■ ■ i ■ ' ■ i H I ■ ■ L WE 9 H^^H >v2 1 1 &**' THE VOYAGE OF H.M.S. CHALLENGER. ZOOLOGY-VOL XXVIII. REPORT SCIENTIFIC RESULTS ^ ~r . OF THK VOYAGE OF H.M.S. CHALLENGER DURING THE YEARS i 8 7 3-7 6 UNDER THE COMMAND OF Captain GEORGE S. NARES, R.N., F.R.S. AND THE LATE Captain FRANK TOURLE THOMSON, R.N. PREPARED UNDER THE SUPERINTENDENCE OF THE LATE Sir C. WYVILLE THOMSON, Knt., F.R.S., &c. RBGIUS PROFESSOR OF NATURAL HISTORY IN THE UNIVERSITY OF EDINBURGH DIRECTOR OF THE CIVILIAN SCIENTIFIC STAFF ON ROARI) AND NOW OF JOHN MURRAY, LL.D.. Ph.D., &c. ONE OF THE NATURALISTS OF THE EXPEDITION Zoology— Vol. XXVIII. IPubltsbeD bp ®toer of %n jfflajrstp's ©oiiernment PRINTED FOR HER MAJESTY'S STATIONERY OFFICE AND SOLD BY LONDON :— EYRE & SPOTTISWOODE, EAST HARDING STREET, FETTER LANE EDINBURGH :— ADAM & CHARLES BLACK DUBLIN :— HODGES. FIGGIS. & CO. l8SS Price Thirty-one Shillings and Sixpence. \ ' *^ A PRINTED BY NEILL AND COMPANY, EDINBURGH, FOR HER MAJESTY'S STATIONERY OFFICE. CONTENTS. Report on the SiphonophorjE collected by H.M.S. Challenger during the years 1873-1876. By Ernst Haeckel, M.D., Ph.D., Hon. F.R.S.E., &c, Professor of Zoology in the University of Jena. (The Manuscript was received in Instalments between 8th February and 5th July 1888.) 7 iiOO t EDITORIAL NOTE. During the voyage of the Challenger, species of Siphonophorse were con- tinually under observation in the living condition, both on the surface of the ocean and in tanks on board the vessel ; none of the naturalists, however, made any special study of this interesting and complicated group of animals ; the specimens collected were merely preserved with as much care and skill as was possible in the circumstances. Probably no class of marine organisms presents, in general, greater difficulties in the way of examination than the Siphonophorse, and these difficulties are increased many fold when the naturalist has to deal with collections preserved in alcohol and other media, which contract and distort the specimens. Indeed it was essential that any naturalist who would undertake a successful examination and interpretation of these varied forms should himself have investigated these organisms during many years in the hiving state. Professor Haeckel, through his long-continued and elaborate investi- gations of living Siphonophorae and Medusa? in the Mediterranean, Indian, and Atlantic Oceans, was in a very special manner fitted to undertake such a task, and it must be regarded as fortunate that he should have been willing to undertake the work on condition that some of his own unpublished observations should be incorporated. This important and masterly Report has thus become a Monograph of the whole class, more complete than any hitherto published ; the classifi- viil THE VOYAGE OF H.M.S. CHALLENGER. cation has been reformed and placed on a new basis. The new order, Auronectffi, discovered by the Challenger in the depths of the ocean, is particularly interesting to naturalists. The Introduction, giving a general account of the morphology of the order, was translated from the German Manuscript by Mr. J. Arthur Thomson, M.A., F.R.S.E. All the remaining part of the Report was written by Professor Haeckel in the English language. This Report on the Siphonophoile forms Part LXXVII. of the Zoological Series of Reports, and consists of 383 pages and 50 chromo-lithographic plates. John Murray. Challenger Office, 32 Queen Street, Edinburgh, October 13, 1888. THE VOYAGE OF H.M.S. CHALLENOEE. ZOOLOGY. REPORT on the Siphonopiior^e collected by H.M.S. Challenger during the Years 1873-76. By Ernst Haeckel, M.D., Ph.D., Hon. F.R.S.E., Professor of Zoology in the University of Jena. PEEFACE. The Sipkonopkorse collected by tke Ckallenger during ker cruise round the world were kanded to me for investigation in 1879. Tke examination of this material, among which were to be found many new and remarkable types (especially abyssal forms), led me to recommence my study of the organisation and development of tke wkole class, wkick kad been interrupted for more tkan ten years. During a residence of tkree months at Puerto del Arrecife, tke karbour of Lanzerote in tke Canary Islands, December 18G6, January and February 1867, I kad tke opportunity of becoming acquainted witk nearly all tke typical forms of tkis most attractive group of animals. Tke facts wkick I tken discovered regarding tke curious, but till tben little known, ontogenesis of tkese Acalepks, were embodied in a prize essay publisked by tke Gesellsckaft fur Kunst und Wissensckaft in Utreckt.1 I tken intended to follow up tkis embryological memoir witk a more extensive comparative anatomical one, or if possible a Monograpb, in wkick also tke systematic arrangement of tke Sipkonopkorse skould be subjected to a tkorougk revision, and tke very intricate literature of tkis class critically elucidated. Tke completion of tkis sckeme was, kowever, deferred owing to various external kindrances, and tke rick material collected at Lanzerote remained for tke most part unpublisbed. 1 Ziir Entwickelungsgeschichte der Siphonophoren, Utrecht, 1869, 14 plates. (ZOOL. CHALL, EXP. — FART LXXVII. — 1888.) Hhhh a ii THE VOYAGE OF H.M.S. CHALLENGER. In the meantime I found many opportunities of continuing my researches into the Siphonophorse during the numerous journeys which I undertook for the purpose of completing my System der Medusen, and to which I made some reference in 1879 in the preface to that monograph. I also collected much valuable material in the different museums which I visited for the same purpose. A further supply of specimens, and among them many new and instructive forms, were received from various travellers, and notably from Captain Heinrich Rabbe of Bremen. Lastly my journey to India furnished me with a long wished-for opportunity of becoming acquainted with the rich though little known Siphonophoral fauna of the Indian Ocean ; on the journey out to Ceylon (via Bombay), on my return home (via Socotra), and especially in the expeditions which I made from Belligemma and Matura, I observed a number of new genera, some of which were exceedingly interesting (November 1881 to March 1882). At the same time I was able to extend my earlier observations upon the development and metamorphosis of the Siphonophorse. Valuable systematic contributions were also derived from the study of the excellent original drawings and the manuscripts of Mertens, of which Brandt gave a short account in 1835. I am highly indebted to Professor Alexander Strauch in St. Petersburg for the opportunity of examining these drawings which Mertens took in 1827 from living Siphonophorse in the North Pacific, and which, unfortunately, have never been published; they are among the best and most lifedike representations of the class that have ever been produced. I take this opportunity of expressing my best thanks to those gentlemen who have assisted me in this investigation, especially to those travellers and directors of museums who have furnished me with material. By their aid I have been enabled to extend our knowledge of this group, and by the discovery of new morphological types to bring it to a certain degree of completeness. I am especially indebted to my honoured friend, Dr. John Murray, who with the greatest liberality offered to me all the Sipkonophorze collected by himself, and gave me permission to incorporate my own rich materials in this Monograph. My sincerest thanks are due also to the excellent artist, Mr. Adolph Gdtsch, who, aided by his morphological interest and unwearying care, executed the chromo-lithographic Plates of this Report with such perfection, that they will give a complete idea of this most wonderful class of animals. CONTENTS. PREFACE, GENERAL INTRODUCTION, DESCRIPTION OF THE FAMILIES, GENERA, AND Order I. DiscoNECTiE, Family I. Discaliib , . Family II. Porpitidre, . Family III. Velellidae, . Order II. Calyconect,e, Family IV. Eudoxida;, . Family V. Ersseidas, Family VI. Monopliyid.r, Family VII. Diphyidas, SPECIES, Family VIII. Desmopliyidie, Family IX. Polypliyidae, Order III. Physonect^e, Family- X. Circalidae, . Family XL Athoridw, . Family XII. Apolemidae, Family XIII. Agalmidae, . Family XIV. Forskalidse, Family XV. Nectalidas, . Family XVI. Discolabida?, Family XVII. Aiithophysida-, Order IV. Auroxectj:, . Family XVIII. Stephalidse, Family XIX. Rhodalida?, Order V. Cysto.nect.e, . Family XX. Cystalidee, . Family XXI. Rhizophysidse, Family XXII. Salacidse, . Family XXIII. Epibulidas, Family XXIV. Physalidse, BIBLIOGRAPHY OF THE SIPHONOPHOR/E, LIST OF FAMILIES, GENERA, AND SPECIES GLOSSARY, .... STATISTICAL SYNOPSIS, PACE i 25 26 41 51 70 89 103 121 125 141 167 173 184 196 200 204 213 236 249 255 268 281 296 300 305 314 317 329 332 338 353 357 373 380 GENERAL. INTRODUCTION. THEORIES AS TO THE ORGANISATION OF SIPHONOPHOR.E. The morphological and physiological peculiarities which distinguish the Siphonophorse from other Acalephse have led, as every zoologist is aware, to very diverse theories as to the real nature and import of their organisation. Two essentially different interpretations still stand in marked antithesis to one another, and these may for the sake of brevity be described as the Poly-organ and Poly-person theories. Of these the poly-organ theory is the older (Eschscholtz 1829, Huxley 1859, P. E. Midler 1871, Metschnikoff 1874). According to this view the adult organism of all Siphonophorae is a simple " Medusa-like animal," which is distinguished from the typical Medusae only in the multiplication and differentiation of its polymorphic organs. More accurately defined, the organism remains an individual Hydromedusoid person (a morphon of the third grade, or a " morphological individual of the third order of individuality"). In contrast to this older interpretation is the poly-person theory (Vogt 1848, Leuckart 1851, Kolliker 1853, Gegenbaur 1854, Claus 1863, Chun 1882). According to this view the adult Siphbnophoral organism is a colony of animals, composed of many polyp- like individuals, which, according to the laws of the division of labour, have to undergo various modifications, both in the way of specialisation and retrogression. More accurately expressed, it is a " swimming Hydropolyp stock or conn," composed of many polymorphic persons, partly polypoid, partly medusoid (a morphon of the fourth grade, or a "morphological individual of the fourth order of individuality"). Since in the course of the last two decennia the ontogeny of Siphonophorae has become better known, and since the latter has in the light of the theory of descent been inter- preted in causal relation to likylogeny, the main antithesis between the two theories has become more marked. According to the poly-organ theory the ontogenetic primitive form of the Siphonophoraj is a simple Hydromedusoid person, and with this is in- volved the phylogenetic corollary that the whole class originally arose from Medusae. According to the poly-person theory, on the other hand, the ontogenetic primitive form is a swimming Hydropolyp stock, and with this is associated the phylogenetic corollary that the whole class sprang from Polyps. (ZOOL. CHALL. ESP. — PART LXXVII. 1888.) IThllh 1 2 THE VOYAGE OF H.M.S. CHALLENGER. According to the present opinion of most zoologists, which we also share, the low sessile Hydropolyp form is the more primitive ; the higher Hydroniedusa form has been secondarily developed at a much later stage, and that by adaptation to a free- swimming mode of life. In this way has the characteristic swimming organ of the Medusse arisen, namely, the umbrella, with its radially constructed gelatinous disc, which was entirely absent in the primitive ancestral forms — the Polyps. The most important structure of the latter, however, the gastral tube (inherited from the Gastrsea), has been transmitted to the Medusae, and has become the " manubrium," in the wall of which the generative products are devekrped. If we apply this fundamental and firmly based conclusion to the two theories of Siphonophore organisation, the following is evidently the antithesis in regard to the question of origin. According to the poly-organ theory the primitive form of Siphonophoras was a simple Medusa and already possessed an umbrella ; from this established swimming organ the various locomotor organs of the Sipkonophora? ( swimming - bells and air-chambers) are derived by multiplication and modification. According to the poly-person theory, on the other hand, the primitive form of the Siphonophorse was a Hydropolyp colony, and possessed no umbrella ; the locomotor organs which are present are therefore new structures, not to • be derived from any pre-existing swimming organ of the primitive form. And this leads to a weighty contrast in regard to the Medusiform larva?, which arise directly from the gastrulse of Siphonophorse. According to the poly-organ theory, such a larva possesses essentially the morphological value of a simple Medusoid person, and as the hereditary repetition of the original primitive form has the greatest palingenetic importance. According to the poly-person theory, on the contrary, it possesses no such importance ; it is merely of subordinate kenogenetic value, and is to be regarded as a peculiarly modified Hydroid polyp. Both these opposing theories have been for forty years supported with much acuteness by distinguished zoologists, but yet without decisive conclusion ; both are in fact partially justified ; both contain a mixture of truth and error. According to my own opinion, which is based on an extensive comparative investigation of the entire class, and on numerous new facts discovered in the process, the truth lies midway between the two interpretations. The poly-organ theory is right in starting in its whole interpretation and rationale of the Siphonophorse from a Hydromedusoid type, in regarding the primary medusiform larva as palingenetic, and further in supposing an extensive multiplication and dislocation of the several Medusa organs. It is wrong, however, in attributing to the fully developed Siphonophoral corni the value only of a person, and in regarding the persons which compose the stock purely as organs in the morphological sense. The poly-person theory, on the other hand, is right in explaining the fully developed Siphonophore as a conn (colony or stock), composed of many poly- morphic persons. It goes, however, much too far, and is in error when it seeks to REPORT ON THE SIPHONOPHOR^E. 3 attribute to the several (morphological) organs of these persons the same individual value {e.g., to the single protective bracts and capturing filaments); is furthermore wrong in denying an extensive secondary multiplication and dislocation of these organs ; and finally is especially at fault in regarding the primary medusoid larva as a merely henogenetic embryonic form, and, by ignoring the palinge'netic value of the latter, in interpreting the Siphonophorae as " swimming Hydropolyp colonies " comparable to Hydractinia. MEDUSOME THEORY. The new theory of the. organisation of Siphonophoros to which I have been led by my investigations on their comparative anatomy and ontogeny may be briefly designated as the Medusome Theory. It seeks to incorporate the elements of truth in the two views above described, but to avoid their errors. The principal positions of this restatement may be briefly resumed: — 1. The primary larva, which arises directly from the gastrula of the Siphonophorae, is always a simple Medusa ; it may be more or less kenogenetically modified, but always retains a predominant palingenetic import, and is explained by the immediate derivation of the Siphonophorae from Hydromedusae. 2. The primary medusiform larva of the Siphonophorse appears in two essentially distinct forms, as a Disconula and a Siphomda, and since this distinction is associated with essential differences in the resulting Siphonophoral colonies, the class may be legitimately divided into two legions or subclasses, — Disconanthae and Siphonanthae. 3. The first legion, or Disconanthae, including the single order Chondrophoridae or Porpitariae, develops from a regular and octoradial Medusoid larva {Disconula), retains the original corona of marginal tentacles throughout life, and produces the persons of the colony by annular budding from the subumbrella. 4. The second legion, or Siphonanthae, on the other hand, includes all the other Siphonophorse (Calycophoridas, Physophoridaa, Pneumatophoridae, Aurophoridae) ; its primary larva is a bilateral Medusa {Sii^honula), which is distinguished by a ventral umbrellar cleft and by the possession of a single tentacle ; it forms the persons of the colony by one-sided linear budding from the stomach-wall or manubrium. 5. The primary larva of the Disconanthse {Disconula) is to be regarded as the ontogenetic recapitulation of a common primitive octoradial ancestor (Archimeda), and its phylogenetic origin is probably to be sought among the Trachomedusae (Trachy- nemidae, Pectyllidee). 6. The primary larva of the Siphonanthas {Siphomda) is to be regarded as the ontogenetic recapitulation of a common primitive bilateral ancestor {Protomecla), and its origin is probably to be sought among the Anthomedusre (Codonidae, Euphysidae). 4 THE VOYAGE OF H.M.S. CHALLENGER. 7. All the parts which arise by budding from the primary larva of the Siphonophorae are either medusiform persons or special organs of the same. 8. All organs which belong originally to one medusoid person are included in the definition of a Medusome, whether they bud out from a common basis on the stem, or arise in different positions, in consequence of kenogenetic migration or dislocation ; the multiplication of individual equivalent portions {e.g., nectophores, bracts, palpons) which often occurs secondarily is simply to be regarded as a multiplication of organs, not of persons or medusomes. 9. The medusomes appear on the Siphonophoral colony in two distinct main forms, which cannot however be sharply distinguished — in the palingenctic medusomes the principal organs have remained more or less in their original connection {e.g., in the gonophores of Eudoxia); in the kenogenetic medusomes, on the other hand, the principal organs are more or less dislocated, e.g., in the sterile medusome of Eudoxia, which consists of a protective piece (umbrella) and a gastral tube (siphon) with a tentacle. 10. The lateral budding of the secondary medusomes (appendages) on the Siphono- phoral stem occurs sometimes singly and sometimes in groups. Those groups which consist of several medusomes we call " cormidia." 11. The cormidia are originally simple segmental repetitions of a medusome-group in metameric succession separated by free internodes {cormidia ordinata), e.g., the Eudoxise of the Calycophoridaa, the Prodoxiee of the Physophoridae {Apolemia, &c). 12. By the breaking up of such original cormidia, those centralised corms arise in which the persons bud in a scattered fashion over the stem, their several organs thus becoming separate from one another {cormidia dissoluta), e.g., Agalmopsis, Mhizophysa. 13. The degeneration of the single medusomes and of the disassociated organs is of the greatest importance in the development of the Siphonophoral colonies, and that the more, the more markedly the corm is centralised and the more intimate the mutual relations between the polymorphic medusomes. DISCONULA LARVA of the DISCONANTILE. Among the different medusiform larvae of Disconanthae (Chondrophoridae or Porpitarise) which I was able to observe, the youngest larvae of Porpitidae (from O'l to 0'4 mm. in diameter) are of special importance. They possess a circular, flatly arched disc, the margin of which bears a circle of eight simple tentacles. From the middle point of the subumbrella hangs a large central gastral tube, and from the base of this siphon arise at equal distances eight radial canals, which run in the concave subumbrella to the margin of the disc, and are there united in a circular canal. Above this there lies in the middle of the gelatinous disc a pneumatophore, composed of a central lens- shaped air-flask and a circle of eight radial air-chambers surrounding the same. Both REPORT ON THE SIPHONOPHOR^. 5 the former and each of the latter possess on the middle of their upper surface an external aperture (the stigma of the exumbrella). Only in the possession of this hydrostatic apparatus is this Disconula distinguished from an ordinary eight-rayed entirely regular Hydromedusa. Certain Trachoinedusae (Trachynemidse, Pectyllidse) present a most striking resemblance. Even in another further advanced larval stage, which we may call a Porpula, the regular structure of a simple Craspedote person is still retained. The Porpula is distinguished from the Disconula in this, that the number of marginal tentacles has increased from eight to sixteen, and several concentric annular air-chambers are apposed to the primary circle of eight. In a stdl later stage the number of concentric air-chambers' is greatly increased, and the tentacles form several rows at the margin of the umbrella. Then between the margin of the umbrella and the central siphon, eight or sixteen conical buds grow out from the subumbrella — the subsequent bearers of the gonophores. They remain closed in the monogastric Discalidse, but acquire a mouth opening in the polygastric Porpitidse and Velellidse. The youngest larvaB of the Velellidse are hardly to be distinguished from those of the Porpitidse ; the difference between them only becomes distinct when in the former the vertical skin-fold of the exumbrella begins to be developed (Bataria), in the base of which is formed the diagonal crest of the pneumatophore which determines the amphithect ground-form of the Velellidse. Their youngest larvae, however, are as octoradial as those of the regular Porpitida3. At an early stage they are distinguished by this, that in the Porpitidse each of the eight radial air-chambers forms a stigma, in the Velellidse only two lying obliquely opposed. The resemblance between these larvae and the Pectyllidse suggests that the Disconanths have been originally developed from this group of Craspedotse. If the eight sac-shaped gonads, hanging down from the subumbrella of the Pectyllidse, were not themselves reproductive organs, but medusoid gonophores, and if above these in the gelatinous substance of the disc, a glandular depression or exumbrellar invagination led to the constriction of an air-sac, then from a Pectyllid or Trachynemid there would be derived a very simple Discalid. SIPHONULA LARVA of the SIPHONANTILE. Entirely different from the first stage in the development of the Disconanthse, is that of the second legion — the Siphonanthse (Calycophoridae, Physophoridse, Pneumato- phoridse, Aurophoridse). Different as are the Siphonophorse of this manifold legion in their fully developed form, their primary larval forms are very similar, at least so far as their ontogenetic history is yet known. The Siphonula — or the primary medusiform larva — is in this legion from the fii*st not octoradial and regular, but bilaterally sym- metrical. The original circle of tentacles on the umbrellar margin has disappeared ; in all cases only a single capturing filament (Fangfaden) persists, the primary unilateral 6 THE VOYAGE OF H.M.S. CHALLENGER. larval tentacle, which in consequence of the ventral umbrellar cleft has been shifted centripetally to the base of the gastral tube. The primary umbrella itself (Protocodon) has in consequence of this unilateral development become bilateral ; it develops in the Calyconect?e (or Calycophoridse) into tbe primary swimming-bell, in the other three orders (Physonectse, Cystonecta;, and Auronecta?) into the pneumatophore. The " swimming-bladder " arises here again as a gland-like invagination of the ectoderm, not however centrally in the apex of the exumbrella (as in the Porpitaria?), but in an excentric position. The primary siphon (Protosiphon) persists in the monogastric Siphonanths as a single stomachic tube ; while in the polygastric forms it develops into the stem from which all the other persons of the colony proceed by lateral budding. The whole structure of the Siphonanthas, as well as that of their Siphonula larvae, suggests the closest relationship with the Anthoniedusse, and especially with the family of Codonidge. Only in these Anthomedusse do the reproductive elements develop in the entire gastral wall (as in the manubrium of the gonophores in all the Siphonanths), without hint of radial divisions. Among the Codonidaj, however, the subfamily Euphysidse (particularly Hyhocodon and Amphicodon) is of especial importance. Here only do three of the four primary tentacles of the umbrellar margin disappear, and one alone remains to attain a proportionately greater development. In this way the bilateral modification of the umbrella is determined. This origin of the Siphonanths is also corroborated by the marked tendency of many Anthomedusse to form Medusae by direct budding from the gastral tube (Codonivm gemmiferum, Sarsia siphonophora, &c). Since these Euphysidse develop from Tubularian polyps of the genus Corymorpha, the latter are probably to be regarded as the older ancestral forms of the Siphonanthas. CORM and COPMIDIA. All Siphonophorae are pre-eminently characterised by the development of a stock (corm or colony), that is to say, of an individual organism which is composed of several polymorphic persons (zooids or "individuals proper"). The laws and modifications of this compositeness or colony-formation have been as yet but slightly investigated, though they are undoubtedly very interesting and important. I distinguish in the first place simple and multiple colonies. The simple colony (Cormus simplex) consists of a single centralised group of persons, as in all Disconanthse and in the monogastric Siphon- anthas. The compound or multiple stock, on the other hand (Cormus compositus), is formed from the union of several individualised groups of persons, or " groiqis of individuals scattered upon the stem," — the Cormidia. Such colonies are represented by the polygastric Siphonanths. Each cormidium usually represents, in its general composi- tion of several polymorphic persons, a simple stock, and is in most cases monogastric. There are, however, polygastric cormidia, e.g., Apolemia, Salacia, Physalia. In all REPORT ON THE SIPHONOPHORSE. 7 polygastric Siphonanthaa the cormidia are lateral, originally metameric branches of a segmented trunk or jointed stem. This appears, however, in very varied form, as the cormidia occur in distinct order or irregularly scattered and dissolved. ORDINATE CORMIDIA. In most polygastric Siphonanths (and therefore in the great majority of now existing genera of Siphonophorse) the cormidia are ordinate, that is, regularly arranged as the metameres of the jointed stem or corm axis ; the internodes, or regular intervals of the stem between each two cormidia, are often quite free, especially in much elongated corms, as for instance in almost all polygastric Calyconectse, in a number of the Physonectse (Apoleniidse, many Agalmidse) and of the Cystonectas (Salacidse, many Rhizophyskke). Not unfrequently in these stocks the elongated stem is so strikingly jointed by the annular strictures separating the internodes at equal intervals, that the uniform and all- pervading metameric structure of the Articulata is quite equalled. This comparison is the more permissible, since the apical portion of the stem (representing the head) is distinguished by the higher morphological differentiation of its group of persons. The poly-organ theory might conceive these regularly jointed forms as Siphonophorse articulata? in contrast to the others or Siphonophorse inarticulatse. But even when the stem is much shortened and the cormidia so closely compressed that the internodes are hardly distinguishable, the cormidia are often arranged with great regularity in a com- pressed spiral row, as in the Discolabkke and Rhodalidse. In others, and often in nearly related forms, the regular arrangement disappears, and gradually passes into the scattered disposition of Cormidia dissoluta. DISSOLVED CORMIDIA. While in the majority of polygastric Siphonanths the corms are distinctly articulate, and the cormidia are arranged in regular succession, this original arrangement is more or less lost in one portion of this group, and in some entirely. The dissolution usually begins in this way, that the siphons and gonophores belonging to one cormidium separate ; the latter bud off directly from the stem, often regularly alternating with the first, as in Polyphyes among the Calyconectaa, in Linophysa, Nectophysa, Rhizophysa, among the Cystonectae, and in many Agalmidse among the Physonectae. In consequence of further dissolution of the stem arrangement, the palpons and the bracts also lose connection with the cormidia, and bud out directly from the stem, as in several Agalmidse and Forskalidae. Finally the ordinate arrangement is quite lost, and the entire stem exhibits hundreds or thousands of different appendages (siphons, palpons, gonophores, bracts, &c.) in irregular grouping, so that it is impossible to distinguish 8 THE VOYAGE OF H.M.S. CHALLENGER. the various connected components of the broken up cormidia (Physcdia, Agahnopsis, and other Agalmidaj). This appearance is of the greatest interest, because, within one and the same family (e.g., Agalrnidse, Rhizophysidse), most nearly related genera exist, of which one possesses perfectly ordinate cormidia, another completely scattered, and a third an exact transition between these two. In this fact lies the direct morphological evidence of the multiplication and dislocation of the portions of the Siphonophoral stock. DISLOCATION and MULTIPLICATION of ORGANS. If our medusome theory is correct, the title of real persons (or " individuals proper") is only to be applied to those portions of the Siphonophoral stock which have originally the morphological value of a medusome-person, and not to those portions which were originally only organs of such a person. If this be so, it is necessary to assume in many cases a far-reaching dislocation and multiplication of the parts which were originally organs of a medusome. This assumption is directly supported by the fact that the replacement of primary organs by equivalent secondary structures does to a very large extent take place. As such vicarious organs (reserve or replacement organs, " Ersatz- organe"), I interpret, for instance, the numerous swimming-bells and covering bracts of many Physonectse, the successive heteromorphic swimming-bells of the Calyconectse, the groups of palpons in many Physonectse. On the other hand, in such an instance as the clustered groups of reproductive members, each separate " gonophore " is to be interpreted as a medusoid person which has lost mouth-opening and tentacles. The morphology and " sociology " of the Siphonophorse must take much more account than heretofore of these important tectological distinctions, and the rank of individuality must be more strictly defined. MONOGASTRIC and POLYGASTRIC CORMIDIA. The ordinate cormidia contain usually only a single siphon, more rarely two or more. The most important forms of monogastric cormidia (with a single siphon) are the following : — 1. The Eudoxome of the Calyconectte (often becoming free as a " Eudoxia"), in which each cormidium consists of two persons, one sterile (siphon with tentacle and covering bract) and one fertile — the gonophore (often with accessory gonophores). 2. The Ersseome of the Calyconectae (often becoming free as an " Ersxa "), in which each cormidium consists of three persons, having a medusoid " special swimming-bell " added to the two persons above mentioned in the eudoxome. 3. The Rhodalome of some Rhodalida?, of Hippopodws, Vogtia, Aurophysa, Canno- REPORT ON THE SIPHONOPHOR^E. 9 physa, Arethusa, &c, in which each cormidium consists of a siphon with a tentacle and of one or more gonophores. 4. The Athorome of Physophora and of the Anthophysidee, in which each cormidium consists of a siphon with a tentacle, one or more palpons, and one or more gonophores. 5. The Crystallome of Crystallodes, Anthemodes and other ordinate Physonectse, in which each cormidium consists of a siphon with a tentacle, one or more palpons, one or more gonophores, and a group of bracts. Much rarer and much less manifold are the ordinate polygastric corrnidia, in which each group of persons contains several siphons (each with a tentacle) ; such occur in Apolemia among the Physonectse, and in Salacia among the Cystonectse. The line of the stem, in which the cormidia originally bud forth in uniform succession, is the ventral median line of the protosiphon ; usually it is rolled up in a wide or narrow spiral, rarely it remains straight (Crystallodes, Stephanomia). STEM or TRUNK. (Ccenosome, Ccenosarc, Axial Body.) The stem of the Siphonophorse, or the central axial body, on which all the various persons and organs of the corm are attached, is generally compared to the stem of a Hydropolyp stock. This comparison is in our opinion in the accurate sense illegitimate ; for in the latter the primary larval body from which the corm develops is a Polyp-person, while in the former it is a Medusa-person. The comparison holds so far, however, inasmuch as the trunk is branched in both cases. It is not logical to describe the stem of the Siphonophorse as " unb ranched " as is generally done. In reabty it is always branched, for all the appendages — whether they be interpretable as persons or as organs — arise as lateral branches of the axial body. Only the forking or dichotomous ramification is here absent. Further, the conventional description which is generally given of the stem or axial body of the Siphonophoras is strictly applicable only to one legion of the class — to the Siphonanthse. For it is here only that the stem (whether it be long or short) is formed from the primary siphon, and has all its appendages (or branches) budded off in a row from its ventral median line. That they are subsequently often radially disposed depends wholly upon a secondary spiral twisting of the stem, and displacement of its appendages. It is quite different in the second legion, the Discon- anthse. Here the proper stem, i.e., the common central portion of the corm, is formed from the primary umbrella, and all its appendages (or branches) bud out from its lower or subumbrellar surface, not in one row, but in concentric circles or rings, which are originally octoradial. On the other hand the primary siphon in this legion has only the value of a central nutritive organ. (ZOOL. CHALL. EXP. — PART LXXV1I. 1888.) Hhhh 2 10 THE VOYAGE OF H.M.S. CHALLENGER. NECTOSOME and SIPHOSOME. In all Siphonophorse, without exception, both in the monogastric and polygastric forms, the developed corm may be distinguished, both from a morphological and from a physiological point of view, into two main portions, the nectosome and siphosome. The nectosome or the swimming body forms in the Siphonophorse which swim quietly on the surface of the sea, the upper, anterior or proximal portion of the stock ; the siphosome or nutritive body the lower, posterior or distal portion. Physiologically the former is the organ of locomotion (and often also of respiration), the latter, on the other band, the organ of nutrition and reproduction. If one compares the Siphonophoral stock with the simple medusoid person, then the nectosome in general represents the umbrella, and the siphosome of the corm the manubrium of the Medusa. In the polygastric Siphonanths the two portions pass through an independent course of development, separated from one another, with distinct points of growth. In those Siphonanths where the stem is spirally coiled, the spiral twisting is usually opposite in the two portions ; the nectosome has usually a left-handed ("Lambda"), the siphosome a right-handed ("Delta"), spiral. NECTOSOMES of the FIVE ORDERS. The swimming apparatus of the Siphonophoree is formed of two essentially different organs, the swimming-bell (nectophore) and the swimming-bladder (pneumatophore). The nectophore is the umbrella of a Hydromedusa, in which both the annular muscular layer of the velum and of the subumbrella, and the primary canal-system (four radial canals united by a marginal circular canal) are perfectly developed. The pneumatophore, on the other hand, is an invaginated and much modified umbrella, in the surface of which (by apical or lateral invagination of a gas-gland) a chitinous gas-filled "air-flask" or pneu- matocyst is formed. The definition of the five Siphonophoral orders is in the first place determined by the different formation of the swimming apparatus. I. The Calyconectye or Calycophoridse possess, only one, two, or more swimming-bells, but no pneumatophore. II. The Cystonectae or Pneumatophoridse bear only a simple large pneumatophore, but no swimming-bells. III. The Disconectse or Chondrophoridse possess an octoradial pneumatophore, usually composed of concentric annular chambers, but no swimming-bells. IV. The Physonectge or Physophoridse bear on the apex of the stem a simple pneumatophore, and under that a biserial or multiserial column of swimming-bells ; sometimes instead of these a corona of numerous bracts. V. The Auronectae or Aurophoridse (a new and hitherto quite unknown group of very remarkable deep-sea forms) possess on the apex of the stem a colossal pneumato- REPORT ON THE SIPHONOPHOR.E. 11 phore, and underneath that a (simple or multiple) horizontal corona of swimming-bells, and in the middle of the latter (in the dorsal median line of the nectosorne, opposite the ventral row of buds) a large aurophore, a new, hitherto unknown organ, which is to be regarded as a swimming-bell modified into a large gas-gland. NECTOPHOEES or SWIMMING-BELLS. (Nectocalyces, Nectozooids, Schwimmglocken , Schvnmmh ohlenstiicke. ) The swimming-bells generally exhibit so clearly the structure of the umbrella of a simple hydromedusa person, that there can be no doubt as to their morphological import as the locomotive organ of such a person — in the sense of the poly-organ theory. The cavity of the gelatinous disc is clad throughout by the circular muscle layer of the suburnbrella, and its distal margin protrudes in the form of a true velum, at the base of which runs the circular canal which unites the four radials. The poly-person theory interprets these facts, however, in this way, that each swimming-bell is to be regarded as a medusoid person or "locomotive zooid," in which the locomotor umbrella is well developed, but the nutritive manubrium degenerated. To this the medusome theory has to reply that such a phylogenetic interpretation is not corroborated by the known ontogenetic facts. The protocodon, or the "primary swimming-hell" of the Siphonula is only an organ of this medusoid person. It does not remain as such in most (if not all) of the Calyconectse, but becomes thrown off, and replaced by metacodons or heteromorphic "secondary swimming-bells." In the monogastric Calyconectse the protocodon is modified into the covering bract of the Eudoxia, whde in the Physonectaj, Cystonectse, and Auronectse it becomes the pneumatophore. The numerous metacodons, which in the Physonectse and Auronectse, as well as in the Polyphyidae, form the two-tiered, many -tiered, or radially disposed swimming column, are either displaced umbrellas of medusomes, which are repeated metamerically in the several cormidia, or are simply vicariae or "reserve bells" ("Ersatzglocken") which arise by the multiplication of such umbrellas, and are often present in great numbers. PNEUMATOPHORE or SWIMMING-BLADDER. (Air-Chamber, Luftkam mer, Schwimmblase.) The hydrostatic organ of the Siphonophorse, which is known as pneumatopliore, is only absent in one order, that of the Calyconectse. In the other Siphonophorse it is generally present and that in two different forms; the Physonectse, Cystonectse, and Auronectse possess a simple swimming-bladder at the superior or apical pole of the stem; 12 THE VOYAGE OF H.M.S. CHALLENGER. the Disconectae, on the other hand, have a complicated apparatus, composed of concentric annular chambers, which occupy the greater portion of the umbrella. In all cases the pneumatophore arises very early in the primary medusoid larva, and that by a gland- like invagination of the exoderm of the exumbrella. In the bilateral Siphonula of the Siphonanths this invagination has an excentric position (being often shifted down near to the umbrellar margin), but in the octoradial Disconula of the Disconanths it is central, in the very apical pole. The marked and apparently considerable differences of the developing pneumatophore, in relations both of time and space, between different (often nearly related) Physonectae, I simply explain as secondary cenogenetic modifica- tions. That portion of the medusoid larval body on which the first trace of it appears always belongs originally to the exumbrella. In the Siphonanthae the invaginated portion of the exumbrella, comparable to a simple, pouch -like, glandular sac, is known as the air-sac (pneumatosaccus); it secretes in its superior or apical half a chitinous membrane, the air-flask (pneumatocystis), while its inferior or distal portion (the "air-funnel") discharges the function of a gas-gland (pneumadenia). The glandular (usually yellowish or greenish) epithelium of this last portion secretes the air, which passes by the basal opening of the air-flask (" funnel aper- ture," " Trichterpforte," or pneumatopyle) into the latter. The Cystonectae or Pneumato- phoridae possess at the apical pole of the pneumatocyst an external air-hole or stigma (the primary opening of invagination), by which the air may be discharged at will. In many Siphonanths the air-secreting glandular epithelium of the air-funnel (or infundibulum) grows through the aperture into the basal half of the pneumatocyst and clothes the latter as endocystal tapetum, or " secondary ectoderm" (Chun). In most Siphonanthae the air-sac becomes subsecpiently united with the peripheral (uninvaginated) portion of the primary umbrella — the pneumatocodon — by a number of vertical radial septa, usually eight, more rarely four or sixteen, and occasionally a variable number. The radial pockets between these septa open inferiorly into the central canal of the stem and represent the radial canals of a simple medusoid umbrella. On this is based the opinion that the entire air-chamber is to be regarded as an " invaginated swimming-bell " (Metschnikoff 1874). In contrast to this the air-chamber is at present regarded by most zoologists as an independent medusoid person (or "medusiform zooid"), and is supposed to originate as a "bud" from the primary larva (Leuckart 1875, Claus 1878, Chun 1887). The antithesis between these two opinions culminates in the inter- pretation of the air-sac, which according to the first theory is the exumbrella of a medusoid disc, according to the second the subumbrella. The latter opinion is according to my conviction entirely erroneous, the former is in a certain sense admissible. The compara- tive ontogeny of the Sipkonophorae appears to me to show conclusively that the air-sac is an apical gas-gland of the exoderm, which in the Disconula of the Disconanthae sinks centrally into the gelatinous disc from the apex of the same inwards, and in the Siphonula REPORT ON THE SIPHONOPHOR^E. 13 of the Siphonanthse lies excentrically beside the apex. It arises in the former, indeed, at the place where some Medusae (e.g., Aglaura) form a suctorial organ of attachment. Only subsequently does this apical air-gland become greatly extended, to occupy the greater part of the original larval umbrella, so that one may in a certain sense speak of an " invagination." At any rate, the " air-sac " is originally a portion of the ectodermic epithelium of the exumbrclla, not of the subumbrella. The whole pneumatophore is not a secondary medusoid person, but the modified disc of the primary medusoid larva. The many-chambered pneumatophore of the Disconanthas is comparable in its first beginning alone with the single-chambered pneumatophore of the Siphonanthse ; there again it is at first nothing more than an apical gas-gland, which occupies only a small area on the apex of the primary larval umbrella. This simple central chamber becomes rapidly surrounded, however, with a girdle of eight radial chambers, and around these again many concentric rings of chambers are apposed. Thus arises the characteristic octoradial, concentrically chambered air-disc of the Porpitidse, of which the amphithect pneumatocyst of the Velellidse represents only a secondary modification. Sometimes its margin is split into eight or sixteen radial lappets. On the upper surface the disc opens by numerous stigmata, of which the central one alone can be compared to the simple apical stigma of the Pneurnatophoridae. From the lower surface proceed numerous small air-tubules or tracheae, which enter into the so-called "liver" or glandular central organ. I regard only the entodermic canal network of this voluminous "central organ" as " liver " (and perhaps " kidney "), and the compact exodermic parenchyma as a strongly developed gas-gland. The lower ends of the tracheae are not closed, as is generally supposed, but open. They take off the secreted gas, and their terminal apertures correspond to the " funnel aperture of the air-flask " in the Siphonanthse. On the other hand, the stigmata of the upper surface (or of the exumbrella) in the Disconanthae do not serve for the reception of atmospheric air from outside (as is generally supposed), but solely for the exit of the secreted gases, and correspond to the simple apical pore of the Cystonectse. AUROPHORE or AIR-BELL. The new deep-sea famUies Stephalidae and Rhodalidse are distinguished from all the other Siphonophorae by the possession of a peculiar organ, belonging to the nectosome, designated the aarophore. Since they are in other respects peculiarly organised, they may be regarded as representing a special order, Auronectse. The aurophore is a voluminous gas-gland, which has developed out of a medusiform swimming-bell. It lies below the large pneumatophore, in the dorsal middle line of the stem, opposite the ventral row of buds. The disc-cavity of the swimming-bell has been modified into a 14 THE VOYAGE OF H.M.S. CHALLENGER. narrow muscular air-duct (auroductus), which opens internally at the base of the pneumatophore, and externally to the outside. The radial canals of the metamorphic nectophore are modified into a complex radial system of wide glandular chambers. SIPHOSOME or NUTRITIVE BODY. The nutritive body exhibits in the different groups of the class Siphonophorae much more manifold and important differences of form and structure than the swimming body. Undoubtedly the most important difference is this, that in the Monosiphonise (or the monogastric Siphonophorae) the archisiphon or protosiphon (the primary gastral tube of the medusoid larva) remains alone as the organ for the reception and digestion of food, while in the Polysiphonise (or the polygastric Siphonophorae), "secondary suctorial tubes" or metasiphons are developed by budding, each provided with a gastric cavity and a mouth opening. In the Polysiphonise the primary mouth opening of the medusoid larva only rarely persists, in all Disconanthse, and in two families among the Siphon- anthae (Stephalidse and Physalidae). In most (perhaps all the rest) of the polygastric Siphonanthse, the primary mouth opening of the protosiphon is probably closed, and the latter persists only as the stem of the stock (truncus or ccenosome). The polymorphic appendages, which bud out as lateral branches from the stem of the siphosome, are partly medusoid (bracts, gonophores), partly polypoid (sip>hons, palpons, cystous, gonostyles). The poly-organ theory regards the former as multiplied umbrellas of one Medusa, or as secondary vicaria of such, the latter as repeated manubria or vicaria of the same. The poly -person theory, on the other hand, regards each single bract as a medusoid person, which has lost all its organs except the umbrella, and each single siphon and palpon as an independent Hydropolyp. Our medusome theory regards in the different cases these polymorphic appendages partly as dislocated organs of medusomes, partly as multiplied reserve organs or vicaria of the same. SIPHONS or SUCTORIAL TUBES. {Polypites, Gastral Tubes, Stomach Sacs, Nutritive Polyps, Eating Polyps, Hydranth s, Gastrozooids. ) The siphons, which have given the name Siphonophorae to the entire class, are the most important and the most constant appendages of their organism. From a physiological point of view they are rightly regarded as organs for the reception of food and digestion ; from a morphological point of view they are sometimes regarded as homologous with an entire Polyp, sometimes with the gastric tube or manubrium of a Medusa. REPORT ON THE SIPHONOPHOR.^. 15 As in the latter, each siphon or " suctorial tube " possesses at the distal end of the sac- shaped body a mouth opening, at the proximal end a stem opening, through which the simple cavity communicates with the cavity of the umbrella or the stem. By one or two (often even three) annular strictures the simple cavity of the siphon is divided into two, three, or four chambers. When four portions can be distinctly distinguished, the first (proximal) is a thin stalk (pedunculus siphonalis), usually bearing the base of insertion of the tentacle ; the second is a thick-walled crop (" Vormagen," basigaster) equipped with masses of urticating cells ; the third is the stomach proper (stomachns) , usually with "hepatic stripes," more rarely with glandular villi; the fourth is an ex- tremely contractile proboscis. The stomach usually passes without sharp boundary into the proboscis, but is generally definitely marked off from the basigaster, often by means of a pyloric valve. The stalk of the crop region is often degenerate or not distinctly defined. The single tentacle springs from this portion in the majority of Siphonanthse. According as the adult Siphonophoral corm has only one or several siphons, we distinguish Monosiphonise and Polysiphonise. MONOSIPHONLE or MONOGASTRIC SIPHONOPHOR/E. Of great importance for the right understanding of the Siphonophorse is the fact, that in all forms the primary medusiform larva (Disconula or Siphonula) bears only a single siphon, and that this remains in one portion of the class, namely in the Monogastricae, while in the other division it is modified into the stem, and is physio- logically replaced by numerous secondary siphons (lateral branches of the latter). Hitherto the monogastric Siphonophorse have been represented solely by the Eudoxise among the Calycophoridse ; but among the new deep-sea Siphonophorse of the Challenger Expedition which I have described in this Report, there are interesting monogastric forms from three other orders (Discahdse, Athoridae, Cystalidae). Since in both legions of the class the Monosiphonise exhibit close resemblance to the known larval forms of the Polysiphonise, they may be regarded as " sexually mature larvse." In more accordance, however, with the phylogeny is the reverse interpretation, that those larvse of the Polysiphonise repeat, according to the fundamental biogenetic law, the structure of their Monosiphonial ancestors. In regard to the position of the stomach on the subumbrella, the Monosiphonise exhibit in both legions very important differences, which are explained by their diphyletic origin. In the Discalida? or monogastric Disconanths, the primary siphon occupies the central point of a regular octoradial umbrella ; in the monogastric Siphonanths, on the other hand, it is more or less excentrically situated at the base of a bilateral umbrella. 16 THE VOYAGE OF H.M.S. CHALLENGER. POLYSIPHONLE or POLYGASTRIC SIPHONOPHORffi. In the great majority of Siphonophorse the adult form possesses several suctorial tubes or polypites as organs for the reception and digestion of food. These arise in the Dis- conanthae by budding from the primary subumbrella, in the Siphonanthse on the other hand by budding from the ventral middle line of the primary siphon, which is modified into the stem. In the former these " metasiphons " surround the primary archisiphon as a corona ; while in the latter they are disposed upon the protosiphon either on the one side only or on all sides, in a spiral line. In the simplest and most primitive case, such as occurs in the polygastric Calyconectee and in many Physonectse, the metasiphons develop in regular metameric succession on the segmented stem, separated by wide internodes. Each individual siphon is (in the Eudoxise and the corresponding simplest Prodoxise of the Physonectse) associated with a covering bract ; both together form a medusome, the umbrella being represented by the bract, the manubrium by the siphon. In most of the Physonectas numerous covering bracts soon develop, which are to be considered as mere multiplications of the primary bract, and therefore as entirely subordinate organs. But when from the base of such a medusome gonophores bud forth — appendages that is to say of the morphological value of a medusoid person — then such a " group of persons " acquires the value of a cormidium. In many Polysiphoniae the metameric arrangement of the cormidia subsequently breaks up, and then the connection between the scattered eiphons and the separated sexual medusomes is often no longer demonstrable. PALPONS or TASTERS. (Feelers, Tasters, Arms, Fluid Receptacles, Hydrocysts, Dactylozooids.) In the great majority of the Siphonophoraa, the siphosome bears, scattered between the siphons, or connected in groups with the latter, the tasters or feelers. These are always simple, thin-walled, very contractile sacs, in which the proximal portion communicates with the cavity of the stem, while the distal end is closed. Morphologically the tasters are to be regarded as mouthless manubria, or as the stomachic sacs of medusomes in which the umbrellas have become modified into covering bracts or are entirely degenerate. The palpons are distinguished from the cystous by the absence of a distal opening, from the siphons not in this alone, but also in the absence of the glandular villi and hepatic stripes in the stomach region. Their function appears to be mainly, if not exclusively, sensory. Their sensitive point probably acts generally as a taste organ, and sometimes also as an eye ; in a (new) Athorybia I observed a lens in this ocellus (a sickle-shaped pigment spot on the upper surface of the sensory apex). In some Agalmidse the distal REPORT ON THE SIPHONOPHORjE. 17 part appears to act as an otocyst, since it is constricted off by a strong sphincter from the wide palpon cavity, and forms a spherical terminal bladder in which a crystalline otolith is kept in rotatory movement by a ciliated epithelium. Palpons are found generally distributed in the Physonectse and Physalidae ; they are absent in the Auronectae, Calyconectas, and Disconectae. Not unfrequently palpons are confused with tentacles, as for instance repeatedly by Claus, even in Physojihora. CYSTONS or ANAL VESICLES. Under this title I distinguish from the other polypoid organs of the Siphonophorae certain vesicular sacs, which have hitherto been generally confused with the palpons. They are indeed very like the latter, but are very essentially distinguished from them by a terminal aperture. By means of this aperture, which the animals can open or close at will, fluid and excretions are emptied from the canal-system, and water may also be taken in. It is therefore to be regarded as an anus. The distal portion of the cystons is frequently pigmented and furnished with special glandular cells, they also sometimes contain definite crystalline excreta. These anal vesicles obviously stand in closer morphological and physiological relations with the siphons than with the palpons ; they are, however, distinguished from the former by the simple structure of the wall. I find these excretory structures widely distributed among the Physonectaa, but they appear to be altogether absent in the other orders. TENTACLES or CAPTURING FILAMENTS. (Senkfaden, Stinging Filaments, Tentacular Filaments, Nematozooids.) Tentacles are present in all Siphonophorae, and are both as capturing organs and as offensive and defensive weapons quite indispensable. Recent reports of their absence in some species are certainly to be explained by the ease with which the delicate threads are during capture detached from their base. The poly-person theory regards the tentacles as independent polypoid persons. I agree, on the other hand, with the poly- organ theory, which explains them as organs of the siphons. In my opinion the tentacles of the two legions of the class are in their nature entirely different. In the Disconanths (or Porpitariae) the primary circle of tentacles on the umbrellar margin of the medusoid larva (Disconula = Archimeda) persists ; there are at least eight present ; their number usually becomes much increased ; they have no individual relations to the separate siphons which bud forth in large numbers froni the suburnbrella. It is (ZOOL. CHALL. EXP. PART LXXVII. — 1888.) HhlA 3 18 THE VOYAGE OF H.M.S. CHALLENGER. quite different in all the other Siphonophorse — the Siphonanthse. Even their primary medusoid larva (Siphonula = Protomeda) always possesses only a single tentacle, and this is excentrically attached to the base of the primary siphon. The phylogenetic cause of this peculiar position is to be sought in the bilateral modification, which the older four-rayed medusoid ancestors of this legion (Codonidte) have undergone through the degeneration of three marginal tentacles (Euphysidse). Only a single tentacle remained persistent, and was all the more strongly developed ; it shifted in consequence of the ventral splitting of the umbrella to the inferior surface of the latter beside the base of the siphon. A similar centripetal migration of tentacles occurs also in Medusae, sometimes on the superior, sometimes on the inferior surface of the umbrella. The fact that each individual siphon of the polygastric Siphonanths bears at its base only a single tentacle is simply to be regarded as the (hereditary) result of a metameric repeti- tion of the primary condition. Only in a few Siphonophorse do the tentacles remain simple cylindrical filaments, as in the Velellidee, in Apolemia, Stephalia, Linophysa, Salacia, and allied genera. In the great majority a row of lateral branches (side-filaments, accessory tentacles, secondary filaments, or Tentilla) is formed, as in the Cladonernidaa (Pteronema, Gemmaria). The terminal equipment of the latter by manifold cnidonodes or " stinging knobs" is often very characteristic of the several genera and species. The Porpitidse are distinguished by the possession of three longitudinal rows of secondary branches. Some supporters of the poly -person theory regard each of these accessory organs as an autonomous person, and Claus even states that the mantle-like covering (involucrum) which in many Physonectae surrounds each stinging knob, apparently corresponds morphologically to the disc of the Medusa (1878). In my opinion all these appendages, however complex their structure, are merely subordinate organs of the second order. PALPACLES or TASTING FILAMENTS. Under this designation I include only the long, extremely contractile, hollow, simple filaments, which occur in the majority of Physonectae at the base of the j>alpons. They have the same morphological and genetic relation to the latter as the tentacles to the siphons. While the tentacles are especially of importance as capturing organs and as weapons, the palpacles act as fine, far reaching, tasting organs auxiliary to the sensory function of the palpons. Each palpon bears always only a single palpacle, and this is always simple, never stalked. . While the supporters of the poly-person theory regard each palpacle as an independent " individual," that is as a distinct person, I must on the other hand agree with the poly-organ theory, which explains them simply as subordinate organs of the palpons. REPORT ON THE SIPHONOPHOR^. 19 BRACTS or COVERING PIECES. (Hydrophyllia, Protecta, Deckblatter, Covering Scales, Phyllozooids). The polymorphic organs which are usually spoken of as " covering pieces" (Deckstiicke) are entirely absent in the three orders Disconectae, Cystonectae, and Auronectaa ; in the former originally, in the two latter probably as the result of degenera- tion. In the two remaining orders the bracts are essentially distinguished in this, that in the Calyconectae they appear singly on each Medusome, in the Physonectae on the other hand there are several. Only the primary larva of the latter (Siphonula) frequently forms a single " primary covering piece " (Protobractea), which is then to be compared with the single bract of the former (Eudoxia), and to the umbrella of the primary ancestral form (Protorneda). The physiological importance of the bracts consists exclusively in their protective function ; they are shields or umbrellar organs, under the shelter of which the other persons of the colony are protected. As regards their morphological import, the poly- person theory regards them as degenerate medusoid persons, which have lost manubrium and tentacles, while the gelatinous disc has been the more developed; the poly-organ theory, on the other hand, regards them as multiplied umbrellas. According to our medusome theory, a distinction must be drawn between primary and secondary bracts. The protobracteae or primary covering pieces, which occur singly, on the larvae of Physonectae and on the Eudoxiae of Calyconectae, are to be interpreted as the umbrella of a primary medusoid person. The metabracteae or secondary covering pieces, however, which usually cover the stem of the Physonectae in large numbers, may have various phylogenetic origins. They may arise as — (1) Displaced umbrellas of secondary medusomes ; (2) Vicariae or multiplied reserve-bracts of the same ; (3) Cleft portions of divided (e.g., quadripartite) umbrellas. The direct transition of nectophores into bracts is to be observed among the Physonectae in the Athoridae and Anthophysidae. In Athoria and Rhodophysa, I find in the distal portion of each bract a small rudimentary nectosac or swimming cavity, sometimes with four cnidonodes or stinging knobs, which may be regarded as the rudi- ments of four reduced tentacles. The highest and most manifold development of bracts is found in the Calyconectae, where the sterile medusome of each single cormidium bears a large hydrophyllium of a peculiar form, often characteristic of the genus. The phyllo- cyst, too, or the canal of the bract, here often exhibits several apophyses, which may be regarded as rudimentary radial canals of the umbrella; e.g., Aglaisma, the free Eudoxia of Calpe, possesses four radial canals in its bract, two paired lateral and two odd sagittal (an ascending and a descending canal). 20 THE VOYAGE OF H.M.S. CHALLENGER. GONOSTYLES or REPRODUCTIVE STALKS. (Blast ostyles, Gonoblastidia, Gonodendra, Klinozooids.) The gonophores or the medusiform "reproductive persons" of the Siphonophoras sometimes develop directly on the stem, that is as buds on the gastral wall of the primary siphon. Such is the case in many monogastric Siphonophorse, and in several polygastric forms (e.g. , in Agalmidse) with broken-up cormidia. Much more frequently the buds arise from the stomach-wall of secondary polypoid persons, for which we shall use Allman's title of Gonostyles or Blastostyles. These sacs are usually mouthless, and are described as " tasters," but are better designated sexual palpons. Sometimes (in the Porpitidse and Velellidae) these gonopalpons possess a terminal mouth, and may then be called sexual siphons. These sacs frequently branch in a very marked tree-like fashion, so that the numerous gonophores attached to them form large clustered masses (reproductive-trees, gonodendra). From a physiological point of view the gonostyles are distinguished both from the palpons and from the siphons very essentially in this, that their sole function is the production of gonophores. From a morphological point of view they may be regarded as sterile Polypo-persons, which are related to their gonophore buds, as the Hydropolyp stock to its medusoid buds (poly-person theory). In this case the succession of generations (strophogenesis) would replace the original alternation of generations (metagenesis). According to the poly-organ theory, on the other hand, each gonostyle is only to be regarded as an organ, corresponding to a Medusa-manubrium, from which sexual medusoid persons of the next generation arise by budding. GONOPHORES or REPRODUCTIVE PERSONS. (Sexual- Medusoids, Gonozooids. ) Among all parts of the Siphonophoral organism, the gonophores are always those members which most distinctly preserve the original character of the medusoid person. The two principal organs, the umbrella and the manubrium, are always present. In the wall of the latter the generative products arise. But only in the Disconectse (and perhaps in a number of Cystonectse) does the oral aperture of the manubrium appear to break through; in the others it remains closed. The reproductive cells arise in both sexes of the Siphonanthse from the entire surface of the manubrium (as in the Codonida?), and indeed from the exoderm. The umbrella of the gonophores has the velum and the circular canal of the umbrellar margin usually well preserved, and the same is true REPORT ON THE SIPHONOPHORAE. 21 of the four radial canals of the subumbrella. Only rarely do these portions become more or less rudimentary, so that the medusoid form is lost and passes over into a more or less degenerate " sporosac." The tentacles on the umbrellar margin of the gonophores have usually disappeared ; sometimes, however (as in some Calyconectse), four rudiments are recognisable, and at other times even a corona of tentacular rudiments (Desmojrfiyes). The Disconanthas, in which free reproductive Medusae in the sexually mature state are not yet sufficiently known, may perhaps bud off four or eight separate gonads either from the manubrium, or from the subumbrella. As to the distribution of the two kinds of gonophores on the corm, it must be noticed that most of the Siphonophorae are monoecious. Some Calyconectse {Mitroj>hijes, Galeolaria) and some Physonectae (A'polemia, Athoralia) are dioecious. Among the monoecious corms both kinds of sexual persons occur, sometimes in separate cormidia (diclinous), sometimes associated in each single cormidium (monoclinous). The gono- dendra, or the clustered groups of gonophores, are usually distylic, all their branches bearing Medusae of the same sex; male and female gonodendra arising separately. But sometimes there are monostylic gonodendra, the basal part of the single branched gonostyle bearing female, and the distal part male, gonophores (as in the Auronectae and Cystonectas, Forsbdia, &c). On the other hand, hermaphrodite persons {i.e., gono- phores which form both spermatozoa and ova) are not known in this class. The male persons are known as androphores, the female as gynophores. Usually the androphores are slender, more oblong, and bear a spindle-shaped or cylindrical spermarium ; the gynophores are thicker, more rounded, and bear an ovate or ellipsoidal ovarium. FUNDAMENTAL FORM (PROMORPH). The promorphology of the Siphonophorae has led to .very different views respecting the ideal geometrical fundamental form (or the " promorph "), which may be recognised by abstraction from the concrete single forms of these Acalephs. One group of authors regards the Siphonophorae as " Radiate animals," whilst another group maintains that they are "Bilateral animals." These opposed views are both right in a certain sense. On the one hand, a typical radial structure — generally a quadriradial form, composed of four equal quadrants — is recognisable in all Siphonophorae (mainly in the gonophores and nectophores), and this is in my opinion produced by inheritance from the older ancestral Medusae. On the other hand, a distinct bilateral structure — or a dipieural fundamental form, composed of two symmetrical halves or antimeres — is recognisable in most Siphonophorae, viz., in all Siphonanthae, whilst it is wanting in the Disconanthae. This bilateral type is partly inherited from the bilateral Medusae, which we regard as the ancestors of the Siphonanthae (Protomeda), and the essential form of 22 THE VOYAGE OF II. M.S. CHALLENGER. which is presei'ved in the bilateral larva, Siphonula ; but partly the bilateral type is produced by adaptation, and mainly by accommodation to the conditions of development of the single forms. In order to get a clear conception of these difficult promorphological relations, we must distinguish, firstly, the fundamental forms of the entire cormus and of the single persons or medusomes composing it ; and secondly, the different promorpho- logical development in the two independent legions of the class, the Disconantha^ and the Sipbonanthse. The promorph of the single medusome has the quadriradial medusoid type usually clearly expressed, when the umbrella is well preserved, as in the gonophores and necto- phores (always with four radial canals and a connecting marginal ring-canal). But also in other parts of the medusomes the radial promorph may be recognised, as in those siphons which possess four, eight, or sixteen hepatic stripes, mouth lobes, &c. Promorph of the Corms in the Disconanthw. — The ideal geometrical fundamental form exhibits in the colonies of Disconanthse two different types ; one of these, represented by the Discalidae and Porpitidae, is the primary and original type; the other, exhibited by the Velellidae, is a secondary modification. All corms of Discalidoe (Pis. XLIX, L.) and of Porpitidse (Pis. XLV.-XLVIII.) preserve a completely regular octoradial structure ; their ideal promorph is a regular octagonal pyramid. The vertical main axis of this pyramid, around which the eight equal parameres are regularly arranged, bears at its superior or apical pole the apical stigma of the central chamber of the pneumatocyst, at its inferior or basal pole the mouth of the central siphon. The eight equal sides of the pyramid are represented by the eight triangular radial chambers of the pneumatocyst, whilst the eight perradial grooves between these, and the eight canals running in the grooves, further the eight primary tentacles at the distal end of the canals, mark the eight edges of the pyramid. The horizontal lines which connect these edges with the vertical main axis are opposed in four pairs, and represent the four primary or perradial cross-axes of the octagonal pyramid ; whilst the four secondary transverse axes alternating with these, and bisecting the eight triangular radial girdle-chambers of the pneumatocyst, are interradial. The perfectly regular octoradial promorph, which is so clearly marked by the structure of the central pneumatocyst, is likewise expressed by the entire structure of the canal-system, the eight primary perradial canals of the exumbrella, the centradenia, the subumbrella, &c, by the regular octoradial corona of the eight primary tentacles and gonostyles, the eight gastral ostia in the fundus of the central siphon, the eight lappets of its mouth, &c. The young larva of all Discabdse and Porpitidse (Disconula, PI. L. figs. 9, 10), and the simplest permanent genus of this legion (Disccdia, PI. XLIX.), exhibit the octoradial type in the same complete regularity as any octoradial Medusa (e.g., Trachynema, Pectyllis). The Velellidae (Pis. XLIIL, XLIV.) differ from the regularly octoradial Porpitidaa and Discalidae in the amphithoct modification of the promorph, which is usually called REPORT ON THE SIPHONOPHOR^E. 23 " bilateral," sometimes even " uniradial " (Chun). But a comparative study of their development, and mainly of the central part of their body, demonstrates that the corm of all Velellidae originally is also octoradial ; this is clearly proved by the girdle of eight radial lobes which surrounds the central chamber of the pneumatocyst ; further, by the eight primary radial canals, which arise from the gastrobasis of the central siphon and give off the ascending branches forming an octoradial liver-star on the upper surface of the centradenia. Rataria (PI. XLIV.), as the simplest form of this family, and the similar Ratarula-larvse of other Velellidae, possess marks of an octoradial type also in other organs (eight primary buds of gonostyles, sixteen primary tentacles, &c). But whilst the eight parameres in the Discalidae and Porpitidse are precisely equal (each composed of two symmetrical antimeres), they are unequal in the Velellidae (each composed of two asymmetrical antimeres). The two horizontal cross-axes, which are perpendicular one to another, and to the vertical main axis, are both equal in the Discalidae and Porpitidse ; whilst they are unequal in the Velellidae. Their sagittal axis (marked by the vertical sail) is longer than the frontal axis (marked by the transverse groove of the pneumatocyst). But the dorsal half of the corm (on one side of this frontal groove) is symmetrically equal to the ventral half (on the opposite side), in the same manner as the right half is symmetrically equal to the left (separated from it by the vertical sail). The fundamental form of the corm in the Velellidae, therefore, is not "bilaterally-symmetrical," but " amphithect," or bilaterally-octoradial, similar to that of the Ctenophorae. The octoradial structure of the Medusae is originally derived from the quadriradial type, as I have demonstrated in my Monograph of the Medusae. The same promorpho- logical law is valid also for the Disconectae. Their octoradial trunk has arisen from the umbrella of some Medusa, the older ancestors of which were quadriradial. Their next ancestral forms may be Trachynemidae, with eight radial canals, eight tentacles, &c. But these again have arisen from the older Petasidae, which possess four radial canals, four tentacles, &c. This older quadriradial structure is still preserved in the medusiform gonophores of the Disconectae (Discomitra), Promorph of the Corms in the Siphonanthse. — The corm of the Siphonanthae, differing from that of the Disconanthae in all respects from the first beginning, is also distinguished completely by its bilateral promorph. The primary larva (Siphonula), which develops the corm of the Siphonanthae by unilateral budding from its manubrium, has already a markedly bilateral fundamental form. Its primary umbrella has a deep ventral cleft, and its only tentacle is attached to one side of the manubrium. The vertical plane, which passes through the median line of these parts, is the sagittal plane, and bisects the entire body ; the two halves separated by it, right and left, are symmetrically equal. That side of the manubrium from which arise the buds of the corm is the ventral side, the opposite the dorsal side. 24 THE VOYAGE OF H.M.S. CHALLENGED In order to get a clear conception of the bilateral promorph of the Siphonanthaj (very unnaturally described by many authors), it is very important to distinguish the three primary dimensive axes, and to compare them with those of man or of some other bilaterally symmetrical animal. The first or principal is the vertical main axis, the longitudinal axis of the tubular stem ; its upper or apical pole bears the pneumatophore in the majority of Siphonanthas, the primary nectophore in the Cabyconectse. The opposite end is the lower or basal pole. The second dimensive axis is the sagittal diameter ; its ventral or anterior pole is marked by the series of buds, and usually by a ventral groove of the tubular stem. The opposite dorsal pole is distinguished in the Auronectae by the aurophore, in Physalia by the crest of the pneumatocyst. The vertical plane, which is determined by the sagittal and the principal axes, perpendicular one to another, is the median or sagittal plane ; it separates the right and left halves of the body. The distinction between right and left halves (often confounded by different authors, and described in striking contradiction) is always clear, when the ventral side is constantly defined in the same sense as that side of the body from which the buds arise. Therefore, the two poles of the frontal diameter, or the third dimensive axis (right and left pole), must be always the same. In the monogastric Calyconectae, for instance (Eudoxia, &c), the single siphon is placed on the ventral side of the bract (or the modified umbrella) ; in the polygastric Calyconectae, correspondingly, the trunk is placed on the ventral side of the first or proximal nectophore (the nectosarc, therefore, on its dorsal side). In the Diphyidse, the ventral sides of the two nectophores are opposed one to another. The bilateral promorph of the Siphonanthse is at the same time quadriradial (or by duplication of the parameres octoradial). This radial structure, inherited from the ancestral quadriradial Medusas, is not only evident in the four radial canals of the gonophores and nectophores, the eight hepatic stripes and mouth-lobes of many siphons, &c, but also in the structure of the primary larval umbrella, and the pneumatophore arising from it. The majority of the Siphonanthas exhibit in the basal part of the pneumatosaccus eight (more rarely four or sixteen) radial pouches, which are separated by vertical septa and comparable to the radial canals of a Medusa. DESCRIPTION OF THE FAMILIES, GENERA, AND SPECIES. SIPHONOPHOR^E. Definition of the Class. — The Siphonophorse are swimming colonies or corms of Hydromedusse, composed of polymorphous medusoid zooids or persons, which arise by 1 mdding from an original simple Medusa. The class is divided into two subclasses : — the Disconanth^e arise by budding from tbe subumbrella, the Siphonanth^e by budding from the manubrium of the original Medusa. The Disconanthce comprise one order only (Disconectae) ; the Siphonanthse comprise four orders — Calyconeetse, Physonectse, Auro- nectas, Cystonectae. Synopsis of the Five Orders of Siphonophorse. I. DISCONANTFLE : Truncus or coenosome formed by the umbrella of the original octoradial Medusa, which includes a polythalamous pneumatocyst; the buds arise in concentric girdles from the subumbrella. Larva octoradial (Disconula), . 1. Disconect.e. II. SIPHONANTH.E. Truncus or coenosome formed by the manu- brium of the original bilateral Medusa ; the buds arise in the ven- tral line of the manu- brium. Larva bilateral (Siphonula). No pneumatocyst or float filled by air. No palpons. Always one or more nectophores are present, 2. Calyconect.e. A monothalam- ous pneumato- cyst or a float filled with air is always present. AVith nectophores or with bracts (often both to- gether present). With palpons. Without aurophore ; common stem thin, tubular, with a simple canal, With an aurophore ; common stem thick, bulbous, with a net- work of canals, 3. Physojstectj;. 4. AUEONECT.E. Without nectophores, and without bracts. Pneumatocyst with an apical stigma, . 5. Cystostect.e. (ZOOL. CHALL. EXP. — PART LXXVII. — 1888.) Hhhh 4 26 THE VOYAGE OF H.M.S. CHALLENGER. Subclass I. DISCONANTHAE. Order I. DISCONECTvE, Haeckel, 1888. (Pis. XLIII. to L.). Clwndrophoras, Chamisso, 1821, 16,1 p. 363. Velellids, Eschscholtz, 1829, 1, p. 165. Cirrhigradx, Blainville, 1834, 24, p. 303. Porpitariee, Haeckel, 18G9, MS. Canar. Definition. — Siphonophorae with a permanent primary umbrella, without necto- phores and bracts. Nectosome campanulate, lenticular or discoidal, including a poly- thalamous, originally octoradial pneumatocyst, which exhibits numerous stigmata on its upper face, and tracheae on its lower face. Siphosome composed of a central primary siphon, and one or more concentric girdles of gonostyles (either secondary siphons, or instead of these palpons), the latter producing gonophores. The primary larva (Disconula) has a regular octoradial umbrella, with eight radial canals and a connecting circular canal at the margin, which bears eight primary radial tentacles. The order Disconectae comprises three families, the Discalidae, Porpitidae, and Velellidae ; all members of this order agree in so many important characters, and differ so widely from all the other Siphonophorae, that I divide the whole class into two subclasses, Disconanthae and Siphonanthae. The first subclass, Disconanthae, represented by the Disconectae only, is developed from the octoradial and octonemal medusoid larva Disconula ; it retains the primary corona of eight or more marginal tentacles, possesses a centradenia, and produces the polymorphous persons by budding from the subumbrella. The second subclass, Siphonanthae, on the other hand, represented by four orders (Calyconectae, Physonectae, Auronectae, and Cystonectae), differs in the bilateral form of its mononemal larva, Siphonula ; this, as well as each of the following siphons, has only a single tentacle ; the centradenia is wanting, and the polymorphous persons of the cormus bud in the ventral line of the primary siphon. History. — Eschscholtz, in his fundamental work,2 separated from the other Siphono- phorae the family Velellidae, comprising the genera Porpita and Velella of Lamarck (1815). So early as 1821 these were united as "Medusae chondrophorae " by Chamisso and Eysenhardt. Eschscholtz added as a third genus Rataria, and found the peculiar character of all Velellidae in the possession of an internal polythalamous cartilaginous shell, the chambers of which are filled with air. Brandt afterwards (in 1835, 25) separated the Porpitidae (with circular shell, without vertical crest) from the true Velellidae 1 The figures in black type refer to the Bibliography at the end of the Report. 2 System der Acalephen, 1829, p. 165, (1). REPORT ON THE SIPHONOPHORjE. 27 (with elliptical or oblong shell, and a vertical crest). Huxley (in 1859) gave the first accurate definition of this group (9, p. 71) : — " Physophoridse without nectocalyces or hydrophyllia ; with short, clavate, simple or branched, submarginal tentacles. A single central principal polypite. Pneumatocyst flattened, divided into chambers by numerous concentric partitions, and occupying almost the whole of the discoidal coenosarc." The description which Huxley gave of the two genera Velella and Porpita is very accurate, as is also that given in 1853 by Kolliker (4). Another description of both has been recently published by Alexander Agassiz (1883, 57). His opinion is that these Siphono- phorse are more nearly allied to the Tubularise, and on the other side to the Hydro- corallinse (Milleporidse, &c). The oldest figures are those of Forskal (11). All authors hitherto agree in the general opinion that the Disconectse are the most highly developed of all Siphonophorse, and that they form the terminal group in their systematic arrangement, as being most closely allied to the Physalidse. In my opinion this general view is quite erroneous. I am convinced, mainly by their ontogeny, that the Disconectse have preserved the original medusoid structure more than any other Siphonophorse (or the Siphonanthae). They retain the original umbrella with its corona of marginal tentacles, and the original manubrium, which is the " principal polypite," or better, " central siphon." The gonostyles, or the polypiform persons which bear the gonophores (either mouthless palpons or mouth-bearing siphons), bud from the sub- umbrella (or the lower face of the disc) ; while the polymorphous persons in all the other Siphonophorse bud from one side of the manubrium (or the primary siphon) in its ventral line. The primary larva of all the Disconectse (Disconula) is a regular octoradial Medusa, with a marginal corona of eight radial tentacles, widely different from the larva of all other Siphonophorse (the Siphonula, with bilateral umbrella and a single tentacle). I suppose, therefore, that there is no direct relation between the Disconanthse and the Siphonanthse, and that the two subclasses or legions have a different origin : the former arising from Anthomedusse (Codonidaa, Euphysidse), the latter from Tracho- inedusse (Trachynemidaa, Pectyllidse). Nectosome and Siphosome. — The two main parts of the fully-developed body, which we separate as nectosome (or swimming apparatus) and siphosome (or nutritive apparatus), have in the Disconectas a mutual relation very different from that of the other Siphono- phorse (the Siphonanthag). The nectosome is represented by the exumbrella, or the upper half of the discoidal body, which includes the polythalamous pneumatocyst ; the sijjhosome, however, is formed by the subumbrella, or the lower half of the horizontal disc, and by the different organs which are attached to its margin and to its lower face. The boundary face between the two main parts is the upper face of the centradenia or the so-called " fiver "; it is in close contact with the lower face of the pneumatosaccus. The constant organs, which depend from the subumbrella, are the following : — (I.) The large central siphon, in the middle of the lower face ; (II.) a simple or multiple corona 28 THE VOYAGE OF H.M.S. CHALLENGER. of gonostyles, which bear the gonophores (mouthless palpons in the Discalidse, mouth- bearing siphons in the Porpitidae and Velellidse) ; (III.) a simple or multiple corona of submarginal tentacles. Umbrella. — That part of the body, which is the true homologue of the umbrella in the common Medusae, is in the Disconectse the most voluminous part of the whole cormus. It includes the polythalamous pneumatocyst, and bears all the other organs on its lower face. The latter corresponds to the subumbrella of the typical Medusae, while the upper face is homologous with their exumbrella. Both faces are separated by the peripheral margin or limb of the umbrella, which constantly bears a corona of muciparous glands. The exodermal epithelium of the umbrella everywhere includes scattered nematocysts. Beyond it there is a nervous plate, mainly developed on the limb and the upper face. The muscles of the umbrella are more developed on the upper than on the lower face ; they are composed of an outer layer of longitudinal or radial fibres, and an inner layer of transverse or circular fibres. Exumbrella. — The superior (apical or proximal) part of the umbrella, which is sepa- rated from the inferior part (or subumbrella) by the glandiferous limb, is sometimes a nearly even horizontal disc, at other times more or less convex. Its upper free face is either smooth or papillate, always pierced by stigmata ; its lower face is in contact with the centradenia. The exumbrella in all Disconectse is composed of two parallel plates, which are separated by the reticulate plexus of the pallial canals ; the outer plate is the pneumatocodon, the inner is the pneumatosaccus. Pneumatophore, — The hydrostatic apparatus, or the swimming-bladder, which we call pneumatophore, has in the Disconectse another and a far more complicated structure than in all the other Siphonophorse (or the Siphonanthas). Only the first beginning can be common to the two subclasses, viz., a simple pneumadenia, or a gas-producing gland of the exumbrella. But this is centrally placed in the Disconanthse, excentrically in the Siphonanthse ; and further, the chitinous pneumatocyst, which covers the inside of the pneumadenia, is a simple monothalamous cyst in the latter, a multiple polythalamous cyst in the former. Another important difference is furnished by the openings of the pneumatocyst ; the simple pneumatocyst of the Siphonanthse is either closed, or has only a simple opening (infundibulum) at the lower pole, and another simple mouth (stigma) on the upper pole of its main axis. The polythalamous pneumatocyst of the Disconanthse, however, has numerous openings on both faces, internal tracheae on the lower face, external stigmata on the upper face. Pneumatocodon. — The pneumatophore as a whole, or the hydrostatic apparatus, is composed of all the above-mentioned parts, and of the two plates of the exumbrella also already referred to. The outer or upper of these is the pneumatocodon, the uppermost lamella of the entire umbrella. It is composed of three layers, an outer exodermal epithelium, a nervous plate (composed of a loose reticulum of branched ganglion cells), REPORT ON THE SIPHONOPHOR.ffi. 29 and a double muscle plate, with outer radial and inner circular fibres. The plexus of pallia! canals separates the pneumatocodon from the pneumatosaccus. Pneumatosaccus. — That lamella of the exoderm which surrounds immediately the chitinous pneumatocyst, is the pneumatosaccus. It represents the invaginated part of the exumbrella, and has taken its origin from a simple bottle-shaped gland in its apex. This gas-secreting gland — or pneumadenia — originally small and occupying only the apical centre of the exumbrella, has afterwards become so extraordinarily expanded that it usually forms the most voluminous part of the entire umbrella. The exodermal epithelium of the pneumatosaccus is a simple stratum of glandular cells, which secrete the cuticular chitinous plate of the adjacent pneumatocyst. The basal part, or the inferior face of the former, is probably also the matrix of the centradenia or the so-called " liver." The exodermal cells, and cnidoblasts, which constitute the solid parenchyma of the latter, are probably derived from the basal part of the pneumatosaccus. Pneumatocystis. — The chitinous polythalamous float filled with air, which we call pneumatocyst (formerly called " inner shell "), exhibits in all Disconectae a rather com- plicated structure. Its general form is circular, and originally octoradial in the Discalidse and Porpitidge, elliptical or nearly quadrangular (parallelogram) in the Velelliclse ; but also in the young larva? of the latter its first rudiment is octoradial. It always commences with the formation of a simple central chamber, which is situated in the centre of the exumbrella, just above the gastral base of the large central siphon. It opens outside by a central stigma in its upper face. Around this primary central chamber (the chitinous lining of a central pneumadenia of the exumbrella) a peripheral corona of eight radial chambers is next formed, each provided with an outer stigma on its upper side, and with an articulate trachea on its lower side. These eight radial chambers are equal and regularly radial in the Discalidse and Porpitidae, while they are more or less amphithect and somewhat bilaterally disposed in the Velellidae. Sometimes in the latter family they are more or less obliterated. In the simplest case (Discalia, PL XLIX.) the formation of the pneumatocyst is complete with the eight radial chambers ; in all the other Disconectae a different number of peripheral concentric chambers is formed around their octoradial corona. All these tertiary chambers are simple rings without radial partitions ; they open outside (in the exumbrella) by a different number of stigmata, inside (in the centradenia) by a number of open tracheae. The rings are circular in the Discalidae and Porpitidae, elliptical or quadrangular (parallelogram) in the Velellidae. In these latter there usually arises after- wards a solid vertical crest, placed diagonally on the upper side of the horizontal disc. The general opinion regarding the physiological function of the polythalamous pneuma- tocyst of the Disconectse may be summed up in the following propositions : — (1) The Disconectse are exclusively pelagic animals, always floating on the top of the ocean, and never sinking below its surface ; (2) the air contained in the pneumatocyst is atmo- 30 THE VOYAGE OF H.M.S. CHALLENGER. spheric air, taken in by the stigmata of the exumbrella ; (3) this air is conducted through the respiratory tracheae to the different appendices of the suburnbrella (central siphon and gonostyles) ; (4) the distal or lower ends of the tracheae are closed. The observations which I have myself been able to make on the structure and function of the pneumatophore in the different groups of Disconectae have conducted me to quite opposite views, viz., — (1) The Disconectae are (all or partly) capable of sinking under water, by muscular compression of the pneumatocyst, and expulsion of air by the stigmata of the exumbrella. (2) The air contained in the pneumatocyst is not atmospheric air taken in by the stigmata, but a gas produced by the exodermal cells of the great "central pneumadenia" (the so-called "liver" or central organ). (3) This gas, therefore, has not respiratory, but hydrostatic functions (like the gas in the swimming-bladder of the fishes). (4) The distal or lower ends of the tracheae are not closed by a chitinous plate, but open into the glandular, gas-secreting, exodermal tissue of the centradenia ; they are comparable to the " pylorus infundibuli " of the Siphonanthae. Central Siphon. — The large central polypite, which is placed in the centre of the suburnbrella in all Disconectae, is the original manubrium of the primary medusiform larva (Disconula) ; its terminal mouth is the permanent primary mouth of the latter. The central siphon is the only organ of feeding and digesting in the monogastric family Discalidae, whilst in the other two families of the order, the polygastric Porpitidae and Velellidae, this function is also executed by the numerous peripheral gonostyles, which are here developed in the form of mouth-bearing siphons or secondary polypites. But also in the young monogastric larvae of these latter, the primary central siphon is alone provided with a mouth. Its size is very variable in the different Disconectae ; generally it is comparatively large in the small Discalidae, and in the smaller forms of Porpitidae and Velellidae, which possess few secondary siphons ; on the other hand, it is relatively small in the larger forms of the two latter families, which possess a great number of feeding peripheral polypites. The form of the central siphon in the Discalidae and Porpitidae is inversely conical, with circular transverse section ; whereas in the Velellidae the inverted cone is stronelv compressed from both sides, so as to be elliptical or lanceolate in transverse section. The wider proximal or upper part, or the true stomach, passes without a sharp boundary into the cylindrical, very contractile, distal or lower part, the proboscis. The latter, as well as its distal opening, the mouth, is very variable in size and form, according to its state of contraction. The surface of the central siphon exhibits in many Disconectae a number of longi- tudinal or radial folds, visible partly outside, partly inside ; usually there are eight or sixteen, sometimes more. The distal mouth correspondingly often exhibits eight or sixteen lobes, sometimes also four larger lobes ; at other times it is simply circular, or REPORT ON THE SIPHONOPHOE^E. 31 in the Velellidae elliptical. In the expanded state the mouth may assume the most different forms, trumpet-shaped, discoidal, &c. The thick wall of the central siphon is always very contractile, and composed of five layers: — (l) An outer epithelium or epidermis, rich in cnidocysts ; (2) a very thick layer of strong longitudinal muscles ; (3) a solid fulcrum or a homogenous structureless elastic supporting plate ; (4) a thin layer of circular muscles ; (5) an inner epithelium or entoderm. The mouth is armed with peculiar cnidocysts and pro- vided with sensillae. The upper or proximal half of the central siphon contains in its epidermis often (but not always) a number of bent tracheae, which end here. Gastrobasal Plate. — The basal or proximal part of the central siphon in all Disconectae is separated from the superjacent centradenia by a strong fulcrum, or a structureless elastic supporting plate (lamina gastrobasalis). This horizontal fulcral plate (also called the roof of the central polypite, " le plancher " of Bedot, 59, 60) is covered on the upper face by the basal surface of the centradenia, on the lower face by the entodermal epithelium of the stomach. It corresponds to the jelly-plate which forms the roof of the manubrium in the Medusae. Its central part is solid, whilst its peripheral part is pierced by eight or more gastral ostia ; these form an octoradial corona in the Discalidae and the smaller Porpitidae, whilst their number is increased in the larger forms of the latter family (sixteen to thirty-two or more). Sometimes the numerous ostia form here vertical lanceolate fissures, and the septa between them form an elegant multiradiate star, composed of numerous vertical lamellae. The Velellidae exhibit instead of this regular star a bilateral arrangement of the gastral ostia ; they form here two opposite longitudinal rows of fissures (usually sixteen) on the two lateral margins of the lanceolate gastrobasal plate. Centradenia or Central Gland. — The central space of the body, between the apical or proximal pneumatophore and the basal or distal central siphon, is in all Disconectae occupied by a peculiar large glandular organ, wanting in all the other Siphonophorae (or Siphonanthae). This interposed central organ is usually called the liver (hepar); but as its structure and function are complicated and not merely hepatic, it may be better called centradenia, or central gland. It is composed essentially of a dense network of entodermal gastral canals, and of a compact parenchyma of exodermal epithelium, with innumerable cnidoblasts, filling up the meshes or intervals of that network. The physiological function of the gastral canals may be partly hepatic (digestive), partly renal (excretory) ; the exodermal epithelium, however, seems partly to perform the function of a pneumadenia (or gas-producing gland), partly to be a large reservoir of cnidoblasts for other purposes. The Form of the Centradenia is in general lenticular or discoidal, sometimes sub- globular or even cylindrical, sometimes more conical or flatly expanded ; its peripheral outline is circular, or sometimes regularly octagonal, in the Discalidae and Porpitidae ; it 32 THE VOYAGE OF H.M.S. CHALLENGER. is elliptical or sometimes lanceolate in the Velcllidse ; often the periphery is more or less lobate, corresponding to the form of the surrounding peripheral part of the umbrella. The upper or proximal surface of the centradenia is attached to the lower or distal surface of the pneumatosaccus, and separated from it by a thinner or thicker fulcrum, a firm and elastic, structureless, supporting plate. But this supporting septum is wanting in the young larva?, where the exodermal epithelium of the pneumatosac is in immediate contact and connection with the upper face of the centradenia, All the various and often complicated apophyses (radial ribs, circular rings, &c.) which are formed by the distal or inferior face of the pneumatocyst, and its enveloping jmeumatosac, are surrounded by corresponding furrows or envelopes of the proximal or superior face of the centradenia ; the latter being always immediately attached to the supporting plate, which separates it from the former in the adult Disconectse. The lower or distal surface of the centradenia is in its central part in contact with the base of the central siphon, in its peripheral part with that portion of the subumbrella which bears the gonostyles. The fundus of the central siphon, or the uppermost part of its cavity, is separated from the attached centradenia by a strong support, that sup- porting plate, which has been described above as " lamina gastrobasalis." This elastic and structureless plate is pierced in the periphery of the basal part of the stomach by eight or sixteen (sometimes more) radial canals, which arise from the fundus of the stomach itself and pass into the vascular system of the centradenia. We call these the primary perradial gastro-canals. There are originally eight equal and regularly disposed radial canals in the Discalidse and Porpitidse. In the Velellidse, on the other hand, they are arranged in a bilateral manner ; two larger radial canals (a dorsal and a ventral) arising from the opposite poles of the longer (sagittal) axis, two smaller (right and left) from the poles of the shorter (transverse) axis, and four other (diagonal) canals in the middle between the latter and the former. But usually the number of intercalated secondary canals is here much larger, and their arrangement more or less irregular. Origin of the Centradenia. — For the right understanding of this peculiar organ of the Disconectse, a knowledge of its origin and development is essential. This question may be solved by the comparison of very young Disconectas and mainly of the Discalidse, These simplest and most primitive Disconectse remain permanently in a lower stage of development, which is transitory in the Porpitidse and Velellicke. In the smallest Disconula-larvse which I observed the centradenia is a small circular, biconvex, lenticular disc ; its upper face is in direct contact with the pneumatosaccus, whilst its lower face is separated from the central siphon by the gastrobasal plate (" plancher" of Bedot). The entire mass of this solid disc is composed solely of exodermal cells and cnidoblasts ; it is not traversed by any canals. The only canals of the centradenia are the eight simple radial canals which run upon its upper face ; they arise from the eight ostia of the basi- gaster, embrace the surface of the lenticular centradenia like eight meridian lines, and REPORT ON THE SIPHONOPHOR^E. 33 unite in the centre of its upper face, forming a typical octoradial " liver star." This is quite regular in the youngest Discalidse and Porpitidse ; it is amphithect in the young Velellidse, where the two opposite canals of the sagittal axis are larger than the six others, three right and three left (compare Bedot, 60, pi. ix. figs. 1, 2). These eight primary liver-canals, running between upper face of the centradenia and lower face of the pneumatosaccus, must be regarded originally as ascending branches of the eight primary radial canals of the subumbrella, which run from the basigaster towards the peripheral limb. All the numerous canals (partly hepatic, partly renal vessels) which traverse the centradenia of the larger Disconectaj in all directions, and form a complicated network, are merely secondary branches and ramifications of those eight superficial canals of the primary "liver-star." They seem to be disposed and differentiated in a variable manner. The majority of Disconectaa exhibit in the adult state a superior system of hepatic vessels (characterised by production of brown or blackish pigment granules) and an inferior system of renal vessels (characterised by the production of greenish guanin- crystals), the so-called " white plate " of its discoverer, Kolliker.1 The cnidoblasts and the intermingled matter, indifferent or interstitial cells of the exoderm, which, densely aggregated, constitute the solid glandular parenchyma of the centradenia, doubtless have their origin in the exoderm. But the locality of their origin, or the point of their exit between the central siphon and pneumatosaccus, has hitherto been doubtful. Bedot, who has given the best and most accurate histological description of the " central organ " of the Velellidse (58-61), assumes that the place of their origin is the subumbrella. According to his description,2 a number of subumbral exoderm-cells immigrate into the interior, passing through numerous pores of the subumbrellar support (his " lamelle aniste externe," loc. cit.). These pores possess, in my opinion, a secondary importance, and are perhaps artificial openings. It seems to me much more probable that the exodermal cells of the centradenial parenchyma may be derived from the basal or inferior part of the pneumatosaccus, i.e., that invaginated lamella of the exumbrellar exoderma which encloses and produces the chitinous pneumatocyst ("couche cellulaire qui tapisse le pneumatophore " 3). This important part of the pneumatosac is, in young Disconectse, in immediate contact with the upper face of the centradenia, as Bedot himself has demonstrated.4 The structureless support, which separates the two organs in the adult (" lamelle aniste interne " of Bedot, loc. cit., p. 238), is not yet formed. This, in my opinion, is right, and the constituting exodermal cells of the centradenial parenchyma are derived from the exumbrellar invagina- tion of the pneumatosaccus ; they may be compared to the glandular gas-secreting cells of the " infundibulum pneumatophori " of the Siphonanthse. I have very often found in well-preserved specimens the cnidoblasts of the centradenia filled with an air-bubble, and 1 Kolliker, 4, p. 59. 2 Bedot, 59, p. 503, pi. xxv. figs. 4 le, 8 o. 3 Bedot, 59, pi. xxv. figs. 4, bin. * Bedot, 60, p. 238, pi. ix. fig. 2. (zool. CHALL. EXP. — paet lxxvii. — 1888.) Hhhh 5 34 THE VOYAGE OF H.M.S. CHALLENGER. this seems to confirm my opinion that the centradenia is not only a hepatic and a renal, but at the same time a gas-secreting gland. Canal System. — The system of gastral vessels or entodermal canals is in the Disco- necta? very different from that of all the other Siphonophorse (or the Siphonantha?) ; it is far more complicated and exhibits from the beginning quite a different type ; it is originally octoradial, determined by eight primary perradial gastral canals, which arise from the periphery of the base of the stomach and correspond to the eight subumbrellar radial canals of a common octoradial Medusa (Trachynema, Pectyllis). This regular, strictly octoradial type is retained through life in the central part of the canal system of all Discalidse and Porpitidse, whilst in the amphithect Velellidae it is found only in early stages, and afterwards becomes bilaterally modified, according to the different growth of the two horizontal cruciate axes. No trace of this octoradial canal system is found in the trunk of any other Siphonophora? ; but a similar type is apj^arently marked by the octoradial structure of the pneumatosaccus in many Physonectse. The peculiar development of the gastro-canals in the umbrella of the Disconectse is produced, firstly, by the voluminous expansion of the pneumatophore and the invagination of the exumbrella connected with it ; and secondly, by the development of the centra- denia between the latter and the subumbrella. The following eight parts of the canal system may be distinguished in the most highly developed Disconecta?, as we ascend from the central siphon to the top of the pneumatophore : — (1) The system of subumbrellar radial canals (originally eight), arising from the fundus of the stomach, and running horizontally and centrifugally or ascending in the subumbrella towards the margin of the umbrella ; they usually form a complicated network of dichotomously branched radial canals. (2) The marginal canal, which connects the distal ends of the latter ; it is placed in the true margin of the umbrella, and corresponds to the circular canal of the Hydromedusse. (3) The system of renal canals or white excretory vessels, containing green crystals of guanin ; it is formed by a network of branches of the subumbrellar canals, which is placed usually in the basal or inferior part of the centrodenia. (4) The system of hepatic canals or brown pigment vessels ; it is formed by the apical or superior part of the canal network in the centradenia, in the surface of which it forms a regular octoradial " liver-star." (5) The system of exumbrellar or pallial radial canals (originally eight) ; these run centripetally and more or less horizontally in the exumbrella towards the centre of the pneumatophore, where they are united by a small ring, surrounding the apical stigma (" mantle-star ") ; they may be compared to the eight radial pouches of the pneumatophore in many Siphonanthae. (G) The system of gono- stylic cavities, or the gastral cavities of the polyps which bear the gonophores (palpons in the Discalidse, siphons in the Porpitidse and Velellidae) ; they arise as simple subumbrellar diverticula from the inferior branches of the centradenial system. (7) The canal system of each single gonophore, composed of four radial canals and a connecting ring-canal, REPORT ON THE SIPHONOPHOR^E. 35 identical with that of a simple Medusa ; the central union of the former is connected by a pedicular canal with the gastral cavity of the gonostyle. (8) The system of the tentacular canals or the simple peripheral vessels which arise from the periphery of the subumbrellar system and pass into the tentacles. Besides these eight parts of the canal system there occurs often (9) a coronal canal, or a horizontal circular ring-canal placed in the coronal groove of the umbrella, between the margin of the pneumatocyst and the centradenia. It represents an inner annular vessel which carries on the circulation in the canals of the centradenia and the sub- umbrella. The entoderm al epithelium exhibits a very different and remarkable shape in these different systems of gastro-canals. The most important seems to be the rich production of black or dark brown bdious granules in the hepatic vessels, and of green guanin crystals in the renal vessels. Eegarding the morphological value of these different parts of the gastro-canal system of the Disconectae, and comparing them with the corresponding parts of a simple octo- radial Medusa (Trachynemida), we may arrive at the following important conclusions : — (1) The eight primary perradial canals, which arise from the base of the stomach (of the central siphon) and run in the subumbrella towards the margin, are homologous with the eight subumbrellar centrifugal radial canals of a simple Trachynemid (e.g., Pectanthis). (2) The marginal ring-canal, which connects the former and runs along the limb of the umbrella inside the series of marginal glands, is homologous with the usual marginal canal of a simple Medusa. (3) The tentacular canals, which arise from the subumbrellar canals and pass into the tentacles, are comparable to the tentacular canals of those Medusae which possess submarginal tentacles (e.g., Drymonema). (4) The gonostylar canals, which arise from the subumbrellar canals and pass into the cavity of the gonostyles, are homologous with the cavities of the eight genital sacs of a Trachynemid. (5) The eight centripetal radial canals, which arise from the coronal canal, run in the upper face of the centradenia to its centre and there unite into a " liver star," may be compared to the centripetal canals of the subumbrella of Carmarina, Pectyllis, and other Trachy- medusae. (6) The coronal canal, or the inner ring-canal, which runs in the coronal groove on the margin of the pneumatocyst, may be compared to the inner ring-canal which develops by anastomoses of the radial canals in some Medusaa. Whilst these parts of the canal system of the Disconectae may be compared to corre- sponding parts of simple Trachyrnedusse, there are other parts which are quite peculiar to the former. These secondary productions are : — (1) The pallial system or the anasto- mosing radial canals of the exumbrella, which form a network on the upper face of the pneumatophore. (2) The internal reticular canal system of the centradenia, forming a hepatic plexus in its upper and a renal plexus in its lower half. The development of the superficial pallial system of canals is a consequence of the invagination of the apical part of the exoderm which produces the pneumatosaccus. The development of the internal 36 THE VOYAGE OF H.M.S. CHALLENGER. centradenial system of canals, on the other hand, is the consequence of the centripetal immigration of exodermal parenchyma between pneumatosac and gastrobasal plate, producing the centradenia. The innumerable ramifying and anastomosing branches of the above-mentioned vessels, which form a very complicated network in the majority of Disconectee, are secondary productions. Their mutual relations are best understood when we compare them with the typical organisation of the simplest forms of that order, Discalia and Disconalia (Pis. XLIX., L.). The canal system exhibits here three typical octoradial horizontal stars : — (l) The pallial star above the pnoumatocyst ; (2) the hepatic star on the upper face of the centradenia ; (3) the gastral star below the gastrobasal plate. The eight peripheral main rays of these stars are united by the coronal ring vessel, which runs in the coronal groove around the equator of the umbrella (PL XLIX. fig. 4). Gonostyles. — The sexual persons of the cormus, or the zooids which produce the medusiform gonophores of the Disconectse, are arranged in a girdle around the base of the central siphon ; they occupy a smaller or broader zone of the subumbrella between it and the corona of tentacles (gonostylar zone). These gonostyles or gonoblastidia are mouthless palpons in the Discalidae, whilst they are either cystons or true siphons, pro- vided with a terminal mouth, in the Porpitid?e and Velellicke. They are originally nothing else than secondary manubria of a single Medusa person, budding from the radial canals of the subumbrella. They may consequently be regarded as secondary persons, compar- able to the buds of the Gastroblasta (described by Keller and Lang), and of other Hydro- medusae budding from the subumbrella. On the other hand they may be compared also to the genital sacs which depend from the radial canals of the subumbrella in the Trachy- nemidae, Aglauridse, Pectyllidae, and in other families of Hydromedusse. If we imagine that these genital sacs, instead of themselves producing ova and spermatozoa in their exodermal wall, produce by budding Medusae which afterwards become sexually mature, we shall understand how the Disconectse have originated from Trachomedusse. The mouthless medusiferous pahpons of the Discaliclse, and the similar secondary siphons of the Porpitidse and Velellidae (differing only in the possession of a mouth at the distal end), have the same structure as the large primary central siphon, but are always much smaller. Their number is originally eight, and they form a regular octo- radial corona in some smaller and simpler genera {Discalia, PI. XLIX. figs. 1-4 ; Porpalia, PI. XLVIII. ; Rataria, PI. XLI V.). Sometimes there are sixteen, e.g., in Disconalia (PL L.) and Porpitella (PI. XLVL). But usually their number is much increased, and amounts in the larger species to some hundreds. These cover the greatest part of the subumbrella, the large gonostylar zone between the central siphon and the corona of submarginal tentacles, often densely crowded without intervals. The form and size of the gonostyles are very variable, owing to their great contractility. Some- times they are more spindle-shaped, at other times more cylindrical or pyriform, with a REPORT ON THE SIPHONOPHOR^E. 37 dilated basal part. Sometimes the proximal half, or the stomach, is separated by a constriction from the distal half or the proboscis. The upper or proximal half only produces by budding the numerous medusiferous gonophores, whilst the distal part, or the contractile proboscis, is armed with cnidonodes. The wall is very contractile, since the longitudinal muscles of the exoderm and the circular muscles of the entoderm are not less developed than in the large central siphon. Sometimes the wall of the gono- styles exhibits eight longitudinal folds or ribs ; and often, too, the patches of cnidocysts (or the cnidonodes) are arranged into eight parallel longitudinal rows along the proboscis.1 The gastral cavity of the gonostyles opens above into a canal of the subumbrella, or of the centradenia. Its lower or distal end is a closed caecum in the monogastric Discalidae, the terminal apex being densely beset with cnidocysts. It opens by a terminal mouth in the Porpitidse and Velellidae ; this mouth is not less contractile and expansible than the larger mouth of the central siphon. Often the former exhibits four cruciate lips, more rarely eight ; sometimes it is circular, without mouth lobes (compare the descriptions of the gonostyles by Kolliker (4), Vogt (5), Leuckart (8), Huxley (9), Agassiz (57), and others). Gonophores. — The medusiform gonophores arise from the proximal part of the gono- styles, rarely isolated, usually crowded in smaller or larger groups or bunches. They are in all Disconectae of the same form, and are detached from the budding gonostyle before coming to sexual maturity. The detached gonophores are very small quadriradial Medusae of very simple structure. Their subumbrella exhibits four regular radial canals which unite above the velum by a circular canal (compare the above-mentioned authors). Tentacles. — The limb of the umbrella is in all Disconectae armed with a corona of tentacles, in the same manner as in all fully-developed Hydromedusae. They are placed not at the margin itself, but more or less inside, at its lower face, the peripheral zone of the subumbrella. They are, therefore, strictly speaking, submarginal tentacles (such as occur also in some Medusae, e.g., Drymonema). Some authors (Glaus, Alexander Agassiz, &c.) regard these organs as self-subsistent persons or zooids, and call them " prehensile polypites," " marginal polypites," " tasters," " dactylozooids," &c. But this conception is quite erroneous, and, in my opinion, there can be no doubt that the sub- maro-inal corona of tentacles in the Disconectae are the same organs as in the common Medusae, both from a morphological and from a physiological point of view. Octoradial Corona of Tentacles. — It is a most important fact, not hitherto pointed out as it deserves, that in the larvae of most Disconectae there occurs a typical stage, with a corona of eight equidistant and regularly disposed tentacles. They are placed at the distal end of the eight primary radial canals which arise from the base of the central siphon, run along the subumbrella, and are connected round the margin by the circular 1 Compare 57, pi. ii. figs. 1-8. 38 THE VOYAGE OF H.M.S. CHALLENGER. canal ; just as in the common octoradial Hydromedusse (e.g., Trachynema, Phopalonema). This regular corona of eight simple radial tentacles is permanent in the simplest and oldest form of the Disconectse (Discalia, PL XLIX. figs. 1-4). In all other genera of this order the number of tentacles is rapidly increased, either by interpolation of eight inter- radial secondary tentacles between the eight perradial primary ones, or by budding of new secondary tentacles on both sides of the base of the primary ones, so that there arise eight bunches of tentacles (Disconalia, PI. L. ; Porpalia, PI. XLVIIL). In the larger Porpitidse their number is afterwards so multiplied, that the margin is armed with a rich corona of many hundreds, or even thousands of tentacles (Porpema and Porpita, Pis. XLVII. and XLV.). They are here densely crowded, and arranged in concentric girdles (sometimes sis to nine or more) ; the uppermost (or proximal) girdle usually bears the smallest, and the lowermost (or distal) the largest filaments. Their number is much smaller in the Velellidee, where they form only a single submarginal series in Rataria (PL XLIV.) and Velella, a double (or rarely multiple) series in Armenista (PL XLIIL). Also in this family the original number seems to be eight, and in some smaller forms are found sixteen ; but in consequence of the bilateral development of the umbrella, their number and arrangement is often modified, bilateral, or irregular. In the young larval forms (Ratarula) often two primray tentacles, situated at the opposite poles of the major axis (or sagittal diameter) of the elliptic disc, appear earber than the others ; this heterochronism is certainly kenogenetic. Structure of the Tentacles. — The tentacles of the Disconectas are very different from those of all other Siphonophoraa ; they are relatively short and thick, rather rigid, and their movements are sluggish, as in most Trachoniedusae. In general they are far less extensile and contractile, and do not exhibit that peculiar development and movement which are obvious in most of the Siphonanthee, and are similar to that of the Anthornedusae. The body of each tentacle in all Disconectse is a hollow cylinder with a very strong muscular wall and a narrow canal, closed at the distal end and opening at the proximal end into the annular canal of the margin, or the marginal zone of the canal network. The wall is composed, as usual, of the following five strata, enumerating them from with- out inwards: — (l) An exodermal epithelium, armed with cnidoblasts, often vibratile in some parts ; (2) a strong layer of longitudinal muscles ; (3) a thin, but firm and elastic structureless supporting lamella ; (4) a thin layer of ring-muscles ; (5) a vibratile entodermal epithelium, lining the central canal, composed of very large vacuolate entoderm cells similar to the axial cells in the tentacles of many Trachomedusse. The armature of the tentacles with cnidoblasts exhibits characteristic differences in the families of Disconectse. Discalia, (PL XLIX. figs. 1-4), as the simplest form of all, and likewise probably the youngest larval stages of all Porpitidae, possess eight simple tentacles, which bear a single cnidosphere (or a spherical knob composed of cnidocysts) at their distal end (PL L. fig. 9). REPORT ON THE SIPHONOPHORiE. 39 The next larval stage of the Porpitidse exhibits four stalked cnidospheres at the distal end of each tentacle ; ! one of these is the primary terminal knob, whilst the three others are lateral branches (PL L. fig. 10). By multiplication of the latter in the prolonged tentacle arise three longitudinal rows of stalked cnidospheres, one of which is situated on the inferior or distal side of the tentacle, the other two opposite on its two lateral sides. These three longitudinal rows of short lateral branches, each of which bears a spherical cnidonode, are very characteristic of all Porpitidse ; each row bears six to nine or more branches in the larger species, their size decreasing from the terminal (oldest) branch towards the basal (youngest). The basal half of the tentacles is usually simple, and bears no branches. The tentacles of the Velellidse are never branched ; they always remain simple cylindrical filaments, and are relatively short and small. The cnidocysts are either irregularly scattered on their surface, or form two opposite ribands on the two lateral sides. Ontogeny. — The individual development of the Disconectse is very incompletely known, but seems always connected with a peculiar form of metagenesis. The first generation is asexual, the complicated cormus above described producing at its sub- umbrella numerous polypites or secondary manubria (mouthless palpons in the Discalidse, mouth-bearing siphons in the Porpitidse and Velellidse). From the gastral wall of these secondary polypites (surrounding the sterile central siphon) there arise numerous medusiform buds of the form Discomitra. These do not become mature whilst sessile and attached to their parents, but are soon detached, and develop into free Hydromedusae, which produce ova and spermatozoa. Some advanced stages of this second sexual generation are described by Gegenbaur as Chrysomitra, and possess eight or sixteen radial canals ; but they have not hitherto been sufficiently examined in the adult state. The origin and structure of the sexual organs of the Disconectse, ovaria as well as spermaria, require a further accurate examination. It is very probable (though not observed) that from the fertilised egg of this second generation arises a young Medusa with eight radial canals, and that this early produces in the top of its hemispherical umbrella the pneuinatocyst, at first a simple central chamber (comparable to the simple pneuinatocyst of the Siphonanthse) and subsequently a corona of eight radial chambers. From this common larval stage probably arise two different lines of individual development. The Discalidse, on one hand, remain regularly octoradial, and develop eight marginal tentacles (with increasing number), and between these and the central siphon eight or sixteen gonostyles, remaining mouthless palpons. The Porpitidse retain the same regular octoradial type, but are further developed, and their gonostyles, at first mouthless, acquire afterwards a distal mouth-opening and metamorphose into secondary siphons. On the other hand, a different course is followed by the bilateral Velellidse. Here 1 Compare A. Agassiz, 59, pi. ix. fig. 1. 40 THE VOYAGE OF H.M.S. CHALLENGER. there arises early a vertical longitudinal fold of the exoderm above the pneumatocyst ; and this commencement of the typical vertical sail determines the amphithect or bilateral development of this family ; the umbrella becomes elliptical, and the gonostyles bud between central siphon and corona of tentacles, not in concentric circles (as in the Porpitidae) but in elliptical or oblongish rings. The ontogenetic metamorphosis of these larvae, developing into the first asexual generation, is not completely known in any Diseonectae. For further particulars regarding the ontogeny reference may be made to the works of Kolliker (4), Vogt (5), Gegenbaur (7), Huxley (9), Pagenstecher (55), Alex- ander Agassiz (57), and others. Phytogeny. — The historical or phylogenetic development of the various forms of Sipkonophorae, which we unite in the legion Diseonectae, may be partly recognised from the critical study of their comparative anatomy and ontogeny. By this means I have been conducted to new views of the origin of this interesting subclass, quite different from those of all naturalists who have hitherto treated the question. The general opinion is that the Diseonectae are the most highly developed Siphonophorse, terminating the series of this class, and most nearly related to the Cystonectae (Physalidae). Even the radial chambers of the pneumatophore of the Diseonectae have been often compared to the crista-chambers of the pneumatophore of Physalia. This comparison, and all the important consequences deduced from it, are, in my opinion, perfectly erroneous. In direct contradiction to it, I am convinced of the truth of the new theory which I have already shortly explained in my propositions ; ! its principal points are here repeated. The Diseonectae (or Disconanthae) have no direct relation to the Siphonanthae (or all other Siphonophorae) ; they have originated, independently of the latter, in a different way and from a different group of Hydromedusae. Whilst the Siphonanthae are probably the offspring of the Anthomedusae (Codonidae), and their cormus developed by budding from the ventral line of the original siphon, the Disconanthae, on the other hand, are probably descendants of the Trachomedusae (Trachynemidae), and developed by budding from the subumbrella. The common ancestral group of all Disconectae is the family Discalidae (most nearly allied to the Trachynemidae). From these, probably, the two other families, Porpitidae and Velellidae, have been developed as two divergent branches, or, perhaps, the latter have been derived directly from simpler forms of the former family. Synopsis of the Three Families of Diseonectae. 1. Umbrella circular and regular octoradial. Blastostyles without mouth, . . 1. Discalidae. 2. Umbrella circular, in the centre octoradial. Blastostyles with a mouth, . . 2. Porpitidae. 3. Umbrella elliptical or bilateral. Blastostyles with a mouth, . . .3. Velellidas. 1 System der Siphonophoren, Jena, 1888. REPORT ON THE SIPHONOPHOR^E. 41 Family I. Discalidse, Haeckel, 1 Discalidx, Hkl., System der Siphonoplioren, p. 29. Definition. — Disconectse with an octoraclial circular permanent umbrella, including a circular, campanulate or cliscoidal pneumatocyst, without vertical crest. No vertical sail upon the umbrella. Submarginal tentacles of the umbrella simple or branched, with terminal cnidospheres. Central siphon surrounded by numerous mouthless palpons, which bear the medusiform gonophores. The family Discalidse comprises some new, small, but very interesting Siphonophoraj from the deep sea, which were found in the collections of the Challenger. They are in general very similar to young Porpitidae, but differ from them in the very important character, that the large central siphon alone possesses a mouth, while the surrounding gonostyles or blastostyles are mouthless palpons. The Discalidse are therefore " Disco- nectie monogastrica}," and become mature in the monogastric state, which is a transi- tional larval stage in the Porpitidse. These latter, as well as the Velellidse, are in the adult state " Disconectse polygastricse" each gonostyle possessing a mouth and repre- senting a peripheral sexual siphon. The Discalidse may be regarded as the simplest and most primitive of all Siphonophorse, since they retain the original character of a simple octoradial Medusa (like Trachynema or Pectyllis) more nearly than all the others. Umbrella. — The complete body of all Discalidse is circular, sometimes more lenticular or discoidal, at other times more campanulate or even subglobular. The vertical or main axis is, therefore, sometimes nearly as great as the horizontal or equatorial axis, at other times scarcely half as great, or even less. The free prominent margin, or the glanduli- ferous limbus umbrellse, marks the boundary between its upper (apical or proximal) face, the exumbrella, and its lower (basal or distal) face, the subumbrella. The former includes the pneumatophore, and represents with it the nectosome ; the latter is the siphosonw, and bears in its centre the large primary sterile siphon, around this numerous sexual palpons (or gonostyles), and towards the margin the corona of tentacles. A vertical meridional section through the umbrella (PI. XLIX. fig. 4) demonstrates that the superior half of the umbrella is occupied by the pneumatocyst (ph), the inferior by the large centradenia (uc), and from this depend in the centre the large central siphon (sa), and around it the corona of gonostyles (gs). A deep circular coronal groove separates this latter from the corona of submarginal tentacles. Exumbrella. — The superior (apical or proximal) face of the umbrella is flat or slightly convex, smooth, and pierced by the stigmata of the pneumatocyst. It is composed of two parallel lamellse, which are separated by the network of the pallia! vessels. The external or superior lamella is the pneumatocodon, which contains numerous cnido- (ZOOL. CHALL. EXP. — PART LXXVII. — 1888.) Hllhh 6 42 THE VOYAGE OF H.M.S. CHALLENGER. cysts, and (usualty) exhibits a distinct octoradial pigment star in its centre (fig. 7). The internal or inferior lamella is the pneumatosaccus, the invaginated plate of the exoderm, which secretes the chitinous pneumatocyst. Its lower face is in contact with the upper face of the ccntradenia. Limbus Umbrella? (PI. L. figs. 1, 9, us). — The free horizontally prominent margin of the umbrella, which separates the exumbrella from the subumbrella, is in all Discalidse more or less octolobate ; the eight interradial convex lobes are the more prominent, the deeper are the eight incisions in which are inserted the marginal tentacles. The whole edge of. the mantle-border is beset with a continuous series of marginal muciparous glands. Subumbrella. — The inferior (basal or distal) face of the umbrella is convex, and bears in its centre the large conical central siphon, around this a corona of gonostyles (eight in Disealia, sixteen in Disconalia), and in the peripheral part a simple or double corona of tentacles (eight simple tentacles in Disealia, PI. XLIX., eight radial bunches of numerous tentacles in Disconalia, PL L.). Pneumatocyst (PI. XLIX. figs. 4, 5, 8, 9). — The float filled with gas, which is in- cluded in the pneumatosaccus, always exhibits in the Discalidse a regular octoradial structure. This is of typical simplicity in Disealia (PI. XLIX. figs. 2-5), composed only of a subspherical central chamber and a surrounding regular ring of eight equal triangular radial chambers. The more advanced genus, Disconalia (figs. 8, 9), exhibits the same biconvex octoradial disc in its central part ; but it is here surrounded by a peripheral girdle of five to ten concentric ring-chambers ; the middle ones of these are far broader than the innermost and the outermost. Pncumoihyrie. — Each of the eight radial chambers of the central disc of the pneuma- tocyst communicates with the common central chamber by an inner opening or pneumo- thyra, placed on the proximal apex of the triangular chamber. Opposite to this lies in the centre of its distal base another pneumothyra, which opens into the first or innermost ring-chamber. An interradial series of similar septal openings, by which every two neighbouring chambers communicate, lies in the centrifugal continuation of the interradial line, which bisects each triangular chamber and connects its apical with its basal pneumo- thyra. They are, therefore, in Disconalia eight regular interradial rows of pneumothyrse (PI. XLIX. figs. 8, 9, pg), and these alternate regularly with the eight perradial grooves which separate the eight triangular chambers from one another, are continued to the peri- pheral margin of the pneumatocyst, and divide the latter into eight ecpial triangular octants. The free margin of the pneumatocyst thus becomes distinctly octolobate. Stigmata (PL XLIX. figs. 2, 5, 8, 9). — The superior (apical or proximal) face of the pneumatocyst bears the stigmata or the short tubular openings which pierce the cx- urnbrella and permit an expulsion of the enclosed air. Disealia (fig. 2) has only nine stig- mata, one central, in the central chamber, and one in each of the surrounding eight radial chambers. In Disconalia (figs. 8, 9) this number is increased hj a variable number of REPORT ON THE SIPHONOPHOR^E. 43 accessory stigmata, lying in one or in several of the concentric ring-chambers. Usually the third or fourth only of these exhibit eight stigmata, placed in the interradial lines, which pass through the former. But sometimes these accessory stigmata are irregularly scattered. The gas enclosed in the pneumatocyst may issue by these stigmata, when the strong muscle-plate of the surrounding pneumatosaccus contracts. Trachese (figs. 5, 9,pt). — The inferior (basal or distal) face of the pneumatocyst bears the aeriferous tubules which receive the gas secreted by the pneumadenia and conduct it into the chambers of the former. The simplest genus, Disculia (fig. 5), possesses only eight short trachea?, which arise from the peripheral part of the inferior side of the eight triangular radial chambers. The more highly developed Disconalia (fig. 9) exhibits, besides these latter, a greater number of peripheral trachese, arising from the lower face of the concentric ring-chambers ; they are more numerous in the innermost than in the middle ring-chambers, and are wanting in the outermost. Their number amounts to from twenty to eighty or more ; their arrangement is variable and irregular. The trachese in all Discalidse are very short and small, their cylindrical articulate tubules composed of ten to thirty small conical segments. They are more or less irregularly curved, and descend in various directions into the solid glandular parenchyma of the centradenia, where their open distal ends are surrounded by exodermal cells. In no Discalidse do the short trachese pierce the entire centradenia and the subjacent gastrobasal plate, nor do they enter into the base of the wall of the central siphon and the gonostyles, as is the case in the Porpitidse. Central Siphon. — The large central polypite, which corresponds to the manubrium or gastral tube of the original Medusa, in the Discalidse is relatively larger and more developed than in the Porpitidse and Velellidse. It is in the former the only organ for the reception of food and digestion, whilst these nutritive functions in the two latter families are executed also by the sexual peripheral siphons. The central siphon of the pyriform Discalia (PI. XLIX. figs. 1-4, sa) is very elongated, and about as long as the greatest diameter of the umbrella, whilst it is much smaller in the discoidal Disconalia (PI. L. fig. 1) ; its diameter (in length and breadth) is here only one-third or one-fourth of the latter. The basal part (or stomach) is ovate or pyriform, the distal half (or proboscis) cylindrical. The thick, very contractile wall is composed, as usual, of a stronger exodermal longitudinal layer of muscles, and a thinner entodermal layer of circular muscles, separated by an elastic structureless support. The fundus of the stomach is separated from the superjacent centradenia by the horizontal gastrobasal plate. The periphery of this sobd circular or octagonal support exhibits eight equidistant openings, the ostia, which conduct into the eight radial canals of the subumbrella. These ostia are prolonged sometimes downwards into eight longi- tudinal grooves at the inside of the stomach, and to these correspond eight longitudinal folds or ribs on its outside. 44 THE VOYAGE OF H.M.S. CHALLENGER. The mouth or the distal opening of the central siphon is either circular or octagonal, with eight short regular lips or radial mouth lobes (PI. L. figs. 1, 3). Its edge is strongly armed with cnidocysts. Centradenia (PI. XLIX. figs. 4; 10, 11). — The large central gland, which is called the " liver " in the Porpitidse and Velellidse, does not exhibit in the Discalidse the coin- plicated structure seen in those two families, but merely a typical and most instructive simple shape. It is a biconvex lenticular disc of circular or octagonal outline, in Discalia (fig. 4) relatively thick and small, in Disconalia (fig. 10) broader and flatter. Its horizontal diameter is two to four times as great as its vertical main axis, and of the same length as that of the pneumatocyst above it. Its peripheral margin is surrounded by the corona of gonostyles, whilst its inferior face is in contact with the central siphon. The entire mass of the lenticular centradenia is composed, in the Discalidae, of numerous densely aggregated exodermal cells and cnidocysts, and many of these are (in the well-preserved spirit specimens of the Challenger collection) filled with an air-bubble ; it is therefore very probable that these cells secrete the gas, which is taken up by the open distal ends of the tracheae, and conducted by these into the chambers of the pneumatocyst. These gas-producing exodermal cells are probably derived from the basal part of the pneumatosaccus, or the invaginated lamella of the exoderm which includes the pneumatocyst. The thin structureless supporting plate, which separates the upper face of the centradenia from the overlying pneumatosaccus, is pierced by numerous pores which permit a direct connection between the two. The solid exodermal parenchyma, in the Discalidee, is only traversed by the trachese, and not by the so-called " liver-canals," which form a complex network in the Porpitidse and Velellidse. These hepatic canals are here confined to an octoradial " liver-star," which lies in the superior face of the centradenia. The eight main rays of it lie in the eight perradial grooves between the eight interradial triangular air-chambers, and are united in the centre of the lower face of the central chamber. They arise from the basal part of the eight subumbrellar radial canals (near their opening into the base of the central siphon), and embrace the surface of the centradenia like eight equidistant meridional arches. They remain single in Discalia (fig. 4), whilst they are forked and branched dichotomously in Disconalia (fig. 10). Gonostyles. — The polypites (or secondary manubria) which produce by budding the medusiform gonophores are in the Discalidse mouthless palpons, and not siphons pro- vided with a mouth, as is the case in the nearly allied Porpitidaa and the more divergent Velellidse ; but also in the latter two families the gonostyles arise from the subumbrella in the same mouthless form, and accpure their mouth opening later. Their structure is the same as in the palpons of the Discalidaa. These are spindle-shaped or pyriform, much smaller than the central siphon ; they form a regular simple corona around the base of the latter. Discalia (PI. XLIX. figs. 1, 3) possesses eight, and Disconalia REPORT ON THE SIPHONOPHORiE. 45 (PL L. figs. 1, 2) sixteen gonostyles. The medusiform gonophores arise from their proximal part either isolated or in clusters (compare p. 37 above). Tentacles. —The corona of submarginal tentacles exhibits in the Discalidce very inter- esting forms, since these organs correspond in the two genera of this family to those of two important larval stages of the Porpitidae. Discalia (PI. XLIX. figs. 1, 4) possesses only eight simple radial tentacles of cylindrical form, which bear no lateral branches, but a simple terminal cnidosphere (or a spherical knob composed of radially disposed cnido- cysts). The larva of the other genus, Disconalia, also exhibits in its young state eight simple tentacles (PL L. fig. 9), and hereafter at their distal end a group of four terminal cnidospheres (fig. 10). These become multiplied in older larvae, and form a subumbrellar corona inside the limb of the umbrella. The adult Disconalia (PL L. fig. 1) possesses two rows of tentacles, in which eight perradial bunches are prominent. Each tentacle is cylindrical, with club-shaped distal end, and bears in its distal half three longitudinal rows of pediculate cnidospheres, an odd inferior and two paired lateral series (compare p. 38). Ontogeny. — The individual development of the Discalidse is not known, but is probably identical with that of the Porpitidae (compare p. 39). Phytogeny. — The phylogenetic value of the Discalidse is, in my opinion, very great. I regard Discalia as a survival of the common ancestral form of all Disconectae, and Disconalia as an intermediate form between this and Porpalia. Synopsis of the Genera of Discalidse. Umbrella with eight simple tentacles. Pneumatocyst without concentric ring-chambers, . 1. Discalia. Umbrella with numerous tentacles, arranged in eight radial bunches. Pneumatocyst with concentric ring-chambers, surrounding the octoradial central disc, . . .2. Disconalia. Genus 1. Discalia,1 Haeckel, 1888. Discalia, Hkl., System der Siphonophoren, p. 29. Definition. — Discalidse, with a lenticular or subglobular umbrella, including a lenti- cular pneumatocyst, which is composed of a central chamber and eight surrounding radial chambers, without concentric ring-chambers. Marginal tentacles eight, simple, with a terminal cnidal knob. The genus Discalia is the simplest and most primitive form of all Disconectae, and may be regarded as one of the common ancestral forms of this order. It may be com- pared to an octonemal Trachomedusa (Trachynema, Mdrmanema, Rhopalonema), which has developed an octoradial pneumatophore in its exumbrella, and in which the eight simple genital sacs of the subumbrella have been replaced by eight palpons or mouthless gonostyles, which afterwards produce medusiform gonophores by budding. 1 Discalia = marine disc, 01'sx.o;, £>.ios. 46 THE VOYAGE OF H.M.S. CHALLENGES. All Porpitidae probably pass during their metamorphosis through a larval stage, which is essentially identical with Discalia, and differs from it only in the absence of sexual palpons or gonostyles, bearing gonophores. The young larva of Porpita linnseana, which Alexander Agassiz has figured,1 exhibits a larval stage which is a little more advanced, and forms the transition to the Disconalia-stage. The phylogenetic importance of Discalia, therefore, is very great. On the one hand, it seems to indicate clearly the origin of the Disconectse from the Trachomedusse (Trachy- nemidae) ; on the other hand, all the other Disconectse may be derived from it as from a common ancestral genus. Two species of Discalia, both inhabitants of the deep sea, were found by me in the Challenger collection. The first species, Discalia medusina, was taken in the centre of the Southern Pacific, at Station 288 (depth, 2600 fathoms), and is figured in PI. XLIX. figs. 1-6. The second species, Discalia primordialis, was captured in the tropical Pacific, at Station 274 (depth, 2750 fathoms). The latter is distinguished from the former by a more flattened umbrella, and by eight longer (perradial) tentacles, between which eight shorter (interradial) were interpolated ; but as its state of preservation was not sufficient, I give here only the description of the first well-preserved species. Discalia medusina, n. sp. (PL XLIX. figs. 1-6). Habitat— Southern Pacific, Station 288, October 21, 1875; lat. 40° 3' S., long. 132° 58' W.; depth, 2600 fathoms. Umbrella (fig. 1, view from below ; fig. 2, from above ; fig. 3, in profile ; fig. 4, in meridional section). — The umbrella, which represents the nectosome, is subspherical, 0"2 to 0"4 mm. in diameter. A deep annular constriction separates the flatter exumbrella from the inflated and highly vaulted subumbrella. The latter bears in its equatorial zone a corona of eight simple tentacles, and beyond it a corona of eight gonostyles, which surround the central siphon. Exumbrella (fig. 2). — The superior or apical face of the umbrella is rather flat, and exhibits an octoradial star of brown pigment, indicating the course of the eight per- radial exumbrellar canals, or the superior vessels of the mantle. The apical stigma, or the superior opening of the central air-chamber, is situated in the centre of the pigment- star, whdst eight ether stigmata, the outer openings of the eight interradial air- chambers, are placed between the eight rays of the dark pigment-star. Limbus Umbrella? (uu). — The circular margin of the umbrella is thickened and divided by eight perradial incisions into eight prominent flat interradial lobes. The whole margin is beset with a corona of the usual marginal glands (fig. 6) (compare p. 42 above). 1 57, pi. ix. figs. 1, 2. REPORT ON THE SIPHONOPHOR^E. 47 Subumbrella (fig. 1). — The inferior or basal face of the umbrella is nearly hemi- spherical. Its centre is occupied by the central siphon (sa), and this is surrounded by a corona of eight gonostyles (gs). The equatorial zone of the subumbrella is beset with eight perradial tentacles, and between these proceed outside the eight interradial lobes of the limbus. Pneumatocyst (fig. 2, pf, from above ; figs. 4, 5 in meridional section). — The central air-chamber (pli) is spheroidal, large, and occupies about half the volume of the float. A prominent tubular stigma (figs. 4, 5, po) opens in the centre of its upper face. The eight radial air-chambers, which surround the former as a regular corona (pq), are crescentric, concave on the axial side, convex on the abaxial side ; each opens by a stigma on the upper face (pe). The pneumatocyst possesses only eight simple tracheae {pi), one arising from the inferior face of each radial chamber ; the eight tracheae descend obliquely and contorted downwards, enter into the centradenia (fig. 4, uc), and terminate in it : but they do not pass into the exodermal wall of the central siphon. Centradenia (fig. 4, uc). — The large central gland is a circular disc, the diameter of which is about three to four times as great as its height, and equals that of the pneumatocyst. The inferior face of the latter is in close contact with the superior face of the former ; whilst the inferior face of the centfiadenia is separated by a thick support, the gastrobasal plate, from the base of the central siphon. The whole mass of the solid centradenia is composed of exoderm cells. Its entodermal system of hepatic vessels is confined to an eight-rayed star on its superior face. The eight radial canals, which are connected in the centre of the latter, arise from the proximal third of the eight sub- umbrellar radial canals which run from the base of the central siphon towards the margin of the umbrella. This simplest form of the gastro-hepatic canal-system gives the explana- tion of its more complex form in the other Disconectse. Central Siphon (figs. 1-4, sa). — The central polypite is a thick-walled contractile tube, inversely conical in the proximal half, cylindrical in the distal half. Its length about equals the equatorial diameter of the umbrella, and is twice as great as the diameter of its base. The thick muscular wall is composed of a strong outer layer of exodermal longitudinal fibres, and a thin inner layer of entodermal circular fibres, both separated by a structureless elastic fulcrum. The distal mouth is octolobate. The gastrobasis, or the horizontal roof of the stomach, is regularly octagonal, separated by a thick fulcrum, the gastrobasal plate, from the centradenia above it. It is pierced in the periphery by the gastral openings of the eight radial main vessels. These give off branches for the gonostyles and tentacles, and run centrifugally in the subumbrella towards the limbus, where they are united by a marginal ring-canal. Gonostyles (figs. 1-4, gs). — The eight sexual palpons are about half as long and broad as the central siphon. They bear clusters of gonophores in their cybndrical basal part, patches of cnidocysts in their spindle-shaped distal part. 48 THE VOYAGE OF H.M.S. CHALLENGER. Tentacles (figs. 1-4, t). — The eight tentacles are cylindrical, somewhat larger than the diameter of the umbrella, and bear a single cnidosphere at the distal end. Genus 2. Disconalia,1 Haeckel, 1888. Disconalia, Hkl., System der Siplionoplioren, p. 30. Definition. — Discalidse, with a lenticular or discoidal umbrella., including a discoidal pneumatocyst, which is composed of a central chamber, eight surrounding radial chambers, and several concentric ring-chambers. Marginal tentacles numerous, arranged in eight radial bunches. The genus Disconalia is closely allied to the preceding ancestral genus Discalia; but it differs from it in the multiplication of the marginal tentacles, which form eight marginal bunches ; and mainly in the peripheral increase of the octoradial pneumatocyst, which is surrounded by several concentric ring-chambers. It corresponds therefore to that larval stage of the Porpitidse which Alexander Agassiz ~ has figured of Porpita linnsena. The young larvae, probably of all species of Porpitidse, after having passed the Discalia- forrn, assume a transitional Disconalia-form. Perhaps even Eschscholtz observed a Disconecta belonging to this genus. The interesting small form, taken in the tropical Pacific, which he figures under the name Porpita ramifera? is either a true Disconalia, or the corresponding larva of some Porpitid passing through this typical stage. The answering of this question is not possible, since the organs of the subumbrella, which would be decisive, are neither figured nor even mentioned in the description given by Eschscholtz. Two species of Disconalia (both deep-sea inhabitants) were found by me in the Challenger collection, one from the Southern Pacific (Station 181), the other from the Indian Ocean, south of Australia (Station 157). The latter (Disconalia pectyllis) had much longer and less ramified tentacles, and a larger pneumatocyst, than the former (Disconalia gastroblasta) ; but its state of preservation was not sufficient for a full description. Regarded from a phylogenetic point of view, Disconalia is of the highest importance as a necessary connecting link between Discalia and Poipalia. It is derived from the ancestral genus Discalia by the multiplication of the air-chambers and the tentacles ; if its gonostyles acquired a mouth, it would pass into Porpalia, Disconalia gastroblasta, n. sp. (PI. XLIX. figs. 7-12; PI. L. figs. 1-10). Habitat. — Southern Pacific, north-east of Australia, Station 181, August 25, 1874 ; lat. 13c 50' S., long. 151° 49' E. ; depth, 2440 fathoms. 1 Disconalia, derivative from Discalia. - 57, pi, ix. lig. 3. 3 1, p. 17, Tat'. 16, figs. 3a, 36. REPORT ON THE SIPHONOPHOR-ffi. 49 Umbrella (PL L. fig. 1, from below). — The umbrella is slightly vaulted, lenticular, nearly cliscoidal, 4 to 6 mm. in diameter ; its vertical axis is about 1 mm. The margin is regularly octolobate, with eight deep margiual incisions, from which arise the eight bunches of tentacles. Exumbrella (PI. XLIX. fig. 7). — The superior face of the umbrella is rather flat, a little vaulted, and exhibits a regular network of brown pigmented canals, the exumbrellar vessels. Eight regular radial canals arise from a central circular canal, in the centre of which is placed the apical stigma of the pneumatocyst. The eight interradial chambers of the latter are separated by these perradial mantle canals, which anastomose outside and form a ring around each chamber. From this ring arise in each octant four to six centrifugal canals, which are forked and prolonged outside the pneumatosaccus into the exumbrellar canals of the limbus. Their dichotomous branches are connected by anastomoses. A few peripheral stigmata are scattered in the outer half of the exumbrella. Limbus Umbrellas (fig. 7, um). — The free margin of the umbrella, outside the corona of tentacles, is distinctly octolobate, and bears a series of pyriform marginal glands on its free edge. The eight interradial convex lobes are more or less prominent between the eight perradial bunches of tentacles. Subumbrella (PL XLIX. fig. 11). — The inferior face of the discoidal umbrella is more convex than the superior, and is divided into four different zones. The central zone is occupied by the large central siphon (sa). This is surrounded by a corona of sixteen gonostyles, and this by an octagonal tentacular zone (fig. 12, tu). Outside the latter is prominent the broad octolobate limb of the subumbrella. Pneumatocyst (fig. 8, half of the inferior face; fig. 9, half of the superior face). — The subspherical central chamber (ph) possesses an apical stigma in the centre of its upper face, and is surrounded by a corona of eight equal triangular radial chambers, each of which bears a circular stigma. This octoradial central disc is surrounded by a peripheral girdle composed of eight or nine concentric ring-chambers ; the second and third of these (pk) are the broadest, much broader than the peripheral chambers. Eight deep perradial furrows of the lower face separate the eight chambers of the octoradial ring one from another, and are prolonged up to the circular margin of the pneumatocyst, dividing it into eight slightly prominent lobes. In the middle between each two furrows (therefore in eight interradial rows) may be seen the pneumathyrae, or openings by which the concentric ring-chambers communicate one with another (fig. 8, pg). The superior or apical face of the discoidal pneumatocyst (fig. 9) is slightly convex, and bears about twenty stigmata, viz. , one central, eight interradial (in the middle of the eight triangular chambers), and eight to twelve peripheral, irregularly placed in the second to the fourth ring-chamber. The inferior or basal face of the pneumatocyst (fig. 8) is slightly concave, and bears (ZOOL. CHALL EXP. — PART LSXVII. — 1888.) Hbhh 7 50 THE VOYAGE OF H.M.S. CHALLENGER. between fifty and ninety short trachea) (six to twelve in each octant). These are placed in the periphery of the octoradial ring, and in the two to four next following chambers ; they are wanting in the periphery. Centradenia (figs. 10-12, uc). — The large central gland, or so-called "liver," is an octagonal, lenticular, biconvex disc, of the same horizontal diameter as the overlaying pneumatocyst. Its thickness is inconsiderable, and decreases towards the periphery. Its convex superior face (fig. 10) is in close contact with the concave inferior face of the latter. The peripheral octagonal margin of the centradenia does not exceed that of the covering pneumatocyst. The inferior face of the centradenia is rather flat, and is separated by a thin fulcrum from the subumbrella. Its central part is in contact with the gastrobasal plate of the central siphon ; its peripheral part with the corona of gonostyles, and its margin with the corona of tentacles. The structure of the centradenia exhibits in Disconalia the same remarkable simplicity and typical form as in Discalia. The solid parenchyma is entirely composed of masses of aggregated exoderm-cells, with innumerable cnidocysts. A regular octoradial star of pigmented hepatic vessels is placed on its superior face (fig. 10). From the centre of the latter arise eight perradial brown " fiver-canals," which branch dichotomously, and run on its upper face towards its peripheral margin. They pass here over into the peripheral canal-network of the subumbrella, from which arise the canals of the tentacles and gonostyles. Central Siphon (PI. L. figs. 1, 3). — The large central polypite has the usual form of an inverted cone. The broad octagonal base occupies the central area of the sub- umbrella, and is separated from the centradenia above it by the gastrobasal plate. The periphery of the base opens by eight perradial ostia (PL XLIX. fig. 12) into the eight radial main vessels which run in the subumbrella towards the limb, and give off the ascending centripetal canals of the " liver-star " (fig. 10). The muscular wall of the central siphon is very thick. It opens by a distal octolobate mouth, the edges of which are strongly armed with cnidocysts (PL L. fig. 3). Gonostyles (PL L. figs. 1, 2). — Sixteen sexual palpons, or mouthless gonostyles, form a corona around the base of the sterile central siphon. They are small spindle-shaped sacs, scarcely half as long as the central siphon. Their wider basal half bears clusters of gonophores (fig. 2, g), whilst the slender distal half is beset with cnidonodes (k) ; the closed distal end is obtusely conical and entirely covered by cnidocysts. Tentacles (PL L. figs. 1, 4-G). — The tentacles are very numerous and densely crowded in the subumbral groove between the corona of gonostyles and the limb of the umbrella. They alternate here in a double row (PL XLIX. figs. 1, 2). The tentacles which arise from the eight perradial corners of the octagonal girdle are much longer than the interjacent interradial, and form therefore eight prominent bunches. REPORT ON THE SIPHONOPHOR^. 51 The structure of the cylindrical tentacles is the same as in all Porpitidse. They bear three rows of pecliculate cniclospheres in their club-shaped distal half, one inferior and two opposite lateral rows (figs. 4-6). Family II. Poepitid^, Brandt, 1835. Pm-pitidx, Brandt, Prodromus, 25, p. 40. Definition. — Disconectse with a circular permanent umbrella, including a campanu- late or discoidal pneumatocyst, which is composed of an octoradial centre and numerous concentric rings, without vertical crest. No vertical sail upon the umbrella. Submarginal tentacles with three rows of pediculate cniclospheres. Central siphon surrounded by numerous peripheral fertile siphons, which bear the gonophores. The family Porpitida?, founded in 1835 by Brandt for the genus Porpita, comprises all those Disconectse polygastricse which have a circular umbrella and a regular originally octoradial ground-form. They agree in their regular octoradiate form with the monogastric Discalidse, their ancestral group, but differ from them essentially in the polygastric structure which they share with the Velellidse. Not only does the primary sterile central siphon possess a mouth opening at its distal end, but likewise also each of the surrounding gonostyles. These peripheral polypites, which bear the gonophores budding from their proximal part, are therefore not mouthless palpons (as in the Discalidse), but mouth-bearing, feeding, and digesting secondary siphons. The Velel- lidse, with the same polygastric organisation, differ from the Porpitidse in the amphi- thect or bilaterally-radial type, and in the development of a vertical sail upon the exumbrella. Eschscholtz, in his fundamental work,1 described five different species of Porpitidse, which he united in the single genus Porpita. Lesson 2 added to this two other genera (each with a single species), Ratis and Acies. His description, however, is very incom- plete, and not illustrated by any figure, so that it is impossible to decide whether they are merely young forms of Porpita (as most authors suggest) or perhaps Discalidse. Some interesting new Porpitidse, found in the Challenger collection, and some other new forms observed by myself on different occasions, have led me to divide this family into four genera (characterised in my System).3 These may again be disposed in two subfamilies, the Porpalidse and the Porpitellidse, each with two genera. The first subfamily, Porpalidse, has a lenticular or subglobular umbrella and a campanulate pneumatophore with lobate margin {Porpalia and Porpema) ; whilst the second sub- family, Porpitellidse, possesses a flat discoidal umbrella and a medal-shaped pneumato- phore with a circular, not lobate margin {Porpitella and Porpita). Umbrella. — The complete body of all Porpitidse is circular, sometimes more lenticular 1 1, 1829, p. 176. 2 3, 1813, p. 592. 3 95, 1888, p. 30. 52 THE VOYAGE OF H.M.S. CHALLENGER. or subglobular (in the Porpalidae), sometimes more flatly expanded or discoidal (in the Porpitellidae). Its vertical main axis is in the first case little shorter or nearly as long as the horizontal or equatorial axis ; whereas in the second case it is much shorter, only one-half or one-third as long, or even less. The free promine ntmargin, or the glandiferous limbus umbrellas, marks the boundary between the exumbrella (upper or pneumatophorous half) and the subumbrella (lower or siphonophorous half). Exumbrella. — The upper, apical or proximal, part of the discoidal trunk, which corresponds to the exumbrella of the Medusae, and physiologically represents the nectosome, is in all Porpitidae circular and composed of the pneumatocyst, filled with gas, and the surrounding pneumatophore. This latter is composed of two parallel membranes, separated by a network of anastomosing radial canals — the outer thicker membrane armed with thread-cells and pierced by the stigmata, is the pneumatocodon ; the inner thinner membrane, immediately including the pneumatocyst, is the pneumato- saccus or the invaginated exoderm. Both membranes are connected by numerous branched radial septa, and the cavities between these, radiating from the centre of the disc, are the radial canals of the exumbrella (pallial vessels) ; they open at the margin of the disc into the large circular canal of the limb. The surface of the exumbrella is in many Porpitidae smooth, in others more or less papillate, owing to conical protuberances of the pneumatocyst. Often an elegant? multi- radiate pigment-star is visible, produced by a dark pigment in the wall of the pallial vessels. The central part of the exumbrella in all Porpitidae is pierced by a central stigma, and a surrounding corona of eight stigmata placed in the walls of the eight surrounding radial chambers. The other stigmata of the exumbrella, in larger species several hundreds, in smaller only few, are sometimes disposed in regular radial rows, at other times more irregularly scattered. Pneumatocyst. — The chitinous polythalamous float filled with gas, which is called pneumatocyst (formerly "inner shell"), is in the Porpitidae always regularly circular, corresponding to the surrounding pneumatosaccus (or the invaginated exumbrella) from which it is secreted. Consequently its general form in the subfamily Porpalidae is highly vaulted, campanulate or cap-shaped [Porpalia, Porpema, Pis. XL VTL, XL VIII. ), whereas in the flatly expanded Porpitellidae it is discoidal, even or slightly vaulted {Porpitella and Porpita, Pis. XLV. and XLVL). The pneumatocyst is always composed of two little distant and nearly parallel lamellae of structureless chitin, which are connected by numerous concentric annular septa. The latter divide the float into numerous concentric ring-chambers, and these open on the upper or proximal face by stigmata, on the lower or distal face by tracheae. The central disc of the pneumatocyst has in the Porpitidae the same structure as in the Discalidae ; it is composed of a spherical or subglobular central chamber (with a central stigma above) and of eight equal triangular radial chambers, each of which bears REPORT ON THE SIPHONOPHOR^E. 53 a stigma above, and a trachea (or a bunch of trachese) below. This evident octo- radial structure of the central part is often lost in the peripheral part of the pneuma- tocyst. The concentric ring-chambers which compose the latter are simple circular rings, without radial septa. Their number is very variable, in the smaller species ten to twelve, or even less, in the larger species forty to sixty, and in the largest more than a hundred. They are usually of nearly equal breadth ; often, however, the chambers of the middle part are wider than the more proximal (near the centre) and the more distal chambers (near the margin). The first or innermost of the concentric ring-chambers is different from all others ; it embraces the octoradial central disc not only from the outer (abaxial) but also from the lower (basal) side (PI. XLVL figs. 3, 6). Pneumothyrse. — It is a general opinion that the concentric circular ring-chambers of the Porpitidse are perfectly separated one from another by solid annular septa. But this is not the case. I found in all members of this family openings of communication between them, which I shortly call pneumothyrse. The thin chitinous annular septa, which separate the ring-chambers, are concave on the axial side, convex on the abaxial side, they are usually thicker in the upper, thinner in the lower, part. Each annular septum is pierced in its basal part by at least eight pneumothyrse, ovate or roundish openings, which are about twice as broad as the stigmata of the upper surface. The pneumothyrse or ring-gates (PI. XLVL figs. 3, 4, pg) be originally in the same radii as the stigmata (pe). There are, therefore, eight interradial rows of pneumothyrse. But in the larger species their number is increased, accessory pneumothyrse being interpolated between the primary ones in the outer chambers. The superior, proximal or apical side of the pneumatocyst is usually flat in the central part, whilst the peripheral part is highly convex or campanulate in the Porpalidse, slightly convex or even in the Porpitellidse. Its surface is sometimes smooth, at other times rough and marked by radial stripes or ribs, aud by concentric circles. It is sometimes, mainly in the central part, spiny, papillate, or armed with conical, irregularly scattered tubercles. The chitinous substance in the central part is often much thickened by apposition of secondary layers, and these may close the stigmata of that part. Those of the peripheral part remain always open. The stigmata, or the pneumatic foramina on the upper face of the pneumatocyst (pe), are much more numerous in the Porpitidse than in the Discalidse and Velellidse. Constantly there is a central stigma in the apex of the central chamber, and around this a regular corona of eight equidistant stigmata in the eight radial chambers which surround the former. The other stigmata are usually not regularly disposed, but scattered in great numbers over the upper surface. When the latter is provided with radial ribs or prominent ridges, the stigmata are placed in the height of the ridges, not in the valleys between them. The stigmata are sometimes simple openings in the upper wall of the 54 THE VOYAGE OF H.M.S. CHALLENGER. ring-chambers. But usually they are dilated in the middle, more or less urn-shaped, and sometimes prolonged into short prominent tubules (PL XL VIII. fig. 8, pe). The inferior, distal or basal side of the pneumatocyst, separated by the exodermal epithelium of the pneumatosaccus from the neighbouring superior face of the centra- denia, corresponds in its general form to the latter, and has therefore a more complicated shape than its superior face. It is always more or less concave, hemispherical or campanulate in the Porpalidae, slightly concave or cap-shaped in the Porpitelkdae. Numerous radial ridges or folds are prominent from the inferior face of the float, and often these arise in the form of vertical radial lamella?, separated by deep valleys. These latter are filled up by corresponding high radial ridges of the upper face of the centra- denia. The number of the radial ridges increases rapidly towards the periphery, numerous secondary and tertiary centripetal lamellae (which do not reach the centre) being interpolated between the centrifugal primary ones. The concave under surface of the pneumatocyst, and the convex upper surface of the centradenia, catching one into another, become very similar to a Fungia (PI. XLVIII. fig. 7). From the combs of the prominent radial ridges arise the tracheae (fig. 6, pt). Tracheae. — The pneumatic tubules or aeriferous filaments, which we call shortly trachea?, are much more numerous in the Porpitida? than in the Discalida? and Velellidae. Each radial ridge bears in the larger species on an average more than one hundred tracheae, and as the number of the ridges there amounts to some hundreds, the number of the tracheae reaches many thousands. The innermost tracheae, nearest the main axis, arise from the eight radial chambers which surround the central chamber. The other tracheae arise sometimes isolated, in irregular radial series, from the crest of the ridges, at other times in bunches, crowded in small groups of three to six or nine, rarely more. The aeriferous tubules are usually simple, very rarely branch, and never anastomose. Their thin chitinous wall is cylindrical, of very different length. Their course is never straight, always more or less curved, and often serpentine. The trachea? of all Porpitidae are distinctly articulate, and seem to be composed of a series of short truncate conical segments ; the distal or inferior end of each segment is wider (PI. XLVI. fig. 8). Sometimes each segment seems to be separated from the other by a transverse septum ; but the apparent septum is merely an annular constriction. The two ends of each ring- segment (proximal and distal) remain always open. The course of the tracheal is difficult to make out. Descending from the lower face of the pneumatocyst, they enter immediately into the centradenia, and run in sinuous curves between the canal-network of this organ. A great part finishes inside the central gland, whilst another part descends deeper and passes into the wall of the central siphon and the peripheral siphons. They end here in the exoderm of the proximal half of the siphons, and do not enter into their distal half. Usually the great majority of the trachea? are much shorter than the vertical diameter of the centradenia ; they must there- REPORT ON THE SIPHONOPHOPwE. 55 fore finish in the solid glandular parenchyma of this latter, and cannot pass through it and enter into the wall of the siphons. A small part only of the tracheae enters into this latter. The distal ends of all tracheae are open, surrounded by exodermal epithelia, as in the other Disconectae. No tracheae are found in the marginal tentacles, nor in the peripheral part of the umbrella outside the centradenia. The margin of the pneumatoeyst is circular and not lobate in the flat discoidal Porpi- tellidae, whilst it is divided into numerous radial lobes in the campanulate Porpalidae ; the lobes are here sometimes small, sometimes widely prominent. Their number is originally eight or sixteen, and by furcation thirty-two or sixty-four (PI. XLVIII. figs. 4, 5). Central Siphon. — The large central polypite of the Porpitidos is not different from that of the Discalidae, a thick-walled and very contractile tube of very variable form. Usually it is inversely conical, its diameter decreasing gradually from the broad proximal base towards the distal mouth. Sometimes the upper half, or the stomach, is ovate and much wider than the lower half, or the cylindrical proboscis. Its transverse section is either circular or octagonal, in consequence of eight prominent radial folds. In some larger species the outer wall exhibits sixteen longitudinal folds instead of eight, and sometimes eight larger (perradial) and eight smaller (interradial) ribs alternate. To these correspond the same number of internal furrows at the inside of the siphon. These furrows lead into the basal openings of the stomach, in which the primary radial canals open (eight or sixteen). The basal ostia form a regular corona ; in some larger species their number is increased, numerous secondary and tertiary ostia being intercalated between the eight primary ones. The base of the stomach is separated completely from the overlaying centradenia by the structureless sobd gastrobasal plate. Centradenia. — The large central gland, or the so-called " central organ" (formerly "fiver"), exhibits the peculiar composition described above (p. 31). In the Porpitidae it is much more voluminous than in the Discalidae and Velellidas, and occupies the whole space between the inferior face of the pneumatophore and the superior face of the sub- umbrella which bears the siphons. The central gland is largest in some lenticular or subglobular Porpalidae, where its weight and volume are greater than those of all other parts of the body together ; it is relatively smaller in the flat discoidal Porpitellidae. The dense network of canals in the central gland is in its upper brown or blackish half composed of hepatic vessels (with biliary epithelium), in its lower green or whitish half of renal vessels (with epithelium secreting guanin crystals). The compact exodermal epithe- lium filling up the intervals of the canal-network contains masses of cnidoblasts and probably secretes the gas, which enters into the open distal ends of the tracheae. Gonostyles. — The numerous polypites of the subumbrella, which produce by budding the medusiform gonophores, in the Porpitidae are not mouthless palpons as in the Dis- calidae, but mouth-bearing siphons as in the Velellidae. They are, therefore, usually called " smaller polyps, sexual polypites, or peripheral siphons" (shortly " perisiphons"). They 56 THE VOYAGE OF H.M.S. CHALLENGER. either form a simple ring of eight or sixteen gonostyles around the central siphon, or they occupy a broader gonostylar zone, often the whole suburnbrella between the central siphon and the submarginal corona of tentacles. At first there is a corona of eight gonostyles only, regularly placed around the central siphon, and this state is permanent in some species of Porpalia (PI. XLVIIL). In Porpitella (PL XLVI.) there is a girdle of sixteen gonostyles. But usually their number is soon increased, and they form several concentric circles, more or less regular. Finally, in the largest Porpitidse their number amounts to several hundreds, and they are densely crowded. Their form and structure have been described above (p. 36). Tentacles. — The corona of submarginal tentacles in all Porpitidse is originally regularly octoradial. In all young larvae of this family there occurs a stage in which the umbrella-margin bears only eight tentacles regularly disposed at equal intervals (PI. L. figs. 9, 10), as permanently in Discalia (PL XLIX. figs. 1-4). But their number is soon increased either by interpolation of eight secondary interradial tentacles (in the middle between the former), or by budding of several secondary tentacles at the base of each primary one, so that eight regular bunches arise {Porpalia, PL XLVIIL). Sometimes sixteen bunches are formed {Porpitella, PL XLVI.). Usually the number is so increased that hundreds or thousands of tentacles, densely crowded, cover the whole zone beyond the margin of the disc. Often three to nine or more concentric circles may be discerned, and then the uppermost are the smallest, the lowermost the longest. Sometimes their bases are so densely crowded in the tentacular area, that after removing the tentacles there appears an elegant reticulated girdle beyond the margin ; each rhomboidal dimple of the reticulum is the place of the basal insertion of a lost tentacle (PL XL VII. fig. 3, tu; PL XLVIIL fig. 2, tu). The, general structure of the tentacles is the same in the Porpitidse as in the other Dis- conectse ; but they differ from those of the Velellidse in their peculiar constant form and structure. The youngest larvae of the Porpitidse bear eight simple radial tentacles with a terminal cnidosphere (PL L. fig. 9) like those of Discalia. The next larval stage exhibits four pediculate cnidospheres at the distal end of each tentacle, one of which is the primary terminal knob, the three others basal branches of it (fig. 10). The number of the latter is soon multiplied, and the older and longer tentacles, which are club-shaped and thiekened at the rounded distal end, bear always three longitudinal rows of pedunculated cnidospheres ; one odd inferior series in the middle line of the lower or distal face, and two paired (lateral) series on the two lateral faces ; the upper or proximal face is always smooth.1 Each cnidosphere is a thin lateral branch of the tentacle, having the same structure as the latter, and bearing a terminal sphere composed of radially crowded cnidoblasts. The peduncles of the latter are sometimes shorter, at other times longer. Their length often increases towards the distal end of the tentacle. Their number is very variable in different 1 57, pis. ix., x. REPORT ON THE SIPHON OPHOR^E. 57 species ; usually each of the three longitudinal rows bears six to nine branches, often twelve to twenty or more (PI. L. figs. 1-6). Ontogeny. — The individual development of the Porpitidae has hitherto been unknown. I conclude, however, from the comparative morphology of the new Porpitidae and Disca- lidee here described, that all members of these two families pass through a larval stage very similar to Discalia, This is subject to a shorter or longer metamorphosis, and passes through a stage similar to Disconalia. The transition from this to Porpalia, the simplest form of Porpitidae, is easy to conceive. The gonostyles acquire a mouth, and so the sexual palpons of the former are replaced by the sexual siphons of the latter. The medusiform gonophores, which are produced from these gonostyles by budding, have the same form and structure as the well-known Discomitra (afterwards CJirysomitra) larvae of the Velellida?.1 They become sexually mature in the free medusoid state, after having been detached from the gonophores. The larvae which arise from the fertilised egg have not yet been observed. Phytogeny. — The comparative morphology of the Porpitidae and Discalidae admits the phylogenetic hypothesis that the former have arisen from the latter. When Disconalia acquires a terminal mouth on the distal end of each blastostyle, it passes over into Porpalia. Synopsis of the Genera of Porpitidae. I. Subfamily Porpalid.e. Umbrella highly vaulted. Pneumato- - cyst campanulate, with a radially lobate margin. Tentacles arranged in eight radial bunches, the eight primary more promi- nent, ...... 3. Porpalia. Tentacles very numerous, in a circular corona, the eight primary not promi- nent, 4. Porpema. II. Subfamily Porpitellid^:. Umbrella flat, slightly vaulted. Pneu- matocyst discoidal, without pro- minent radial marginal lobes. Tentacles arranged in sixteen radial bunches, the eight primary and the eight secondary prominent, . . 5. Porpitella. Tentacles very numerous, in a circular corona, the eight primary not promi- nent, ...... 6. Porpita. Subfamily 1. Porpalid.e, Haeckel. Genus 3. Porpalia,3 Haeckel, 1888. Porpalia, Hid, System der Siphonophoren, p. 30. Definition. — Porpitidae with a lenticular or subglobular strongly vaulted umbrella, in- cluding a campanulate pneumatocyst with radial marginal lobes. Tentacles numerous, arranged into eight or sixteen prominent radial bunches. -"6^~ """' ^6' 1 57, pi. x. 2 Porpalia = Marine ring of a buckle, wo'jxu, &Km. (ZOOL. CHALL. EXP. — PART LXXVlL — 1888.) Hhhh 8 58 THE VOYAGE OF H.M.S. CHALLENGER. The genus Porpalia and the following Porpema together make up the subfamily Porpalidse, characterised by a strongly vaulted, lenticular or nearly spherical, umbrella, which includes a mitriform or campanulate pneumatocyst ; the distal margin of the latter being divided into radial lobes. The chitinous substance of this lobate pneumatocyst in the Porpalidse is much thinner, more delicate, and softer than in the Porpitellidse (Por- pitella and Porpita), sometimes it resembles crumpled tissue-paper. The numerous mar- ginal tentacles in Porpalia are arranged in eight regularly disposed radial bunches, whilst in Porpema they are equally distributed along the margin of the umbrella. The genus Porpalia is founded upon a new species, Porpalia prunella (PI. XLVIIL), well-preserved specimens of which were found in the Challenger collection, taken in the tropical Pacific (Station 222). The incomplete description of another species of this genus, founded upon a single specimen, was given in 1829 by Eschseholtz, under the name Porpita globosa.1 He took this specimen in the tropical Atlantic, near the Cape Verde Islands. The exumbrella is in this Atlantic Porpalia globosa much smaller, but the tentacles larger than in our species from the Pacific. The phylogenetic position of Porpalia is clearly indicated by its morphological relation to Disconalia on one hand, and to the Porpula larvae of the other Porpitidte on the other. It may be regarded as the common ancestral form of this family derived from Disconalia by the formation of a mouth on the distal ends of the gonostyles. Porpalia prunella, n. sp. (PI. XLVIIL). Habitat. — Tropical Pacific, north of New Guinea, Station 222 ; March 6, 1875; lat. 12° 15' N, long. 146° 16' E. Surface. Umbrella (fig. 1, as seen from above ; fig. 2, in profile ; fig. 3, in meridional section). — The umbrella is biconvex lenticular ; its diameter amounts to 4 or 5 mm. (without ten- tacles and siphon). A deep submarginal ring-furrow separates the flat exumbrella from the biconvex body, the superior face of which is more strongly vaulted than the inferior. The equatorial diameter of the biconvex lens is twice as great as its vertical main axis. Exumbrella (figs. 1-3, uc). — The superior or apical face of the umbrella is rather flat, with a shallow ring-furrow separating the central disc from the elevated peripheral margin. The central disc exhibits an elegant pigment-star with eight dark brown rays. Numerous stigmata are disposed in irregular rows between them. Limbus Umbrellte (figs. 1-3, um). — The free prominent border or limb of the umbrella is as broad as the radius of the central disc of the umbrella, and therefore half as broad as the equatorial radius of the lens. Its upper face is concave, the lower convex. The thickened margin is reflected upwards, and contains a single series of the usual muci- parous glands (compare above). 1 1, p. 178, Taf. 16, fig. 4. REPORT ON THE SIPHONOPHORiE. 59 Subumbrella (figs. 1-3, iv). — The inferior or basal face of the umbrella exhibits beyond the deep submarginal ring-furrow a broad naked girdle, in which the radial canals ascend from the tentacular zone to the margin of the umbrella. The tentacular zone begins in the equator of the biconvex lens, and occupies the peripheral half of its lower convexity. Its breadth equals the radius of the central disc of the exumbrella. Pneumatocyst (fig. 4, from above ; fig. 5, from below; fig. 3,p, in vertical meridional section). — The float, filled with air, is campanulate, and envelops the subspherical centra- denia, with exception of its basal centre, which is occupied by the central siphon. The equatorial diameter of the pneumatophorous bell is twice as great as its vertical main axis. The basal opening of the bell, closed by the base of the central siphon, is scarcely greater than half its height. The concave inferior face (or the cavity of the bell) is in close contact with the centradenia, and the thirty-two prominent radial crests of the former fit into thirty-two deep radial furrows of the latter. The convex superior face is flatly conical above the equator (itg) ; its peripheral girdle, beyond the equator, is regularly divided into sixteen furcate radial lobes (or thirty-two smaller lobes). The central chamber of the pneumatocyst is surrounded by a regular girdle of eight radial chambers, each of which possesses a stigma on its upper, and a trachea on its lower side. This octoradial girdle is surrounded by eight to ten complete, concentric, circular ring-chambers, the outermost of which touches the equator of the lens. Outside follows the peripheral girdle of thirty-two lobes, which is recurved inwards and down- wards (horizontally expanded by pressure, see fig. 4). The superior or convex face of the pneumatocyst bears in its superior central part (which is in contact with the exum- brella) numerous radial rows of stigmata. From its inferior or concave face, which is in contact with the centradenia, hang down thirty-two prominent radial crests, or rather lamellar pouches (fig. 6) ; and from the lowermost part of the latter arise numerous long tracheae which enter into the glandular tissue of the liver (fig. 3, uc) ; the majority of them terminate here with open distal ends, whilst a small part of the tracheae enters into the basal part of the wall of the siphons, and ends in its exodermal epithelium. The chitinous substance of the thin wall of the campanulate pneumatocyst is very delicate and richly folded, like crumpled tissue-paper. Centradenia (fig. 3, uc). — The large central gland is nearly spherical, and entirely fills up the subumbrellar cavity of the pneumatocyst. The circular polar area only of its lower pole is in contact with the base of the central siphon. Besides this smooth inferior area, the entire surface of the centradenia is traversed by numerous deep meridional grooves, which are filled by the lamellar radial pouches of the basal face of the pneuma- tocyst. The dense parenchyma of its exodermal cellular tissue is traversed by numerous bent tracheae, and by a loose network of hepatic canals. The latter arise from a regular octoradial star of superior liver-canals, which unite in the superior or apical pole of the centradenia. These eight radial main canals embrace the greater part of its surface like 60 THE VOYAGE OF H.M.S. CHALLENGED. eight meridional arches, and open on the periphery of its inferior polar area into the gastral cavity of the central siphon. Their dichotomous branches run along the convex outer surface of the centradenia, enclosed in the numerous radial folds which fit into the corresponding meridional furrows of the subumbrellar face of the pneumatocyst. Central Siphon (fig. 2, in profile ; fig. 3, in longitudinal section). — The large central polypite is a thick-walled cylindrical tube, the length of which surpasses the vertical main axis of the umbrella. Its distal end opens by a four-lobed mouth, whilst its proximal or basal part is conically dilated and closed by the supporting plate separating it from the centradenia (fig. 3). In the periphery of the latter open the eight radial main canals. Sexual Siphons (figs. 2,3, sx). — A corona of eight sexual polypites surrounds the base of the central siphon, and separates it from the tentacular zone. These are much smaller than the central siphon (about half as long, and many times thinuer). Their thin cylin- drical basal part is beset with numerous gonophores ; their spindle-shaped distal part opens by a four-lobed mouth. Tentacles (figs. 1-3, t). — The tentacles are very numerous, and occupy a broad convex zone of the subumbrella, between its equator and the girdle of sexual siphons. After removal of the tentacles, their insertion forms an elegant reticulate girdle, with rhom- boidal meshes (fig. 2, to). The tentacles are arranged in four to five transverse rows, and in eight prominent radial bunches ; the longest of each bunch surpassing the diameter of the umbrella. Genus 4. Porpema? Haeckel, 1888. Porpema, Hkl, System der Siphonophoren, p. 30. Definition. — Porpitidae with a lenticular or subglobular strongly vaulted umbrella, including a campanulate pneumatocyst with radial marginal lobes. Tentacles very numerous, equally disposed in several concentric circles, not forming radial bunches. The genus Porpema has the same strongly vaulted umbrella and the same campanu- late and radially lobate pneumatocyst as the preceding genus Porpalia, from which it differs in the equal distribution of the tentacles along the whole margin of the umbrella. The tentacles are very numerous and densely crowded in several parallel circles (as in Porpita), and they are not grouped into radial bunches. The new genus Porpema was established for an Indian species, Porpema lenticula, which I observed in 1881 in the Indian Ocean, between Aden and Bombay. Another species, described in the following pages as Porpema medusa (PI. XL VII.), was found in the Challenger collection ; this was taken in the South Atlantic, Station 327 (between Buenos Ayres and Tristan da Cunha). A third species [Porpema pileata) was sent me from Chili ; it will be described in my Morphology of the Siphonophorse. • 1 Porpema = Mantle, cloth, wopx/ipx. REPORT ON THE SIPHONOPHORjE. 61 Eegarding the phylogenetic origin of Porpema, we may simply assume that it has been derived from the ancestral genus Porpalia, by multiplication and equalisation of the submarginal tentacles, which form circular girdles, without octoradial arrangement. Porpema medusa, n. sp. (PI. XLVIL). Habitat.— South Atlantic, Station 327; March 4, 1876; lat, 36° 48' S., long. 42° 45' W. Surface. Umbrella (fig. 1, from above; figs. 2 and 3, in profile, fig. 2 with tentacles, fig. 3 after removal of them ; fig. 4, meridional section). — The umbrella is nearly spherical, and has a diameter of 4 to 6 mm. (without the tentacles and the siphon). In some specimens the vertical main axis is somewhat shorter than the equatorial diameter, in other specimens a little longer. Exumbrella (figs. 1, 2, 3, ue). — The superior or apical face is slightly convex, cap- shaped, with a flat annular furrow which separates the central disc from the elevated margin, like a flat hat with a recurved brim. The central disc exhibits a dark pigment- star with sixteen broad rays, and a greater number of finer rays between the dark prominent main rays. A great number of stigmata is arranged in radial rows. Limbus Umbrellas (figs. 1-4, uu). — The free prominent border or margin of the umbrella is nearly horizontally expanded, and slightly reflected upwards, like the broad brim of a flat hat ; it is concave above, convex below. The breadth of the margin equals the diameter of the exumbrellar central disc, and half the ecmatorial diameter of the subglobular umbrella. The thickened edge of the margin is beset with a series of the usual muciparous glands (compare above). Subumbrella (figs. 1-4, w). — The inferior or basal face of the umbrella exhibits beyond the deep submarginal ring-furrow a narrow, smooth zone, which is radially striped ; these stripes are the radial canals ascending from the corona of tentacles to the margin of the umbrella. The broad tentacular zone of the subumbrella (fig. 3) is a circular girdle, which embraces the equatorial zone of the whole umbrella ; its vertical height equals the radius of the latter. After removal of the tentacles it appears elegantly panelled (fig. 3, tu). Pneumatocyst (fig. 7, from above ; fig. 8, from below ; fig. 4, p, in vertical section). — The float, filled with air, is campanulate or nearly spherical, with a central opening at the basal pole of its vertical main axis. The diameter of this inferior opening measures 1 mm., and is about one-third as great as that of the float (3 mm.). The spherical outer surface of the pneumatocyst is in close contact above with the exumbrella, beyond the equator with the tentacular zone of the subumbrella. The inner cavity of the 62 THE VOYAGE OF H.M.S. CHALLENGES. campanulate double-walled float is filled up by the ceutradenia (uc), which passes through its basal aperture (fig. 4). An exquisite regular octoradial structure is visible in all parts of the pneumatocyst. The central chamber (ph) is nearly cylindrical, with an apical stigma (po), and sur- rounded by a regular corona of eight radial chambers (p>q), each with a stigma above and a trachea below. The deep radial furrows between these eight lobes (fig. 7, pr) are prolonged to the periphery of the pneumatocyst, which is divided into eight large main lobes. Each of the latter is again divided by a median incision into two secondary lobes, and these again into four tertiary lobules. Thus the entire peripheral part of the float, beyond its equator, is radially lobate. The deep radial furrows of the superior face of the pneumatocyst correspond to high prominent radial crests or lamellar pouches of its concave lower face, and from the height of these lamellar crests arise numerous tracheae entering into the centradenia. The double wall of the concentric ring-chambers presents therefore a very large surface, by means of the strong development of these radial folds. Centradenia (figs. 6 and 9, transverse sections ; fig. 4, uc, meridional section). — The large central gland exhibits in this Porpitid a strange cylindrical form, due to the unusual development of the vertical or longitudinal axis ; the latter attains 3 to 4 mm., whilst the horizontal diameter of the cylindrical centradenia is only 1 to 1*5 mm. Its middle part is constricted by the embracing inferior margin of the campanulate pneu- matocyst. The radial lamellar crests of the latter fit into corresponding meridional furrows on the surface of the upper head-like part of the centradenia. The lower cylindrical part of this latter is surrounded by the corona of sexual siphons, and its circular basal face is in contact with the basal plate of the sterile central siphon. From the periphery of this plate arise eight radial canals, which are soon forked ; an outer main branch runs outwards to the tentacular zone and the umbrella margin ; an inner main branch ascends vertically between the convex outer surface of the centradenia and the concave inner surface of the pneumatocyst. These eight centripetal liver-canals unite in the apical pole of the central gland, and form here a regular "liver-star" (fig. 8, cm) they give off numerous branches, which form a network in theexodermal parenchyma of the liver ; between its branches numerous bent tracheae are visible. Central Siphon (sh, figs. 2 and 3 in profile, fig. 4 in vertical section, fig. 5 from below, fig. 10 in transverse section). — The large central polypite has an upper pear-shaped part (stomach) and a lower cylindrical part (proboscis). Its length is, in the contracted state, equal to the diameter of the spherical umbrella, and twice as great as its greatest hori- zontal diameter (near the base). Its thick muscular wall exhibits eight longitudinal exodermal furrows on the outside, and eight corresponding radial entodermal folds on the inside (fig. 10). Between these folds open in the basal part the eight radial canals (fig. 9, sf). The distal mouth has eight radial lappets (figs. 3, 4, so). REPORT ON THE SIPHONOPHORvE. 63 Sexual Siphons (sx, figs. 2-5). — A great number of sexual polypites, densely- crowded and arranged in four to five concentric rings, occupies the broad basal zone of the subumbrella, between the central siphon and the corona of tentacles ; their number may be between sixty and ninety in the smaller, two hundred to three hundred in the larger specimens. The form of these contractile gonostyles is very variable, usually spindle-shaped or pear-shaped. The large sterile central siphon is eight to nine times as long and five to six times as broad as each of the small fertile peripheral siphons. Their mouth is small and exhibits eight radial lappets. The basal part is densely beset with medusiform gonophores. Tentacles (figs. 2-4, t). — The corona of tentacles occupies a broad equatorial zone, nearly half the height of the entire subumbrella. After removal of the tentacles (fig. 3) this zone appears as a convex elegantly reticulated girdle, the concave inside of which embraces the inferior half of the campanulate pneumatocyst (fig. 4, p). Each rhomboidal mesh of the reticulum is the basal insertion of one tentacle. There are eight to ten transverse rows of tentacles, one alternating with the other, and each row represents a ring composed of fifty to sixty teDtacles, so that their whole number may be four hundred to six hundred. The length of the longest (in the middle zone) surpasses the greatest diameter of the umbrella, whilst the length of those placed in the superior and inferior rows decreases towards the limits of the tentacular zone. The tentacles are slender cylindrical filaments, club-shaped at the distal end, and beset with three rows of cnidospheres (compare above, pp. 38, 39). Subfamily 2. Porpitellid^e, Haeckel. Genus 5. Porpitella,1 Haeckel, 1888. Porpitetta, Hid., System der Siphonophoren, p. 30. Definition. — Porpitida? with a flat discoidal umbrella, including a circular discoidal pneumatocyst without marginal lobes. Tentacles numerous, arranged in eight or sixteen prominent radial bunches. The genus Porpitella and the following Porpita together make up the subfamily Porpitellidae, characterised by a discoidal or slightly vaulted umbrella, which includes a discoidal pneumatocyst, the distal margin of the latter being circular, not divided into radial lobes. The numerous marginal tentacles in Porpitella are arranged in eight or sixteen regularly disposed radial bunches, whflst in Porpita they are equally distributed along the margin of the umbrella. The new genus Porpitella is founded for those species, formerly placed among Porpita, which are distinguished by the possession of sixteen radial bunches of marginal 1 Porpitella = Small ring of a buckle, dmiimitive of Porpita. 64 THE VOYAGE OF H.M.S. CHALLENGER. tentacles. The first species observed was taken by Eschscholtz in the tropical Pacific, and described in 1829 as Porpita ccerulea.1 It is probably the species of which Huxley afterwards gave a very accurate anatomical description. Another species of this genus, also taken in the tropical Pacific (near the Marianne Islands), is Porpita radiata of Brandt.2 I have been able to compare the excellent (hitherto unpublished) figure and description left by its discoverer, Mertens. Two specimens of a third species, described in the following pages, were found by me in the Challenger collection, from Station 244. It exhibits in several points a remarkable similarity to the interesting deep-sea Medusa, Pectanthis asteroides.3 I call it therefore after this Trachynemid. The phylogenetic origin of Porpitella is to be found in Porpalia. It may be derived from this ancestral form by flattening of the highly vaulted umbrella and the pneumatocyst, which thus become more or less discoidal. Porpitella pectanthis, n. sp. (PL XLVL). Habitat. — Station 244, Northern Pacific, between Japan and Honolulu ; June 28, 1875 ; lat. 35° 22' N., long. 169° 53' E. Surface. Umbrella (fig. 1, from above; fig. 2, from below; fig. 3, meridional section). — The disc is a biconvex lens, the horizontal diameter of which amounts to 12 or 15 mm., and is three times as great as its vertical axis (4 to 5 mm.). The meridional section (fig. 3) demonstrates that the larger superior half of the umbrella is occupied by the discoidal pneumatocyst (p), the smaller inferior half by the lenticular centradenia (uc). Exumbrella (fig. 1). — The superior or apical face \ of the umbrella is slightly convex, and exhibits an elegant radial striation, crossed by numerous concentric rings. Sixteen more prominent ribs bear a number of conical papillse, and between these numerous more delicate, also denticulate, ribs radiate from the centre. The stigmata open at the apex of the conical papillse. An elegant pigment-star with sixteen brown rays indicates the course of the main ribs. Limbus Umhrellee (figs. 3, 7, um). — The free margin of the umbrella is prominent at the periphery of the lenticular body, and about one-third or one-fourth as broad as the length of its radius ; it is circular, divided by sixteen slight incisions into sixteen flat marginal lobes (fig. 1). Its thickened edge includes a series of marginal glands (fig. 7, us). Subumbrella (fig. 2). — The inferior or basal face of the umbrella is slightly convex, and divided into four different parts of nearly equal breadth. The central part is occupied by the large sterile central siphon ; this is surrounded by three circular concentric zones ; the first zone is occupied by sixteen smaller sexual siphons (bearing 1 1, p. 179, Taf. 16, fig. 5. 2 25, p. 40. 3 Zool. Cliall. Exp., part xiL p. 20, pis. vii., viii. EEPOET ON THE SIPHONOPHOE^. 65 clusters of gonophores) ; the second zone is marked by the corona of tentacles, divided into sixteen bunches ; and the third, outermost, zone is formed by the inferior face of the free peripheral limb. Pneumatocyst (fig. 4, superior or apical view; fig. 5, inferior or basal view; figs. 3, 6, 7, p, meridional sections). — The float, filled with gas, is a circular disc, the horizontal diameter of which (9 to 12 mm.) is ten to twelve times as great as its vertical diameter (1 to 1"2 mm.). Its thickness is nearly equal throughout, or increases a little towards the margin. The superior or exumbrellar face (fig. 4) is slightly convex, and exhibits numerous denticulate radial ribs, sixteen of which are more prominent (fig. 1). Numerous stigmata (pe) open by the conical papillae of the radial ribs. The inferior or subumbrellar face (fig. 5) is slightly concave and radially folded, numerous radial furrows (sixteen of which are deeper) corresponding to the ribs of the upper face. Numerous articulate tracheae (figs. 5, 6, pt ; fig. 8) arise from conical protuberances of the inferior ribs, which are prominent between every two radial sulci. The central chamber of'the pneumatocyst (figs. 3, 6, p>h) opens above by a central apical stigma (jpo), and is surrounded by a regular corona of eight triangular radial chambers ; each of these bears a circular stigma (fig. 4, pe) on its upper face (placed in variable points), whdst from its lower face arise one or two short tracheae. The corona of con- centric ring-chambers, which surround the girdle of eight radial chambers, is composed of nine to twelve rings. The height (or vertical diameter) of these rings increases from the centre towards the periphery, whilst the breadth (or horizontal diameter) decreases (figs. 3, 7, pk). The inner or proximal wall of each ring is concave, the outer or distal wall convex. The concentric rings communicate one with another by eight radial rows of apertures (figs. 3, 4, pg), which are placed interradially between the perradial sulci. The innermost or first ring embraces the octoradial chamber-girdle not only from the distal, but also completely from the basal side (figs. 3, 6, pk) ; the other concentric rings embrace one another only from the distal side. Centradenia (figs. 3, 6, 7, tic). — The large central gland (or the so-called liver) is a circular biconvex lens, the horizontal diameter of which (6 to 8 mm.) is three to four times as great as the vertical axis (2 mm.). Its upper surface is in close contact with the concave lower surface of the pneumatophore, its lower surface with the subumbrella. Numerous radial ribs of the upper surface fill up the radial sulci of the lower surface of the pneumatophore. The brown liver vessels are very densely crowded in the upper half, loosely scattered in the lower half of the pneumatophore. ATI these hepatic canals unite on the lower surface of the centradenia into descending canals, which pierce the fulcrum of the subumbrella ; eight of these (forming an inner girdle of openings) open into the peripheral part of the base of the central siphon ; sixteen others (forming an outer girdle) open into the sexual siphons (sx). The majority of the canals of the central gland contains in their epithelium that black pigment which is regarded as " hepatic (ZOOL. CHAIX. EXP.— PAKT LXXVII. — 1888.) Hhkh 9 66 THE VOYAGE OF H.M.S. CHALLENGER. granules " ; whilst another part is colourless or more or less filled up with guanin- crystals ("renal granules"). The uppermost canals (between the pouches of the pneumatocyst) are colourless. Central Sijjhon (figs. 1, 2, 3, sa).— The large sterile central polypite has the usual form of a thick-walled cylindrical or inversely conical tube. Its dilated base occupies the centre of the subumbrella, and is separated from the centradenia by a thick fulcrum or supporting lamella, the gastrobasal plate ; it exhibits a peripheral ring of eight openings, leading into the eight radial main vessels (compare p. 31). Sexual Siphons (fig. 2, fig. 3, sx).— The sixteen peripheral polypites form a regular corona around the base of the large central polypite, and are separated by a circular furrow from the corona of tentacles. They bear numerous gonophores on their thinner basal part, whilst their spindle-shaped distal part opens by a four-lobed mouth (compare p. 37). Tentacles (figs. 1-3, t). — The numerous tentacles form a submarginal corona, which occupies about one-third of the subumbrella. They form six to eight concentric rows, and exhibit a very different length. Seen from above or below the corona presents sixteen elegant rays, each of which is composed of an adradial bunch of larger tentacles. The longest tentacles (in the middle of each radial bunch) surpass the diameter of the umbrella in the contracted spirit specimens, and may be much longer in the living animal. The form and structure of the tentacles, with three rows of cnidospheres, is that which is usual in all Porpitidae (compare pp. 38, 39). Genus 6. Porpita,1 Lamarck, 1816. Porpita, Lmk., Hist. nat. anim. s. vert., t. ii. p. 483. Definition. — Porpitidae with a flat discoidal umbrella, including a circular discoidal pneumatocyst without marginal lobes. Tentacles very numerous, equally disposed in several concentric circles, not forming radial bunches. The genus Porpita possesses the same flat discoidal umbrella, and the same circular, not radially lobate, pneumatophore, as the preceding genus Porpitella ; but it differs from the latter in the ecpial distribution of the tentacles along the whole margin of the umbrella. The tentacles are very numerous, and densely crowded in several parallel circles (as in Porpema), and they are not grouped into radial lobes. The genus Porpita, hitherto regarded as the only representative of the famfly Porpitidae, was founded in 1816 by Lamarck upon the first known species, which Forskal had observed in the Mediterranean, and described very accurately as Holothuria denu- data.2 Eschscholtz, the founder of the class Siphonophoras, in 1829 placed Porpita in his family Velellidae, and gave the following definition :3 — " Corpus orbiculare, supra 1 Porpita = Ring' of a buckle, iripirn. * 11, p. 103, Taf. 26, fig. L 1. 3 1, p. 176, Taf. 16. REPORT ON THE SIPHONOPHORiE. 67 inerme. Tentacula marginalia trifariam glaudulosa." He described five different species, Two of these only appertain to the genus Porpita as above defined [Porpita mediter- ranea and Porpita umbella). The best and most accurate anatomical descriptions of the genus are those given in 1853 by Kolliker of Porpita mediterranea (4), and in 1881 by Alexander Agassiz of the North Atlantic Porpita linnmana (57). Closely related to the latter is probably the tropical Atlantic Porpita umbella of Eschscholtz (1). The collection of the Challenger contains several specimens of a new species, taken in the Northern Pacific, and figured in PI. XLV. as Porpita fungia. Three other distinct species seem to be Porpita liitkeana (25), from the Indian Ocean ; Porpita pacifica of Lesson (22) ; and a new southern form, Porpita australis. The phylogenetic origin of Porpita is to be found in the preceding genus, Porpitella, from which it is derived by multiplication and equalisation of the tentacles, not forming separated radial bunches. Porpita fungia, n. sp. (PI. XLV.). Habitat. — Stations 253 to 255, Northern Pacific, between Japan and Honolulu; 14th to 19th July 1855; lat. 38° 9' N. to 32° 28' N, long. 15G° 25' W. to 154° 33' W. Surface. Umbrella (fig. 1, meridional vertical section). — The disc is flat, circular, of equal thickness throughout nearly its whole extent, like a coin. Its diameter is usually 20 to 25 mm., but amounts in the largest specimens to 30 mm. ; its thickness (without the siphon and tentacles) is 4 to 6 mm. A meridional section (fig. 1) demonstrates that three-fourths of the thickness belong to the centradenia, one-fourth to the pneumatocyst. Exumbrella (fig. 2). — -The superior or apical face of the umbrella is flat or slightly convex, a little depressed in the centre. It is covered with numerous, irregularly scat- tered, conical tubercles, arising from the upper face of the pneumatocyst. Their number and size increase towards the centre. The peripheral half of the exumbrella is pierced by numerous irregularly scattered stigmata (pe). The regular octoradial striation and the numerous concentric rings, which shine through the silvery exumbrella, are due to the structure of the subjacent pneumatocyst. An elegant network of anastomosing canals, with narrow irregular polygonal meshes, is expanded everywhere in the exumbrella. Limbus Umbrellss. — The free membranous border or limbus of the umbrella overlaps the tentacular zone to the extent of 1 to 2 mm. It is flatly expanded or reflected upwards, and includes a series of marginal glands of the usual form (compare above). The numerous canal-branches of the exumbrella anastomose richly, form an irregular network, and unite in a circular canal inside the glandular corona ; and from this arise the vessels of the centradenia and the subumbrella. 68 THE VOYAGE OF H.M.S. CHALLENGER. SubumbrcUa (fig. 1). — The inferior, distal or basal face of the umbrella is divided into four different zones. The central zone, with a diameter of 3 to 4 mm., is occupied by the gastrobasis or the fundus of the large central siphon. The broad middle or gonostylar zone (4 to 6 mm. in breadth) is occupied by the numerous sexual siphons or gonostyles. The third or tentacular zone (2 to 3 mm. broad) bears the numerous tentacles. The fourth or outermost zone (1 to 2 mm. broad) is formed by the inferior face of the free mantle-border. Pneumatocyst (fig. 1, fig. 5, pf). — The float, filled with air, is a flat circular disc of 15 to 20 mm. diameter. Its thickness increases considerably from the centre towards the margin, owing to the height of numerous radial folds. The form of the pneumatocyst presents a great resemblance to a regular Fungia, the disc being folded radially in a very regular manner. The vertical folds arise in the upper surface in eighty or ninety high radial ridges, with deep valleys between them, whilst branched radial lamellae are pro- minent on the lower face. The octoradial structure of the pneumatocyst is very distinct on its upper face, since the eight equidistant primary rays are more prominent than all the others. They arise from the septa of the eight radial chambers which surround the central chamber. The octant between every two primary folds is bisected by a secondary ray, somewhat less prominent, but also stronger than all the others. Sixteen tertiary or adradial ridges arise in the middle between the eight primary and the eight secondary rays. The other rays are developed in the peripheral half of the disc only. The total number of radial ribs is about eighty in a specimen of 1 5 mm. diameter, one hundred and twenty in a larger specimen of 20 mm. The most elevated part of each radial rib bears a radial series of stigmata. The number of concentric annular septa is about fifty in the former, eighty in the latter. These are wider (about twice as broad) in the middle part of the disc than in the central and the peripheral part. The thickness of the chitinous wall of the air-chambers is much greater in the upper than in the lower face, and it increases from the periphery of the disc towards the centre. It equals in the central half of the disc the height of the air-chambers. Numerous strata of chitinous substance are here deposited as secondary supports of the disc-wall, and have closed the stigmata (fig. 1 , p2). The conical tubercles of the surface of the disc are pro- duced by stronger deposits at separate places (p1). The stigmata of the air-chambers remain open in the peripheral part of the disc only (_pe); they are urn-shaped, very numerous, and irregularly scattered in radial rows (fig. 2, pe). The lower face of the pneumatocyst, which is in close contact with the centradenia, is more strongly folded than the upper face ; the folds are here ramified centrifugally, and form high triangular lamellae, the height and number of which increase from the centre towards the periphery (fig. 1, £>4). Eight primary radial lamellae, arising from the septa of the eight pericentral radial chambers, are more prominent than the eight secondary REPORT ON THE SIPHONOPHOR^. 69 alternating, and these more than all the other lamellae. The centrifugal dichotomous ramification of the lamellae increases gradually towards the periphery, and new peripheral folds are interposed between the centrifugal folds, so that their number at the margin of the disc is much greater than that of the simple radial ridges of the upper surface. The crest of the folds is elegantly denticulate, and from it arise two opposite radial rows of tracheae. The tracheae or aeriferous tubules in this Porpita are exceedingly numerous, amounting to many thousands. The majority of them are short and end half way up the centradenia. A smaller number pierce the latter and the subjacent support, and project into the exodermal wall of the central siphon and the surrounding gonostyles. They terminate here in the basal half of the latter, inside the exodermal epithelium (fig. 8, £>£). The circular concentric ring-chambers of the pneumatocyst communicate one with another by radial apertures or " pneumothyras," oblique elliptical openings, which are placed in the radial sulci between the folds of the lower face. The central chamber possesses, as usual, a central stigma in its upper face, as does also each of the surrounding eight radial chambers. From the basal part of each of the latter arises a bunch of three or four short tracheae. In the basal sulci between these are eight openings, which lead into the first ring-chamber. Centradenia (fig. 1, uh, meridional section; fig. 3, inferior face; figs. 6-8, un, parts of sections). — The large central gland (or the so-called "liver") is a biconvex lenticular disc, the horizontal diameter of which is four times as great as its vertical axis ; the latter measures in a specimen of .20 mm. diameter (umbrella) 3 mm., the former 12 mm. Its superior face has a regular radiate appearance ; innumerable radial folds (thirty-two of which are stronger) fitting into the radial grooves which lie between the lamellar pouches of the inferior face of the pneumatocyst. The height of these folds increases from the centre towards the periphery. The circular margin meets the line which marks the boundary between the gonostylar and tentacular zones. The inferior face of the centradenia is less convex than the superior, and is divided into a white central disc (kidney) and a peripheral brown radiate zone (liver). The former covers the roof of the central siphon (sa), and surrounds it like a broad, white girdle (fig. 3, un) ; the peripheral part of the brown liver (fig. 3, sh) surrounds again the former as a dark radiate ring. The gonostyles arise as well from the former as from the latter. The meridional or vertical section of the centradenia (fig. 1) demonstrates that its solid exodermal parenchyma is traversed by a dense network of innumerable canals, the greater part of which are darkly coloured by a brown (hepatic) pigment. This is want- ing in a part of the superficial canals, and in the inferior white central disc (" kidney"), which contains masses of guanin-crystals. Innumerable tracheae terminate between the canal branches ; their distal openings being surrounded by (air-secreting ?) exoderm cells (compare the explanation of figs. 6-8). 70 THE VOYAGE OF H.M.S. CHALLENGER. Central Siphon (fig. 1, sm, fig. 4). — The large sterile central polypite is inversely conical, in the contracted state about as long as the diameter of its broad base, which is a quarter or a fifth of the diameter of the umbrella. Its thick muscular wall exhibits sixteen longitudinal folds, and between these run sixteen deep grooves, which open at the bottom of the stomach into the sixteen radial main canals. The distal mouth of the central siphon is octolobate. Its opposite proximal roof is separated by a strong fulcrum or supporting lamella, the gastrobasal plate (fig. 1, st), from the superjacent kidney. The sixteen radial main canals, which arise from the periphery of the base of the stomach, run along the subumbrella towards the peripheral edge of the limb, where they are united in a circular marginal canal. They give off innummerable branches ; of these the ascending ones enter into the centradenia, the descending partly into the gonostyles, partly into the tentacles. Numerous superficial mantle- vessels arise from the marginal canal and enter centripetally into the exumbrella, where they form an irregular dense network above the pneumatocyst. Sexual Si£)hons (figs. 1,8, sx). — The gonostyles cover the greatest part of the sub- umbrella, and are densely crowded in the broad zone which lies between the central siphon and the tentacular zone. Their number amounts to several hundreds. They are covered with bunches of gonophores in the cylindrical proximal half, with patches of cnidocysts in the spindle-shaped distal half. Their distal mouth is often divided into four or eight small lobes. Tentacles (fig. 1, t). — The tentacular zone of the subumbrella, inside the free pro- minent limb, is about as broad as the base of the central siphon. It presents an elegant reticulum, with rhomboidal meshes, when the tentacles are removed ; each mesh being the base of a detached tentacle. These are arranged in six to eight concentric rows. In the largest specimens their number exceeds one thousand. Their length may exceed the diameter of the umbrella ; the majority, however, are mueh shorter. The structure of the club-shaped tentacles is the usual one, with three rows of stalked cnidospheres (compare p. 39). Family III. Velellid,e, Eschscholtz, 1 829. VeleUidx, Eschscholtz, System der Aealephen, p. 165 (sensu restricto). Definition. — Disconectse with an elliptical or amphithect, often nearly quadrangular umbrella, including a polythalamous pneumatocyst of the same form, which is composed of numerous concentric rings, and usually bears in its diagonal a vertical crest. Always a membranous vertical sail upon the umbrella. Marginal tentacles simple, without cnido- spheres. Central siphon surrounded by numerous peripheral fertile siphons, which bear the gonophores. The family Velellidae, founded by Eschscholtz in 1829, comprised originally all Dis- REPORT ON THE SIPHONOPHOR.E. 71 conectaa, or all the " Medusas chondrophorse " of Chamisso and Eyssenhardt. After- wards (in 1835) Brandt separated the true Velellidaj (with elliptical disc and vertical crest) from the Porpitidse (with circular disc, without crest). The Velellidae, thus restricted, agree with the Porpitidse in the polygastric structure, not only the large sterile central siphon possessing an open mouth at its distal end, but also each of the surrounding gonostyles. These peripheral polypites, which bear the gonophores budding from then proximal part, are therefore not mouthless palpons (as in the Discalidse), but mouth-bearing, feeding, and digesting secondary siphons. On the other hand, the Velellidse differ from the Porpitidse (as well as from the Discalidse) in the peculiar amphithect or bilaterally-radial ground-form of the body, produced by the prolongation of the sagittal axis, and shortening of the frontal axis ; and, further, in the development of a peculiar vertical crest or sail, which arises from the exumbrella either in the sagittal or in a diagonal axis. Eschscholtz, in his fundamental work,1 divided his Velellidae into three genera : — Rataria, Velella, and Porpita. The genus Rataria, however, was considered by the ma- jority of subsequent observers to consist only of young forms of Velella. A new species of Rataria, found in the Challenger collection, proved to be an autonomous genus, provided with gonophores. Although the different Ratarize, described by some authors, may be merely young Velellse, there nevertheless exist also Velellidae which become mature in the permanent Rataria-fovm. The numerous species of the true Velella were disposed in my System (p. 31) in two genera : — Velella (sensu restricto), with an ebiptical umbrella and simple corona of tentacles ; and Velaria, with quadrangular umbrella and double or multiple corona of tentacles. For this latter the old name Armenista (or Armenistarium of Carburius, 1757) may be better retained. Umbrella. — The common trunk of the cormus, which corresponds to the umbrella of a hydromedusoid person, is in all Velellidse a flat, horizontally expanded, and floating disc, distinguished from the circular disc of the Porpitidse and DiscaUdae, firstly, by its bflateral form, and secondly, by tke vertical crest or sad. Tke latter is originally notking more tkan a simple longitudinal fold of tke exumbrella, and may be regarded as tke primary cause of tke peculiar fundamental form. Tke tkree genera wkick we kave distinguisked represent tkree different degrees or historical steps in its peculiar development : — Rataria, next to tke ancestral group Porpitidse, kas an elliptical disc, and tke sail placed in its sagittal plane, in tke longer axis ; in Velella the sad turns a little around the vertical main axis, and comes to be situated in a diagonal axis of the disc ; finally, in Armenista the disc becomes quadrangular or parallelogram-shaped. Besides, a vertical crest of the pneumatocyst, as a support of the soft sail, is developed in the two latter genera, but is wanting in Rataria. Amphithect Fundamental Form. — The peculiar fundamental form presented by the 1 1, 1829, p. 1G6. 72 THE VOYAGE OF H.M.S. CHALLENGER. umbrella of the Velellidse may be called either amphithect or bilaterally-radial. It is characterised by three unequal dimensive axes, each perpendicular to the other two ; one being allopolar (with unequal poles), the other two isopolar (with equal poles). The allo- polar axis is the vertical main axis, its upper or apical pole determined by the top of the sail, its lower or basal pole by the mouth of the central siphon. The greater isopolar axis is the sagittal axis of the disc (the major axis of the ellipse), which divides it into two equal symmetrical halves, right and left. The smaller isopolar axis is the transverse or frontal, which is perpendicular to the latter, and separates the disc also into two equal halves, dorsal and ventral ; it corresponds to the minor axis of the ellipse. The octoradial character of their amphithect ground-form is always indicated by the origin and course of the eight primary gastro-canals, which arise from the base of the central siphon ; secondly, it is marked by the centre of the pneumatophore, which exhibits more or less distinctly eight radial chambers around the simple central chamber ; thirdly, it is often indicated by the radial arrangement of eight primary marginal tentacles and of other parts. The umbrella of the Velellidse is therefore composed originally of eight equal parameres (or homotypical radial parts), like that of the Porpitidse and Discalidse ; but it differs from that of the latter in the fact that the parameres are dis- posed in pairs, so that the disc may be regarded also as composed of four paired quadrants ; and sometimes the limits of these are sharply marked off by four marginal incisions of the pneumatophorous disc. The quadriradial appearance, the body seeming to be composed of four parameres (as in the most Medusa?), is mainly obvious in the more highly developed Velellidse. But regarding their phylogenetic origin from the Porpitidse, we must assume that this funda- mental form has arisen from a true octoradial type composed of eight parameres. In no case can we regard this fundamental form as uniradial, a point of view which is repre- sented by Chun.1 I agree perfectly with the explanation of the origin of the sail which Chun there gives (loc. cit., p. 15) ; but I cannot agree with his promorphological deductions. I cannot concede at all that uniradial fundamental forms exist anywhere. A trice asymmetry, mentioned by many authors in the umbrella of the Velellidse, does not exist ; for in every case the two halves of the body, which are separated by any possible meridional section (through the vertical main axis), are perfectly equal ; and the dorsal and ventral halves cannot be distinguished by any character ; the right and left halves being likewise equal one to another. The so-called "asymmetry" in the disc of many Velellidse is only the beginning of a spiral twisting, caused by a slight dislocation of the vertical sail, turning more or less around the vertical main axis. But in the elliptical Rataria (and also in the similar Ratarula-larvse of Velella and Armenista) the sail is placed in the sagittal plane, and its later dislocation from this plane is a secondary alteration, comparable to the turning of the sail in a boat sailing before the wind. 1 Chun, Fauna and Flora des Golfes von Neapel, i., Ctenophoren, 1880, p. 14. REPORT ON THE SIPHONOPHOR^E. 73 This sail, or the vertical crest of the Velellidse, is the most characteristic part of their trunk, and its gradual development is the primary cause of their peculiar amphithect form. The sail is originally nothing other than a small fold of the exumbrella, or the soft upper lamella of the disc. We may assume that the elevation of such a fold in any ancestral Porpitidse was very useful as an aid to the locomotion of the pelagic animal floating on the level of the sea by its horizontal disc. Natural selection, therefore, will have increased the height of that crest, driven by the wind, and the immediate effect must have been the transformation of the circular disc into an elliptical one, the sail occupying the major axis of the ellipse (Rataria). Afterwards the soft and contractile sail becomes supported by the development of an inner chitinous crest, arising from the pneumatophore ( Velella), and finally the whole outline of the disc, and the arrangement of its marginal parts, assumes the form of a parallelogram, and the sail is placed in its diagonal axis (Armenista). A continuous series of intermediate transitional forms conducts us from the elliptical Rataria (with sagittal sail), through different forms of Velella, to the most specialised parallelogram-shaped Armenista (with diagonal veil). The special form of the sail in the two latter genera depends upon that of the supporting firm crest of the pneumatophore ; in Rataria, however, where the skeleton-crest is wanting, its form differs greatly according to its varying state of contraction. Exumbrella. — That part of the trunk which includes the pneumatocyst, and cor- responds to the invaginated exumbrella, is composed in the Velellidse, as in the other Disconectge, of two parallel membranes connected by branched septa. The outer mem- brane, or the permanent exumbrella (the uninvaginated part), is the pneumatocodon ; its exodermal epithelium is armed with many cnidoblasts, and beyond it is placed a strong muscular plate, composed of longitudinal or radial fibres. The inner membrane, or the invaginated part of the exumbrella, is the pneumatosaccus ; its thinner exodermal epithelium envelops like a complete sac the whole surface of the pneumatocyst, and this is nothing more than the hardened chitinous cuticula secreted by the former. A great number of septa connect both membranes, and between them occur the canals of the exumbrella, more radially in the horizontal surface of the disc, more longitudinally and parallel in the vertical surface of the sail (on both sides of it). These pallia! canals open in the periphery of the horizontal disc into the marginal canal, and along the free margin of the sail into a crescentic canal, running along the whole margin. Pneumatocyst. — The chitinous polythalamous float filled with air, which we call pneu- matocyst (usually called the " inner shell "), always assumes the form of its surrounding matrix, the pneumatosaccus. It is, therefore, in the Velellidaa elliptical or cmadrangular, and very different from that of the Porpitidae and Discalidse, where it is always circular and regularly octoradial. Since, however, the former have arisen originally from the latter, their pneumatocyst also must be regarded as an amphithect or bilateral modifica- tion of that circular float of the latter. Indeed, in all Velellidae there are traces, more or (ZOOL. CHALL. ESP. — PART LXXVIJ. — 1888.) Hhhll 10 74 THE VOYAGE OF H.M.S. CHALLENGER. less distinct, of the same octoradial composition of the centre of the float, which is immediately evident in all Porpitidse and Discalidse. The pneumatocyst of Rataria consists only of a horizontal, slightly campanulate or flatly conical disc of elliptical outline, whilst that of Velella and of Armenista bears a solid vertical triangular crest, placed in a diagonal axis of the disc. The pneumatocyst of Rataria is of the highest interest, since it offers all desirable intermediate forms between the bilateral crested float of the other Velellidaj and the circular crestless float of the Porpitidse. This is the more important, as the young Ratarula-larvse of Velella and Armenista exhibit the same structure of the pneumatocyst (during a longer or shorter time) which is permanent in the autonomous genus Rataria. A central chamber, placed in the centre of the exumbrella, is here always surrounded by eight radial chambers, just as in the Porpitidae and Discalidse. But whilst in these two families each radial chamber (as well as the central chamber) possesses a stigma, or a free opening on its upper surface, there are in the similar pneumatocyst of Rataria three such openings only, one in the upper face of the central chamber and one on each side of it, in the two chambers, which are directed towards the two sagittal poles, or the poles of the major axis of the ellipse. The stigmata of the six other chambers seem to be lost by phylogenetic reduction (PI. XLIV. fig. 8). Each of the eight radial chambers usually possesses a trachea or a bunch of descending air-tubules on its lower face (fig. 9). The octoradial structure of the pneumatocyst, which is obvious in Rataria, is also recognisable in the similar Ratarula-larvse of Velella and Armenista. But it is not ecmally distinct and well preserved in all species of these two genera. The peripheral wall of the central chamber, which separates it from the eight surrounding radial chambers, is pierced by eight openings, connecting the former with the latter. These openings of communication afterwards often become so enlarged that the eight chambers appear only as radial lobes of the central chamber, and sometimes the eight radial septa between the former are so reduced that the eight lobes nearly disappear. This fact explains the striking contradictions of the former observers, some of whom interpret the Ratarula-larvse as young forms of Velellidse (Huxley, Bedot, &c), some as larvae of Porpitidse (Agassiz, Burmeister, &C.).1 Indeed all these larvse belong to the Velellidse, since the larvae of the Porpitidse never develop a vertical sail. But the octoradial struc- ture of the central part of the pneumatocyst, which is always very distinct in the latter, is of very variable distinctness in the former ; sometimes it is well preserved, at other times not, and sometimes the cenogenetic modification is so strong that it becomes quite lost. The concentric ring-chambers of the pneumatocyst, which surround its octoradial central part, are not circular in the Velellidse (as in the Porpitidse and Discalidse), but more or less elliptical, with prolonged sagittal axis, and shortened transverse or frontal 1 Compare Pagensteclier, 55. REPORT ON THE SIPHONOPHORiE. 75 axis. The ellipses are short and nearly circular in the young Rataria; they become longer and sometimes slightly quadrangular in Velella, and completely parallelogram- shaped in Armenista. The number of these concentric ring-chambers is very variable in the different species, two to eight only in the smallest species (Rataria), twenty to thirty or more in the larger (Velella), and fifty to eighty or more in the largest forms (Armenista). Pneumothyrse. — The concentric ring-chambers of the pneumatocyst are all in open communication one with another. Each annular elliptical septum between two chambers is pierced by two opposite openings or septal pores, one in the dorsal half, the other in the ventral half of the disc. All these septal pores are placed in a straight line, cutting the major axis of the disc at a very small angle, as was first shown by Kolliker.1 There are, therefore, not eight radial rows of pneumothyrae as in the ancestral Porpitidae, but only two opposite radial rows. In some species of Velellidse, however, there seem to exist two other opposite rows, placed in a transverse line near the frontal groove. Stigmata. — All Velellidae possess a relatively small number of superficial openings on the ring-chambers, which lie at their apical or superior surface, and serve for the emission of gas. The number of these stigmata is much smaller than in the Porpitidae. Usually only every third or fourth ring possesses a pair of opposite stigmata ; sometimes rings with and without stigmata alternate regularly ; rarely each ring possesses one pair. All the gas-pores or stigmata discovered by Kolliker lie in a straight line, just as do the septal pores, and the line of the former is placed between the line of the latter and the base of the vertical crest. Half the stigmata, therefore, lie in the ventral half of the disc on the right side of the crest, and the other half in the dorsal half on the left side, or vice versa (compare PL XLIII. fig. 5, pe). The direction of the series of stigmata in all Velellidae is already marked by the three first stigmata of their Ratarula-larvae. These, as well as the mature Rataria (PI. XLIV. fig. 8), possess only three stigmata, which are placed in a line, cutting the sagittal axis of the horizontal disc at an acute angle ; one of the three pores lies near the centre (j)o) ; the second on the left (pe), and the third on the right side. All the following stigmata develop in the direction of that line. In many Velellidae the stigmata lie so near the crest-basis that they are difficult to find. Tracheae. — The number of tracheae arising from the lower face of the pneumatocyst in the Velellidae is also much smaller than in the Porpitidae, but they are longer than in the latter and usually branched. In many species eight tracheae arise from the periphery of the eight radial chambers which surround the central chamber ; these are usually branched (PI. XLIII. fig. 6 ; PI. XLIV. fig. 9). A small and variable number of other tracheae arise here and there, irregularly scattered, from other parts of the pneumatocyst. The vertical crest of the pneumatocyst, or the sail-skeleton, wanting in Rataria, exhibits different degrees of development in the various species of Velella and Armenista. 1 Kolliker, 4, Tat xi. fig. 11. 76 THE VOYAGE OF H.M.S. CHALLENGER. At first it is a low thin crest in the base of the soft muscular sail, and afterwards arises as a hioh vertical lamella, usually of more or less triangular form ; the top of the triangle is the highest point of the body, and placed in the upper or apical pole of the vertical main axis. The broad base of the triangle, placed in one diagonal Hue of the parallelo- gram-shaped disc,' is usually about one and a half times as long as each of the two equal lateral sides. The substance of the crest is a thin chitinous plate, secreted from that part of the pneumatosaccus which arises from the diagonal of the disc as a vertical fold. It is perfectly homogeneous, solid, and structureless, and contains no canals nor air-chambers. A number of lines or thin ridges, parallel to the two ascending edges of the triangular crest, and visible on both flat sides of it, indicate its successive growth. Central Siphon. — The large central polypite of the Velellidae differs from that of the Porpitidse and Discalidse in its bilateral compression ; the transverse section of its basal part is circular in the two latter families, elliptical in the former ; the major axis of the ellipse corresponds to that of the umbrella. Its general structure and shape are other- wise the same as in the other Disconectse. The central siphon is comparatively large in the small Rataria, where the number and size of the peripheral siphons is small, whereas in the larger species of Velella and Armenista, where the peripheral siphons are very numerous, the central polypite is less preponderant. The thick wall of the central siphon is very contractile, and composed of two strong muscle-plates, an outer longi- tudinal and an inner circular, both separated by a strong elastic fulcrum. The wall often exhibits prominent radial or longitudinal folds, eight in the smaller, sixteen or more in the larger forms. Correspondingly, the terminal mouth is often four-lobed or eight-lobed (PI. XLIII. figs. 4, 8, so ; PI. XLIV. figs. 2-5, so). Centradenia. — The large central gland, or the so-called " central organ " (formerly " liver"), exhibits the peculiar composition described above (p. 31). Bedot has given an accurate histological description of its structure (59, 60). In the Velellidse it is more developed than in the Discalidse, but much less than in the Porpitidse. It does not usually occupy the greater part of the superior face of the subumbrella as in the latter, but only its central part, between the centre of the float above and the base of the large central siphon below. The outline and the horizontal section of the centradenia are not circular, as in the Porpitidse and Discalidse, but elliptical or lanceolate, the major axis of the ellipse corresponding to that of the umbrella. Its superior or apical face is more or less conical, and fills up the concave inferior face of the pneumatocyst. Its inferior or basal face is even, separated by the gastrobasal plate from the base of the central siphon. The difference between the hepatic vessels in the superior half of the central gland, and the renal vessels in its inferior half, seems to be usually not so striking in the Velellidse as in the Porpitidse. The canal-plexus, as well as the compact parenchyma of exoderm cells, which fills up the interstices of the canal-network, and probably secretes the gas, is in the former far less developed than in the latter. This weaker development of the REPORT ON THE SIPHONOPHOR^E. 77 centradenia in the Velellidae is probably due to their pelagic habit of life, and the development of the vertical sail as an excellent means of passive locomotion. The Porpitidas, on the other hand, are probably inhabitants of different depths, and only occasionally come to the surface ; their hydrostatic apparatus and the gas-secreting gland are therefore more developed. Gonostyles. — The numerous polypites of the . subumbrella, which produce by budding the medusiform gonophores, are in the Velellida± mouth-bearing siphons, as in the Porpitidae, not mouthless palpons as in the Discalidae. They are, therefore, usually called "smaller polypes," " sexual polypites," or "peripheral siphons" (shortly "peri- siphons "). They occupy usually, densely crowded in great numbers, a broad gono- stylar zone, and often the whole space of the subumbrella between the central siphon and the submarginal corona of tentacles. But at first their number seems to be restricted to eight or sixteen ; at least this is the case in Rataria cristata (PI. XLIV. fig. 2), and also in some similar young larvae (Ratarula) of Velella. Their form and structure are the same as in the Porpitidse, already described above (p. 36). Tentacles. — The corona of submarginal tentacles is in the Velellidas far less developed than in the Porpitidae ; their structure, too, is simpler than in the latter. The corona is simple, composed of a single series of filaments in Rataria and Velella ; it is double or multiple, and composed of two or three (seldom more) series in Armenista. Their number is probably originally eight, and this occurs in some Ratarula-larvaa ; but there are other similar larvae in which the corona bears a variable number of tentacles irregu- larly disposed. Some very small and young larvae of the Ratarula-form exhibit only two tentacles, at opposite poles of the major axis of the ellipse, and corresponding to the two primary stigmata of the pneumatocyst.1 The only Rataria which 1 have observed in the adult state (with gonophores) possessed sixteen tentacles, rather regu- larly disposed along the limb (PI. XLIV. figs. 1, 2). In Velella this number is soon increased, and amounts in the simple series usually to fifty to eighty, often more than a hundred. In A rmenista there are usually some huudreds, or sometimes thousands ; the smaller tentacles of the outer (or distal) series alternating with larger filaments of the inner (or proximal) series. Form of the Tentacles. — The general structure of the tentacles in the Velellidae is the same as in the Porpitidae, but their form is much simpler, and the three rows of stalked cnidospheres, which are characteristic of the tentacles of the latter famdy, are wanting. The submarginal filaments of all Velellidae are simple cylindrical tubes, sometimes slightly compressed, usually gradually tapering towards the rounded distal end, seldom somewhat club-shaped ; in very young larvae they are conical and pointed. Their arma- ture with cnidoblasts is very different from that of the two other families. Usually each tentacle bears only two lateral ribands of cnidoblasts, sometimes four (two stronger 1 Compare Bedot, 60, 1884. 78 THE VOYAGE OF H.M.S. CHALLENGER. lateral and two weaker medial, an upper and a lower). The cnidoblasts are often more crowded towards the distal end. Sometimes they are irregularly scattered. Ontogeny. — The larvse of the Velellidse were described by Eschscholtz,1 who founded upon them the new genus Rataria. We retain here this genus for those simplest forms of the family, which, at this same stage, develop a corona of sexual siphons around the central siphon, and gonophores from the gastral walls of these gonostyles. The larval forms of Velella, described usually as Rataria, lack the gonostyles, and are better designated as Ratarula. Pagenstecher (55) has given a careful historical and anatomical description of these larval forms. But I suppose that the numerous larvae examined by him, and all figured as Rataria, may be mixed larvae of Velella and Porpita. The regular octoradial pneumatocyst, figured in his Taf. xli. fig. 1 , seems to belong to Porpita, since the same organ in Velella is always more or less amphithect or bilateral. . But no larva of Porpita has the vertical sail or crest shown in his figs. 3-5, &c. ; these belong to Velella. The medusiform gonophores, which bud in clusters from the gonostyles of the Velellidse, and afterwards become detached (Discomitra, Chrysomitra), are described by many authors (mainly by Kolliker, 4, Vogt, 5, Gegenbaur, 7, Leuckart, 8, Huxley, 9, Alex- ander Agassiz, 57, and others (compare on the metagenesis of the Velellidse, p. 39 above). Phylogeny. — The Velellidse have probably descended from the elder forms of Porpitidse (or directly from the Discalidse) ; their peculiar amphithect or bilateral develop- ment is due to the production of the vertical sail as a longitudinal fold of the exumbrella. Synopsis of the Genera of Velellidse. 1. Umbrella elliptical, with integral margin. Pneumatocyst without crest. Corona of tentacles simple, ......... 7. Rataria. 2. Umbrella elliptical or quadrangular, with integral margin. Pneumatocyst with a crest. Corona of tentacles simple, . . . . . .8. Velella. 3. Umbrella quadrangular, with a lobate margin. Pneumatocyst with a crest. Corona of tentacles double or multiple, ....... 9. Armenista. Genus 7. Rataria? Eschscholtz, 1829. Rataria, Esch., System der Acalephen, p. 166. Definition. — Velellidse with an elliptical umbrella and a vertical membranous sail placed in the major axis of the ellipse. Margin of the umbrella not lobate. Pneumato- cyst flatly campanulate or more discoidal, elliptical, without vertical chitinous crest. Submarginal corona of tentacles simple. The genus Rataria is the simplest and most primitive form of the Velellidse, and 1 1, p. 166. 2 i?atoria= A float. EEPORT ON THE SIPHONOPHORJE. 79 comprises those small forms of the family, the vertical sail of which contains no chitinous crest. The pneumatocyst, therefore, is the horizontal elliptical chitinous disc alone ; no vertical skeleton crest is developed upon its surface, as in Velella and Armenista. The soft vertical sail is placed in the major axis of the ellipse, not obliquely in a diagonal. The species of Rataria upon which Eschscholtz founded the genus were probably young larval Velellse, perhaps the same as those of which Bedot has during the last few years given an anatomical description (58-62). Pagenstecher gave in 1863 a very accurate description of Rataria, with historical remarks on the Velellidse in general, and discussed the question whether these forms of Disconectse were only a larval stage of Velella (or perhaps of Porpita) or an independent peculiar genus of this group. As a fact both alternatives are true. Rataria cristata, from the Tropical Atlantic, described in the following pages, and found in the Challenger collection (Station 348), is a VelelHd which produces gonophores in the form of Rataria, and therefore is the representative of an independent genus. The larvse of Velella and Armenista, on the other hand, all pass through a larval stage similar to the former. But no Porpitidse have a similar larval form, since the vertical sail is completely wanting in this family, and is only to be found in the family Velellidse. Regarded from a phylogenetic point of view, Rataria is a necessary intermediate link between the older Porpita and the more modern Velella. Rataria cristata, n. sp. (PL XLIV.). Habitat.— Station 348, Tropical Atlantic ; April 9, 1876 ; lat. 3° 10' N., long. 14° 51' W. Surface. Umbrella (fig. 1, from above ; fig. 2, from below ; fig. 3, half lateral, half apical view; fig. 4, lateral view ; fig. 5, frontal section). — The horizontal disc of the umbrella is elliptical, 4 mm. long and 3 mm. broad in the expanded state. The vertical sail which arises in its sagittal or longitudinal axis, is very contractile, and therefore exhibits very different forms (figs. 3, 4). The vertical transverse section of the umbrella (fig. 5) demonstrates that the greatest part of its volume is occupied by the pneumatocyst (.])/), and the centradenia (uc) which lies between this and the central siphon (sa). Exumbrella (figs. 1, 3, 4). — The superior or apical face of the umbrella is divided into three parts — the vertical sail (velarium), the campanulate part, which includes the pneumatocyst (pneumatophore), and the broad horizontal free border (limbus). Limhus Umbrellm (uu). — The peripheral border of the horizontal disc, which surrounds the campanulate pneumatosaccus, is an elliptical ring of 0'5 mm. in breadth ; its outer edge is densely beset with a series of marginal muciparous glands (us), and on the inside of this glandular corona runs the elliptical marginal canal (fig. 10, cc), into which open the numerous radial canals of the umbrella (ce). 80 THE VOYAGE OF H.M.S. CHALLENGER. Velarium (figs. 3-5, uf). — The vertical sail is a simple fold of the exumbrella, which arises in the longitudinal or sagittal axis of the elliptical disc, along the median line of the campanulate pneumatosaccus. Its form and size are very variable, since it is very contractile ; if highly expanded, it appears nearly rectangular, with a median notch at the central top (fig. 3), and its vertical diameter is greater than the major axis of the horizontal disc ; if strongly contracted it appears much smaller, and has the form of a shallow vertical elliptical disc (fig. 4). At other times it is more crescentic in form, or cordate with a deep constriction in the middle of the top. The velarium is composed of an elastic support, or a thin vertical plate of jelly in its median plane (fig. 5), and a thin mantle-plate of the exumbrella, covering both sides of this fulcrum. This mantle-plate exhibits beyond the exodermal epithelium a double strong muscular layer, composed of two different strata, an outer thinner layer of trans- verse or horizontal muscles, and an inner thicker layer of longitudinal or vertical muscles ; the bundles of the latter are parallel, of equal breadth, about forty or fifty on each side. The parallel vertical bands of equal breadth ascending in the velarium to its top (twenty to thirty in the sagittal diameter) are simple vessels, which arise from the ex- umbrella of the campanulate pneumatosaccus. They are united at the free superior edge of the sail by a marginal sail canal. This runs along the whole free edge of the sail and opens at its base into the two sagittal canals of the exumbrella, which run to the two opposite poles of its major axis. Subumbrella (fig. 2). — The inferior or basal face of the umbrella, beyond the series of marginal glands, consists from without, inwards, of the following parts : — (l) The inferior or subumbral side of the broad mantle border (urn) ; (2) the elliptical corona of tentacles (t) ; (3) the corona of gonostyles (gs) ; and (4) the large central siphon (so). Pneumatocyst (fig. 8, from above; fig. 9, from below; fig. 3,pf, in profile; fig. 5,^/*, in vertical section). — The chitinous thin-walled float, filled with gas, is flatly campanulate with an elliptical quadrilobate outline. Its length (or principal axis) is 2 mm., its breadth or transverse axis 1*5 mm., and its height (or sagittal axis) also 1"5 mm. It is composed of an elliptical central chamber, an inner corona of eight radial chambers, and an outer corona of four to eight concentric elliptical ring-chambers. These latter are divided by two crossed centripetal furrows (a sagittal and a frontal notch) into four rounded lobes (figs. 8, 9). These are systematically arranged in pairs. The left anterior and right posterior lobes are smaller than the right anterior and left posterior. Stigmata (fig. 8, pe). — The convex superior face of the pneumatocyst, which is covered by the exumbrella, exhibits only three stigmata. These lie in a straight line which crosses the major axis of the elliptical disc at a very small angle. The subcentral stigma (_po) lies nearly in the top of the central chamber, and opens on the left side of the base of the sail ; the second lies on the surface of the left anterior quadrant, and REPORT ON THE SIPHONOPHOR^E. 81 the third on the right posterior quadrant of the pneumatocyst, each at the base of a radial chamber (fig. 8, pe' anterior, pe" posterior stigma). Tracheal (fig. 9, pt). — The concave inferior face of the pneumatocyst, which is in close contact with the superior convex face of the centradenia, gives off eight small bunches of tracheas, each composed of two to four. These are simple undivided tubules, which arise from the periphery of the eight radial chambers of the pneumatocyst, and enter into the centradenia, where they end in its exodermal parenchyma. Centradenia (figs. 3, 4, uc, in profile ; fig. 5, uc, in vertical transverse section ; figs. 6 and 7 in horizontal section, fig. 6 through the inferior, fig. 7 through the superior part). — The central gland, or the so-called " liver," has the form of a flat cone, strongly com- pressed from the two lateral sides, or of a thick vertical triangular lamella, which is somewhat sigmoidal, or slightly bent in an S-form. Its upper face is completely covered by the campanulate pneumatocyst ; its lower face is in contact with the gastrobasal plate of the central siphon (fig. 5, st), and the surrounding corona of sexual siphons. The glandular parenchyma of the centradenia is composed in the superior half more of entodermal canals, in the inferior half more of exodermal cells and cnidoblasts. The vascular reticulum of the superior face exhibits a star of eight radial main vessels ; two of these, opposite in the longitudinal axis of the disc, are longer and stronger than the six others, which are disposed symmetrically on both sides of the former, three on the right, three on the left ; the middle of these, or the frontal canal, is the shortest, whilst the two diagonal canals of each side are longer than the latter, shorter than the sagittal canals.1 The numerous smaller vessels, which arise from these eight superior hepatic vessels, form an irregular network, which is rather dense in the superior half of the centradenia, loose in its inferior half ; the axial part of the latter is chiefly composed of exoderm-cells. The descending superficial canals of the centradenia pass over at its basal periphery into four different groups of canals : — (l) The numerous radial canals of the free mantle border (or pallial canals, fig. 10, ce) ; (2) the canals of the tentacles (fig. 5, t) ; (3) the canals of the gonostyles (gs) ; (4) the eight innermost canals which open into the base of the central siphon. The pallial canals (fig. 10, ce), which run nearly horizontally from the periphery of the base of the liver to the edge of the mantle border, are very numerous, and so densely attached one to another, that their intervals are smaller than their lumen. Their distal ends, which open into the annular elliptical marginal canal (cc), are partly simple, partly forked ; usually a simple and a forked vessel regularly alternating. Central Siphon (fig. 2, so ; fig. 3, sa ; fig. 5, sa, in longitudinal section). — The large central polypite is a slender inverted cone, strongly compressed from both sides. Its ellip- tical basal plate is in contact with the basal face of the centradenia, and separated from it by the strong fulcrum of the gastrobasal plate (fig. 5, st). The periphery of this is 1 Compare the figure of Bedot, 59, pL ix. fig. 1. (ZOOL. CHALL. EXP. — PART LXXV1I. — 188-'.) Hbllh 11 82 THE VOYAGE OF H.M.S. CHALLENGER. pierced by eight openings which conduct into the eight primary radial vessels, arising from the original manubrium of the Medusa. The thick wall of the central siphon exhibits eight strong radial longitudinal folds, and its slender proboscis has eight lips around the terminal mouth (figs. 3, 5, so). Gonostyles (figs. 2, 3, 5, gs). — The sexual siphons, sixteen in number, form an elliptical corona around the central siphon and separate it from the tentacular zone and the mantle-border. Their slender cylindrical basal part bears clusters of medusiform gonophores, whilst their dilated spindle-shaped distal part opens by a four-lobed mouth. Tentacles (figs. 1-5, t). — The sixteen tentacles form an outer corona around the inner corona of gonostyles, and alternate regularly with the insertions of the latter. They are simple, cylindrical, very contractile, and beset with two opposite rows of cnidoblasts. Genus 8. Velella,1 Lamarck, 1816. Velella, Lmk., Hist. nat. anim. sans vert., t. ii. p. 481. Definition. — Velellidse with an elliptical or slightly quadrangular (parallelogram - shaped) umbrella, and a vertical triangular sail, placed obliquely in a diagonal line of the ellipse. Margin of the umbrella not lobate. Pneumatocyst discoidal, of the same form as the surrounding umbrella, with a vertical chitinous crest supporting the sail. Sub- marginal corona of tentacles simple. The genus Velella is, next to Physalia, the oldest and best known form of all Siphonophorse, since it is generally distributed in all warmer seas, often very common, and in some countries used as food. Owina; to its striking form and sailing movement it is well known to the fishermen. The Italian naturalists Ferrante Imperato (1599) and Columna (1616) mention it under the names Vela or Velella, Carburius (1757) and Dana (1776) under the names Armenista or Armenistarion. The first good description and figure were given in 1776 by the excellent Swedish naturalist Forskal;2 he named the Mediterranean species Holothuria spirans. Afterwards Lamarck erected the genus Velella with three species (the Mediterranean Velella limhosa, the North Atlantic Velella mutica, and the South Atlantic Velella scaphidia). Esckscholtz3 in his fundamental work described ten different species, and Lesson4 as many as sixteen. But the greater part of these descriptions are worthless and quite insufficient, since they are founded only upon the different coloration and other characters of little value (for example, the different direction of the sail). Two species of Velella only are hitherto completely known ; the Mediterranean Velella spirans, very accurately described by Kolliker5 and by Vogt,6 and the North Atlantic Velella mutica by Alexander Agassiz.7 1 Velella = Diminutive of Vela, sailing-boat. 2 11, Taf. xxvi. fig. K. 3 ^ p 168j Taf. xv. 4 3, p. 563. 6 4, p. 46, Taf. xi. figs. 9-15. u 6, pis. i., ii. 7 57, pis. i.-vi. REPORT ON THE SIPHONOPHOR^E. 83 The collection of the Challenger contains different species of Velella, collected in widely distant parts of the Atlantic and Pacific Oceans. The comparison of these species and of others which I have seen in many museums, has led me to the opinion that about a dozen different " good species," at least " geographical species," may be distinguished, and that these may be placed in two different genera, Velella (sensu stricto) and Armenista.1 The genus Velella (sensu stricto), the type of which is the well-known Velella spirans of the Mediterranean, has an elliptical umbrella with an integral, not lobate margin, and a simple series of tentacles ; it stands nearer to the ancestral form Rataria. The second, more differentiated nd larger genus, Armenista, has a more quadrangular or parallelogram-shaped umbrella, with a broad lobate margin, and a double or multiple series of tentacles. Perhaps even four genera may be distinguished when the numerous local varieties are better known. The following species may be distinguished of the true Velella (sensu stricto) : — (1) Velella spirans, Eschscholtz, 1 (Mediterranean). (2) Velella caurina, Eschscholtz, 1 (North Atlantic, observed by me in 1866 in the Canary Islands). (3) Velella scaphidia, Peron (12), collected by the Challenger in the Guinea Current (Stations 346 to 352). (4) Velella oblonga, Chamisso (21), collected by the Challenger off Volcano Islands, south of Japan (Stations 229 to 237). (5) Velella patella, Brandt (25), collected by the Challenger in the Tropical Pacific (Stations 265 to 279). Nearly related to the latter seem to be Velella pacifica of Eschscholtz and Velella cyanea of Lesson (South Pacific). The other species of Velella described seem to belong to the following genus, Armenista. Genus 9. Armenista? n. gen. Velaria, Hkl., System der Siphonophoren, 1888, p. 31. Definition. — Velellidse, with a quadrangular or parallelogram-shaped (sometimes slightly elliptical) umbrella, and a vertical triangular sail, placed oblicjuely in a diagonal line of the parallelogram. Margin of the umbrella lobate. Pneumatocyst discoidal, of the same form as the surrounding umbrella, with a vertical chitinous crest supporting the sail. Submarginal corona of tentacles double or multiple. The genus Armenista (enumerated in my "System" as Velaria, a name already 1 The name Velaria adopted in my System der Siphonophoren, 1888, p. 31, had previously been employed for a Sponge (Aplysinidae). 2 Armenista or Armenistarium (Carburius, 1757) is the Cephalonian term for sail= Velella. 84 THE VOYAGE OF H.M.S. CHALLENGER. employed for a Sponge) comprises those Velellidse which differ from the true Velella by the broad lobate margin of the umbrella and by the double or multiple corona of tentacles ; usually also by the quadrangular form of the umbrella, which in the former is more elliptical. Although a sharp boundary line between the true Velella (s. str.) and Armenista cannot be drawn, the differences of the extremely divergent forms are very striking ; and when the intermediate forms are neglected, they represent two widely different genera. The elliptical umbrella with integral border of the mantle, and the simple corona of tentacles, brings the true Velella nearer to the ancestral Rataria. On the other hand, the larger Armenista, with the broad lobate border of the mantle of the quadrangular umbrella, and the double or multiple corona of tentacles, is a more recent and more differentiated form. To this genus belong probably the following species : — (1) Armenista sigmoides, n. sp., from the Southern Tropical Atlantic, figured in PI. XLIIL, and taken by the Challenger at Station 346. (2) Armenista mutica, Lamarck, from the North-western Atlantic, very good figures of which were published in 1883 by Alexander Agassiz (57). (3) Armenista antarctica, Eschscholtz (1), and (4) Armenista indica, Eschscholtz (1), from the Indian and Antarctic Oceans, and the Cape of Good Hope. (5) Armenista lata, Chamisso (21), from the Northern Pacific. (6) Armenista lobata, n. sp., from the Southern Pacific. Armenista sigmoides, n. sp. (PI. XLIIL). Velella sigmoides, Hid., 1881, MS. Habitat.— Station 346, Tropical Atlantic; April 6, 1876 ; lat. 2° 42' S:, long. 14° 47' W. Surface. Umbrella (fig. 1, from above; fig. 2, from below; fig. 3, half lateral, half superior view). — The umbrella is nearly rectangular, with rounded edges, about once and a half or twice as long as broad ; the longitudinal or sagittal diameter of the horizontal disc is in the largest specimens 80 to 90 mm., the transverse or frontal diameter 30 to 40 mm. This latter about equals the height of the triangular vertical crest or sail. Exumbrella (fig. 1). — The superior or apical face of the umbrella is divided into the broad lobed mantle-border and the colourless pneumatic disc, both separated by a sharp line which is not parallel to the margin of the border. The vertical diagonal crest divides the exumbrella into two equal halves, an antero-dextral and a postero-sinistral. When we regard the disc in profile from the broad side (its major axis perpendicular to the axis of vision), the crest or sail runs from our proximal and left to our distal and right side ; but seen from behind, from the smaller side (the major axis of the disc REPORT ON THE SIPHONOPHOR^E. 85 parallel to the axis of vision), the vertical sail runs from our proximal and right to our distal and left side (" Velella sinistra," Chamisso). Limbus Umbrella} (fig. 1). — The broad soft border of the horizontal disc is widely prominent over the edge of the pneumatophore, and has four deep marginal incisions. Two opposite ones of these correspond to the diagonal axis of the sail (the macrodia- gonal), and touch the two ends of its base ; the two others correspond to the diagonal axis of the transverse or frontal groove (the brachydiagonal). The border of the mantle is divided by these four marginal incisions into four large lobes, which are disposed in pairs. The two lateral lobes (right and left) are nearly rectangular, and much smaller than the two crescentic sagittal lobes (anterior and posterior). Besides these four deep marginal incisious, which correspond to those of the edge of the pneumatophorous disc, the margin of the mantle-border often exhibits four smaller incisions alternating with the former ; but these are not constant. Subumbrella (fig. 2). — The inferior or basal face of the umbrella exhibits the smooth and lobed mantle-border to the same extent as the upper face. A deep elliptical incision, the tentacular or submarginal furrow (ut), separates it from the elliptical central area of the subumbrella. This area is bisected by the longitudinal axis of the spindle- shaped, long and narrow centradenia and the base of the central siphon. Numerous sexual siphons occupy the greatest part of that subumbrellar central area, whilst a double or triple corona of tentacles surrounds it. Margin of the Sail. — The triangular vertical sail is composed of an inner chitinous crest arising from the pneumatocyst, and of a canaliferous plate of the exumbrella, covering the two sides of the former. This plate is prolonged over the free margin of the crest, and borders it as a soft contractile limb, which is the uppermost part of the original longitudinal mantle-fold. Pneumatocyst (fig. 5, from above; fig. 6, from below; fig. A,p, in vertical section). — The pneumatocyst is composed, as in all species of Velella and Armenista, of two different parts, the horizontal discoidal float filled with air, or the pneumatodisc, and the vertical triangular crest, which is the inner skeleton-plate of the soft sail, perpendicular to the disc, and placed in its greater diagonal. The pneumatodisc, or the horizontal float, is sometimes elliptical or nearly rec- tangular in outline, at other times distinctly S-shaped. It is two and a half times as long as broad ; in the largest specimen 50 to 60 mm. long, 18 to 21 mm. broad. The four characteristic marginal incisions, which are described above of the mantle- border, are marked also in the periphery of the pneumatodisc (fig. 5) ; the two notches of the macrodiagonal touch the two ends of the base of the crest, whilst the two notches of the brachydiagonal approach the two ends of the frontal furrow. The four cmadrants of the disc, which are separated by these four diagonal notches, are disposed in pairs opposite ; the two lateral pairs form an oblique triangle with concave base, and their 86 THE VOYAGE OF H.M.S. CHALLENGER. surface occupies less than half the area of the two opposite sagittal pairs (posterior and anterior) ; these form a much broader oblique triangle with a prominent convex base. The superior or apical face of the disc is slightly convex, the inferior concave ; the convexity is much stronger in younger specimens (fig. 4) than in older (fig. 3). The central chamber of the pneumatocyst (fig. G, ph) exhibits nearly the same form as that figured in Rataria cristata (PI. XLIV. figs. 8, 9). It is elliptical and dis- tinctly octolobate. The deep frontal furrow proceeds near to the central chamber, so that the octoradial ring is bisected by it ; its ventral half is composed of two anterior and two left chambers ; its dorsal half of two posterior and two right chambers. Each of the eight radial chambers possesses a branched trachea, arising from the periphery of its basal side. The apical side of the octolobate ring, however, exhibits three stigmata only, placed near the macrodiagonal, one subcentral, an anterior upon the left ventro-lateral, and a posterior upon the right dorsolateral chamber. The concentric elliptical ring-chambers, which surround the octolobate central part of the float, are sixty to eighty in number in the largest specimens ; those of the central and those of the peripheral part are half as broad as the intermediate chambers placed between them. Each elliptical ring-chamber is deeply bisected by the frontal furrow of the brachy- diagonal. All elliptical ring-chambers communicate with one another by two opposite openings or pneumothyrae, which are placed one on each pole of the sagittal axis (fig. 5,pg). The stigmata, or the external openings of the exumbrellar face of the float, are placed in one straight line, very near the base of the vertical crest, and this line bisects the latter at a very small angle (fig. 5, pe). There are in the largest specimen (besides the central stigmata) about thirty to forty stigmata opposite in pairs ; these are situated, the half in the right ventral quadrant of the disc (to the right of the basis of the crest) ; the other half in the left dorsal quadrant (to the left of the basis of the crest). The tracheae, which arise from the inferior face of the disc, are not numerous. Besides the eight above-mentioned tracheae, which usually arise from the octolobate ring (fig. 6, pt), there is in some specimens another corona of sixteen tracheae, which arise from the sixth or ninth ring, or between these. But in other species there is no regular arrange- ment, a few scattered tracheae arising here and there from one of the middle chamber- rings. In a few specimens I found an oblique double series of tracheae arising along the frontal furrow, on both sides of it. The number, size, disposition, and branching of the tracheae seem to be subject to many variations in this as well as in other species of the Velellidae. Some tracheae are very long (about equal to the diameter of the centradenia), and bear ten to twelve or more irregular and curved branches (fig. 10) ; others are much smaller, and bear only a few short branches. The majority of tracheae run more horizontally beyond the pneumatocyst, and finish in the glandular tissue of the centradenia ; but a few larger tracheae pierce this latter, and enter partly into the exodermal wall of the central siphon, partly in that of the gonostyles, where they end in their proximal half. REPORT ON THE SIPHONOPHORiE. 87 The Centradenia (fig. 4, uc; fig. 7), or the so-called liver, is a lanceolate gland, the upper convex surface of which is flatly conical, and fills up the inferior concave face of the pneuniatocyst (fig. 4, #/). Its lower flat horizontal face forms the roof of the central siphon, and is separated from its base by the gastrobasal plate, a thick structureless fulcrum (fig. 4, st). The periphery of this plate is pierced by sixteen radial main canals, which arise from the lanceolate base of the central siphon in two longitudinal rows, and run along the subumbrella towards its margin ; they give off at their base sixteen ascending liver-canals, which run in the convex upper surface of the centradenia towards its conical apex, where they unite and form a central " liver-star." Very numerous lateral branches, which arise everywhere from these main canals, penetrate into the sobd exodermal parenchyma of the centradenia, and form a very dense network by innumer- able anastomoses. The vessels in the upper and outer (hepatic) part of the central gland are dark brown or black, thickly filled with pigment-granules (fig. 4, uh), except the uppermost ; whilst the canals in the lower and inner (renal) part are white or colourless, and contain numerous small greenish guanin-crystals (fig. 4, un). Central Siplion (fig. 2 ; fig. 4, sa ; fig. 7, sa). — The large central polypite is inversely conical, sharply compressed from both sides, very contractile, and variable in form and size. Its lanceolate roof is nearly half as large as the adjacent basal face of the centra- denia, and separated from it by the structureless fulcrum or the gastrobasal plate. Its muscular wall is very thick, and exhibits in some specimens outside eight or sixteen thick longitudinal ribs, which are prominent inside as deep grooves. The opening of the mouth has often eight triangular lips (fig. 4, so). The periphery of the lanceolate base of the central siphon exhibits sixteen openings, which conduct into the sixteen radial main vessels of the subumbrella ; two opposite of these gastral canals lie nearly in the sagittal axis of the subumbrella (one ventral and one dorsal), and these are much wider and longer than the fourteen other vessels which arise symmetrically from both sides (fig. 7). Canal System (fig. 7). — The reticular system of anastomosing gastro-canals, which arise from the branches of the sixteen above-mentioned radial main canals, is in this, as in other large Velellidae, very complicated. The following principal branches may be distinguished : — (l) The sixteen ascending superficial canals, which pass off from the former immediately at their origin, and end centripetally in the upper face of the centradenia ; (2) very numerous canals which form a dense reticulum in the sub- umbrella ; (3) numerous vessels which arise from the peripheral part of the centradenia, and enter into the gonostyles; (4) a smaller number of vessels, outside the latter, which open into the tentacles ; (5) a corona of radial submarginal vessels which run in the peripheral part of the subumbrella, between the tentacular zone and the margin ; these open into (6) a marginal ring-canal, which runs inside the series of marginal glands ; (7) a coronal canal, or inner submarginal ring-vessel, runs in the furrow between 88 THE VOYAGE OF H.M.S. CHALLENGER. border and tentacular zone ; from this arise (8) the numerous mantle-vessels of the exumbrella, which form a dense network in the latter, and from these arise (9) the mantle-vessels of the vertical crest, which form a dense network of different form on its two sides ; four of these crest-canals are much stronger than the others, two opposite longer, which run along the free upper edge of the crest, and two opposite shorter, which bisect the triangular sides of the crest, and arise near the centre of the frontal groove ; (10) a double series of elegant pinnate sail-canals, which arise from the large superior edge-canal of the crest, and run in the soft sail border towards its free margin ; they are here united by a small outermost canal running in the free edge of the sail. The special form and reticulation of these numerous canals is about the same as figured by Alexander Agassiz ' in Velella (Armenista) mutica. Gonostyles (fig. 2 ; fig. 4, gs ; fig. 8, gs). — The small sexual siphons, or "gonoblastidial polypites," are very numerous (many hundreds in the larger specimens) and densely crowded, occupying the entire broad gonostylar zone between the base of the central siphon and the corona of tentacles. Their upper or proximal half is cylindrical, and covered with bunches of medusiform gonophores (g) ; their lower or distal half is a contractile muscular stomach, beset with sixteen longitudinal rows of cnidonodes ; its proboscis opens below by a very extensile mouth, and this is sometimes distinctly lobed (fig. 8, so). The special form and structure of the gonostyles is the same as figured by Alexander Agassiz 2 in Velella (Armenista) mutica. Tentacles (figs. 1-4, t). — The corona of tentacles, placed in the elliptical groove between the outer edge of the gonostylar zone and the inner edge of the broad mantle- border, is composed of two or three rows of tentacles ; those of the innermost row are the largest, and in the expanded state about as long as the transverse diameter of the umbrella. The tentacles of the outer row, alternating with the former, are only half as long ; and when there is a third outermost row, these are very short. The form and structure of these cylindrical tentacles are the same as in other Velellidge ; they are beset on both sides with two lateral rows of sessile cnidocysts.3 Armenista sigmoides, as described above from the Tropical Atlantic (Station 346), is perhaps identical with a Velella of which I have obtained several specimens from Cape of Good Hope, collected by Dr. W. Bleeck. The short description and the figures which Eschscholtz (1) has given of Velella indica (also occurring at the Cape) are not sufficient to decide the question of the identity of these species. Further accurate examinations and comparative studies are required to enable a better distinction of the different species of Disconectse. 1 57, pis. iv., v. 2 57, pi. ii. 3 Compare Kolliker, 4 ; Vogt, 6 ; Huxley, 9 ; Agassiz, 57. REPORT ON THE SIPHONOPHOR^E. 89 Subclass II. SIPHONANTH^. Order II. CALYCONEOLE, Haeckel, 1888. (Pis. XXVII.-XLII.) Dijihyidx, Eschscholtz, 1829, 1, p. 122. Calycophoridx, Leuckurt, 1854, 8, p. 256. Definition. — Sipkonophorse with one or more nectophores, without pneumatophore, and without palpons. Nectosome always without pneumatocyst, represented only by the nectophores. Siphosorue either a single siphon (Monogastricaa) or a long tubular stem, bearing a series of equidistant siphons, separated by free internodes (Polygastricse). Each siphon with a single tentacle bearing a series of tentilla or lateral branches ; each tentillum with a cnidosac and a simple terminal filament. The order Calyconectse, hitherto usually called Calycophoridae, comprises all Siphono- phorse without pneumatophore. No doubt this large group is very natural ; it differs from all the other Siphonophorse, not only in the complete absence of the important pneumatic apparatus, but also in a number of other peculiarities. The primary larva develops first a nectophore (in the others a pneumatophore). The siphons have always the same structure, and so also the single tentacle, which is attached to each siphon ; it bears a naked cnidosac (without involucrum) of a constant structure. The long tubular stem of the polygastric Calyconectse always bears a single row of ordinate cormidia, separated by naked free intervals. Each cormidium constantly bears a single siphon only, and a single tentacle. Palpons and palpacles never occur in the Calyconectse. History. — Eschscholtz, in his fundamental work,1 divided the order Siphonophorse into three natural families ; the first of these, Diphyidse, differs from the two other (Physophoridse and Velellida?) in the complete absence of any float filled with air. He distinguished six different genera ; three of these (Eudoxia, Erssea, Aglaisma) are monogastric ("with a single suctorial tube") ; the three others (Abyla, Cymba, Diphyes) are polygastric (" with a long digestive tube bearing numerous suctorial tubes "). The oldest known of these six genera was Diphyes, founded by Cuvier in 18 17,2 upon an apparently double animal, or twin-zoophyte, of which Bory had given the first figure in 1804 under the name Biphora bipartite (13) ( = Diphyes dispar, Chamisso). Lesson adopted in 1843 (3, p. 424) the system of Eschscholtz, and divided the family Diphyidae also into Monogastricse and Pol ygas tricse. But he distinguished a greater number of genera and subgenera (partly constituted already in 1827 to 1833 by Quoy and Gaimard, 2 and 20). Lesson collected all the scattered descriptions of Diphyidee 1 System der Acalephen, 1829, p. 122. - Regne animal, edit, i., tome iv. p. 61. (ZOOL. CILALL. EXP. — PART LXXVII. — 1888.) Hhhli 12 90 THE VOYAGE OF H.M.S. CHALLENGER. published up to his time ; but he could not distinguish the natural groups critically. Marked progress in our anatomical knowledge of the Calyconectae was made by the excellent descriptions of different Diphyidae which were published in the third period of our knowledge of Siphonophorse (from 1853 to 1859, 4-10) by Kolliker, Leuckart, Vogt, Gegenbaur, and Huxley. Two of these celebrated zoologists simultaneously and inde- pendently discovered, in the spring of the year 1853, that the monogastric Diphyidae, or the so-called Eudoxise, were the isolated individual groups (or cormidia) of the polygastric Diphyidae, detached from the common stem, and that the former were connected with the latter by a regular metagenesis. Gegenbaur observed in Messina the detached Eudoxise of Abyla pentagona.1 The same observation was made at the same time in Nice by Leuckart, who further demonstrated that the monogastric Eudoxia campanula was the detached sexual zooid of his Diphyes acuminata (5, pp. 41, 69). Leuckart in the next year (8, p. 256) replaced the name Diphyidae by the more con- venient term Calycophoridae, and united in this family the true Diphyidae (with two nectophores, loc. cit., p. 257) and the Hippopodidae (with a biserial nectosome, composed of four or more nectophores, loc. cit., p. 298). The latter were formerly regarded as a separate family of Physophoridae, though they possess no float filled by air. Huxley in his great work (9, 1859) adopted the main group Calycophoridae, and opposed it to all other Siphonophorae or Physophoridae. He gave the first exact descrip- tion of many hitherto incompletely known forms, mainly Abylidae. He was also the first to describe a very remarkable Calycophoridj which possesses only a single permanent nectophore, under the name Sphieronectes kollikeri, and rightly regarded it as the type of a new family, Sphaeronectidae.2 Fifteen years later a very similar species of the same genus was described by Claus under the name Monoplxyes gracilis (70, pi. iv.). He observed its metagenesis and connection with that Eudoxia which Gegenbaur had described in 1854 as Diplophysa inermis.3 The peculiar family represented by these Calycophoridae, the Sphaeronectidae of Huxley, was called by Claus Monophyidae, in opposi- tion to Diphyidae. Following the systematic manuals of recent years, I adopt the term Monophyidae for all those polygastric Calyconectae which possess only a single permanent nectophore, while I restrict the term Diphyidae to those forms which have two permanent nectophores. A third family is formed by the Hippopodidae,4 which possess numerous (at least three or four) nectophores arranged in a biserial nectosome ; they were afterwards named Polyphyidae by Chun (86, p. 12). The Polyphyidae differ from the other Calycophoridae in the lack of bracts. A new group, described in the secpael as Desmophyidae, is intermediate between the Diphyidae and Polyphyidae, having in common with the former the possession of a bract on each eudoxome, with the latter a biserial nectosome, composed of numerous nectophores. 1 7, p. 295 ; 4, p. 78 ; 31, p. lot!. " 9, pp. 29, 50, pi. iii. fig. 4. 3 7, Taf. xvi. fig. 3. * Kolliker, 4. p. 28. REPORT ON THE SIPHONOPHORiE. 91 111 the numerous voyages which I have made during the past twenty-five] Jyears, in order to complete my System der Medusen (compare the Preface to that Monograph), I have found also many opportunities of examining a great number of Calyconectse belonging to all the genera described in the following pages. The richest harvest was reaped during my residence in the Canary Islands (December 1866 to February 1867), where I was able to examine accurately a great number of new or imperfectly known Monophyidse, Diphyidae, and Polyphyidae. I there traced the complete metagenesis of many typical genera and the ontogenetic connection of monogastric and polygastric Calyconectse. Supported by this rich personal experience, I have in the following pages attempted to give a more correct definition of genera and a more natural arrangement than has hitherto been possible. Nectophores. — The Calyconectse differ from all other Siphonophorse in the complete absence of any pneumatophore, so that the nectophores are the only organs of locomotion. The polygastric Calyconectse bear on the top of the long tubular stem either one or two large nectophores (Monophyidse and Diphyidae), rarely a biserial nectosome, which is composed of two opposite rows of nectophores, four to six or more (Desmophyidas and Polyphyidae). The locomotion of the monogastric Calyconectse is effected usually by the subumbrella of the gonophores (Eudoxidaa) ; but in the Ersseidae the first-formed gonophore loses its sexual functions, and acts as a " special nectophore." The larva of the Calyconectse, which develops from the fertilised egg, and which we call Calyconula (PI. XXVII. figs. 8-11), is a bilateral medusoid person, the manubrium of which Hes outside the campanulate umbrella, apparently protruded through a ventral fissure of the latter.1 The " primary nectophore " of this Calyconula often (perhaps always) becomes detached, and replaced by a secondary nectophore, which is often heteromorphous. The number, arrangement, and form of the nectophores in the different Calyconectse are very variable, and serve mainly for the distinction of genera. Eegarding the general form of the nectophores, we may distinguish two different main groups, Sphsero- nectarise and Cyrubonectarise ; the umbrella of the former consists of a very soft jelly, and is subspherical, mitrifurm or reniform, always devoid of distinct edges, with rounded surface. The umbrella of the latter, on the contrary, has always a definite geometrical form, and is polyhedral, either pyramidal or prismatic, with polygonal faces and sharp, often denticulate, edges ; its jelly-substance is rather hard and firm, often cartilaginous. Nectosac and its Canals. — The original form of the muscular subumbrella in the Calyconectse is hemispherical, but usually in correlation with the fundamental form of the nectophores (primary as well as secondary umbrella) it is more or less bilateral, and at the same time quadriradial ; the latter structure (inherited from the ancestral Anthomedusse) is indicated by the constant four radial-canals of the subumbrella ; 1 7, Taf. xvi. fig?. 12-21 ; 85, Taf. vi., vii. ; 86, Taf. xviL figs. 6, 7. 92 THE VOYAGE OF H.M.S. CHALLENGER. and these are always so differentiated, that two opposite lateral vessels (right and left) are symmetrical, different from the two sagittal vessels (shorter ventral and longer dorsal canal). The sagittal plane, which divides the bod)'' of the nectophore into right and left halves, is therefore precisely marked by the two latter opposite canals. Usually both halves are symmetrical, but sometimes more or less asymmetrical. The velum which surrounds the ostium of the muscular nectosac is usually rather broad and strong. The four radial canals are connected above the insertion of the velum by a constant circular canal. Pallial Canals. — Besides the four constant radial canals, there are in many Caly- conectae one or two pallial canals or nutritive vessels of the jelly-substance of the nectophore. An apical pallial canal runs in many Diphyopsidse from the top of the nectosac to the apex of the umbrella. Praya and others have two pallial canals at the ventral side, an ascending and a descending. Number and Arrangement of the Nectophores. — The number of the swimniing- bells, and their arrangement on the top of the stem, are employed by modern authors to divide this order into three families : — Monophyidae, Diphyidae, and Polyphyidae. The Monophyidae (or Sphaeronectidae) possess a single large nectophore on the top of the stem ; this is smooth and without distinct edges in the true Sphaeronectidae, sharp-edged or pyramidal in the Cyrnbonectidae. The Diphyidae, comprising the great majority of genera and species (among the living Calyconectae), have always two large nectophores on the top of the stem. These are opposite, of nearly equal size and similar form, with rounded exumbrella, in the subfamily Prayidae. They are also of nearly equal size and similar form in the subfamily Diphyopsidse ; but in this group the exumbrella is sharp-edged, pyramidal, and the two nectophores are not opposed, but one placed behind or below the other. The remarkable subfamily Abylidse is distinguished by two nectophores of very unequal size and dissimilar form ; the first or anterior being much smaller than the second or posterior. The form of the two very different nectophores becomes in this group extraordinarily strange and complicated. The Polyphyidae (or Hippopodidae) and the closely allied Desmophyidae are dis- tinguished from the other Calyconectae by the possession of a biserial nectosome, similar to that of most Physonectae. There are here at least four to six nectophores (some- times ten to twelve or more), arranged in two opposite series. They have here no definite edges, and are mitriform or reniform in Desmophyes, Hippopodius, and Polypkyes ; whereas they are angular (pentagonal), prismatic, or pyramidal in Vogtia. It may be that all the Calyconectae with rounded nectophores represent a natural suborder of this order (Sphaeronectariae), and all the others (with edged or pyramidal nectophores) another suborder (Cymbonectariae). The first suborder (Sphaeronectariae) comprises the Sphaeroncctidae, Prayidae, Desmophyidae, and Hippopodidae. The second EEPOET ON THE SIPHONOPHOE^E. 93 suborder (Cymbonectaxise) would include the Cyinbonectidae, Diphyopsidae, Abylidae, and Vogtidae. Hydrcecium or Infundibulum (funnel cavity, house-room, Gehauskammer, Trickter- hohle, Stammbehalter). — All polygastric Calyconectse possess a protective cavity, into which the contracted siphosome may retire. This hydrcecium (or infundibular cavity) is always an external space, filled with sea-water and invested by the exoderm. In the Monophyidae it is originally an open groove or fossula on the ventral side of the single nectophore {Monophyes, Cymbouectes, PL XXVII. ). This open groove, or the " hydroecial sulcus," becomes a closed cylindrical or conical canal, by concrescence of the two opposite margins, or by deeper invagination of the exodermal fossula, in another part of the Monophyidae {Sphzeronectes, Muggizea, Cymba, PL XLL). The singular genus Mitrophyes (PL XXVIII.) has no hydrcecium, but it is replaced here by a pouch- like space between the permanent secondary nectophore and a mitre-shaped or scutiform bract, which is the remnant of the reduced primary nectophore. The hydrcecium of the Diphyidae exhibits various degrees of development. In the Prayidae it is an incomplete canal, formed by two opposite ventral grooves of the two nectophores, fitting one into another. Diphyes, Diphyopsis, and the Abylidae possess a conical infundibular cavity at the ventral side of their first or apical nectophore, and this continues into an incomplete hydrcecial canal, formed by a longitudinal groove at the ventral side of the second or basal nectophore. The two margins of this groove are often partly united by concrescence, so as to form a shorter or longer canal. In other cases the two opposite margins of the hydrcecial groove are developed in the form of two broad dentate plates (right and left) which overlap one another. The genus Galeolaria has no hydrcecium, the siphosome hanging freely down between the distal end of the first and the proximal end of the second nectophore. The hydrcecium of the Desmophyidae and Polyphyidae is an infundibular cavity between the two opposite rows of nectophores, almost as in the Prayidae. Somatocyst or Acrocyst (top-cavity of the stem, coryphal cavity ; Saftsack, Saftbe- halter, cs). — The single nectophore of the Monophyidae, and the first or proximal nectophore of the Diphyidae, contains a remarkable cavity, the acrocyst or somatocyst, at the ventral side of its nectosac and at the top of its hycLrcecium. This is the uppermost part of the common trunk included in the jelly-substance of the first nectophore ; it may be com- pared to the apical canal or peduncular canal of the ancestral Medusa, which was con- nected by it with its hydropolyp-parent. The somatocyst is usually spindle-shaped or ovate, at other times subspherical or cylindrical ; its upper or apical end is blind, whilst its lower or basal end passes directly into the small apical central cavity, from which arises the central canal of the stem and the pedicular canal of the nectosac. The narrow cavity of the somatocyst, or the acrocyst-canal, is invested by very large entoderm cells, usually vacuolated and polyhedral owing to mutual compression. Its 94 THE VOYAGE OF H.M.S. CHALLENGER. uppermost part usually includes a smaller or larger oil globule, the oleocyst (co). The physiological function of the somatocyst may be hydrostatic (as a float) and nutritive (as an accumulation of nutritive, strongly refracting albuminous globules). Its morpho- logical nature is explained by the medusome-theory which compares it with the apical canal or original peduncular canal of a Medusa-person. Trunk or Ccenosarc. — The common stem in all polygastric Calyconectse is a long cylindrical and highly contractile tube, very long and thin in the expanded state, short and thick in the contracted state, when it is retracted into the hydroecium. The cormidia are always ordinate, arranged in a single series on the ventral side of the articulated stem ; they are separated by free naked intern odes of equal length. Very rarely (in Polyphyes) the cormidia begin to be scattered. The number of the cormidia is in the smaller cormi ten to twenty (rarely less), usually forty to eighty or more, sometimes several hundreds. In the largest species (mainly of Praya) the expanded stem attains a length of more than a metre. The structure of the stem-wall is that usually found in the Siphonanthse ; the tubular fulcrum (or structureless supporting plate) is invested on its inner side by a thin layer of entodermal circular muscles, on the outside by a strong layer of exodermal longi- tudinal muscles ; these are arranged, as usual, in parallel bundles along the lamellar radial folds of the fulcrum. Cormidia. — The aggregation of different medusoid persons, by which the cormus of the Calyconectse is formed, follows certain simple and regular laws, but is different in the two kinds of cormidia, which we distinguish as Eudoxomes and Ersseomes. The cormidia of the great majority of Calyconectse are Eudoxomes, or in the free inde- pendent state " Eudoxise " or " Diphyozooids " ; each Eudoxome is a twin-group, composed of two medusoid persons, a fertile and a sterile medusome. The sterile medusome is composed of a bract, a siphon, and a tentacle. The fertile medusome is represented originally by a single medusiform gonophore, but afterwards this is often replaced by a cluster of several gonophores. The Ersseomes (or the monogastric generation of Lilyopsis and Diphyopsis) differ from the Eudoxomes in the fact that the primary gonophore loses its sexual manubrium, and is converted into a so-called " special nectophore " ; its sexual function is replaced by a secondary gonophore. The Ersseome, therefore, is composed of three medusoidal persons, a sterile medusome (bract, siphon, and tentacle), a sterile nectophore, and a fertile gonophore. Afterwards the latter is often replaced by a cluster of several accessory gonophores. The sessile gonophores of the Eudoxomes and Ersseomes attain sexual maturity, whilst attached to the trunk, in Mitrophyes and Cymbonectes among the Monophyidse, Praya and Gaholaria among the Diphyidoe, probably in all Desmophyidse and Polyphyidse. This is not the case in the majority of Monophyidse and Diphyidse. Here the cormidia become detached from the common stem before reaching maturity, and swim freely REPORT ON THE SIPHONOPHOR^E. 95 about under the individual form of self-subsistent monogastric Calyconectae (Eudoxia and Erssea, Families IV. and V.). Bractese or Hydrophyllia (Protective persons or shields, Protecta, Pkyllozooids — " Deckstiicke, Decksckuppen " of German authors). — Three families of polygastric Calyconectae, tke Monopkykke, Diphyidae, and Desmophyidae, possess constantly a single bract on each cormidium ; it is wanting only in the fourth family, Polyphyidse, where it kas been lost by reduction. Tke single bract of eack cormidium is tke reduced umbrella of tke Medusa-person, tke manubrium of which is the single siphon of the former. This is very obvious in Praya, Ccdpe, and some other genera, where the bract still possesses four radial canals. Usually some of these canals are lost, or they have disappeared altogetker. Eack bract kas a convex exumbrella and a concave subumbrella, both separated by the basal margin of tke umbrella. Tke form of the bract is very various, and characteristic of the single genera of monogastric Calyconectae ; usually it corresponds more or less to tke form of tke first nectopkore. It is hemispherical, mitriform or subspherical, with a smooth exumbrella, in the polygastric Spkaeronectidae, Prayidae, and Desmophyidae. The bract is pyramidal, spathiform or conical, with sharply edged exumbrella, a ventral fissure, and a pointed apex in tke Cynibonectidae and Diphyopsidae ; it is prismatic with polygonal faces and sharp edges in the Abylidae. The bracteal cavity corresponds with the subumbrellar cavity of the ancestral Medusa, but has lost its important muscle-plate ; it embraees the siphon, the single tentacle, and the gonophore ; the last is placed at the ventral side of the siphon, the tentacle at its dorsal side. The nutritive canals of all the organs unite in the top or the centre of the bracteal cavity, where it communicates also with the central canal of the common stem by a short bracteal canal. From tke same point arises also tke pkyllocyst, Pliyllocyst. — Tke apical cavity or corypkal cavity of tke bract, which we call shortly " pkyllocyst," corresponds to tke acrocyst or somatocyst of tke nectopkore. Its form and tke number and disposition of its apophyses are often very characteristic of the individual genera of Calyconectae. Usually tke pkyllocyst is an ovate or spindle-skaped sacculus of tke same structure as tke acrocyst, filled witk large polyhedral vacuolated cells, and often containing also an oleophore or an apical oil-globule. It arises usually more or less vertically from the top of the subumbrellar cavity, and projects into the thick jelly- substance of tke bract. From its base arise sometimes four radial canals, wkick correspond to tke four original subumbrellar radial canals of tke nectopkorcs, in Praya (PI. XXXII. figs. 8, 9) and in Calpe (PI. XL. figs. 14-18). These are so arranged that two paired canals lie on both sides of tke bilateral bract (rigkt and left), and two odd in tke sagittal plane (dorsal and ventral). Tke majority of tke Calyconectae do not now possess tke four original canals. Cyniba and Abyla have only two lateral canals; Bassia a single basal canal, arising from tke base of the pliyllocyst ; often they are entirely lost (Diphyes, Monophyes. &c). 96 THE VOYAGE OF H.M.S. CHALLENGER. The genus Mitrophyes among the Monophyidse (PI. XXVIII.) differs from all other Calyconectse in the remarkable peculiarity that the convex exumbrella of the single nectophore is covered by a large concave scutiform bract, and that the contracted siphosome is retracted into the space between the former and the latter. We may assume that here the primary nectophore of the larva, instead of being lost, is converted into a permanent bract. Siphons or Polypites. — Each cormidium of the Calyconectse possesses constantly only a single siphon, and this is the manubrium of the medusome, the modified umbrella of which is the bract. A single long tentacle is always attached to the base of the siphon, between its dorsal side and the concave bract- wall. When the cormidium is detached from the common stem, and represents an independent free Eudoxia or Erssea, it is therefore called a " Monogastric Calyconecta." The siphons of all Calyconectae are of similar form, of moderate or small size, very contractile. The four segments or parts of each siphon are usually very distinct, viz., (1) a short peduncle to which is attached the tentacle ; (2) a hemispherical or subspherical basigaster with a thick wall ; (3) an ovate or ellipsoidal stomach ; and (4) a very mobile proboscis with the terminal mouth. (Compare PI. XXVII. fig. 8, &c.) The pedicle of the siphon (sp) is usually very short, sometimes rudimentary, and connects as a narrow cylindrical canal the gastral cavity of the former with the common alimentary cavity of the stem. The tentacle arises from it sometimes more proximally, at other times more distally ; sometimes even more from the following part. The basigaster (sb), or the bulbous basal portion of the siphon, is usually subspherical or ellipsoidal, and distinguished by a much-thickened exodermal wall ; this is filled with innumerable densely crowded nematocysts. Its outer and inner surfaces bear a vibratile epithelium. Its small ovate cavity is separated from the stomach by a pyloric valve, which can be closed completely. The stomach (sin) is ovate or ellipsoidal, with a thin, non-ciliated exoderm ; its entoderm, however, is very thick, glandular, and often exhibits eight or sixteen hepatic stripes (often of a yellow, red, or orange colour), sometimes instead of these hepatic villi ; some of the entoderm cells are usually vacuolate. The proboscis (sr) or the distal portion of the siphon, is very contractile and extensile, with a strong muscular plate ; its free distal end opens through the mouth. The very mobile mouth may be expanded in the form of a suctorial disc, circular, or polygonal ; it is often octolobate (more rarely with four, twelve, or sixteen lobes). Tentacles. — The capturing filaments or tentacles of the Calyconectse exhibit in all members of this order essentially the same structure, and are very uniform, compared with the manifold and richly varied forms which we find in the similar tentacles of the following order — Physonectse. In all Calyconectse, without exception, each siphon bears at its base a single, tubular, very long and extensile tentacle, which is beset with a single REPORT ON THE SIPHONOPHOR^E. 07 series of lateral branches or tentilla. Each tentillum is a thinner cylindrical tubule, and is constantly composed of three different parts — (1) a thin pedicle or proximal portion (tp), (2) an inflated cnidosac as a dilated middle part (tk), and (3) a thin distal portion, the terminal filament ((/'). The closed distal end of the latter is sometimes vesicular; so also the distal end of the pedicle is sometimes club-shaped. The structure of the thin cylindrical pedicle and of the long terminal filament is simple and always the same, whilst the cnidosac, placed between them, exhibits a complicated, and more or less varied structure (compare PL XXVII. fig. 7 ; PI. XXVIII. fig. 8 ; PI. XXXII. figs. 12-14 ; PI. XXXIV. fig. 18 ; PI. XXXVI. fig. 26 ; PI. XXXVIII. fig. 16 ; PI. XL. figs. 19, 20). The cnidosac (or sacculus) of all Calyconectas is originally nothing more than an inflated dilatation of the middle part of the simple cylindrical tubular tentillum. This dilatation seems to be produced by the stronger development of larger cnidocysts on one side of its middle part. This side, which contains the so-called " cnido-battery," is the convex dorsal side of the cnidosac, whilst the opposite ventral side is usually more or less coneave ; it contains two parallel elastic bands, which seem to form together a noose at the distal end of the cnidosac, the so-called " cnido-band or angle-band." The whole cnidosac, therefore, has a bilateral form, usually more or less ovate, pyriform, or kidney- shaped. In most Calyconectse it is more or less compressed from both sides, so that the sagittal axis is larger than the frontal, but smaller than the principal axis. The differentiated cnidocysts, or thread-cells, which are crowded in great numbers in the '" cnido-battery " at the convex dorsal side of the cnidosac, occur generally (probably in all Calyconectte) in three different forms — (1) very numerous, small, and palisade- shaped (paliformes) ; (2) few, large, and sabre-shaped (ensiformes) ; (3) small and pear- shaped, in a distal group (pyriformes). These are so arranged that several parallel rows of paliform or medial cnidocysts occupy the whole convex dorsal side of the cnidosac (Jem), two bilateral groups of a few large ensiform cnidocysts are placed at each side of its proximal basal part (kg), and an odd distal group of small pyriform cnidocysts (kp) lies at its distal end, at the base of the terminal filament (tf). The small palisade-shaped or medial thread-cells (Cnidocysts paliformis, Am) are always very numerous, usually some hundreds in number, and comprise far the greatest part of the crescentic cnido-battery, occupying the dorsal half of the sacculus. They are arranged so regularly that they form together a certain number (usually four, six, or eight) of parallel longitudinal columns, each composed of a single series of numerous (twenty to fifty, or more) cnidocysts. All the latter are cylindrical or spindle-shaped bacilli, of similar form and equal size, three to six times as long as broad, sometimes straight, at other times slightly curved. Their axis is perpendicular to that of the sacculus. The convex dorsal surface of the sacculus, therefore, appears elegantly panelled or facetted, the outer or abaxial ends of the paliform cnidocysts forming regular transverse rows (crossing the longitudinal rows). (ZOOL. CHALL. EXP. PART LXXVII. — 1888.) Hhllll 13 98 THE VOYAGE OF H.M.S. CHALLENGER. The large sabre-shaped or lateral thread-cells (Cnidocysts ensiformes, kg) form constantly two small lateral groups in the basal or proximal part of the sacculus, covering both the right and left sides of the base, of the cnido-battery. Their number and form are often characteristic of the individual species of Calyconectae. Thus, for example, Mitrophyes peltifera (PL XXVIII. fig. 8) has only two, Oymbonectes huxleyi (PI. XXVII. fig. 7) three, Diphyopsis compressa (PI. XXXIV. fig. 18) six large lateral thread-cells on each side of the base of the sac ; their number is larger in the Abylidae (usually eight to twelve), and especially in the Prayidae (sixteen to twenty, or more, on each side). Their form is usually slender, spindle-shaped or sabre-shaped, four to eight times as long as broad, straight, sometimes slightly curved ; but sometimes they are more rounded, ovate (as in the smallest Calyconectse). They are always placed in a single longitudinal row, parallel to one another, and also (more or less) to the axis of the sacculus. Usually the axis of the ensiform cnidocysts is so directed obliquely that the dorsal end is more proximal and medial, the ventral end more distal and lateral. The small pear-shaped or distal thread- cells (Cnidocystse pyriformes, hp) always form an odd group at the distal end of the cnidosac and touch the base of the terminal filament (tf). Their number is very variable, usually between twenty and sixty. The group formed by these pyriform cnidocysts has usually the form of a rounded cap, covering the distal end of the cnido-battery, and is evidently sensitive to a remarkable degree, since long cnidocils arise from these thread-cells. Sometimes the group is tri- lobate, with an odd middle and two lateral lobes ; and in some species it is even divided into three separate parts, an odd medial group being separated from two paired lateral groups, as in Praya (PI. XXXII. figs. 12-14) and in Bassia (PI. XXXVIII. fig. 16). The elastic angle-band, composed of two very long linear and parallel ribbands, is closely coiled up spirally in the thin-walled ventral pouch of the closed cnidosac. But when this becomes opened (by rupture of the thin ventral wall), then the angle-band is expanded to a great length, often folded in a zigzag. Its proximal end remains in connection with the pedicle (tp), its distal end with the terminal filament (tf). The cnido-battery becomes hung out, and is freely prominent, and its distal end only re- maining in connection with the proximal part of the terminal filament and its junction with the elastic angle-band (PI. XL. fig. 20). Gonophores. — The sexual persons of the Calyconectse are always quadriradial Medusae, with a well-developed umbrella and a manubrium, in the thickened wall of which the sexual cells are produced from the exoderm. The cavity of the manubrium has, however, no mouth opening, and the margin of the umbrella bears no tentacles. Originally each cormidium possesses only one gonophore, and in many Calyconectse never more than a single sexual medusome is attached to the siphon. But when this gonophore is mature, it usually detaches itself from the cormidium (Eudoxia) and may be replaced by a secondary or vicarious gonophore. In many species (mainly of Abylidae) REPORT ON THE SIPHONOPHORiE. 99 two or three gonophores are developed simultaneously in a single cormidium, and in some genera (IAlyopsis, Desmophyes) a bunch of numerous clustered gonophores ; then usually their umbrella is more or less rudimentary, whilst still a " special nectophore " is developed as a swimming organ. Each single gonophore is always gonochoristic, either male or female. Those cor- midia, which bear two or more gonophores, are usually also gonochoristic (diclinic) ; but in some cases they are hermaphrodite, one of the gonophores being female, the other male (monoclinic cormidia). The corms of the polygastric Calyconectse are usually monoecious and bear cormidia of both sexes, sometimes regularly alternating (as in Cymbonectes, PL XXVIL, and Desmophyes, PL XXX.). At other times the distal (older) cormidia are males, and the proximal (younger) cormidia females. A few genera of polygastric Calyconectse are dioecious, each cormus bearing only gonophores of one sex, either male or female (Mitrophyes, Galeolaria). But the question of the relation of these different forms of sexual differentiation requires a further examination. Umbrella of the Gonophores. — The calyx or umbrella of the sexual persons has in all Calyconectse the same structure as in a common simple quadriradial Hydromedusa. Its form is very rarely quite regular, hemispherical (as in Diplophysa and some Poly- phyidae) ; usually it is more or less bilaterally symmetrical, in adaptation to its place inside the bracteal cavity, at the ventral side of the siphon ; sometimes it is strongly asymmetrical, as in Amphiroa (PL XXXVI. figs. 12-25) ; here two gonophores are developed simultaneously on each side of the siphon (right and left) ; each corre- sponding to an antimere, so that the two together form a symmetrical pair. The Exumbrella of the Gonophores corresponds in its general form to a certain degree with that of the nectophores. It' is therefore evenly convex and smooth, without pro- minent edges, in most Sphseronectarise (Sphseronectidse, Prayidse, Hippopodidse). On the other side, it is quadrangular, pyramidal, or prismatic, in most Cymbonectarise (Cymbonectidaa, Diphyopsidse, Abylidae). Usually in the latter the four edges of the exumbrella are sharp and prominent, often elegantly denticulate, and wing-like dilated in the distal part ; their distal ends are often prominent as four strong triangular teeth over the basal ostium of the umbrella. Often two paired dorsal edges and their terminal teeth are much more developed than the opposite two ventral edges. More rarely a single wing is stronger than the three others, or even five wings may be developed instead of four (compare Amphiroa, PL XXXVI.). The uppermost part of the exum- brella is usually prolonged into an apical horn, or a hook-shaped apophysis, which serves for its insertion into the bracteal cavity. The Subumbrella of the Gonophores has always a well-developed layer of ring-muscles (iv), and, on its distal prolongation, a circular velum (v) which closes the peripheral part of the opening of the subumbrellar cavity. Four radial canals (cr) always run in the subumbrella from its apex, diverging to its basal ostium, where they are united, above 100 THE VOYAGE OF H.M.S. CHALLENGER. the insertion of the velum, by a ring-canal (cc). From their apical junction arises a peduncular canal (cp) which runs through the pedicle of the gonophore (or the apical horn) to its insertion into the bracteal cavity and opens here into the phyllocyst. Gonads. — The gonad, or sexual gland of each gonophore, is represented by the manubrium, which depends from the apex of the subumbrella into its cavity. In the mature state it usually fills about the apical half of that cavity, but in many cases the entire cavity, and sometimes by further growth it becomes much larger than the latter, ami is widely prominent through its distal opening (for example, in Lilyopsis, Des- mophyes, Vogtia, PI. XXX.). Each gonad is a simple, cylindrical, ovate, or spindle- shaped sac, and contains a central cavity, which is closed at the distal end, whilst it opens at the proximal end into the peduncular canal. The thick wall of the cavity consists of three different layers, outside a covering exodermal epithelium, and inside a vibratile entodermal epithelium, which includes the cavity (spadix) ; between these is a layer of sexual cells, which are originally derived from the exoderm. They produce a number of large ovules (usually between twenty and forty) in the female, and in- numerable small zoosperms in the male gonophores. The ovaria or female gonads are usually somewhat rounded, ovate or club-shaped, colourless (PI. XXVIII. fig. 4 ; PI. XXXIV. figs. 13, 14, &c). The spermaria or male gonads are rather elongated, cylindrical, or fusiform, often vividly coloured (yellow, orange, red) (PI. XXVIII. fig. 6 ; PI. XXXIV. figs. 11, 12, &c). Ontogeny. — The development of the Calyconectse from the fertilised egg has hitherto been very little known. The first observations were made in the spring of 1853 by Gegenbaur in Messina.1 He observed the segmentation of the egg of Diphyes sicboldii, and the development from it of a larva, which is a peculiarly modified medusome, composed of a simple nectophore and a cylindrical sac-shaped larval body, which is attached externally to the ventral side of the nectophore. In my opinion the mouthless larval body is the original siphon, protruded through a ventral fissure of the nectophore. From its base the primary tentacle arises afterwards. I call this larva Calyconula. A similar Calyconula is developed from the egg of Galeolaria aurantiaca ( = Epibulia aurantiaca), which Metschnikoff described in 1874.2 The Calyconula of Hippopodius gleba, described by the same author,3 exhibits still more distinctly the dislocation of the siphon, the axis of which is perpendicular to that of its nectophore, in the subumbrellar cavity of which it was originally placed. The remnant of the ventral fissure of the bilateral umbrella is yet partly visible. The Calyconula of a Monophyid (Muggiwa kochii), and its development from the egg as well as its metamorphosis, were described in 1882 by Chun.4 This larva developed 1 7, p. 332, Taf. xvi. figs. 12-21. 2 85, p. 39, Taf. vi., vii. 3 Loc. cil, p. 46, Taf. xi. figs. 5-S. 4 86, p. 9, Taf. xvii. figs. 6, T. REPORT ON THE SIPHONOPHORvE. 101 directly into Cucubalus eschscholtzii ( = Eudoxia eschscholtzii), the monogastric generation of Muggisea kochii. Scarcely different from this is the ontogeny of another Monophyid (Cymbonectes huxleyi (PI. XXVII. figs. 8-11), which I observed in December 1881 during my residence at Belligemma, in Ceylon. The Calyconula here presents distinctly the character of a bilateral Medusa, through the ventral fissure of which the siphon was protruded ; its distal end opens through the mouth rather early. Its similarity to the young Eudoxia of the same species is interesting. Metamorphosis. — All Calyconeetse seem to undergo a metamorphosis, since the gastrula, developed from the fertilised egg, develops into a larva differing more or less considerably from the adult state. But the metamorphosis of the larva is very little known, and has been observed in a few species only. Chun, who has observed accurately the metamorphosis of some Monophyidse and Diphyidse (86-88), holds the opinion that the primary umbrella of the medusiform larva is always lost and replaced by a hetero- morphous secondary umbrella. More extended researches are required to prove whether this supposition is generally true. Metagenesis. — The majority of Calyconeetse are subject to a regular metagenesis, two different generations alternating regularly, as in the majority of the Anthomedusaa. The first generation is a monogastric and sexually developed cormidium — Eudoxia or Erssea. From its fertilised egg arises a larva, which is transformed by metamorphosis into the second generation — Monophyid or Diphyid. Its body is a polygastric corm, the tubular stem of which produces numerous cormidia by budding asexually. Each cormidium, provided with a single siphon, afterwards becomes detached from the stem, and maturing as a Eudoxid or Ersasid returns to the first generation. Hypogenesis. — The minority of Calyconeetse develop by hypogenesis (not by metagenesis) ; the cormidia arrive at full sexual maturity whilst sessile on the common stem, and are not detached from the latter. There is here, therefore, no free and independent monogastric generation ; neither free Eudoxidse nor Ersaeidse arise. The ontogeny of these polygastric corms is in a strict sense a strophogenesis.1 The Calyconectaa which are developed by hypogenesis are the following : — All the Polyphyidae and Desmophyidaj, some Diphyidse (Galeolaria, Lilyopsis, and Praya ?) and several Monophyidse (Monophyes partly ?, Mitrophyes, and Cymbonectes). All the other Calyconectae develop by metagenesis. There seem to be, however, some inter- mediate forms of ontogeny, in some species the male gonophores becoming detached, whilst the females remain attached to the stem, or inversely. These relations, as well as the whole ontogeny of the Calyconeetse, require further accurate examination. 1 On the difference between metagenesis and strophogenesis, see my Generelle Morphologie, 1866, Bd. ii. pp. 104-109, and on Hypogenesis, op. cit., p. 99. 102 THE VOYAGE OF H.M.S. CHALLENGES. Metagenesis Calyconectarum. Synopsis of the alternation of generations in the genera of monogastric and polygastric Calyconectse (compare the descriptions of the individual genera and species in this Report). I. Monogastric Generation. Diplophysa inermis, Ggbr., Mediterranean. Diplophysa kollikeri, HkL, Tropical Pacific, Indian Ocean. ? Eudoxella didyma, Hkl., Atlantic. ? Eudoxella galea, Hkl., North Atlantic. Cucubalus eschscholtzii, Huxley, Mediterranean. Cucubalus pyramidalis, Will, North Atlantic. ? Cucubalus cordiformis, Quoy, Tropical Pacific. Cuboides vitreus, Quoy, Mediterranean. Cuboides erysfallus, Hkl., North Atlantic. Cuboides nacella, Hkl., Indian Ocean. Cuboides vogtii, Hkl., Tropical Pacific. Cucullus campanula, Hkl., North Mediterranean. Cucullus gegenbauri, Hkl. ( = Eudoxia messanensis, Ggbr.), South Mediterranean. Cucullus subtilis, Hkl., Mediterranean. ? Cucullus elongatus, Hkl., North Atlantic. Cucullus lessoni, Lesson, Pacific. Cucullus gracilis, Hkl., Indian Ocean. Amphiroa trigona, Hkl, Mediterranean. Amphiroa carina, Hkl., Tropical Atlantic. ? Amphiroa alata, Huxley, Indian and Pacific Ocean. I Amphiroa angulata, Huxley, South Pacific. Sphenoides tetragona, Hkl., Indian Ocean. Sphenoides obeliscus, HkL, North Atlantic. ? Sphenoides perforata, Hkl., Tropical Atlantic. Sphenoides australis, Huxley, South Pacific. Aglaisma eschscholtzii, Huxley, Mediterranean. Agldisma gegenbauri, Hid., Tropical and Subtropical Atlantic. ? Aglaisma elongata, Huxley, Tropical Pacific. 1 Erssta gaimardi, Esch, Mediterranean. Erssea compressa, Hkl., Tropical Atlantic. Erssea dispar, Hkl., Tropical Pacific. II. Polygastric Generation. Spheerunectes gracilis, Hkl., Mediterranean. Sphwonedes kollikeri, Huxley, Tropical Pacific, Indian Ocean. ? Praya eymbiformis, Leuck., Atlantic. ? Praya galea, Hkl., North Atlantic. MuggUea kochii, Chun, Mediterranean. Muggixa pyramidalis, Busch, North Atlantic. ? Muggista chamissonis, Hkl., Tropical Pacific. Cymba enneagonum, Esch., Mediterranean. Cymba crystallus, Hkl., North Atlantic. Cymba nacella, Lesson, Indian Ocean. Cymba vogtii, Hkl.. Tropical Pacific. . Diphyes acuminata, Leuck., North Mediterranean. Diphyes sieboldii, Koll. ( = Diphyes gracilis, Ggbr.), South Mediterranean. Diphyes subtilis, Chun, Mediterranean. ? Diphyes elongata, Hyndman, North Atlantic. Diphyes appendicidata, Esch., Pacific. Diphyes gracilis, Hkl., Indian Ocean. Abyla trigona, Quoy, Mediterranean. Abyla carina, Hkl., Tropical Atlantic. ? Abyla alata, Hkl., Indian and Pacific Ocean. f Abyla. leuckarti, Huxley, South Pacific. Bassia tetragona, Hkl., Indian Ocean. Bassia obeliscus, Hkl., North Atlantic. ? Bassia perforata, L. Ag., Tropical Atlantic. Bassia quadrilatera, Quoy, South Pacific. Calpe pentagona, Quoy, Mediterranean. Calpe gegenbauri, Hkl., Tropical and Subtropical Atlantic. ? Calpe huxleyi, Hkl., Tropical Pacific. ? Diphyopsis campcmuliferd, Hkl., Mediterranean. Diphyopsis compressa, Hkl., Tropical Atlantic. Diphyopsis dispar, Hkl., Tropical Pacific. ? Lilaa medusina, Hkl., Indian Ocean. ? Lilyopsis medusina, Hkl., Indian Ocean. REPORT ON THE SIPHONOPHOR^E. 10J Synopsis of the Families of Calyconectse. I. Suborder Calyconect-e monogastric^:. Cormus represented by a siDgle corinidium, with a single siphon and a single tentacle. Cormidium composed of two medusomes,a sterile and a fertile, without special nectophore, Cormidium composed of three medusomes, a sterile, a fertile, and a special nectophore, II. Suborder Calyconect^: polygastric/e. Cormus represented by a long tubular stem, which bears < numerous ordinate cormidia, separated by free internodia; each cormidium with a siphon and a tentacle. One or two nectophores I A single nectophore, on the top of the < stem. ( Two nectophores, . A biseiial nectosome on the top of the stem, composed of four to ei"ht or more necto- phores, pairs. opposite in Each cormidium with a bract, . . Cormidia without bracts, . 4. Eudoxida?. 5. Ersseidse. 6. Monophyida?. 7. Diphyidse. 8. Desmophyida?. 9. Polyphyids. Family IV. Eudoxida, Haeckel, U Eudoxida;, Hkl., System der Siphonophoren, 95, p. 32. Definition. — Calyconectse monogastricse, representing a single cormidium, which is composed originally of two persons ; a sterile medusome (siphon with tentacle and bract) and a fertile medusiform gonophore (male or female). The family Eudoxida? comprises those monogastric Calyconectse which present in the fully developed and sexually .mature state only a single cormidium, composed of two different medusomes, a sterile and a fertile. The sterile medusome is a bilateral medusoid person with three essential and constant organs, a bract (umbrella), a siphon (manubrium), and a tentacle (capturing filament). The fertile medusome is a gonophore with umbrella and manubrium, but without tentacles ; the sexual cells are produced in the wall of the mouth less manubrium. Eschscholtz (1, p. 124) in his System of the Diphyidaa, distinguished first two main groups in this family — I. Monogastricse (" with a single suctorial tube "), Eudoxia, Ers&a, Aglaisma ; and II. Polygastricee (" with a long digestive tube bearing numerous suctorial tubes or lateral branches"), Abyla, Cymba, Dij)hyes. Lesson (3, p. 437), adopting the division of Eschscholtz, and collecting the descriptions of other observers (mainly Quoy and Gaimard, 2 and 20), described a greater number of "Diphyidaj monogastricae," with not less than nine genera (Joe. cit., pp. 453-462). Some other species were described by Will (65) and Busch (67). The true nature of the monogastric Diphyidse (called usually Eudoxise. sensu ampliori) was not recognised before 1853. In the spring of that year Gegenbaur discovered that 104 THE VOYAGE OF H.M.S. CHALLENGER. the cormidia of Abyla penfagona, detached from the common stem, live independently and develop into a sexual Eudoxia.1 Leuckart, who had made the same observation simultaneously, demonstrated further that the monogastric Eudoxia campanula is nothing more than the detached cormidium of the polygastric Diphyes acuminata (5, pp. 41, 69 ; 8, pp. 268, 277). Huxley, in his excellent description of Diphyidse, separated the monogastric forms or Eudoxias under the title " Diphyozooids." 2 He pointed out, with full reason, that it is necessary on practical grounds to retain generic and specific names for the single Diphyozooids, though they may possibly be only detached and independent portions of " Polygastric Diphyidse." He gave a full anatomical description of many new or imper- fectly known Eudoxise, and comparing their structure with that of the sessile cormidia of corresponding polygastric Diphyidse, he suspected an ontogenetic connection between these two forms (op. cit., 1859). During my residence in the Canary Island, Lanzerote (in the winter of 1866-67), I had occasion to examine nearly all the interesting genera of Diphyozooids which Huxley had accurately described. I was able to confirm most of his suppositions concerning their connection with certain polygastric Diphyidse, observing directly the development and detachment of the former from the latter (compare above, pp. 101, 102). The metagenesis which connects the monogastric and polygastric Diphyidse is usually combined with a peculiar metamorphosis, some interesting cases of which have recently been described by Chun (86-88). The detached and independent cormidia of Calyconectse — of the " Diphyozooids " — arise in two different main-forms, Eudoxia and Erssea, which we regard here as repre- senting two different families, Eudoxidse and Ersaeidas. Each Eudoxia is composed of two medusoid persons, a sterile medusome (bract with siphon and tentacle) and a fertile medusome (gonophore). Erssea differs from Eudoxia in the possession of a sterile " special nectophore," and is therefore composed of three medusomes. The sterile medusome has in all Eudoxidse the same characteristic composition of three essential parts — a bract (covering scale or hydrophyllium), a siphon placed in the dorsal part of the bracteal cavity, and a tentacle attached to the base of the siphon. The fertile medusome, or the gonophore, occupies the ventral part of the bracteal cavity. Bract or Hydrophyllium. — The protectum or bract of each Eudoxia ("Deckstiick or Deckblatt " of German authors) is the modified umbrella of the sterile medusome. This is very obvious in the genera Eudoxella (PL XXXII.) and Aglaisma (PI. XL.), where the four radial canals of the subumbrella are preserved by heredity, whilst its muscle-plate is lost by adaptation ; furthermore, the jelly cap is strongly developed, forming a thick and firm protecting shield or cap (" covering scale "). Regarding the characteristic form of the bract, I distinguish two subfamilies among 1 4, p. 78 ; 7, p. 205, Taf. xvi. figs. 1, 2. - 9, pp. 57-66, pis. iii.-v. REPORT ON THE SIPHOTSTOPHOR^E. 105 the Eudoxidaa. The first group, Diplophysidse, possess a smooth umbrella with rounded surface, never prismatic; hemispherical or mitriform in Diplophysa and Eudoxella, conical or spathiform in Cucubcdus and Cucullus. The second subfamily, Aglaismidae, has a poly- hedral or prismatic umbrella with angular surface ; it is more or less cuboidal in Cuboides and Af/laisma, wedge-shaped or similar to a prompter's box in Sphenoides and Amphiroa. The cuboidal form (Pis. XL., XLII.) is of special interest as a reminiscence of the quadri- radial structure in the orisjmal umbrella of the ancestral Medusa. Bracteal Cavity. — Whilst the superior or proximal face of the bract is convex, and corresponds to the exumbrella of the Medusa, its inferior or distal face is concave and comparable to the subumbrella. In the fundus of this cavity hangs the siphon, like the manubrium or gastral tube of the Medusa. Its point of insertion is usually dislocated towards the dorsal side. The single tentacle, which arises from the base of the siphon, is placed between this and the dorsal wall of the subumbrellar cavity. The greater ventral part of the cavity is occupied by the gonophore. The form of the bracteal cavity has sometimes preserved the original hemispherical form of the subumbrella ; but usually it is more campanulate or conical, and often at the same time bilateral, more rarely asymmetrical. Its basal opening, or the bracteal mouth, is usually oblique, more or less bevelled off, sometimes armed with prominent teeth. Phyllocyst.- — The central part of the subumbrella, where the siphon is inserted, con- tains in each Eudoxia a caecal diverticulum of the entoderm, which is in direct communi- cation with the basal part of the siphon, and in the young sessile Eudoxia with the central canal of the stem. This phyllocyst (bursa centralis bractese) is comparable to the apical canal, which in certain Medusae (Codonidae, &c.) ascends vertically from the base of the manubrium, and ends blindly in the jelly of the umbrella. The cavity of the phyllocyst is usually small, lined by large clear vacuolated entoderm- cells, which are polyhedral from mutual pressure. The apical part of the phyllocyst usually contains an oleocyst (co), an oil-globule, which has a hydrostatic function. The phyllocyst of the bract is, therefore, similar to and comparable with the somatocyst of the nectophore. The phyllocyst is sometimes a simple cylindrical caecal canal or an ovate sac, placed in the vertical axis of the bract, or somewhat excentric ; as in Diplophysa, Cucubalus, and Cucullus. But usually some nutritive canals arise from its base, which enter into the jelly mass of the bract. The number and disposition of these phyllocyst canals are characteristic of the different genera, and of interest as remnants of the four radial canals, which run in the subumbrella of Medusae towards its margin. Eudoxella (PL XXXII.) and Aglaisma (PL XL.) still possess all four canals ; two of them are placed in the sagittal plane (one dorsal and one ventral), two others symmetrically on both sides (one right and one left). The latter are preserved too in Cuboides (PL XLII.) and in Amphiroa (PL XXXVI.), whilst the two sagittal canak are lost. Sphenoides (PL XXXVIII.) possesses only a single canal, descending on the dorsal side ; the three others have disappeared. (ZOOL. OHALL. EXP. PART LXXVII. — 1888.) Hhhh 14 106 THE VOYAGE OF H.M.S. CHALLENGER. Siphon. — The single polypite of Eudoxia, or the manubrium of the sterile medusome, always exhibits the same structure as in other Calyconectse. It is composed of four con- stant parts: — (1) A short pedicle (sp) ; (2) a thick-walled, usually hemispherical, basi- gaster, with masses of cnidoblasts (sb) ; (3) a dilated stomach, separated from the latter by a pyloric valve, and usually provided with liver-stripes (sm) ; and (4) a very extensile and contractile proboscis ; the latter opens through the mouth, which often exhibits four or eight lobes. (Compare PL XXXII. fig. 8; PL XXXIV. figs. 9-11 ; PL XXXVIII. fig. 12 ; PL XL. figs. 13, 14, &c.) Tentacles. — The single tentacle, which arises in each Eudoxia from the pedicle of the siphon, has usually the structure which is characteristic of the alyconectse. The long cylindrical and very contractile tubule is beset with numerous lateral branches or tentilla, separated by equal intervals. Each tentillum is composed of three parts, a slender pedicle, a reniform or ovate cnidosac, and a slender terminal filament. On the structure of the cnidosac compare above, p. 97. Gonophores. — Each Eudoxia produces originally only a single gonophore, and this fertile medusome is placed in the ventral part of the bracteal cavity of the sterile medusome, before its siphon. In many species there is constantly found only a single gonophore, and when this becomes detached, it is usually soon replaced by a secondary or vicarious gonophore ("Ersatz-Geschlechts-Glocke"). But in some species usually two, or even three, gonophores are found at the same time developing in one bracteal cavity ; they are usually of different sizes and ages, one placed at the right of the siphon, the other at its left (Pis. XXXIV, XLIL). EudoxiaB with a single gonophore are always gonochoristic, either male or female. When, however, two or three gonophores are developed simultaneously, then they are either of the same sex (Eudoxiie diclinicse, PL XLIL figs. 10, 14) or of different sexes {Eudoxise monoclinicse, PL XL. fig. 14). Each single mature gonophore of the Eudoxidse is a well-developed quadriradiate Medusa, without mouth and tentacles, with sexual cells in the wall of the manubrium (spadix). The umbrella is rarely rounded, and without sharp edges, as in the gonophores of Dip>lophysa aaA Eudoxella; usually it is four-edged, prismatic or pyramidal ; commonly two dorsal edges are more strongly developed, and more prominent over the basal ostium, than the two ventral edges. The apex of the umbrella is always prolonged into a beak- shaped apophysis or apical horn, the pedicle by which the gonophore is attached to the bracteal cavity. The quadriradial form of the umbrella in the single gonophore is usually symmetrical (with equal right and left halves) ; but when two gonophores are placed symmetrically on the two sides of one siphon, they are often modified by mutual pressure, and the single umbrella assumes an asymmetrical (sometimes pentagonal) form. (Compare p. 115, and PL XXXVI. figs. 12-25.) Compare on the development of the Eudoxidas above, pp. 100-102. REPORT ON THE SIPHONOPHOR^E. 107 Synopsis of the Genera of Eudoxidse. I. Subfamily Diplophtsidj;. Bracts not polyhedral, never prismatic, without com- plete sharp edges and poly- gonal faces. (Exumbrella rounded and smooth, not facetted.) Bract hemispherical or mitri- forra, rounded, without sharp edges, and without pointed apex. ' Phyllocyst simple, radial canals, without II. Subfamily Aglaismid.e. Bracts polyhedral, more or , less prismatic, with many ' sharp edges and polygonal faces. (Exumbrella facet- ted and angular.) Bract conical, spathiform or pyramidal, with a pointed , apex, and often with some ' incomplete edges. Bract more or less cuboidal, with an inferior opening into its cavity at the basal face. Bract not cuboidal, with an anterior opening into its cavity at the obliquely bevelled ventral face. Phyllocyst with four radial canals arising from its base, .... Phyllocyst simple. Bracteal exumbrella smooth, with no sharp edge, . Phyllocyst simple. Bracteal exumbrella with three or five edges, Phyllocyst with two lobate lateral canals. Bract with- out caudal apophysis, Phyllocyst with four cruciate radial canals. Bract with a pyramidal caudal apo- physis, . Phyllocyst descending, with two superior lateral canals. Bract with five odd trape- zoidal and two paired pentagonal faces, Phyllocyst ascending, with an odd inferior canal. Bract with four odd and four paired faces, 10. Diplophysa. 11a. Ewloxella. 11b. Cucubalus. 12. Cucullus. 13. Cuboides. 16. Aglaisma. 14. Amphiroa. 15. Sphenoides. Genus 10. Diphphysa,1 Gegenbaur, 1854. Diplophysa, Ggbr., Zeitschr. f. wiss. Zool., Bd. v. p. 291. Definition. — Eudoxidae with a hemispherical or subspherical bract, without sharp edges, in the vertical axis of which is placed a simple ovate or cylindrical phyllocyst. (Eudoxise of the genus Sphieronectes.) The genus Diplophysa comprises those monogastric Eudoxidse which arise from the polygastric Monophyid genus Sphieronectes (Genus 18). It was founded in 1854 by Gegenbaur for a Mediterranean Eudoxia, distinguished by a subspherical bract with a simple phyllocyst.2 It is very similar to another Mediterranean species, described in 1844 by Will under the name Erssea truncataJ Thirty years later Claus (70) demon- 1 Diplophysa = Double vesicle, Bi's-Xoa, Qiao.. - 7, p. 291, Taf. xvi. fig. :!. 3 65, p. 82, Taf. ii. fig. "28. 108 THE VOYAGE OF H.M.S. CHALLENGER. strated that this monogastric Calyconecta is the detached free Eudoxia of a polygastric Monophyid, which was first described in 1859 by Huxley as Sphaeronectes kolliheri (9, p. 50). He called it Monophyes gracilis (compare below the descriptions of Genera 18 and 19). The peculiar metagenesis of these two forms of Calycophorida? was afterwards accurately described by Chun (86, 87). A few Eudoxia? of this genus were found by me in a bottle in the Challenger collec- tion, taken in the Tropical Pacific, Station 274. Since the same bottle contained a specimen of Sphseronectes kolliheri, it is probable that they were detached from the latter. I call them Diplophysa kolliheri. Diplophysa differs from all other Eudoxidse in the hemispherical, or nearly spherical, form of its bract, and the simple ovate or cylindrical shape of its phyllocyst, which ascends vertically in the axis of the bract. In the centre of the shallow subumbral cavity of the bract is suspended a hemispherical gonophore, of about the same size, and between these two, at the ventral face of the latter, is a simple siphon with its tentacle. Genus 11a. Eudoxella,1 Haeckel, 1888. Eudoxella, Hkl., System der Siphonophoren, p. 32. Definition.— Eudoxidae with a mitriform, helmet-shaped, or hemispherical bract, the exurnbrella of which is smooth and without sharp edges. Phyllocyst with four radial canals arising from its base. (Eudoxia? of the genus Praya ?) The genus Eudoxella, as defined in my System, comprises free Eudoxise, the bract of which is essentially identical with that of the sessile cormidia of Praya (PL XXXIL). 1 suspect, therefore, that this genus is the monogastric generation of a true polygastric Praya. But the well-known species of this latter (Praya galea, Praya maxima, Praya cynibiformis) are generally assumed to possess eudoxomes, which become mature whilst still attached to the common stem. The question, however, whether this holds good as a general rule, requires further accurate examination. A bottle in the Challenger collection, containing surface animals taken in the Tropical Atlantic (Station 343, April 10, 1876), contained a few specimens of a Eudoxella, which is very similar to the sessile eudoxome of Praya galea (PI. XXXIL). But the helmet- shaped bract was more highly vaulted, and the four radial canals, arising from the base of the phyllocyst, were shorter and more equally disposed. The main difference was indicated by the remarkable fact, that each Eudoxia exhibited no trace of the stem of a Praya, and contained a large, completely mature gonophore. It is therefore probable that this Eudoxia, which I call Eudoxella didyma, may be the detached cormidium of some unknown species of Praya (Praya cymbiformisf), the gonophores of which are not brought to maturity whilst sessile on the stem. 1 Eudoxella^- Small Eudoxia. REPORT ON THE SIPHONOPHOR^E. 109 Genus 11b. Cucubalus,1 Quoy et Gaimard, 1824 (MS.). Cucubalus, Blainville, Manuel d'Actinologie, 1834, p. 130. Definition. — Eudoxidse with a conical, spathiform or cordiform bract, obliquely bevelled on the ventral side, evenly convex on the smooth dorsal side, pointed at the apex. Phyllocyst simple, ovate. (Eudoxise of the genus Muggisea.) The genus Cucubalus was founded by Quoy and Gaimard (2. pi. iv. figs. 24, 27) upon a Eudoxia from the Tropical Pacific, which Blainville afterwards described as Cucubalus cordiformis (24, p. 130). It may be, perhaps, the monogastric generation of Muggisea chamissonis ( = Diplujes chamissonis, Huxley, 9, pi. i. fig. 3). The best known representative of the genus is the Mediterranean Cucubalus eschscholtzii, which was described by Busch as Eudoxia eschscholtzii.2 Its development from the polygastric Muggisea kochii, and the regular metagenesis of these two forms, were afterwards described very accurately by Chun (86, 87). A similar and nearly allied species was observed by me in 1866 in the Canary Island, Lanzerote, and I found it again in some bottles from the Challenger collection taken in the Tropical Atlantic (Stations 348, 352). This Atlantic species may retain the name of Cucubalus pyramidalis, and its polygastric generation Muggi&a pyramidalis, with respect to the original names of Busch and Will (compare Chun, 85). The genus Cucubalus is closely allied to the following Cucullus, the monogastric generation of Diphyes. It differs from the latter in the smooth and edgeless shape of the bract, which is more rounded and not pyramidal. Genus 12. Cucidlus,3 Quoy et Gaimard, 1824 (MS.). Cucullus, Blainville, Manuel d'Actinologie, 1834, p. 131. Definition. — Eudoxidse with a cowled or cucullate, sometimes three-sided pyramidal bract, open at the bevelled ventral side, evenly convex, or sometimes with three incom- plete edges on the dorsal side, pointed at the apex. Phyllocyst simple and conical. (Eiidoxige of the genus Diphyes.) The genus Cucullus contains those monogastric Eudoxidaa which arise from the polygastric genus Diphyes, in the restricted sense in which recent authors have employed it, and which we have accepted. The bract is irregularly conical, or pyramidal, with a deep ventral fissure, which passes into the obliquely truncated cavity. Its dorsal face is either evenly convex, or slightly three-sided, its apex pointed. The inferior half of the spathiform bract embraces the included siphon and gonophore like a mantle, and 1 Ci«:w&aZMS= Name of a plant. ! 67, p. 33, Taf. iv. figs. 7-10 ; and Taf. v. figs. 1-9. 3 Cucullus = Cowl. 110 THE VOYAGE OF H.M.S. CHALLENGER. often one of its free ventral margins overlaps the other ; the superior half is comparable to the cowl of the mantle, and includes the simple ovate ascending phollocyst. The free inferior part of the mantledobes often exhibits two small prominent teeth. The genus Cucullus was founded by Quoy and Gaimard upon an Australian Eudoxia, which they figured first under the name Diphyes cucullus (2, p. 92), synonymous with Eudoxia lessonii of Eschscholtz (1, p. 126). Blainvdle called the former in 1834 Cucullus doreyanus (24, p. 131), and Lesson himself separated the latter as Cucullus lessonii (3, p. 459). Indeed the genus Cucullus of the French authors (published in 1834) is identical with Eudoxia of Eschscholtz (1829). But since Gegenbaur and Leuckart, in 1853, demonstrated that the Eudoxise are the isolated cormidia of Diphyidse, the term Eudoxia is generally used for all these " Diphyozooids " (Huxley, 1859, 9, p. 57). It seems, therefore, best, to avoid further confusion — to employ the term Cucullus for the monogastric generation of those Diphyidas which constitute the genus Diphyes in the strictest sense (compare below, Genus 27). The typical and oldest species remains Cucullus lessoni, as the free Eudoxia of the Pacific Diphyes appendiculata, Esch. Scarcely different from Cucullus is the genus Cucubalus of Quoy and Gaimard. This genus, however, maybe retained for the Eudoxise of theMonophyid Muggisea, which are very similar to those of Diphyes. (Compare the preceding genus.) Many bottles of surface animals in the Challenger collection contained specimens of Eudoxise belonging to Cucullus, and in some cases it was possible to demonstrate their connection with a species of Diphyes contained in the same bottle as the former. Supported by the comparative study of these forms, and of similar forms compared in other collections, and partly observed living by n^self, I have been led to state the ontogenetic connection of the following species of monogastric Cucullus with corre- sponding species of polygastric Diphyes : — 1. CucidJus lessoni, Lesson (3, p. 459), synonymous with Eudoxia lessonii, Eschscholtz (1, p. 126), is the free Eudoxia of Diphyes appendiculata, Esch. (1, p. 138), taken in the Pacific. 2. -Cucullus gracilis, Hkl., synonymous with Eudoxia bojani, Esch. (1, p. 125) and Huxley (9, p. 59), is the free Eudoxia of Diphyes gracilis, Hkl., taken in the Southern Pacific and in the Indian Ocean. 3. Cucullus elongatus, Hkl., is the free Eudoxia of Diphyes elongata, Hyndman (64, p. 166), occurring in different parts of the Northern Atlantic, Gulf Stream, &c. 4. Cucullus gegenhauri, Hkl., synonymous with Eudoxia messanensis, Ggbr. (7, Taf. xvi. fig. 4), is the free Eudoxia of Diphyes sieboldii, Koll. (4, Taf. xi.), Mediterranean and Tropical Atlantic. REPORT ON THE SIPHONOPHOR^E. Ill 5. Cucullus campanula, Hkl., synonymous with Eudoxia campanula, Leuckart (5, p. 43), is the free Eudoxia of Diphyes acuminata, Leuck. (5, Taf. iii.), inhabitant of the Mediterranean. 6. Cucullus subtilis, Hkl., synonymous with Ers&a elongata, Will (65, p, 82, Taf. ii. fig. 30), is the free Eudoxia of Diphyes subtilis, Chun (88, p. 687), also Mediterranean. Genus 13. Cuboides,1 Quoy et Gaimard, 1827. Cuboides, Quoy et Gaimard, Ann. d. Sci. Nat. (Zool.), t. x. p. 19. Definition. — Eudoxidse with a cuboidal bract, bounded by six quadrangular faces, twelve edges, and eight angles. Phyllocyst composed of a slender vertical canal, and two broad horizontal lateral lobes at its base. (Eudoxise of the genus Cymba.) The genus Cuboides (PI. XLII.) comprises those very remarkable monogastric Eudoxidas, which arise from the polygastric Monophyid genus Cymba (Genus 23, PL XLL). Its bract has the extraordinary form of a subregular cube, and is distinguished by a very characteristic phyllocyst. This is composed of a slender, vertically ascending canal, which usually includes an oleocyst ; and a wide basal diverticulum, which is divided into two broad, horizontally diverging, and ventrally directed lobes. We may compare these latter with the two symmetrical lateral canals of a bilateral Medusa, wlrilst the ascending canal corresponds to the odd ventral vessel. The fourth, dorsal and descending canal (preserved in Aglaisma) is lost in Cuboides. The basal diverticulum of the latter is the proper phyllocyst. The genus Cuboides was founded by Quoy and Gaimard (loc. cit.) upon an isolated Eudoxia, found in the Straits of Gibraltar, and called Cuboides vitreus. This is probably the detached monogastric form of the polygastric Cymba enneagonum found by the French authors at the same locality. It seems different from the species occurring in the Eastern Tropical Atlantic, and figured by me in PL XLII. as Cuboides crystallus. The figures and descriptions of the French authors are too insufficient to identify the two forms. The first accurate description of this peculiar Eudoxia, its cuboidal hydrophyllium and bilobed phyllocyst, was given in 1859, by Huxley (9, p. 63, pi. iv. fig. 5). He took it twice, once on the east coast of Australia, and once on the south coast of New Guinea, called it Cuboides vitreus, and suspected, rightly, that it might be derived from his Abyla vogtii [loc. cit., pi. ii. fig. 3). Different from this Australian species is an Indian species (Cuboides nacella, Hkl.), and the Atlantic species, which I shall describe in the sequel. It occurred in a bottle of the Challenger collection, from Station 352, near the Cape Verde Islands. I myself examined this species living in the Canary Islands in February 1867, and observed directly its origin and detachment from the polygastric Monophyid Cymba crystallus (compare below, Genus 23). 1 Cuboides — Cuboidal, xvpou ant. 112 THE VOYAGE OF H.M.S. CHALLENGES. Cuboides crystallus, n. sp. (PI. XLIL). Habitat. — Tropical and Subtropical Atlantic, Station, 352; lat.' 10 °55' N., long. 17°46'W. Surface. Canary Islands, Lanzerote, February 1867 (Haeckel). Bract (fig. 9, u, lateral view from the right side; fig. 10, u, dorsal view; fig. 13, u, basal view ; fig. 14, w, apical view ; figs. 11, 12, a younger specimen, seen from the left and from the dorsal sides respectively). — The hydrophyllium or bract is very large (10 to 12 mm. diameter), nearly cuboidal, with six flattened, slightly concave, quadrangular faces, twelve serrate, slightly concave edges, and eight prominent, three-sided pyramidal angles. The six faces are of nearly equal size, the apical face (ua) somewhat smaller, and the basal face a little larger than the four lateral faces (ud, dorsal ; uv, ventral ; ux, right ; id, left). The basal, inferior or distal face is occupied by a wide, flatly conical, subumbrellar cavity (ui), its apex reaches the centre of the cube. Phyllocyst (cs). — The apical canal of the bract, or the phyllocyst, consists of two very different parts, the axes of which are nearly perpendicular one to another. The slender apical part is subvertical, spindle-shaped, includes a fusiform oleocyst (co)', and touches with its apex the centre of the concave apical face of the bract (ua). The dilated basal part of the phyllocyst, however, is divided into two broad ovate horizontal lobes (cs", right ; cs', left) ; these are filled up with large clear entoderm cells, directed towards the ventral face of the bract (uv), and embrace the uppermost part of the peduncle of the siphon' (s). Siphon (s). — The siphon or polypite is pyriform, very contractile, suspended in the top of the subumbrellar cavity, and exhibits a pale yellowish colour. Its short pedicle bears at its thickened base, on its dorsal side, the tentacle (t) ; on its ventral side are placed one or two gonophores (g). Tentacle (t). — The single tentacle of this cuboidal Eudoxia is usually coiled up and hidden in the dorsal part of the infundibular cavity, behind the siphon. The tentilla or secondary filaments (fig. 16) have a thin pedicle (ts) and a simple terminal filament (tf), and between them a large reniform sacculus. This cnidosac has a leather-yellow or reddish-yellow colour, and contains on the convex dorsal side numerous (six to nine) rows of medial cnidocysts (km), and on each side of this battery, in the proximal part, a series of six to nine very large, spindle-shaped, lateral cnidocysts (kg) ; at the distal end a small group of small pyriform cnidocysts. Gonophores. — The Eudoxise develop in the bracteal cavity, on the ventral side of the siphon, either male or female gonophores. The young Eudoxise (figs. 11, 12) exhibit only one gonophore, the older two or three (figs. 9, 10, 13, 14). The umbrella of the gonophores (or the gonoealyx) is quadrangular, slightly bilateral, a little asymmetrical. REPORT ON THE SIPHONOPHOK-ffi. 113 Its superior or proximal half has four smooth edges, is pyramidal, and forms a prominent ventral crest. Its inferior or distal part is a quadrilateral prism with four strong serrate edges, prominent below as four three-sided pyramidal apophyses, the two dorsal far larger than the two ventral (fig. 1 1 ). Canal of the Gonophores (fig. 17, cp). — This is long, peduncular, and divides at the apex of the subumbrella («>) into four rather regular radial canals (or). These exhibit an elegantly pinnulated appearance, produced by alternately prominent small diverticula. They are connected at the base of the large velum (v) by a circular canal (cc). The ovaria (fig. 17, o), as well as the spermaria (figs. 9-14, km), are spindle-shaped manubria, which hang freely in the upper half of the umbrellar cavity of the gonophores. They contain a large central cavity (spadix). Genus 14. Amphiroa,1 Blainville, 1834. Amphiroa, Blainville, Manuel d'Actinologie, p. 133. Definition. — Eudoxidse with a bilateral, prismatic bract, bounded by five odd trape- zoidal faces and two paired pentagonal faces. Phyllocyst composed of an ovate de- scending dorsal sac, and two slender horizontally diverging lateral canals at its apex. (Eudoxise of the genus Abyla.) The genus Amphiroa (PL XXXY1.) comprises those monogastric Eudoxidse which arise from the polygastric Diphyid genus Ahyla (Genus 29, PL XXXV.). Its bract has a peculiar and very complicated form, which may best be compared with a prompter's box. It is composed of two four-sided prisms, which are united at right angles. The vertical prism covers the dorsal side of the Eudoxia, and is nearly filled up by a very large sac- shaped phyllocyst. From the apex of the latter arise two divergent lateral canals which run horizontally in the second prism, covering the apical face of the Eudoxia. The genus Amphiroa was first observed and named by Lesueur (probably in 1803), but first published from his drawings by Blainville, in 1834."' A very accurate description of it was given in 1859 by two excellent observers, Huxley (9, p. 64) and Gegenbaur (10, p. 17). These two authors, independently of each other, arrived at the same opinion, that Amphiroa is the free Eudoxia of Abyla triyona. The latter, therefore, called it Eudoxia trigones. We retain this name for the Mediterranean species, whilst Amphiroa. alata of Huxley is the Eudoxia of Abyla alata, inhabiting the Tropical Pacific and Indian Ocean. A third species, described by Huxley as Amphiroa angulata (9, pi. v. fig. 2), is probably the Eudoxia of Abyla leucharti, Huxley. Different from these three species is Amphiroa carina, which occurred in a bottle in the Challenger collection, taken at Station 348. I myself observed this species living, and detaching itself from Abyla carina, in February 1867, in the Canary Islands. 1 Amphiroa, ' ApQiQu, a daughter of Okeanos. - 24, p. 133, Atlas, pL iv. tig. 1. (ZOOL. CHALL. EXP.--PART LXXV1I. — 1888.) Hhhll 15 114 THE VOYAGE OF H.M.S. CHALLENGER. Amphiroa carina, n. sp. (PL XXXVL). Habitat. — Tropical and Subtropical Atlantic, Station 348 ; April 9, 1876 ; lat. 3° 10' N., lone. 14° 51' W. Surface. Canary Islands, Lanzerote, February 1867 (Haeckel). Bract (figs. 14, llJ, and 21, ventral view; fig. 15, dorsal view ; fig. 16, apical view; figs. 17 and 20, basal view; figs. 13 and 18, lateral view from the left side; fig. 12, lateral view from the right side. Figs. 12 and 13 taken from immature Eudoxomes, sessile on the stem). —The bract or hydrophy Ilium has a diameter of 6 to 8 mm., and in general the form of a prompter's box ; it may be described as a bilateral poly- hedron, which is composed of two four-sided prisms united perpendicularly ; the superior or horizontal prism includes the two diverging horizontal phyllocyst canals ; the posterior or vertical prism includes the large vertically depending phyllocyst. The obliquely bevelled ventral face of the bract (the inferior face of the first, and the anterior face of the second prism) is deeply excavated, and encloses the siphon and the gonophores. The horizontal apical or superior face (fig. 16) is trapezoidal, the ventral edge three times as long as the parallel dorsal edge ; the two equal lateral edges are concave, as long as the former, and diverge ventrally. The vertical dorsal or posterior face (fig. 15, ud), through which the phyllocyst (be) shines, is also trapezoidal, the basal edge twice as long as the parallel apical edge and half as long as the two lateral edges, which are slightly concave, and diverge towards the base. The two lateral faces of the bract (figs. 12, 18, between ventral and dorsal faces) are concave, irregularly pentagonal, nearly vertical, and divergent from the dorsal to the ventral side. The superior horizontal and the posterior vertical edge of the pentagon are the longest, of nearly equal size, and meet at right angles (in the apical corner of the dorsal face). The vertical anterior and the horizontal inferior edge are only half as long- as the former. The ventral (or antero-basal) edge of the pentagon, between the anterior and inferior edges, is deeply emarginate and serrate. The horizontal basal or inferior face of the bract (fig. 17, ub) is nearly square, and the smallest of all its six faces. Its ventral edge is bisected by the prominent median groove of the bract-cavity. The ventral or anterior face of the bract (figs. 14, 21) has the most complicated form of all six faces, and is deeply excavated by the cavity which includes the siphon and the gonophores. The ventral opening of this cavity has again a trapezoidal outline ; its upper horizontal edge is three times as long as the parallel basal edge, and somewhat longer than the two dentate lateral edges which diverge upwards. The superior part of the ventral face, above the opening of the cavity, is formed by a broad frontal face through REPORT ON THE SIPHONOPHOR^E. 115 which the two phyllocyst-cauals shine. This facet is nearly rectangular, its horizontal inferior edge somewhat longer than the parallel superior, and three times as long as the two short lateral edges which diverge downwards. The bract of the young Eudoxomes, sessile on the stem (fig. 13), exhibits the characteristic form of the adult to a much less degree, and has a more irregular cuboidal form. The bract of the young, just detached Eudoxia (figured in fig. 12, b), forms the transition from the latter to the former. Bracteal Cavity (figs. 12, 18, bh). — The subumbrellar cavity of the hydrophyllium is deeply campanulate or nearly conical, somewhat compressed from both lateral sides. Its posterior or dorsal wall is nearly vertical and touches the siphon and the tentacle. Its anterior half is filled up by one or more gonophores. The inferior and anterior trape- zoidal opening of the cavity has been already described. Phyllocyst (figs. 12-18, be). — The canal system of the bract is composed of a very large dorsal sac-shaped phyllocyst, which stands vertically, and of two slender horizontal canals, which arise from the apex of the former and diverge laterally. The large phyllocyst occupies nearly the dorsal half of the bract, and is a vertical cylinder with rounded basal faces, or nearly ellipsoidal, three times as long as broad. Its small internal cavity is surrounded and almost filled with very large vacuolate entoderm-cells. The two paired lateral canals of the bract (figs. 12-18, ex right, cl left), which arise from the apex of the phyllocyst at right angles, are slender and run divergently towards the two frontal corners of the bract, in their first half horizontally, in the second curved upwards. Their blind distal ends (at the junction of the apical, ventral, and lateral faces) are somewhat club-shaped. Siphon (figs. 12, 13, 21, s). — The polypite is hidden in the dorsal part of the bracteal cavity, and exhibits the usual structure of the Calyconectae — a short pedicle, a thick- walled basigaster, a wide stomach with hepatic stripes, and a proboscis with a four- or eight-lobed mouth. Tentacle (figs. 12, 13, t, 26). — The single tentacle, which arises from the pedicle of the siphon, exhibits the same structure as that of other Abylidse. Each tentillum bears on its pedicle (fig. 26, ts) a slender reniform cnidosac, and in the proximal part of this six to eight large spindle-shaped cnidocysts at each side of the cnido-battery (kg). The terminal filament (tf) is slender and long. Gonophores (figs. 22-25, isolated; figs. 12-19, included in the cavity of the bract). — Each ripe Eudoxia usually bears a pair of gonophores in the ventral part of its cavity, before the siphon. The two gonophores, right and left, are so transformed by mutual pressure that their umbrella or gonocalyx (originally a tetragonal prism) assumes an asymmetrical pentagonal form. The inner or medial faces (in which the two neigh- bouring gonophores meet in the sagittal plane of the bract) arc flattened, whilst the outer or lateral faces are prominent above, in form of a lateral horn. One of the five edges 116 THE VOYAGE OF H.M.S. CHALLENGER. (n7) is dentate. The five teeth of the basal mouth are very unequal, two much larger than the three others. The two paired gonophores together form a double body of bilaterally symmetrical form. Sometimes they are of the same sex, at other times of different sexes. Genus 15. Sphenoides,1 Huxley, 1859. Sphenoides, Huxley, Oceanic Hydrozoa, p. 61. Definition. — Eudoxidse with a bilateral prismatic or sphenoidal bract, bounded by eight faces, twelve edges, and twelve angles. Phyllocyst composed of a large apical sac and a slender odd dorsal canal descending from the base of the sac. (Eudoxiae of the genus Bassia. ) The genus Sphenoides (PI. XXXVIII.) comprises those monogastric Eudoxidas which arise from the polygastric Diphyid genus Bassia (Genus 30, PI. XXXVII.). Its bract has a very complicated sphenoidal form, and is distinguished by a large ovate sac of the phyllocyst in the apical half of the umbrella, whilst a long slender odd dorsal canal (similar to a basal spur) descends into the basal half. The original main axis of the umbrella is strongly curved, so that its ventral part is shortened and its dorsal part correspondingly expanded. The twelve edges of the wedge-shaped umbrella are produced into twelve three-sided pyramidal teeth, five of which belong to the ventral (if-if), seven to the dorsal half (uc'-uV2). The eight faces of the umbrella are four larger paired lateral (a superior pair quadrangular, an inferior pair hexagonal) and four smaller odd frontal faces, two superior triangular (one ventral and one dorsal) and two inferior (a triangular dorsal and a hexagonal basal). The comparison of the young bract (PI. XXXVIII. fig. 13) and the adult (fig. 14) exhibits the curious development of this cuneiform hydrophyllium. The genus Sphenoides was founded by Huxley (he. cit.) upon an Australian species, which he rightly suspected to be the Diphyozooid of Abyla bassensis ( = Bassia quadri- latera). This species is rather different from the Atlantic species described in the following, the development of which from Bassia I observed in the Canary Islands ; and from another species (Sphenoides tetragona) which I observed in the Indian Ocean. A fourth species may be the Eudoxia of the South Atlantic, Bassia. perforata (10). Sphenoides obeliscus, n. sp. (PI. XXXVIII.). Habitat. — Northern Atlantic; Canary Islands, Lanzerote, February 1867 (Haeckel). Bract (fig. 12, u, lateral view from the left side, with the included parts ; figs. 13 and 14, lateral view from the right side; fig. 13, of a younger attached Eudoxia; fig. 14, of Spherwides= Wedge-shaped, stpvntllrii, REPORT ON THE RIPHONOPHOR^. 117 an older free Eucloxia). — The hydrophyllium or bract lias a diameter of 10 to 12 mm., and a sphenoidal or wedge-shaped, rather complicated form, which is best understood if we imagine the true axis of the umbrella standing vertically, its upper or apical pole marked by the top, the lower or basal pole by the centre of the aperture of the cavity. The latter (or the original ostium subumbrellse) is surrounded by five prominent points, an odd dorsal basal point (w8) and four paired lateral ; two* of these are dorsolateral basal points (it8 right, u7 left), the two others ventro-lateral basal points (u4 right, ub left). The opposite apical or superior part of the ripe bract (fig. 14) has the form of a house- roof or of a triangular prism, inclined ventrally. The apical crest (nk) is strongly inclined towards the ventral side, so that the dorsal apical point (u12) may be regarded as the original apex of the bilateral umbrella. The opposite ventral apical point (w1), on the contrary, marks the upper pole of the curved axis of the phyllocyst (be). The ventral face of the bract is pentagonal and marked by five prominent points, one odd and four paired. The odd is the above-mentioned ventral apical point (ul). The upper pair of points are the ventro-lateral apical (ir right, u3 left) ; the lower pair of points are the ventro-lateral basal points (u4 right, us left). The dorsal face of the bract is quadrangular in the upper part, crest-shaped in the lower part, and is marked by five points, three odd and two paired. The uppermost is the odd dorsal apical point (vP), the lowermost the odd dorsal basal point (u8), the hmdermost (between the former) the odd dorsal caudal point (a9). At the right side (u10) and left side (u11) the two dorso-lateral points are prominent. The wedge-shaped dorsal crest (ng) is in the ripe bract (fig. 1 4) opposite and parallel to the apical crest (nk). All crests and edges of the bract are elegantly denticulated. The mature bract of the free Eudoxia (fig. 14) differs from the young bract of the sessile Eudoxoma (fig. 13) mainly in the development of a pair of arched prominent transverse ridges or girdle-wings (u\ uw, right ; ii\ u11, left). These separate more distinctly the thicker apical half of the hydrophyllium from its thinner ventral half. An accurate comparison of fig. 13 (young) and fig. 14 (adult) will explain the other differences of these two forms, which are produced by further growth. Bracteal Cavity (figs. 13. 14, bh, seen from the right side).— The subumbrellar cavity has the form of a, helmet or a Phrygian cap ; it is eampanulate, with strongly curved axis, concave at the ventral, convex at the dorsal side. The cavity occupies in the young bract (fig. 13) about two-thirds of the whole body; in the adult (fig. 14) only the basal half. Its basal aperture, or the original mouth of the umbrella, is armed with five large triangular, pyramidal points or teeth, one odd and four paired ; the posterior odd tooth (it8) is the dorsal basal point ; the two ventro-lateral basal teeth (u4, us) are larger and wider apart than the two dorso-lateral basal teeth (mb, v7). A deep ventral groove, through which passes the stem of the cormus (fig. 13, a), is placed in the sagittal plane of the bract, immediately to the ventral side of the subumbrellar cavity (/>//). 118 THE VOYAGE OF H.M.S. CHALLENGER. Phyllocyst (fig. 13, be). — The canal system of the bract is composed of three parts of a large ovate phyllocyst (be), a short peduncular canal, and a long dorsal csecum. The small peduncular canal arises from the common stem (a), and runs immediately to the ventral base of the large ovate phyllocyst (be). This latter is placed in the apical half of the bract, and contains in its uppermost part a spherical oleocyst (eo). Its rounded apex does not reach the apical crest (nk). The dorsal canal (cd) is a slender caecum, which arises from the base of the phyllocyst. and runs along the dorsal median line of its cavity, separated from it by a thin jelly-plate. The dorsal canal is therefore spur-shaped and crooked, concave on the ventral, convex on the dorsal side. Its blind lower end does not reach the basal edge (ng). Siphon (fig. 12, s). — The polypite or siphon is suspended in the top of the subum- brellar cavity of the bract, and may be completely hidden in it. From its short pedicle arises the tentacle (t). A pyloric valve separates the thick basigaster (sb) from the stomach, which exhibits eight yellow striae hepaticfe. The proboscis (sr) is very con- tractile ; its mouth exhibits sixteen short indentations. Tentacle (fig. 1 2, t). — The single tentacle of the siphon is distinctly segmented by numerous constrictions, and from each constriction arises a tentillum. The pedicles (ts) and the terminal filaments (tf) of the tentilla are long and slender. The reniform cnidosac exhibits on its convex side six to eight longitudinal rows of medial cnidocysts (fig. 16, km), on both sides of its base a group of large spindle-shaped lateral cnidocysts (kg), and at the distal end a group of small pyriform cnidocysts (kp). Gonophores (fig. 12, h; fig. 15). — The sexual Medusa which arises from the stem at the ventral side of the siphon, has a four-winged umbrella, which is pyramidal in the smaller apical half, prismatic in the larger basal half. Its four prominent and denticulate wings are unequal in pairs in the younger gonophores, the two dorsal wings (fig. 12, hit) being much larger than the two ventrals ; afterwards (fig. 15) they become nearly equal. A long peduncular canal (fig. 1 5, cp) runs through the pedicle of the umbrella to the top of the subumbrella («'), and divides here into four equidistant radial canals. The spindle- shaped spermaria of the male gonophore (fig. 12, hs), and the ovate or subspherical ovaria of the female (fig. 15, o), occupy the apical half of the subumbrellar cavity. Genus 16. Aglaisma,1 Eschscholtz, 1829. Aglaisma, Eschscholtz, System der Acalephen, p. 129. Definition. — Eudoxidae with a bilateral prismatic bract, which is cuboidal in the larger upper part, obliquely pyramidal in the smaller lower part. Phyllocyst with four cruciate radial canals, two odd sagittal (an ascending and a descending) and two paired lateral canals. (Eudoxiaj of the genus Calpc.) 1 Aglaisma, dyhaiafnt^Omaxaeat REPORT ON THE S1PHONOPHOKJE. 119 The genus Aglaisma (PI. XL.) comprises those monogastric Eudoxidse which arise from the polygastric Diphyid genus Calpe (Genus 31, PL XXXIX.). Its bract is nearly cuboidal, similar to that of Cuboides (Genus 13), but differing from it in the possession of a pyramidal, trigonal, or pentagonal apophysis, which descends vertically from the dorsal and basal part of the cube. It is further distinguished by the possession of four radial canals, arising from the small subcentral phyllocyst. Two of these four cruciate canals are slender and vertical (an ascending and a descending), two others sacculate and hori- zontal (a right and a left). The genus Aglaisma was founded in 1829 by Eschscholtz for a Tropical Atlantic species of monogastric Diphyidse, which he called Aglaisma baeri (1, p. 129) ; it is possibly identical with our Aglaisma gegenbauri. Afterwards Huxley,1 in 1859, gave an accurate description of two nearly related species, which he called Aglaismoides t'sclischoltzii and Aglaismoides elongata. He suspected rightly that the former might be the Eudoxia of the common Mediterranean Abyla pentagona. The other species, Aglaismoides elongata, might be the Eudoxia of Calpe huxleyi. During my stay in the Canary Islands (1867) I observed the development of the species described in the following pages, from Calpe gegenbauri. It occurs also in a bottle of the Challenger collection, taken at Station 346. Aglaisma gegenbauri, n. sp. (PI. XL.). Habitat.— Tropical and Subtropical Atlantic, Station 346 ; April 6, 1876 ; lat. 2° 42' S., long. 14° 41' W. Surface. Canary Islands, Lanzerote, February 1867 (Haeckel). Bract (fig. 13, b, lateral view from the right side ; fig. 14, b, from the left side; fig. 15, basal view ; fig. 16, apical view; fig. 17, dorsal view ; fig. 18, ventral view). — The hydrophyllium or bract is subregularly cuboidal in its apical or superior half, in its basal or inferior obliquely pyramidal, bevelled off at the ventral and basal face. The length or height of the bract is 1*2 to l-5 mm., the breadth 0"6 to 0-8 mm. The apical or superior face of the bract (figs. 13-16, ua) is nearly square, slightly concave, with four equal concave edges. The phyllocyst and its two lateral casca shine through it in the apical view. The ventral or anterior face (fig. 18, uv) has the same square form, but is somewhat longer, and dilated towards the concave basal edge. The dorsal or posterior face of the bract (fig. 17, ud) is more concave, and has the form of a bilateral pentagon, which is twice as long as broad. Its odd superior edge is horizontal, and half as long as the two divergent superior lateral edges. These are somewhat longer than the two convergent inferior lateral edges, which meet in the basal apex of the 1 9, p. 60, pi. iv. tigs. 2, 3. 120 THE VOYAGE OF H.M.S. CHALLENGER. pyramidal caudal apophysis. Whilst the superior half of the dorsal face is nearly vertical, square, aud concave, its inferior peutagoual half is convex aud prominent, directed more dorsally downwards, and bisected by a sagittal crest which becomes obsolete towards the upper half. This crest bears about the middle of its length an odd prominent tooth (figs. 13, 14). The two-paired lateral faces of the bract (right, fig. 13 ; left, fig. 14) are hexagonal, square and vertical in their upper half, irregularly hexagonal and more prominent in the lower half. The superior horizontal edge of each lateral face is half as long as the ventral, and one-third as long as the basal and the dorsal edge. The vertical ventral edge is concave. The opposite dorsal edge bears a prominent tooth between its middle and lower thirds, and so also does the obliquely descending basal edge. The basal or inferior face of the bract is occupied by the square opening of the bracteal cavity (fig. 15, middle portion), and behind this by the triangular ventral face of the caudal apophysis (fig. 18). The caudal apophysis, which distinguishes the cuboidal bract of Aglaisma from the similar bract of Cuboides (PI. XLIL), is a bilateral oblique pyramid, which descends from the prolonged dorsal and basal part of the hydrophy Ilium. The pyramid is five-sided, directed downwards and dorsalwards up to the apex, which is slightly curved inwards. The five angles of its base are marked by the five above-mentioned teeth, viz., the odd dorsal tooth of the sagittal crest (in the inferior half of the dorsal face), the paired teeth of the latero-dorsal edges, and the paired (somewhat inferior) teeth of the lateral basal edges. All the edges of the polyhedral transparent bract are elegantly denticulate, and more or less curved. The bracteal cavity (figs. 13-18, ui) opens in Aglaisma (as in Cuboides) more on the basal or inferior face ; whereas in Amphiroa and Sphenoides the aperture is rather on the ventral or anterior face. It occupies the greater part of the inferior half of the bract, and is obliquely campanulate. The apex of the cavity is directed dorsally. Phyllocyst (figs. 13-18, be). — The phyllocyst is a small subspherical cavity, placed immediately above the basigaster (sb), and gives off four large canals, two odd and two paired, which are comparable to the four radial canals of a bilateral Medusa. The superior odd canal (cv), which corresponds to the original ventral canal of the Medusa, ascends vertically near the median line of the dorsal face, and ends above in a spindle- shaped diverticulum which includes an oleocyst (co). The inferior odd canal (cd) is somewhat longer, also placed in the median plane, and descends obliquely downwards ; it runs parallel to the neighbouring dorsal crest, and corresponds to the dorsal radial canal of a Medusa. The two paired lateral canals (ex right, el left) are much shorter, wider, and sac-shaped, as in Cuboides. They are geniculate and filled by large vacuolate entoderm-cells ; their dorsal half is placed nearly horizontally, whilst their ventral half is REPORT ON THE SIPHONOPHOR^E. 121 curved downwards. Their blind ends are directed towards the ventro-basal angles of the lateral faces. Siphon (figs. 13, 14, s). — The single polypite occupies originally the dorsal half of the bracteal cavity (fig. 13), but, when two gonophores are developed, it becomes placed between them more ventrally (fig. 14). Its thick-walled basigaster (sb) is nearly spherical. The spindle-shaped stomach (sm) bears eight yellow hepatic stripes. The contractile proboscis (sr) opens by an octolobate mouth (ss). Tentacle (figs. 13, 14,/). — The single tentacle exhibits the same structure, which is accurately described in the case of Calpe pentagona by Kolliker, Leuckart, and others. It arises from the short pedicle of the siphon, between the basigaster and phyllocyst. It bears numerous tentilla, each with a reniform cnidosac and a long terminal filament coiled up spirally in fig. 19. Fig. 20 exhibits its elastic band expanded. Gonophores (fig. 13,/; fig. 14, hf). — The bracteal cavity contains sometimes a single gonophore (fig. 13), at other times two or even three gonophores associated, besides some buds of vicarious function. Fig. 14 exhibits a well-developed monoclinic Eudoxia, with a male (h) and a female (f) gonophore. The sexual manubria of both are club-shaped, and fill about the half of the subumbrellar cavity. The umbrella is in both sexes of the same form, distinctly bdateral, with four paired denticulate edges, prolonged at the basal ostium into four strong triangular pyramidal teeth. The two dorsal teeth are much larger than the two ventral. The apex of the umbrella is a curved horn, and contains the pedicular canal ; this gives off at the apex of the subumbrella the four radial canals, as usual. Family V. Ees^id^, Haeckel, 1888. Erseeidx, Hkl., System der Siphonophoren, 95, p. 33. Definition. — Calyconectae monogastricae, representing a single cormidium, which is composed originally of three persons ; a sterile medusome (siphon with tentacle and bract), a sterile special nectophore (without manubrium), and a fertile medusiform gonophore (male or female). The family Ersaeidse comprises those monogastric Calyconectae which differ from the preceding Eudoxidse in the possession of a so-called " special nectophore." Their cormidium is therefore composed not of two, but of three, medusomes or persons of the morphological value of a Medusa. The Ersseidse, as independent forms, occur much more rarely than the Eudoxidse. I have observed two genera only, both rather rare, Erssea and Lilsea. The sterde medusome of the Ersseidse is composed (as in the Eudoxidse) of three organs — viz.,bract(umbrella), siphon (manubrium), and tentacle (cnido-filament) ; all three (zool. chall. exp. — part lxxvii. — 1888.) Hhhh 16 122 THE VOYAGE OF H.M.S. CHALLENGER. organs possess essentially the same structure as in the preceding Eudoxidse. The bract occurs only in two forms, one of which (Lilsea) is equal to that of Praya, or Lilyopsis ; it is rounded and has a phyllocyst with four radial canals ; the other (Ersasa) is similar to that of Muggisea or Diphyes; it is pointed, cucullate or spathiform, and has a simple phyllocyst, without radial canals. Gonophores. — The sexual medusomes of the ErsEeidse have the same form and structure as those of the Eudoxidse ; and mainly of those species, where two or more gonophores are produced in a single cormidium (compare above, pp. 100 and 121). The special nectophore, which is characteristic of the Ersaeidse, and distinguishes them from the Eudoxidse, has been originally the first gonophore. This has completely lost its manubrium, and has, therefore, given up its sexual function ; it has been developed merely as an organ of swimming. Ontogeny. — On the development of the Ersseidae, compare above, pp. 100-102. Synopsis of the Genera of Ersseidae. Bract spathiform, with an apex. Phyllocyst simple, without radial canals, . . . 17a. Ersxa. Bract rounded, without apex. Phyllocyst with four radial canals, . . . .17b. Lilsea. Genus 17a. Erssea?- Eschscholtz, 1829. Erssea, Eschscholtz, System der Acalephen, p. 127. Definition. — Ersseidae with a campanulate or spathiform bract, divided by a collar constriction into a proximal cowl, with an apical point, and a distal mantle, with a wide ventral fissure. Phyllocyst large, simple, without radial canals. (Cormidia of the genus Diphyopsis.) The genus Erssea was founded by Eschscholtz, in 1829, to include two Atlantic monogastric Diphyidae, differing from Eudoxia in the possession of a special nectophore, which is prominent from the cavity of the bract (" Tubulus suctorius unicus ; pars corporis nutritoria cavitate parva natatoria, tubuli instar prominenti instructa"). The two species, however, which Eschscholtz there described [Erssea quoyi and Erssea gaimardi),2 are too incompletely represented to determine with certainty their position and their connection with any species of Diphyopsis. Possibly Erssea gaimardi may be the free cormidium of Diphyopsis campanulifera, and Erssea bojani ( = Eudoxia hojani, Eschscholtz), the cormidium of Diphyopsis disp>ar. I myself observed in 1866, in the Canary Island Lanzerote, numerous free-swimming specimens of the species figured in PI. XXXIV. as Erssea compressa, and afterwards recognised its ontogenetic connection with Dij)hyopsis compressa (PI. XXXIII.). 1 Erssea = f{<7«?«, covered with dew. - 1, p. 128, Taf. xii. figs. 3, 4. REPORT ON THE SIPHONOPHOR^E. 123 Ersssa, as the monogastric generation of Diphyopsis, differs essentially from Cucullus as the free Eudoxia of Diphyes. The latter never possesses the individual " special nectophore," which is characteristic of the former. This swimming organ must In- regarded as the first sterile gonophore, which has lost the manubrium, the muscle-plate of the subumbrella being the more strongly developed. Erssea convpressa, n. sp. (PI. XXXIV.). Habitat. — Tropical and Subtropical Atlantic; Stations, 348, 349, 352, 353, &c. Canary Islands, Lanzerote, December 1866 and January 1867 (Haeckel). Bract (figs. 9, 10, 11, b). — The hydrophyll or bract is in general irregularly conical or campanulate, with a deep fissure along the truncate ventral side ; the broad basal or inferior aperture is also obliquely truncate. An annular collar constriction, corresponding to that part of the umbrella which was attached to the stem of Diphyopsis, divides the bract into a smaller apical (proximal or superior) part (fig. 11, bs) and a larger basal (distal or inferior) part. The latter encloses the siphon and its tentacle, and partly the gonophores, like a mantle, while the former is comparable to the cowl of the mantle. This cowl contains in its solid dorsal half the phyllocyst (cs), with a central cavity and a globular oleocyst on the apex (co) ; its ventral half is excavated and originally embraces the stem of siphosome (fig. 9, a) ; the two ventral wings of this cavity overlap one another in the middle part. The length of the bract is 6 to 8 mm., its breadth 3 to 4 mm. The campanulate mantle, or the larger distal half of the bract, exhibits around its wide basal opening four acute triangular teeth, comparable to the four corners of the original medusa-umbrella. The pair of dorsal teeth is larger and more prominent than the pair of ventral teeth. The two opposite free ventral margins of the bract are smooth. Siphon (figs. 9,s, 10, s, 11, s). — The polypite or siphon occupies originally the axial part of the bracteal cavity, between the dorsal tentacle (t) and the ventral nectophore (tin). The basigaster (sb) is separated by a sharp pyloric constriction from the stomach, the wall of which exhibits eight longitudinal hepatic ridges. The mouth of the muscular proboscis is surrounded by sixteen short lobes (so). Tentacle (figs. 9, t, 10, £,11, t). — The long cylindrical tentacle is distinctly articulated, and from the constriction between every two segments arises a thin tentillum or secondary filament (fig. 18). The terminal filament of the latter (tf) is about as long as its pedicle (ts). The ovate cnidosac placed between them exhibits on its convex side six to eight longitudinal rows of small medial cnidocysts (km), on its base two paired groups of large lateral cnidocysts (six to eight spindle-shaped cnidocysts, kg, in each), and at the distal part a trilobate group of pyriform cnidocysts (hp). 124 THE VOYAGE OF H.M.S. CHALLENGER. Nectophore (figs. 9, 10, 11, nn). — The special nectophore, which contains no manu- brium, occupies the ventral fissure of the bracteal cavity, in which only its basal part is enclosed. The umbrella is a slender four-sided pyramid, distinctly bilateral, since the two dorsal edges are stronger, and more prominent at the base, than the two ventral edges. All these edges, and also the quadrangular basal edge, with the four prominent teeth, are elegantly denticulate (fig. 17). The basal and ventral faces of the exumbrella are concave, the dorsal and the two lateral faces convex. Nectosac. — The peduncular canal, which arises from the base of the phyllocyst, enters through the obliquely truncate top of the nectophore, and divides on the top of the muscular subumbrella (w) into four subregular radial canals (or). These unite at the ostium of the umbrella by a circular canal (cc) above the broad velum (v). Gonophores (fig. 11, hm; figs. 12-14). — The young Erssea compressa possesses, besides the sterile nectophore, only a single gonophore, the older form two or three gonophores. These are usually of the same sex (the cormidia therefore diclinic) ; but sometimes a male and a female gonophore are present in the same Eudoxia (the cormidium there- fore monoclinic). The umbrella (u) of the male, as well as of the female gonophores, is slender ovate, with four edges which are smooth in the proximal half, elegantly denticu- late in the distal half. They are prolonged over the mouth of the umbrella, and form four prominent triangular teeth, the two dorsal teeth being larger than the two ventral. The subumbrella (w) exhibits in the gonophores the same four subregular radial canals, connected by a circular canal, as in the nectophore (figs. 9-11, nn). The male gonophores (fig. 11, hm; fig. 12) have in their umbrellar cavity a large spindle-shaped or ovate spermarium (hm), with a wide central spadix (hx). The female gonophores (figs. 13, 14) exhibit an ovarium of very variable size (frn). Sometimes the entire umbrellar cavity is filled with eggs (fig. 13), whilst, at other times, oidy a few ova (four in fig. 14) occupy its proximal half. Genus 17b. Lil&a,1 n. gen. Definition. — Ersseidse with a hemispherical or mitriform rounded bract, without sharp edges and without pointed apex. Phyllocyst small, with four radial canals arising from its base. (Cormidia of the genus Lilyopsis f) The genus Lilsea comprises those Ersaeidse which possess a mitriform rounded bract, without sharp edges or apical point. Its form agrees with that of the eudoxomes of Praya and Lilyopsis. A further resemblance to the latter is given by the fact, that each cormidium possesses a sterile nectophore, the mouth of which bears a circle of rudimentary tentacles and four red pigment-spots (ocelli) at the distal end of the four radial canals. They are very similar to the medusiform special nectophores of Desmo- 1 Lilxa (Ai'huici), name of a Naiad. REPORT ON THE SIPHONOPHOR.E. 125 phyes annectens (PL XXX.). The latter agrees also in the formation of the gonophores. Each connidiuiu possesses, besides the large vigorously swimming special necto- phore, a cluster of small gonophores, all either male or female. The umbrella of the male gonophores is more developed than in Dcsmophyes. Among six specimens examined four were males and two females. I observed half a dozen of these small interesting Ersaeidae, which will be more accurately described on another occasion, hving in the Indian Ocean, during my voyage from Aden to Bombay, in November 1881. I suspect that they are the detached cor- midia of some species of Lilyopsis (Genus 25). They may bear provisionally the name Lilsea medusina. Family VI. Monoph yidje, Claus, 1874. Morwphyidx, Claus, 1874, 70, p. 29. Sphxronectidx, Huxley, 1859, 9, p. 50. Definition. — Calyconectae polygastricae, with a single nectophore at the apex of the long tubular truncus. Cormidia ordinate, eudoxiform, separated by equal free inter- nodes ; each siphon with a bract. The family Monophyidae comprises a small number of little known polygastric Calyconectse, which are rather rare, of small size, but very interesting on account of their simple structure. They all possess only a single permanent nectophore, and differ in this character from the nearly allied Diphyidae, with which they were formerly united. Eschscholtz, in his fundamental work (1829, 1, p. 134), described only a single form' of Monophyidae, Cymba enneagonum, the free Eudoxia of which is his Cymba cuboides. Both forms were discovered in the Straits of Gibraltar by Quoy and Gaimard, who called the first (polygastric) form Enneagonum hyalinum,1 and the second (monogastric) form Cuboides vitreus. A second species of Monophyidae, inhabiting the Gulf of Trieste, was described in 1844 by Will,2 under the name Diphyes kochii, and in 1851 by Busch,3 as Muggisea pyra- midalis. The same form was afterwards, by combination of both names, called Muggisea kochii, by Chun (86, p. 3). He described its metagenesis and development from the mouogastric Eudoxia eschscholtzii. Huxley, in his excellent work on Oceanic Hydrozoa (1859, 9), described not less than four different Monophyidae, viz. : — (1) Sphseronectes kbllikeri, pi. iii. fig. 4. (2) Diphyes mitra, pi. i. fig. 4 (now Cymbonectes mitra). (3) Diphyes chamissonis, pi. i. fig. 3 (now Muggisea chamissonis) . (4) Abyla vogtii, pi. ii. fig. 3 (now Cymba vogtii). 1 20, pi. ii.D, figs. 1-6. 2 65, p. 77, Taf. ii. fig. 22. 3 67, Taf. iv. figs. 3-6. 126 THE VOYAGE OF H.M.S. CHALLENGER. Huxley gave an accurate description of the only nectophore observed of these four " Diphyidae " ; he supposed (as did also his predecessors) that the second nectophore had been accidentally lost ; it does not, however, exist at all. The genus Sphasronectes was rightly regarded by Huxley as the type of a peculiar family — Sphaeronectidae. During my residence in the Canary Island of Lanzerote (December 18G6 to February 1867), I observed four different species of Monophyidae, viz.: — (1) Monophyes hydrorrhoa. (2) Mitrophyes peltifera (PL XXVIIL). (3) Muggisea pyramidalis. (4) Cymba crystallus. I was able to examine the complete metagenesis of the latter species, and the develop- ment of its Eudoxia, Ouboides crystallus (Pis. XLL, XLII.). A Mediterranean species of Sphieronectes, very similar to the Australian form dis- covered by Huxley, was described in 1874 by Claus, and called Monophyes gracilis (70. pi. iv.). He observed the development of its Eudoxia, which was formerly described by Gegenbaur as Diplophysa inermis (7, Taf. xvi. fig. 3). Another Mediterranean Mono- phyid described by him, Monophyes irregularis, may remain the type of this genus. Claus replaced the term Sphaeronectidae of Huxley by the name Monophyidae, which was accepted as more significant, in opposition to Diphyidse. The metagenesis of the two Mediterranean Monophyidae was afterwards very accu- rately examined by Chun (86-88). Compare his memoirs also for the history of this family. During my voyage through the Indian Ocean (November 1881 and March 1882) and in Ceylon, I had the opportunity of examining some very interesting new forms of Mono- phyidae and their development, viz., Monophyes princeps and Cymbonectes huxleyi (PI. XXVII. ). Supported by these observations, and by some specimens found in the Challenger collection, I was able to give the following description of Monophyidae. Nectophore. — The single nectocalyx of the Monophyidae exhibits differences in form and structure similar to the first or proximal of the Diphyida?. Accordingly, I divide the family Monophyidae into two subfamilies ; the first of these, Sphaeronectidae, has a smooth hemispherical or mitriform nectophore, without sharp edges, and is allied to the Prayidae among the Diphyidae. The second subfamily, Cymbonectidae, has a pyramidal nectophore with five prominent edges, and is nearly allied to the Diphyopsidse and Abylidae. The single nectophore of the Monophyidae is relatively large, of a bilateral and quadriradial fundamental form, sometimes symmetrical, at other times asymmetrical. The first nectophore of their larva is replaced by a permanent, often heteromorphous, swimming-bell. llydrcecium. — Since the single nectocalyx of the Monophyidae corresponds to the first apical or proximal nectophore of the Diphyidae, it possesses a hydrcecium for the reception of the retiring siphosome. This is an open hydrcecial groove, protected by REPORT ON THE SIPHONOPHORjE. 127 two overlapping ventral wings, in Monophyes and Cymbonectes (PI. XXVII.) ; it is a conical or campanulate hydrcecial cavity in Muggisea and Cymba (PI. XLL), a cylindrical canal in Sphmronectes. These closed hydrcecia are secondary cavities, produced by concres- cence of the two parallel ventral wings, which overlap the hydrcecial groove of the former. The singular genus Mitrophyes (PI. XXVIII.) is distinguished by the lack of a hydrcecium. It is replaced by an apical scutiform bract, which covers the nectophore and protects the siphosome hidden between them both. The apical bract is probably the remnant of the original primary nectophore. Nectosac. — The subumbrella in most Monophyidse occupies the dorsal part of the nectophore, whdst the hydrcecium is placed in its ventral part. The nectosac of Sphmronectes and Mitrophyes is placed rather basally (as in the ancestral Medusa?), in the other genera rather dorsally. The four radial canals are in the former genera rather regularly disposed, but usually more bilaterally, the ventral canal being shorter, and the dorsal longer than the two paired lateral canals. The ring-canal of the margin, which unites them, is placed above the velum. Somatocyst. — The acrocyst or somatocyst (" Saftbehalter ") in most Monophyidse is of moderate size, placed in the apical prolongation of the stem ; its cavity is narrow, usually filled by large vacuolate entoderm-cells, and its apex mostly contains an oleocyst. It is directed sometimes vertically upwards, at other times more obliquely. Its structure is the same as in the other Calyconectse (compare above, p. 93). Siphosome. — The long tubular stem exhibits in the Monophyidse the same structure as in the Diphyidas. The median ventral line of the common stem is beset at regular intervals by the cormidia, whose number is very variable. The contracted siphosome may be retracted into the hydrcecium more or less completely. Cormidia. — Each cormidium (Diphyozooid or Eudoxia) is composed in the Mono- phyidse (as in the most Diphyidse) of two medusomes ; the sterile medusome has a bract, a siphon, and a tentacle ; the fertile medusome is a gonophore, the umbrella of which has the usual medusoid structure, whilst the manubrium produces the sexual cells (compare above, p. 94). Sipjhon and Tentacle exhibit no important differences in the cormidia of the various Monophyidse, whilst the bracts or hydrophyllia are of very different form and structure, characteristic of the genera (compare above, pp. 95, 96). Eudoxise. — The cormidia of some Monophyidse arrive at sexual maturity whilst attached to the stem ; they remain sessile eudoxomes. This is the case in Mitrophyes and Cymbonectes, and probably also in Monop>hyes. The cormidia of the three other genera are detached from the stem, and become mature as free-swimming Eudoxise; those of Sphieronectes are described as Dyplophysa; those of Muggisea as Cucubalus, whilst the free Eudoxise of Cymba belong to Cuboides. Ontogeny. — On the development of the Monophyidae, compare above, pp. 100-102. 1-2S THE VOYAGE OF H.M.S. CHALLENGER. I. Subfamily Sph^ronectidj:. Nectophore hemispheri- cal or mitriform, with rounded surface, with- out sharp edges. II. Subfamily CYMBONECTID2E. Nectophore pyramidal, with five prominent sharp edges. Synopsis of the Genera of Monophyidae. Hydroecium a ventral groove of the nectophore, incompletely closed by two overlapping wings, ..... Hydrcecium a complete cylindrical canal in the ventral wall of the nectophore, with a basal opening, .... Hydroecium wanting. Nectophore protected by a cap-sbaped dorsal bract ; between them is the siphosome, Hydrcecium a ventral groove of the nectophore, incompletely closed by two overlapping wings. Bracts spathiform, Hydrcecium a complete coni- cal or campanulate cavity in the ventral wall of the nectophore. Bracts of the cormidia coni- cal or spathiform, with a deep ventral fissure, . Bracts of the cormidia cu- boidal, with six square faces and a basal cavity, . 18. Monophyes. 19. Sphseronedes. 20. Mitrophyes. 21. Cymboneetes. 22. Muggixa. 23. Cym.ba. Genus 18. Monophyes,1 Claus, 1874. Monophyes, Claus, Die Gattung Monophyes, &c, 70, p. 29. Definition. — Monophyidse with a rounded, edgeless, mitriform nectophore, and an open hydroecial groove on its ventral side ; the latter includes the siphosome. which is incompletely protected by two overlapping lateral wings. Bracts mitriform or hemi- spherical, with rounded surface and a simple phyllocyst. The genus Monophyes was founded in 1874 by Claus for two different Mediterranean species of Calyconectse, which bear a single mitriform nectophore on the top of the stem. One of these two species, Monophyes gracilis, belongs to the following genus Sphwronectes, which possesses a closed tubular hydroecium, open only at the distal end. The other species, Monophyes irregularis, may be retained as the type of the present genus ; it differs from the former in the bilateral arrangement of the four radial canals of the sub- umbrella, and mainly in the shape of the hydroecium, which is not a tubular canal, but an open groove or infundibular cavity.2 This peculiar character is more developed in two other species, which I have myself observed, Monophyes princeps, from the Indian Ocean (PI. XXVII. figs. 13, 14), and Monophyes hydrorrhoa, from the Atlantic Ocean (Canary Islands). The hydroecial groove extends here along the whole ventral side of the bilateral nectophore, and its two edges are prominent as two free wings, one of which overlaps the other more or less. The Atlantic species (Monophyes hydrorrhoa) is very similar to a small Mediterranean form figured in 1885 by Chun, who supposed it to 1 Moiiophyes = Single animal (fiovoa<£«, utiktyh). 2 9, p. 50, pi. iii. fig. 4. REPORT ON THE SIPHONOPHORJE. 131 tained a few detached Eudoxiae of this species, which fully developed represent the monogastric Diplophysa kbllik&ri, Haeckel. Sphasronectes may be derived from Monophyes by concrescence of the two parallel crests or wings, which arise from the ventral side of the nectophore. The hydrcecial groove of the latter becomes converted by this process into a closed tubular hydrcecium, which includes the siphosome. The cormidia, which are attached to the common stem at regular intervals, possess a subspherical bract with a simple vertical phyllocyst, and detached from the stem represent the genus Diplophysa (compare p. 107). Genus 20. Mitrophyes,1 Haeckel, 1888. Mitrophyes, Hkl., System der Siphonophoren, p. 34. Definition. — Monophyidse with a rounded, edgeless, hemispherical or mitriform nectophore, without a true hydrcecium. Trunk free between the exumbrella of the nectophore, and a scutiform or cap-shaped bract, depending from the junction of these two pieces. Bracts spathiform or semi-ovate, without phyllocyst. The genus Mitrophyes was founded by me for an Atlantic Monophyid, which I observed living in the Canary Sea, in January 1867. I observed there two complete specimens, a male and a female. A third specimen (female) was found in the Challenger collection, among other pelagic animals from Station 352. The latter specimen was well enough preserved to enable me to identify it with the former. Mitrophyes differs from all other Monophyidse in the possession of a peculiar bract, which covers the single nectophore like a shield or cap, and in the absence of a hydrcecium, the trunk depending freely between those two pieces and arising from their junction. It may be compared to a Praya or a similar Diphyid, the first nectophore of which is rudimentary and transformed into a bract. Mitrophyes peltij "era, n. sp. (PI. XXVIII.). Habitat.— Tropical and Northern Atlantic, Station 352; April 13, 1876; lat. 10° 55' N., long. 17° 46' W. Surface. Canary Islands, Lanzerote, January 1867 (Haeckel). Nectophore. — The single large nectophore is nearly hemispherical, somewhat oblique, its nectosac being higher in the ventral than in the dorsal half ; it is 6 to 8 mm. long, 4 to 5 mm. high. The voluminous jelly-mantle of the umbrella is twice as thick in the dorsal part as in the ventral. The equatorial diameter of the smooth rounded exumbrella is nearly twice as great as that of the subumbrella, and as the height of the nectophore. 1 Miirop%es = Mitrophorous, animal provided with a mitre, ph^cc, tpvhs. 132 THE VOYAGE OF H.M.S. CHALLENGER. The obliquely egg-shaped subumbrella is much more vaulted in the ventral than in the dorsal half; the diameter of its aperture is about equal to its height. The velum (figs. 1, v, 2, v) is small. The four radial canals of the subumbrella are arranged symmetrically ; the dorsal canal (figs. 1, 2, cd) is shorter, the ventral (cv) longer than the two lateral canals {ex right, cl left). The apical canal (fig. 7, en), which arises from their proximal junction, is short, and passes obliquely through the jelly substance of the top in ventral direction to the somatocyst (cs). Apical Bract (figs. 1, 6, 2, b, 7 b). — The peculiar organ, which we call the apical bract, distinguishes at once Mitrophyes from all other Monophyida?, and from all Caly- conectee in general. It covers the exumbrella of the nectophore in the same manner as a cap covers the head of a man. The bract is a circular concave-convex shield, the diameter of which nearly equals that of the subjacent nectophore. Its morphological signification is difficult to make out ; it may be nothing other than the reduced umbrella of a medusome, which has lost all its other parts ; but it may also be the remnant of the primary nectophore, the place of which is taken by the permanent nectophore afterwards developed. These two pieces are connected by a narrow short pedicle (figs. 2, 7) which is probably the uppermost part of the original trunk. The jelly substance of the bract is rather thin in the peripheral, thicker in its central part, and includes here three short blind radial canals, a longer dorsal (fig. 1, cb) and two smaller lateral canals ; from their junction arises a short bracteal canal (fig. 7, cb) which passes through the pedicle to the somatocyst {cs). Somatocyst (figs. 1, 2, 7, cs). — The somatocyst, or the coryphal cavity of the stem, is a slender conical canal, placed nearly horizontally in the gelatinous umbrella of the nectophore, in its sagittal plane, and directed towards the dorsal side. Its direct continuation towards the ventral side is the axial canal of the trunk. Its proximal apex is closed. From its distal base arise two lateral branches nearly opposite ; proximally the peduncular canal of the bract (fig. 7, cb), and distally the peduncular canal of the nectophore (fig. 7, en). Hydrcecium. — Mitrophyes does not possess a distinct hydrcecium, but it has a very small cavity, which may be considered as the rudimentary homologue of such a " funnel cavity" (figs. 1, 7, ui). This rudiment of a hydrcecium is placed nearly in the apex of the nectophore, and represents a very small funnel-shaped foveola of its exumbrella, which surrounds the origin of the free trunk (a). Siphosome (figs. 1, 2, 7, a). — The common trunk or stem of the siphosome in the expanded state (fig. 1) is a very long thin cylindrical tube, attaining a length of 20 to 30 mm. The internodes between the ordinate cormidia are twice as long as these. In the contracted state (fig. 2) the internodes disappear, and the convoluted stem becomes hidden between the nectophore and bract. The number of the cormidia in the largest specimen observed was between thirty and forty. REPORT ON THE SIPHONOPHOR^E. 133 Cormidia (figs. 3, 5). — The cormidia, or the single "groups of persons," disposed regularly in metameric order, are sessile eudoxomes, the sexual organs becoming ripe on the stem. There are no free Eudoxise developed. In two of the three observed specimens all the eudoxomes were female (figs. 3, 4), in the third specimen male (figs. 5, 6). Mitrophyes, therefore, is one of the rare dioecious Siphonophorae. Each eudoxome is composed of two medusomes, one sterile (siphon with tentacle and bract) and one fertile (the gonophore). Lateral Bracts (figs. 3, b, 5, b). — The bract of each cormidium is an oblongish scale, nearly of the form of a bisected egg. Its proximal part is rounded and attached to the stem (a), its distal part is obtusely pointed. The convex umbrella is smooth. Its sub- umbrellar cavity covers the included siphon and gonophore only partly. There is no phyllocyst or bracteal canal. Siphon (figs. 3, s, 5, s). — The siphon of each cormidium is placed between bract (dorsally) and gonophore (ventrally). Its pedicle is very short, the basigaster (sb) very thickened, nearly spheroidal, with a dense accumulation of cnidocysts. The stomach (sm) is ovate, thick-walled, and includes numerous scattered large cnidocysts (kc) in the exoderm ; its entoderm possesses hepatic strise. The proboscis (sr) is very muscular, cylindrical, with a simple circular mouth-opening (so). Tentacle (figs. 1, 3, 5, t). — The single long tentacle which arises from the pedicle of each siphon bears a great number of tentilla. The cnidosac of each tentillum (fig. 8) is kidney- shaped, and bears at its proximal base only two pairs of large ovate cnidocysts (kg). The terminal filament is about as long as the pedicle of the tentillum (figs. 5, 8). Gonophores (figs. 3, /, 4, female ; figs. 5, h, 6, male). — Each cormidium bears only a single gonophore without accessory sexual bells. They possess the usual shape of medusoid gonophores in Calyconectse, and are about as large as the siphon. The sper- maria (figs. 5, 6, hs) are more longish than the ovaria (figs. 3, o, 4). The umbrella possesses in both sexes four regular radial canals, which are united by a ring-canal at the basal ostium (uo). Genus 21. Cymbonectes,1 Haeckel, 1888. Cymbonectes, Hkl., System der Sipliortophoren, p. 34. Definition. — Monophyidse with an angular, pyramidal nectophore, and an open hydrcecial groove on its ventral side ; the latter includes the siphosome, which is incompletely protected by two overlapping lateral wings. Bracts spathiform, with an open ventral fissure, and a simple ovate phyllocyst. The genus Cymbonectes has hitherto been known by a single species only, described in 1859 by Huxley as Diphyes mitra, and taken only once in the Indian Ocean.2 1 Cyiribonectes = Swimming boat, Kv^onif-Tn;. "■ 9, p. 36, pi. i. fig. 4. 134 THE VOYAGE OF H.M.S. CHALLENGER. That author observed only a single nectocalyx, and supposed that it might be a young and imperfect Diphycs. But I find the same form in different bottles from the Challenger, taken in the Pacific, and also in the collection of Captain Eabbe. One nectophore only (the first or proximal) is always present, whilst there is no trace of a second or distal nectophore. I am therefore convinced that this form is a true Monophyid, not a Diphyid, and this the more as the peculiar character of this group is yet more distinct in another Indian species, Cymbonectes huxleyi; I observed this species, described in the following pages, during my stay in Ceylon. A third species, Cymbonectes cymba, inhabits the Atlantic Ocean, and will be described afterwards. Cymbonectes has no complete hydrcecium, but an open infundibular groove on the ventral side of its nectosac ; it agrees in this respect with the genus Monophyes (sensu stricto, compare p. 128), but it differs from this in the pyramidal form of its angular nectophore. AVhUst in Belligemma I succeeded in observing the development of the fertilized egg of Cymbonectes huxleyi; it is very similar to that of Galeolaria aurantiaca, described by Metschnikoff (84, Taf. vi., vii.). The four most important stages of it are figured in PI. XXVII. figs. 9-12. Cymbonectes huxleyi, n. sp. (PI. XXVII. figs. 1-12). Habitat. — Indian Ocean; Belligemma, Ceylon, December 1881 (Haeckel). Nectophore (fig. 1, lateral view from the right side ; fig. 2, dorsal view ; fig. 3, ventral view ; fig. 4, transverse section through the middle part). — The single nectocalyx is helmet-shaped or slenderly campanulate, 6 to 7 mm. long, 2 to 3 mm. broad ; it is some- what broader in the upper than in the lower half. The exumbrella has five prominent, elegantly denticulate edges which unite above in the pointed apex, and end below in the median crests of five triangular teeth surrounding the basal mouth. The five edges of the nectophore are arranged as in Dvphyes, one odd running along the dorsal median line (near the nectosac), two lateral corresponding to the two lateral canals of the latter, and two ventral forming the edges of the hydrcecial canal. From the base of these latter arise in the lower half of the nectophore two broad triangular wings, the larger left of which overlaps the other and thus incompletely closes the hydrcecial groove (fig. 4) ; the free edges of these wings are strongly dentate (fig. 3). The bases of the wings are continued above the basal ostium of the nectophore, and here form on its ventral side two broad ovate basal lobes with elegantly denticulate edges. These lobes support the siphosome proceeding from the basal mouth of the hydrcecial canal. Nectosac (fig. 2, w). — The subumbrellar cavity is ovate, twice as long as broad, and REPORT ON THE SIPHONOPHORJE. 135 broader in the upper than in the lower half. It occupies the two distal thirds of the nectophore, whilst its proximal third is taken by the soniatocyst (cs). Its ventral side is separated by a thin frontal septum from the adjacent hydrcecium. The basal mouth of the nectosac is obliquely truncate and surrounded by five strong triangular teeth, the distal ends of the five exumbrellar crests. The two ventral teeth (as the terminal lobes of the two hydrcecial plates) are about four times as large as the three other teeth, which form an odd dorsal point and two paired lateral points. Canals of the Nectosac. — The nectocalycine duct, which arises from the top of the stem (aa), descends nearly to the middle of the dorsal median line of the subumbrella, and divides here into four radial canals, two odd and two paired. The odd ventral canal (cv) is the shortest, and descends immediately straight to the ostium of the nectosac. The odd dorsal canal (cd) is the longest ; it ascends, in the dorsal median line, to the apical top of the nectosac, and then descends downwards along its whole ventral median line. The two paired lateral canals (right ex, and left cl) are intermediate in length between the former and the latter, and have a strongly bent course. They run firstly ascending towards the dorsal side, form in the upper half of the nectosac a nearly circular loop, and then are turned ventrally and downwards. The four radial canals are united at the ostium of the nectosac by a circular canal, which embraces the velum (v). Hydrcecium (figs. 1-4, ui). — The funnel-cavity of the nectophore, into which the contracted siphosome may be partly retracted, is a long, nearly cylindrical and slightly bent canal, which occupies the two distal thirds of the ventral half of the umbrella. It is separated from the dorsally adjacent nectosac by a thin frontal septum, and incom- pletely closed on its ventral side by the two triangular ventral wings of the exumbrella overlapping one another (figs. 3, 4, nx right, nl left). Its basal ostium is protected at the dorsal side by the two terminal lobes of the ventral crest. The apex of the hydrcecial canal touches the base of the somatocyst. Soniatocyst (cs). — The axial canal of the tubular stem is prolonged above its apex into a pyriform cavity, nearly filled up by large vacuolated entoderm-cells. This is the pyriform somatocyst, which encloses in its dilated uppermost part a hydrostatic oil- globule (co). It occupies the uppermost or apical third of the nectophore, and is far prominent over the top of the nectosac. It is twice as long as broad, and nearly one- third as long as the nectophore. Siphosome. — The common tubular stem, which arises in the closed apex of the hydrcecium, at the base of the somatocyst, proceeds through the basal ostium of the former to a considerable length. It is beset by a series of numerous buds in the upper part, and eight to twelve fully-developed cormidia in the lower part. These are alternating male and female, separated by equal free internodes (fig. 1). Cormidia (figs. 6, 7). — Each cormidium, or " group of individuals," is a eudoxome, composed of two medusomes, a sterile and a fertde. The sterile Medusa exhibits a 136 THE VOYAGE OF H.M.S. CHALLENGER. spathiform bract, and hidden in its cavity a siphon with its tentacle. The fertile medusome, placed at the ventral side of the former, is a medusiform gonophore, either male or female. Bract (figs. 5, 6, b). — The protecting rudimentary umbrella or hydrophyllium is similar to that of Diphyes, spathiform, or irregularly conical, with a deep ventral fissure, and an obliquely truncate base. Its structureless jelly-plate is very thin in the lower half, thick in the upper half, which encloses a large pyriform phyllocyst. This contains large vacuolate entoderm cells, and an oil-globule in its apex. The truncate base of the bract has four corners, two ventral smaller and two dorsal larger triangular lobes. Siphon (figs. 5-7, s). — The single polypite which is attached on the top of the sub- umbrellar cavity of the bract has the formation usual in the Calyconectse, a short pedicle, an ovate basigaster with very thick exoderm, full of cnidocysts (sb), a utricular stomach with thick entoderm (sm), and a very protractile proboscis (sr) ; the distal mouth- opening of the latter may be expanded in form of a circular suctorial disc (fig. 5, ss). Tentacle (figs. 5-7, t). — The single tentacle which arises from the pedicle of the siphon is of medium size, and bears a row of ten to fifteen or more tentilla. Each of these lateral branches bears upon its club-shaped pedicle a reniform cnidosac (fig. 8, km). This includes on its convex side about four to six longitudinal rows of smaller paliform cnidocysts, and at its base, on both sides, three very large ensiform cnidocysts ; the terminal filament (tf) is usually coiled up, and armed at the distal end with a hemi- spherical group of pyriform cnidocysts, provided with long cnidocils (kp). Gonophores (figs. 5, 6). — The sexual medusoids alternate regularly in the cormidia, so that each two neighbouring ones form together a diclinic pair. The male eudoxomes (fig. 6) are nearly of the same shape as the female (fig. 5). Each gonophore is an ovate or club- shaped sac, placed at the ventral side of the siphon. The rudimentary umbrella, which possesses the usual four radial canals, and the uniting marginal canal, embraces closely the manubrium. The spadix or central canal is large in the spermaria (fig. 6, hx), small or rudimentary in the ovaria (fig. 5,f). Genus 22. Muggisea,1 Busch, 1851. Muggisea, Busch, "Wirbellose Seethiere, &c, 67, p. 48. Definition. — Monophyidse with an angular pyramidal nectophore and a complete infundibular hydrcecium on its ventral side. Bracts spathiform or conical, with a deep ventral groove, a bevelled basal face, and a simple ovate phyllocyst. The genus Muggiwa was founded in 1851 by Busch (67, p. 48) for a Mediterranean Monophyid, which Will had figured seven years before uuder the name Diphyes kochii (65, p. 77, Taf. ii. fig. 22). The identity of these two forms was demonstrated in 1882 1 Mtiggixa = Inhabitant of the port of Muggia, near Trieste. REPORT ON THE SIPHONOPHORJE. 137 by Chun, who called it Muggisea kochii (86, p. 1157, Taf. xvii. fig. 2). Chun demon- strated that this poly gastric Calyconecta is a true Monophyid, and that the cormidia, arising from the common stem, become detached and sexually developed as the free monogastric generation, which was described by Busch under the name Eudoxia eschscholtzii.1 Chun also gave the full description of its ontogeny, and found that the larva, arising from the fertilised egg of Eudoxia eschscholtzii, does not possess the pentagonal pyramidal nectophore of Muggisea, but the edgeless campanulate nectophore of Monophyes ; the latter afterwards buds from the base of the former, and remains when the former is detached. Chun supposed, therefore,, that three different genera- tions should be distinguished in this species — (l) Monophyes pyramidalis (85, fig. 1), (2) Muggisea kochii (fig. 2), and (3) Eudoxia eschscholtzii (fig. 3). I cannot agree with this opinion, but I regard the first form (fig. 1) only as the larva of the second (fig.. 2). The primary edgeless nectophore of Monophyes is only a provisional larval organ, and the fact that it is afterwards replaced by the secondary five- edged pyramidal nectophore of Muggisea may be explained by the fundamental law of biogeny — by the hypothesis that Monophyes is the original ancestral form of Muggisea. The mature Eudoxia of Muggisea is very similar to the monogastric genus Cucullus, the Eudoxia of Diphyes. It differs in the rounded and edgeless surface of the conical or spathiform bract, which has three or five edges in Cucullus. In respect of this difference, the name Cucubalus (given in 1824 by Quoy and Gaimard, 24) may be retained for it. The spathiform bract is obliquely conical, with a deep ventral groove, rounded dorsal convexity, pointed apex, and simple phyllocyst (compare above, Genus 11b, p. 109). The free Eudoxia of the Mediterranean Muggisea kochii may, therefore, bear the name Cucubalus eschscholtzii. A second species, slightly differing from the Mediterranean one, was observed by me in the Canary Island Lanzerote, and may retain the name Muggisea pyramidalis; it differs from the former mainly in the size of the conical hydrcecium, the top of which attains half the length of the nectosac. The free Eudoxia of this Atlantic species has a conical bract, with a blunt apex and an ovate larger phyllocyst ; it rnay be called Cucubalus pyramidalis. A third species of Muggisea is probably the Tropical Pacific form, described by Huxley in 1859 as Diphyes chamissonis.2 It agrees with Muggisea pyramidalis in the size of the high hydrceciurn, but differs from this Atlantic and from the Mediterranean species in the more campanulate form of the nectophore, the denticulate shape of its edges, and the stronger teeth of its mouth. The free Eudoxia of this Pacific species may, perhaps, be Cucubalus cordiformis of Quoy and Gaimard.3 Muggisea differs from the preceding 1 67, p. 33, Taf. iv. figs. 7-10 ; Taf. v. figs. 1-9. 2 9, p. 36, pi. i. fig. 3. 3 2, p. 94, pi. iv. figs. 24-27 ; 24, pi. vi. fig. 1. (ZOOL. CHALL. EXP. — PART LXXVII. — 1888.) Hhhh 18 138 THE VOYAGE OF H.M.S. CHALLENGER. Cymbonectes mainly in the complete infundibular cavity of the hydroecium, and, there- fore, bears to it the same relation that Sphasronectes exhibits to Monophyes. Genus 23. Cymba,1 Eschscholtz, 1829. Cymba, Esch., System der Acaleplien, p. 133. Definition. — Monophyidse with an angular, pyramidal nectophore, and a complete in- fundibular hydroecium on its ventral side. Bracts cuboidal, with six quadrangular faces, and two lateral lobes arising from the base of the phyllocyst. The genus Cymba of Eschscholtz comprised in the system of its founder three different species, the first of which [Cymba sagittata) belongs to a different genus of Calyconectae (it is an Abylid). The second species, which I retain as the type of our genus, is the Mediterranean Cymba enneagonum, Esch. ( = Enneagonum hyalinum, Quoy et Gaimard, 2, pi. v. figs. 1-6). The monogastric generation, or the Eudoxia of this polygastric Monophyid, is the third species of Eschscholtz, Cymba cuboides ( = Cuboides vitreus, Quoy et Gaimard, not Huxley !). Different from this typical and oldest known Mediterranean form is a second, Austra- lian, species, which Huxley described very accurately in 1859, under the name Abyla vogtii (9, fig. 46, pi. ii. fig. 3). He rightly suspected that his Cuboides vitreus (not identical with that of Quoy et Gaimard) might be the detached Eudoxia of the former ; he found both together at the same place on the south coast of New Guinea. A third species, Cymba nacella, was found by me in 1881 in the Indian Ocean, and will be described on another occasion. It is more similar to Cymba vogtii ( = Abyla vogtii, Huxley) than to the two western species. Its Eudoxia is Cuboides nacella. The fourth species, here described as Cymba crystallus, was observed living by me in the Canary Islands in 1867, and from these living specimens are taken the figures on Pis. XLI. and XLII. (compare above, p. 111). The same form occurred in the Challenger collection, having been taken in the Guinea current at Station 348. The remarkable cuboidal form of the bracts, and the bilobate horizontal diverticulum of their phyllocyst, distinguishes Cymba at once from all other Calyconectae. Cymba crystallus, n. sp. (Pis. XLI., XLII.). Abyla crystallus, Haeckel, 18G7, MS. Canar. Habitat. — Tropical and Subtropical Atlantic, Station 348; April 9, 1876: lat. 3° 10' N., long. 14° 51' W. Surface. Canary Islands, Lanzerote, February 1867 (Haeckel). 1 Cymba = Boat, xv/*ISri. REPORT ON THE SIPHON OPHORiE. 139 Nectophore (fig. 1, basal view; fig. 2, apical view; fig. 3, dorsal view; fig. 4, ventral view ; fig. 5, lateral view from the left side). — The single nectophore (or the apical nectocalyx) has a diameter of 10 to 15 mm., and a complicated polyhedral form. In that position which is regarded as the normal in figs. 3, 4, and 5, the two parallel axes of the nectosac and of the hydrcecium stand vertically. The umbrella of the nectophore appears composed of a quadrilateral pyramid in the apical half, and of a poly- gonal prism in the basal half. The apical view of the nectophore (fig. 2) exhibits a rather regular quadrilateral pyramid. Its four perradial edges (the dorsal nd, the opposite ventral or coryphal nk, and the two lateral, right and left) meet in the central apex of the pyramid at right angles. The four faces separated by them are irregularly rhombic, the two dorsal (left ud', and right ud") and the two apici-ventral faces (left ua', and right ua"). Each of these four faces is slightly concave, and in the middle between each two pyramid, edges is pro- longed into a descending pyramidal apophysis ; these four basal apophyses alternate regularly with the four pyramid edges, and are of course interradial. The basal view of the nectophore (fig. 1) shows that these four interradial basal apophyses are three-sided pyramidal, and that their inferior crests or basal edges are directed radially towards the centre of the basal face. The centripetal end of the basal edge of the two dorsal apophyses passes over directly into the two lateral teeth of the mouth of the nectosac, whereas the centripetal end of the basal edge of the two ventral apophyses finally passes over into the two smaller dorsal teeth of the mouth of the hydrceciuni. The four basal faces of the nectophore, which are separated by those four interradial basal crests, are also concave, and have a bilateral-rhombic, or, strictly speaking, pentagonal form. The ventral basal face (uv) is somewhat smaller than the dorsal basal face (ug), and the two pentagonal lateral basal faces are intermediate between them. The ventral view of the nectophore (fig. 4) therefore exhibits the smaller ventral basal face (uv) surrounded by four larger faces (fore-shortened) ; the two ventral basal faces (ux right, ul left) and the two ventral apical faces (ua" right, ua' left). The somato- cyst (cs), with its apical oleophore (co), and beyond it the hydrcecium (ui), with the included siphosome, appear through the ventral wall of the nectophore in this view (fig. 4). The basal opening of the hydrcecium (ui) is surrounded by four serrate trian- gular teeth, two smaller ventral and two larger dorsal. The dorsal view of the nectophore (fig. 3) exhibits the two large quadrilateral dorsal apical faces (ud' left, and ud" right), and beyond them the pentagonal dorsal basal face of the nectosac may be seen through the dorsal wall of the nectophore, and above it the oleocyst (fig. 3, co). Nectosac (fig. 6, w, lateral view, right side ; fig. 5, lateral view, left side ; fig. 3, w, dorsal view ; fig. 2, w, apical view ; fig. 1, iv, and 7, basal view). — The nectosac, or the muscular subumbrella of the nectophore, includes a slenderly ovate cavity, which in the 140 THE VOYAGE OF H.M.S. CHALLENGER. contracted state is three to four times as high as broad. Its height equals half the height of the complete nectophore. Its dorsal face is more convexly vaulted than the ventral, which is nearly in contact with the dorsal wall of the hydrcecium. The canal of the nectophore (fig. 6), which arises from the constriction between somatocyst (cs) and hydrcecium (ui), descends nearly vertically to the centre of the ventral median line of the nectosac, and here gives off the four radial canals, which have very different lengths. The shortest is the ventral canal of the nectosac (fig. 6, cv) ; it extends only through the basal half of the nectosac. The longest, on the contrary, is the opposite dorsal canal (cd) ; it runs from the basal circular canal (cc) along the whole dorsal side (in the dorsal median line of the nectosac), and then from its top downwards through the apical half of the ventral line. The two symmetrical lateral canals (ex right, and cl left) are equal and form an S-shaped loop ; they ascend from the basal circular canal (cc) in the dorsal half of the nectosac vertically, send off into its apical third a slender caecal sac (cy), and then descend in the ventral half, meeting each other and the two other canals near to the centre of the ventral line. The mouth of the nectosac (figs. 1, no, and 7, no, basal view; fig. 8, no, lateral view from the right side) is surrounded by a broad velum (v) and armed with five serrate teeth of nearly equal size, the odd dorsal tooth (fig. 8, nd) is perradial ; the two paired dorso-lateral teeth (n1 left, n2 right) give off the two dorsal basal crests of the nectophore ; finally, the two paired ventrolateral teeth (fig. 8, n3 left, n4 right) may be regarded as separated branches of a forked (originally odd) ventral tooth. Hydrcecium (figs. 1 and 7, ui, basal view; fig. 2, ui, apical view; fig. 4, ui, ventral view ; fig. 5, ui, lateral view from the left side ; fig. 6, ui, from the right side). — The hydrcecium or the funnel cavity of the umbrella completely includes the retracted siphosome ; it is slenderly campanulate or nearly cylindrical, and about as large as the nectosac, which is placed at its dorsal side. The vertical main axes of these two organs are nearly parallel ; but the apical half only of the hydrcecium is in immediate contact with the basal half of the nectophore, whilst the basal half of the former projects freely beyond in the form of a short and wide campanulate tube (figs. 5, 7, 8, ui). The basal mouth of the latter is the lowermost base of the entire nectophore. and is armed wTith two pairs of serrate teeth, the two teeth of the dorsal pair being larger than those of the ventral pair (figs. 7, 8). Somatocyst (fig. 4, cs, ventral view ; figs. 5, 6, cs, lateral view). — The somatocyst or the coryphal cavity is ovate or spindle-shaped, about half as long and broad as the hydrcecium. It is placed at the apex of the latter, in the apical prolongation of its vertical main axis, and is nearly filled with large polyhedral entoderm cells. Its dorsal wall is in contact with the upper half of the ventral wall of the nectosac. Its apex bears a vertical oleocyst, containing a large ovate oil-bubble, about one-third as long and broad as the somatocyst. REPORT ON THE SIPHONOPHOR^E. 141 Siphosome (figs. 4-6, i). — The common trunk bears In its upper part very numerous buds of cormidia (i), in its lower part ten to twelve or more well-developed eudoxomes. These are soon detached from the stem and swim about as free Eudoxise of the special form represented in PI. XLII. as Cuboides crystallus (compare their description above, p. 112, Genus 13). When fully contracted the entire siphosome, with all cormidia, is hidden in the hydrcecium (figs. 4-6). Family VII. Diphyid^e, Eschscholtz, 1829 (sensu restricto). Diphyulse, Esch., System der Acaleplien, 1, p. 122. Definition. — Calyconectae polygastricsB, with two nectophores at the apex of the long- tubular truuk. Cormidia ordinate, eudoxiform, separated by equal free internodes, each siphon with a bract. The family Diphyidse, as defined in my system, comprises only those polygastric Calyconectse which bear two permanent nectophores on the top of the stem. I exclude, therefore, those forms, formerly united with them, which possess only a single nectophore (Monophyidse). I exclude, further, on practical grounds, the monogastric independent forms, which are connected with the Diphyidae by metagenesis, the so-called Diphyo- zooids, the families (IV. and V.) Eudoxidse and Ersseidas. The Diphyidae are very common in all the seas of the world, far more frequent than all other Siphonophorae, and richer in different species than the other Calyconectae. The first Diphyid was described in 1804 by Bory under the name Biphora bipartita (13, vol. i. p. 134). Cuvier founded for it the genus Diphyes (in 1817, 91), and Chamisso figured the same as Diphyes dispar (16, p. 365, Tab. xxxiii. fig. 4). The naturalists of the "Astrolabe." Quoy and Gaimard, discovered in 1826, in the Straits of Gibraltar, a greater number of Diphyidse, and distinguished in this family six different genera: — 1. Diphyes (campanulifera) ; 2. Calpe (jpentagona) ; 3. Abyla (trigona) ; 4. Cymba (sagittata) ; 5. Enneagonum (hyalinum) ; and 6. Cuboides (vitreus) (20). These and some other Diphyidse were described and figured by the same authors in 1833 in the Zoophytes de l'Astrolabe (2, pp. 81-106) as fifteen different species of the genus Diphyes, nearly every species of which is now the type of a special genus. Eschscholtz (1, p. 122) gave the first accurate description and a better systematic arrangement of the Diphyidse ; they form, in his System der Acalephen, the first of the three large families of Siphonophorse. He distinguished six genera ; three of these (Eudoxia, Ersasa, Aglaisma) are monogastric, and form now our family Eudoxidre (p. 103); the three others (Abyla, Cymba, Diphyes) are polygastric ; one of these (Cymba) is according to my observations a Monophyid, so that only Diphijes and Abyla remain as types of true Diphyidae. 142 THE VOYAGE OF H.M.S. CHALLENGER. Blainville, in his Actmologie (24, pp. 125-140), and Lesson, in his Acalephes (3, pp. 425-465), collected the scattered descriptions and figures of the older observers, and distinguished a greater number of genera, but without any clear anatomical under- standing and without critical judgement. The excellent naturalists, who, in the sixth decade of our century (1853 to 1859), did so much for the anatomical knowledge of Siphonophorae, gave also the first accurate description of the typical Diphyidse, mainly Diphyes and Abyla (4-10). Kolliker (4, Tab. ix.-xi.) gave an excellent description of three Mediterranean types — Praya diphyes, Diphyes sieboldii, and Abyla pentagona. Vogt (6, Tab. xvi.-xxL) gave beautiful drawings of the same forms, and also of Galeolaria aurantiaca. But the greatest progress in the knowledge of Diphyidse, mainly regarding their development and connection with Eudoxidse, was made by Gegenbaur (7 and 10) and by Leuckart (5 and 8). The former described, too, a greater number of new species (of Praya, Diphyes, and Abyla, 10). The most complete anatomical and systematic description of the polygastric Diphyidse, and of their offspring, the monogastric Eudoxidse, as alsD the best and fullest account of the whole family up to our days, was given in 1859 by Huxley (9, pp. 30-66, pis. i.-v.). He restricted the family Diphyidae to the genera Diphyes and Abyla in the sense of Eschscholtz, and separated them from Praya, as the type of another family, Prayidse (Kolliker, 4, p. 33). He gave, further, the first accurate descrip- tion of numerous Diphyozooids (or Eudoxidse), of seven different genera, and indicated probable ontogenetic connection with different forms of Diphyes and Abyla. During my residence in the Canary Islands, from December 1866 to February 1867, I had the opportunity of examining typical representatives of all the eight genera of true polygastric Diphyidae which are described in the sequel, and there I drew from nature the figures, which will be seen in Pis. XXXI. to XLII. of this Report. The greater number of the Diphyidse, there observed by me, were afterwards found again in the collection of the Challenger, mainly in bottles containing surface animals, which were taken in the Tropical and Subtropical Atlantic (Stations 334 to 354; March 14 to May 7, 1876). In Lanzerote I observed directly the metagenesis of Diphyes (with Cucullus), Diphyopsis (with Erssea), Abyla (with Amphiroa), Bassia (with Sphenoides), and Calpe (with Aglaisma). Nectophores. — The two nectocalyces, which, in all Diphyidse, are placed at the top of the stem, appear in three different stages of phylogenetic development, and these deter- mine the division of the family into three divergent subfamilies. The first subfamily, Prayidse, has two nectophores of nearly equal size and similar form, opposed to one another ; sometimes the first is somewhat smaller than the second ; their surface is rounded, the jelly-substance very soft. Their shape is either mitriform or reniform [Praya, PI. XXXI.), or more hemispherical (Lilyopsis). REPORT ON THE SIPHONOPHORiE. 143 The two nectophores of the second subfamily, Diphyopsid.se (PL XXXIII.), are also of nearly equal size and similar form ; but they are pyramidal, pentagonal, and placed one behind the other. Their junction is very loose in Galeolaria, whilst in Diphyes and Diphyopsis the apical part of the second is hidden in the hydrceciurn of the first nectophore. The sharp edges are often elegantly denticulate, and the hyaline jelly-substance of the umbrella is rather hard and firm, cartilaginous, as also is that of Abylidse (Pis. XXXV. -XL.). The differentiation of the two nectophores attains the highest degree in the third sub- family, Abylidse. The first (proximal or apical nectophore, often also called superior or anterior) is here always symmetrical and much smaller than the second (distal or basal nectophore, often called inferior or posterior) ; this is more or less asymmetrical. Both nectophores are here polyhedral, prismatic, or truncate-pyramidal, with numerous poly- gonal faces and sharp prominent edges. The form of the second nectophore and its basal ostium is especially characteristic ; it offers three prominent wings in Abyla (trigona), four in Bassia (tetragona), and five in Calpe (pentagona). Canals of the Nectophores. — Each of the two nectophores constantly possesses four radial canals in the subumbrella, which are united above the velum by a circular canal. The size, course, and form of the four vessels are very variable, according to the place of their apical junction, where the nectocalycine duct, coming from the top of the stem, enters into the subumbrella. This point of junction is usually placed not at the apex of the nectosac but in its ventral median line, more or less dislocated downwards, so that the ventral canal (cv) is shorter, and the dorsal (cd) longer than the two symmetrical lateral canals (ex right, cl left) ; the latter are often more or less curved or loop-shaped. The ventral canal is very short, rudimentary, or even lost, in the first nectocalyx of Galeolaria, because here the point of junction has quite descended, and the nectocalycine duct enters into the base of the subumbrella, instead of into the apex. The opposite dorsal canal is so much the longer. Hydrceciurn. — The differences which the hydrceciurn offers in the Diphyidse, have been already mentioned above (p. 93). The Prayidse are distinguished by a cylindrical hydrcecial canal open at both ends, composed of the ventral grooves of the two opposite nectophores, fitting one into another. Galeolaria has no true hydrceciurn, since the apex of the second nectophore is simply attached to the base of the first, and the siphosome is suspended freely between them. All other Diphyidse have a conical or campanulate hydrcecial cavity on the ventral side of the first nectophore, and as its continuation, a hydrcecial canal on that of the second ; this is sometimes an open groove, protected by two overlapping wings, at other times a closed canal, produced by concrescence of the two wings. Somatocyst (cs). — The acrocyst or somatocyst (" Saftbehalter ") is wanting in the Prayidse and in Galeolaria ; it may be replaced in the former by the ascending pallial 144 THE VOYAGE OF H.M.S. CHALLENGER. canal. All other Diphyidse possess a somatocyst in the first or apical nectophore ; it must be regarded as the uppermost part of the original common stem, overgrown and enclosed by jelly-substance of the first nectophore. The somatocyst is usually rather large, spindle-shaped, or ovate, sometimes more cylindriaal, at other times more ovate. Usually it ascends from the apex of the hydrcecium ; but in Abyla it descends along its ventral side. The structure is the same as in the other Calyconectse (compare above, p. 93). Siphosome. — The tubular trunk or common stem of the Diphyidae is very contractile, and beset at regular intervals with the cormidia, the number of which is very variable. The stem is very long in the lower and older forms of the family, in the Prayidse and Galeolaria, where it sometimes attains a length of more than one metre, and bears more than one hundred cormidia. Their size arjd number are much smaller in the specialised Abylidse, intermediate in the Diphyidae. The contracted stem may usually be retracted more or less completely into the hydrcecium. The structure of the stem is described above (p. 94). Cormidia. — The cormidia of the Diphyidse, or the Diphyozooids of Huxley (9, pp. 57-66), occur in two different principal forms, eudoxomes and ersseomes. The majority of the genera possess eudoxomes ; each cormidium is composed of a sterile medusome (bract with siphon and tentacle) and a fertile medusome -(gonophore). The two genera Lilyopsis and Diphyopsis possess ersseomes, a sterile special nectophore, as locomotive person, being added to the euxodome. Bracts. — The bracts or hydrophyllia are of very different form and structure, characteristic of the single genera and even of the three subfamilies. The bracts are mitriform and rounded in the Prayidse, spathiform or conical in the Diphyopsidse, prismatic or polyhedral in the Abylidse. Besides, the form and place of the phyllocyst, and the number, form, and course of the radial canals which arise from its base, exhibit characteristic differences in the various genera. Siphon and Tentacle. — The form and structure of the polypites exhibit no important differences in the cormidia of the various Diphyidae. The structure, too, of the tentacles is in general the same ; but the special form of the tentilla, and especially the composition of the cnidosacs and the arrangement of their different cnido- cysts, are subject to many specific variations. Eudoxiee. — The minority of the Diphyidse produce sessile eudoxomes, which maturate whilst attached to the stem. This is the case in some of the Prayidse and in Galeolaria. In all the other Diphyidas they become early detached from the stem, and maturate as free Eudoxise (compare above, p. 101). Ontogeny. — On the development of the Diphyidse, compare above, pp. 100-102. REPORT ON THE SIPHONOPHORiE. 145 Synopsis of the Genera of ' Diphyidse. III. Subfamily Abylim:. Two nectophores of very different size and form, one placed behind the other ; both prismatic, with polygonal faces and sharp edges. I. Subfamily PRAYID.E. Two nectophores of nearly equal size and similar form, opposite to one an- other, edgeless, rounded. II. Subfamily Diphyopsid.e. Two nectophores of nearly equal size and similar form, one placed behind the other, pyramidal, five-edged. Bracts of the cormidia hemispherical, cup- shaped or helmet- shaped, edgeless, round- ed. Phyllocyst with four radial canals. Bracts of the cormidia pyramidal, conical, or spathiform, with a pointed apex, and an obliquely bevelled basal face and a deep ventral fissure. Phyllocyst ovate, simple or rudi- mentary, without radial canals. Bracts of the cormidia six- sided prismatic. Phyl- locyst descending, with two lateral canals at its apex. Bracts of the cormidia four-sided prismatic, below wedge-shaped.