 54. Hab. Taipea, fculta.) S. Xagasawa, anno 1902. DisTUiiî. In Japonia culta. I.K-^s. 23. Myriogyne l 1. M. minuta Less, in Linnaea VI, j». 219; DC. Prodr. Y J, p. COMPOSITE FORMOSAN-E. 2S 139; SiEB. et Zucc. FJ. Jap. Fain. Xat. p. 187; 1>i:xth. F^J. Hongk. p. 18G ; et FJ. Austral. Ill, p. 553 ; Skk.mann, Fl. Vit. p. 144 ; Fiî. et Sav. Eniim. PI. Jap. I, ]). 241; Clarke, Comp. Ind. p. 151 ; Forb. et Hemsl. Ind. F1. Sin. I, p. 440 ; Henry, List PJ. Formes, p. 54. Centipcda orbicularis Lour. FJ. Cochincli. ed-2, p. 602 ; ^Iiq. FJ. Ind. liât. II, p. 89 ; Franxh. PI. David, p. 1G7. Spliacromorpliaca Centipeda DC. \I, p. 140. Jap. Icon. Somolvudsusetz, voJ. 1(3, fol. 29. Hab. Sensoreisho, Xaihosha, leg. H. Kawakami, anno 1895; Kieriing, leg. T. Makix<^, anno 1896 ; Shu-shu-gai, liinpoho, leg. C. Owatari, anno 1898. et leg. '[\ XiiXAMi, anno 1899. DiSTRiB. India, Malaya, Australia et Japonia. 24. Artemisia l. Conspectus Spccieruin. Folia capillaria 1. J. cajjillaris, FoJia [11-ofunde piunatifitla 2. A. cuhjaii^ cur. indkcu 1. A. capillaris Thuxb. Fl. Jap. p. 309 ; DC. Prodr. VI, p. 126; Maxim, in Mel. Liol. VIII, p. 524; Fr. et Say. Enum. Y\. Jap. I, p. 239 ; F^)RB. et Hemsl. Ind. Fd. Sin. I, p. 442 ; Henry, List PI. F'omos. p. 54. Icon. Jap. Somokiidsusetz, vol. 16, 28 ; ILjndsodsufu vol. 14, fol. 2, rect. Hab. Pikaku, leg. T. Makixo, anno 1896. Daitocho : Pinnang, Lodvo-tzun, Kuchu, Sliintenkoe, leg. K. Miyake, anno 1899. DiSTRiB. Mandshuria, Kamtschatka, et Japonia. 24 AKT. 8 — B. HAYATA. 2. A. vulgaris L. Var. indica Maxim. Trim FJ. Amur. p. 536 ; Fr. et Sav. Ennm. PI. Jap. I, p. 259. A. indica Willd. Sp. Fl. Ill, p. 1S4G ; DC. Frodr. VI, p. 114. Hab. Taipea, Tamsid, Sintekku, leg. T. Makino, anno 1S96; Fansini, leg. K. Hr)xi)A, anno 1EXTii. Fl. Hongk. p. 189 ; H.h.k. f. Fl. Prit. Ind. Ill, p. 335. Gijniira Indhosa Hook, et Aiixx. Pot. lîeecli. A'ov. p. 194 ; DC. Prodr, A^I. p. 301. Hal). Sha-ryo-to, leg. T. Makixo, anno 1596 ; Gi.sliun, Garanbice, leg. Y. Tashiro, anno JS97; Daito-clio: Piyo-ko-sha, \k^g. K. ^Iiyake, anno 1899. DiSTHiB. in Cliina et India. 2. G. bicolor DC. Prodr. VI, i>. 299 ; MiQ. Fl. Ind. Pat. II, ].. 99 ; Hook. f. Fl. Prit. Ind. Ill, p. 335 ; F..ul'.. et Hemsl. Ind. Fl. Sin. I, p. 447 ; Hexky, List 1*1. Fornios. ]•. 55. Hal). Kwatana, leg. H. Kawakami, anno 1895; Taiko, leg. C. 0\YATAia, anno 1897 ; Daito : prope Kwarenko, Kai-ko-kwa, leg. K. ]\Iiyake, anno 1899. DiSTRiB, ^Molucca et Japonia. 3. G. elHptica Y a be et Hayata, cam tab. II. Plerba ; cauli.s glaber, striatum, erectas, crassiusculus. Folia magna, petiolata, elliptica, v. oYato-elliptica, apice acuta v. obtusa, basi leviter attennata, subintegra, y. remote serrata, 5-8 cm. longa, petiolis 2 cm. longis,auriculis,di.stincti-:,maJLisculis, grosse pa iici-dentatis,pinninerYia, costis secundariis ntnxpie latere 5-6. Corymbus laxus. 5-6-cephaliis, 26 ART. 8.-B. HAY A TA. pedicelli.s brovibus i— H cm. longis. Ca|)itiil:i boiiiogaina cyliii(lric<3- caiin'aunlutîi, l-J cm. louga, Üoribus (jmnibns o fertilibus. Iii- V(.)]ucrum sub-campanulatum, bracteis .sub-2-seriati.s, interioribiis 1 1^-1 0-] ill vlli.s, lineari-laiiceolatis, acutis, apice margiiieque scariosis, I cm. longis, aequalibus, plus miiius cohaerentibus, additis exterioribus parvis h cm. loijgis. Keceptaciilum planum breviter timbrilliferum. Corollae tiavae tenuiter tubulosae limbo parum ampliatena. 27. Emilia (\\s. 1. E. SOnchifolia DC. Prodr. V[, p. ;502; Miq. Fl. Ind. Kat. U, p. 101 ; ILxiKKii, Fl. Xiger p. do!) ; (tUISEBach, Fl. lîrit. West Ind. Isl. p. 3S1 ; Fii. et Sav. Enum. PI. Jap. I, p. lM5 ; Clarke, Comp. Ind. p. 174; Oliver, Fl. Tropic. Afric. Ill, p. 405 ; Hemsl. Voy. Chall. Pot. I, part o, p. 161 ; Maxlm. in Engl. Jahrb. VI, p. (][) ; FoRB. et Hemsl. Ind. Fl. Sin I, p. 440 ; Henry, List PI. Formos. p. 55; Diels, Fl. Centr. Chin, in Engl. Jahrb. XXIX, p. (519. Hab. Sensoreisho, leg. K. Kawaka?\ii, anno li()]. A'lII, p. 15 ; Lei)ei5. Fl. lîos.s. 11, p. ()4G ; Fu. et S.w. Euuni. PI. Jap. 1, p. 251 ; FuAXCH. FJ. David, p, 174 ; JiAKEU et MooKE, in JoLirn. Linn. Sue. XVll. ]>, o83 ; Fokb. et Hkmsl. IîkI. Fl. Sin. I, p. 450; Dikes, FJ. Centr. Chin, in Engl. Jalirb. XXIX. p. G20. Senecio Fierotil Miq. in Ann. Mu.s. l)Ot. Lugd.-But. 11, \\ 182. Jap. Icon. Somokudsusetz, XYII, fol. 10. Jiab. Taichud^en : Taiko, leg. Y. Tashiuo, anno lSi)7. DiSTRiB. Europea occidentalis, Siberia et Japonia. 2. S. Kaempferi DC. Prodr. VI, p. 363 ; Maxim, in Md. Biol. VIII. p. 14; Fk. et Sav. Enum. 11. Jap. I, 247 ; Maxim, in Engl. Jahrb. ^'I, p. G9 ; Forb. et Hemsi.. Ind. FJ. Sin. I, p. 454 ; Hkxry, List. PI. Forinos. }). 55. Fur/iKjiuni Kaempferi lîKXTii. Fl. Hongk. \\ DJl, Ligularia Kaempferi Sieb, et Zucc. Fl. Jap. I, p. 77, t. 35 et 36 ; J]ot. Mag. t. 5302. Ligularia gigantea Sieb et Zucc. ¥\. Jap. I, p. 79. t. 36. Jap. Icon. SomokudsLisetz, XVII, fol. 27, Hondsodsul'u, XVIII, fol. 8 et 9. 28 AUT. 8.— B. HAYATa. H:>b. Kieruiig, leg\ T. Makixo, aunu 189() ; et leg. C. UwATAKi, aiiiio 181)0; Kn-choo, Shinten-koe, Peruiig, Tuito, Keibi, ShiiikoKae, leg. K. ^Iiyake, anno 1899 ; in niontibns Morrison 5000 ped. alt., leg. II. Touii, anno 1899; 'raicliuken : Taiko, leg. Y. Tasiiiro, anno 1897. DjsTKIB. Japonia, Mandsliuiia. 3. S. nemorensis L. var. octoglossus Lkdeb. Fl. Ross. II, p. ()41. S. nclufjlossiis D(\ I'rodr. \ 1, ]•. o54. Hab. ill montibns ^lorrison ôOOO ped. alt., leg. Iv. T<»Kii, anno 1,S99. DisTh'iiî. in Siberia, (ierniania et Japonia. 4. S. Krameri Fu et ;Sav. Enum. l*l. Jap. I, p. 248. flap. [con. Soniokudsusetz, X \ I, fob 11 ; Hc^ndzodsufu, vol. ,S, fob 14. liab. liivosoku, leg. 'W I'ashiuo. l)isri;iiî. Japonia. 29. Blumea do. Conspectus Specierinii. O Capitula parva, -J ciu. — -l ciu. in diaiuetro ao(|uantia, dense paniculato-coi-ymbo.sa, saepe in glomenilos contractos disposita. □ Pappus albus. Z\ Folia glabrata 1. li memhiaiutcea. z^Z\ Folia villosa v. sericea t Capitula in paniculas spiciformes dense dis- posita 2. B. lucent. COMPOSIT.E FOKMOSAX.E. 29 if Capitula in glomernlas axillares disposita 3. fj. si'iiaiiix. DD Pappus rnfns 4. 11. hal.smnifera. CO Capitula magna, 1 cm. — li cm. in tliametro aequantia, laxe paniculato-coiymbosa. D Pappus albus. A Caulis scandens , ö. />'. cliiiit'iisis. AA Caulis flexnosus ß. H. wahiJunicd. □ Pappus rnfiis 7. /?. s/icct'iliili.s. 1. B. membranacea D^'. Pi-o-lr. W p. 440; Hook. f. FI. ]-rlt. Tnd. Ill, p. i(i5 ; Henry, Li.st. PI. Form. p. 52. Conij:a difusa Roxb. Fl. Ind. HE, p. 42!l, fide Hook. f. Hal). Xiki, Siiichorü, leu'. (\ ()\v.\rAiii. anno ISOS; Koto.sho, prope Gishun, ]eg. K. Mivakk, anno ISOîi. DisïiîiB. Malaya Peninsula. 2. B. lacera P>G. Prodi-. V. ]». 4;)(: : r.EN-rn. Fl. Hongk. p. I7S ; Hook. f. Fl. Prit. Ind. Ill, p. 208 ; Four., et Hemsl. Fnd. Fl. Sin. T, p. 21 ; Heniîy, List PI. Formos. ]>. 52. Hab. Hôsoan, leg. H, K.vwakami, anno 1805. Pikaku. leg. T. ]\[aki\<), anno I8!l(i ; Daito : Pernng. DiSTHiB. Asia tro])i<;'a, Africa, et Australia. 3. B. SericanS Hook. f. Fi. Prit Ind. in, p. '2i\'2 ; FouB. et Hemsl. Ind. Fl. Sin. I, p. 422 ; Hexiîv, List Pi. Formos. p. 5o. Hab. Shinteku, Taipe, leg. T. ]\[akixo, anno lNi)() ; Ku-ehu, Shintem-Kiie, Koto-sho, leg. K. Mfyakk. anno ISDi) ; Hösoang, leg. V. TAsniiio, anno ISD" : Scnton, ipse, anno 1000. DisTKTB. India. 30 ART. 8.— B. HAYATA. 4. B. balsamifera ^^C. Prodr. V, p. 447 ; Clarke, Comp. Ind. p. 81) ; Hook. f. Fl. Brit. Tnd. TU, p. 270 ; Hemsl. Chall. Voy. ]U. I, part 3, p. 160 ; Fort., et Hem.sl. Ind. Fl. Sin. T, p. 420 ; Hkxry, List Formes, p. 52. Coin/:a Jxdsamifera Lixx. S]\ PI. ed-2, p. 1208 ; ]\IiQ. Fl. Ind. Pat. II, p. 5 5. Bacchan's Salcia Lour. Fl. Cochinch. ed-2, p. GOo. Hab. flori-slia, leg. Y. Tashiko, anno 1897. iJisTRiB. Tropica India et ]\Ialaya. 5. B. chinensis I>C. Prodr. V, p. 444 ; P>EXTH. Fl. Hongk. p. 177 ; Hook. f. Fl. Prit. Ind. Ill, p. 268 ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 420 ; Hkxuy, List l'l. Foj-nios. p. 53. Coni/:a clumnisis Lrxx. Sp. PI. ed-2, p. 1208 ; Miq. Fl. Ind. Bat. IL p. 52. Hab. Penang, leg. 4". Makixo, anno 1896; l'achina, leg. 1\ XiiXAMi, anno 1897; 4'aipe, leg. K. Hoxda, anno 1897; Taitocho : Leikokwa, Sliinlenkiie, Shokakuto, leg. K. MiVAKE, anno 1899. DisTUiiî. Java et India orientalis, 6. B. malabarica Hook. f. Fl. Prit. Ind. Ill, p. 267. Hab. Pekak, leg. 4\ Makixo, anno 1896. DisTRTB. India. 7. B. spectabilis DC. Prodr. V, p. 445 ; Hook. f. Fl. P>rit. Ind. III, |.. 269 ; Clarke, Comp. In^. in Linnaea VI. p. 150; et Synop. Comp. p. 207 ; DC. ?rodr. V, p. 451 ; Iîexth. F]. Hongk. p. 179 ; et Fl. Austral. Ill, p. 527 ; Clat;ke, Com posit, [nil. p. 9o ; He^lsl. Voy. Chall. Bot. I, part 8, p. IGO; Hook. f. Fl. Brit. Iii. 427 ; Hexry, List Pi. F(jrmos. p. 51). GiiaphaJiuiii hitm-alhinii, Ltxx. rar. rnuitici'px H'>i)K. f. FI. Brit. Ind. in. p. 2SS. 32 ART. e— B. HAYATA. Giuiplialiwn javainiiu, G. minisennu, et G. coiifusum DC. ProJf. VI, p. 2^2. Icon. Jap. Somokudsusetz, XVI, fol. 32 ; Honzodsufu, XVIII, foJ. II, recto. Hab. Hüsoan, leg. H. Kawakami, anno 1895 ; Shintek, Tanisui, leg. T. ]\Iaki\o, anno 180G ; in montibus Morrison 5000 ped. alt., leg. K. Toiiii, 1890; Sho-to- kakn, Shitemkoé, leg. K. Miyakt:, anno 1899. DiSTKiB. India borealis, et Japonia. 2. G. luteo-album Linn. Sp. PI. ed-2, p. 1196; Less. Synoj^ Comp. p. oöl ; DC. Prodr. A7, p. 230 ; Hooker, FJ. Niger, p. 143 ; Miq. Fl. Ind. Baf. II. ]>. 91 ; Bexth. Fl. Austral. III. 653 ; Soxdek, Fl. Capens. Ill, p. 262 ; Hooker, Handb. X'ew Zeal. Fl, p. 154; Clarke, Comp. Ind. p. 114; Hemsl. Biol. Centr.-Americ. II, p. 136 ; Hook. f. Fl. P^rit. Ind. Ill, p. 288; Hillebraxd, Fl. Hawai. p. 120; Henry, List PI. Formos. p. 53. Hab. Shoku, Shinteku, leg. K. Miyakk, anno 1899. DiSTRiB. In calidis et temperatis regionibiis late diffusa. 3. G. japonicum Thixb. Jap. p. 31 1 ; Bexth. Fl. Austral. III. p. 653 ; Sikb. et Zucc. Fl. Jap. Fam. Xat. p. 187 ; DC. Prodr. VI, p. 237 ; Fr. et Sav. Enum. PI. Jap. I, p. 241; Haxce in Journ. ]>ot, 1880, p. 'Kji ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 427 ; Dielr, Fl. Centr. Chin, in Engl. Jahrb. XXIX, p. 613. Gnaplialiinn incoliicratuin FoRST. DC. l^'odr. VI, p. 235 ; ]>ot. Maçr. t. 2582. Ic. Jap. Somokudsusetz, XVI, fol. 33. COMPOSITE FORMOSAN^. 33 Hab. Taipea,, leg. T. Making, anno 1896. DisTRiB. JaponiiJ, Australia et Nova-Zealandiu. 1. G. indicum, Linn. Sp. PI. ed-2^ p. 1200; Lour. Fl. Cochinch. ed-2, ]). ()08 ; DC. Prodr. VI, p. 231 ; Miq. PJ. Tnd. Bat. n, p. U2 ; Benth. Fl. Hong. p. 188; et FJ. Austral. Ill, p. 655 ; Clarke, Com}). Ind. p. 114 ; Olivek, Fl. Tropic. Afric. Ill, p. 345; Hook. f. Fl. Prit. Ind. Ill, p. 289; Foub. et Hemsl. Ind. Fl. Sin. I, p. 427 ; Henry, List PI. Formos. p. 53. Hab. Taiko-küe, leg. Y. I'asiiiho, anno 1897 ; Shu-shu-koe, Hinpoho, leg. C. Owataki, anno 1898. DiSTRiB. Tropicîi Asia, Africa et Australia. 32. Carpesium li NN. 1. C. abrotanoides Linx. Sp. PI. ed-2, p. 1204; Thunb. Fl. Jap. p. 312; DC. Prodr. YI, p. 282; Clarke, Comp. Ind. p. 131; Hook. f. Fl. Brit. Ind. Ill, p. 301; Maxlai. in Mel. Biol. IX, p. 290; Forb. et Hemsl. Ind. Fl. Sin. I, p. 430; Henry, List PI. Formos. p. 54; Diels, P'l. Centr. Chin. Eng. Jahrb. XXIX, p. 615. Ic. Jap. Somokudsusetz vol. 15, fol. i)'2, sub. Hondzo- dsufu vol. 16, fol. 5. verso. Hab. in montibus Morrison 5000 ped. alt., leg. li. Torii, anno 1899. Distrib. Eûropea orientalis, India borealis et Japonia. 33. Echinops l inn. 1. E. dahuricUS Fisch.; DC. IVodr. \l, p. 523 ; Francii. PI. David, p. 17() ; Henry, List PI. Formos. p. 55. 34 ART. 8.-B. HATATA. Echitiops Gmelini Ledeb. Fl. Ross. II, p. 65o; Maxim. Prirji. FI. Amur. p. 167. Echinops Sphaerocephalns Miq. in Ann. Mus. Bot. Liigd-1'at. II, p. 182 ; Fr. et Sav. Enum. PJ. Jap. I, p. 25P). Hab. Shintek, leg. T. Makixo, anno 180(J; Chüko, leg. V. Honda, anno 1899. DiSTRiB. Siberia et Japonia. 34. Cnicus linn. ( 'onspccfiis Specicnmi. Caulis 2-3-pedes lonj];as, involucri bracteae angustiores....!. C. japonicii.s. Caulis pede liumilior, involucri braeteae exteriores latiores. (ovato-lanceolatae.) 2. C. hrt^ricaiilis. 1. C. japonicUS Maxiaf. in Mél. lîiol. IX, p. 322 ; Fr. et Sav. Enuni. PI. Jap. I, p. 2()() ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 461 ; Henry, List PI. Formos. p. 55; Cirshun japomcwn DC. i'rodr. VI, p. 640 ; Hook, et Arn. Pot. Peech. Yoy. p. 2()i^. Canhus japoniciis Fuaxch. PI. David, p, 178. Cirsiuni Maackil Maxim, l'rim. Fl. Annn-. \). 172. Hab. Naigi, leg. C. Owatari, anno 1898 ; in montibus Morrison 5000 ped. alt., leg. P. Torii, anno 1899. DiSTRlB. Japon ia. 2. C. brevicaulis A. Gray ; Maxim, in Mél Piol. L\, y. 324 ; Henry, List. PI. Formos. p. 55. Cirsmin hrevicavlc A. (jIkay, Pot. Ja]), p. 396. COMFOSIT.E FORMOSANiE. 35 Hab. loco non indicnto. DiSTRiB. Japonia. 35. Saussurea dc. 1. S. affinis Spreng. ; Clarke, Comp. Ind. p. 232 ; Hook. f. Fl. Brit. Ind. Ill, p. 37a ; FoRB. et Hemsl. Ind. Fl. Sin. I, p. 463 ; Henry, List PI. Formos. p. 55 ; Diels, Fl. Centr. Chin, in Engl. Jahrb. XXIX, p. 624. Saussiimi carthainoides Bexth. Fl. Ilongk. p. 168, et Fl. Austral. Ill, p. 456. Aplotaxis caiiltamoides DC. Prodr. VI, p. 540. Semitida carthainoides Blch. ex Roxb. Fl. Ind. Ill, p. 407. Aplotaxis imdticaulis DC. Prodr. VI, p. 540 ; MiQ. in Ann. Mus. Bot. Lugd.-Bat. II, p. 183. Aplotaxis Büwjei, DC. l^rodr. A'l, p. 539. Hemistcpta lijmta Bunge ; Maxim, in Mil. Biol. IX, p. 334. Saiisswea Buivjei Benth. et Hook. f. ex. Fr. et Sav. Enum. PI. Jap. I, p. 255 ; Franch. PI. David, p. 182. Hab. Anping, leg. H. Kawakami, anno 1896 ; Bo-riyo, leg. C. Owatari, anno 1898; Daito: Suibi, leg. K. Miyakk, anno 1899 ; Daichu : Daiboho-shö, leg. H. Satake, anno 1899; in montibus Morrison 5000 ])ed. alt., leg. II. ToRii, anno 1899. DlSTKiB. In India or., liirnia, Japonia et Australia. (gis ART. 8.— B. HAYATA. 36- Crépis l. Conspectus Specierum . Folia lyrata v. siibintegra, ,tenuia 1. C. jajwnica. Folia integerrima, crassiuscnla , ...2. C. inte(ira. 1. C. japonica Bexth. FL Hongk. p. 194; et FI. Austral III, p. 670 ; MiQ. in Ann. Mus. Bot. Lugd.-Bat. II, p. 190 ; Maxim, in Mél. Biol. IX, p. 346 ; Fr. et Sav. Enum. PI. Jap. I, p. 271 ; Franck. PJ. David, p. 185 ; Hook. f. FI. J^rit. Ind. Ill, p. 395 ; Hemsl, in Voy. Chall. I, part 1, p. 46 ; Hillebrand, FI. Hawai. p. 233 ; FoRB. et Hemsl. Ind. FI. Sin. I, p. 475 ; Henry, List PI. Formos. p. 55; Diels, FI. Centr.-Chin. in Engl. Jahrb. XXIX, p. 632. Prenanthes japonica Linn. ; Thunb. Fl. Jap. p. 302 ; Hook, et Arn. Bot. Beech. Toy. p. 266. Youngia japonica DC. Prodr. VII, p. 194. Crépis lijvata Benth. ex Maxim, in Mél. Biol. IX, p. 346. Youngia Thunhergiana DC. Prodr. YII, p. 192. Prenanthes lyrata Thunb. Fl. Jap. p. 303. Youngia rucinata, Y. napifolia, Y. Poosia, Y. anihujua. Y. fastigiata et Y. striata DC. Prodr. VII, pp. 192-193 ; ex Hook. f. FI. Brit. Ind. Ill, p. 395. Icon. Jap. Somokudsusetz vol. 15, fol. 30. Hab. ïaipea, leg. T. Marino, anno 1896 ; et C. Owatari, anno 1897 ; Taichn-ken : Toseikaku, leg. R. Satake, anno 1899. DisTRiß. India, Japonia, et Australia. COMPOSIT.E FORMOSAN^. 37 2. C- intégra ^Iiq. in Ann. Mus. ßot. Lugcl.— Iî:it. IT, ?. U)Ü; .AIaxi.m. in Mt'l. liiol. IX, p. :U8. Forb. et Hemsl. Ind. Fl. Sin. I, p. 475. Frenanthcs inieçfra Tnuxn. Fl. Jap. p. oOO ; Hhok. et Arx. lîot. l>eecli. Voy. p. liG'i. Youngia lanccolaUi. DC. Prodr. VII, p. 193. Prenanthes Ianceolata HorxT.; Willd. Sp. PI. III. p. 1534. Yomujia iiite'ifd A. (tRay, lîot. Jap. in ]\Iem. Anier. Acad. N'l, p. 39(3. Icon. Jap. Soinokudsiisetz vol. 15, fol, 'Hi. Hab. (Tisliun, lei;. K. Mivakk, anno 1899. Distrib. Ja[)onia. 37. Taraxacum iiaii. 1. T. officinale Web.; Ledeb. FJ. Ross. II, p. 812; Guise- bach, Fl. West Ind. Isl. p. 384; Sowerby, English Bot. t. 1304; Fr. et Sav. Enum. PI. Jap. I, p. 1^09 ; Clarke, Comp. Ind. p. i^VS ; Hemsl. ]>iol. Cent.-Amei-ic. II, p. l^»il, et IV, p. ()5 ; He.msl. Voy. Ciiall. I, part 1, p. 4G ; Hook. f. Fl. P.rit. Ind. Itl, p. 401 ; Baker et S. Moor, in Journ. Linn. Soc. XVII, p. 383; Foub. et Hemsl. Ind. Fl. Sin. I. p. 478, A. Gray, Synop. Fl. X'. Americ. ed. 2, I, part 2, [». 440 ; Hexry, List PI. Formos. p. 55 ; Diels, Fl. Centr. China, in Engl. Jahrb. XXIX, p. 630. Tamxacinn Dcns-leonis Dlsf.; DC. Prodr. VH^ p. 145 ; Benth. F\. Hongk. p. 192. Taraxacum coniiculatiun DC. Prodr. VII, P. 146 ; Fraxch. PI. David, p. 187. 3S AKT. 8.-B. HAYATA. Icon. Jap. Somokiulsusetz vol. 15, fol. 10. Hiib. Sbintek, ]eg T. Makixo, auno 1896. DisTHiB. In regionibns teniperatis frigidi.^que fere orbis late diffusa. 38. Lactuca J> L\N. Conspectus Specieruiu . O Involnciuiu angnstiun, bracteis 3-seriatis valde inaequalibus imbiicatis \. L. JjicvinMris. OO Involncium Jatiii&ciilniii, bracteis 1-seiiatis, aequalibus, adclitis exteiioribus minimis adpiessis. □ Capitula milita, parva, 0-8 mm. longa. z^ Caulis validiuscnlns, l-2-pe(les lougiis. 1 Pnppiis rnfo-albns !>. L. Tlninhei(ji. 479; Hknuy, List PI. Fornios. p. 55. Lavtuca sjiuirrosa MiQ. in Ann. Mus. WiA. JAigd.-lîat. fl, p. 189; Maxlm. in Mel. JJiol. IX, p. 353; Fkaxch. Fl. David. ]). 187; Lactuca aiirureiisis Rkgl. ; Maxim. Prim. V\. Amur. pp. 178-473, tide Hemsl. Icon. Jap. Somokudsusetz vol. 15, fol. I'J. I lab. Taipea, leg. T. Makixo, anno 1890; Kotosho, leg. K. Mivake, anno 1899 ; Senton, ipse, anno 1900. COMPOSIT.E FORMOSAN^. 39 DisTiUB. Maiidsliuria et J.-qxinia. 2. L. Thunbergiana .AIaxim. in Mel. liiol. LX, p. 3G1 ; FoiiB. et Hemsl. lud. Fl. Sin. I, p. 484 ; K. Miyabk, FJ. Kiirile Is], ill Mem. Boüt. Soc. Nat. Hist. R', j), 245. Ixeris Tliunhey-jiana A. Gray, 15ot. Jap. in Mem. Amer. x'\cad. YI, p. 397. i^rcnautltes ddutata Tirxi;. Fl. Jap. p. .SOI. Lactuca OldhiUiii Maxim, in Mt'I. l>iol. IX, p. o()o. Icon. Jap. Somokudsusetz V(j]. 15, fol. 15. Hab. Taipca, Shintek, Sha-ryo-to, leg. F. Marino, anno 1(S!:)G ; Cho-ka-ehirai, Kacliinrö, leg. C. Owatari anno 1898; Ku-chfi, Shitemkue, Faito-tea, Dai-ko-shn, Kotosho, \eg. K. MiVAKK, anno 1899 ; leg. T. Shixke anno 1899; Senton, ipse, anno 1900. DisTiiir.. in Japonia, et Fhilippinis insulis. 3. L. versicolor SciiULTZ-Hii'.; Maxi.m. in Mel. l3iol. IX, p. o62; IjAKER et Mookk, in Journ. Linn. Soc. X^'ll, p. 383; Franch. PI. David, p. 188 ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 485 ; Henry, List I'l. Formos. p. 5(3 ; Diees, FI. Centr.-Chin. in Engl. Jahrb. XXIX, p. 031. Ixeri.'i ccrdcolor DC. l*rodr. \\\. p. 151 ; IjEXTII. V\. Hongk. p. 193; MiQ. in Ann. Mus. lîot. Lugd.-liat. 11, p. 191 ; Ledeb. Fl. Koss. II, ]). 817. Prenantlies chincNsis Tiiunb. Fl. Jap. ]). 301. YowKjia chinen.sis DC. Prodr. \'II, p. 194. Icon. Jap. Somokudsusetz vol. 15, fol. 19. Hab. Taipea, FamsLii, leg. F. Makixo, anno 1896 ; Faiko, leg. C. Owatari, anno 1897. Distrib. Dahuria et Japonia. 40 ART. 8 — B. HAYATA. 4. L. gracilis DC. l'rodr. VIT, p. 140; Clarke, Comp. Ind. p. 271; Hook. f. Fl. F.rit. Ind. III, p. 410 ; Forh. et Hemsl. Inü. Fl. Sin. I, p. 482; Dikls. Fl. Centr. Chin, in Engl. Jahrb. XXIX, p. 6?,!. Hab. Taitocho: Perung-, leg. K. Miyakk, anno 1899. DrSTRiR. in India or.. 5. Li, debilis Maxim, in Mél. Biol. IX, p. 365 ; Forb. et Hemsl. Ind. Fl. Sin. l, p. 480 ; Henry, List PI. Fortnos. p. 50. Ixeris dehilh A. Guav, Pot. dap. in Mem, Amer. Acad. n. s. VI, p. 397 ; L>ENTH. Fl. Hongk. p. 193 ; Miq. in Ann. Mus. I>ot. Lngd- Bat, II, p. 191. Youngia dehilU DC. prodr. A II, p. 194. Prenauthcs dehilh Thunb. Fl. Jap. p. 300. Icon. Jap. Somokudsusetz vol. 15, fol. D'. Hab. loco non indicato. DiSTRiB. Japonia, 6. L. repens Maxim, in Mél. Biol. IX, p. 364 ; Forb. et Hemsl. Ind. Fl. Sin. 1, p. 4S3 ; K. Miyabe, Fl. Kurile Isl. in Mem' Bot. Soc. Nat. Hist. vol. IV, p. 245 ; HexrY List PI. Formos. p. 56. Ixeris rcpcus A. (Iray Bot. ffap. in ]\Iem. Amer. Acad. n. s. VI, p. 397. Bexth. Fl. Hongk. p. 194. Icon. Jap. Somokudsusetz vol. 15, fol. 18 ; Ilondsodsufu vol. 49, fol. 22, recto. Hab. Tamsui, leg. T. Marino, anno 1896. Distrib. Mandshuria, flaponia et Kamtschatka. 39. Sonchus Lixx. Conspectus Speciennn. Folia grosse dentata v. pinnatifida 1. S. oleraceus. Folia remote serrata 2.5. arvensis. COMPOSITE FORMOSAN^. 4| 1. S. oleraceus Lixx. Sp. V\. ed-2, p. lUG ; RoxB. Fl. Ind. Ill, p. 402 ; Less. Synop. Comp. p. 140 ; Hooker, Fl. Niger, p. 144; Miq. Fl. Ind. Bat. II, p. 112; Benth. Fl. Hongk. p. 194 ; Sonder, Fl. Capens. Ill, p. 528 ; H(>r)K. Fl. N. Zeal. p. 166 ; Bexth. Fl. Austral, III, p. 679 ; Grisebach, Fl. Brit. West. India. Isl. p. 384 ; Fr. et Sav. Enum. PL Jap. I, p. 272 ; Clarke, Comp. Ind. p. 276 ; Oliver, Fl. Tropic. Afric. Ill, p. 457 ; Hook. f. Fl. P.rit. Ind. Ill, p. 414. Hemsl. Yoy. Chall. I, part 1, p. 46; Forb. et Hemsl. Ind. Fl. Sin. I, p. 487 ; A. Gray, Synop. Fl. X. Americ. ed~2, I, part 2, p. 444 ; Hillebrund, Fl. Hawai. p. 233 ; Henry, List PI. Formos. p. 56. .S'. ciliatus Lam. DC. Prodr. VII, p. 185. Icon. Jap. Somokudsusetz vol. 15, fol. 6. Hondsodsufu vol. 49, fol. o, recto. Hab. Taipea, leg. T. Makixo, anno 1896; Gishun, leg. Y. Tashiro, anno 1898. DiSTRiB. In temperatis et siibtropicis regionibus late diffusa. 2. S. arvensis Lixx. Sp. PI. ed-2, p. 1116; DC. Prodr. A'll, p. 187 ; SowERBY, English Bot. t. 1308 ; Fr. et Sav. Ennm. PI. dap. I, p. 273 ; Clarke, Comp. Ind. p. 276 ; Hook. f. Fl. Brit. Ind. Ill, p. 414 ; A. Gray, Synop. Fl. X. America ed-2, I, part-2, p. 444 ; Forb. et Hemsl. Ind. Fl. Sin. I, p. 487 ; Henry, List PI. Formos. p. 56 ; Diels, Fl. Centr.-Chin. in Engl. Jahrb. XXIX, p 631. S. nliglonocus BiEBERST ; DC. Prodr. Vll, p. 186 ; Maxim. Ind. Fl. Pek. in Prim. Fl. Amur. p. 473, Franch. PI. David, p. 189, fide Hemsl. S. hmchyotiis DC. Prodr. YII, p. 186. 42 AET. 8.-B. HAYATA. Jap. Icon. Soniokudsusetz vol. 15, fol. 7. Hab. Taipea, leg. T. Marino, anno 1896 ; Koto-sho, leg. K. MiYAKE, auno 1899. DiSTRiB. In temperatis et subtropicis regionibus late diffusa. ÏNDEX NOMINUM ET SYNONYMORUM. (SYNONYMA LITTERIS CURSIVIS TRADUNTUR.) 43 Adenostemma Forst 6. ,, visCOSUm Forst ... 6. Ageratum Lixx 7. ,, conyzoides Linn 7. Aplotaxis Bidifjei, DC 35. J, caitlunii aides DC 35. ,, rnidticauUs DC 35. Artemisian L 23. ,, capillaris Thine 23. ,, illilini WiLLD 24. ., vulgaris J; var. in- dica Maxim 24. Aster Linn 12. (t(iei atoides TvRCZ 13. ,, altaicus Willd 12. ,, baccharvides Steetz 13. ,, indicus I inn 14. ,, Oldhami Hemsl 14. ,, fitridtiis Fr 13. ,, trinervius Eoxb 13. y liicc/un is Sal cid Lour 30. BidenS Linn '20. ,, bipinnata Linn 21. ., chillCllsis WlLLD 21. ,, hinceohita Willd 21. ,, pilosa Linn 20. ,^ pilosa rur. bipinnata Hook. f. 21. WaJlichii DC 21. Blumea DC 28. ,, balsamifera DC 30. ,, chineiisis DC 30. lacera DC 29. Blumea malabarica Hook, f 30. ,, membranacea DC 2'J. ,, sericans Hook. f. 29. ,, spectabilis DC 30. Bolt nia indica Benth 14. Cacalia hidhosa IjOVR. Fl 25. Calinieris altaica Nees 13. ,, ciliata A. Gray 13. Ca id tins japonicus Fraxch 34. Carpesium Linn 33. ,, abrotanoides Linn... 33. Centipeda orbicidaris Lour 23. Centiatlternni chiiiensis Less 5. Chri/san(Jii'inii)n L 22. ,, Ser/etionlj 22. Cirsitim hiecicaule A. Gray 34. ,, j(/po)iiciim DC 34. ,, Maackii Maxim 34. Cnicus LiNx 34. ,, brevicaulis A. Gray 34. ,, japonicus Maxim 34. Conyza Less 15. ,, aegyptica Ait 15. ,, halsaini/i-ra Lixx 30. ,, ihiniiisis Lixx 30. ,, japonica Less 15. ,, patiila Willd 5. ,, ni noniaie/'olia Wall 15. Crepis L ^^^■ ,, intégra Miq 37. ,, japonica Bexth 3G. Ivraia Benth Ö3. 44 CroSSOStephilim Less 24. ,, artemisoides Less 24. Ci/a)iopis pubesceiis DC 5 , Dichrocephala DC ll. „ latifoliaDC li Diploppapiifi aspen ini us DC 13 EchinopS Linn 33 ,, dahiiricus Fisch 83 „ (ryndini Ledeb 34 ,, SpJtae>-ocej)h(itiis Miu 34 Eclipta Linn 17 ,, alba Hassk 17 ,, I'ifcta Linn 18 ,, prostrxtft DC '. . 18 Elephantopiis Lixn 5 ,, scaber Linn 0 ,, spicatus B. Juss... 6 Emilia Cass 2G ,, sonchifolia DC ... 2() Erigeron Linn 14 ,, linifolius Willd 14 )) jn/rifolius Benth 10 Elipatorilim Linn 8 ,, chinense var. tripiirtitimi MiQ 9 ,, japonicum Thu.nb 8 ,, Lindleyanum DC 9 ,, Reevesii Wall 8 ,, scavdi'iia Linn 10 ,, Tashiroi Hayata 9 fr^////c/m DC 8 FitrfiKjiinn Kdciiipfcii Benth 27 Glossogyne Cass 22 „ chiiioisis Less 21 ,, tenuifolia Cass 22 Gnaphalium Linn 31 „ confiisuni DC 32 ,, indicum, Linn 33 ,, involuciatntii Forst ... 32 Gnaphalium japonicum Thunb ... 32. ^^ jiH/niiim DC 32. ,, luteo-album Linn... 32. ,j hiteo-nlbum, Linn. ivir. muJticeps Hook, f 31. , , multiceps Wali 31. ,, ravnsentm, DC 32. Grangea Adans 12. ,, maderaspatana PoiR ... 12. Gynura Cass ; 24. ,, ail) icalatd DC 25. ,, bicolor DC 25. ,, hiilhosa Hook, et Arn 25. ,, elliptica Yabe et Hayata. 25. ,, ovalis DC 25. ,, l'üeiid(?cJti)l)i DC 25. He))iif>t('ptti Ii/rata Bunge Hisiitsua serrata Hook, et Aru Lceris debilis A. Gray 40. ,, repeiis A. Gnw 40. ,, Thunheyfjiaiia A. G's.w 39. ,, versicolor DC 39, LactUCa Linn 38. ,, a )ii u ren.si.'y FiKiiL, 38. ,, brevirostris Champ 38. ,, debilis Maxim 40. ,, gracilis DC 40. Oldluniii Maxim 39. ,, siiiiaroK.sd Mu} 38. , , repens Maxim 40. ,, Thunbergiana Maxim ... 39. ,, versicolor Schultz-Bip... 39. [Aiiitlarui (/iij/inter) Sieb et Zucc 27. ,, Kaeinpferi, Sieb, et Zucc... 27. Mikania Willd 10. ,, SCandenS Willi) 10. ,, roliibili.s Willd 10. Microglossa DC 16. ,, volubilis DC 16. Myriagyne Lk^^« 22. 45 Myriagyne minuta Less 22. PlucheaCAss 31. ,, indica Less 31- Frenanthee chiiiensis TnvyiB 39. „ debilis Thunb 40. rrenanthes de)itata Tuvhb 37. „ inteijra Thunb 37. „ jiqwnica Linn.; 36. ,, lanceolate Houtt 37. ,, I i/rata Thvsb 36. Saussurea DC 35. ,, affinis Spreng 35. „ caithamoides l^E'STH 35. „ Biinrjei Benth. et Hook. f 35. Senecio L 27. ,, campestris DC 27. ,, Kaempferi DC 27. ,, Krameri Fk et Sav 28. , , nemorensis L. var octo- gloSSUS Ledeb 28. , , octot/lossiis DC 28. „ Fierotii Miq 27. Serratula carthamoides BrcH 35. Siegesbeckia Linn 17. ,, orientalis Linn 17. Solidage Linn 10. „ cantonieiisi'i IjOVR 11. „ dccuirens IjOVV. 11. , , Virga-aurea Linn lo. SonchuS Linn 40. ,, arvensis Linn 41. ,, brach i/ot US DC 41. „ ciliatus Lam 41. ,, oleraceUS Linn 41, „ iilii/lnnociis Bieberst 41. Sj)haeroiiiorj>Jiae(( Centipeda DC 23. SpilantheS Linn 20. Spilanthes Acmella Linn 20. ,, tinctoi ills IjOvv. 7. Tagetes Linn 22. ,, patula Linn 22. Tanacetnm chineuse A. Gray 24. Taraxacum Hal] 37. ,, corniciilatiDn. DC 37. ,, l^eiis-letniis Disk 37. ,, officinale Web 37. Yerhesina calendidacea. Linn 19. ,, piostrata Hook, et Arn... 19. Vernonia Shreb 4. ,, AnderSOni Clarke 4- ,, chinensis Less 4. ,, cinerea Less 5. ,, I 'tnni)i(jiinia Benth 4. Wedelia Jacq I8. ,, biflora Benth 19. ,, calendulacea Less 19. ,, prostrata Hemsl 18. WoUastonia hißora DC 19. ,, piostrata Hook, et Arn... 18. Xanthium Linn 16. „ indicinn lio^m 10. ,, Strumarium Linn 16. YoiDKjia ainbii/iia, DC 36. „ chinensis DC 39. debilis DC 40. ,, fasticiiata DC 36. ,, inter/ra A. Gray 37. 5) jcponica DC 36. „ lanceolata DC 37. „ napifolia DC 36. „ Foosia, DC 86. ,, riicinata DC 36. „ striata DC 36. ,, Thindierifiana DC 36. B. HAY AT A. Compositae Formosanae. PLATE I. Eupaloriiini Ta^ihirn/ IIayata. EXPLICATIO TABULAE. Eiipatoriuni Tosliiroi, Hayata. Planta tabulae ex specimine exsiccato clepicta. Fig. 1. Fragmeiitnm rami. mag. iiat. Fig. 2. Capitulum. mag. anct. Fig. 3. Flos. mag. auct. Fig. 4. AcLaeninm, cum pappo supenie amorto^ validissime anctum. Fig. 5. Ejus verticalis sectio, validissime aucta. Fig. 6. Acbaenii horizontalis sectio, validissime aucta. Rayata, Composit.Formos. Jour. Sei. Coll. Vol, XVIII, Art. 8, PI. I. E up a.t orram Tasliir oi, ±i ay at a ■ F. Mi Stic 3 3c.'jl|it B. HAY ATA. Compositse Formosanae. PLATE II. Gymira elliptica Yabe et Hayata. EXPLICATIO TABULAE. Gtjnura ellijitica, Yabe et Hayata. Planta tabulae ex specimine exsiccato depicta. Fig. 1. Pars plantae. mag. nat. Fig. 2. Flos. mag. anct. Fig. 3. Flos, corolla et stamina amorta, stylus videtur. mag. auct. Fig. 4. Stamina, mag. auct. Fig. 5. Styli rami, validissirae aucti. Fig. 6, Pappi seta, validissime aucta. Fig. 7. Acbaonii horizontalis sectio. mag. auct. Hayalob, Camposit. Forinos. Jour. Sei. Coll. Vol. XVIII, Art. 8, PI. II. E.Kayat.a de Gynura elliptica,Y3L"be et HayaLa, Viahraa 3culpl i UH ITKB M