" s N X o- X- 5i i o ! m ; _D m a N k 5; NATIONAL ACADEMY OF SCIENCES. VOL. V. FOURTH MEMOIR. THE EMBRYOLOGY AND METAMORPHOSIS OF THE MACKOURA. \ V VOL. FOURTH MEMOIR. NATIONAL ACADEMY OF SCIENCES. THE EMBRYOLOGY AND METAMORPHOSIS OF THE MACROURA. 319 THE EMBRYOLOGY AND METAMORPHOSIS OF TBTE MACEOURA ^V. Iv. BROOKS, PROFESSOR OF ANIMAL MORPHOLOGY IN THE JOHNS HOPKINS UNIVERSITY, AND F. H. HERRICK, PROFESSOR OF BIOLOGY IN ADELBERT COLLEGE, LATE FELLOW IN THE JOHNS HOPKINS UNIVERSITY. WITH FIFTY-SEVEN PLATES. S. Mis. 94 21 CONTENTS. I. Introduction. II. Tbe life history ofStenopus hi.spiilus. Srrtion 1. Natural history "I Steuopus. Section 2. Segmentation and tlie early Section 3. Metamorphosis of the larva. Section 4. The adult. Remarks. List of species. Literature of Steuopns. III. The habits aud metamorphosis of Gouodactylus chi- ragra. .Section 1. The structure aud habits of the adult. Section '2. Metamorphosis. IV. The metamorphosis of Alpheus. Section 1. The metamorphosis of Alpheus minus. Section 2. The metamorphosis of Alpheus hete- rochelis in the Bahama Islands. Sections. The metamorphosis of Alpheus hete- rochelis at Beaufort, North Carolina. Section 4. The metamorphosis of Alpheus hete- rochelis at Key West, Florida. Section 5. The larval development of Alphens saulcyi. V. Alpheus : A study iu the development of the Crus- tacea. Introduction. Methods. PART FIRST. Section 1. The habits ami color variations of Alpheus. [With fifty-seven plates.] PART FIRST — Continued. Section 2. Variations in Alpheus heterochelis. Section 3. The abbreviated development of Alpheus and its relation to the environment. Section 4. The adult. Section f>. Variations from the specific type. Section G. Measurements. Section 7. The causes and significance of varia- tion in Alpheus saulcyi. PART SECOND. Development of Alpheus. Section 1. Structure of the larva of Alpheus saulcyi. Section '2. The origin of ovarian eggs in Alpheus, Homarus, and ralinurus. Section 3. Segmentation iu Alpheus miuox Section 4. The embryology of Alpheus. Stages 1-13. Section 5. Notes on the- segmentation of Crusta- cea. Section 6. Cell degeneration. Section 7. The origin aud history of wandering cells iu Alpheus. Section 8. The development of the nervous sys- tem. Section 9. The eyes. Section 10. Summary. Section 11. References. Explanation of figures (accompanying each plate). 323 CHAPTER I. INTRODUCTION. By W. K. BROOKS. No great group of animals is more favorable than the Crustacea for the study of the history and significance and origin of larval forms, for these animals possess a number of peculiarities which serve to render the problem of their life history both unusually interesting and significant, and at the same time unusually intelligible; nor are these peculiar features exhibited, to the same degree, by any other great group of animals. The body of an arthropod is completely covered, down to the tip of each microscopic hair, by a continuous shell of excreted matter, and as this chitinous shell is not cellular it can not grow by the interpolation of new cells, nor can it, like the excreted shell of a mollusk grow by the dep- osition of new matter around its edges, for there are no such growing edges, except in a few exceptional cases, such as the barnacles. Once formed and hardened the cuticle of an arthropod admits no increase in size, and as soon as it is outgrown it must be discarded and replaced by a new and larger one. The new shell is gradually excreted, in a soft condition, under the old one, and as soon as this is thrown off the new one quickly becomes fully distended and solid. As a result, from the very nature of the chitiuous shell and the method of renewal which its structure entails, the growth of an arthropod, from infancy to the adult condition, takes place by a series of well-marked steps or stages, each one characterized by the formation of a new cuticle and by a sudden increase in size. In most arthropods the newly-born young are very different in structure from the adults, and growth is accompanied by metamorphosis. As the changes of structure are necessarily confined to the moulting periods, the stages of growth coincide with the stages of change in organization, and there is none of the indefiniteness which often characterizes the different larval stages of animals with a more continuous metamorphosis. On the contrary the nature of each change is as sharply defined and as characteristic as the structure of the adult itself. As the moulting period is fre- quently a time of inactivity the animal may then undergo profound changes without inconvenience, and the successive steps in the metamorphosis of an arthropod are not only well marked, but often very profound as well. In these features all the other arthropods are like the Crustacea, but another consideration, the fact that, with few exceptions, the higher Crustacea are marine, renders the problem of their life history much more intelligible than that of any other class of animals. So far as the ontogenetic history of the metamorphosis of a larva is a recapitulation of ances- tral stages in the evolution of the species its retention at the present day must depend to a great degree upon the persistency of those external conditions to which the larval stages were originally adapted. This is true at least of all free larvse, which have their own battle to fight and their own living to get, and while a larva inside an egg or within a brood pouch may possibly recapitulate obsolete ancestral stages, the survival of a free larva depends upon its adaptation to its present environ- ment. As compared with the ocean the inorganic environment of terrestrial or fresh-water animals is extremely variable, and changes in climate, elevation, and continental configuration are accom- panied by corresponding changes in enemies, competitors, and food, so that the conditions which 325 326 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. surround a modern terrestrial larva must, in nearly every case, be very different from those under which the remote ancestors of the species passed their life, but while this is also true, to some degree, of marine animals their inorganic environment is comparatively stable, and the persistence of so many ancient marine types shows that the changes in the organic surroundings of marine animals take place much more slowly than corresponding changes on land. This fact, joined to the definite character~bf the changes which make up the life history of a marine crustacean, renders these animals of exceptional value for the study of the laws of larval development, and for the analysis of the effect of secondary adaptations, as distinguished from the influence of ancestry ; for while Clans has clearly proved that adaptive larval forms are much more common among the Decapods than had been supposed, his writings and those of Fritz Muller show that no other group of the animal kingdom presents an equal diversity of orders, families, genera, and species in which the relation between ontogeny and pliylogeny is so well displayed, but, while proving this so clearly, Claus' well known monograph also shows with equal clearness that this ancestral history is by no means unmodified, and that the true significance of the larval history of the higher Crustacea can be understood only after careful and minute and exhaus- tive comparison and analysis. Greatly impressed by this fact, I began nearly ten years ago to improve the opportunities that were offered by the marine laboratory of the Johns Hopkins University, for obtaining more complete and detailed knowledge of the larval stages of a number of Macroura, and this work has been prosecuted at every opportunity up to the present time. Some of my results have been pub- lished in my monograph on Lucifer, in the Phil. Trans. Royal Soc. for 1882, and others are incor- porated in my report on the Stomatopoda collected by H. M. S. Challenger. This memoir contains the life histories of a number of additional species based in part upon my own studies at Beaufort, North Carolina, and at Green Turtle Key and New Providence in the Bahama Islands, but chiefly upon the researches which one of my students, Mr. F. H. Her- rick, has carried on under my general supervision. In 1SSO he undertook, at my suggestion, the study of the embryology and metamorphosis of the Macroura, and devoted three years to this sub- ject under my direction, and the results which follow are almost entirely due to his zeal and energy. He has completed the study of several subjects upon which I had previously made a beginning, so that my own unfinished notes have been incorporated with his researches, and our respective shares in the work are as follows: The chapter on Gonodactylus is entirely based upon my own researches; the chapter entitled " Alpheus, a study in the development of the Crustacea," is entirely the work of Mr. Herrick ; the one on the metamorphosis of Alphens is based upon our combined studies, and that upon Steuopus is almost entirely the work of Mr. Herrick, as my own contributions to this life history are of minor value except so far as they supplement his work. I shall now give a brief outline or summary of the chief results which are described in detail in each chapter. THE LIFE HISTORY OF STENOPUS HISPIDTIS. During the six seasons which I spent at Beaufort, North Carolina, I captured in the tow-net, at different times, some six or seven specimens of a remarkable pelagic crustacean larva, all of them well-advanced and in nearly the same stage of development. Nothing was learned of the earlier larval life nor of the adult form of the animal, although enough was made out to show that it is one of the few Macroura which, like Peneus and the Ser- gestidai, have retained tbe primitive or ancestral metamorphosis, and that its secondary modifica- tions are very slight as compared with those of ordinary macrouran larvse, and also that the Beaufort larva? are new to science. (See Pis. is and x.) These larvse have the full number of adult somites and appendages, and in side view they are very suggestive of the Sergestid;e. They are very much larger than ordinary pelagic larvae and are quite different from any known forms of Macronra. The chief locomotor organs are the last pair of thoracic legs, which are extremely slender, as long as the entire body of the larva, ending in flattened elliptical paddles, which are used as sweeps for rowing through the water. They are stretched out in front of the body near the MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 327 middle line and are then swept backwards and outwards, describing at each stroke a circle equal in diameter to about twice the length of the body. By the vigorous use of these oars the larva skims rapidly through the water, and its movements are not unlike those of a Gerris upon the surface of a fresh-water pond. Notwithstanding the importance of a complete knowledge of the life history of the animal to which this sergestid-like larva belongs, I was unable to complete the study at Beaufort, although I made careful drawings of two stages and filed them away for future use. Immediately upon our arrival at Green Turtle Key, in the Bahama Islands, early in June, 1886, our attention was at once attracted to a small, graceful, brilliantly colored prawn which was found in abundance among the coral. (See PI. v.) It proved to be Stenopm hispiilus, a species which is chiefly known to naturalists through specimens from the Indian and South Pacific oceans. .It has been recorded as occurring in the tropical Atlantic, but our knowledge of the adult has been very scanty and imperfect, and nothing whatever has been known regarding its life history until Mr. Herrick devoted himself to its thorough investigation. It is an active, timid animal, and is one of the most brilliantly colored of the Crustacea. As it is also one of the most widely distributed, it is noteworthy that while its color markings are so prominent and conspicuous they are extremely well fixed and constant; so much so that the speci- mens from the Indian Ocean and the South Pacific agree with those from the West Indies dowu to the most minute markings. The adults are found in pairs, a male and a female swimming together side by side and exhib- iting evidence of strong conjugal attachment to each other. The most noteworthy fact in its history is its world- wide distribution, and the question whether this can be a result of any peculiarity in its structure or habits at once suggests itself. We should expect, on general principles, to find the least specialized species the most widely diffused; and one which holds its ground in so many parts of the world, and without any change of structure finds a safe and congenial home in seas so widely separated, might be expected to be of indefinite or slightly specialized habits, but this is not the case. In structure, in habits, in color, and in external appearance, and also in its metamorphosis, Stenopus is one of the most highly specialized of che crustacea; and it owes its ability to survive in many seas to the accuracy and delicacy of its adjustment to a narrow range of conditions, rather than to indefinite and vague adaptation to many conditions. Its antenna? are unusually long and slender, and the acuteness of its senses, together with its very remarkable alertness; the quickness with which it perceives danger, and the rapidity with which it escapes; have undoubtedly aided it in holding its own wherever it has gained a foothold in a suitable locality, and no crustacean, with the exception, possibly, of Gouodactylus chiagra, is better adapted for life in a coral reef. It is well protected from enemies by a thorny armor of hooked spines, which cover all the upper surface of its body and limbs, and as all the hooks point forward the attempt of an enemy to swallow a Steuopus must be difficult and painful. These facts no doubt account for its survival, and the length of its pelagic larval life is beyond question an aid to its wide dispersal and to the discovery of new homes. While we cannot state that the adult will not at sometime be found upon the Atlantic coast of our Southern States, there is no evidence that this is the case, and the larva? which were obtained at Beaufort, North Carolina, were undoubtedly hatched from eggs which were carried upon the abdominal appendages of parents in the West Indies or on the Florida Keys ; and these larva' had therefore wandered more than six hundred miles from their birthplace. The, species might therefore be diffused through a chain of coral islands six hundred miles apart, from a single start- ing point, in a very small number of generations. The eggs, which are very small, are laid at night, and the segmentation, which Professor Herrick has thoroughly studied by sections, is entirely confined to the nuclei, the yolk remaining undivided; Steuopus therefore presents a most pronounced type of centrolycethic segmentation. The great mass of the egg consists of a homogeneous mass of yolk grannies, which takes no part in the process of segmentation and probably contains no protoplasm. This yolk is aggre- gated around a central nucleus, which divides, probably indirectly, into two, four, eight, sixteen 328 MEMOIRS OP THE NATIONAL~ACADEMY OF SCIENCES. nuclei, and so on until the number is very great. As this process of division goes on the nuclei, each with an investing layer of protoplasm, gradually migrate to the surface of the yolk, and at last form a superficial investing layer around a central yolk, out of which all the protoplasm has been withdrawn. The yolk does not divide up into typical yolk pyramids, although the outlines of the blasto- meres are sharply indicated by transitory superficial furrows. The embryonic area is soon marked out as a region where the nuclei are densely crowded, and the point of invaginatiou is indicated by a solid ingrowth which penetrates the yolk to form the inner layers of the embryo. The subsequent stages of embryonic development were not followed in detail. The larva hatches in the afternoon, and during the following night the parent moults and lays another brood of eggs. At the time of its escape the larva is a Protozoea, and its later history is of great interest, since it unites features of resemblance to Lucifer, Sergestes, Peneus, and to the prawns in general, with individual peculiarities in which it differs from all of them. At the time of hatching (PI. VII and PL XI, Fig. 25) it has sessile eyes, locomotor antenna?, an enormous mandible, a deeply forked telson, a long rostrum, and a complete series of append- ages as far as the first pereiopods, which are essentially like the third maxillipeds. The long hind body has no appendages and is only vaguely divided into somites. Five or six hours after hatching it changes into a true zoea, much like that of an ordinary rnacrouran (PL viii). The carapace becomes much enlarged ; the rostrum is shortened to less than half its former length, the mandible becomes small, the forks disappear from the telsou, the eyes become stalked, the antenna? are shortened like those of a zoea, and the maxillipeds become the chief .locomotor organs. As these larva?, could not be reared in captivity the later stages were studied from captive specimens, but Professor Herrick has proved that the Beaufort larva; are either young Stenopi or else the larva? of some closely allied species which is at present unknown. A specimen a little older than the oldest Beaufort specimen was captured at Nassau (PL xn). It is in the Mastigopus stage, with greatly elongated eyes, and with antenna? which are gradually approximating to those of the adult. The third maxillipeds are now extremely long and are the largest of all the limbs, while the huge, oar-like fifth pereiopod of the preceding stage is now reduced to a rudimentary bud, and the fourth is also reduced to a two-jointed rudiment. It thus appears that, as in the Sergestida?, the last two pairs of " walking legs'' are shed alter the My sis stage, to be again reconstructed in the Mastigopus stage. After several moults the Mastigopus larva gradually assumes the adult form, the principal changes being the shortening of the eyes and the reacquisition of the fourth and fifth pereiopods. ALPHEUS. The genus Alpheus includes a large number of small, brilliantly colored crayfish-like Cru- stacea, which are widely distributed, although all are essentially tropical. Two species range as far northward as the coast ot Virginia, but the true home of the genus is the warm water between tide-marks or near the shore in coral seas, and they occur in the greatest abundance and variety in all the sounds and inlets among coral islands. They are well adapted, in structure as well as in habits, for a life among the coral, and of all the Crustacea which abound upon the coral reefs the genus Alpheus is one of the most common and most thoroughly characteristic (PL i, n, and iv). Nearly every mass of sponge or alga? or of copal rock or living coral which is fished up from the bottom and broken to pieces contains specimens of one or more species of Alpheus, and pieces are often found which fairly swarm with these little animals. A few of the species wander over the bottom, and wandering individuals of other species are found occasionally, but their true home is in the fubes of sponges and the holes and crannies in the porous coral limestone, or under the broken shells and fragments of limestone which lie upon MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. 32(,> the button) in shallow water. Occasionally they inhabit short, vertical burrows, which they con struct for themselves in the sandy uiud, but most of the species pass their life hidden, in the shelter which they find upon the reef. The most conspicuous characteristic of the genus is the great enlargement of the claws of the first pair of walking legs. Both claws are large, but one of them is enormous, and it serves as a most formidable, weapon of offense and defense. In some species this large claw nearly equals the body in size, and it is usually carried stretched out in front of the body, but one species carries it folded down under the body and hidden, ready to be instantly pushed out to make a rapid thrust at any enemy. In nearly all the species the, large claw terminates in hard, powerful forceps. The claw or dactyle is provided with a plug, which fits into a well or socket in the other joint and probably serves to prevent dislocation. When the forceps are opened the dactyle is raised so that the plug just rests in the mouth of the socket. As soon as the claw is released it is suddenly and violently closed, as if by a spring, and the solid stony points striking together produce a sharp metallic report, something like the click of a water hammer, and so much like the noise of breaking glass that I have often, when awakened at night by the click of a little Alpheus less than an inch long, hastened down to the laboratory in fear that a large aquarium had been broken. In the open water the report is not so loud as it is when the animals are confined in small aquaria, but Al- pheus is so abundant in all the Bahama Sounds that a constant fusilade is kept up at low water all ' along the shore. The animals are remarkably pugnacious and they will even attack bathers. They are known to the inhabitants of the out islands as "scorpions," and are much dreaded, although their attacks are harmless to man. The snapping propensity is exhibited both in the water and out by both sexes, and if two males or two females, either of the same or different species, are placed together in an aquarium, a most violent combat at once takes place, and quickly ends by . the destruction of one or both. Some species appear to pinch with the large claw, but it is more frequently used like a saber for cutting a slashing blow. The edge of the movable joint is sharp and rounded, and the animal advances warily to the attack with the claw widely opened and stretched out to its full length. Watching its opportunity it springs suddenly upon its enemy, instantly closing its claw with a violent snap and a loud report, and cutting a vertical sweep witli its sharp edge. I have often seen Alpheus heterochelis cut another completely in two by a single blow, and the victim is then quickly dismembered and literally torn to fragments. The abundance of these animals in coral seas is well shown by the fact that of the twenty species which are known to inhabit the shores of the North American continent we found twelve, or more than half, upon a little reef at Dix Point, a few rods to the eastward of our laboratory at New Providence, in the Bahama Islands. Of the thirteen species which we found in the islands several are new, and as none of them have ever been adequately described, an illustrated, systematic description of all the species is now in preparation by Mr. Herrick. The present memoir deals only with the embryology and metamorphosis of the genus. This is a new field, for nothing whatever has as yet been published upon the embryology of any species of the genus, and all our knowledge of the metamorphosis is contained in two short abstracts without illustrations on the metamorphosis of a single species, Alpheiifi lieterochelis. Eggs have now been obtained from all thirteen of the Bahama species, and the first larval stages of most of them have been reared from the eggs in aquaria in the laboratory, and the metamorphosis has been traced from actual moults. THE METAMORPHOSIS OF ALPHEUS. One of the most remarkable results of our study of the various species of the genus Alpbeus is the discovery that, while there is such a general similarity as we might expect between the larval stages of the different species, the individuals of a single species sometimes differ more from each other, as regards their metamorphosis,' than the individuals of two very distinct species. This phenomenon has been observed by us and carefully studied in two species — Alpheus hetcro- chelis and Alpheu-g mtulcyi— and it is described in detail, with ample illustrations, in the chapter on the metamorphosis of Alpheus. In the case of the first species the difference seems to be geo- graphical, for while all the individuals which live in the same locality pass through the same series 330 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. of larval stages, the life history of those which are found at Key West is very different from that of those which live on the coast of North Carolina, while those which we studied in the Bahama Islands present still another life history. In the case of the second species — Alpheus saulcyi — the difference stands in direct relation to the conditions of life. The individuals of this species inhabit the tubes and chambers of two species of sponges which are often found growing on the same reef, and the metamorphosis of those which live in one of these sponges is sometimes different from thai of those which inhabit the other. In this species the adults also are different from each other, but as we found a perfect series of transitional forms there is no good reason for regarding them as specifically distinct, and in the case of the other species — Alpheus heterochelis — we were unable, after the most thorough and minute comparison, to find any difference whatever between adults from North Carolina and those from the Bahama Islands, although their life histories exhibit a most surprising lack of agreement. In fact, the early stages in the life of Alpheus hete- rochelis in the Bahama Islands differ much less from those of Alpheus minor or Alpheus normani than they do from those of the North Carolina Alpheus heterochelis, and, according to Packard, the Key West heterochelis presents still another life history. In the summer of 1881 1 received the American Naturalist with Packard's very brief abstract of his observations at Key West upon the developmeutof Al/>hens heterochelis, and read with great surprise his statement that this species has no metamorphosis, since, while still inside the egg, it has all the essential characteristics of the adult. As I had under niy microscope at Beaufort on the very day when I read his account a newly hatched larva of the same species and was engaged in making drawings to illustrate the metamorphosis of which he denies the existence, and as my experience in the study of other Crustacea had taught me that all the larvje of a species at the same age are apparently facsimiles of each other down to the smallest hair, Packard's account seemed absolutely incredible, and I hastily decided that, inasmuch as it was without illustra- tions and was written from notes made many years before, it involved some serious error and was unworthy of acceptance. This hasty verdict I now believe to have been unjust, since my wider acquaintance with the genus has brought to my notice other instances of equally great diversity between the larva? of different specimens of a single species. The phenomenon is, however, a highly remarkable one and worthy the most thorough exami- nation, for it is a most surprising departure from one of the established laws of embryology — the law that the embryonic and larval stages of animals best exhibit their fundamental affinities and general resemblances, while their specific characteristics and individual peculiarities make their appearance later. As in most animals the adult life is most important, the adults have a more diversified envi- ronment than the young, and the divergent modification which is continually taking place to perfect the adjustment between such organism and its conditions of life chiefly affects the adults, so that specific characters and the slight differences between varieties or races are usually con- fined to the adults, while the embryos and larva? are, as a rule, more generalized. This is true to a marked degree of those animals whose young are nursed or protected or cared for in any way by their parents, and while it is less true of those animals whose independent life begins very early, yet the same law holds with them also ; and the chief scientific value of embry- ology lies in the fact that a knowledge of the early stages in the life of animals enables us to trace their broad affinities and to distinguish them from more recently acquired differences; for the early stages of two related forms of life share in common their more fundamental characteristics and are essentially alike, while the adults differ from each other and exhibit the divergent speciali- zations which are of more recent acquisition. It sometimes happens, however, that the early stages of two closely related species differ greatly. This may occur when the larva? of the one species lead a free, independent, life, while the young of the other species .are protected in someway by the parent. For example, the compli- cated metamorphosis which is so characteristic of starfishes is almost totally absent in those star- fishes which are provided with brood-pouches. The same relation may also bo exhibited when the larviB of one species of a genus have become adapted to a mode of life very different from that of the larva? of the other species of the genus. Thus those species of ^Egiuidie whose larva? are para- MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 331 sitic multiply asexual ly during the larval life, and build up complex communities, while nothing of the sort occurs in those species with free larva/. Many similar cases might be given, but we must bear in mind that they are all very different from the one now under examination. In all such cases the difference is between the larva- of two distinct species, while in Alpheus we have a similar difference between the. larvir of individuals of a single species. Among other animals it is not very unusual for certain individuals which arc placed under conditions exceptionally favorable for embryonic development to be born in a more advanced stage than the normal for the species, and in such cases the larval metamorphosis is abbreviated by the omission of the earlier stages. This abridgment of the larval life is not common, but many cases are known, and if the his- tory of Alpheus were simply another illustration of this process of abbreviation it would not be at all anomalous, although the existence of three well marked and fixed grades of abridgment in Alpheus heterodielis, in three widely-separated localities, would still be remarkable and interesting. The life, history of the North Carolina form of this species is more abbreviated than that of the Bahama form, and the metamorphosis of the Key West form is still more shortened, but, in addition to the abridgment, the different forms also present most important differences in structure and in the order in which the, appendages are developed; differences which are much more funda- mental and profound than the mere length of the larval life. The various larval forms are described with so much detail in the chapter on the metamor- phosis of Alpheus that it is not necessary to-repeat them here, but the following very brief outline will serve to call attention to a few of the most conspicuous features : As several distinct species of the genus Alpheus pass through a long metamorphosis, each stage of which is almost exactly the same in all the species, we may safely assume that this is the primitive or ancestral metamorphosis which was originally common to all the species. It has been traced in Alp/tens minor by me at Beaufort, North Carolina, and by Mr. Herrick in a similar species at New Providence. Mr. Herrick has also traced it at New Providence for Alphatx normatil and Alj)lieus keterochelis. In all these forms the larva hatches from the egg in a form which is very similar to Fig. 13 of 1*1. xvi, and very shortly after hatching it moults and passes into the second larval stage, which is the one from which Fig. i' was drawn. This larva has all its appendages fully developed and functional as far backwards as the, third pair of maxillipeds. Following these are three bud like rudiments, to represent the first, second, and fifih thoracic limbs, and posterior to these a, long, tapering, imperfectly-segmented abdomen, ending in a flat triangular telson. The locomotor organs are the plumose antenna; and the exopodites of the three pairs of max- illipeds. After the second moult the larva passes into the third stage, which is shown in PI. xvi, Fig. 1, and PI. xvn, Fig. 1. The first and fifth thoracic limbs are now functional; all the abdominal somites are distinct and movable, and the uropod, or sixth abdominal appendage, has appeared, and its exopedite is functional and fringed with plumose hairs, while its endopodite is rudimentary. The five abdominal appendages have not yet appeared. The first thoracic leg, which was represented by a bud in the preceding stages, has now acquired a flat, basal joint and a swimming exopodite like those of the maxillipeds, but its endo- podite is rudimentary. The fifth thoracic limb is fully developed and is the most conspicuous peculiarity of the larva at this stage of development. It has no cxopodite; its basal joint is not enlarged nor flattened, and its long, cylindrical, slender shaft is prolonged at its tip into a long lance-like, hair, which projects beyond the tips of the antenna-. After its third moult the larva passes into the fourth stage, which is shown in PI. xvnr. Fig. 3. The carapace now begins to extend over the eyes, and the ears make their appearance, in the basal joints of the antennules. There are now five pairs of plumose locomotor exopodites, belong- ing to the first, second, and third maxillipeds, and the first and second thoracic limbs. Between the latter and the elongated fifth thoracic limb are buds to represent the third and fourth. The telson has become narrow and elongated, and the uropods are fully developed, although there is as yet no trace of the other abdominal appendages. 332 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. After the fourth moult the larva passes into the fifth larval stage, when it resembles Fig. 1 of PI. xxi, so far as concerns the anterior end of the body, from which it differs greatly as regards the telson and uropods. The eyes are now well covered by the carapace, and the swimming organs are the seven pairs of exopodites belonging to the three maxillipeds, and the first four thoracic limbs. The first five pairs of abdominal appendages are now represented by buds, like those shown in PL xxi, Fig. 1, but the telson and uropods are nearly like those of Fig. 3, in PL xx. The telson is narrow and much elongated, and its marginal spines are very small. During the moults which follow, the abdominal appendages become fully developed, the eyes become completely covered by the carapace, the autennule develops a scale, the antennae elon- gate, the swimming appendages of the midbody disappear, these appendages assume their adult form, and the animal gradually becomes like the young Alpheus shown in PI. xx, Fig. 2. This life-history is common to Alpheus minor at Beaufort and New Providence and Alpheus normani and Alphem heterocheUs at New Providence, although the latter species presents a totally different life-history at Beaufort. Before it hatches, this form, as shown in PL xx, Fig. 1, reaches a degree of development which bears a general resemblance to stages two and three of the Bahama form, with certain differences which are pointed out in the sequel. Immediately after hatching it assumes the form which is shown in PL xix, Fig. 2, and Fig. 1. The animal has now all the somites and appendages of the adult, but all behind the maxillipeds are rudimentary, and there is little power of locomotion. The first moult occurs in a few hours, and the larva assumes the form shown in PL xx, Fig. 3, when it is no longer a larva but a young Alpheus. The eyes are almost completely covered by the carapace, the ear is well developed, the flagellum of the antenna has elongated, and the other appendages have assumed the adult form. An older specimen is shown in PL XX, Fig. 2, and a still older one in PL xvii, Fig. 3. Careful com- parison will show that no exact parallel can be drawn between any larval stage of this form and a sluice of the first form, and that we have to do with something more profound than simple accele- ration of development. The Bahama heterocheUs has, at first, three, then four, then five, and then seven fully developed and functional exopodites, while the North Carolina form never has more than three. As regards the thoracic region and the first five abdominal appendages the Beaufort larva, at the time of hatching, is more advanced than the fourth larval stage of the Bahama form, while the nropods are like those of the Bahama form at the time of hatching. In the latter the first and fifth thoracic limbs are the oldest, and the others appear in succes- sion, while all five pairs appear together in the Beaufort form. In the Bahama form the uropods appear before and in the Beaufort form after the others, and many minor differences show that we have to do with profound modification of the life history rather than with simple acceleration. Packard's short account of the development of those specimens of this species which occur at Key West shows that these differ from the Beaufort specimens about as these latter differ from the Bahama specimens. The second species is probably A. naulcyi, although Gue'rin's figure and description of this form are not in accord with it in some important points. It is found in the Bahama Islands, living in the tubes and chambers of two species of sponge, a green one and a brown one. Those found in the green sponges have many small eggs, while those found in the brown sponges have only a few largo eggs. The eggs from the green sponge hatch in the stage shown in PL xxi, Fig. 1. It has rudimentary gills, the eyes are imperfectly covered, the antennules and antenna; are beginning to assume their adult form, and the exopodites of the three pairs of maxillipeds are the chief organs of locomotion, although all the appendages are represented. The abdominal feet are rudimentary, however, and the nropods are covered by the cuticle of the telson. Very soon after hatching the larva moults and assumes the form shown in PL xxi, Fig. 2. The eyes are more completely covered, the antennules and antennre are elongated, the thoracic limbs have the adult form and the pleopods are all functional. In twenty-five or thirty hours after hatching it moults for the second time and passes into the third stage, which is shown in Fig. 8. It is no longer a larva, but a young Alpheus, with all the structural characteristics and pugnacious instincts of the adult. In a few cases the development of this species is still more accelerated, and a few eggs from animals taken from the brown sponge hatched in the stage shown in Fig. 8, instead of the stage MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 333 shown in Fig. 1. The following notes on the variations in the coloration and habits in Alplicus, particularly in A. saulcyi, are taken from a paper by Mr. Herrick published in the Johns Hopkins University circulars. VARIATIONS IN THE HABITS AND COLORATION OF ALPHEUS. Some of the species of Alpheus are usually or even universally found living as parasites within the water tubes of sponges, and it is extremely interesting to find that individuals of the same species, living in different species of sponges, may themselves differ greatly in color and in habits. A large brown sponge, Hircinia arcuta, which is not to be mistaken, grows on the shallow reefs and off the shores of many if not all the Bahama Islands. This is called the "loggerhead sponge" by the fishermen, and is found from just below low tide mark out to one-half a fathom or more of water, where its great size and sooty brown color distinguish it at once on the white bottom. If a sponge colony of this kind be pulled and torn apart, one is certain to find it swarming with a small species of Alpheus, which quarter themselves in the intiicately winding pores of the sponge. Hundreds, or even thousands of individuals might be collected from a single large specimen. These animals vary from one-eighth to three-fourths of an inch in length. They are nearly colorless, excepting the large chehe, which are tipped with brown, reddish orange, or bright blue. The females are so swollen with their eggs or burdened with the weight of those attached to the abdomen that they can crawl only with great difficulty if taken from the water. The eggs are few in number and of unusually large size, their diameter varying from one twenty-second to one twenty-fifth of an inch, and their number from six to twenty. These are most commonly yellow, but may be either bright green, olive, flesh color, brown, or dull white. Another quite different sponge grows on all the reefs in from one to two fathoms or more of water. There are several varieties of this, which may be told by their olive-green color, yellow flesh, and clumpy, irregular shape, as well as by the putrescent mucus which some of them pour out when broken open. In nearly nine out of ten of these sponges one will find a single pair of Alphei which resemble those living in the brown sponge in most particulars, although they differ from them in several important points. They are distinguished by their large size, and by their peculiar and very uniform color. They vary in length from two-thirds to one and two-thirds inches. The females exceed the males greatly in bulk owing to the great size and number of their eggs. Both sexes are nearly transparent and colorless excepting the large claws, which are bright vermilion-orange (PI. iv). The female is practically inert during the breeding season (which lasted during our stay, March to July), and at such times is well protected in her sponge, or against any green surface, by the bright green ovaries which fill the whole upper part of the body, and by the mass of similary colored eggs attached to the abdomen below. Only two pairs, or four individuals, out of a hundred or more which were examined showed any variation from these colors. In these the eggs were yellow, and the pigment on the claws more orange than red. The table which fol- lows shows the variations between two large females taken respectively from the brown and green sponges, and between the size, number, and color of the eggs. Habitat of Alpheus. Length of $? Number of eggs. Diameter. Color. Color of adult. Inches. Inches. Brown sponge 1 19 A Yellow (variable). Large clielie, red, blue, or brown. Green sponge 1A 347 A Usually green ; in this Large cueko, always case yellow. orange-red. These two forms, apparently distinct, are seen however, by closer examination, to belong to the same species, although they show very interesting variations. The Alpheus living in the brown sponges tends to vary in several ways, chiefly in size and in the color of the body and eggs. The rostrum usually has three spines, but occasionally only two are present, the median one being lost. It is evident that these animals are perfectly protected from outside enemies while within the tortuous mazes of the sponge, as their numbers would show. Parasites such as Isopods, how- 334 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. ever, are not uucommon. There has thus been no chance or need for natural selection to act along the line of color. On the other hand, possibly, the Alpheus of the green sponge does require color protection, since the females are very sluggish during the breeding season, which extends over a good part of the year. Tins animal is certainly well protected against any green surface, as already stated. But as will be shown, natural selection has probably nothing to do with it. The bright col- oring of the tips of the claws, which only are protruded from the place of concealment, recall the similarly colored heads of boring annelids which abound on the reef, and may have a protective significance. This evidence, however, is not very reliable. The colors of certain Crustacea, and also the color of their eggs, are known to vary greatly with the surroundings. In the Alpheus parasites in the brown sponges these colors vary consid- erably where the surrounding conditions are the same. However, the color of the ovarian eggs is always the same as that of those already laid, and although these animals were kept for several days at a time in differently colored dishes, we never observed any very marked change in the color of the eggs, but these experiments were not continued long enough or carefully enough to be conclusive. The eggs of Alpheus hecterochelis are almost invariably of a dull olive color, while as in the case of the parasite of the green sponge, about one in a hundred has bright yellow eggs. lu the first case at least this is possibly an instance of reversion to one of the original colors from which the green was derived by natural selection. In most species of Alpheus the color of the eggs is fixed and uniform, and as already suggested may have a protective significance, but in a few other cases where this is not true, the color is not only variable in different individuals, but probably also iu the same individual. lu order to explain the variations which we find iu these two forms, we must assume either (1) that the parasites of the green sponge are a fixed variety with distinct habits, or (li) that they repre- sent individuals which have migrated from the brown sponges and adapted themselves to their new surroundings, or further (3) that only those chance individuals with orange-red claws and bright-green eggs, which occasionally ocmr in the brown sponge, find their way to the smaller green species, where they acquire great vigor and size. This last supposition is evidently untena- ble. If moreover the two forms, which were at first supposed to be specifically distinct, represent fixed varieties, -we ought to find the young or at least adults of all sizes in both sponges, whereas it is only iu the large brown variety that any small or undersized individuals occur, while a single pair, of large and tolerably uniform size, is invariably found in the exhalent chambers of the green sponges. These and other considerations render it probable that the second (li) proposition above stated is the correct one, viz, that the parasites of the green sponges were born in the brown variety, and after attaining considerable size migrated thither, where they adapted themselves at once, to their slightly different surroundings, growing to three or four times their former size, and the females acquiring bright green eggs, which become a source of protection in their new habitat. This view implies the greatest variability in color and in size of the individual, and iu the color of the egg, which is more remarkable from the fact that it is quite uuusual in this genus. THE EMBRYOLOGY OF A1.PHEUS. At my suggestion Mr. Herrick undertook, iu 1886, the study of the embryology of Alpheus, and devoted a considerable part of his time for three years to this subject, and while he carried on the work under my general supervision the results which he has reached are entirely his own, and my share in the chapter which is devoted to this division of the subject is only that of an instructor. I must call attention, however, to the fact that Mr. Herrick's studies were begun at a time when our knowledge of the embryology of the higher Crustacea was far less complete than it is at the present time. From time to time brief abstracts of the progress of the research have been written by Mr. Herrick and published in the Johns Hopkins University circulars, and the following cor- rected summary of his results contains the substance of these preliminary reports. The work was begun at Beaufort, North Carolina, and the eggs of the two species of Alpheus which occur there were carefully examined aud preserved for laboratory research, but much better and simpler material was afterwards obtained at the Bahama Islands, the early stages were much more thoroughly studied, and the development of the animals was traced in detail, step by step, MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 335 from the first nucleus of the fertilized egg, through all the embryouic and larval stages, up to the adult condition. The eggs of each of the thirteen species which occur in the Bahamas were ob- tained and studied sufficiently to ascertain what are the specific differences in development, and four species were studied exhaustively, in detail. These four are Alpheus heterochelis, Say ; A. mi- nus, Say; A. saulcyi, and the Bahama heterochelix. Unless otherwise stated the following notes on the early stages refer to the last species. The development in the egg is the same for all, except- ing A. minor, which will be referred to separately. This prawn has proved to bo a good subject in which to study the origin aud role of certain much disputed bodies, which are met with in several Crustacea, the "secondary mesoderm cells." SEGMENTATION OF THE NUCLEUS AND OF THE YOLK. The egg when laid, is enveloped by a single membrane, the chorion or shell, to which is added the secondary membrane of attachment. If ihe nucleus is unfertilized, it is not able to initiate the process of segmentation. The fertile nucleus divides, aud its products pass towards the surface, until a syucytiutn of eight nuclei is formed. Either just before or after the division of these, the yolk undergoes segmentation simultaneously over the whole surface into a similar number of partial pyramids. Each yolk pyramid has a large nucleus at its base, while its 'apex fuses with the common yolk mass in the interior of the egg. The process is now a regular one until 128 to 250 small segments are formed. The rate of cell multiplication is then retarded over one-half of the egg, while it still continues and perhaps is accelerated over the remaining portion of it. The egg thus loses its radial symmetry and becomes two-sided. It is important to notice that no products of the segmentation nucleus are left in the interior of the yolk. The superficial pyramidal structure is lost; the primitive blastoderm is established, and there now takes place a general migration of nuclei from the surface to the yolk within, but principally, as would be expected, from, that part of the egg where the blastoderm cells are most numerous, corresponding to the future embryo. This is followed by a partial secondary segmentation of the food yolk into balls. The yolk-ball is apparently formed about the migrating nucleus, but as the latter is moving, this segmentation is irregular. Mr. Herrick has been able to follow very closely the entire process of segmentation in Stenopus, where it is substantially the same as that just described, except that there is no general migration of cells from the surface, prior to iuvagination. This is also true of Poutouia domestica, and it is quite probable that the majority of inacroura pass through the same phases in their early development. Alpheus minor is anomalous from the fact that the products of the first nucleus instead of multiplying by regular binary division, multiply indirectly, and give rise to numerous nuclei, many of which degenerate, before the blastoderm is formed. THK INVAG1NATION STAGE. A slight invagination occurs where the superficial cells are thickest, aud the egg becomes •what has been generally regarded as a modified gastrula. The depression is shallow, aud does not form an inclosed chamber within the yolk. The included cells multiply rapidly, and form a mass of nearly similar elements, some of which pass into the yolk. The protoplasm surrounding the nuclei of these cells is prolonged into a reticulum, which encloses myriads of small yolk frag- ments, and probably digests them by an intracellular process, after the manner of feeding amoeba;. The thickening in front of aud surrounding the pit, which is now obscured, is the rudiment of the abdomen. Anteriorly the " procephalic lobes " or more properly the optic disks make their appear- ance on either side of the long axis of the embryo, as circular patches of ectoderm. Meantime nuclei wander from the cell mass below the abdominal plate to all parts of the egg. Some pass to the opposite side, and take up a position beside the flattened epithelial cells, of what was the prim- itive blastoderm. The majority, however, pass forward and upward in divergent lines from the sides of the abdominal plate, aud eventually large numbers of these wandering cells settle down over the dorsal surface of the embryo. 336 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PRODUCTS OP CELL DEGENERATION. At the beginning of the egg-nauplius period, when numerous wandering cells have passed forward and joined the inner surface of the embryonic ectoderm, certain new bodies begin to appear in great numbers. They vary in size from small refringent particles to bodies nearly as large as ordinary nuclei. The latter stain deeply and nearly uniformly, but with high powers it is often possible to demonstrate a clearer zone about them, which might be mistaken for a layer of cell protoplasm. How do these bodies, the so-called "secondary inesoderm cells," originate and what is their function? As to their origin there can be no doubt whatever. They arise by a process of degeneration from the embryonic cells or nuclei, chiefly from those wandering cells ju.st described. Many of the latter may be seen to be swollen out and their chromatiu divided into coarse grains and balls of various sizes. The wall of the cell breaks down and thus sets the chro- matiu granules free, or, more correctly, the products of the degenerating chromatin. These degenerating bodies are most marked in the fully developed egg-uauplius, where there is a large accumulation of them around the oesophagus and at the bases of the rudimentary appendages. After this stage they generally disappear from these regions. Somewhat later, however, when there is a well developed nervous system and six pairs of post-uaupliar appen- dages, a patch of ectoderm cells on the surface of the egg opposite the embryo proper becomes noticeable. It reminds one of a median unpaired "dorsal organ." A slight invagiuatiou appar- ently takes place at this point, but at any rate a. number of cells pass into the surrounding yolk, and these give rise in the way described, to a swarm of minute particles of chromatiu products. Before any pigment is deposited in the eyes, it is easy to demonstrate the presence of blood corpuscles in the stream of plasma which bathes the nervous system. They have the adult characteristics, that is, they possess a deeply staining nucleus and a clear irregular body. In the nauplius stage, moreover, some of the larger "secondary mesoderm cells" have a similar appear- ance, but there is no evidence that they ever become blood cells. Mr. Herrick's study of these bodies has shown that Eeichenbach's views on the function of secondary mesoderm cells of Astacus are probably erroneous. According to this naturalist they arise from the nuclei of the eudoderm cells, forming the ventral wall of the primitive stomach, and are converted into mesoderm. THE GERMINAL LAYERS. The apparent plasticity of the embryonic cells and layers and the comparative tardiness with which they are clearly differentiated can not fail to impress anyone who follows closely the early stages of development. The cell mass developed around the invaginate area, forming the thoracic- abdominal process, can not be artificially divided into layers. It certainly represents very largely the primitive mesoderm, but some of its elements pass to the opposite pole of the egg and become almost indistinguishable from the superficial ectoderm, although it will be shown that they do not pertain to this layer. A part of this mass remains as the mesoderm of the rudimentary abdomen, while many of the cells which migrate from it degenerate and perform a nutritive function. The endoderm does not appear as a definite layer until the egg-naupiius stage. It' arises from wandering cells which assume a peripheral position, and, joining the cells of the hindgut, form the walls of the mesenteron. THE EYK. The optic discs appear as patches of ectoderm, one cell thick on either side of the long axis of the embryo in front of the rudimentary abdomen. Before the appendages are definitely formed, these have become thickened ectodermic discs. This thickening is due (1) to delamination, or to a division of cells in a plane parallel with the surface; (2) to emigration of cells from the surface, due to crowding or to a division of superficial cells in a plane at right angles to the surface. A disc of cells is thus formed which gives rise chiefly to the eye and its ganglia. The cord of cells uniting the two optic discs represents mainly the future brain. The eye proper is due to the differentiation of the outer layer of the cells of this disc, while the ganglion is developed from the inner layer. For fuller results of later studies not represented by these partial and preliminary notes, reference must be made to Mr. Herrick's completed paper and to the summary of the whole history of the development in the egg given at the end. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 337 GONODACTYLUS CHIRAGRA. ''There are few orders of animals of which we arc more ignorant than we are of the Stomato- l>ods. They are well known as museum specimens, and every natural-history cabinet contains one or two, which have been brought home as rare curiosities from distant seas; but we know hardly anything of the habits of the living animals. They are abundant and widely distributed, but like most rapacious animals they are very alert, taking alarm at the slightest disturbance and retreat- ing to the depths of their burrows at the bottom of the ocean, where they are so completely hidden tioui observation that their capture is difficult, and any attempt to study them in their homes is almost out of the question. The habits of Squilla? are tolerably well known, and in my report on the Stomatopoda, collected by II. M. S. Challenger, I have given an account of the habits of Lysiosquilla based upon observa- tions made at Beaufort, North Carolina; but, except for a few scattered and fragmentary notes in the various descriptive papers, this is the whole of our knowledge of the order. During the sea- sons of 1886 and 1887 I was so fortunate as to find in the Bahama Islands Gonodactylus chiragra living in localities which were peculiarly favorable for observing its habits, and I am now able to supplement my report upon the Challenger collections by an account of this interesting species, of which little had hitherto been known, except the fact that it is the most cosmopolitan of the Macroma abounding on the shores and islands of all tropical and subtropical seas. I also obtained its eggs in abundance and succeeded in rearing the young from them iii aquaria, and am now able to make a contribution to a subject upon which there were hitherto no direct observations, for it is a noteworthy fact that while the older larvae of Stomatopoda have long been known, and while many genera and species of them were carefully figured and described and named by the older naturalists before their relationship to the adult Stomatopoda was sus- pected, not a single species in the whole order has, so far as I am aware, been reared from the egg and in this way identified with its specific adult. While the adults usually inhabit burrows in the bottom the larvae swim at the surface of the ocean, and as none of the animals which are captured in the surface net exceed them in beauty and grace, their glass-like pelagic larvae are familiar to all naturalists who have had an oppor- tunity to study the surface fauna of the ocean. Their perfect transparency, which permits the whole of their complicated structure to be studied in the living animal, their great size and rapacity, the graceful beauty of their constant and rapid movements, can not fail to fascinate the naturalist. Unfortunately they are as difficult to study as they are beautiful and interesting, and notwithstanding their great abundance and variety, only two or three of them have been traced to their adult form. Unlike most Malacostraca the Stomatopoda, instead of carrying their developing eggs about with them, deposit them in their deep and inaccessible burrows under the water, where they are aerated by the currents produced by the abdominal feet of their parents. The eggs quickly perish when deprived of this constant current, and as it is very difficult to obtain them at all, I know of no Stomatopod which has ever been reared from an egg under observation. The older larv;e are hardy and are easily reared, but they are seldom found near shore, and microscopic research is so difficult at sea that I know of only two cases in which they have been kept until they assumed the adult form. As I have stated in my report on the Challenger Stouiatopoda, I have reared a young Lysiosquilla excavatrix from an old larva which was captured at the surface, and Faxon has in the same way obtained the young Squilla enipusa. The young larvae are common near shore, but as they seldom survive a uioult in captivity they can not be identified in this way. The growth of the larvae is slow and the larval life long, and as they are as independent and as much exposed to changes in their environment and to the struggle for existence as the adults they have undergone secondary modifications which have no reference to the life of the adult, and are therefore unrepresented in the adult organism. The larvae have been arranged in genera and species, but their generic characteristics are quite different from those upon which the adult genera are based, and this is true in a still greater degree of their specific characteristics. As they undergo great changes during their growth different stages have been described as distinct species or even genera, and it is not easy to select from the rich gatherings which are brought home by S. Mis. 94 22 338 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. collectors the successive stages in the history of a single species. Like the adults, they are widely distributed, and a gap in a series from the North Atlantic may be filled by a specimen from the coast of Australia or the Sandwich Islands, and the collection from a single locality may contain the larva} of several widely -separated species of adults in all stages of growth as well as the larvae of deep-water species which are as yet entirely unknown. The attempt to unravel the tangled thread of the larval history of the Stoiuatopods is there- fore attended with very exceptional difficulties, and the earlier writers were content to rest after the bestowal of generic and specific names upon the larva?. As I found after the Challenger collec- tion was placed in my hands that it was very rich in larva?, I attempted to determine, by compari- son, the larval series for each genus, and the methods which I employed for making the comparison are fully stated in my report. As one of the results of this comparison 1 ventured to describe the general characteristics of the larva of the genus Gonodactyhis (p. 113), and in PI. xn, Fig. 5, of that report I figured a larva which I ventured to call the larva of Gonodactylus. A comparison of that figure with PI. xv, Fig. 11 of this memoir will show that this determination was correct, for the larva of Qonodactylus chirayra which is here described is so much like the one figured in the Challenger report that they belong, in all probability, to the same species. CHAPTER II. THE LIFE HISTORY OF STENOPUS. By FRANCIS H. HERRICK. This paper is the result of observations made at Beaufort, North Carolina, in 1881 ami 1883, and at Nassau, New Providence, iu 1887. The marine laboratory of the Johus Hopkins University was stationed at the latter point in the Bahama Islands from March until July of that year, and with the means thus generously afforded, I was able to considerably extend my studies upon the Crustacea of these coral islands. Professor Brooks found a number of peculiar pelagic larvre at Beaufort, and it is very probable that they represent a part of the life history of Stenoims liispidiis. Plates ix and x, illustrating two important stages of these very interesting larva-., are contributed by Professor Brooks, and the descriptions of these stages are based entirely upon his observations. While the material gathered in a sojourn of a few months at the seashore is in many instances incomplete, it seems worth while to bring out this sketch of the Stenopus, inasmuch as nothing was previously known of its development, and indeed but very little concerning the adult form. Menopus hispidus is, in fact, generally known to naturalists as occurring only in the Indian and South Pacific oceans. 'It was at first quoted from the Atlantic (Cuba) by Von Martens (7) in lS7ii, and it has not since been reported from the Western Continent, so far as we are aware, until we rediscovered it at Abaco, Bahama, iu 1880, but any assiduous collector on West Indian coral reefs must somewhere have hit upon it (v. Appendix i). As the eggs are quite small, as is the case, iu all Crustacea with a protozoa stage, they are not particularly well suited for study by means of sections, and no special attempt has been made 1o trace out the history of the germinal layers, a subject, which can be dealt with to better advantage iu other species. The Stenopi breed readily iu aquaria, aud several series of eggs, illustrating fully the segmentation, and some early phases of development were prepared, and the sections were afterwards made, in Baltimore. These are given on PI. vi. They are especially interesting, since the segmentation is like that of Peu.Biis studied by Haeckel, who relied wholly upon surface observations. The ova were immersed iu Kleinenberg's picrosulphuric acid and afterwards hardened in alcohol. This answered sufficiently well for the purpose in hand, although it rendered the esrgs more resistant than is desirable. I. — THE NATURAL HISTORY OF STENOPUS. The Bahaman Steuopus (PI. v) measures from 1£ to 1.^ inches in length. All the appendages are long aud generally quite slender and delicate, especially the antenna;, which give to this form a very characteristic appearance in the sea. These are snow-white. They are carried widespread and arch outwards in graceful curves. The flagella of the second or outer antennae are two and a half times the length of the body. In the act of swimming these are bent, backward and outward, while the outer division of the first or inner anteuu;e is carried upward, aud their inner branch is directed forward. The body is pure white or nearly so, excepting three broad transverse bands of reddish scarlet. The first or most anterior of these color bands covers the front of the animal, involving the eyes and bases of the antenna), aud in some cases it extends behind the rostrum as far as the mandib- 339 1340 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. ulur or "cervical" groove. The second is nearly confined to the broad tergal surface of the third abdominal segment, while the third zone crosses the last abdominal somite and impinges on the tail fin. The appendages are all colorless excepting the third pair of legs which carry the large pincers. These are similarly marked with four bauds of the same bright color. As shown by the colored plate two of them encircle the great claws, a third belongs to the carpus, and the fourth to the uieros or fourth segment of the limb. The bases of the third and sometimes of the foiuJi and fifth thoracic legs are tinged with bright blue. The ovaries often give to the dorsal surface of the females a light-greenish cast. There is but little variation in the size and character of these markings in the same sex or m different sexes, but it is most remarkable to observe how constant these colors are in individuals of the species from different parts of the world. We possess two colored drawings of this spe- cies* (which will be referred to again), one by Adams (5), from a living specimen taken in the China Sea, and the drawing of Dana (6), who found the species on the coral reef of Earaka, one of the Paumotu Islands, and at Balabac Passage, north of Borneo. Both ot these, and especially the Sameraug plate, essentially agree with our Bahaman specimens, which in color seems to be the more faithful copy of nature. Here the basal joints of the thoracic legs are colored bine as in the Nassau form. Why should Steuopus, coming from different seas, retain the same colors and markings, to a nicety of shade and pattern, while a cosmopolite like Oonodactylun chiragra (a Stouia- topod) presents such wide color variations as to be as unlike as possible, so that scarcely any two taken from the same place have a similar color pattern i To this question we can not at present give a satisfactory answer. Alcohol soon removes all trace of color from the body, but the spots on the legs remain for a longer time as light orange red. Both sexes are of nearly the same size, and, as already stated, alike in color. Being thus brilliantly decorated with the American colors, our crustacean soon acquired with us the name of the " Bandanna Prawn." As we see this animal swimming deliberately in the water we are reminded of some strange and fantastically colored insect. It is by far the most showy, and for its size the most attractive, member of that giant family, the Crustacea, which have their dwelling on the reef. One day, when out upon a wading and diving expedition, a pair of these prawns was discovered by turning over a plate of loose coral, and was easily captured by slowly raising the slab from the water into the boat; for this species, unlike some shrimps, is quite helpless when once out of its element. More frequently, however, they led us to a long chase. There seems to be considerable attachment between the sexes, since they are invariably found in pairs, the male and female swimming side by side. On a still day they may be found clinging to the mosaic of sponges and living coral, which form the reef bottom, but if disturbed they sud- denly become very active, and, darting backward by sudden jerks, dive into some chink, out of reach of the hand net. Several females both hatched and laid eggs in aquaria in the month of June, but the breeding season, as inferred from the capture of locomotor larvae, probably extends throughout the spring aud summer months, if not throughout the entire year. The eggs are very numerous. They are nearly spherical and measure one-fiftieth of an inch in diameter. They were always of the same light opalescent-green color. The ova are laid at night, but the process was not observed. Three different females hatched their broods on the afternoons of June 4, 14, aud 24, respec- tively, and moulted and laid eggs during the following nights. As these animals invariably moult just before laying their eggs, the latter are probably fertilized at the time they are laid. The hatching of one brood lasted about 9 hours, from 2 o'clock in the afternoon until well into the fol- lowing night. By 10 o'clock the same evening some of the larva; had moulted for the first time. The eggs are closely felted to the abdomen, aud, as in all Decapods, they are cemented together by a secretion which possibly comes from the oviducts during ovulatiou. They are fastened by the same substance to the hairs which fringe the bases of the pleopods, chiefly to those of the first aud second pairs. * Besides Milne-Ed wards figure (4), evidently made from aspecimeu in which the natural colors had been removed by alcohol. (See remarks, etc., under Section iv.) MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 341 After a moult the colors are, as is usual, very bright, and the moulted skin, as it stands intact supported by the an tenure, may easily be mistaken for the living animal. These prawns make no sounds and appear to be very timid. The surface of the whole anterior body and of the large claws is thickly beset with tooth-like spines, the points of which are bent forward, and these may be regarded as an admirable protection against being swallowed head first by an enemy. It is also interesting to notice that the spines of the hinder part of the body project backward, and may thus be of service to Steuopns when attacked from the rear. Their long sensitive an tenure or "feelers" and well-developed eyes doubtless warn them of approaching enemies, which, by their rapid angular movements, they may easily escape. The extraordinary development of the eyes in the older larva? (PI. n) is remarkable. The geographical distribution of Stcnopu* Mxpidus is very interesting.* H. Milne-Edwards, in his "Histoire uaturelle desCrustaceV' (:i), gives the habitat of Stenopus hispidm (Latreille) as the "Indian Ocean," following Olivier (1) and the older writers. In the " Kegne Animal" of Cuvier, third edition, "Les Crustace's," p. 137, lie says: "We know of only one species, reported from the Australian seas by Perou and Lesneur." The Samaraug naturalists (5) met with it on the coasts of Borneo and at the Philippines in 1843-'4G. Dana, in 1838-'42, on the Wilkes Expe- dition (6), found it in the South Pacific at the points already noticed. In 1872 E. von Martens (7) describes the species for the first time from the Atlantic, in a collection of Cuban Crustacea made by Dr. J. Gundlach, and de Man in 1888 (9) quotes it from Amboina in his monograph on the Decapoda and Stomatopoda collected in the Indian Archipelago by Dr. J. Brock. We can now add to this list the Bahama Islands (Abaco and New Providence). We have also the interesting fact that the larva was taken on our coast at Beaufort, N. C., whither it had probably been carried by the warm waters of the Gulf Stream. We may thfrefore expect to find the adult Steuopus on the Florida Keys, but not much farther north, since this is essentially a tropical form. We thus have in Stenopus hispidus another instance of not only the same genus, but also the identical species, occurring on the eastern shores of two continents. It seems not impossible that the prolonged larval period which this animal possesses may have played an important part in its geographical distribution. This may be also true of Oonodactylus chiragra, but on the other hand it can not be asserted of Limulus. In the last case the Asiatic and American forms are specifically distinct. II. SEGMENTATION AND EARLY PHASES OF THE EGG. The prawn, which hatched her zoe'a brood on the 4th of June, laid eggs the next morning prob- ably at about G o'clock, and as soon as discovered some of these ova were hardened at intervals of a few hours during the next two days. In this way a complete history of the segmentation was obtained. First stage.— The first eggs preserved (probably 5 to 6 hours after ovulation) are perfectly opaque, nothing but the light-green yolk corpuscles showing through the shell or egg-envelopes. Thin sections prove that the segmentation nucleus has divided, and that its two products lie remote from each other. Physiologically speaking, we now nave two cells, each consisting of a deeply staining nucleus and perinuclear protoplasm. The first segmentation is evidently central. What takes place is briefly as follows : Primitively we have a central nucleus, about which protoplasm is gathered. Around this again is the great mass of yolk, and the whole is encapsuled by the protective chorion and the secondary membrane of attachment. The first division involves only the nucleus and surrounding protoplasm. The products as independent bodies now leave their central position and seek the surface of the egg. In one instance one of these has reached the surface (shown in PI. vi, Fig. 1), while the other is only halfway there on the opposite side. The superficial cell, as seen by the figure, has the same characters as when burie'd in the yolk. In * The reason for considering the Bakanian form identical with the Hispidn.s of Olivier, Latreillo, Milne-Edwards, Adams, Dana, and others are given on page 351. 342 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. another egg of the same phase neither cell is quite at the surface, so that the example given in Fig. 1 may be taken to illustrate a tendency, not a rule.* The yolk (Fig. 1, Y. C.) consists here, as in subsequent stages, of homogeneous and tolerably uniform green corpuscles. No vacuolar cavities are to be seen. Second stage. — Four and one-half hours later the second segmentation is going on or is already completed. The egg from which Fig. 2 was taken contains four nuclei with perinuclear protoplasm. It was cut into a series of 50 sections, of which the second represented is the twenty-first. Nuclei occur in sections 21, 25, 2!), and 35, none being as yet superficial. A portion of section 21 (Fig. 2) is shown under a higher power in Fig. 3. Third xta(/e. — After three hours and twenty-five minutes have passed the third phase is reached and we have eight cells, around which the yolk is superficially constricted into eight corresponding lobes or segments. A surface view of this entire egg and also a section through it are shown in PI. VI. Figs. 5 and 6, and a tangential section of one of the nuclei and lobes is given with more detail in Fig. 4. The constriction furrows appear to be considerably deeper than they actually are, and we might predicate a total segmentation of this egg without the knowledge which the section affords. AYe have here a merely superficial indentation of the yolk, the great central mass of which is undi- vided. It is a close approach to the yolk pyramid stage seen in Astacus, Alpheus, Ilippa, Pahie- monetes, and many other Decapods. The dividing planes, Figs. 7 and 8 (unless artificially pro- duced), do not penetrate into the egg. The furrows extend iuward to a plane just below or on a level with the nucleus. Each nucleus with its outer protoplasm maybe spoken of as the cell, and it is hardly probable that there is any protoplasm like that surrounding the nucleus in the other parts of the egg. The nuclei increase gradually in size, as seen by comparing the figures of successive stages, and the surrounding plasm, which they manufacture out of the yolk, is also of greater bulk. Each is a flattened, oval disc, shown well in transverse section in Fig. 5 at a, and tangentially in Fig. 4. It contains coarse grains and granules of chromatin, and the enveloping protoplasm radiates visibly but a short distance between the yolk spherules. The long axis of each nucleus lies in a plane parallel with the surface. Cell multiplication is in all cases indirect, as my observations show to be the case with several other related forms, and this is undoubtedly the rule not only with the seg- meni ing eggs of the Decapod Crustacea, but with those of all the Metazoa. There seems to be an exception in the case of Ali>hcux minor. Fourth utinje. — After another interval of an hour and five minutes there are sixteen cells re- sulting from the fourth segmentation. The blastomeres are less sharply marked at the surface and more distinctly polygonal. Six nuclei are cut by the section given in Fig. 7. They are nearer to the surface than in the former stage. Fifth stage. — The egg represented by Fig. 8 is three hours older than the last and has thirty-, two cells and the same number of superficial segments. Up to this time the egg has exhibited radial symmetry. The nuclei are quite near the surface of the egg. They are more spherical and the investing protoplasm is less conspicuous than formerly. The fissures between contiguous blastomeres are becoming less and less prominent. i>n.i xtage. — After it had been kept three days this larva passed through -a nionlt, by which only slight changes were introduced. The fourth pair of walking legs is now distinctly jointed, the fifth remaining as a bud. The llagellum (endopodite) of the second pair of antenna? uncoils and speedily lengthens. The terga of the first and second abdominal somites bear on the lower lateral margins of each side a short tooth. The larva lived four days without further ecdysis. Figs. 20, 27, 30, and .°>1 are from this stage. G. MfiKtif/opiix stage [PL xi, Figs. 28, 29, 32-34, PI. xnj, (Length = 9""").— An older larva, caught in the net on May 7, is shown on PI. xn. The most striking features of this form are the long trailing antenna? (tlagella of the outer pair), the actual length of which is about 1 inch, which is more than twice the length of the larva. The remarkable eyes which this animal possesses- give it a very odd appearance. They are placed at the extremity of club-shaped stalks, each of which is nearly 2""" long. The distance between the eyes is 4.7""". In passing to the adult stage the eye-stalks are much reduced. The outer antenna? have a short peduncle; along scale, armed with stiff hairs on the inner margin, and a long fiagellum, all very much as in the adult prawn. (PI. xii, and PI. xm, Figs. 40, 41.) The first pair of antenna? are much less like the adult form. (PI. xn and Fig. 40). The stalk is longer and more slender than in the full-grown condition. The flagella are short, the inner one still rudimentary, and the sensory hairs are retained. The carapace has developed on it a lateral furrow, which is surmounted by a conspicuous spine placed on either side at a point one third the distance from the rostrum to the posterior end of the carapace. • The rostrum is short and stout, bent upward, and does not reach beyond a line passing through the vesicula auditoria. The front of the carapace bears also a short dentiform prodftss on each side below the rostrum. These are the only indications of the future spinous armature of this region of the body. The abdomen and abdominal appendages are about as repre- sented in Fig. 27. The telson is a short triangular plate, garnished with short bristles, and is terminated by a pair of very small spines. The uropods are provided with a close, fringe of inter- locking, plumose hairs, which are longest on the inner margin. The outer lamella is one-third longer than the inner and three times as long as the telson. The first ami second maxilla? of this larva are represented in Figs. 2$ and 20. In Figs. 12, 19, 29, and 38 we have four stages in the evolution of the first maxilla, and we see that it undergoes comparatively little change. No trace of a palpus (endopodite) was seen in the specimen exam- ined (Fig. 29), the appendage consisting of*a small inner (coxopodite) and a larger outer knob- shaped branch (basipodite), each armed with short tooth-like spines. The second maxilla, has also the adult character. (Figs. 28, 42.) It consists of an elongated outer plate (scaphognathite), fringed with a single row of plumose hairs; a palp-like endopodite, and an innermost lobulated division (basipodite and coxopodite), each part carrying a few bristles. The first maxilliped of stage F is given in Fig. 30. It-consists of a basal portion (coxopodite), which bears an inner and larger lobe (basipodite), having bristles on its proximal border; an exo- podite tipped with a pair of bristles, and of an intermediate bud (endopodite) bearing a single bristle. Part of the second maxillipeds is shown in PL xi and also in Fig. 31 (St. F.). The exopodite is rudimentary. The outer segments are covered with spinons bristles. We see already a resemblance between these appendages and their adult forms. (Figs. 43, 45.) The third pair of maxillipeds are still the largest limbs. (PL xi, Mxp. m.) The terminal joint bears several long spines. Compare with the adult limb seen in Fig. 40. The pereiopods are slender appendages, of which the third pair are longest, as in the adult; the second are longer than the first; the fourth and fifth are rudimentary. One of the first pair of pereiopods is represented in Fig. 33. This appendage is nonehelate, unlike the adult stage; all its segments are armed with long spines, and there is a cluster of serrate bristles on the inner side of the proximal end of the terminal segment, and near it, a similar cluster on the next. Similar tufts of hair are found on the adult appendage. (PL xm, Fig. 47.) The terminal joint of the sec- ond thoracic limb is shown in Fig. 32; the basal extremity of the third, the, fourth, and fifth are given in Fig. 34. The second and third pair are. chelate; the fourth is rudimentary; the fifth is still a bud. The abdominal appendages, excepting the sixth pair described above, are all uuira- roous. (Fig. 27.) 348 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. This larva is colorless, excepting large spots of reddish pigment, distributed much as iu the previous stage. There is a spot near the extremity of the eye-stalk and similar ones on the abdo- ineu. Some of the appendages are also tipped with brownish red. The attainment of the adult characters is now mainly a question of the further growth of parts already present. The above outline gives us a pretty complete history of the metamorphosis of Stenopus. Between stages B and C a lacuna occurs, but it is not difficult to bridge over this gap. The development of Stenopus is especially interesting, inasmuch as it combines certain features of the metamorphosis of Pena-us, Sergestes, Lucifer, and the Prawns in general, but it differs essentially from any of them. Detailed comparisons are purposely omitted in this paper, but we will cull attention to the apparent similarity of tbe second larva of Steuopus (PI. vni, Fig. 17) to the zoea of Callianasxn xubterranm, figured by Glaus.* The length in each case is 5""u. He says, p. 54: Die jnngen Calliauassa hirvm besitzen beira Verlassen der Eibiillcn oinc ausehuliche Griisse, sind sehr laug- gestreckt und tragen droi spaltiistigo Fiisspaare, von denen eich das Voi-dern scbou wescntlioh der Formgestaltung des spateren Maxillarfiisses niihert. Der lange Stirnschnabd, sowir dio Bestachelung don Abdomens, dessen zweites Segment mit oinem besondera langen Riickendorn bewatl'ni't 1st errinern an die oben beschriebene larve. which applies perfectly to the Steuopus zoea, except that the latter has the first thoracic segment with its appendages, while, according to Clans, the first zoea of Callianassa has not, although his figure is not clear on this point. The rostrum, eyes, antenna;, second maxillie, and maxillipeds are nearly identical in the two forms. The differences are in the shape of the telson and in the condition of the thoracic appendages. The tail tin has a convex posterior edge, a median and two lateral, short spines, and eleven intermediate pairs. The rudiments of the sixth pair of abdomi- nal appendages show through the integument. Behind the maxillipeds, already "die kurzen, schlauchformigen Anlageu samm timber Thoracalfiisse nnter dem Integument bemiirkbar sind." Among Ihe Prawns, Peureus has apparently preserved most completely the ancestral history of the Decapod Crustacea, and for this reason a thorough knowledge of the development of related species is very desirable. IV. THE ADULT. STENOPUS (Latreille). Cancer (Herbst). Palcemon (Olivier). Stenopus (Latreille) Le'ach, Desmarest, Eoux, Milne, Edwards, Adams, Dana, etc. Diagnosis of Stenopus h/spidus (Latreille).— Body nearly cylindrical. Carapace with prominent rostrum and distinct transverse groove. Outer antennae with long, bristle-bordered scale bent under the inner antennas toward the middle line. Second maxillipeds with epipodite and long exopodite. Third maxillipeds very long and appendicular, with a rudimentary exopodite at base. First, second, and third pairs of pereiopods chelate. The first and second pairs quite slender, ending in small shears. Third pair longest, bearing the large claws. Fourth and fifth pairs of pereiopods slender and uouchelate. Carpus and propodus of the same articulated into numerous rings. First pair of pleopods uniramous in both sexes, all the others biramous. Special description. — Length, 37-44mm (li-lf inches). There is little difference in size between the sexes, but the females are usually a trifle the larger. Color: Body invariably white, crossed by three bands of reddish crimson. Appendages col- orless, excepting the third pair of pereiopods, which are encircled by four wide zones of the same color. These markings are not of uniform tint, but vary from bright scarlet to mottled orange red. The basal joints of at least the third and fourth pairs of thoracic legs sky blue. Antenna? snow white. For further particulars under this heading, see PI. v, and Sec. i. The carapace (Fig. 37) presents a marked transverse fossa. It is covered with short dentiform spines, largest on the front. The rostrum- is elevated, extending hardly beyond the basil joint of the inner antenna?. It ends iu a sharp terminal spine and carries six to seven stout, curved teeth on the dorsal median line, besides a single spinule projecting downwards near the tip. From the single dorsal row of teeth two similar rows diverge, extending back to the transverse furrow. The *C. Clans : " Untersuchungen zur Erforschung der geuealogischen Grundlage des Crustaceen-Systems." Wien, 1^711. Taf. vni, Fig. 1 ; also Figs. 2-7. MEMOIRS OF TOE NATIONAL ACADEMY OF SCIENCES. 349 rostrum also bears oil each side a single row of about four teeth projecting forward. The under side is unarmed. Several large spines occur in the angle behind the eye and on the basal joints of the antennae. The epidermic spines, which are characteristic of the Hispidus, though not confined to this species, are found upon the dorsal surface of the entire body, on the third pair of pereiopods and on the bases of the appendages generally. The first, second, fourth, and fifth pairs of thoracic legs are destitute of conspicuous spines. The spines of the carapace and anterior abdominal terga are bent forward; those of the fourth, liftli, and sixth abdominal somites and of the tail tin are oppressed, stouter, uondentiiate, and point backwards. The telson is arrowhead-shaped; its free edges are garnished with short, closely set hairs; it has a median groove, bordered on either side by a longitudinal elevated ridge, bearing spines; it hardly surpasses the uropodal lamellae. The eyes project at right angles to the long axis of the body. They have dark brownish black pigment and are mounted upon short, stout stalks, covered with small prickles. The labrum consists of a semicircular bar, the convex surface ot which points forward and bears two nearly median spines projecting downward. From its concave border is suspended a Ungulate appendage, which is supported by a thin, median, and vertical plate. The inner antenna; (Fig. 40) bear very long flagella, the disposition of which has already been noticed (Sec. i). The segments of the stalk are armed with stout denticles, and each division of the proximal portion of the outer tiagellum or exopodite bears externally a sharp spine. The outer antenna) (Fig. 41) possess at their base a long, narrow scale (exopodite), "which is traversed by longitudinal grooves. Their inner borders, which meet in the middle line, are fringed with closely set hairs. The stalk or protopodite is spiny, and the flagellum or endopodite is two and a half times the length of the body of the animal. The mandibles (Fig. 39) bear very large palpi, and have blunt interlocking teeth ; a transverse furrow divides the cutting surfaces of ea.ch. The first pair of maxilhe (Fig. 38) consist of an inner (coxopodite) and outer branch (basipodite), with a slender endopodite. The outer division or coxopodite is thickly beset with strong spines. The second pair of maxilla? (Fig. 42) are furnished with an elongated plate, the "bailer" or scaphoguathite, which is fringed with hairs, an inner lobulated portion (basipoditeaud coxopo- dite), and an intermediate eudopodite, which bears several plumose hairs at its distal end. The first pair of maxillipeds (Fig. 43) consist of an inner lobulated portion (coxopodite and basipodite), thickly studded with short bristles, an outer triangular plate (epipodite), and two intermediate appendages. The innermost of the latter (eudopodite) terminates in a stout spine. It consists of two segments armed without by a row of long plumose hairs. The whip-like, appendage exopodite next this is twice as long and is bordered with short hairs. In the second pair of maxillipeds the basil portion consists of several lobules, tufted with hairs, and a small, external epipodite. There is a stout incurved eudopodite, with hirsute terminal joints, and a Jong slender exopodite. A transparent lamella springs from the outer side of the proximal half of the the eudopodite, and bears plumose hairs on its free margin. The third pair of maxillipeds (Fig. 46) are long and conspicuous, somewhat less slender than the first or second pairs of thoracic legs. The inner and outer borders are fringed with long hairs. The outer border is denticulated ; the distal extremities of the segments, as of the ischiopodite, produced into a sharp spine. The basipodite is small, bearing the persistent and rudi- mentary exopodite, which is a slender palp equal in length to the ischiopodite. The first pair of dereiopods (Fig. 47) are small, slender, and chelate. The second pair of pereiopods are similar to the first pair, but longer. The third pair of pereiopods, the " great chela;," differ somewhat in size, the right being some- times larger and sometimes smaller than the left. The chela is compressed and slightly twisted. There is a single row of stout regular denticles, forming a saw-tooth edge on either margin of the " palm," and several rows of lesser spines on the broad sides. There is also a longitudinal groove extending to the base of the dactyle. The carpus is prismatic and bears about five rows of large teeth. The ischium is more cylindrical, but similar. The dactyle and propodns possess each a prominent tooth, which fits into a corresponding depression. 350 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. The fourth pair of pereiopods (Fig. 48) end iu short bifid dactyles, the terminal claw bearing a shorter proximal oiie below. The propodus is superficially segmented into from five to seven rings, which vary in size. The right propodus may have five rings, the left seven. The carpus is articulated into ten to twelve segments, commonly twelve, of variable or equal size. The fifth pair of pereiopods is similar to the fourth, but shorter. The propodus bears from six to seven seg- ments, the carpus from eight to twelve. In the fourth and fifth thoracic legs the number of rings into which the propodus aud carpus are divided differ within the above limits in different individuals of either sex and on the right and left sides of the same individual. The pleopods are all biramous, excepting the first pair, iu which the endopodite is suppressed as shown in Fig. 44. This pair of appendages is much smaller in the male. In the female the first swimmerets are nearly as long as the following pairs and are fringed with long setae. Measurements (in millimeters). [Locality: Nassau, New Providence, Bahama Islands.] 43 16 Id '10 44 44 40 8 7 r> 4 ~> g 1 2 r> ID 1 :. ;i r :, :, :t 3 a s 3 4 f> .1 !> 8 3 H :i 4 2 6 r. 2 7 45 1 73 86 98 !i 1 * 1 5 1 r> i" ii9 0 4 I.-1 5 2 5 5 1 5 • 20 5 5 1 2 1 «> 6 •27 6.5 3 1.5 9 7 48 19 5 3 8 7 3 •> 11 8 53 22 6 4 Q Leu«"tli of left third pere*opod . ... y 1 2 5 9 5 12 3 4 n 46 3 4 12 51 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. (in millimeters) — Continued. 351 Sex - . & 9 j 9 ,t IS 19 5 5 4 3 2 9 11 4 4 Greatest breadth of same without Hnines .. - . 3 3 12 12 5 36 1.5 1.5 5 5 5 5 14 18 12 12 10 11 Greatest breadth of meros 1.4 30 5 6 G t; « 15 16 13 12 9 5 11 3 7 iilns (Lair.): Distribution : («) Indian Ocean, Borneo, and Philippines (Adams). (6) Pauuiotu Islands aud Balabac Passage, north of Borneo (Dana). (c) Amboyna, Cuba (Von Martens). (d) Abaco aud New Providence, Bahama Islands. (c) "Red Sea, Indian Ocean, Indian Archipelago, New Guinea" (do Man). (2) Stenopus spiaonni (Hisso) : Mediterranean (Heller), teste Von Martens and de Man. (3) Stenopus ensiferus (Dana): Fiji Islands. (4) Stenopus aemUeemt ( Ton ilartens) : (Out) specimen in the Berlin Zoological Museum, purporting to have come from the West Indies. Length 12""". Von Martens.) (5) Stenopus teniiirostrit (de Man) : Amboyua: Length 24ulm. (More closely allied to Stenopus spiuosus of the Mediterauean than to Stenopus hispidus, and is the representative of the former in the Indian Oceau ; de Man.) STENOPUS LITERATURE. (1) Olivier: Encyclopedic Me'thodiqtie, Hist. Nat. Insectes, t. viii, p. 666, 1811. (2) Latreille: Encyclopedic Methudique, Hist. Nat. Crustacea, Arachnidses, et Insectes, t. 10, Paris, 1682. (:!) Milne Edwards, H. : Hist. Nat. des Crustaces, t. 2, p. 40K, 1837. (4) AlUne Edwards, H. : Le Rcgne Auiuial, Cnvier ; Les Crustacea, with Atlas, by Milne Edwards, p. 137. (5) Adams and White: The Zoology of the Voyage of H. M. S. Samaraug, 1843-6, p. 61, London, 1850. (6) Dana, J. D. : U. S. Exploring Exped. U. S. N., 1838-1842, vol. xiii, pt. 1, Crustacea, p. 607. (7) Miirtcny, E. v. : Uober Cubanische Crustaceeu ; nach den Sanmihmgen Dr. J. Guudlach. Archiv. f. Naturgesch., 3S. Jahrg., Bd. 2, 187-2, p. 143. (8) Heller: Crustacean des siidlichon Etiropa, S. 299. (I have seen only references to this paper.) (9) De Man, J. G.: Bericht liber dieim iudischeu Archipel von Dr. J. Brock gesamnielteu Decapoden und Stomato- podeu. Separat-Ausgabe aus deui Archiv. f. Naturgesch., 53. Jahrg., pp. 21ii-l>00, 17. Taf., Berlin, 1888. CHAPTER III. THE HABITS AND METAMORPHOSIS OF GONODACTYLUS CHIRAGRA. By W. K. BROOKS. (With PI. i, in, xiv, and xv.) THE STRUCTURE AND HABITS OF THE ADULT. This well-known species is found along the shores aud islands of all tropical and subtropical sea£, and our collections contain specimens from the Atlantic, the Pacific, and the Indian Oceans. Among the many localities where its presence has been recorded the following may be named: Bermuda, Florida Keys, Bahama Islands, Cuba, St. Thomas, Brazil, Mediterranean, Cape St. Eoque, Samboanga, Samboanga Banks. Nicobars, Red Sea, Amboina, Indian Ocean, New Guinea. It is subject to but little variation, notwithstanding its very wide distribution, and also notwith- standing the fact that there are several other distinct species of Gonodactylus extremely similar to cbiragra, aud distinguishable from it by only very minute differences. There is a well-marked chiragra-like group of species all so close to each other that their divergence from each other must have been comparatively recent, and in view of this fact it seems remarkable that one of these species should so persistently retain its identity when exposed to such a wide diversity of conditions. The species may be thus characterized : Stomatopoda with the sixth abdominal somite sepa- rated from the telsou by a movable joint; the hind body convex; and the dactyle of the raptorial claw without spines and enlarged at the base ; rostrum consisting of a transverse proximal portion more than twice as wide as long, with subacute antero lateral angles and a slender, acute median spine which does not quite reach to the bases of the eyes ; carapace nearly rectangular, three fifths as long as wide, leaving the dorsal surface of the second thoracic somite completely exposed ; au- tero-lateral angles semicircular and projecting beyond the median gastric area, which is nearly flat, and bounded by two nearly parallel gastric sutures, which are continued to the posterior edge of the carapace, which is nearly transverse with rounded postero-lateral angles; the transverse cervical suture is faintly marked, distant from the anterior margin about two-thirds of the length of the carapace; second thoracic somite, somewhat narrower than the carapace, with acute lateral angles; the eight following somites equal in width and wider than the carapace; the third, fourth, aud fifth thoracic somites about equal in length; the lateral margins of the third are straight, with rounded angles, aud as wide as the dorsal portion ; the fourth is narrowed a little towards the lateral edge, and the fifth still more so ; dorsal surfaces of the free thoracic somites and of the first five abdominal somites smooth ; hind body convex ; all the abdominal somites have marginal lateral cariuff, which are nearly linear, with the anterior end only a little wider than the posterior end; postero-lateral angles rounded in the first four abdominal somites, rectangular in the fifth, aud acutely pointed in the sixth; there are no dorsal carimB on the first five abdominal somites, and no median dorsal carina on the sixth, which carries three pairs of swollen convex lateral cariiue, which are equal in length and end posteriorly in acute spines, which are occasionally wanting on the submediau pair; the external carina is much less swollen than the others, and it unites at its posterior end with the laterial marginal carina; the spines of all the carinse project beyond the posterior edge of the somite aud lie in the same transverse plane. 353 S. Mis. 94 23 354 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. The fifth abdominal somite is somewhat louger thau those in front of it, aud about twice as long as the sixth. The telson sometimes presents slight variations, but most of its characteristics are well marked, so that there is usually no difficulty in distinguishing the species by examining it. It is considerably wider than long, and its median portion is occupied by a rounded prominence, which consists of three broad, convex rounded cariiue, none of them ending in spines ; the median one is louger than the others aud spatulate at its posterior end, while the others have both ends obtusely rounded aud alike ; external to the proximal end of each lateral carina, and almost directly under the tip of the second, or intermediate dorsal carina of the sixth abdominal somite there is a small polished hemispherical tubercle. The edge of the telsou is folded into six teeth, of which the sub-medians are largest aud project farthest backwards; the tips of the intermediates are distinct and reach about halfway to the tips of the submedians; the laterals are obsolete on the dorsal surface, al- though thin, small tips are distiuctly visible on the flat ventral surface of the telson ; each of these six teeth carries a dorsal carina ; that of the lateral is marginal and nearly linear, while the otliers lie in the dorsal axes of the teeth and are thick and convex; that which lies above the subuiedian tooth is short, and lies in the same longitudinal plane as the external cariua of the median promi- nence of the telson, while that which lies above the intermediate tooth runs nearly to the anterior edge of the telson; the median edges of the submedian teeth are minutely serrated, slightly con- cave, and meeting at an acute angle. There is a minute, nearly obsolete, tooth in the angle be- tween the submedian and the intermediate, aud the tips of the submedians are occasionally, but exceptionally, tipped by movable acute spines. The dorsal surface of the basal joint of the uropod ends posteriorly in an acute spine with a small lobe on the outer side of the base: its ventral sur- face ends posteriorly in a curved process divided into two acute curved spines, of which the outer is much the stouter and usually considerably louger than the inner, although they are occasionally nearly equal ; the outer one has no marginal tooth. The paddle of the exopodite is about halt' as long as the second joint, which carries a central terminal immovable spine, and usually eleven — rarely twelve, aud still more rarely ten — movable spines, of which nine are marginal aud the tenth and eleventh terminal, largest, and central to the paddle. The eyes are cylindrical, with rounded cornea}; and the first and second antennae are about equal in length, and more than half of t!>e second joint of the shaft of the first antenna is exposed in front of the eye. In the Bahama Islands this species presents two well-marked color variations, which occur side by side, specimens of both sorts being often found in burrows less than au inch apart. In the one form the color is a uniform dull-olive without spots or markings of any sort, as shown in PI. in; while the other form, which is copied iu PI. 1, Fig. 2, is more transparent aud is delicately mottled over the entire dorsal surface iu au iutricate but constant pattern of greyish-green pigment so distributed as to form three transverse bands across the carapace and the large joints of the raptorial claws and line transverse bands across the telson, while over the rest of the dorsal surface it forms a complicated reticulum. This difference is not sexual, for I found both males and females of each color; nor is it distinctive of age, for, while all the largest specimens were of the uniform green color, I found specimens of each color of all sizes except the largest. It is not probable that there are two constant color varieties living side by side in the Bahama Islands, and 1 am disposed to think that the mottled transparent specimens are those which have recently moulted, aud that the color becomes more uniform as the cuticle hardens. Iu the Bahama Islands this species inhabits burrows which it constructs iu the coral rock or in masses of coral in shallow water, ami, as nearly all the localities where its presence has been recorded are in the coral area, it is probable that this habit is pretty generally retained by the species all over its habitat. I have found it most abundant in lagoons and sounds on shelving- beaches which are bare or nearly bare at low tide; and when a beach of this description is over hung by a limestone cliff, from which fragments fall into the water, each fragment is honeycombed by their burrows. A crack or natural depression in the rock seems to be selected by the animal when about to construct a new burrow, for most of the burrows opened into such cracks. The mouth of the burrow is nearly circular and only a little larger thau the body of its inhabitant, but just within it widens out into a flask-shaped cave (PI. in), with smooth, even walls aud regular curvature, aud large enough for the auimal to coil up or turn around inside it. Most of the burrows are horizontal, but many are vertical with the opening below, and a few are vertical with the opening above. 1VIEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 355 The animals usually rest coiled up, with tin- eyes :i.iul antenna- directed outwards, just within the month of the burrow. They are always on the alert and reach out and snap at every small animal which approaches, even when it is two or three times larger than the Gouodactylus. They rarely pursue their prey, at least in the day time, and while a bait held near the mouth of the bur- trow will usually tempt them as far out as the. body can be stretched without leaving the burrow they seldom go any further. In aquaria, they are much more active at night than in the daytime, and they may possibly wander more in search of prey at night than I have ever seen them do in the daytime. They are solitary in their habits, and I have never found two in the same burrow. They are pugnacious to an astonishing degree, and their fighting habits, as I have observed them iu aquaria, are so fixed and constant that they must be constantly exercised by the animals when at home. When two .specimens are placed together in an aquarium they at first appear to be un- conscious of each other, but more careful examination will show that their eye stalks are in con- stant motion following each movement of the enemy. They soon assume a position in which they are face to face, although they may be on opposite sides of the aquarium, and the constant motion of their eye stalks shows how intently each movement is watched. Soon one attempts to get be- hind the other, but each such attempt is frustrated, until finally they are brought close together, face to face, and soon one springs suddenly upon the other and attempts to pinch some unprotected part. They then spring apart and eye each other again to repeat the attack at short intervals until one is disabled ; the other then springs upon him and soon tears him limb from limb, dis- jointing all the free somites of the body and tearing out and devouring the flesh. I was not able to learn how the burrows are made, for none which I kept in captivity made burrows. The regularity and smoothness of the burrows and their adaptation to the shape and size of the body indicate that they are constructed by the animals themselves. The habit of bur- rowing iu hard rock instead of soft mud is a fortunate one for the naturalist ; for, while it is almost impossible to obtain the eggs of an ordinary Stomatopod without using a steam dredging machine; it is easy to get those of Gouodactylus by breaking up the rock iu which it lives. While adult Stomatopods are abundant and widely distributed, their eggs are almost unknown, for most of them inhabit deep burrows under the water, where it is no easy matter to capture the adults, and even when these are caught they do not carry eggs even in the breeding season, for the eggs are not fastened to the appendages as they are in most Crustacea, but are deposited at the bottoms of the inaccessible burrows. As they are dependent upon the aeration which is pro- duced by the current of water which the parent pumps through the burrow by means of the valve- like paddles of the abdominal feet, they die wheu deprived of this current. The eggs are sometimes obtained1", but unless they are found in an advanced stage of development it is difficult to rear them, and I know of no Stomatopod which has been reared from the egg tinder observation except the Bahama Gonodactylus chiragra. As the pelagic larvae are large and conspicuous they are. often captured at the surface of the ocean iu the tow net, and the number of genera and species of Stomatopod larv;e which have beeu described is nearly equal to the number of adult species which are known, and the opportunity to identify even one of these larvaj by actually rearing it from the egg is a most noteworthy and important occasion. The habits of the Bahama Gonodactylus afford this opportunity; for the nature of the rock which it inhabits prevents the construction of a deep burrow, and as the fragments of rock may easily be carried ashore and broken up the eggs can be obtained without difficulty. At the time of my first visit to the Bahamas I was engaged in correcting the proofs of my report on the Chal- lenger Stomatopods, and one of the motives of the expedition was the hope that I might possibly ohta.ii! Stomatopod eggs. A day or two after our arrival Dr. E. A. Andrews brought me a Gono- dactylus and a bunch of yellow eggs, which he had picked out of a rock which he had broken to pieces while searching for Annelids. The eggs were newly laid, and, while they were obviously those of some crustacean, there was no evidence that they belonged to Gonodactylus except the fact that they were found among the fragments of a rode which also contained this animal. As soon as I saw the eggs and heard how they had been obtained I started for a point where the beach was covered with fragments of coral rock. It was then late in the afternoon and growing dark, but I waded into the water and carried ashore as large a rock as I could lift. After I had thrown this on to a larger rock and broken it to pieces there was just daylight enough to show me 356 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES* the Gouodactyli .scattering in all directions, and the masses of yellow eggs which were spattered over the large rock which I had used as an anvil; but the problem was solved, and I went home and to bed, confident that I should next day get all the embryological material I needed. As shown in PI. in, the animal molds or shapes the mass of eggs into a hemispherical cap, which fits over the convexity of the hind body and lies between it and the stone wall of the bur- row. The parent reaches out to snatch at passing prey, but so long as she is undisturbed she remains in the burrow. When the burrow is broken open she quickly rolls the eggs into a ball, folds them under her body in a big armful, between the large joints of her raptorial claws, and endeavors to escape with them to a place of safety. The promptness with which this action is per- formed would seem to indicate that it is an instinct which has been acquired to meet some danger which frequently presents itself. It would seem as if a cave in a solid rock were a pretty safe refuge from all euemies except a naturalist with a geological hammer, and it is difficult to say what the accident is which has thus been provided against. The larger heads of growing coral are often broken off by the waves, and loose fragments of rock are overturned by severe storms, and it is possible that, when alarmed by a violent shock, it liees from its cave to escape the danger of being crushed when the rock is torn from its place and turned over. At any rate its habit is the reverse of that of most burrowing animals, for they usually retreat to the depths of the burrow when alarmed. This is true of all the Stomatopods which 1 have had an opportunity to observe except this species, and the chief use of the burrow of tiquilla cmpusa is for refuge in danger, while LysiostjuiUa c.cc/inttrix darts down its burrow at the least alarm and can not be driven out even when the sand has been dug up on all sides of it. THE METAMORPHOSIS OF GONODACTYLTJS CHIRAGRA. That feature of the life of Stoiuatopods upon which new data are most to be desired is the history of the early larval stages, and an abundant supply of the eggs of Gmtodactylus chiragra rendered it an easy matter to obtain this history for that species. I also obtained a complete scries of eggs for studying the embryology, but, as a few preliminary sections showed that this was of slight interest and that there is no essential difference from other Macroura as regards the egg embryology, this subject was not studied. Most of our knowledge of the metamorphosis of Stomatopods is based upon the comparative study of collections of alcoholic specimens, and the direct observations on living larva; are very scanty. In 188U Faxen published an account (Selections from Embryological Monographs com- piled by Alexander Agassiz, Walter Faxon, and E. L. Mark, I Crustacea, Cambridge, ISSL', Hull. Mus. Cornp. Zoiil., Vol. ix, No. 1, PI. vin, Figs. t! and 3) of observations made, three years before upon a young tiquilla empusa which he had reared from an Alima larva; and in a paper which was published in 1879 I described (On the larval stages of Squill a cmpuna) a series of similar larv;e which I had studied while they were alive, and which was sufficiently complete to warrant the statement that they were the young of Nqmllacmpuxit, and that this species probably hatches from the egg in the Alima stage. In my report on the Challenger Stomatopods (Report of the Scientific Results of the Voyage of H. M. S. Cluillenycr during the years iS7;5-'7(i, xvr, part XLV, 1880) I have given an account of the metamorphosis of Lyxiosquilla (witrithi.v which 1 had reared at Heaufort, N. C. : but except for these observations our knowledge of Stomatopod metamorphosis rests upon the comparative study of preserved specimens, and, while the series which are picked out from miscellaneous collections sometimes present pretty satisfactory evidence as to the adults which they represent, this sort of indirect evidence can not be conclusive. Large and varied collections of larvae have been compared for the purpose of selecting those which form stages in the same series, and of ascertaining as accurately as possible the adult affini- ties of tli coldest larva;, by Glaus (Die Metamorphoseu der Squilliden, AbhandL d. /.-. Gesellseh. , and in side view in PI. xiv, Fig. 5. The rostrum is now greatly elongated and reaches to the tips of the antennules. Small anterolateral spines have made their appearance, as well ;is a small spine external to the base of each posterolateral. These latter are greatly elongated aud very slightly divergent. A great change in the shape of the carapace has taken place, as will be seen by comparing Fig. 7 of PI. xv with Fig. it. Its lateral margins are nearly parallel, aud its greatest width only a little exceeds that of the abdomen. Its posterior border is now nearly transverse and crosses the middle line above the last thoracic somite. The MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 359 sixth abdominal somite bas separated from the telson, but its appendages are not yet developed. The scale of the antenna is now fringed with hairs, and the eyes are divergent, with well devel- oped stalks. The raptorial claws have greatly increased in size and are beginning to approximate to the adult form, while at the earlier stage they closely resembled the chela' of the third, fourth, and fifth pairs of maxillipeds of an adult Stomatopod. From this time on to the end of its larval life the yonug Erichthus of Gonodactylus ch iragra presents the characteristics of that larval type for which I have proposed the provisional name Gouerichthus; and, while the resemblance grows stronger as the larva grows older, it is unmistakable even now, and still clearer after the next molt, wbeu it assumes the form sbow/n in PI. xiv, Fig. (5, from above, and obliquely from below in PL xv, Fig. 10. The autennulary flagella are now beginning to elongate, and that of the antenna is now rep- resented by a bud, but there are no new appendages, although the sixth abdominal somite is uow indicated. Although it is very much younger than £he Gonericbthi shown in my Challenger report in PI. XV, Figs. 1, 5, 6, and 11, it resembles these larvae in the following features as well as in many minor points : The rostrum is long and reaches beyond the tips of the antennules, and it lias four or five median teeth on its ventral surface. The anterolateral angles of the carapace end in acute spines pointing forwards, and the anterior edges are inclined towards each other, so as to make at the base of the rostrum an angle a little greater than a right angle. The lateral borders of the carapace are nearly parallel, and the posterolateral spines long, slightly divergent, and with a small acute spine external to the base of each. The carapace covers all the five thoracic somites and all or nearly all of the first abdominal, and its posterior border is transverse. The median dorsal spine, which was carried on the posterior edge of the carapace of the younger larvae (Figs. 3, 4, and 5 of PL xiv), has disappeared, although it persists until a much later stage, in the Iarva3 shown in Figs. 1, C, and 11. of the Challenger report. The hind body is now nearly three-fourths as wide as the carapace. The lateral margins of the telson still carry, as they did during the earlier stages, four nearly equal marginal spines on each side; of these the most anterior is the external, the next the inter- mediate, the third a secondary spinule, and the fourth, which, at the stage shown in Fig. 6, PL xiv, forms the posterolateral angle of the telson, is the submedian. The posterior border between the submedians is very slightly notched and nearly transverse. All the Challenger Gonerichthi are very much older than this larva, and their telsons are more developed. The spines especially are much more elongated ; but in Figs. 5 and 0 of PL XV of the report the secondary spine can be clearly recognized about halfway between the snbuiediau and the intermediate. With the assumption of the form shown in PL xv, Fig. 10, the habits of the larva undergo a sudden change. Up to this time, while able to 'swim briskly about by the use of their abdom- inal appendages, they spent most of their time near the bottom of the aquarium, seldom going up more than an inch or two, although they are quite able to reach the top of the water, which was about 10 inches deep, and when masses of Nudibranch eggs were suspended hear the surface of the water they quickly discovered and fastened upon them. Up to this time, also, they were peaceful and did not attack each other. Several hundred sur- vived the molt which precedes the beginning of their pelagic life, but all of them soon died and none passed this stage, which is the one shown in Fig. 10. They now left the bottom, and became rest- less, swimming continually at all levels in the water. They refused to touch the eggs of which up to this time they had been so fond, and I could find nothing else which suited their appetites, but their proper food is, beyond question, small swimming animals of some sort, for they now began fighting among themselves, and when two met they would seize each other with their raptorial claws, and then tumble over and over together, until they struck the bottom, when both died. The survivors would not touch the dead bodies, although most of them soon shared the same fate, and the rest became weak and soon died. At the same time that I was studying the growth of the captive larvrel captured several older ones in the surface net, and one of them somewhat older than Fig. 10 is shown in Figs. 11 and 12. The third, fourth, and fifth maxillipeds are now developed and are like those of the adult; and 360 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. tlie three pairs of free thoracic legs, and the uropods are represented by buds. An umber of moults and probably aii interval of many weeks intervenes between this stage and the one shown in PI. xv, Fig. Jl of the Challenger report. The life history of this species of Gonodactylus, in the Bahama Islands at least, is thus seen to bo extremely simple. It hatches as an Erichthus and remains an Erichthus until it assumes its adult form ; and as the successive appendages make their appearance they have from the first the structure which they are to retain through life. The statement which I made in my Chullcnt/cr report (p. 55), that Gouodactylus hatches from the egg in the Krichthoidina stage and subsequently changes into an Erichthus, is an error, at least so far as Gonodactylus chiragra is concerned, although it is possible, in view of the great variation which we have observed in a single species of Alpheus, that in other regions, where the adults have different habits, the larva may hatch in a younger stage. Coral-dwelling Crustacea seem to exhibit a tendency towards the abridgment of their metamorphosis, and it is not at all improbable that other species of Gouodactylus may have an Erichthoidina stage. The Challenger collection contains a bottle of very minute and young larva- in the Erichthoi- dina stage, and one of these is shown iu Fig. 3 of PI. XII of my report. Comparison between this and the newly hatched Erichthus of our species, PI. xiv, Fig. .'3, will show many points of resem- blance, and future research may possibly prove that it is the larva of Gouodactylus, although the statement that all Gouodactyli hatch as Erichthoidiua; is au error. CHAPTER IV. THE METAMORPHOSIS OF ALPHEUS. By W. K. BROOK AND F. H. HERRICK. (With Pis. I, II, IV, XVI to XXIV. ) SECTION I. — THE METAMORPHOSIS or ALPHEUS MINOR FROM BEAUFORT, NORTH CAROLINA. This small species is found in abundance at Beaufort, North Carolina, and in the Bahama Islands, and it is no doubt widely distributed along our southern coast. At Beaufort it is found in shallow vertical burrows in the sandy mud which forms the bottom of most of the land-locked sounds between tide marks. It is also met occasionally in shells, and under loose stones and oyster shells. During its development, between the time when it-natches from the egg and the time when it acquires the adult form, it passes through a lung metamorphosis, divided into many stages. Its life history has been traced by one of the authors at Beaufort, and by the other at Nassau, and the individuals from both these localities pass through exactly the same series of changes. As we also find that other species, such as Alpham normani, pass through the same metamorphosis, the life history of Alplieux minor may bo regarded at the primitive or ancestral life history of the genus, which originally characterized all the species ; although it is now retained in its perfect form by only a few, and has undergone secondary or recent modifications in the others. THE FIRST AND SECOND LARVAL STAGES. The stage in which the larva hatches from the egg is of very short duration, as it molts and passes into the second stage within a few hours after hatching. No drawings of it were made before the change, but this is very slight, and the description of the second stage holds true in all essentials of the first stage, except that the tips of the exopodites of the three pairs of maxillipeds, and the plumose hairs on the autennules and antenna5, are not fully extended until after the change. The second larval stage is shown in PI. xvi, Fig. 2, and in PI. xvn, Fig. 2, and various organs of the larva during the first stage are shown in PI. xvi, Figs. 4, 6, 7, and 8, and PI. xvin, Fig. 4. In PI. xvi, Fig. 4, is the antenna of the first larval stage, Fig. G, the first maxilla, Fig. 7, the second maxilla, Fig. 8, the mandible, and Fig. 4 of PI. xviii, the first maxilliped. As shown in PI. xvil, Fig. 2, and in PI. xvi, Fig. 2, the locomotor organs of the larva during the first and second stage are the plumose exopodites of the anteiuiie and of the three pairs of maxillipeds. There are no functional appendages posterior to the maxillipeds, and the large eyes are freely movable and entirely uncovered. The larva has all its appendages fully developed and functional as far back as the third pair of maxillipeds. Following these are three bud-like rudiments of the first, second, and fifth pairs of thoracic limbs, and posterior to these a long tapering abdomen, divided into six segments, there being at this time no joint between the telsou and the sixth abdominal segment. During the first 361 362 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. stage there are uo traces of any abdominal appendages, but in the second stage, the outlines of the sixth pair are faintly visible under the cuticle of the telson, as shown in PI. xvi, Fig. 2. The stomach is almost completely free from yolk, and the surface of the body is marked by red and yellow pigment spots, which are very constant in position and number, and are well shown in the figures. As shown in PI. xvi, Fig. 2, the antennule consists of a stout shaft composed of a long basal portion with no trace of an ear and a ranch shorter distal joint, which carries externally a ranch shorter and smaller joint with four sensory hairs, and internally a long slender plnrao.se. hair, which is not fully extended until after the first moult. At this stage this hair is almost sessile upon the shaft, although its base is destined to give rise to the long flabellum of the antennule of the adult. The antenna has a large exopodite, which is fringed with plumose hairs, and is an efficient organ of locomotion. During the first larval stage, this exopodite, which is destined to become the flat scale of the adult antenna, is cylindrical and distinctly annnlated, as shown in PL xvi, Fig- 4. At this stage it is divided into a basal portion and live movable joints, about equal in total length to the basal portion. After the lirst molt the aunulations become less distinct, although the "scale" is still cylindrical, as shown in PI. xvr, Fig. 2. The basal joint of the antenna is about equal in length to the " scale," undivided, and it carries upou the inner edge of its distal extremity a small, short, movable joint, with a single, long, plumose hair, which is "telescoped" before the first moult, but fully extended afterwards. This short joint is the rudimentary anteunal fla- gellum, which in the adult is equal in length to the entire body of the animal. The mandible is shown in Fig. S. It is deeply cleft into two branches, the outer one with two rows of large, strongly marked dentations, and the inner one with a rudimentary palpus, two rows of hairs, and a, finely serrated cutting edge. The first maxilla is very small, but it does not appear to be rudimentary. It is shown in PI. xvi, Fig. C. No exopodite could be made out. There is a small endopodite, with one long, plumose hair, and two basal joints, one with two sharp cutting hairs and the other with one. The second maxilla is shown in PI. xvi, Fig. 7. The two basal joints arc- feebly indicated, and each carries three slender, simple hairs. The endopodite carries two terminal hairs, and the flat exopodite is fringed by seven. I could not determine whether these hairs are plumose or not. The three pairs of maxillipeds are functional and they present features which are characteristic of the genus Alpheus (see PI. xvi, Fig. 2). Each has a large, flattened, polygonal, basal joint, which carries upon its inner edge a few short, sharp teeth, and upon its outer edge a long, flat exopodite, with plumose swimming hairs, and an endopodite which presents several peculiar features. The endopodite of the first maxilliped is very short and two-jointed, that of the second is somewhat longer and five-jointed, while that of the third is very greatly elongated, without traces of joints, and ending in a long, simple hair which, as shown in PI. Vvm, Fig. 4, is telescoped before the first moult, but immediately afterwards becomes lengthened, as shown in PI. xvi, Fig. 2, until it reaches forward beyond the tips of the antennules and antennas. Following the maxillipeds are three pairs of buds to represent the first, second, and fifth pairs of thoracic limbs. The first bud consists of a single branch, which is shown by its subsequent history to be the exopodite. The second has two branches, a short exopodite, and an extremely short endopodite, while the third consists of a somewhat longer, but still rudimentary, shaft, which represents the endopodite of the fifth thoracic limb, and has uo trace of an exopodite. The hind body is divided by joints into five abdominal somites, behind which is a long undi- vided region to represent the sixth abdominal somite and the telson. Before the first monlt none of the abdominal appendages are present, but after this molt the sixth pair are faintly indicated under the integument of the telson, as shown in PI. xvi, Fig. 2. The telson itself is broad, sub triangular, with its posterior border nearly straight, and transverse. It carries eight pairs of stout plumose hairs, of which three pairs are much longer than any of the others and nearly equal. This set of three spines is placed at the angle of the posterior edge on a lobe or elongation, which, while it is so slightly marked as to scarcely interrupt the triangular outline, is still very distinct and easily recognizable. Of the remaining five spines on each side of the middle line, one is on the external border, and the other four on the posterior edge between the group of three and the middle line. The internal one is very small and might easily be overlooked. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 363 THE THIRD LARVAL STAGE. (PI. xvi. Fig. 1.) After molting the second time the larva assumes the form shown in PI. xvi, Fig. 1. It is also shown, much less enlarged, in side view in PI. xvii, Fig. 1. The first and fifth thoracic limbs are DOW functional, the second is represented by a bud, all the abdominal somites are distinct, and the sixth abdominal appendage has made its appearance. The first five abdominal appendages are still unrepresented, and the endopodite of the sixth is rudimentary, although its exopodite is fully developed and functional. Those appendages which were present in stage two have undergone little change. The external branch of the antenuule has, in place of the four seuse-hairs of the earlier stage, only two, which are much longer than before. The long terminal hair of the inner branch has lost the marginal hairs of the earlier stage aryl is now simple, while two plnino.se hairs have made their appearance on the lower surface of the distal joint of the shaft. The scale of the antenna is still cylindrical, but the annulations which marked it during the earlier stage have disappeared. The flagellum still consists of only one short joint, and the long terminal hair which it carried at the earlier stage has disappeared. The mandibles, maxillae, and maxillipeds are about as they were before, but the eudopodite of the third maxilliped has almost completely lost the long terminal hair of stage two, and has also become relatively shorter, and is now divided into four joints. The first thoracic leg, which was rudimentary in stage two, has now acquired a flat basal joint and a plumose exopodite, like those of the preceding appendages, but the endopodite is represented only by a rudimentary knob or bud upon the anterior edge of the basal joint. The second thoracic limb is, as it was at the earlier stage, a two-lobed bud. No buds have as yet appeared between it and the base of the fifth thoracic appendage, which is now fully developed and forms the most conspicuous peculiarity of this stage in the development of Alpheus. It has no exopodite, its basal joint is not enlarged nor flattened, and its long, slender, cylindrical shaft, made up at this stage of four joints, is prolonged at its tip into a long, slender, tapering, simple hair, the end of which reaches beyond the tips of the antenna? when the appendage is in the posi- tion shown in the figure PI. xvi, Fig. 1. The appendage seems to have little power of motion and it seldom deviates much from the position shown in the drawing, being usually carried closely pressed against the ventral surface of the body between the liases of the other appendages, with its tip directed forward. All six abdominal somites are distinct and movable, but the first five have as yet no traces of appendages. The first four somites are short and equal, the fifth is nearly as long as the first four together, and the sixth is very narrow and almost twice as long as the fifth. The eudopodite of the sixth abdominal appendage is present and of considerable size, but it is not as yet functional, although the exopodite, which is not very much larger, is fringed by six long, plumose, swimming hairs and is used in locomotion. The two spines which are carried upon the lateral margins of the telsou at an earlier stage have disappeared, and there is less dif- ference than before in the relative sizes of the others, but the general form is the same. FOURTH LARVAL STAGE. The subsequent history of Alpheus minor was traced by one of the authors at Beaufort and by the other at Nassau, but as the stages which follow were found to be almost exactly like the corresponding stage of other species which had already been drawn, it did not seem to be advisa- ble to make new figures, and in the remainder of the description the illustrations which are referred to actually represent the larva; of other species. After its third molt the larva of Alpheus minor passes into its fourth stage, when it becomes almost exactly like the fourth larval stage of Alpheus heternchelis, shown in PI. xvm, Fig. 3. There is little change at the anterior end of the body, except that the carapace now begins to extend over the eyes, and the ears have made their appearance in the basal joints of the antennnles. The mandible has lost its outer branch, and the basal joint of the second maxilla, PI. xvi, Fig. 5, carries on its inner edge three hairy lobes. There are now five pairs of swimming appendages in place of the three of stages one and two, and the four of stage three. These five are the exopodites of the first, second, and thud maxillipeds 3G4 MEMOIttS OF THE NATIONAL ACADEMY OF SCIENCES. and those of the first and second thoracic legs. The eudopodites of the maxillipeds are as before. The eudopodite of the first thoracic leg, which was represented in stage three by a rudimentary bad, now appears to be entirely wanting. The second thoracic linib, which in stage three was represented by a bilobed bnd, now consists of a basal joint, with a large, functional, plumose exopodite and a rudimentary, bud-like endopodite. Between this appendage and the base of the fully developed fifth thoracic limb there is a row of buds to represent the third and fourth thoracic limbs, which became developed after the next molt. The fifth is about as it was in the preceding stage, and it carries no trace of an exopodite. The abdomen is about as before, except that the endopodite of the- sixth abdominal appendage, the only one yet represented, is now fully devel- oped and fringed like the exopodite by long, plumose, swimming hairs. The telsou has becomjs elongated and narrow, and the spines upon its posterior end are much smaller than before. THE FIFTH LARVAL STAGE. None of the figures of the larva; of other species exactly represent the larva of Alpheus minor after the next molt. The eyes are now partially covered by the carapace, and the swimming organs are the seven pairs of fully developed exopodites belonging to the three pairs of ruaxillipeds and the first four pairs of thoracic legs. At this stage these four pairs of appendages reacquire their endopodites, and the anterior end of the body is similar to that of the larva shown in PI. xxi, Fig. 1, from which, however, it differs greatly as regards the telsou and the sixth abdominal ap- pendage. The first five abdominal appendages are nou represented by buds like those shown in PI. xxi, Fig. 1, and in PI. xix, Figs. 1 and 2, but the terminal portion of the abdomen is nearly like that of Fig. 3 in PI. xx. The telson is greatly elongated, narrow, and its terminal spines are very small. THE OLDER LARVAL STAGES OF ALPHETJS MINOR. During the successive molts the abdominal appendages become fully developed, the eyes be- come completely covered by the anterior edge of the carapace, the antennre become elongated, the autennnle develops a scale, the swimming exopodites of the maxillipeds and thoracic legs disap- pear, these appendages assume their adult form, and acquire gills, aud the animal gradually be- comes like the one shown in PL xx,,Fig. 2, which is a young Alpheus of another species. THE METAMORPHOSIS OF ALPIIETJS HETEROCHELIS FROM THE BAHAMA ISLANDS. Ill the Bahama Islands this species passes through a series of stages which, except for a few minor differences of detail, are exactly like those in the life history which lias just been described. This fact is remarkable when it is known that the life history of the same species is very different at Beaufort, North Carolina, and that Packard has described still another life history for specimens of the same species which he studied at Key West. FIRST LARVAL STAGE. The Bahama specimens hatch from the egg in the stage shown in side view in Fig. 1 of Pl.xviil. As this larva agrees in all details of its structure with the first stage of Alpli-eus minor shown in PI. xvii, Fig. 2, already described, no further description is necessary. THE SECOND LARVAL STAGE. Like Alpheus minus the Bahama specimens of Alpheun keterochelis molt within a few hours after hatching, but they undergo no essential change, and PI. xvi, Fig. 2, exhibits all the essential characteristics, although this figure was drawn from a specimen of Alplieus minor. The most noteworthy specific difference is in the relative length of the marginal spines of the telson. In the first and second larval stages of both species there are eight pairs of spines, one pair on the outer edge and seven on the posterior edge, as shown for Alplieits minor in PI. xvi. Fig. 2, and for Alphtus JieterncJielis in PI. xvi, Fig. 3. In both species the pair next the median line are rudimentary and the next pair very small, but the three which arise from the rounded angle of the MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. 365 telson are much more nearly equal to the others in .!//>/«•».« heterockelis than in Athens minor. If, as seems probable, the triangular telson of the macrouran zoiia is a secondary modification of the deeply furcated telson of a more ancient proto/,oea, then the first larval stages of Alpheus minor are in this respect more primitive or protozoeau than those of Alpheus keterochelis. THE THIRD LARVAL STAGE. This is shown from below in PI. xvin, Fig. 2, and a comparison with Fig. 1 of PI. XVI will show its very close resemblance to Alphata in in its at the same stage. The only essential difference between them relates to the- rudimentary thoracic limbs. In both species the first thoracic limb has a functional swimming exopodite and a rudimentary eudopodite, and in both the fifth thoracic limb has a greatly elongated jointed cylindrical endopodite and no exopodite, but between these limbs Alpheus hetcrochelis has buds to represent the other three pairs of thoracic limbs, while Alpheus minor has buds for only one pair, and the other buds do not appear until after the next molt. THE FOURTH LARVAL STAGE OF ALPHEUS IIETEROCIIKLIS. This is shown from below in PI. xvin, Fig. 3, and there are no noteworthy differences between it and Atyheus minor. THE LATER STAGES OF THE BAHAMA ALPHEUS HETEROCUELIS. % The transformation of the larva into the adult Alphens occupies a number of molts, and the general character of the changes will be understood by the study of PI. xix and xx, although these plates were drawn from Beaufort specimens of the species. THE METAMORPHOSIS OF ALPHEUS HETEROCHELIS FROM BEAUFORT, NORTH CAROLINA. As shown in PI. xx, Fig. 1, this, before it hatches from the egg, reaches a stage of develop- ment which somewhat resembles stages two and three of the Bahama specimens. There are many important differences however, and the stage in which it hatches is not directly comparable with any stage in the life of the Bahama form, nor in that of Aipheus minor. Just before hatching it has. like the Bahama form immediately after hatching, three pairs of fully developed swimming maxillipeds, but it also has buds to represent all five pairs of thoracic legs. The antennary scale and flagellum are much more advanced than they are at a much later stage in the Bahama form, and the abdomen is much more distinctly segmented. The larva, immediately after hatching, is shown in side view in PI. xix, Fig. 2, and in ventral view in Fig. 1. The autennule and antenna are shown on a larger scale in Figs. 3 and 4, and the mandible and first and second maxilhe in Figs. 5, 6, and 7 of the same plate. The animal now has all the appendages which are present in the adult, but all behind the maxillipeds are rudimentary, although they all become functional after the first molt, as shown in PI. xx, Fig. 3. The antennule, PI. Xix, Fig. 3, has a long cylindrical shaft made up of three joints fringed with plumose hairs and terminating in an exopodite with sensory hairs and an eudopodite or flagelluni, which is short and rudimentary but much longer than it is in the younger stages of the Bahama specimens. The antenna, Fig. 4, presents even greater differences. The flagelluni is about as long as the scale, and two jointed, while the scale itself is flat, although its tip still pre- sents traces of a primitive segmented condition. It is, however, of little use in swimming, and in fact the larva has at this stage only very feeble locomotive power. The eyes are stalked and movable and almost completely uncovered. The mandible is simple and without a palpus, as shown in Fig. 5. The first maxilla, Fig. 6, is very small, but apparently it is not rudimentary as its two lobes carry cutting hairs. The second maxilla, Fig. 7, is a broad flat plate, very much more developed than that of the newly hatched Bahama specimen shown in PI. XVI, Fig. 5. The three pairs of maxillipeds (PI. xix, Fig. 1) are almost exactly like those of the newly hatched Bahama larva (PI. xvm, Fig. 1) or those of the Alpheus minor at the same, stage (PI. xvi, Fig. 2), but the thoracic appendages (PI. Xix, Fig. 1) are entirely different, and the ventral surface of the body is covered by a mass of limbs closely crowded, all pretty well developed, but all as yet 366 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. fuuctiouless. Careful examination shows that there are tive pairs (the five pairs of thoracic limbs), ami that all but the last pair are biramous. In all, the exopoilites are longer than the cndopodites, which decrease in length from in front backwards, while the eudopodites increase in length. The later history of these limbs shows that the exopodites never become functional, aa they do in the Bahama form. All six abdominal somites are distinct, although the line separating the sixth from the telson is faintly marked. The first five pairs of abdominal feet are represented by five biramous buds projecting beyond the outline of the body, while the sixth pair are only faintly outlined under the cuticle of the telson, which itself presents a most important difference from that of the young Bahama larva, as it is not triangular, but spatnlate ; and of the eight pairs of setaj the three pairs which in Al/iltcnK minor He on the lobe at the angle of the telsou are not on a distinct lobe, nor do they differ in size from the adjacent seta?. This larva, molts a few hours after hatching, and at once undergoes the most profound changes, and assumes the form shown in PI. xx, Fig. 3. It is no longer a larva, but a young Alpheus. The eyes are almost covered by the carapace, the car is well developed, and all the, appendages are present and functional and essentially like those of the adult. The autennule has two Hagclla, each with several joints. The flagellutn of the antenna is more than twice as long as the scale ami is composed of twenty-two joints, while the scale has -its final form. The first maxilla (Fig. 5) has a large club-shaped lobe, fringed with short hairs, and a rudi- mentary eiidopodite, while the second maxilla (Fig. (>) is a broad flat plate with cutting lobes and a short, rod like endopodite. The three pairs of maxillipeds (Figs. 7, S, and 9) have assumed the char- acteristic Macrourau form and are no longer concerned in locomotion, while the thoracic limbs have elongated into the tive pairs of ambulatory appendages of the adult, although they still retain their rudimentary exopodites. The abdomen is now like that of the adult, and the telson (Fig. 4) is long and narrow. An older specimen is shown in Fig. li and a still older one in PI. xvn, Fig. 3. Comparing the history of the Bahama form with that of the North Carolina form, the most conspicuous peculiarity, aud that which first attracts attention, is tl.e great abbreviation of the latter. The Beaufort specimens hatch in a much more advanced condition than the Bahama speci- mens, and, while the latter pass through many larval stages, the former quickly assume the adult form. This is not all, nor is it even the most fundamental difference between them. The develop- ment of the Beaufort specimens is not simply accelerated; it is profoundly modified, so that nc exact parallel can be drawn between any larval stage of the one and a stage of the other. The statement that the Ueaufort specimens pass, before leaving the egg, through stages which are exhibited during the free life of the Bahama specimens would do violence to the facts; for the difference between them is very much more fundamental than this statement would imply. For example, the Bahama form has at first three, then four, then five, aud then seven schi/.opod feet with functional swimming exopodites, while the Beaufort form never has more than three. As regards the thoracic region aud the first five abdominal appendages the Beaufort larva, at the time of hatching (PI. xrx, Fig. 1), is more advanced than the fourth larval stage of the Bahama form (PI. xvm, Fig.;-!), while the sixth pair of abdominal appendages are like those of the Bahama form at the time of hatching ( PI. xvi, Fig. 3). In the Bahama form the first and fifth thoracic limbs are the oldest, and the others appear in succession from in front backwards ; all live pairs make their appearance together in the Beaufort form. In the Bahama form the sixth pair of abdominal feet appear before and in the Beaufort form alter the others. Many minor differences of the same general character show that we have to do with profound modification of the life history rather than with simple acceleration. THE DEVELOPMENT OF ALPHEUS HETEKOCHELIS FROM KEY WEST. According to Packard's account the specimens of ,1 l/ikcx n Jit-terochi'lin which occur at Key West differ from those which occur at Beaufort in about the same way that the latter differ from those from the Bahamas, as the metamorphosis appears to be entirely absent in the Key West speci- mens. Packard states that, while still inside the egg, they had all the appendages of the adult in essentially the adult form. There were tive pairs of thoracic legs and the first pair had large MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 367 cbeliie, aud the eyes were nearly sessile. In this case also there seems to be modification as well as acceleration, as Packard says that there were only five 'pairs of abdominal feet and that these were well developed. It may seem to some that the fact that these three forms present such great aud constant differences in development is a reason for regarding them as three distinct species, but, whether we hold that they belong to cue, two, or three species, they will still furnish proof of the existence of profound modifications in the life histories of adults which have remained almost exactly alike. Careful and minute comparison between adult specimens from Beaufort and Nassau showed the closest agreement in nearly all particulars (v. Chap, v, Pt. First, Section n), and it has there- fore seemed best for us to regard them as belonging to a single species; the more so since our discovery that different individuals of another species found at Nassau (Alpheus saulcyi) differ from one another during their larval §tages in somewhat the same way that the Beaufort specimens of heterochelis differ from the Bahama specimens. Alpheus minor aud Alpheus heterochelis are very distinct species. The adults have diverged from one another so far that one could not possibly be mistaken for the other ; yet the life history of the Bahama heterochelis is so exactly like that of Alpheus minor, both at Beaufort and in the Bahamas, that the same figures of the early stages will serve for both ; for the larval stages of heterochelis have undergone local modifications, while the adults have remained almost absolutely unchanged, except as regards the reproductive elements aud their product. t SECTION V.— LARVAL DEVELOPMENT OF ALPHEUS SAULCYI. An egg of Alpheus saulcyi just ready to hatch is shown in PI. xxi, Fig. 5. The large claws are plainly visible through the transparent shell. The auteume are folded back alongside the body, while the abdominal aud closely packed thoracic appendages are directed forward. The telson overlaps the head. First larva (length, = -^ inch). — Fig. 1 shows the larva as it is just hatched. It belongs to the variety found in the brown sponges. The various parts may be seen more highly magnified in PI. xxi, Figs. 4, 6, 7,9, and PI. xxri, Figs. 1-8, 12. In both varieties the animal hatched as a schizo- pod, loosely infolded in a larval skin, but uot invariably, as I have noticed that in one or two cases, where females of the lougicarpus with very few, perhaps half a dozen eggs, produced young, the metamorphosis was completely lost, the larva; being in a stage corresponding to that usually at- tained after the second molt aud represented in PI. xxi, Fig. 8 This is referred to again at the end of the section. To return to the first larva (PI. xxi, Fig. 1) ; this is fifteen one-hundredths of an inch long. It is semi-transparent and colorless, except for spots of characteristic red and yellow pigment sprinkled freely on the abdomen, the telsou, and appendages. Rudimentary gills are present and a remnant of uuabsorbed green yolk is conspicuous in the stomach. The carapace covers the bases of all the thoracic appendages but the last pair. It is produced forward into a short simple spine, the ros- trum, which extends between the eyes. There is a rudiment, on either side, of the ocular spines (PI. xxii, Fig. 6), which soon grow forward and give to the front the characteristic trident shape. The eyes project forward, only the extreme base of the stalk being covered by the carapace. A median eye or ocellus is present just below and between the bases of the lateral eye stalks. Both pairs of auteunre are biramous and jointed. The auteuuules (Fig. 8) consist of a stout peduncle, a short eudopodite, and a shorter bud or outer branch, which bears several bunches of sensory filaments. The peduncle is composed of three segments, as in the adult; the basal joint being four times the length of either of the other two, and bearing on its outer side a rudimentary aural scale. The upper margin of each joint carries one or more plumose hairs. The auteuure (PI. xxu, Fig. 7) are formed on the adult plan. There is an inner auteimal stalk consisting of two joints, bearing a rudimentary flagellant, and an outer scale or exopodite. The distal margin of the exopodite is garnished with plumose hairs and carries a short outer spur. The mandibles (Fig. 12, drawn from a larva after the first moult) arc deeply cleft, as in the adult. The outer branch is deutated at its distal end aud carries a palpus. The first maxilhe (Fig. 0, 368 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. shown with more detail in Fig. 3, PI. XXII) have adult characters. They are biramous. The endo- podite is stout and toothed at its apex. The more slender outer division bears a short spine near the distal end. In the second 7iiaxill;e (Fig. 6, PL xxi) the scaphoguatliite or respiratory plate is most prominent. This is now composed of an anterior portion, bordered with from six to twelve long plumose hairs and a posterior, rudimentary, and hairless lobe. The inner division (eudopodite) has the adult form, while the innermost lobes of the adult appeudage (PI. xxiv. Fig. 9) are unrepresented. The maxillipeds are all birainous appendages, and their exopodites are the principal swim- ming organs. The endopodite of the lirst pair is short and stout and divided at its tip. That of the third pair is three-jointed and equal in length to the exopodite. In the first pair of thoracic legs (PI. xxi, Figs. 4 and 1) the inequality of the cliche is very marked, and, as we have already seen, it is so for some time before hatching. Individuals differ .somewhat in this respect. The articulations of the carpus and meros are distinct. The exopodites of this and of the three suc- ceeding pairs of thoracic limbs are tipped with rudimentary in vagina ted hairs. The second pair of pereiopods (PI. xxn, Fig. 1) are chelate, but the articulations of the carpus are not distinct. The third pair of pereiopods (Fig. -') end in bideutated dactyles and have short exopodites. The fifth pair are without swimming organs. All the abdominal appendages are present and functional, excepting the sixth pair. They have only very short hairs until after the lirst moult. The first pair (L'l. xxn, Fig. 5) consist of a larger outer and smaller inner blade. This endopodite remains rudimentary in the adult male, but nearly equals the exopodite in length in the female, as will be seen by reference to PI. XXIV, Figs. 1 and 5. This convenient sexual mark probably appears early, but can not be relied upon at this stage. The second (I'l. xxn, Fig. 1) and three succeeding pairs of pleopods have a stout base, an outer blade like that of the first pair, and a shorter endopodite which bears on its inner margin a lobule or palp. The sixth pair, or uropods (I'l. x\i, Fig. !>), are not yet free. The inner and smaller divisions point forward, meeting on the middle line. The telsmi, which termi- nates the body, covering the outer uropodal limits, is a r ided, spatula te pi ale, with a median notch. Its tree posterior edge is fringed with seven pairs of plumose spines, the first or median pair being rudimentary, and the next four succeeding pairs long and nearly equal. Sci-ainl larva (length, ,',,"„ inch). — The lirst moult takes place either immediately or very soon after hatching. The animal as it now appears is shown in PI. xxi, Fig. U. The principal external changes thus produced are the. following: (1) The rostrum and ocular arches extend farther over the eyes. (2) Both divisions of the autennules are considerably extended. The llagella, of I he antennae are from three to four times their former size and are articulated into twenty to thirty rings, the scale sill I not passing the peduncle. ('.}) The thoracic appendages have more of the adidt characteristics. The articulations of the carpus of the second pair are distinct. The exo- podites of the lirst four pairs are functional, and the last pair has grown forward. (4) The pleopods presently acquire swimming hairs; the telson plate is free and the uropods are func- tional for the first time. (5) The last thoracic segment is still uncovered and the eyes are incompletely hooded. Third l/irni (length, about '; inch). — The third larva as it appears after the second moult, which takes place in twenty-five to thirty hours after hatching, is represented in PI. xxi, Fig. 8. It has now the general adult character, and can not be called a larva in the strict sense. At even this early age the pugnacious instinct is strong, and although only about one-sixth of an inch long, it snaps audibly the fingers of its large "hand," which is carried extended forward. It also swims on the bottom of the jar in all respects like the adult. Only a few globules of yolk remain in the stomach. The gills are now quite prominent. They are evidently functional to some degree, and were so, possibly, at an earlier date. The yellow and red pigment cells have nearly all disappeared or are temporarily withdrawn from view. A most prominent change at the second moult is the extension forward of the rostrum and the ocular spines, which form a hood over each eye. The autennal peduncle surpasses the scale, and its fhigellum nearly equals the carapace in length. As in the adult, the large chehe are very prominent. The exopodites of the thoracic appendages have dwindled to rudiments. The view of the head of a four-days old Alpheus is shown in Fig. 3, PI. xxi. MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. 369 The fourth form (after third moult).— When six or seven days, old the third moult is passed, but only slight changes are introduced. The small chela aud the inner and outer anteuuie of this phase are given in Figs. 9, 10, 16, PI. xxn. The inner branch of the antennules is still relatively short; the basal or aural spine extends to nearly the end of the first joint. The bristle-bordered plate of the autenute has now developed a considerable spine near its outer extremity, a rudiment of which appears in the first larva (Fig. 7). This represents the squamal spine, to which the plate is ordinarily attached, in the adult. The spine is here developed from the plate. The latter may disappear, as we shall see further on, to be finally regenerated from the base of the spine. The small chela has the adult form. The fifth form (after fourth moult). — These animals moulted the fourth time ten days after hatching. Very little change was apparent, except in size, aud beyond this point we did not follow them. METAMORPHOSIS OF ALPHEUS SAULCYI FURTHER ABBREVIATED. As was stated above, the metamorphosis of Alpheus saulcyi may be still further accelerated so as to practically disappear altogether. This fact is illustrated by a young Alpheus hatched in a glass dish April 25 (Fig. 17, PI. XXii). The prawn (var. longivarputs) was taken from a brown sponge. The eggs, half a dozen in number, were slow in developing. The small chela is shown in Fig. 15. This phase corresponds with that usually attained after the second moult (shown in Fig. 8, PL XXI), with which it corresponds in size and color. All the thoracic aud abdominal appendages have nearly the adult form, the exopoclites of the former being rudimentary, as in Fig. 8. The large chela is most prominent, being nearly as large again as the smaller one. The eyes are partly hooded, but not so much as the four-day old prawn represented by Fig. 3. The Alpheus had to be held in a compressorium in order to be drawn, so that the parts are slightly distorted by pressure. At the time of hatching most of the hairs on the appendages generally are in a rudimentary condition. S. Mis. 94 .24 V. ALPHEUS: A STUDY IN THE DEVELOPMENT OF CRUSTACEA. By FRANCIS H. HERRICK. CONTENTS. Introduction. Methods. PART FIRST : I. The habits and color variations of Alpheus. II. Variations in Alpheus heterochelis. III. The abbreviated development of Alpheus and its relation to the environment. IV. The adult. V. Variations from the specific type. VI. Measurements. VII. The causes and significance of variation in .11- pheus saulcyi. PART SECOND : I. Structure of the first larva of Alpln-ns sanlcyi. II. The origin of ovarian eggs in Alpheus, Homarus, and Palinurus. III. Segmentation in llpheus minim. IV. The development of Alpheus. First stage: Segmentation to formation of blas- toderm. Second stage: Migration ofcells from blastoderm to the interior. The iuvagination stage. Third sta/je: Optic disks and ventral plate. Fourth stage: Thickening of optic disks. Ku- dimeuts of appendages. Fifth stage: Rudiments of three pairs) of ap- pendages. Optic disks closely united by transverse cord. Degenerative changes. Sixth stage: The egg-naupliua. Seventh stage: Seven pairs of appendages formed. [With thirty PART SECOND — Continued. IV. The development of Alpheus — Continued. Eighth stage: Nine pairs of appendages present. \intli stage: Eye-pigment formed. Tenth stage: Ganglia of ventral nerve-cord distinct and completely separated from the skin. Klerenlh stage: Embryo about to batch (Al- pheus heterochelis'). TiettJ'lh xlage: First larva (Alpheus saulcyi). Thirteenth xlaut the literature of the Arthropods is very great, commensurate indeed with the size of the group. During the progess of this work a number of important papers have appeared which are referred to either in the text or in notes. While much is known of the Arthropods as a whole and of that large division of them included under the Crustacea, it is probably true that a great deal of this knowledge is of a very fragmentary and unsatisfactory nature. There is great need for detailed and full accounts of the development and organogeuy of many forms iu order that the relations of the various members of the Arthropod type may be clearly established. The present work may be regarded as a contribution toward supplying the need just men- tioned, but how imperfectly it is unnecessary to say. The plan of making observations upon other Crustacea for comparison with the more detailed studies of Alpheus has been as yet only partially carried out. The early stages of Nlenopun hispidm, Romania Americanus, and Pontonia domestica have, however, been followed, and less completely those of Hippa talpoideis and Palannonetex rulij/iria. Spence Bate (3) states that the shortened development of Alpheus was first described in his memoir, with drawings, communicated to the, Royal Society in 1870, from a specimen procured in the Mauritius. He named his specimen Hamaralpheus, "from the impression that species producing a Megalopa could uot be placed in same genus as those producing a Zoi'-a." He says: "The orig- • inal of my drawing is Li""" in length and was procured -from a specimen 14mm long, resembling the figure that I have given of Alplti'i/s minim, Say. An inspection of this drawing (3, PI. cxxil, Fig.l) leaves some doubt as to whether there was not an error in referring this form to the genus. The general shape is unlike that of Alpheus, the abdomen being three times as long as the carapace, ' and there appear to be only three pairs of thoracic appendages behind the chelipeds. Packard (40) in 1881 was the first to describe a shortened metamorphosis for Alpheux hrtcrochelis. In some brief notes published in the American Naturalist ot that year, he states that both this and the small green Alphens(A.7Kt/ms)occuriu abundance at Key West, Florida, in the excurrent open- ings of large sponges. This fact is interesting, and probably significant also, as will be later shown. Packard describes the first larva of this Florida form as much further advanced toward the adult state than is the first zoea of the Beaufort species, according to the observations of Brooks. In fact it more nearly agrees with the first larva of a Bahaman Alpheus soon to be described, in which the metamorphosis is nearly lost. The Nassau form of Alpheux lietcnx-lielin has, as I have recently ascertained, a complete metamorphosis. The bearings of these facts will be discussed further on. The larval "development of the Beaufort Alphei was studied by Brooks (7) and a short abstract of his results was published iu 1882. This is all, 1 believe, that has been previously done on the embryology of these Crustacea. Several abstracts of the present work appeared in 18S7-'8S (20-22). METHODS OF WORK. Several species of prawns, such as Stenopus and Pontonia, repeatedly laid eggs while kept in aquaria, and doubtless I should have succeeded equally well with Alpheus, if sufficient pains had been taken. As it was, only two or three individuals gratified me in this respect, but in each case, the ova failed to develop. The animals were therefore taken from the sea with eggs in the earliest phases of development, and were kept ittider observation in an aquarium for the length of time required. The ova were then carefully removed from the pleopods, and were hardened at intervals of thirty minutes or one hour or a longer timr, according to the phase or age of the embryo. By obtain- ing a number of series in this way the whole life history within the egg could be followed, and by 372 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES, this means I was able to observe the peculiar movements of the wandering cells and the formation of germ-layers, which are ofteu very difficult to interpret, when we rely upon material taken by chance. Experience with the use of Perenyi's fluid in preparing the eggs led me to discard this reagent altogether, and to substitute for it Kleiueuberg's picro-sulphuric acid, made np either with water or 30 per cent alcohol. The alcoholic solution works equally well and economizes time. The Pe- reuyi is too violent and uneven in its action. While it serves fairly well in some cases, it generally swells out the membranes or shell by the rapid eudosmosis, and distorts some part of the egg or embryo in consequence. The egg is frequently deformed and the shell ruptured. The ova should be transferred directly from the killing fluid to 70 per cent alcohol, and they will then generally retain their normal shape, and can thence be removed to alcohol of a higher grade for permanent keeping. If, however, they are carried from the Kleineiiberg fluid to a weaker alcohol (30 per cent), distortion is sure to follow, the capsule bursting and the egg sometimes exploding. Preparations of the entire embryo as well as sections were made, but very little was attempted with the living egg. For surface preparations the hardened ova. were first punctured to allow the fluid to penetrate the shell more easily and they were then stained entire, in Kleiuenberg's hu'inotoxylon. They were afterwards shelled, when this was possible ; saturated with paraffin by the turpentine-paraffin method, and were then mounted. While the paraffin was congealing they were carefully placed in position with a hand lens. This last important and often troublesome process was rendered easy by the differential property of the stain, which affects only the embry- onic cells, leaving the ylok, which in preserved eggs is of a light straw color, unaltered. The embryonic tissues are thus made to appear nearly black on a light background. The embryo was then cut away from the rest of the egg by the microtome razor; attached to the slide by collodion and mounted in balsam; or the egg was cut in two and both halves were similarly treated. All drawings which represent surface views excepting Fig. 10 were made from objects thus prepared. In general, Kleinenberg's hiemotoxylou proved to be the best staining fluid, and it is especially useful in this case, where the massive yolk contains numerous elements, the relations of which it is important to determine. The carmines are less serviceable, since the food yolk is also affected by them. Soda carmines (Beccari's formula) proves to be incapable of removing pigment from the eyes, although it is specially recommended for this purpose. This may be easily effected by soak- ing the entire tissues in very weak solutions of nitric acid for a considerable length of time. Gaule's quadruple stain of ha-inotoxylon, eosiu, satl'ranin, and nigrosin was also tried with excellent results, but this method is very laborious, and since our inquiries do not extend in most cases, to cell structure it is unneccessary.* Perenyi's fluid is sometimes available for swelling the chorion and thus aiding in its removal, although the embryo is liable to injury. It is also helpful in studying the egg with low powers. The food yolk, which is often dark green, is affected less actively by this reagent than the embry- onic tissue. The latter is turned to a waxy whiteness and is thus clearly defined for a short time, but the yolk is soon decolorized unless the eggs are transferred to water, becoming pink, and liutilly light yellow after preservation iu alcohol. PART FIRST. I. — THE HABITS AND COLOR VARIATION OF AiPHEUS. Some facts of general interest have been gathered from a study of the Alpheus iu its natural environment on the coral shores and reefs of the Bahamas, and iu giving these we will limit our- selves mainly to the three species which have contributed the material for the history of the em- bryo, viz: Alpha/* niiiniH Say, from Beaufort, N. C., A. ht'tcrocltelis Say, from Beaufort, N. C., and Nassau, New Providence, and ^4. sanlcyi, from Nassau, New Providence. The genus Alpheus comprises numerous species spread over a large part of the globe, many of which are closely connected by intermediate forms. From North America upwards of twenty species of Alpheus have been described; five are known to inhabit the eastern coast of the United States, three from Florida, and two others (A. minus and A. hcterochelis), which range from Panama to as * In studying the development of the lobster, which has also a largo egg, I have found it necessary to adopt new methods, especially iu the treatment of the eggs for surface preparations. Iu most cases the egg-membranes are best removed by the aid of hot water. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 373 far north as Virginia. From Florida and Cuba nine species are recorded. I have found twelve species of tbis prolific genus, or about one-half the number described for the whole American con- tinent, inhabiting the beautiful little reef of growing coral called Dix Point to the eastward of Nassau Harbor and along the margins of the little bay which was just in front of our laboratory. Another species (A. icebsteri Kingsley), first reported from Florida, was also discovered on Greeii Key reef, a few miles from Nassau. From collections which I made at Abaco and Andres Islands, I am led to believe that the different species are quite generally distributed in the Bahamas, and as these islands have prob- ably been largely populated from the South, we may expect the same forms to occur at Cuba and at other West Indian Islands. This genus, however widely distributed, is essentially lropic,.il and abounds in all coral seas. Of the great family of the Crustacea which make their home on the submerged reefs of growing coral, Alphcus is perhaps the most prominent and thoroughly charac- teristic. They pop out of almost every rock which is brought up from the bottom, and every loose head or block of growing coral, with its clusters of alga-, sponge, and sea fans, which you pull from the reef, resounds with the click of their little hammers. Some of. these animals lead a semi-parasitic life in sponges, or seclude themselves in the porous limestone which forms the solid floor of the beach, and others, again, live under loose shells and stones in the white coral sand. Some are highly and beautifully colored, and with few exceptions the pigment is characteristic of the species for any locality. In all cases the claws of the first pair of walking legs are enormously enlarged and serve as formidable weapons of defense so re- markable in this genus, and in most there is the greatest disparity in the size of these claws, one, either the right or left, being the larger. One species, the habits of which are peculiar, carries the larger of these claws so folded under the body as to be completely concealed. It can, however, quickly withdraw this weapon and make a rapid thrust, when an enemy comes near.* By the scis- sor-like blades of the large claws a sharp metallic report is produced. This is true of nearly all the species, and so abundant are many in these islands that a constant fusilade is kept up along some of the shores at low tide. This snapping propensity is shared by both sexes whether in or out of the water, and it is undoubtedly correlated with their pugnacious habits. If two males or females of the same or different species are placed in the same aquarium, they will dismember each other in a very short time, and one is usually literally torn to pieces. The sounds emitted by Alpltcus hcterochelis are the loudest I have heard from any member of this genus. We frequently kept this species in glass dishes in our room for several days at a time, and sharp reports like the explosion of a small torpedo or pop gnn were heard at intervals through the day and night. It sometimes, swims with its large claw so widely opened as to suggest dislo- cation. This weapon then reminds one of a cocked pistol, and the report apparently follows in the same way that the click follows the impact of the hammer on the lock. I have given this mattter no closer attention, but find that Mr. Wood-Mason, who is quoted in a notice on " Stridulating Crustacea"! in "Nature," (05). has offered another explanation. According to this observer the sound always accompanies a sudden opening of the claws to their fill lest extent, and may be caused either by impact of the dactyle upon the joint to which it is articulated or "by forcible withdrawal of the huge stopper-like tooth of the dactylopodite from its pit in the immovable arm in the claw." It seems most probable to me that the sound. is caused by impact, and most likely by the rapid closure of the finger into its socket. :f * This species is entirely new. The large concealed claw suggests a poison .apparatus. The " fingers " .are ex- ceedingly slender and sharp at the points. Although kept for over a, week in an aquarium it emitted no sounds. t According to Wood-Mason sound-producing organs in Crustacea were first brought to notice by Ililgrndorf, in V. der Decker's "Roiseu in Ost-Afric.a (Criistaceen)," and were afterwards observed by himself in his dredging ex- pedition to the Andaman Islands. The Btridnlating organs— scrapers and rasps — may bo either on the carapace and appendages or on the appendages alone. t Hoth Kent and Wood-Mason speak of the sounds emitted by the Alphei as if produced by the extension or opening of the claw. As pointed out above, it is just the other way, the sound following upon the impact of dactyle and propodus, when the tooth of the dactyle is not pulled out of its socket but driven into it. None of the conditions of piston movement are present. The walls and floor of the pit are relatively soft, while the tips of the elaw are dense and stony. The "click" can be artificially produced when the claws are clamped with rubber, whether the " stopper " is present or not. 374 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. In Alpliem heterochelis the dactyle of tlie large pincers is a curved blade which shuts down into a groove on the occludent margin of the " thumb," and closes over the latter like a pair of shears. The huge stopper like tooth is borne on the inner and proximal edge of the dactylopodite and fits neatly into a corresponding pit in the •' thumb," in line with the groove just mentioned. The object of this plug is evidently to steady the movable dactyle and to prevent lateral strain and the dislocation .which might result, and thus to give it a strong grip on any object which it has seized. In alcoholic specimens in which the relations of the parts are well preserved the stop- per works freely in and out the well, and not like a "tightly packed piston from a cylinder closed at one end." (The claw is widely opened, before the sound is produced, but the sound is not produced while the claw is open, but at the instant when it is violently and suddenly closed. It is due to the impact of the "thumb" and "finger," and 1 have freqncnth -seen specimens of A. licterochelix, when prepared for combat, facing each other for several seconds with claws distended to the utmost. In these cases the "snap" does not come until tlie claw is closed. In lighting the claw is not used as a clasper, but as a saber. The sharp external edge is a weapon of such efficiency that I have seen individuals killed and almost cut in two by a single blow. — \V. K. E.) A large brown sponge, Hi win in in-cut a, which is not to he mistaken, grows on the shallow reefs and off the. shores of all the Bahama Islands which I visited. It is found from just below low tide mark out to one-half a fathom or more of water, where its great size and sooty brown color distinguish it at once on the white bottom. These '• loggerheads" are. round and much flattened (the smaller ones more vase-shaped), and of a coriaceous texture; they sometimes measure 1A feet in diameter. There is commonly one, sometimes two, large e.xhalent chimneys into which small tish, young spring lobsters, and other Crustacea, oft en beat a, hasty retreat. It is easily broken open since ithas no consistent skeleton. If a sponge colony of this kind is pulled and torn apart, one is certain to find it swarming and crackling with a small specie.^ of Alphens, which quarter themselves in the intricately winding pores of the sponge. The sounds emitted from every fragment of these mutilated sponges remind one forcibly of " those made when sparks are taken by the nuckles from the prime conductor of a small electrical machine," as Wood-Mason remarks. Hundreds of indi- viduals may be collected from a single large specimen. These animals have an average length of about, 12mm. They are nearly colorless, excepting the large chehe, which are tipped with brown, reddish orange, or bright blue. The females are so swollen with their eggs or burdened with the weight of those attached to the abdomen that they can crawl only with great difficulty, if taken from the water. The eggs are few in number and of unusually large size, their diameter varying from one-twenty second inch to one-twenty fifth inch, and their number from six to twenty. These are most commonly yellow, but may be either bright green, olive, greenish white, brown, brownish yellow, or dull white. The ova and ovarian eggs have, always the same tint in the same individual. Although translucent and apparently colorless, upon close inspection the body is seen to be sprinkled with cells of reddish and yellow pigment. Another quite different sponge grows on all the reefs in from one to two fathoms or more of water. There are several varieties of this, which may be told by their olive green color, yellow flesh, and clumpy, irregular shape, as well as by the putrescent mucous which some of them pour out when broken open. In about nine out often of these sponges one will find a single pair of Alpheus (rarely more than this), which resemble those living in the brown sponge, but differ from them in several important points. -We are concerned at the. present with the color variations only. They are distinguished by their large size (averaging about L'.'1.' in length) and uniform color. The females exceed the males greatly in bulk, owing to the large size and number of their eggs. In both sexes the large claws are bright red (v. PL iv, and for details section iv). The female is practically inert during tlie breeding season, and at. such times is well protected in her sponge or against any green surface by the bright green ovaries which fill the whole upper part of the body and by the mass of similarly colored eggs attached to the abdomen below. Only two pairs, or four individuals, out of a hundred or more which were examined showed any variation from these colors. In these the eggs were yellow, and the pigment ou the claws was more orange than red. The table which follows shows the variations l-.etwecn two large females taken, respec- tively, from the brown and green sponges, and between the size, number, and color of the eggs. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 375 Habitat of Alpheua. Length of $ . Numlicr of "Rgs. Diameter. Color. Color of adult. Indies. Inches. Brown sponge. . . i 19 A Yellow (variable). . . . Large chelre, red (blue or brown in others. ) Green sponge 1A 347 A- Usually green ; in Large chela?, orange- this case yellow. red. These two forms, although apparently distinct, are seen, however, by closer study to belong to the same species; but besides the more superficial variations just mentioned, there are others of a more remarkable character, the morphological significance of which is considered in sections IV and v. Of this species, Alpheus saulcyi, Guerin, it is necessary, for descriptive purposes, to distin- guish two varieties, viz : Alpheus saulcyi, variety longicarpus (from brown sponges), Alpheus saulcyi, variety brevicarpus (from green sponges). These two varieties shade completely into each other by numerous intermediate forms. The longicarpus varies greatly in size and in the color of the body and eggs (besides the other more profound variations mentioned in section v), while the brevicarpus type from the green sponges is more uniform in size and stable in color and other characters. The former variety is well pro- tected from outside enemies while in the tortuous mazes of its sponge, as its great numbers would show, if any evidence under this head were needed. The enemies which invade them successfully seem to be parasites.* Possibly the variety inhabiting the green sponge does require color-protection, especially since the females are very inert during the breeding season. They are, indeed, admirably protected when exposed on the green surface of sponges, alga;, etc. The bright color on the tips of the large claws, which only are protruded from the places of concealment, recall the similarly colored heads of boring annelids, which abound on the reef, but this fact may have no significance. It seems quite probable that if we have in this Alpheus a case of protective coloring, it is due very largely to individual adaptability. This view implies great individual plasticity, which does not appear in any of the species of Alpheus known to me within- a restricted area. The colors of certain Crustacea, and also the colors of their eggs, are known to vary greatly with the surroundings. In the Alpheus, parasitic in the brown sponges, these colors vary consid- erably where the surrounding conditions are the same. However, the color of the ovarian eggs is always the same as that of those already laid, and, although these animals were kept for several days at a time in differently colored dishes, I never observed any very marked change in the color of the ovary, but these experiments were not continued long enough or carefully enough to be con- clusive. The eggs of Alpheus heterocJieUx are almost invariably of a dull olive color, while, as in the case of the parasite of the green sponge, about one in a hundred has bright yellow eggs. In the first case at least this may possibly be an instance of reversion to one of the. original colors from which the green was selected. In most species of Alpheus the color of the eggs is fixed and uniform for any locality, and, as already suggested, may have a protective significance; but in a few other cases, where this is not true, the color is not only variable in different individuals, but probably also in the same individual. Alpheus heterochelis from Beaufort, N. C., is uniformly of a dark olive-green color, with some red and blue on the appendages. It lives iu the beds of oyster shells, which are more or less *A parasitic Isopod, probably a Hop?/ run, is found on both the varieties, but is most, common with the dweller in the brown sponge. It appears as a tumid bunch, (irmly rooted in the branchial cavity or to the tinder side of the abdomen. In this connection I will mention another curious parasite which was found infesting the eggs of a single female taken from a brown sponge at Ab.ico. This is a large, spherical, unicellular organism in the encysted state. The egg, with the embryo, is packed full of them. (v. Fig. I'M and section IV, Part Second.) In looking over a collection of unpublished drawings of Crustacea, made by the associates of Louis Agassiz and deposited iu the library of the Museum of Comparative Zoology of Harvard College, I find a sketch (by H. J. Clark, December 23, 18f>7) of a Bopyrut taken from the brauchial cavity of Mjiheus helerochflis. 376 MEMOIRS OF THE NATIONAL ACADEMY OF "SCIENCES. exposed ;it low tide. Alpheus minus has a similar environment and is similarly colored. Alpheus hetcrochelis from Nassau, New Providence, on the other hand, lives under loose stones, ainid the white coral sands of the beach, and is noticeably transparent, looking as if the color had been bleached out of it. The body is sprinkled with dots of brown pigment. The claws and legs are pale greenish. Young and old are invariably colored alike. In a collection of adult Alpheus of either sex of the same or of several species, where there is a difference iu size of the large claws, it is noticed that either the right or the left, indifferently, may be the greater. As we will see, this differentiation of the chela* begins in one instance before the animal is hutched. Is this right and left handed condition to be explained by inheritance from the parents ? In about forty lame of a small brood of Alpliens saitlcyi, all invariably had the left claw enlarged, and in a smaller number (all that were preserved), from another female of the same species, the left chela was also in each case the larger. This would indicate that the young of the, same mother have always the same claw, either right or left, the greater, and that this phe- nomenon is one of direct heredity from the parents. But to prove this it is only necessary to trace right and left handed broods to parents which are themselves right and left handed, respectively. This, unfortunately, I have not done, as my attention was not called to the subject while at the seashore.* The breeding season of Alpheus begins at Beaufort, N. C., about April I. It covered the period of our stay at Nassau (March to July), and probably began earlier and lasted considerably later. t There the temperature is high and remarkably constant, the annual range being about 15° (temperature of air 70° F. in March, 80° in June), and in consequence the early phases of devel- opment are rapidly passed. Not one prawn in a hundred was found with eggs iu an earlier stage than that of yolk segmentation. II. — VARIATIONS IN ALPHETJS HETEROCHELIS. A renewed comparison of Alphcux hcterix-lirlix with the Nassau form lends support to the con- clusion already reached that we here have to do with two varieties of the same species. There are certain differences, which systematic zoologists might regard as of specific value, but they are 110 greater than we have proved to exist among individuals of the same species living in the same sponge, (v. Section V.) The Nassau specimens average smaller, but the chief difference lies iu the shape of the small chela. The propodus of this appendage in the Nassau form is relatively shorter and thicker in both sexes. Both lingers are nearly cylindrical, and covered with hairs, which are distributed either singly or in tufts. In the Beaufort hc.tcroi-he.Ux there is a striking variation in the small chela which appears to have escaped detection. Judging from the small collection at my command it is a sexual variation. In the females the small chela is like that of the Nassau form, but is usually longer and slenderer. The dactyle is about one-half the length of the propodus. In the males the dactyle is relatively much shorter, and has a median longitudinal carina which is continued into the apex of the claw. In transverse section the dactyle is trihedral, with two con- cave sides, corresponding to the deep groove on either side of the keel. These grooves are fringed with a row of stout plumose setre. Similar rows of seta; occur on the sides of the opposing " thumb.1' Perhaps the most interesting variation which I have observed iu the Beaufort IttlcrocIteUs has reference to the size of the egg. The eggs in this locality have an average diameter ot about one " Mr. .1. .1. Northrop, of Columbia College, while at Nassau in the winter and .spring of 1S90, kindly offered to collect for mo some specimens of Alplieus sauJcyi with, young. Ou February 10 he rollivieil six females, five from green sponges, one of which had a brood of sixteen young, and one small female with three lnrv:r from the "loggerhead"' sponge. In the first instance the left chela was the largest in the mother and in rnrli <>f Hi/1 sixteen young. In the latter, two had the right claw enlarged and one the left. The inference is suggested that when the claw of the same side is invariably the greater in all the young, this character is doubly inherited from both father and mother, but the data are insufficient to settle this point. t Professor H. V. Wilson found this species breeding around Rreen Turtle Key from July until December. Mr. Northrop fouud newly hatched young early in February. It therefore breeds the year through, which is probably true of many of the Crustacea. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 377 twenty-fourth ineli, but two females were found which carried a few bundles of very small eggs, nor- mally Allied to the anterior swimmerets. These eggs measured only one fifty-third to one sixty -fifth inch in diameter, that is, the contents of the smaller was about one-twelfth that of the larger egg. This occasional production of very small eggs exhibits a tendency, which is still present iu the species of this locality to-day, to revert to its old metamorphosis long since laid aside. III.— THE ABBREVIATED DEVELOPMENT OF ALPHEUS AND ITS RELATION TO THE ENVIRONMENT. Belated species, as a rule, resemble each other more iu their early stages of development than in their adult state. This is not, however, invariably true, since all animals, whether young or adult, must adapt themselves to their environment or be destroyed. It is probable that animals in all stages of growth are equally plastic and tend to vary with the varying conditions of life. The early life in large classes of the animal kingdom," as tishes, birds, and mammals, is spent either in the protecting membranes of the egg or within the body of the parent, and is thus lint slightly affected by external conditions, and suffers -little change in consequence. In other groups, on the contrary, and in the Crustacea in particular, the case is very different. Mere the young are usually hatched in a very immature condition, and lead a life of their own at the surface of the ocean, wholly independent of their parents. They have accordingly adapted themselves to this mode of life, and the variations thus entailed have led to the production of the zoea, a locomotor larva, fundamentally different from the adult. We may regard the zoea as a secondary, adaptive form, directly descended from an ancestral protozoeau type. After passing a longer or shorter period (usually of several weeks) at the surface of the sea, the adult state is gradually reached through a complicated series of changes, and the animal adapts itself to new conditions on the sea bottom or ou the shore. Now, if the habits of the adult and larva should tend to converge, if the adults should adapt themselves to an entirely new environment, which it is necessary for the young to become fitted for at once as soon as hatched, we would expect that the zoeal stages, formally assumed to bridge over a gap which no longer exists, would be dropped or shifted to the egg. This seems to have actually taken place, and is illustrated in a remarkable manner in the genus Alpheus. Most of the species, which are very numerous, inhabit the shores, in common with many related forms, and, as already stated, they abound on coral reefs. They all, as a rule, hatch as zoea-like and have a complicated metamorphosis. Two species have been discovered, however, which have adopted a parasitic life, and iu each the larval period is accelerated. In one, which is semiparasitic, the metamorphosis is partially abbreviated ; in the other, which is completely para- sitic, the metamorphosis is completely lost. Still more interesting and significant is the fact that one of the species in one locality is nonpuranitic and lias a complicated metamorphosis, while the same species from another locality is parasitic and has the metamorphosis abridged. We will now consider more particularly the history of these two forms, in order to make a clearer comparison. The species are — f Alpheus lieterochelis, from Nassau, New Providence. (1) 1 Atyhcus lieteroeltcUfi, from Beaufort, North Caralina. ( AlpheiiK hctcrochclin, from Key West, Florida. (2) Alpheus saulcyi, from Nassau, New Providence. ALPHEUS I1ETEROCHELIS FROM THE BAHAMAS. t This species, found at Nassau, exemplifies the development common to the genus, as seen, for instance, in A. normani (Kingsley), which is closely associated with it, A. minus Say, and in many other Bahainan forms It is one of the common species at Dix Point, and may be found iu abun- dance on the shore of the little bay, in pools left by the ebb tide, under shells or loose fragments of coral. First larva (length = J- inch). — The three pairs of maxillipeds, eacli with long exopodites ending in feathered hairs, are the principal locomotor organs. Two pairs of rudimentary thoracic legs are present. All the abdominal segments, but none of their appendages, are formed. 378 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. The antennules consist of a stout jointed stalk, the terminal segment of which bears four sensory filaments. A long plumose spine springs from the extremity of the second joint on the inner side. The an tenure are biramous, the two branches arising apparently out of a common basal segment. The outer division is a scale-like plate. It bears on its extreme inner border a row of plumose hairs, eight to ten in number. The endopodite is slender and shorter than the scale. It terminates in a short spine or denticle, from near the base of which springs a long plumose hair. The eyes are large and uncovered. The mandible is simple. The telson is a broad triangular plate, the terminal side of which is garnished with the charac- teristic; zoeal spines, the number, relative size, and position of which vary slightly in the different species. There are in this case eight pairs of these spines. The first or median pair is rudimen- tary; the second is not half the length of the third. The lateral angles of the plate are each prolonged into a short lobe, bearing three spines. There is no marked median notch, but there is a slight median depression. The rudiments of the sixth pair of abdominal appendages are plainly seen. Second larva. — With the first molt the endopodites of the maxillipeds lengthen. Three pairs of rudimentary pereiopods are now present, the last of which are the longest. Of the pleopods only the sixth pair are represented, whether free or not was not observed. ALPHKUS HETEROCHELIS FROM BEA0FORT. The peculiar metamorphosis of the Beaufort Hetcrochclix was described in 1884 by Brooks, who also showed th.it in this respect it departs widely from the associated Alphcusminnn. The form in question is hatched as a larva with preparations for the schizopod stage. It has the usual swimming organs, but all the thoracic legs are present in the condition of rudimentary buds. The abdoininal segments are formed, and the buds of the first five pahs of feet belonging to them. The eyes are not completely covered by the carapace. At the first molt the rudiments of the sixth pair of abdominal feet are added, and the larva, undergoes profound changes. All the ap- pendages arc now functional and the eyes are nearly hooded. With later molts the adult char- acters become more pronounced, but the marked difference of the great claws appears only after several months. ALPITEUS HETEROCI1ELIS KUl.lM FLORIDA. The short description, given by Packard in 1881, of the first larval stage of this species from Key West, where it inhabits sponges, has already bei-u alluded to. From this we infer that the development is considerably more abridged than in the Beaufort case. This is also indicated by his figure of one of the abdominal appendages. lie says : The eyes are yearly sessile, the yolk nearly absorbed, although the embryo (in the egg) was near the time of hatching. The antenna? are " well developed." All the thoracic legs are present, their joints distinct, " the first pair about twice as rhick as the others, the claws rather large, but not so disproportionately so as in the adult form, but as much so as in the larva in the second stage of the lobster. Abdomen broad and flat, spatu- late at the end, much as in the adult. There were five pairs of abdomi ual feet or swimmerets, each with endopodite and exopodite, like those of the second larval stage of the lobster." ALPHEU9 SAULCYI FROM THE BAHAMAS. In this form, the metamorphosis of which is fully described in another paper, we have either an abridged development in which the general adult characters, very marked in the first larva, are all acquired in twenty four hours after hatching, or a case where the short metamorphosis is done away with entirely, so that the animal leaves the egg in the full adult form. Comparing the histories just given with the one before us, we find thaf the first larva of Alpheus saulcyi is about equivalent to the third larva of the HeterocJieUs from Beaufort, and rather more advanced than the first larva of this species from Florida. The eggs of the Alphei, with the development unabridged, are invariably small and quite numerous. In the two species, however, with shortened metamorphosis, the ova are fewer and maiiy times larger. Moreover, as would be expected, the degree of abbreviation is correlated with the MEMOIRS OF TOE NATIONAL ACADEMY OF SCIENCES. 379 size and number of the eggs. These and the other facts which we have beeu considering arc given in tabular view below: Species. Habits. Metamorphosis. Nnnilin of eft:*. ^Diameter of egg. Lcll^tli c,l IrlMIllo. Complete - • • "500 Inch. ', Incln'R. il-l do do :w M 1-500 M "A , I ; flo Abridged 150-300 .1, .'. 1-1J do loo-:!50 yV I is .species live, as ;i rule, in tlie interior of the green and brown sponges, a few undersized individuals may be found, by careful .searching, among the loose blocks of porous coral which are scattered over the reef, and it^ frequently happens that these individuals possess intermediate characters hetween the two varieties just described. MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 385 claw. (1) With respect to the first point, we meet with a perfectly graduated series between the. two extremes (Figs. 11, IS, PI. xxn, Figs. 4, PI. xxin). (2) The same is true, of the relative lengths of the atitenual spines, the scale, and peduncle (Figs. 11, 13, 14, PI. xxn). In Fig. 11 there is no evident scale and the spines are nearly equal. In Fig. 14 the spines are markedly unequal, and there is a rudimentary scale. In Fig. I.'! this scale is further developed. (.'?) Great variation is seen in the small chela. The fingers of this claw may each end in two or three prongs, or one in two, the other in three, or the tips of the fingers may be simple or merely notched. The tuft of peculiar seta1 on the dactyle may be reduced or wanting. (4) Various stages between the long and short carpus arc observed, and (5) slight variations not easily described are constantly seen in the relative size, shape, and other characters of the largo chela. These variations are shown in a general way in Table I. (For detailed measurement, v. Table II.) The fifteen cases here recorded were selected from upward of seventy-five, in a large number of which the variable parts were drawn for more careful comparisons. TABLE I. — Showing variations in Alphcus saulcyi and the intermediate stayes between the varieties brevicarpus and longicarpus. No. Sex. Habitat. Length. Aural spine. Sijuanions spine. 1 •2 3 4 5 (i 7 8 9 10 11 12 13 14 15 3 9 3 i 9 9 9 3 i £ i $ i 9 Green sponge do »i »i . . •„'!!. 7 9 9 5 Extends J leug iiient nt antfi To. i 1 . 2d segm do th 2d seg- ii ii l.i, stalk . eat Extends nearly to end of antennu- lar stalk. Do. Do. Do. Extends nearly to end of antennal stalk. Not nearly to end of antenna! stalk. Do. £ 1. antennal stalk. More than £ anteunal stalk. Nearly to end anteuual stalk. sl antenna! stalk. More than j- auteuual stalk. Do. J 1. antennal stalk. l 1. antennal stalk. Brown spoug Green sponge Rocks, Dix 1 do rln 't. reef.. 17 10 4 2d segment. . do -.. ....do Rocks; Hog Id. ivi-f. . do 11.6 10 10 17. r. 13 9. r. 1. of 1st segniei Over i Lot' 1st Nearly to end 1 J 1. 1st segment Nearly to end 1 do t «-gment st segment. . Rocks; Gret reef. IJrowu snoug do n Key ) . st segment- . Rocks; Dix I Brown apoug Reef rocks 't. reef- . 41 do n Nearly to end 1st segment. . No. Inferior basal spine. S no tuft, and the carpus is short. In No. 9, which is of the same sex, the same length, and from tbe same locality as No. 8, the small chela has the characters of the variety lonyicarpus. Nos. 5 to 12, in tbe middle of the table, show iu one way or another intermediate characters between the extremes, Nos. 1 to -1 arid Nos. 12 to 15. VI. — MEASUREMENTS IN MILLIMETERS. TABLE n. [Locality: Nassau, N. P., Bahama Islands.] l ; i rrn sponge. $ Green sponge. Rocks: Uix Pt. Eeef. Rocks : Green Key Reef. Rocks : Green Key Reef. Brown sponge. Rocks : Dix Pt, Reef. gex cf ? I\ I'l. Reef. Rooks: Own Ki'V Eeef. Rooka ireeu Key Keef. Brown sponge. Rocks : Dix Ft. Eeef. Sex 9 11 6 1 5. 1 •i •i i) l.i. r, 2 r> •' G r, 4 3 2.8 r> r> r, 3 2.4 5.3 2. r, 6 3.1 2 2 2 1 0.8 2.5 1 2 Length of carpus of lar^e chcliped, ou upper nio 1 3 1.7 2.2 1.8 0.9 0.6 2.0 2.0 1.5 :; C.. 7 4 2.4 1 8 2 1.8 0.8 4.5 3.4 o 1.4 1 2 1.3 1 Lenwth of saino to articulation of dactyle . Greatest depth of same :i 1 •J.:i G '> 7 3 O.'J •2 r> 2 5 1 1.9 3 0 8 1.5 0. G 2 1 2 7 4 5.5 1.8 0.8 0. G 1.7 1.7 1 2 2.2 r. i Length of second, third, and fourth segments of 2 1 1 1 5 2.8 6 2 4.6 2.7 5 8 6 1.2 1 !> 1.5 6 2 5 8 3.5 1.6 4 3.7 3.7 2.9 4.2 5 3.5 VII. — THE CAUSES AND SIGNIFICANCE OF VARIATION IN ALPHEUS SAULCYI. If we consider Nos. 1 or 2 of Table 1 as representing the nearest approach to the mean of the species, Nos. 5 to 15 must stand for individuals which have fluctuated farthest from the mean. The individuals given in the table were chosen without reference to sex, yet it appears that nearly two-thirds of the number are males. In examining a larger collection of these aberrant forms 1 find a still greater percentage of males. There can be little doubt that in those cases the males are in the excess, although I have not tested this point carefully. It is also evident from observation and detailed measurements that the average size of these abnormal individuals is less than one-half that of the brevicarpus type. The brown sponges teem with a population of under- sized forms, nearly all of which are aberraut, and none of those which were examined exceeded the length of 17.5mm, which is considerably less than the average for the type. How far are these variations individualistic and how far are they confined to the race or species as a whole! In other words, is the individual plastic, departing from the standard of the species and becoming different at different periods of its life, or do individuals deviate from the mean of the species, each along its own line ? Further, are the variations congenital ? While we 388 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. are not prepared to answer these questions as fully as we could wish, yet the facts are sufficient to throw some light upon the subject. We have abundant evidence that there is considerable fluctuation in the life of the individual, as regards the number, color, and shape of pigmentcells for instance. In all larvre of these prawns the external antenna; have a well developed scale, and it is thus clear that this organ may degenerate and apparently disappear, to be reconstructed again at a later period. The variety Jinif/icurpiis (No. 15, Table I) has no " squanie," although it is present in the young (Fig. 7, PI. xxn), and the cases in which the organ is seeu in various stages of development (Figs. 13, 14, PI. xxii) support and illus- trate this conclusion. This, however, is not a rule with the species as a whole, as it is in the some- what analogous case of the loss and subsequent reconstruction of the last two pairs of thoracic legs in the larvre of Steuopus and Sergestes. The question as to how far the characters which distinguish such forms as Nos. 1 and 15, Table I, are congenital can only be answered by a careful study of their development. My attention was not directed to this subject while at the seashore, and in this connection some interesting experi- ments remain to be performed. The evidence we have goes to show that the young in any given case share the peculiarities of the mother, and this is probably true of such details as the, right and left handed condition of the large chelipeds. The following examples illustrate this fact: (1) The adult female in this case has the characters of No. 15, Table I. The autennular or aural spine is nearly three-fourths the length of the first antenuular segment. The aural spine has a correspond- ing length in the larva} of this prawn at the time of hatching. In the adult the fingers of the small chela end in prongs; there is a tuft of peculiar setse on the dactyle. In the first larva the fingers of the small chela also end in prongs, and there is a tuft of rudimentary setn- on the dac- tyle. In the adult the carpus of the small cheliped is relatively very long. In the first larva the carpus of this appendage is about one-third the length of the propodus (relatively a little shorter than in the, adult). Fig. 11, PI. xxn, may be taken to represent the mother (rostrum here wanting), and Fig. 17 the youug. The small chela of the mother is shown in Fig. 2, PI. xxiv, that of the young in Fig. L">, PI. xxti. Another case exactly like this was observed, where the embryo was taken from the abdomen of the female. (2) The adult in this case has the characteristics of No. 2, Table I (var. brericarpus) (PI. IV, Figs. 1, 2). The larva? are shown in PI. xxi, Figs. 1, 2, 3, 8. The aural spine, at first short, is nearly as long as the first anteunular segment when the larva is a week old (Fig. 10, PI. xxii). In both the parent and young the carpus of the small cheliped is relatively short. The lingers of the small chela end in simple tips; there is no tuft on the dactyle (see Fig. 16, PI. xxii). These facts indicate that the young share the peculiarities of the parent, but exactly how far the individual may depart from this standard in its own life, or how strictly the law of inheritance applies in all cases, my observations do not warrant a decisive answer. A few experiments could be easily made upon this Alpheus which would throw light on some inteiesting questions in heredity. The females with ova are easily obtained; the young are readily hatched and kept alive in glass dishes until they have reached the adult state. In this species the change of environment, due to the adoption of life in sponges, has probably aeted as a direct stimulus to variation. These auimals tend to vary most along certain definite lines, as, for instance, the relative lengths of the, antennular segments and aural spine vary much, while those of the segments of the carpus of the second pair of thoracic legs are practically invari- able. Homologous pa^ts vary alike, unless specially differentiated in different ways, as in the chelipeds. There is no diversity of life between males and females, and both sexes vary alike, but aberrant males are probably the more common. The occurrence of large numbers of individuals showing variations of the same kind, but of different degree, render it plausible at least that the same variations may occur in a large number of individuals simultaneously, but the reason why this or that part has varied most is wholly obscure. The aberrant forms (variety longicarpus) which have adapted themselves to life in the brown sponge thrive and produce young which, in the early stages certainly, share in the peculiarities of the parent. The variety brevicarpus is similarly adapted to its environment and its young resemble MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 389 it. Has natural selection, then, acted so far as to differentiate the species in more than one direc- tion ? There are some facts which favor the view that it has done so, but before the question can be definitely settled we must determine more precisely how far intermediate or aberrant forms represent phases of the individual and of the race. It is not probable that we are here dealing with the hybrids between two originally distinct species. PAET SECOND. THE DEVELOPMENT OF ALPHETJS. I. STRUCTURE OF THE I.ARVA. (PI. XLIX, Fig. 174. PI. LIII, Fig. 196. Pis. LIV-LVII.) These studies in the embryology of Alpheus begin with the growth of the ovarian egg and the early phases of segmentation and extend to the larval and adult periods. lu order that the prog- ress of development may be followed in the light of the structure which the embryo finally attains, we will start with a general survey of the anatomy of the first larva of Alphcust xaulcyi. A fuller description of the histology and histogenesis of the tissues will be given in the parts which treat of the different organs in detail. A profile view of the larva as it appears while still inclosed by the eggshell and of one imme- diately after hatching is seen in PI. xxi, Figs. 1 and 5, and the brief and insignificant metamor- phosis which is required to provide it with the adult characters are illustrated and described in a separate paper (Pis. xxi-xxiv). Most noteworthy are the large, stalked, compound eyes, the segmented abdomen provided with its full number of appendages, the short, stumpy antennae, and the swollen chehe or pincers of the first pair of thoracic legs. At this stage this Alpheus is a larva, but in a restricted sense, since many adult characteristics are present. It is a larva, with preparations for immediately assuming the adultstate. Some of the larval peculiarities are the spatnlate telson, the biramous or schizopodal pereiopods (first to fourth pair, inclusive), the rudimentary pleopods, the unabsorbed food yolk, and the uncovered, stalked eyes. The, structural details are now very great, so that it is often impossible to interpret the parts seen in a single section, and it is only by comparing sections made in different planes that the relations of the organs can be successfully made out. In Fig. 196 (PI. LIII) the plane of section is nearly vertical and median throughout, except for the posterior half of the abdomen. The supra-iBSOphageal ganglion, which is usually spoken of as "the brain "(s.o. #.), is a complex organ, composed of internal, medullary masses (punktsubstanz balls), and cellular tissue which completely invests them. It is made up of the fused ganglia of at least two segments, those of the first and second antenme. This fusion is complete from the early stagi-s of development, and the relations of the parls are now extremely complex. They are best illus- trated by a comparison of the series of transverse sections (Pis. LIV, LV, Figs. 211-219) with those made in a horizontal plane (PI. LVII, Figs. 238-243), and itVill be seen that there are four pairs of fibrous masses in the brain, intimately connected together. These compact and finely granular masses in the interior of the ganglia of invertebrates were, described by Leydig twenty-five years ago under the name of Punktsubstanz ^uid later by Dietl (1876) as Marksubstanz. As Krieger remarks, the latter name is bad, since it confuses this tissue with the spinal marrow of vertebrates, with which it has nothing to do. It is essentially a./Hf of very fine fibers. We will therefore speak of it as the Pwnktsubstanz, or, to use a more descriptive term, the fibrous substance of the ganglia. The first pair of these, the anterior or optic fibrous masses (PI. LV, Figs. 212-215), are the largest. They are completely fused on the middle line and form a single compact mass, which is slightly constricted laterally (PI. LVII, Fig. 242, of.) and which is divided in front (PI. LIV, Figs. 210, 211), where it gives off two diverging stems of fibrous tissue (sometimes called optionerres) to the optic ganglia in the stalks of the compound eyes (see also PL LVII, Fig. 240 of.). 390 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. Next in point of size are a pair of large lateral balls, which appear kidney- shaped in transverse section (PI. LV, Fig. 216, /./.). Each is virtually segmented at the lower surface into two lobes (PL LVII, Fig. 242, If.). These lobes are closely united to each other, and by a pedicel or stalk of fibers to the lower posterior extremity of the anterior, optic mass. A third pair of fibrous masses (PL LV, Figs. 215, 216, af.) fuse with the anterior mass at the same point. Each of these balls is also bilobed, and from them issue the fibers of the auteuuular nerves. (PL LVII, Fig. 243, n. ««., also PL LV, Figs. 212-214, a. o., nau.) The nerve of the first pair of antenna? consists of cells and fibers, which pass to a mass of deeply staining cells (a. o.), the ear, and to the tissues of the antenuular stalk. The fourth pair of fibrous masses (PL LV, Figs. 217, 218, c.) borne, on a median papilla, which projects downward between the eye stalks. The details of the structure of the eyes are given in Section ix. The alimentary tract of the larva is a somewhat complicated structure, and the relations of its parts are best understood by reference to sections taken in more than one plane. We can recognize five well-defined portions: the (esophagus, the masticatory stomach, the midgut, the hiudgut or intestine, and the appendages of the midgut. These are shown in a semidiagramrnatic way in the cut (Fig. 2), and the longitudinal section (PL LIII, Fig. 196) and series of transverse and horizontal sections (Pis. LV-LVII) illustrate the structures in more detail. It is interesting at this point to compare the larva shown in Fig. 196 with the longitudinal section of an advanced embryo (PL XLVIII, Fig. 16$). In both we recognize the foregut, a tube bent on itself, consisting of the (esophagus and masticatory stomach (m. ft.). In the embryo the latter is closed on the side of the food yolk. In both we also see a vertically directed fold of endoderm (/., overlying mi/3 in Fig. 196) and behind this the large lumen of the hindgut, which gradually tapers into that of the narrow, tubular^ intestine. Between this fold on the one hand ard the stomach on the other we find in the embryo an enormous space filled with yolk, which is partially walled iu MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 391 with endodermal epithelium near the point where it communicates with the cavity of the gut. This in tlie larva corresponds to the midgut (Fig. 190, mg.) and its diverticula. The oesophagus (Figs. 196, 218-220) is a straight, vertical tube, with very thick walls, which are throwu into longitudinal folds. There is an anterior and posterior fold and two lateral ones, which give to the lumen of the oesophagus the shape of the hitter X when seen in transverse section (PI. LVII, Figs. 241, 242). The walls of the masticatory stomach resemble those of the oesophagus, and the folds of the latter are continuous with the valvular structures of this region. The lateral and median thickenings (PI. LV, Fig. 221, p. v.) at the point where this portion of the stomach passes into the midgiit may be regarded as a rudimentary pyloric valve. The pouches formed between the median ventral fold (Fig. 221) and the lateral folds (p. v.) correspond to the gastrolith sacs in the crayfish embryo (54), but no gastroliths are found in Alplieus. The midgut appears in the longitudinal section (Fig. IOC, mg.) as a short, lestricted cavity. It is, however, a spacious chamber, as we see by examining a series of sections made in other planes (Pis. LV-LVII). It consists of seven parts or divisions: a dorsal, unpaired, median division (mg. in all the figures), and, opening from this, a pair of anterior lobes (mg.*), a pair of posterior (mg.'J) and a pair of ventral lobes (mg.2). All these parts are lined with a peculiar columnar epithelium, composed of endoderm cells, derived primarily from the wandering cells, excepting a part of the median and the anterior divisions, where the endodermal wall is absent or only imperfectly formed. The epithelium of the midgut passes imperceptibly into that of the intestine, since the cavity of the hiudgut is in communication with the food yolk from the very early stages of the embryo, and since also the eudoderin is formed very gradually and first appears in the region where the hind gut communicates with the yolk. On the other hand, the demarcation between the wall of the masticatory stomach (of ectodermal origin) and that of the midgut (Fig. 19G) is most pronounced. Correlated with this distinction is the fact that the foregut is a blind sac and completely cut off from communication with the yolk until very late in embryonic life (PI. XLVIII, Fig. 168). The anterior lobes contain the remnant of unabsorbed yolk (Figs. 218, 237, ?/.), and in cases where the lining epithelium is unformed, the food yolk is in contact with the brain. These lobes are sepa- rated by a median vertical partition (mp.), composed of connective tissue and muscle cells, which suspend this portion of the digestive tract to the anterior dorsal wall of the body. In a very late embryo which is about ready to hatch we find that the partition separating the anterior lobes is incomplete. The dorsal half of it consists of a downward-growing fold of endoderm cells, with a mesodermic core. The ventral and lateral walls of these diverticula are devoid of epithelium, so that the eudoderm extends itself most rapidly forward, on the dorsal median line, and thence spreads to the ventral floor. The posterior lobes (mg.*) are the first to develop (see P. 1 LI, Fig. 185, mg."). They lie to one side of and below the hindgut (PI. LVI, Figs. 226-230, mg\, #<;.' J). Up to this stage their position is never dorsal to other parts of the digestive tract. It is from these lobes that the gastric gland or so-called " liver" arises. Each lobe is simple until a short time before the embryo hatches, but in the newly born larva it is divided into three lobules. This division is effected in this manner: The lower median part of the primary lobe (Fig. 228, , mg.:!), is imperfectly divided into three lobules, as in the early larva. They extend backward, below and to one side of the gut. The two remaining 392 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. pairs pass forward on either side of the masticatory stomach to a point about on a level with the first maxillary segment. The ventral is the larger and longer of these, and two lobules are con- stricted oft' from it near its extremity. They correspond to the ventral lobes of the midgut (»ig.2 rut, Fig. 2). The dorsal pair represent the anterior lobes ( w/.1), which are now entirely withdrawn from the head region, and naturally contain no food yolk. The gastric caeca are all tilled with a coagulable fluid which stains feebly in carmine. The gastric epithelium for a short distance behind the point where glands communicate with the stomach has marked histological peculiarities. The internal absorbent surface is increased by folds which nearly obscure the lumen of the tube. The cells are columnar and resemble the glandular cells of the liver and probably have the same origin as the latter. In the masticatory stomach there is a strainer of hairs developed on the ventral and lateral walls which are greatly thickened, as we saw in the larva. The dorsal wall is thin, but there is a large valvular fold on the ventral side. The vascular system of the adult is already outlined in the larva in all its essential character- istics. The walls of the blood vessels are exceedingly delicate, so that it is not easy to ascertain their distribution by means of sections alone. The heart (PI. LLII, Fig, 190, H.) is a short tubular chamber, flattened between the dorsal body wall aud the enlarged section of the hind gut. It is suspended in the pericardial sinus (p. a.) to the body wall and surrounding organs by means of strands of connective tissue (al:e cordis). The walls of the heart are, quite, thin, aud its cavity is partially divided into three compartments by the growth downward from its roof of two sheets of mesoderm cells (PI. LVI, Fig. 231, aud PL LI, Fig. 18(1). Of the several arteries which lead from the heart, three, and possibly five, can be distinguished. Posteriorly the heart is continuous with the large xii/ininr ulxlominnl artery, which traverses the abdomen close to the dorsal wall of the intestine (Figs. 1!H!, 232, 235, «.. *. ft.). Near its origin from the heart, the sternal artery (Fig. lid! shows a trace of this vessel between ganglia 12 and 13, to the left of pi:) is given off, and passes directly downward to the ventral nervous system, which it penetrates at a point between the third and fourth thoracic ganglia. This is continued backward under the nervous system and forms the infn-inr nlnlominul artery (Figs. 229-234, a. i. a.) Anteri- orly the heart gives off the unpaired ophthalmic artery (Figs. 19G, 215-229, <>p, a. op.), which runs for- ward to the region of the eyes aud brain. It is not an ophthalmic- artery, strictly speaking, but from the tirst, supplies arterial blood to the brain and anterior cephalic region generally. In Figs. 215, 210, it is seen cut in partial longitudinal section, where it evidently communicates with the blood space surrounding this part of the brain. The antenna! arteries can not be clearly distinguished in sections, but in a much earlier stage trains of cells are seen at the surface of the egg passing forward on either side of the middle line toward the eye stalks, which possibly represent the anteu- nal vessels. Besides the sinuses already mentioned, there is a large sternal sinus (Fig. 190, xls. K.}. This occupies the extensive space between the thoracic ganglia and the alimentary tract and " liver/' and, like all other similar spaces, is more or less completely tilled with serum and blood corpuscles. Five pairs of gills are present at this stage. They are developed from simple pouches or folds of the skin on the bases of the thoracic appendages (Figs. 193, 230-233, br 2 5). The outer sur- face of this primary fold soon becomes divided into a number of secondary folds or gill plates, and in a larva which has moulted twice aud is twenty-four hours old, the branchia has the structure shown in Fig. 195. The adult gill is precisely similar to this, except that it has a greater number of plates aud more definite branchial vessels. In the early larval stages the skin aud especially the branchiostegites (Fig. 193, bg.) probably serve as important respiratory organs. In respect to its muscular system the first larva appears to differ but little from the adult. The flexor and extensor muscles of the abdomen are most prominent (Fig. 190, inu.f., mu. e.). The former consists of a double rope of ti bers, fuse 1 completely together and very much twisted. They extend from the sides of the thorax to the terminal telson (Fig. 227-235. »»«../.). The extensor muscles (MIM. e.) are smaller, but otherwise similar to the latter, both in origin and extent. They lie above or to the sides of the digestive tract. Their attachment to the carapace is showu in Figs. 227, 228. The next most prominent muscles are the adductors of the mandibles and great chehe. The former consists of a large baud of fibers which pass from oue side of the body to the other directly MEMOIRS OF TtlE NATIONAL AC A UK. MY OF SCIKNCES. 393 over the nervous system (Fig. 2lil, ad. in.]. Closely associated with it are the inn.M-lex of the maxilla^. Tlie, large flat tendon to which the adductor muscle of the forceps is attached, is well developed at the time of hatching. It is Conned by the infolding of a sheet of ectoderm cells a I I lie point of articulation of the tinkers ol'the claws, and in a. plane at right angles to their plane of act ion. The outer ends of the. cells of this infolded sheet now oppose each other and secrete the chitinons tendon, while to their morphologically iinirr ends the muscle libers are attached. The connective tissues invest the organs and seem to bind them together and to suspend them to the outer ectodermal wall of the body, but in some cases the ectoderm of the surface is apparently replaced by mesoderm cells, and often muscle libers appear to be attached to the tergum of the somite (Fig. 190). This may be explained by the intimate fusion of the ectoblast and mesoblast at these points. 'The green gland (PI. LIII, Fig. 198, ay.) at the base of the second antenna is a well delined structure. It consists of a blind tube, which passes up close to the brain as far as the anterior sacs of the midgut, and of a solid, disc, shaped body. The walls of the tube are composed of a single layer of large cubical cells. These thin out at the lower end, and to the outer wail is applied the solid nodular body. Neither at this stage nor at any previous one have I been able to detect an opening to the exterior. In the adult the tubular portion of the gland grows to very great length, coiling itself in all the available space in the anterior region of the body in front of the mandibles. It surrounds the brain and oesophagus and passes down to the labriun and into the eyestalks. The solid almond shaped body (probably the end-sac) becomes a spongy mass of tissue. Its function is plainly different from that of the epithelium which forms the wall of the tube and to whicli the secretive product ot the gland is due. The reproductive organs, or what I regard as such, are difficult to tind, owing to their very rudimentary condition. They consist of a small cluster of large cells on either side of the middle line between the digestive tract and the anterior end of the heart (Stage x, Fig. 173, 1\. O.). With this sketch of the structure of the larva we are ready to trace the history of develop- ment from the earliest stages and to ascertain the manifold changes through which the unicellular egg with its great store of yolk passes, before it attains to the wonderful complexity of the larval and adult forms. (a) Alpheus. — The ovaries of Alpheus are paired cylindrical bodies which extend 'between the alimentary tract and dorsal blood vessels, from just behind the eyes to the end of the third or fourth abdominal somite. Owing to the transparency of the skin in this species (A. xanlci/i) the.x are extremely conspicuous, giving to the female an intense green or yellow hue, according to the color of the egg (PI. iv). The oviducts open in the usual way by means of a slit-like'valve on the basal joint of the third pereiopods. In PI. xxvi, Fig. 11, the condition of the adult ovary is shown, as it appears two or three days after the eggs then carried on the abdominal appendages had been laid. The ovarian ova are ripe by the time the young are ready to leave the shell, and the new ova are laid in a few hours after the hatching of the larval brood. Thus there is a constant succession of young, and females are not commonly found without either attached or large ovarian eggs. The. breeding season of this species extends, as we have seen, throughout the entire year. The structure of the ovary is quite simple (Fig. 11). It is essentially a sac lined with ger- minal epithelium. The external layer of the sac (O. W.) is muscular and contains numerous nuclei. Between the epithelium and fibrous coat there is a wide space tilled with blood. This may be unnaturally large in the preparation owing to the disturbing effects of the reagents employed, but it is not wholly abnormal. The germinal epithelium consists, for the most part, of a single layer of large cubical cells. The nuclei are large and granular, and the cell outlines are often distinct. The function of these epithelial cells is twofold: (1) They give rise to ova; (2)"They form the epithelium of the egg follicle. There is no germogeu or poly nuclear mass of protoplasm from whic.li the ova arc developed, but the eggs appear to originate directly from epithelial cells. The new eggs begin to develop, 394 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. • while the ovarian lobe is yet crowded with ripe ova ready to be laid, on the ovarian wall next the middle line of the body. The process seems to be as follows : The nuclei of the cells of the ger- minal epithelium increase in size along a certain tract. The cells grow rapidly and are slowly dehisced or pressed into the cavity of the sac. Each is surrounded by a coat of follicle cells. This is formed by the ingrowth of the germinal epithelium about the egg. Sometimes .several ova occupy a common pouch (Ger.) which is separated from the rest of the ovary by sheets of follicnlar tissue (F. E.), but eventually each egg has a covering of its own. Between very young ova (e) no larger than the epithelial cell, aud the maturer egg (el) every stage can be traced. The yolk appears very early as a fine granular deposit in the protoplasm of the cell. In this species the development is nearly direct, there being no zoeal stage, and the egg contains more than nine times as much yolk as the egg of Alphcun minus, in which the first larva is a zoea-liko form. The materials for the yolk must be derived directly from the blood, and in this form the germinal epithelium is bathed with the blood current. Where there is an enormous food yolk blood must be supplied to the developing ova in more than the usual quantity. This is often accomplished by reentrant blood sinuses which penetrate all parts of the ovarian stroma, as in the lobster (Homariia) aud in the cephalopods, which are .precocious in development and conse- quently deposit a great store of yolk in the egg. In the latter the follicular epithelium is folded in a remarkable manner about the egg to increase its nutritive surface. The germinal vesicle (Fig. 11, G. V.) is filled with coarse chromatin grains, and in the early phases grows relatively faster than the rest of the egg. In the egg, (e,1 to the left) which is ^s inch in diameter, the diameter of the germinal vesicle is one-half that of the entire egg. The chromatin grains increase in size until there are formed, as in an egg like the last, six or more large masses of chromatin, or uucleoli. The older eggs are spherical ; their food yolk is often vacuolated, as in later stages, and they are invested by a single membrane, the chorion, which is a chitenous secre- tion of the follicular cells. In the ripe ovary of this Alpheus the mature eggs fill the ovarian sac, except at the lower portion next the middle line, where, as already stated, the young ova first make their appearance. These mature eggs are closely crowded aud irregular in shape, and their bulk greatly distends the body of the prawn. The choriou is now fully formed and closely invests the vitellus. The yolk is in the form of spherules, usually fused aud always vacuolated in preparations which have been subjected to alcohol aud turpentine. In the ripe egg the nucleus was not seen, but it is quite probable that careful sectioning would show that it lay at the surface, as is th» case with the ripe ovarian egg ef the lobster, which is often left in the ovary, after the bulk of the eggs are laid. We thus conclude that the extrusion of polar cells may be internal, that is, may take place within the ovary, as is sometimes, if not always, the case with Homurux. (b) The*Lobster (Homarus americanus). — The ovaries of the lobster consist of two lobes or rods of tissue, united by a short transverse bar. When filled with eggs their color is a dark olive green, except in young females, where the color of the immature ovary is variable. Each lobe is composed of an outer wall, which is a felt of muscle and connective tissue fibers with very small nuclei, and of a loose framework of germinal epithelium, which penetrates all parts of the lobe. The latter is a syncytium and consists of a matrix in which great numbers of small nuclei are embedded. These nuclei, with surrounding protoplasm, give rise (1) to ova and (2) to cells of the egg follicle. The growth of the ovarian egg from the epithelial nucleus is illustrated in PI. xxv, Figs. ,'5, 0. Fig. 6 is from a section through the posterior end of an ovarian lobe of a lobster obtained from the Baltimore markets in January. Fig. 3 shows the central portion of this section greatly enlarged The diameter of the entire section is about twice that of the part represented in Fig. 0, and the oldest eggs lie at the periphery. The germogens, the centers of dispersion of new eggs, lie nearly in the long axis of the lobe. We can therefore trace in a single good section at this stage the de- velopment of the egg through every stage, from the indifferent nuclei of the ovarian stroma to the large peripheral ova. The ovary is supplied with blood by means of sinuses which penetrate to all its parts (IJ1. S.). The sinuses are definite reentrant channels with thin membranous walls. The ovarian tissue (Ct. S.) consists of a fibrous matrix in which numerous oval granular nuclei are embedded. The process of the conversion of the epithelial cells into eggs is shown in Fig. 3. The epithelial nucleus (O, Oa) swells out, becomes spherical, and its chromatiu has the charac- MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 395 teristic granular appearance of the germinal vesicle of the young egg. The iirst trace of (he yolk (O:!, O4, O5) appears in the outer granular layer which surrounds the germinal vesicle. This layer represents primarily the cell protoplasm, in which the yolk is formed. The cell takes on a definite shape and is very early invested with a follicular coat (F. C.). In an egg a little older (O7) the nucleolus has appeared, and in still older eggs (Fig. 6, O, O1) a delicate choriou (Ch.) can be seen. This is secreted by the cells of the follicular envelope (F. C.). The growing eggs pass out from the central to the peripheral parts of the lobe in the sheets of stroma between the blood sinuses. Distinct yolk spherules are very early seen (O7) and are of uniform size, but in maturer eggs (Fig. 6, O, O1) the germinal vesicle is sometimes surrounded by a central layer of small spherules and a peripheral layer of larger ones. The germinal vesicle is centrally situated and always contains a single excentric uucleolns, besides stellate masses in the chromatin reticnlum.* (c) The Spiny Lobster (Palinurus). — In the spiny or rock lobster from the Bahamas the ova originate exactly as in Elomarus, and the structure of the ovary is essentially the same. There are several uucleoli, as in Alpheus. The ovary is not nearly so richly supplied with blood sinnses as in the cases just considered. This is perhaps correlated with the fact that the amount of yolk * Since the above accouut was written I have been able to study the structure of the ovary more thoroughly, and the subjoined notes are largely extracted from a preliminary notice on "The Reproductive Organs and Early Stages of Development of the American Lobster." (23.) The structure of the mature ovary is somewhat peculiar. The free, nnextruded eggs till the lumen of tin- ovarian lobes. The lobe or tube itself consists of the proper ovarian tissue and the outer muscular wall, which is very thick. The stroma is characterized by the presence of gland-like structures, blood sinuses, and immature ova. The glands are in close relation with the growing eggs. They are plaited or folded structures, and consist of a single layer of columuar cells, the boundaries of which are iudistiuct. The lumen of the fold usually contains a granular residue, but often yolk and degenerating nuclei. It seems possible that these structures are comparable to yullc (/lands, and that their function is to supply the growing ova at this stage with a part of their massive fond yolk. Three days after the extrusion of the eggs tho glandular c;eea have much thicker walls; the rapidly dividing cells are smaller, and their nuclei lie at various levels. In another ovary of about the same age the glands are relatively very large. Tho columnar cells are greatly elongated, their nuclei lie at the deeper or outer ends of the cells, and the Ininen of the gland is often completely obscured. The gland forms a kind of egg tube, abutting upon and partly inclosing the growing egg. The columnar cells stop short at the sides of the egg, so that the glandular cjeeum resem- bles a narrow bag with an egg pushed into its mouth. The glandular cells are directly continuous with those of the follicle. The axial portion of this ovarian lobe is composed of hollow spaces, blood sinuses, and loose stroma, in which very young eggs occur. Degenerating cells occur not only in the strouia, but probably in the developing ova also. In Peripatus Norir Zntlaniliw the yolk is described by Lilian Sheldon as arising not only from tho egg proto- plasm, but also from the follicle cells (.r>7). When ten to fifteen days have elapsed after egg-laying (eggs in egg-nauplius stage), the gland-like bodies have almost wholly disappeared. The wa,lls of the caeca are shrunken and crumpled, and the latter have been crowded to the extreme periphery of the ovary. The ovary now contains a solid core of immature eggs, stroma, and bloodvessels. This is continuous with radial sheets of similar tissue which extend from the center toward the pe- riphery. The outer and more mature masses of ova are thus divided into more or less continuous, longitudinal bands. At a still later period (eggs with eye pigment, four to five weeks' old ) the glands are present merely as shriveled remnants. Later still (lobster taken August 21 ; egg embryos in a late stage) there is no trace of gland-like structures. In the ovary of a lobster (taken June 30), with eggs about to hatch, the condition is similar to the last. It is now about eleven months since the eggs were laid, yet the diameter of the largest ovarian ova is only about our- half that of the mature eggs. The ovarian wall is thinner than in previous stages, and in the axis of the. lobe there are still sheets of very small, immature ova. It seems that the bodies which have been described as probable yolk glands are present in the peripheral parts of tho ovaries only during the limited period of from two to three weeks after the eggs are laid, and when I In- orgius are recovering from tho changes which follow this event. Their structure is quite unlike that of bloodvessels or sinuses with which they are intricately associated, and their relation to the growing eggs seems to imply thai they have some function to perform in tho nourishment of the peripheral ova. Their short existence, on the other hand, might lead us to suspect that they were more or less rudimentary structures, or that they were concerned with the secretion of the gluey substance with which the eggs are coated at the time they are laid. Their true function, however, remains to be determined. Ovaries which I have examined, taken in summer (July) from lobsters ( " paper shells " ) which have reeenl ]y moulted and which do not carry eggs, present very thin walls, and the largest ovum measures in diameter about one- half that of the mature egg. These lobsters have probably hatched moulted. (Compare the ovary of the lobster taken June 30 above.) Some allowance is to be made for individual difference, but the slow growth of the ovarian egg, which we, have traced from the summer when eggs were laid to the following summer when these rg^s wen' hatched, is very noteworthy, and shows conclusively that the lobster is not an anuual breeder. 396 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. in eacb egg is very small, although the number of eggs produced by this auimal is enormous. At Nassau, Palinurus begins to spawn in June. (d) Cknnpiii-inon. — Ishikawa describes very fully the ovaries and ovigenesis of the prawn At- i/t'/il///i-n miiqircxm and concludes that the ovum "originates from the inner lining of the ovary and is at the beginning a cell with a nucleus and one, two, or rarely three nucleoli.' The mature egg, according to this observer, has two membranes, one of which is due to the " hardening of the peripheral protoplasm of the egg," while the other (secondary egg-membrane) is secreted by the epithelial cells of the oviduct and added at the time the eggs are laid. There seems to be au error here in regard to the origin of the chorion. In the Decapod Crus- tacea it is the rule that the choriou it secreted gradually during the growth of the egg by the cells of the egg follicle. The large glandular cells found in the oviducts of Atyephyra possibly secrete the viscid fluid by which the eggs are attached to the swimmerets, yet this point needs cornfir- ination. The chorion was found in the ovarian egg of Pagurus by Mayer (.'»9), who says: Dan EiiTstoc.ksei von Pa^'mis jHt in dor crsten Zeit seines Bestekens eine eehte Zelle mit Protoplasina, Kern und Kern-Korperchen. Spiiter findeteine Einlagerung von Deutoplasuia und die Bilduug einer Hiille aus Cbitiu statt. Eudlieli \vird iler Kern unsiulitbar ; das Ei wtcllt danu eine Cytode vor. Das t'erlige Ei verliisst den Leib des Krebses ohne Kern und mit einer Hiille versehen. This description answers for Alpheus in all essential points. The ripe egg of the crayfish (Astacus flumatilis) is inclosed by ;» single envelope, the chorion. According to Ludwig and Waldeyer it is not known whether this is a product of the egg or of the follicle cells. Huxley (2G) merely states that '" a structureless vitellme membrane is formed between the vitellus and the cells which line the ovisac." The ovisacs burst and the ova pass through the ovary into the oviduct. When laid, the eggs "are invested by a viscous, transparent substance which attaches them to the swimmerets of the female and then sets." Here as in other forms the choriou is clearly the secretion product of the ovisac. In C run ;/<> it nili/aris Kingslcy (ol) finds that the late ovarian ova resemble the newly laid eggs. There is a thin structureless envelope (chorion), but no trace of au inner vitelline membrane. Lud wig's general statement that the egg cells of ail Arthropods are surrounded by a vitelline membrane (Dotterhaut), the product of the egg itself, is certainly erroneous. He divides the egg membranes into primary eyg membranes, those which are derived from the protoplasm of the egg itself or from its follicle cells, and srrt>niltiri/ <•//// mcmbrn-nes, those formed by the wall of the oviduct or otherwise. Balfour, following Van Beueden, restricts the term vitelline membrane to structures derived from the protoplasm of the ovum, and chorion to those formed by the cells of the follicle or oviduct. In the category of secondary structures would fall also those secreted by special glands, found, according to Ludwig, in Trombidium, Chilopoda, and nearly all Crustacea, and the winter eggs of Daphnia and Tardigi ada, which is due to a moult or direct separation of epithelium from the body of the mother. In speaking of the vitelline membrane Van Beneden and Bessels, in their monograph on the formation of the blastoderm (GO), thus define it: Nous entendous la membrane vitelline dans le sens oil M. (.'lapan-de fa si nettement ddnnie dans son travail sur les vrrs Ne'matodes : C'est la couclie exterue ilu protoplasma de 1'u'iif, qni, ayant acqnis uno deusit.6 plus grande qne la masse sons-jaconte, se se'pare de cellc-ci par un contour net et tranche'. Elle est iX IVeufce que la membrane cellu- laire ost ;\ la cellule ; elle se forme de la meme maniere. According to this view the ovum is morphologically a cell, the vitelline membrane is the cell wall. The origin and growth of the egg in Arnphipods (Gammarus locusta) agrees quite closely with what takes place in Alpheus and Ilomarus. According to Van Beneden and Bessels (GO) the young ova are at first protoplasmic cells, the nucleus of which becomes the germinal vesicle. The ovarian egg is a cell without a membrane, and in the cell protoplasm refringent vesicles are devel- oped which form the yolk elements. According to these authors the mature ovarian egg consists of a viscous, finely granular, and contractile liquid, which represents the primitive cell protoplasm and holds in suspension the germinal vesicle, and, secondly, of nutritive yolk elements (called by .them ilcitlopldxm because of secondary origin), which are also suspended in the protoplasm of the egg. MEMOIRS OF THE NATIONAL ACADEMY OK SCI ENCKS. 397 In insects it appears that a clioriou is always present in ovarian eggs, while, on the other hand, arachnids possess a vitelline membrane and the eggshell is secreted in tlie oviduct. .There is no simple rule to express the appearance of egg membranes in a diversified group like the Arthropods, and, considering that these structures are purely secondary to the cell and expensive products however formed, this is what we should expect. Their function is chiefly pro- tective, and where a choriou is present in the ovary a yolk membrane is not developed, but the latter is present, as in spiders, when the shell is a later product. Ercll (15) describes three egg- membranes for the lobster, but it is clear, as Mayer lias already shown, that the inner, delicate membrane which has been described for the decapod egg, is a secretion product of the blastoderm. III. SEGMENTATION OF THE EGG OF ALPHEUS MINUS. In the small green Alpheus of the southern coast we have a peculiar modification of the usual process of segmentation, which seems to be anomalous. The fertile egg- is pervaded wi'th a remarkably fine reticulum, which incloses \<>lk spherules of minute and uniform size. The nucleus is central, or nearly so,* and consists of an ill-defined mass of protoplasm, in which a fluechromatiu network is suspended. In the next phase (I'l. xxvr, Fig. 14) the nucleus is elongated and about to divide. Division appears to bo direct and irregular. At a somewhat later stage the phenomena of the most interest occur (Figs. 12, 13). Each product of the first nucleus has developed a swarm of nuclear bodies (S. S.), which seem to arise by fragmen- tation. These bodies take the form of spherical nuclei in clear masses of protoplasm. The yolk frequently has a tendency to segment about the nuclear masses, in the same way that it divides about a single nucleus to form a yolk pyramid. This yolk segmentation seem.s to be normal, but it is very irregular. lu one case there were two large segments, whicb nearly divided the. egg in two, besides several smaller ones. Nuclear matter consists either of small particles or of indefi- nite reticulated masses, resembling the first nucleus (Fig. 14). Clear areas are sometimes found with nuclei which appear to be breaking down. About eight nuclear swarms or clusters are present in the stage shown in Figs. 12, 13. The nuclei vary in size from refriugent particles to bodies of ordinary nuclear appearance. Figs. 25 and 20 represent two sections of one of the clear areas in the same egg from which Fig. 12 was drawn. This clear field has several degenerating nuclei near its border. The largest one (S. C.) is included in both sections. A small chromatiu mass with indistinct body lies next it (S. C.2), and other similar bodies occur in different sections. The cell S. C. contains two chro- matiu balls, and in Fig. 26 (the next section but one in the series) this body appears to be dis- charging through its broken-down wall numerous minute elements (S.) into the clear field. In Fig. 22 a small protoplasmic area occurs, in which a single nucleus lies. This body is granular and contains a large chromatiu ball. Figs. 5 and 23 are also from the same egg. Here we see structures similar to the cell just mentioned. They are surrounded by yolk and consist of a deli- cate reticulum in which usually one large uucleolus is suspended, besides great numbers of small chromatin particles. Various stages of growth are here represented, and it might appear at first sight that we have a case of endogenous cell formation. I see no reason to suppose that the eggs examined are abnor- mal, and I conclude that we rather have in this species a remarkable modification of the usual indirect cell division, attended by an equally remarkable degeneration of nuclear material. In the last stage obtained (Fig. 29) the whole egg is filled with several hundred very large elements, which are descended more or less directly from some of the nuclear bodies just consid- ered, but the intermediate stages have not been traced. This probably corresponds to stage vi ol A. sanlcyi, at the period just before invagiuatiou, but it is quite unlike anything which I have seen in other species. The yolk is now irregularly segmented into blocks or balls, but probably not with reference to these cells. This case is interesting when we compare it with the degeneration of cells to be described in another section, and from a cytological point of view it deserves careful study. * In single sections the nucleus is strictly central, but whether it is NO with respect to the entire, e^y it is imi easy to determine. Minot states that the egg nucleus is always eccentric. — Am. Saluni/ixt, Vol. xxm, 1889. 398 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. IV. — THE DEVELOPMENT OF THE EMBRYO. STAGE I. — SEGMENTATION TO FORMATION OF THE BLASTODERM. These observations relate to Alplieus JieterocJiclis of tbe Southern States, to a Bahauian form which hatches as a zoea but which otherwise resembles this species very closely, aud to Alpheus Kii/i/i-i/ij also the from Bahamas, which lias large eggs and a nearly direct development. Except where it is necessary to meutiou specific differences, these three species will be treated as one form. In June two Alphei (A. saulcyi) laid eggs in an aquarium, but the ova were in each case unfertilized, aud for the most part failed to adhere to the swimmerets. One of these eggs, hardened at an interval of five hours after it was laid, is shown in section on PI. xxvn, Fig. 17. I regard the nucleus of this egg as the female prouuclens. It consists of clear protoplasm, which .stains feebly aud sends out processes on all sides into the yolk, and of an indefinite chromatin network suspended in it. The massive yolk is composed of corpuscles of uniform size, excepting at the periphery where they are much smaller. Numerous small lacuna} occur, representing parts of the yolk which were soluble in the reagents employed. The chorion, or inner egg-membrane, is transparent, tough, and very distensible. It frequently splits into thin layers when subjected to the hardening aud embedding process, thus- showing the manner in which it is formed in the egg-follicle. It thus appears that the unfertilized egg of Alpheus is incapable of segmentation. The n'rst segmentation nucleus has been observed iu a few cases. That shown in Fig. 10 is possibly preparing for division. It possesses a flue reticulutu ; it is lenticular in shape, and granular in appearance, and is surrounded by protoplasm which spreads into the yolk. The once divided nucleus and the next phase with four cells were not obtained in Alpheus, but the hitter was seen iu an allied prawn (Pontonia domestica), and is shown in Fig. 27. One of the three cells present is in the aster stage of karyokiuesis and has a well-marked equatorial plate. The third segmenta- tion phase is illustrated in Figs. 9, 28, and 30. In the section through the entire egg, three of tin- eight cells present are met with, and one of these (x) is shown with greater detail in Fig. 30. A cell in process of division is represented in Fig. 28. In another egg with eight cells present, two are undergoing division in different plaues. As before, the cells consist of a chromatiu network of various shapes surrounded by a clear protoplasmic body, which sends out processes between the surrounding yolk spheres. 'It is important to notice that the products of the segmentation of the first nucleus pass gradually and uniformly to the surface of the egg. At this stage they have not reached the surface but are visible through the egg shell. The yolk in these specimens consists of spherules or angular blocks (Fig. 28, Y. S.), which are largest in the center of the egg, aud con- tain very few vacuoles. The fourth phase of segmentation is attended by the cleavage of the superficial parts of the yolk (Fig. 10) around the nuclei, thus giving rise to sixteen blastomeres or partial yolk pyramids. The division of the yolk proceeds but a short distance below the level of the nuclei, so that all the yolk-pyramids open by their inner ends into the common yolk mass which fills the segmentation cavity of the egg. The base of the pyramid, which abuts on the surface is polygonal in shape, aud at its middle point some distance below the surface, the nucleus is seen with its investments of protoplasm. The nucleus is large and granular, and the protoplasm which surrounds it reaches out on all sides into the yolk. We may look upon the yolk pyramid as a cell in the strict morpho- logical sense, its protoplasm being concentrated about the nucleus. The blastoderm or primitive egg envelope arises through the multiplication and consequent reduction in size of these huge yolk elements. The surface has then the usual appearance of a fine mosaic of hexagonal plates or cells, the nuclei aud surrounding protoplasm of which lie at the surface. The fifth segmentation phase is shown in Figs. 15 and 31. The septum between the pyramids extends farther iuto the yolk, while the nuclei are slightly nearer the surface, and the long axis of each is parallel with it. This particular egg was taken in a" period of "rest," but in others the nuclei are in karyokiuesis, the division being always radial or in a plane at right angles to a surface tangent. The segmenting egg of Hippo, talpoides is shown in Fig. 1 (32-cell stage;, and a section of a later phase in Fig. 4. The egg appears to have undergone a total segmentation when seen from the surface, but this is not quite so marked as represented in the sketch. The yolk pyramids MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 399 (Fig. 4, T. P.) agree with those in Alpheus ami are probably formed in a similar way. In Puln-- monctes vulgaris the history of segmentation appears to be essentially the same. The nucleus and base of one of the yolk pyramids of this form is shown in Fig. 24. Here we see that the peri- mielear protoplasm has a rayed appearance, being produced in all directions into very delicate threads which ramify among the yolk spherules. Some of these threads moreover unite with a thin septum (Sep.) which forms the boundary wall between two adjacent yolk pyramids. Segmentation proceeds with a regular rhythm up to the fifth stage, but beyond this it soon becomes irregular. A blastula is thus formed, consisting of a single layer of cells or blastoderm, and the inclosed central yolk. All the nuclei reach the surface and take part in forming the blastoderm, so that all the protoplasm of the egg which is at first central or internal, comes gradually to assume in the course of segmentation an external position with respect to the food yolk. The blastodermic cell is the direct descendant of the yolk pyramid. It is improbable that the yolk con- tains any active protoplasm, excepting that which radiates from the uuclei, and which is descended from the periuuclear protoplasm of the first segmentation nucleus. STAGE II.— THE BLASTODERM AND INVAGINATION. The prawn when discovered with eggs in the fifth stage of segmentation (Fig. 15) was kept in an aquarium, and the ova were preserved at intervals of several hours. Thus it has been possible to follow the changes which take place between segmentation and iuvagiuatiou with considerable detail. The egg represented in Fig. 47 is about 15 hours older than last described (Fig. 15). Cell division, which is now irregular, has become accelerated over a part of the egg so that a germinal area or disk (G. D.) representing the future embryo is formed. The side of the egg shown in Fig. 47 corresponds to that occupied by the germinal disk. In reverse view there are much fewer uuclei. The egg has thus lost its radial symmetry and become two sided. Imagination soon follows this stage at a certain point in the germinal area (G. D.). The superficial cells of the blastoderm (Fig. 48) are about one-third their former size, but they still have the characteristics of the yolk pyramids. The cell is polygonal in surface views; the nucleus is surrounded by yolk and the cleavage planes between adjacent cells (Fig. 38, Sep.) are still present. This stage is characterized by the passage of large numbers of cells from the surface to the central parts of the egg just before the invagination takes place. This process is well illustrated by a series of consecutive sections (Figs. 38-44) taken from the same egg. In all the.se sections the cleavage of the yolk can still be seen. Many of the blastoderm cells (Fig. 39, a.) are in different phases of division, the dividing plane being always perpendicular to a surface tangent. It is probable, therefore, that the uuclei with their perinuclear protoplasm, leave the yolk pyramid and pass by-aimi'boid movement into the interior. It is, therefore, evident that while morpholog- ically the yolk pyramid is a cell, the elements which pass into the egg have also the value of cells in a physiological sense. Six nuclei are met with in Fig. 38, one of which has wandered some distance from the surface. In the next (Fig. 39) two cells (a. a1.) are in the aster phase of division ; one (a2) has passed just below the surface, and another (a3) is near the center of the egg. These cells (a, a2, a3,) are sectioned again in the following figure (Fig. 40). Various phases of the process of migration* are seen in other members of this series. In an enlarged portion of a similar section (Fig. 37) several cells are met with, some at the surface just beneath the shell, and others at some distance below it. The protoplasm about the nucleus has no defi- nite bounds, and is often filled with fine particles of food yolk. It is thus evident that these cells feed essentially like amoeba?, by taking the food directly into the protoplasm of the cell. The critical stage at which cells begin to pass from the superficial to the central parts of the yolk was obtained in an egg just thirteen hours older than the fifth segmentation phase, shown in Fig. 15. There are about a dozen yolk cells in this egg, and one of these is in karyokinesis. The remainder lie very near the surface, but for the most part are separated from it by a thin layer of yolk spheres. It is thus clear that the migration of cells to the central parts of the egg begins * In the lobster the primary yolk cells arise by iU.-luiuiuat.inu, anil as suggested iu Section V, this is possibly true of Alpbeus. 400 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. before all the protoplasm, that is, the nuclei and periuuclear protoplasm of the \<>IU pyramids, has reached the surface. In the slightly older phase, shown in ri. xxx. Fig. 40, all the proto- plasm which does not pass inward is strictly superficial. The yolk has the same appearance as in previous stages, and, as already noticed, the cleavage planes (Sep.) between yolk pyramids are .still met with. Very soon, however, the central portion of the yolk segments into bulls or angular blocks (Fig. 46, Y. B.), apparently with reference to these wandering cells. A seution through the germinal disk of an egg seven and one-half hours older than that shown in Fig. 47 is given in Fig. 57. The cells in the area of the germinal disk are quite closely crowded, and the superficial seg- mentation of the yolk is still apparent. We now have a primitive epiblast, or external layer of cells, and a primitive hypoblast, composed of yolk cells which have migrated from the blastoderm. The invagination stage immediately follows that last described (PI. xxxi). A slight depres- iou occurs at a point on one side of the germinal disk, where the cells are multiplying most rapidly, and numerous cells pass downward into the yolk. The iuvaginatioiris nearly solid, and the segmentation cavity is still filled with the great mass of yolk and with primitive hypoblast. In the crayfish (Axtticiix JlupiatHis) the invaginate cavity becomes a closed chamber within the yolk (54), and this is eventually converted into the midgut, but in most decapods the pit is very small and the mesenteron is formed independently at a later period. A line drawn through the pit and the middle of the germinal disk marks the long axis of the embryo, and the point of ingrowth is at the posterior end. . The structure of the embryo is illustrated by a scries of transverse sections (PI. xxxi). The cells in the center of the egg represent the primitive hypoblast or yolk cells. The nuclei are large and granular, and sometimes occupy the center of a yolk ball. In Fig. 4!) the posterior edge of the embryo is sectioned, and the three following sections (Figs. 50-.">.j) pass through the re-ion corresponding to the invaginate area (Ig.). Fig. 52 represents the entire section, of which Fig. 51 is a part. The pyramidal cells, which form the Moor of the depression, contain at their peripheral ends no unabsorbed yolk, but at the deeper ends of the cell, below the level of the nucleus, the cell boundaries are lost, and the protoplasm of the cell blends off into the yolk and ingulfs its finely divided particles (Fig. 50). Numerous cells (Figs. 5i>-54, b, b1 r') have already wandered from the point of invagiuation into the. egg and a considerable distance forward under the germinal disk (Fig. 54, G. D.). These cells are more or less intimately united by pseudopodal extensions of the protoplasm. A coarse reticulum is thus formed, the meshes of which are filled wTth yolk, lii front of the invaginate cells, the germinal disk (Fig. 55, G. D.) is still one cell thick. At the close of the invagiuation stage the primitive hypoblast has received a consid- erable accsssion of wandering cells. This stage is usually described as the "egg-gastrula," in accordance with the theory that it represents an ancestral condition, and that the cavity formed at the surface is the remnant of the primitive, digestive tract. The discovery of delami- nation, the actual separation of the inner ends of the cells of the blastula by karyokinesis before any invagination occurs, as I have described in the lobster, and the occurrence of this or of mul- tipolar emigration in Alpheus, together with the fact that in the typical decapod the invagination has no direct relation to the digestive tract or to to the mouth and anus, point to the view already expressed in a preliminary paper upon the lobster (-•'*), that the invagination stage has no refer enee to an ancestral invaginate gastrula. It seems to me more probable that the egg with primary yolk cells corresponds to the. ctelenterate planula stage, and that these, yolk cells, which originate from the blastula and which partially or entirely degenerate, represent the remains of a primitive hypoblast. According to this view the invagination is a secondary process, which became so indelibly impressed upon the ancestors of the Decapods that it has remained in the ontogeny of present forms. The conditions which are found in the crayfish can not be regarded as in any sense general or typical. STAGK III. — OPTIC DISKS AND VENTRAL PLATE FORMED. This stai:e (Fig. 5S) is characterized by a thickening of epiblast which gives rise to the tho- racic-abdominal or ventral plate in front of and around the point of ingrowth, by the simultaneous spreading of invaginate cells below the surface, and by the appearance of the optic disks (O. D.), MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 401 two patches of ectoblast on either side of the middle line in front of the ventral plate. When the living egg is examined by reflected light these structures appear as misty white patches sep- arate from each other. (See Fig. 2.) The principal cell mass is the thoracic-abdominal plate (Fig. 58, Ab. P.), from which the thoracic-abdominal process is developed. The position of the pit (Ig.) is faintly marked by the arrangement of the cells at that point. The optic disks are due to the rapid multiplication of epi- blastic cells around definite .centers. Kadi is joined to the ventral plate, by a lateral band or cord of cells (L. Cd.), on which the appendages are subsequently budded off. A transverse cord (T. Cd.) soon bridges over the space between the optic disks, thus inclosing a triangular area, which corresponds largely to the sternal region of the adult. The extension of the, in vagi n ate cells below the surface is only partially indicated by the shaded nuclei. They advance forward and backward from the point of ingrowth, but principally upward, that is, toward the center of the egg, along the lines joining the optic disks to the ventral plate (Figs. 59, CO). The embryo covers nearly one hemisphere of the egg. It is V shaped, but the angle between the arms of the V varies much in different eggs. There is a marked contraction of the embryo which takes place immediately after this phase, that is, the area of the surface occupied by the embryo becomes appreciably smaller. A similar contraction of the embryo has been observed in Astacus (54) and Crangon (31). With the extension of the epidermis there has been a corresponding activity among the wandering cells. Their relations are well shown by sections through the entire egg (Figs. 56, 59, 60), in which we can still distinguish the primary yolk cells (P. Y. C.) from the cells derived from the imagi- nation (S. Y. C.). The cells of the first have large, granular nuclei and send out processes iuto the yolk. The others are smaller and are probably multiplying more rapidly. It soon becomes impossible to find any distinction between these wandering cells. The yolk is irregularly seg- mented into balls (Figs. 60, 03, Y. B.), inside of which the migrating cells are usually found. The epiblastic cells of the surface, which are the direct descendants of the yolk pyramids, have definite boundaries, but some of the cells of the ventral plate (PI. ix, Figs. 61, 63) tend to form a syn- cvtium, as already seen, while the wandering cells are independent, free-moving elements. The lateral section (Fig. 56) passes through the outer edge of the ventral plate, and the next toward the middle line (Fig. 60) encounters a sheet of wandering cells. We see at a glance that the migrating cells pervade the greater part of the egg, and that they pass out in all direc- tions from the region of the ventral plate (Ab. P.). Fig. 59 represents a median longitudinal section through the embryo and eiit ire egg, and Fig. 63 a part of a section highly magnified through the \entral plate and region of ingrowth. The cells immediately below the surface (S. Y. C.) are characterized by large and very granular nuclei, which stain with much less intensity than the superficial cpiblast. This shows that they are multiplying rapidly, and the finely divided yolk in their neighborhood shows also that the cell protoplasm is rapidly absorbing food. A series of trans- verse sections of this embryo is given in PI. xxxin. The plane of section in Fig. 61 is oblique and passes in a posterior direction. In Fig. 62 the lateral cords (L. Cd.) are crossed and numerous wandering cells arc encountered, while anterior to this (Figs. 68, 69) the optic disks are cut. The optic disks (Figs. 64-07) consist of a single layer of epiblast. Their cells are flat and polygonal, cell boundaries are distinct, and the long axis of the oval nucleus is parallel with the surface. When the cell divides this position is reversed, the plane of division being perpendicular to a tangent at the surface. From the optic disk the eye and its ganglia are developed. STAGE IV. THICKENING OF THE OPTIC DISKS AND RUDIMENTS OF THE APPENDAGES. An embryo a few hours older than the last described is shown in Fig. 72. On the thickened cords of cells (L. Cd.) unit ing the optic, disks to the ventral plate the traces of two pairs of append- ages can be made out— the first pair of antenna-, A (I.), and, close to the ventral plate, the mandibles (Md.). Some of the central cells (('. M.) of the optic disk have large, granular nuclei. These mark the area of most active cell division, and form an ingrowth or thickening, which is the rudi- ment of the optic ganglion. Glancing at a scries of longitudinal sections through this egg (Figs. 70-71, 73-75), we notice S. Mis. 94 26 402 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. several important changes since the last stage. The ventral plate is more extensive and the wanderiug cells are more numerous. The primary and secondary wandering cells can no longer be distinguished, but all the cells within the yolk are similar in character. They have large, granu- lar nuclei, and multiply by ordinary karyokiuesis (Fig. 75, Y. C2.). Some pass forward underneath the, embryo (Fig. 74, Y. C.) and to other parts of the egg. In Fig. 71 two yolk elements are found near the lateral cord, and in Fig. 70 one is seen just under the optic disk. In Fig. 75 the section is median and corresponds closely with the similar longitudinal section of the last stage (Fig. 59.). The transverse cord (T. Cd.) is seen to consist of a single layer of flat epiblastic cells. The structure of the ventral plate is shown in detail in Fig. 85, which is a section just in front of the point of invaginatiou. This illustrates the character of the syncytium beneath the surface of the plate and the tine degree of fragmentation which the yolk suffers in the presence of the cell protoplasm. At the surface, the cell walls are quite distinct, as already seen, but as soon as the cells pass below it, the protoplasm extends itself in pseudopodia-like processes and incloses particles of yolk. Under these favorable conditions of nourishment these elements, which must be regarded as the mother cells of the mesoderm and the eudoderm, multiply rapidly and spread to all parts of the egg. If this section is compared with that of the invagiuate stage (Fig. 54), and with a similar section of Stage in (Fig. 01), it is easy to understand the relations of the ventral plate to the wandering cells, and the way in which the thickening of the plate is brought about. At the second stage comparatively few cells take part in the iuvaginatiou, and the most of these pass directly into the yolk. But, almost simultaneously with this migration of cells there occurs a migration of cells from the surface of the ventral plate. Thus this becomes thickened and cells continue to be supplied to the, yolk. This thickening of the plate is possibly due to cell division in both planes ; that is, to delauiination and emigration. (Compare cells EC, EC1 3, Fig. 85.) The process by which the optic disk becomes thickened at this stage is quite similar, although there is no true invaginatiou concerned in it. This is shown by a series of connective sections (Figs. 70-83) passing through the entire disk. The anterior rim of the disk is cut in Fig. 70, and in Fig. 83 the rudiment of the first antenna. This thickening is mainly the result of the migration of epiblastic cells from the surface. After leaving the surface, the cell wall usually becomes indefi- nite. The relations of the optic disks to the entire egg are seen in Fig. 84. The central yolk is segmented as shown in Fig. 74, and a yolk cell usually lies within the segment. STAGE V.— RUDIMENTS OF THKEE PAIRS OF APPENDAGES— CELL DEGENERATION. The embryo represented in Fig. 93 is, approximately, three days old (temperature at Nassau 78-80° F.). It occupies nearly one entire hemisphere of the egg, the opposite side of which is covered with Hat epithelial cells like those seen at the periphery of the figure. The shape of the embryo proper is nearly that of an equilateral triangle, one angle of which corresponds to the ventral plate, and the other two to the optic disks. A line drawn through the first of these angles and the middle of the opposite side would therefore correspond with the longitudinal median axis of the embryo. The rudiments of the second pair of antenna?, A (II), have now appeared, and we therefore have present at this stage buds of three pairs of appendages, namely, the first and second pairs of anteume and the mandibles. All are developed nearly simultaneously, but the second pair of anteniue .seem to lag a little behind the rest. In respect to the order of appearance of these appendages, allied species of Crustacea differ slightly. The central parts of the ventral plate ( Ab. P.) and optic disks (C. M.) are areas of rapid cell division, and are characterized by the presence of large granular nuclei and by the irregular arrangement of the cells. In all other parts of the egg the superficial cells form a uniform stratum one cell deep. This irregularity is due to the gradual migration from the surface of individual cells in these three places. The first pair of antenna* are closely associated with the optic disks, and the mandibles abut against the ventral plate on either side of the, middle line. The appendages start on the lateral cords from definite centers of cell division, and the cells tend to assume a radial and concentric arrangement around eacli center. The space between the optic disks is now completely bridged over by a sheet of closely MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 403 crowded cells (T. Cd.), and the backward extension of this, and tlie approximation of tlie lateral cords lias quite closed over the central or sternal region of this part of the embryo (St. A.). Cell outlines arc very distinct at the surface in preparations, and they are sometimes well defined in cells which have passed from the surface to parts below it. in both the region of the optic disk and that of the ventral plate (Fig. 80, EC.), but elements closely associated with yolk are usually amu'- boid. The nucleus of the epiblastic or epithelial cell on the confines of the embryo, or on the extra- embryonic surface of the egg, has the shape of a flattened, round, or oval disk. Epiblastic nuclei in the appendages and other parts of the embryo, where there is rapid cell division, are angular in consequence of crowding, and deep-lying nuclei are generally spherical. The arrangement of the embryonic cells of the superficial epiblast in beautiful curves and rings around definite, centers— orthogouic systems of curves — is not nearly so pronounced as in the embryo crayfish (Astacusfluviatilis), according to the delineations of Keichi'iibach and Winter. Eeichen- bach states that in the crayfish the superficial embryonic cells multiply about a given center, like that of the "head fold" (optic disk), or " thoracic abdominal rudiment," according to definite laws. This was discovered by Sachs in the growing tips of plants. According to Sachs, Iteich- eubach, and others, the cell nuclei always divide in one of two opposite planes; that is, they either separate along a' radius drawn from a given center, thus giving rise to radial strings of cells, or in a plane at right angles to this, producing new strings. Thus there is developed about the initial cell a series of concentric circles and radiating lines of cells. The positions of the equa- torial plates of dividing nuclei, which some eggs of Alpheus show in abundance, do not indicate the prevalence of such a law in the earlier stages. The early embryo of Alpheus is much less diffuse than in the crayfish, and the different cell groups soon impinge on each other, and their relations are disturbed. Several transitional stages between the last two embryos figured (Figs. 72 and 93) will now be examined. The first is represented by three longitudinal sections (PI. xxxvi, Figs. 88-90), and is about seventy hours old. It is from the same prawn as the segmented egg shown in PI. xxvn, Fig. I"). These sections give some interesting facts with reference to the role of the wandering cells. The first (Fig. 88), which is nearly median, cuts the ventral plate and below it the cells which are migrating from it into the yolk. A continuous layer of cells extends anteriorly to the transverse cord (T. Cd.). In this region a wandering, mesoblastic cell (Y. C.) is nearly in contact with the superficial epiblast. The next section touches the outer edge of the ventral plate (Fig. 89), which is marked by large granular nuclei, and crosses the lateral cord and rudiments of the ap- pendages (A. I, A. II, Md.). The folds of the latter arise through the ingrowth of superficial cells. Here another cell (Y. C.1) is close to the outer surface of embryo; another (Y. C.2) is in a distant part of the egg and is in the aster stage of karyokinesis ; others still ( Y. C.3) have wandered in a diametrically opposite direction. In Fig. 90 we see still more of these wandering cells, in this instance, chiefly above the embryo. In the middle of the optic disk some of the large granular cells visible from the surface (Fig. 93, C.M.), are met with, and one of them (Fig. 90, EC. dotted line extended) has just passed below it. Figs. 80 and 87 are parts of longitudinal sections of an embryo six hours older than the last. The first exposes the optic disk (O. D.) ami gives evidence of the further increase of the latter by the emigration of cells from the surface, and it is quite probable that some of the wandering, meso- blastic cells (Y. C.) have already attached themselves to it. (Compare Fig. 90.) In the next (Fig. 87) the inner edge of the optic disk (O. D.) and the outer border of the ventral plate (Ab. P.) are involved. Here the epiblast consists for the most part of a single layer of cells. Two large wandering elements (Y. C.1) are in contact with the surface cells of the embryo in the neighbor- hood of the appendages. Fig. 91 is from an embryo twelve hours older than the last. The plane of section passes obliquely through the optic disk, cutting the anterior half of one (to the left) and the posterior part of the other. A study of this section and of the series to which it belongs, shows us beyond a doubt what is the fate of large numbers of wandering cells, present at this time. As has been already shown by preceding figures (see Figs. 73, 88, and others), the cell mass constituting the thoracic-abdominal plate is now the principal source of the wandering yolk elements, and, as has been also shown, they migrate into all parts of the egg, multiply by karyokinesis, and settle 404 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. upon the optic disks, the bases of the appendages, and other parts of the embryo. They also pass to the extra-embryonic surface of the egg. In Fig. 91 one of these wandering cells (Y. C.) is approaching the surface, while another (Y. C.1) nearly touches it; and in PI. xxix, Fig. 34, which represents a portion of a section from a later series (Fig. 106) greatly enlarged, we tiud two yolk elements (Y. C.) quite at the surface. They are triangular in outline, oue of the flat- tened .sides being applied to the surface of the egg. Histologically their nuclei are more gran- ular and stain with less intensity than the nucleus of the ordinary cpiblastic cell (Ep.), which appears spindle-shaped in vertical section. But between a cell like that seen in Fig. ill (Y. C.1), where the long axis of the nucleus is at right angles to the surface, or cells like those shown in Fig. 34, where the nucleus is flattened against it and the ordinary epiblastic cell, a variety of tran- sitional phases can be found. This is most clearly illustrated in the next stage. The egg already described (Fig. 93) shows some important changes. The structure of the optic disks and ventral plate is readily seen in the transverse (Figs. 92, 94, 95) aud lateral longitu- dinal sections (Figs. !)(i, 97). The optic disk, which in stage in consisted of a single stratum of cells, (Fig. 69), is now a thick cell-mass two-thirds the size of the ventral plate. In it we still distin- guish a small area of cells with large granular nuclei (Fig. 93, C. M.), which, as we see in Figs. 96, 97, C. M., O. I)., is clearly differentiated. It occupies a position just without (or external with respect to the longitudinal median axis) the center of the disk. The nuclei of surrounding cells are not mure than half their size. These large cells do not all lie at the surface, but form a solid mass extending into the yolk. The evidence of karyokinetic figures shows that these cells are dividing, and usually in planes perpendicular to the surface. This results in the crowding of the cells, and also in their migration from the surface to parts below it. lu Fig. 90 there is a cell (ec.) whose nucleus has sunk below the level of the surrounding cells, but the cell protoplasm still reaches up to the surface. Such cases render oue cautious in pro- nouncing positively upou the emigration of cells, but sections like that given in Fig. 94, and the i'act that cell divisions seem to be for the most part in one plane, convince me that the thickening is partly, if not largely, due to this cause. In Fig. 95, a large cell, the polar star of which is just below the surface, is delamiiiating. Cases of this kind were rarely noticed, but were observed at a later stage (PI. xxxix, Fig. 102, ec.). That wandering cells attach themselves to the optic disk, there is little doubt. They can be traced in all stages of progress from the region of the ventral plate to the neighborhood of the disk (Fig. 90, Y. C.) until they finally come in contact with it. The thickening of the optic disk described in Stage in (PI. xxxni. Fig. 69), is therefore effected: (1) partly, perhaps largely, by emigration of cells from the surface; (2) partly by delamination ; (3) by the accession of wandering cells; (4) by the indirect cell division of the elements constitut- ing the deeper part of the disk. Chromatin grains make their appearance somewhat abruptly at this stage (Fig. 96, S.1 ~'!.) and they serve to explain in some degree the peculiar granular nature of nuclei in earlier stages. They originate by the degeneration of cells in the ventral plate and in other parts of the embryo, and probably correspond to what Keicheubach has called in the crayfish "secondary rnesoderrn cells." Their history has been fully traced aud will be discussed in Section vi. The structure of the ventral plate (Fig. 92) resembles that of an earlier stage (Fig. 85) with the difference that the nuclei are considerably larger aud contain from one to several large spore- like masses of chroiuatin. This cell mass has not increased in bulk at a rate corresponding with the growth of other parts, since elements are being continually subtracted from it aud added to the yolk. This loss on the part of the ventral plate is made good not only by cell divisions, but also by continued emigration from the surface (Fig. 98, Ab.). Au embryo six hours older than the last is represented by three longitudinal sections (Figs. 98-100). The optie disks resemble in size and general shape the three pairs of rudimentary limbs. The abdominal fold is not yet formed and the labrum is undeveloped. The stomoda-um (Fig. 98, Std.) is just making its appearance as a slight invaginatiou of epiblast on the middle line between the first pair of antenuic. The number of chromatin balls (S) aud degenerating cells (S. C.) has greatly increased. We see them in all parts of the embryo. Sporelike nucleoli are seen in cells of the epiblast aud are confined to no part of the embryo, but they are most characteristic of the ventral plate and optic disks. MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 405 In Fig. 98 several independent chromatiu balls (S) are seen in the yolk, and the granular cells of I In- ventral plate are very marked. A large nucleus of one of the cells (ec.), which contains sc\cral sporelike bodies, is very irregular in shape, owing to the pressure to which it is subjected, and it lias evidently been crowded down from a level nearer the surface. The cell protoplasm, however, shows that it still belongs to the surface tier of cells. This is also true of the small len- ticular nucleus next it on its anterior side. Just anterior to this cell is another (Ec.1) which is in the act of dividing in the usual way. It seems probable that had this division been effected one or more of the adjacent cell.s must have been crowded quite below the surface. It is difficult, however, to always determine whether cells whose nuclei are a considerable, distance below the surface do not SIMM! it]> stands of protoplasm to meet it. We find in the ventral plate, inde- pendent cells multiplying by karyokinesis, and, as evidences of delamination in this region are rare or altogether wanting, we are convinced that, as in the case of the optic disks, this thickening is largely due to the migration of cells from the surface. There is at this stage a fairly well-defined sheet of cells (Figs. 99, 100, Mes.) extending forward from the ventral plate on either side. The nuclei are oval or elongated, and their long axes are parallel with the surface, that is, at right angles to the major axes of the superficial epiblast cells. This layer of cells is most marked at the bases of the appendages (Fig. 100, Mes.) and extends from the optic disk on either side to the ventral plate. The question of the origin of these cells is not difficult. They are wandering cells which have settled down on these parts of the embryo. They form a part of the future mesoblastic tissues ; exactly what part they play will be discussed later on. They multiply by indirect division and extend into the folds of the appendages, while some, on the other hand, degenerate. STAGE VI. — THE EGG-NATJPLIUS. The fully developed egg-uauplius is represented in Fig. Ill, but before this condition is reached there are several intermediate stages to be considered. A series of longitudinal sections (PI. xxxix, Figs. 101-105) illustrates the structure of an embryo twelve and a half hours older than the one last described. The thoracic-abdominal fold (Fig. 104, Ab.) can now be recognized, and the stomodieiim (Fig. 105, Std.) has the form of a straight, narrow tube, between the buds of the first and second pairs of antenna?. The space between these two structures is tilled with yolk fragments, among which are scattered, numerous chrouiatiu particles (S) and cells derived from the thoracic-abdominal fold. The epiblast of the sternal region (Fig. 105, 98, St. A.) is no longer a simple layer, but is slightly thickened. This thickening seems to be partly due to rapid cell division in one plane. The cell nuclei are elongated or wedge-shaped and stand perpendicular to the surface. As will be seen later it is also due to the accession of wandering, mesoblastic cells. The stomodaniiu is a relatively long straight tube with very slight lumen, and is surrounded with chromatiu grains and scattered cells. It is formed at a considerable distance in front of the point of invagination, and one to two days earlier than the proctodajnm. The thoracic-abdominal fold seems to arise by the sinking down of the epiblast along a defi- nite, line. There is thus formed a narrow transverse pocket (Fig. 10G, Ab. C.), which is quite deep and perpendicular with the surface. Numerous cells continue to pass from the thoracic-abdom- inal fold to various parts of the embryo, and to join the sheets of cells (Fig. 1 <>.">, Mes.) already mentioned. In Fig. 103 the four segments of the embryo are well shown. Th'is section crosses the, optic disc (O. G.), the buds of the three appendages, and the edge of the thoracic-abdominal fold. The sheet of mesoblastic cells (Mes.) is most marked opposite the folds of the appendages. The optic disks are now large masses of cells united by a transverse cord which is thickened slightly on the middle line. In Kig. Klli, three cells are met in karyokinesis, one in the abdominal area and two side by side in the optic disks. The former exemplifies the common method of cell division, while the latter is a good example of the less common delamination. As has been already seen, karyokinetic figures of dividing cells are commonly met with in Alpheus at all stages, except- ing the, species Alpheus minor, where the division is at first probably direct. I have seen unclear figures at, the yolk segmentation stage of ('nnii/on, also in Hippa, Pontonia and Homarus, and Rei- cheubach found them in abundance in Astacus. Indirect cell division is undoubtedly the rule in 406 MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. the developing egg8 of the Crustacea and probably of all the Metazoa. Since we ofteu study only the rapidly achieved result, the phases of nuclear division may be easily missed. Fig. 100 is a median longitudinal section of an embryo preserved a few hours later than the last. This shows the thoracic-abdominal fold and cleft (Ab. Ab. C.) previously referred to. The sternal area between the latter and the stomodaMim now lies nest the egg shell, but soon changes its position (Fig. 125, St. A.), owing to the forward growth of the thoracic-abdominal fold. The. labrum, here undeveloped, soon grows backward towaid the latter, helping to bend the lesophagcal tube and probably slightly altering the position of the mouth (Fig. 125). The proctodseum arises as a, solid invagination of the epiblast, at a cousiderable distance behind the abdominal cleft (Fig. 100, IM.), in a stage intermediate between the embryos represented by Figs. 105, 101). A transverse section through the. point of invagination is shown in Fig. 120, Pd. The relation of the. embryo to the rest of the egg can be seen in Fig. 108, which belongs to the same species as Fig. 100. Besides the shell, which is unnaturally distended, the egg is sur- rounded by a delicate embryonic membrane (Mb.). This membrane is secreted early by the super- ficial epiblast, as shown by the fact that it does not conform with the thoracic-abdominal fold. Wandering cells (Fig. 108, Y. C.) become gradually less abundant, but still continue to pass to the outer surface of the egg next the epiblast. The transition from the wandering, amoeboid cell to the llalteued mesoblast cell, lyiug close to the surface, can be best followed at this stage. The fully developed egg-nauplius (Fig. Ill) is about a week old. Embryos from the same prawn vary slightly in size and in the degree of development, and also in the general character of the, cells. In some, the cells are larger and fewer in number, in others they are smaller and much more numerous. The embryo is usually at one pole of the obloug egg. That represented in Fig. 1 1 1 is about one eightieth inch long, and if the entire end of the egg were shown the drawing would be. nearly twice as large. The relation of the embryo to the whole egg can be determined from Fig. 127, where the plane of section is through the long axis of the egg, but through the short axis of the embryo. In the course of development the egg increases appreciably in size and also changes its shape, at ftrst being spherical, but gradually becoming oblong. At this period the long axis of the embryo (using this term to apply to the more obvious embryonic- tissues of the egg) is parallel with the short axis of the egg, while in the course of growth the embryo spreads over one side of the egg, until its long- axis coincides with that of the latter. The optic disks have become large oval masses of cells which project from the surface, and may now, for the first time, be appropriately called lobes. They represent the eye aud eyestalk. The blocks of cells (S. O. G.) in intimate relation with the optic lobes are the ganglia of the antenna-, and represent a large part of the future brain. The appendages are all simple, but a bud soon grows out from the posterior sides of the second pair of antenna;. The right antenna is already bifid near its tip. A little later it has the appearance shown in Figs. 109 and 110. Rudiments of the first pair of maxilla; (Mx. I.) are also present. The ganglia of the second pair of antennae are developed in close union with the ganglia of the antennules. Together they form the supra-oesophageal ganglion or "brain." The stomoda-um (Std.) appears Irom the surface as a distinct mass of cells extending behind the labrum (Lb.). The thoracic abdominal fold, at first vertical to the surface, bends up aud grows forward toward the labrum, and a shallow groove which marks the median notch of the telson plate, (Fig. 110) is formed at its extremity. The anus passes forward (backward in a morphological sense) from its dorsal position through the median notch until, at a considerably later period, it comes lo lie on the ventral side of the fold near its apex. The mass of cells (H.) behind the anus probably represents the heart. Near the mandibles and maxilla-, cells are seen with large granular nuclei. These are cells which have migrated from the yolk to this part of the embryo. Nuclei of epithelial cells are sprinkled over the entire surface of the egg, but increase in number as we approach the embryo. The section through the entire egg (Fig. 127) shows some of the general characteristics of the egg- uauplius. The thoracic abdominal fold is here cut on a level with the anus, and lies in a wide shallow groove. The yolk is composed of irregular blocks, only a few of which are represented, and con- tains numerous wandering cells. Some of these (Y. C.'-) are settling down on the ventral nervous MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 407 plate, while others (Y. C. Y. O.1) have wandered far and wide through the yolk. The embryo is raised considerably above the general surface of the egg and the shell more closely invests the egg than shown in the drawing. The embryonic membrane is not represented. Fig. 125 is a nearly median longitudinal section, and shows the relations of the thoracic-alt dominal fold, the oesophagus, and the ventral or sternal surface between them. The loose and irregular arrangement of cells immediately below the surface is most marked, and also the granu- lar nature of the nuclei which is such a constant character. Numerous degenerating cells (S. S. C.) are seen near the oesophagus, and amu-bilbrm cells can be traced from the thoracic abdominal fold to the surface immediately behind it. The structure of this embryo is illustrated more completely by a series of transverse sections (Figs. 114-124), the first two of which (PI. XLI, Figs. 114. 115) traverse the optic lobes, and the third cuts the brain. The central mass of large cells which was noticed in the optic disks can no longer be distinguished. The lobe (O. L.) is composed of similar cells with granular nuclei, the superficial tier being somewhat the larger and columnar. In the brain the central cells are smaller and stain more intensely than those at the surface (Fig. 116, S. O. G.). Wandering cells (Y. C.) can be traced in their passage from the yolk to the optic lobes, brain, ventral nervous plate, and other parts of the embryo. Between the lateral halves of the brain there is a shallow median fur- row (Fig. 110, M. F.). This is continued backward into the much broader and deeper depression in which the convex ventral side of the abdomen fits (PI. XLII, M. P.). The three following .sections (Figs. 117-1 19) .pass through the (esophagus, and the ventral nerve thickening immediately behind it. About the oesophagus (Fig. 117, Std.) numerous chromatin balls (S., S. C.) are seen in the yolk, and a mass of cells (Mes.) is met with at the base of the appendage and within its fold. These elements are derived from the wandering cells and must be regarded as mesoblast. The fold of the appendage consists essentially of a single layer of cells. Those elements which enter it undoubt- edly go to form the musculature of the limb while the cells of somewhat similar appearance, which are derived from the ectoblast, represent the ganglia and nerves. Ou either side of the oesophagus the yolk has undergone important physical and chemical changes. The yolk spheres or blocks are full of vacuoles and have a corroded and granular appearance, while in contact with the embryonic cells there is a residue of small retractive granules. These vary considerably in size, and some of them stain lightly in bsematoxylin and represent the last stages in the degeneration of chromatin. The eroded and altered yolk (A. Y. S.) is represented in many of the sections. Between the (esophagus and bases of the antenna1 the yolk is absorbed, leaving a protoplasmic reticulum (Fig. 117, Ret.). In Fig. 118 the mass of cells representing the tnandibular ganglion (Md. G.) is sectioned, and in the following figure, the mandibles themselves (Fig. 119, Md.). Numerous cells, both in this and the following sections, are seen in the course of their migration from the yolk to join the ventral nerve-thickening. The latter, which is the rudiment of the nervous system, is at this .stage scarcely thickened at all, on the middle line, below the level of the mandibles. Thus in Fig. 121 the buds of the first maxilla; (Mx. I.) are united by the primitive layer of epiblast. To this a single migrating cell has attached itself on the middle lino. Migrating mesoblast cells (Mes.) also pass into the fold of the appendage, and others (Fig. 120, Mes.) take up a position against the epiblast of the body wall. The nuclei of the latter stain intensely and become flat, spindle shape in section, and probably represent the nuclei of muscle cells. The structure of the abdomen at this stage is shown in Figs. 120-123. The body wall of the dorsal side, consists of a single layer of columnar epiblast, while the ventral wall is thickened. The hind gut (Fig. 122, Hg.) is a tube, the wall of which consists of a single layer of cells. It is laterally compressed so that the lumen is hardly appreciable. The intervening cells (Mu.) largely represent the rudimentary flexor and extensor muscles of the abdomen. A comparison of Figs. 12:; and 125 shows that cells extend from the thoracic-abdominal fold on all sides into the yolk. The cells at the surface in Fig. 124 have come mainly from the yolk (H.) and are in the posi- tion where the heart is subsequently developed. Cells approaching the surface in this region are very clearly shown in Fig. 125 (Y. C., Y. C.1), which is a section through a somewhat younger embryo. The wandering cells, as we have seen, abound in the parts of the yolk nearest the embryo. The 408 MEMOIRS OF TFTE NATIONAL ACADEMY OF SCIENCES. nuclei have the usual characteristics — irregular shape and graunlar contents. They are sur- ronmleil by a small irregular body of protoplasm which does uot readily stain and which is ofteu difficult to observe. In PI. xxix, Fig. 33, several of these wandering cells are seen greatly enlarged. Sometimes, as here, the perinuclear protoplasm appears to join that of neighboring cells, but in most cases there appears to be no such connection between them. The eudoderin, though not represented in the drawings, makes its appearance as a distinct cell layer at this time. In an embryo a few hours older than that represented in Fig. 125 the eudo- derm has the form of a narrow sheet of rather large, cells, between the yolk and the rudimentary heart, near the body wall. In the space corresponding to the heart, blood corpuscles can already be detected, besides scattered mesoblastic cells. Both the latter and the eutoblast are derived from the wandering cells which come out of the yolk. STAGE VII. — RUDIMENTS OF SEVEN PAIRS OF APPENDAGES. Fig. 110 represents a phase intermediate between the egg-nauplins (Fig. Ill) and the present stage (Fig. 130), and is of special interest for the light which it throws on the history of the wan- dering cells. The structure of this embryo is illustrated by PI. XLIV, and Figs. 136, 137, 144, 145. Fig. 137, which represents a section just behind the base of the first antenna; (A. I.), maybe com- pared with Fig. 117. Numerous yolk elements are found in the vicinity of the u-sophagus, where, as will be seen (Fig. 134, Mn.), they become speedily converted into muscle cells and somatic meso- blast. In Fig. 136 several wandering cells attached to the body wall, have, all the characteristics of blood corpuscles, a deep staining granular nucleus, and a clear irregular cell body. The blood cell and muscle cell are both derived from wandering mesoblastic cells, and in the early stages of their metamorphosis they resemble each other, so that it is not always possible to distinguish them. Undoubted blood cells, however, have already made their appearance. Figs. 144, 145 show that the yolk' is pervaded by a great number of cells. These originated in the way described, chiefly by migration from the ventral plate and thoracic-abdominal fold, and also by subsequent multiplication in the yolk. At this period some of the cells migrate to the pole of the egg opposite the embryo and apparently assist in forming a conspicuous dorsal plate (Dp.). In this embryo they have not quite reached the surface. A structure is eventually formed which reminds one of the "dorsal organs" of various Crustacea. This and the wandering cells will re- ceive further consideration later on. The embryo of Stage VII is represented in Fig. 130, PI. XLIV. Figs. Ill, 109, 110, and 130 form a consecutive series of embryos, each but a few hours older than the preceding. In the first (Fig. Ill) the optic lobes, first and second pair of an tenure, and mandibles are all simple appen- dages, and are quite similar in general appearance. The abdomen and part of the thorax are represented by a simple fold, the thoracic-abdominal fold. In the second embryo (Fig. 109) the parts are more compact and the second pair of antenna; are forked at their tips. In the third phase (Fig. 110) the optic lobes and abdomen exhibit the most rapid growth. The former are drawn closer together and arch outward from the middle line. The anus is dorsal. The abdomen extends forward until it nearly meets the labrnm and has a slight groove or depression at its ex- tremity. All the appendages have assumed a more oblique position with respect to the long axis of the body, and the second antennae are now the largest. In the fourth phase (Fig. 130) we see the same changes carried still further. The optic lobes are large convex disks which join each other on the middle line and are utimately united to the brain. The anus is terminal. On at least the first pair of antenna' hairs are developed, although there is not perhaps so marked a contrast between the first and second antenna? in this respect as would appear from the figure. The first and second maxillae and the first and second maxillipeds are present as rudimentary buds. The general structure of the embryo (Figs. 129, 131) agrees with that of preceding stage. The embryonic tissues consist of cells of various sizes and shapes, from quite large cells down to particles no larger than the balls of chromatiu which are suspended within the nucleus, and from the spherical to the lens-shaped, spindle-shaped, and wedge-shaped forms. Generally all the nuclei agree in containing coarse grains of chromatiu or uucleoli. These vary much in size and number in different nuclei according to the condition of the cell. In degenerating nuclei, the chromatiu residue is aggregated into fewer and larger masses. MEMOIES OF T1IK NATIONAL ACADEMY OF SCIENCES. 409 Wandering cells are now scattered throughout the entire egg. They occur in abundance both in proximity to the embryo proper and on the sides of the body walls, and especially in the region immediately behind the thoracic-abdominal fold. Fig. 131 is a median longitudinal section through an embryo like that shown in Fig. 130. ' The outer or superficial cells are generally columnar and have distinct boundaries. Their nuclei are spherical, elongate, or wedge-shaped. They divide in both planes, but most commonly in the plane perpendicular to the surface. When we compare this section with the. similar one (Fig. 125) of the preceding stage the most striking difference is the great increase in the length of the thoracic- abdominal fold and the ventral thickening of the nervous system. Some of the cells of the latter next the yolk (Mes.) are flattened and spindle-shaped. They have recently come from the yolk and joined the nervous system, and it is possible that these and similar cells represent primitive connective tissue envelopes. In a more lateral section (Fig. 129) the optic lobe is seen to consist of two portions, a super- ficial layer of large cells (O. E.) and a deeper layer (G. L.). The first is continuous with the general epiblast over the surface of the body. Its cells multiply in both planes, and some of them pass below to join the deeper layer. The cells of the latter multiply by the usual mitosis and also receive accessions from the yolk. The plane of section passes through the equatorial plate of one of these dividing cells. Some of the cells nest the yolk (Mes.) are flattened like those just described in the ventral nervous thickening, but this condition appears to be somewhat transitory. The outer layer of the optic lobe may be regarded as a retinogen, since from it, or from a layer corre- sponding to it, the visual apparatus of the eye is developed, while from the deeper layer or gangliogen the optic ganglia of the eyestalk are formed. A comparison of the transverse sections (Figs. 128, 132-135) with corresponding sections of the previous stage (Figs. 115, 117. 121, 124) shows some interesting changes. The brain is larger and more compact, and some of the cells next the yolk are flattened (Fig. 132, Mes.) and bear a resemblance to muscle or connective tissue cells. They originate from the cells marked Ct. S. in Fig. 110, and come from the yolk. Like the cells already mentioned in the optic lobes and ventral nervous system, they seem so represent a rudimentary perineurium,but, as a well developed covering of the nervous system is not present until a considerably later stage, they are probably transitory. Fig. 134 corresponds closely with Fig. 117. It shows the section of the oesophagus and of the ganglia of the second antennae. In the younger stage the ganglion (seen to the left in Fig. 117, at the base of the appendage) is a small, loose mass of cells associated with the surface epiblast, while in the older embryo it is in contact with the wall of the oesophagus, is more compact, and its cells are somewhat differentiated. The central nuclei are smaller and stain most intensely. The esophagus (Std.) is suspended to the body walls by rudimentary muscles (Mu.), the cells of which are much elongated. They are derived from migrating mesoblastic cells like those seen in the earlier stage (Fig. 117, S. C.), or like the one shown in this figure to the left of the (esophagus. Fig. 137, which is from a stage intermediate between the two just considered, gives additional evidence of this role of the wandering cells. Fig. 128 furnishes a very interesting comparison with Fig. 121. In the latter, cells abound in the yolk adjoining the rudiment of the ventral nervous system, which is represented by the primitive epiblast on the middle line. In the older stage scattered mesoblast cells are greatly reduced in number and the ventral thickening is very marked. Cells of recent derivation from the yolk (Mes.) at the base of the appendage can also be distinguished. In Fig. 13J, as in Fig. 124, the plane of section is just behind the thoracic-abdominal fold. Here we recognize a tier or plate of tall, columnar cells (End.), the nuclei of which lie at the deeper ends of the cells or on the side away from the yolk. In the presence of these bodies the food yolk (Fig. 135, A. ¥. S.) is absorbed or converted into a granular residue. This layer represents the I'titobhifit or the epithelium of the mesenteron already described. Numerous wandering cells are encountered (Figs. 124, Mes.; 135, Y. C.), which take up a peripheral position, and from the first are closely associated with the epithelium of the hindgut. They unite the mesenteron to the hiudgut, and it is impossible to say exactly where the one begins and the other ends. Between this ente- blastic plate and the surface epiblast (Ect.) numerous cells are interpolated (Figs. 133, 135, Mes.), which are undoubted mesoblast. They are directly continuous with the layer of mesoblast (Fig. 410 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 128, Mu.) which extends throughout the abdomen between the hindgut and body wall. Mesoblast cells derived from the yolk (Fig. 129, Mu.) are also seen underneath the epiblast on either side of the body. The nuclei of these cells stain very deeply, and the cell protoplasm is prolonged into a short fiber or forms an irregular body. In the latter case the cells have the appearance of blood corpuscles. The. heart originates in the mesoblast (Fig. 135, Mcs.), between the entoblast and outer wall of the body, just behind the thoracic- abdominal flexure. At this time fibrous substance (punktsubstanz) is first seen in the braiu. It appears as two small masses joined by a transverse commissure, in a plane just anterior to the roots of the first antennae. It is distinctly fibrous and apparently originates from the protoplasm of surrounding cells. STAGE vm. — SEGMENTATION OF THE NEKVOTJS SYSTEM — AT LEAST EIGHT PAIRS, OF APPEN- DAGES PRESENT. We have two longitudinal sections (Figs. 138, 130) to illustrate this stage. If we compare the latter figure with the corresponding one of the previous stage (Fig. 131) we see at a glauce. that a long step forward has been taken in development. Between these, 1 have obtained one intermediate phase, which is a trifle older than the embryo given in PI. XLIV, and can be best described by showing in what respects it differs from it. The rudiment from which the nervous system is formed (Stage vn) is a plate of cells extend- ing from the optic lobes to the apex of the abdomen. Anteriorly in the brain (Fig. 132) it is very thick, but gradually thins out as it is followed backward, until it consists of a single layer of epi- blast at the very tip of the abdomen. In the phase intermediate between Stages vn and vm the portion of nervous thickening between the (esophagus, which passes through it, and the thoracic- abdominal fold is partially segmented into cell masses, the primitive ganglia. The cells of this cctoblastic thickening may be roughly divided into a surface layer, the nuclei of which are large and contain diff use granules of chromatin, and a deep cell mass with smaller nuclei which stain more intensely. Similarly in the optic lobe we find a thick pad of uniform, deeply stained cells (ganglioyen) next the yolk, and separated from this a well-marked surface layer (reti- nogen) of larger ceils. The superficial, epiblastic cells on the inner or ventral side of the thoracic- abdominal fold are large and columnar. The nuclei are very much elongated and closely crowded together, and lie at all levels. This implies rapid cell division in this layer in a plane perpen- dicular to the surface, and as a result of this, the thickening of the ectoblastic plate in this region, such as we see in the next phase (Fig. 139). Near the apex of the abdomen there is a transverse zone of very large cells, and the smaller superficial cells adjoining it arc arranged in parallel lines. Something resembling this was noticed in Stage vi (PI. XLII, Fig. 120, B. Z.). It corresponds to the budding zone (Knospungszone) which Reichenbach figures and describes in the crayfish. He detects it in a very early stage (Stage E, embryo with rudiments of the mandibles, corresponding nearly to Stage v of this paper), and finds that it consists of a transverse zone of cells containing large nuclei at the extremity of the abdo- men, below the notch of the telson. From it the segments following the mandibular segment are gradually budded off. The present stage is characterized by the segmentation of the nervous system and the great development of the optic lobes. The ventral nerve plate is blocked out by lateral constrictions into its component ganglia. There is also a median vertical ingrowth or constriction which tends to divide the plate into a double cord. It is, however, discontinuous at the middle of each block, so that the ventral nervous system consists of a double chain of ganglia, each pair of which are united by a transverse commissure or band of cells, and each ganglion is similarly joined to the one behind or in front of it. Distinct commissures pass from the brain around the oesophagus and unite the former to the ventral nerve chain. The first pair of post-oral ganglia contain two masses of fibrous substance united by transverse fibers as in the brain. The ganglia following these also contain punktsubstanz. It is developed as a small isolated mass on the dorsal side of eacji ganglion, toward the middle line. As development proceeds these masses increase in size and are grad- ually united by transverse commissures in each pair of ganglia (PI. XLVI, Figs. 150, 151, PI;.). A mass of fibrous or granular substance appears in each optic lobe in the gangliogen next the brain. Fibers pass from it to the punktsubstan/ of the brain, which sends fibers down to the MEMOIRS OF TUB NATIONAL ACADEMY OP SCIENCES. 41 1 circum-cesophageal commissures to the first post-oral ganglia. The fibrous masses unite on either side of the ventral chain to form a double longitudinal commissure. Thus at an early stage tin- optic lobes, brain, and ventral nervous system are intimately connected by fibrous tissue. All the segments of the body are now marked oft' as seen in Fig. l.'J9. The lirst pout-oral segment is the mandibular (g. IV), and following it are the segments of the maxilhe the maxilli- peds and the first pereiopod. The second thoracic ganglion lies in the angle made by the thoracic-abdominal flexure. The third, fourth, and fifth thoracic, and the six abdominal segments constitute the thoracic-abdominal fold. In the superficial parts of the brain large elements are met with which multiply by mitosis, but have a distinct histological character. They consist of a large spherical nucleus containing a. diffuse chromatin network, and a granular cell body without definite wall. They are the ganglion cells, which are abundant in all later stages. The optic lobe consists of two sharply distinguished parts already mentioned, the retinal and jjanglioiiic portions. The retiuogeu which forms the eye is a superficial disk of cells resting like a cap on the other part of the lobe, thickest on the outer side of the lobe and rapidly thinning out toward the middle Hue to a single layer of cells. The nuclei are elongated and wedge shaped, and cell division takes place commonly in a plane perpendicular to the surface, corresponding to the long axis of the vertical nucleus. The gangiiogeu consists of a deeper portion next the brain, containing a ball of fibrous tissue, and a part next the retiuogen or eye rudiment. Below the thick- est portion of the eye the cells have large nuclei, which show a tendency to arrange themselves in rows radiating from the deeper half of the lobe. These large, clearer cells also extend down to the food yolk, and in lateral longitudinal sections (Fig. 138) form the inner stratum of the lobe. The cells which lie between them and the eye, here one cell thick, are smaller and stain intensely (v. Section ix). The heart (Fig. 139, H) is now a broad and greatly flattened chamber between the body wall and endoderm (End). It extends forward a considerable distance between the epiblast and yolk, and is continued backward into the superior abdominal artery (A. a. s.). It is tilled with serum and blood corpuscles. The endoderm is a more conspicuous layer (End.), and the wandering cells are reduced in num- bers. They are still seen in all parts of the egg, approaching the body wall, the nervous system, the eudoderin plate, and other parts of the embryo (Fig. 139, y. c.). STAGE IX.— EMBRYO WITH EYE PIGMENT FORMING. A sketch of the embryo at the time when eye pigment has already formed is shown in PL xxvii, Fig. 158. The optic lobes are huge pear-shaped masses meeting on the middle line in front and arching outward and backward on either side of the brain. The ocular1 pigment appears as a thin, dark-brown crescent near the outer surface of the lobe. Pigment is first formed at the posterior end of the lobe nearest the base of the anteuuules, and spreads upward over its larger convex end. The brain is constricted into two portions corresponding to the auteunular and anteunal segments. The segments of the abdomen are faintly marked off at the surface, and the telson plate which overlaps the mouth, is deeply forked at its extremity. (Compare with spatulate telson of the first larva, PI. xxi, Fig. 9.) The plumose seta- which garnish the posterior edge of the telson are now represented by short stumps. The first pair of antenna1 are stout, simple appendages, tipped with seta1 and folded backward along the sides of the body. The second pair of antennie just inside of the latter, are biramous. They are also hairy at the tips, and the embryonic membranes surround them like the fingers of a glove. The present stage is illustrated by Pis. XLVI and XLVII. The drawings are made from different embryos, all of which are of the same age, excepting those represented by Figs. 152, 15S, 15!>, and 101, which are a trifle more advanced. In the first series (PL XLVI, Figs. 146-151) the pigment cells are .just forming in the eye. They are first developed in the thickest part of the retiuogeu next the food yolk. A single section, like 412 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. Fig. 140, sbows the complete history of development of the retinal layer from its one-celled con- dition toward the median line (already seen at an earlier stage, Fig. 136, Stage VII) to the point farthest from the middle line, where pigment is formed. The fibrous nerve tissue of the gan- liogen now consists of three musses, a ball nearest the brain, which is the first to appear, and two smaller masses between it and the retinogen. Huge ganglion cells (gc.) are of frequent occur- rence, especially at the surface of the eye stalk next the brain. (The details of the development of the eye are reserved for a special section.) The brain at this time (Figs. 140-148) differs from that of the previous stage chiefly in point of size. It is composed of nerve cells and large ganglion cells (go.), which occur chiefly near the outer surface, and central fibrous tissue. It rests against the food yolk, and in the living embryo it is bathed with a current of blood (B. S.). Ihefibrous tissue of the brain has the form of a letter H with a wide and short transverse bar. In front of the transverse commissure (Fig. 147, Tc.) the fibrous substance is prolonged on either side into the optic lobes; behind, it extends down to the ventral nerve cord, on the inner side of the cesophageal ring (Fig. 148, ocm.). The ventral nervous system, like the brain, is bathed with blood, which fills a large sinus between it and the yolk. This communicates with extensive blood sinuses extending along the sides of the body (Figs. 148-154, fi. S.). In some cases the food yolk, usually in an altered or finely divided state, is in close contact with the nerve chain (Fig. 157). Cells extremely flattened and spindle shaped in section, are found in small numbers closely applied to the nervous system (Figs. 152, l.r>7, j»\), and forming a rudimentary perinenrium. In most cases they are undoubtedly iso- lated cells, and do not constitute a membrane. They originate from the wandering cells and correspond to cells similar in shape and origin which appear between the yolk and nervous system at a. much earlier period (Stage VII, Fig. 131, Mes.). The brain and ventral cord are not as yet differentiated from the superficial epiblast, but anteriorly, flattened epiblastic cells begin to appear between the nerve cells and cuticle. Ganglion cells (Fig. 150) also make their appearance in the cord, commonly at either .side, close to the surface. There is direct continuity of fibrous substance in the optic lobes, the brain, and the ventral cord as far back as the abdominal ganglia. In the latter this substance has not been developed. In each single ganglion there is a ball of this tissue which is united to its fellow in the same segment by a transverse commissure (Fig. 151), and to the preceding and following ganglion by longitudinal com- missures. It is as a rule completely inclosed by the ganglion cells, but is separated from the yolk or blood sinus often by a unicellular layer (Fig. 150), and in theantennular and antenna! segments of the brain (Figs. 147, 148) the cells next the yolk are discontinuous. In the circumo3sophageal cords the fibrous tissue also is without a cellular cortex on its inner or central side (Fig. 14!),/s.). With slight changes these relations are maintained in the hatched larva (see PI., LV., Figs. 220-222). The foregut is at this time a tube with definite walls and wide lumen (Figs. 148. 152, //>.) we noticed an extraordinary migration of wandering cells to the pole of the egg opposite the embryo. The.se cells eventually reach the surface and rein forcing the primitive epiblast, give rise to a conspicuous dorsal plate, which is shown in Fig. 153 (Dp.). This is from an embryo intermediate between Stages vin and ix, in which eye-pigment 414 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. is not yet formed. The plate is slightly thickeued at its center, where there is an inconspicuous pit marking the point of ingrowth. As the invaginated cells pass into the yoik they degenerate, giving rise to spore like particles which spread iu incredible numbers through a large part of the egg. Some of the wandering cells in this region doubtless degenerate before reaching ihe surface. A part of a similar section is shown in more detail in I'l. v, Fig. 30. The particles vary consider- ably in si/e, stain uniformly and intensely and the yolk about them is granular or finely divided. At a corresponding stage iu the lobster (Homarus americanus), I have observed a large difiiiM- patch of cells which probably answers to the structure just described. In this case the embryo rests on the side of the oblong egg and the cell plate is at oiie end of it, at a point about 90° behind the embryo. This position seems to be quite constant, while in Alpheus the plate is nearly opposite the embryo, at the stage when it is most conspicuous. STAGE X. — EMBRYO WITH EYE-PIGMENT STRONGLY DEVELOPED AND THE POSTERIOR LOBES OF THE GASTRIC GLAND FORMING. All the thoracic limbs and the sixth pair of abdominal appendages are present in a rudimentary condition. The abdomen has grown forward until the tip of the telsou now extends beyond Ihe optic lobes. The embryonic telsou is fringed with seven pairs of seta-, and resembles the larval tclson, except that the median notch is deeper. Seen from the exterior the eye-pigment lias the form of an oval disk. The longitudinal section, PI. XLVIII, Fig. 168, shows most of the important changes \\ liidi have occurred since the last stage. These chiefly concern the eye, the nervous system, and the midgut. The eutodermal pigment cells (retinular cells) of the retinogen have spread inward until they cover its whole inner convex surface (PI. XLVIII, Fig. 167). Near the outer surface of the eye the crystalline cone mother cells (cc) can be recognized, and between the eye and the ganglia of the optic lobe there is a narrow space which communicates freely with the blood sinus ( /,'. iS'.) on the outer side of the lobe. Wandering cells are frequently seen rear this blood sinus, and in the sp.ice between the eye and ganglia flattened cells also occur, which tind their way in thither from the yolk. In the optic lobe another fibrous mass has developed near the eye (Fig. 104-7). In hori/ontal section (PI. XLVII, Fig. 159) the relations of the fibrous tissue of the brain and optic lobes is clearly .shown. In each lobe there is a chain of four fibrous masses united by a stalk of fibers to the anterior or optic swelling of the brain (of). The structure of the brain (Figs. 159, 109, 170) begins to approach iu complexity that of the larva, which was described in the first section of Part n of this paper. The lateral fiber-balls, so conspicuous in the later stages have now appeared (Fig. 159 and LM. xnx, Fig. 174, //'.). They are developed in close union with the large central fibrous mass, which supplies the optic lobes, and probably belong primarily to the antenuular segment. Below this and nearer the middle line there is a less definite fibrous center (///'.) which supplies the antennal segment. With this, the uiso- phageal commissures are directly continuous (Figs. 171, 174 ocm.). The complete chain of ventral ganglia can.be seen in Fig. 10S. This section is not perfectly median, but cuts a fiber-ball of each ganglion. The skin or hypoderuiis is now differentiated from the nerve elements and consists of a thin layer of flat cells. The fibrous masses of the gan glionic chain are also imperfectly surrounded by peculiar cells, the nuclei of which are spindle shaped in section. These also occur iu the brain, and in either case must be regarded as metamor- phosed ectoderm cells, or more probably as intrusive uiesoblast. A transverse section of the nerve cord in the thoracic region is shown in Fig. 1713, and corresponds very nearly in plane of section with Fig. 151 of the last stage. The fusion of the ganglia is now more complete, and the fibrous balls and commissure are relatively larger. (Compare with this, Fig. 176, a section of the thoracic ganglion of the larva.) In the abdominal ganglia the fibrous elements have essentially the same relations, but lie at a deeper level, being separated from the adjoining tissues by at most a single layer of nerve cells. In Fig. 168 we see that the yolk comes in close relations with the nerve cord behind the oesophagus to the eudodermal fold (/) near the point of union of the mesenterou and hindgut. Wandering cells approach the cord and become flattened against it, as already ob- served iu much earlier stages. MKM01K3 OF THE NATIONAL ACADEMY OF SCIENCES. 415 The two divisions of tbo foregut, oesophagus, and masticatory stomach, have tin- relations already described. The wall consists of a single layer of tall columnar cells. In the masticatory division the wall has begun to thicken next the yolk screen, its cells being cuneiform, and the nuclei elongated and crowded below the surface. From the anterior wall, muscle or connective tissue cells extend forward under the brain. HG FIG. 1. — Diagrams of transverse sections through the alimentary tract of an embryo of Alplicux aaulcyi u liich is nearly ready to hatch, to show tin- origin of I In- gastric uland It tun tlie postero lateral lobes of the in id gut. Si (tion I cats tbo hindgut and lobes of the "liver," Section ill the. hinduut. \\-hero it merges into the mcsenti ion. :i:i'. ;/;/% Secoudai y lobules of tii'J ' . UO, li indent ; nKj1 , posti i o lateral lol" s ol inidgllt. The development of the meseuteron can be understood by reference to Figs. 162-165, 168, and 185. The endodermal epithelium spreads by the division of its own cells and by accession of cells from the yolk, both forward over the nervous system and upward against the sides of the body. This is shown in the series of horizontal sections (Fig. 162-165). Fig. 168 which is from an embryo a little more advanced, shows that the endoderm is rising from the nervous cord near its point of flexure, into ti transverse vertical fold. Simultaneously with the upward growth of this ventral fold, two dorsal longitudinal folds grow downward, and finally unite with the ventral fold and with each other, thus constricting oil' from the alimentary tract two lateral pouches, the pri irary lobes of the " liver." The folds grow forward and the constriction proceeds gradually with the growth of the embryo. This process is illustrated by the diagrams (Fig. 1) which were drawn from tin embryo near the point of hatching. The histology of the endoderm as shown in Fig. 173, is essentially the same as in the previous stage. The cells are prismatic, and the. undent spherical, and, as in all stages, filled with numerous nucleoli or chromatiu balls. The cell walls are very delicate ?.ud the protoplasm often contains large vacuoles. US. I''n:. 2.— .Semidia:;raiiin>atic representation of tho alimentary tract and it.-, :i|>poudii};e» in the rilM larva ol .1 ljili< '"-v .s-rtf'/'V/i. Tin- middle 1 lie- nl' the body is also sliinyn. /-'N, I'ure^ilt ; (J(J 1-3, secondary lobules of iiostcrolatei.il lobu of midmil ; lift, hind-ill • /«;/, mid^ilt ; 1113 1-3, an- terior, lateral, and postero- lateral divisions of niidgut; mo, mouth. 410 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. Wandering cells still occur in all parts of the yolk, though in far less abundance than in earlier stages. We find numbers of them moving toward the periphery, or next the body wall to take part in forming the endodertu. The epihlast is conspicuous in Fig. 108 just in front of the optic lobes. This corresponds in position with the dorsal plate (Fig. 153 hc-iui HI i nor (from Beaufort, North Carolina) agree in having a relatively smaller yolk and in hatching as /, be constricting into two parts. The swarm of small nuclei, like that shown in Fig. 13, 1 interpret, as already stated, as arising by a kind of budding of the large nuclear mass and the subsequent constricting off of the buds. It is very interesting to find that in this case, possibly the only case of amitotic cell division in the segmentation of the ovum yet recorded, the indirect division of the nuclei is followed to some extent by a degeneration of these bodies. This lends some weight to the view that indirect division is connected with the senescence of nuclei.* The nuclei of the large endodenu cells lining the mesenteron in the crayfish appear to divide directly, and in this case the process is again followed by dissolution. In may, however, prove that we have here a case of multiple karyokinesis, similar to that which I have recently observed in the superficial cells of the lobster embryo, where nests containing from four to sixteen closely packed nuclei are very characteristic of certain early stages. CELL DEGENERATION IN ASTACUS AND HOMARUS. Homarus americanm. — I find certain bodies in the lobster essentially similar to those which characterize the Alpheus embryo. If a longitudinal section of the egg nauplius of Homarus be compared with Fig. 1_;5, which represents a similar section of a similar stage of Alplieits saitlcyi, we find not a few chromatin balls, but a meteoric swarm of granulated bodies and naked chromatin grains coextensive with the embryo, and reaching a considerable distance into the yolk amid the scattered mesodermic cells, but perhaps most abundant, as in Alpheus, in the. neighborhood of the stomodiiMim. A long nebulous train of yolk spherules and granules extends forward a consider- able distance in front of the mouth, and is especially marked in front of the optic disks. The labruin and the folds of the appendages which contain yolk abound, in these peculiar granulated bodies. In less number they occur in connection with the endoderm cells, which have at this stage extended through a greater portion of the egg and form a series of irregular sacs filled with yolk. These yolk masses, with their surrounding sheet or advancing column of cells, correspond to the endoderm sac of the crayfish. In the latter the peculiar cell bodies also occur. It' one now examines very thin sections under high powers, we find that the granules and the granulated bodies correspond in general to the structures we have found in Alpheus. The chroma- tin grains appear sometimes as naked masses in the yolk ; sometimes they possess a distinct proto- plasmic body. The degenerating chromatin stains either very intensely or faintly and is often vesiculated; that is, it appears as a hollow shell. Under favorable conditions it is easy to demon- strate the fact that these bodies surround particles of yolk, and occasionally they have a crescentic shape, when it can be clearly seen that they are enwrapping a yolk spherule. This vitellophilns * While this memoir was in press a paper was received on Amitosis in tlie Embryonal Knrrloiirn of llir Xi'iirpion, by H. P. Johnson, (Bulletin of the Museum of Comparative Zoology, Vol.xxn, No. 3). Only two instances of direct cell division in the embryo of Arthropods are recorded : that found liy Carnoy ill the ventral plain of Hydrophilua ;mr»s and the case which Wheeler has described for the blastoderm of Blatta germaniea. Mr. Johnson finds lli.it degen- eration does not always follow upon indirect cell division, as iu the case of amitosis in the testieular cells of certain Isopods. 428 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. characteristic is certainly not so apparent iti Alplieus, yet I believe that in this form it is present in some stages, though in a less marked degree. Astacus (?) — I have studied several critical stages in the development of the crayfish, critical at least so far as the appearance of degenerative products is concerned, with an eye to comparison with Alpheus and Hoinarus.* The youngest set of embryos corresponds nearly to Reicheubach's stage E, but differs from it iu some details. Rudiments of five pairs of appendages are present, the two maxillae being seen between the maudibles and the thoracic-abdominal process. None of the appendages, however, are folded. The mouth is seen on a line between the first and second pair of antenna?. The bodies which Reichenbach calls "secondary mesoderm" occur iu abundance in or near the wall of the endoderm sac next the embryo. They also abound iu the yolk under the ectoderm, and are most numerous in the area extending from the optic invaginations to the mouth or slightly behind it. In this respect they recall the distribution of similar bodies in Alpheus and Ilomarus. I wish to call attention to the fact that at this stage none of my sections show a cavity in the endoderm sac, as is represented by Reichenbach (compsfre Taf. vm, 54), and the endodermal yolk segments or pyramids do not always possess completed walls. To what extent this appearance is normal, and to what extent due to the action of reagents, I can not at present say. These eggs were treated with hot water and corrosive sublimate. The eudodermal nucleus is surrounded by a thin layer of protoplasm, which works its way amid the yolk so as to practically surround a pyramidal mass. This strongly recalls the serpentine manner iu which the endoderm cells creep through the yolk iu Ilomarus. Whether these cells in Astacus are simply migrating iu a column or sheet, spreading gradually towards the periphery of the egg, as in the lobster, cannot be decided from the material at my command, but it is a point of considerable interest iu its morphological bearings. The eudodermal cells probably multiply indirectly, but I saw no nuclear figures in my sections, and they appear also to divide directly, independently of the yolk pyramids as Reicheubach has described, giving rise to the chromatiu balls and granulated elements (compare Fig. 20) but, as pointed out above, this appearance may be very deceitful. .This process is most marked in the endodermic area noticed above, underlying the anterior half of the embryo. Here we see great numbers of the bodies of varying size, both within and without the domain of the endoderm cells. They closely resemble the vitellophagous elements which I have described for the lobster, and possibly they attack the yolk in a similar way. Where they are thickest the yolk is comminuted and shows traces of profound chemical change. In the midst of the altered yolk one can discover very faint outlines of vesicular bodies which exhibit but slight reaction to the stain. These I regard as degenerated cells. The next stage of Astacus which I have studied corresponds nearly to Reichenbach's stage G. Eight pairs of appendages are present, and there are rudiments of a ninth pair. The first and second maxillae appear as distinct buds, while the third pair of maxillipeds is represented by a proliferating cell area only. The extremity of the abdomen is not bifid. In the central part of the endodermal sac there is a coagulable fluid which comes iu close rela- tion with the posterior end of the embryo. The endodermal cells laterally have quite or nearly reached the ectoderm, while dorsally they fall a little short of the surface. The yolk within the (tontines of the endoderm has an irregular, pyramidal, or radial cleavage. Centrally the yolk blends with the serum-like fluid, in which occasional granules or balls of chromatiu may be found. Small spherical elements (like those represented iu Fig. 18, «, fc, c, or A-, Ar,1 Fig. 20), containing a single chromatiu ball or several balls, occur not only in the yolk underneath the ectoderm and in the vicinity of the endodermal nuclei, but also in the central yolk of the endoderm sac, at various levels below the endodermal nuclei. This is a point of some iuterest in connection with the fate of these bodies. They wander not only peripherally but centrally. Rarely we meet one which is three or four times the average size, having a small chromatin spherule iu its center. In later stages they are present in far less numbers. * For tlio opportunity of studying tlu< crayfish ilovelopiuent at this time I am indebted to the kindness of my friend, Dr. William Patten, who sent rne a number of important Nta.urs collected at Milwaukee. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 429 Reicheubach thus summarizes his observations on the "sekundiire Mesodermzelleu :" Die fraglichen Elemeuto sind als Zelleu zu deuteu, dereu Kerno uicht iuiuier die BesohaffeDheit gowohnlicher Zellkerue liabou, diesclbo aber friiher odor spiitor orlaugen (Fig. 'M, in, in-, Plato xxvn). Sio ucliiuen ibrou Ursprung iuuorhalb dorjonigeu Eutoderuizellen, welche dio ventralo Waud des Urdarmaackchens zusamiiiensotzeii durch eino niiher zu erforgchendo Art eudogener Zellbidung, bei wolcher die in dor Mehrzahl in don Eleineuteu des Entudorins vorhandeuon Kt-rno cine wichtige Eolle zu spielen scheiuen. In don doin Stadium D vorangehenden Entwicklung- speriodou bat jodo Eutodermzelle uieist nur einen Kern ; diea trifft auch uoch zum Thcil fiir Stadium D zu. Bald ver- meliren sich :iber die Entoderiukorne ganz erhoblich mid oudlicli beginuen die sekundiirou Mesodermzellen aufzutreteu. Weiin eine grossore Zahl der sekuudareii Mesodermzelleii in den Entodermelementeu liogeu, so scbeint das Kernma- terial verbraucbt zn sein. Es wandern nun aller Wahrscheinlichkeit nacb dieso Zellun, deren Kerue anscheineud noch in dor Metamorphose sich befinden, aus deui Eutoderm aus uud begebou sich uuter die Embryoualanlage. Die tetreft'eudou Contouren. des Entoderms lassen oft uoch Spuren dieaer Wanderung erkeuneu. Ob sie wirklich aktiv aiiswaudern odor anch ausgestosson werdeu, ist nicht festzustelleu gewosen. Sie begobeu sich nun unter die ubrigcn Mosodermzelleu uud Bind bald uicht mehr von ihuen zu uuterscheidou. Aus dieseiu Grinul fiihrto ioh liir sie deu Nameu "seknndare Mesodertnzelleu" eiu, wiihrend die altereu Urmesodermzelleu als primilre bezeichuet werdeu. Da die letztoron dio Tendenz zeigen, zu kompakteren Massen zu verwachsen, so darf man wohl vermuteu, dasa die sekuudiiren Mesodermzellen die Blutzellen liefern werden (54, p. 36). It is interesting to notice that in Alpheus, Astacus, and Hotnarus degenerating cells appear in greatest force at about the egg-nauplius stage, and from that time ou their numbers begin to waue. In Astacus, Eeicheubach first noticed the " sekuudiire Mesoderuizelleu " in stage D (that is, when the optic disks, the thoracic-abdominal plate, aud the mandibles are outlined), which nearly corresponds to Stage IV of Alpheus. In stage U the bodies in question arc most abundant under the optic disks (Kopflappeu) and in the region of the upper lip, but become more generally dis- tributed in the egg-nauplius. According to Reichenbach, " gastrulation" takes place after the optic disks are formed, but unfortunately his paper is incomplete at a very important period, namely, from the late yolk-pyra- mid stage of segmentation, when the protoplasm is at the surface, to the time when an embryonic disk or plate (Eutodertuhiigel or Entodermscheibe) has been formed. It is impossible, therefore, to follow the history of the so-called " white yolk elements." He says of the latter : • Sie bestehen aus protoplasmatischer, feinkornigor Substauz uud euthalten vacuolenartige Einschliisse, die ilmeu ein schanuiiges Ausseheu gebeu ; ich babe sie als woisse Dotterelemepte bezeicunet. Sie liegeu entweder dicht unter dem Blastoderm oder im Centrum des kugligeu Eies uud verscbwiudeu sebr bald (Op. oit., p. 7). According to my view these bodies ^correspond to the vesiculated elements (m, Fig. 20), and both represent cells in process of dissolution. If Figs. 18 aud 20 are compared (the latter being a copy of Reichenbach's Fig. 67) we will find a striking correspondence between these peculiar cell products iu both Astacus and Alpheus, a correspondence which is even more marked when the comparison is made with the lobster. Reichenbach emphasizes the statement that naked balls of nuclear material never occur free iu the yolk outside the endodermal sac. In this yolk they always have a "cell body" as K, Fig. 20. It is an easy matter, however, to distinguish naked masses of chromatin, minute spherules, or smaller granules iu the yolk of both the lobster and Alpheus. My studies of the lobster are not yet completed, but from the observations which have already been made I draw the following conclusions: In Alpheus, the lobster, and the crayfish similar bodies make their appearance at nearly similar times and play a similar role. They are derived from all three layers of the germ, and in Alpheus minor degenerative products make their appear- ance in the segmentation stages. They tend to break up aud ingest the yolk and to produce iu it a chemical change, possibly in order that it may be more easily assimilated by the other embry- onic cells. Having performed this task they degenerate ; they are converted into a substance resembling yolk aud function as nutritiou. That any play a formative role, giving rise to blood cells for instance, as Reichenbach supposes, there is no direct evidence. The vitellophagous func- tion seems to be iu abeyance in Alpheus, but iu all cases the yolk is comminuted and chemically changed in the neighborhood of these bodies. Nusbauin (45), following Moriu, believes that the "white yolk elements." arise from the segmentation nucleus and migrate to the surface of the egg; that they give rise to the "secondary mesoderm." which are taken up along with the yolk by the anueboid, endoderinal cells. This is reversing the account, and, iu so tar as the origin of the "sec- ondary mesoderin" is concerned, it is not supported, so far as I am aware, by a single observation. 430 MEMOIRS OF THE NATIONAL ACADEMY CF SCIENCES. Ishikawa (27) finds in Atyephyra after the close of the invagination stage, certain proto- plasmic elements under the ectoblast, which he thinks may correspond with the" white yolk ele- ments'' just referred to, and he also identifies "secondary mesoderm cells," but does not trace their origin or function. They are " small granules," easily stained by logwood solution, and some are of considerable size and have a clear cell outline. "These are mostly aggregated in the cephalic region between the involutions of the ectoderm cells, but are also found in all places." In time of appearance and ill their position, he says they seem to correspond to the " secondary meso- derm cells" of Astacus. This short notice with his figures leaves little doubt that these bodies are similar to those just described in Alpheus and Homarus. Fig. G2 of his paper represents a longitu- dinal median section of the egg-nauplius, and may be compared with the same stage of Alpheus (Figs. 104, 105), with respect to the general character and appearance of the degenerating cells. I have noticed similar nuclear fragments in the egg-nauplius of a crab (Fig. 113, PI. XL), and Lebedinski (34) has described "secondary mesoderm" in the embryo of the Mediterranean, sea crab, Eriphia njiini/'roiis. According to this observer they are found in all stages from the "gastrula" on, to the egg-uauplins ; they are derived from ectoderm, and probably give rise to blood cells. In the stage with one pair of maxillipeds these elements are in active proliferation : M;iu iindet, die Ze.lleu desselben bieten verschiedeue Mouieuto and Zustande des Zerl'allens dar; dieses Zerfalleu iler /.ellon steht in geuauou Zusamiuonhange mit der Eutstehnng dor Blutkiirptr. lie further says : Ueber dio Bildiing des Blnteskanu ich nichts bestimmtes luittheileu. Ini Stadium drsi'iNtcii 1'aars kiefori'iisscheu siud die ersten Blutkoiperchen vorhandou, welclie zuiu ersteu Mai iui Beroiche des Herzens vorkommen, wosiuh auch am friihesten das sekunduro Mesoderm riickzubilden beginut. From these quotations it appears that the " secondary mesoderm " shows .signs of degetiera- tiou, and its conversion into blood cells is au unverified inference. It seems more probable that the structures in question correspond with similar bodies already noticed in Alphens, riomarus, and other Decapods, and that in all cases they have to do primarily with the dissolution and not with the construction of cells. \Vheeler (67) in his careful paper on the development of the Cockroach and Potato beetle (Blatta germanica and Doryphora decemlineata) describes au interesting case of the decomposition of nuclei, which bears a close analogy to what takes place in Alpheus and probably also in Astacus. In Doryphora two masses of endoderm are found, one under the stoinodaeum the other under the caudal plate. At both these places numerous cells which originate in the endoderm pass into the adjacent yolk and disappear. The process of dissolution is described as follows : The karyochyletua becomes vacuolated, probably with substances absorbed from without, to judge of the larger size of some of these nuclei, while the chromatin ceases to present the threadlike coil and becomes compacted into irregular masses between the vacuoles. Finally the vacnoles fuse and the masses of chromatin, formally numerous, agglomerate to form one or two largo irregular masses which usually apply themselves to the wall of the clearly vesic- ularnucleus * » * In the last stages seen the masses of cbromatiu lie between the yolk bodies, all other portions of the nucleus having disappeared. They still take the deep red stain, but finally become comminuted and disappear in the iutervitelliue protoplasm. The vesiculated elements recall similar bodies which appear in Reichenbach's plates. Thus the element t, Fig. 88 of Wheeler's paper, where the chromatin is applied to the walls of the nucleus, strikingly resembles nucleus i, Fig. 20 (see this paper), where the chromatin is similarly disposed around the wall of a vacuole. Bruce (10) figures certain yolk cells undergoing what he considered to be endogenous cell division in an advanced embryo of a spider, and compared it with the endogenous cell division which Reichenbach describes as taking place in the endoderm cells of Astacus. The disintegration which has been attributed to the leucocytes of the mammalian blood affords au interesting comparison with the phenomena which have been described for the Arthropod embryo. Howell's careful observations (25) support the view that the multiuucleated leucocytes are disintegrating cells. " The leucoblasts enter the lymph stream, and Eventually reach the blood as unicellular leucocytes." Here they undergo changes, acquire amojboid movements, while the nuclei elongate, become constricted, and finally fragmented. " The mnltinuclear stage * * * is probably followed by a complete dissolution of the cell." Howell adopts the highly reasonable MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 431 view that the unclear fragments persist for a while iu the circulation as the blood plates, and considers it probable that the latter take, some part in forming the paraglobulin of the blood. If the blood plate is then a degenerate body, it may be compared to the spore-like masses of chro- matin, which are discharged from the disrupted cells in the lobster or craytish embryo. VII. — THE ORIGIN AND HISTORY OF WANDERING CELLS IN ALPHEUS. The wandering cells iu Alpheus have a triple origin, from the blastosphere, from the invagina- tion, and from the thoracic-abdominal plate. Those which arise from the blastula at the close of segmentation are, perhaps, the representatives of a primitive endoderm. Following the iuvagina- tion, a thick pad of cells is formed, the ventral or thoracic-abdominal plate. From this plate a, general migration of cells occurs on all sides into the yolk (Pis. xxxn-xxxv). While there is a great tendency to migrate to parts underlying the embryonic area, the cells nevertheless wander to all parts of the egg, even to those most remote from the embryo. The first of these wandering bodies which originate from the blastosphere have been called "primary yolk cells." The latter classes may be called "wandering cells." Since these classes can not be distinguished after a certain period, I refer to all cells which move about iu the yolk and have no direct connection with the thoracic-abdominal plate, and the parts of the embryo in front of it as wander ing cells. I have been somewhat at a loss to find a suitable term for these bodies, since there are obvious objections to the use of "yolk cells" or "yolk nucleus." Where these terms have been employed in the present paper, they must be understood to refer to the wandering cells which have been defined above. The term "embryonic nuclei" is used for convenience merely to discriminate the remaining nuclei of the egg from those of the wandering cells. The object immediately in view is to determine the fate of these wandering cells, to ascertain what formative role they play while the mesoderm, and more particularly the endoderm, are being differentiated into definite cell layers. In the following account, cells which have parted all con- nection with the thoracic-abdominal plate and have entered the yolk are enumerated as wandering cells. In an earlier part of this paper I gave an account of the origin and supposed fate of the wandering cells, the general conclusion being that iu the early stages (Stages m-V) they pass from the yolk to the embryo and to the extra-embryonic parts* of the egg, and contribute to both niesoblastic and entoblastic structures. A number of friends to whom I showed my sections objected to this interpretation on the ground that these wandering cells could be regarded with equal probability as originating, in souie measure at least, in the opposite way, that is, as budding from superficial cells not concerned with the thoracic-abdominal plate, and migrating into the yolk. A careful study of successive stages would not support this idea, but the objection could not be satisfactorily answered, and neither view could be readily proved. I therefore undertook a renewed aud more precise study of the wandering ceils iu Alpheus, and I think that their fate has been definitely settled. The number of wandering cells which occur iu the yolk, and the number of " embryonic cells" (that is, all the other cells of the egg) have been enumerated in five different stages, including seven different embryos, from the period of delauiination at the close of primary yolk segmentation to the early egg-nauplius condition (Stages II-V.) This covers the most important period so far as the wandering cells are concerned. The rate of increase of both wandering cells and embryonic cells has also been determined for the successive stages, aud the data are given iu Table i. * There is a certain convenience in thus referring to the ombryo proper aud to the less differentiated regions, while it is understood that all the cells constitute the embryo. 432 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. TABUE I. — Showing the number of nuclei in yolk, and the number of other " embryvit ic >i m-lei," and the relative increase and decrease in these bodies from the close of yolk-segmentation to tltv eyy-nauplius period. Stage. •8 a - jjj.2 a:. si- SM O a "jj B.S 3.2 s S3 °'S, a"q" • t, M 1C tt ° M ft Sc [_, »*J ^ a a a D 's 3 a *>£. O 0 (. V-. a ®*M ""S 0 a t. o •- : o g^g a-s 0 ° .a« «~ aj— - ^li a' i' i"-> - . : |So eS B t^ 2^.2 - - - 2 § S a g*"? & I*, o B jjj »— ( o a H o fl H a^ t-H O Otu O o goa i~i II. Dblamiuation (Figs. 38-45, 46, PI. XXX) *34 2S4 318 )Fi"s 49-55 PI \\xi »37 3 421 137 458 140 8 'i"i II. Invagmation?,j \Fi5s 4t)-r)ti' PI xxvi "37 430 146 467 149 3 34 III. Optic disks (Pis. xxxn, xxxm) 199 162 t736 306 \ 935 468 si 4" IV. Firstantenuic,u]audibles(Pls. xxxiv, xxxv) 240 41 t915 179 1 155 "20 17 "0 (o) Figs. 34,Pl.xxix;l 101-105, 107, PlA 199 41 t2 989 2 074 :f l-s 2 033 •'1 69 V. Early egg-nau- XXXIX ) plius (6) Fi^s 34 PI xxiX'J 101-105, 107, Pl.^ 128 112 t2 230 1 315 2 358 1 2Q3 871 59 xxxix j * Primary yolk cells. t Number obtained by method described bolow. The distribution of the wanderiug cells and of the embryonic uuclei is also illustrated by a series of curves, constructed in the following manner : Each vertical row of the smaller squares corresponds to a section of the ovum (Sections Nos. 1, 2, 3, etc.), while each square in this row represents a single nucleus. The number of nuclei in a section is shown by the number on the line where the curve makes a bend or intersects the middle of the lower side of a square (nuclei Nos. 1, 2, 3, etc.). Thus where the descending line of the curve in Fig. 5 stops in section No. 2, making an angle in the fifth vertical square, counting from the upper base line, this implies that in the second section of the egg, corresponding to this vertical area, there were five nuclei which did not appear in the following or preceding section. The number of nuclei were determined in the following way : Camera-lucida drawings were made of every section of a given series on thin paper and each nucleus was marked. Then, by superimposing upon each drawing the drawing of the section immediately following, every nucleus new to that section could be determined in the early stages with absolute certainty. The number of primary yolk cells and wandering cells were thus counted in all stages. In the older embryos (see numbers marked with dagger in Table i, Stage m-v) where this method became impracti- cable with reference to the total number of embryonic nuclei, their number was estimated in a difierent manner. The uuclei appearing in each section were counted and the total number of nuclear sections was thus obtained for the whole series. Then the percentage which the actual number of nuclei in the egg bore to the total number which appear in sections could be determined approximately by the method described above applied to a number of lateral sections, that is, by actual count of uuclei in a favorable part of the series. The percentage which the actual number of wandering cells bears to the total number appearing in section could be exactly determined and compared with the former estimated percentage. The percentage which involved the wanderiug cells was in the average of all stages a trifle the smaller, showing that the yolk cells were, on the average, a little larger than the other embryonic nuclei. In Stage n before invagiuation the super- ficial nuclei are the larger, while after invagiuatiou the difference is at first very slight indeed. In Stages in and iv the wandering cells are markedly the largest in the egg, while in Stage v they either equal or fall slightly below the size of the other embryonic nuclei. This method presupposes a perfect series of sections of uniform thickness. These conditions were approximately fulfilled. The egg (imm, or ^ inch in diameter) was cut, on the average, iuto 57 sections, each section being yirmm in thickness. The size of the egg, neither too small nor too large, rendered this species (called throughout this paper the Bahaman variety of Alpheus heter- most favorable for study so far as technical difficulties were involved. MEMOIRS OF THE NATIONAL ACADEMY CF SCIENCES. 433 STAGE II. — Close of yolk segmentation — Formation of yolk cells, followed by imagination. — A surface view of this egg is given in Fig. 47, PI. xxx. The curve (Fig. 5 of text) shows that the blastodermic cells are distributed very uniformly. In other words the embryonic area is not as yet marked oft'. The distribution of the primary yolk nuclei, of which there are exactly thirty-four, is shown in the constructed figure (Fig. 3), which represents the egg as it would appear if the yolk were transparent and the nuclei opaque. The distribution of these nuclei through the egg is given more completely by the curve (Fig. 4). The only questions which need detain us in this stage are, how do the primary yolk cells arise and from what part of the surface do they come ? Karyo- kinetic figures, which abound among the surface nuclei, ought to furnish au immediate answer to the first question. In this egg no less than sixteen nuclei are met in various phases of division, fif- teen of these belonging to the superficies and one to the central FIG. 3.— Diagram ..r,-^ in ,i,hn,,m;iti,.n , . . sta^i- constructed from serial sections, portion of the yolk. Clusters of two and rarely ot three nuclei also ^.^ .iU tl)t. iirim.(ry yiilk , ,.,,s F,ir occur at the surface, showing that cell division is active. In every details, *«• Tabi,. 1. st^e n (iMami.m. case the cleavage is radial or perpendicular to the surface, and in no instance have I seen au unambiguous case of delamination (v. PI. xxx). It is possible, however, FIG. 4,-Curve construed from serial sections, slum-in- tlio distrilmtiun of tin- primary Yolk nuclei in tbe ens represented l,y Fiji. •'•• For further details, compare Table i, Stage II (Delamination). ,i = Anterior; 1'= Posterior. ;A//V< •H-H-h : FIG. 5._Cnrve showins tbe distribution of nuclei ,,t tlie surface ttbat is. nuclei of the embryonic cells, exclusive uf primary yolk . .11-, of the egg represented by Figs. 3 and 4. For details, sec Table l. Stajze II (Delamination ). S. Mis. 94- 28 434 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. that the primary yolk cells are formed iu this way rather than by emigration, and that my failure to detect the actual process is due to the fact that I did not sec- tion exactly the right stage, the egg shown in Figs. 38-45 being a trifle too old. In Homarus the primary yolk cells arise hv delamination, as 1 have already shown in a preliminary paper iu the development of this form (!'.'!, Fig. .">). Sections of this stage show conclusively that the primary yolk nuclei do not come from any one point on the surface, but that the majority of them may come from a restricted area of the egg. In Fig. 3 about two-thirds of the nuclei present are con- fined to the lower (ventral ?} surface of the egg. They are in various degrees of progress from the surface toward the central parts, which the majority have already reached. The formation of primary yolk cells is followed by the in vagination and ingrowth of certain cells at the surface. The his- tology of the embryo at this phase is given in PI. xxxi, and Fig. (> (of text) is constructed from the entire series of sections to show a" "ie primary yolk nuclei present. The plane of the paper (sup- posing the drawing to represent, a sphere) nearly passes through the point of invagination (in.). In order to test the accuracy of the method, two eggs of this stage were studied («, n and b, u ot Table i), aixl the results show a remarkable agreement. Thus there are exactly thirty-seven primary'yolk nuclei in each egg, and the total difference in the number of embryonic nuclei in the two eggs is only nine. Curves were constructed to show the number aud distribution of yolk nuclei and embryonic nuclei in both eggs, and the two are introduced here because of the striking simi- larity. Figs. 7 and S are constructed from the egg seen in Fig. G(n, n, of Table i). Figs. 10 and '.) represent corresponding curves constructed from the second egg (u, /*). The two sets of curves tell exactly the same story in each case, and it is not necessary to dwell upon it. I1'!'' (>- Pia^runi nl r-j-r in m\ ;i^in;il mi st.'ijjr-, rmistriirtril i'roiu serial N.I linns, t. sliow ;ill tin- jiriinary yolk Jiinl.-i pi. s. nl FOr (IftjIlN M|' lt,is |'M.ri S(,,. -|-;,|l]f I (II. "I Cnvagiiiation. In ]'«'inl <»(' in\ ;i^in;itioji tii'iU'lx iii Llie j)l;iiH- nt' thr pa JUT. t / £ : t/ , r> 4 . i •1 •, 6 ' A /\ \ ' < \ './ •> \l / L Fio. 7.— Curve coDstruotcil from serial sections, sli (See Table i, Invagination stajti-, u, a.i wiu^ (lie di.^trilmtion of tbe primary yolk nuclei iu. the e"" represented bv Fie 6 The bulk of the primary yolk nuclei are placed near the center. Of the thirty-seven nuclei in «•, ii, twenty-two are iu the ventral (?) hemisphere of the egg, aud fifteen in the dorsal. In egg ft, II, twenty-one nuclei are situated in the ventral half and sixteen in the dorsal. Thus these yolk nuclei incline toward the ventral side of the egg, aud hence, as already inferred, are probably derived in great measure from that part which corresponds to the embryonic area. The curves showing the relations of the embryonic nuclei (Figs. 8 aud 10) read from end to end of the embryo (posterior to anterior), the sections being transverse to the longitudinal axis. The greatest depression naturally occurs in the region of the thoracic-abdominal or ventral plate (Ab. P.), near the center of which is the point of iuvaginatiou. In front of this there is a more extensive, but less depressed portion, corresponding to the embryonic area (E. A.). The number of cells entering into the ventral plate at this time are shown in Table n. MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. 435 4S « SO A Abp. EA. FUi. g.— C'urve showing distribution of all nuclei, exclusive of primary yolk nm-loi, in tlie same egg as represented by Figs. 0 anil 7. (See Table i, In valuation stage, II, a.) A, Anterior; P, Posterior; Abp, Ventral plate; KJi, Embryonic area. r * I - ,f "1 V i /5 /.(' /* 1 '/ •a a ^ y g 7 a a * >/ ,' ' [3 w w / ^ i * >J '•:' Af £ TV s\ \ - r A / * 4* 3 4 •r V - - i F &<• G. -/ 9 0 ~ Ci A in a re si 01 rii I 'g ti e i lis -f tr bi (V iti /(-, on <> Ft /-( In /•J P /•, rir 1:1 ry /,(• y< >ii< t 11 •le y i 'J1 n ',1 ti e in 'ft va gii M ia j/ .io 7"1 ii n aU "•i.j Sl If \ u Ig f 7 t If Vi w . I *• I, ;/ b, <• X 7.) 4 u C, e n iT_ /\ \ A -f /\ ^ r s T -t- 3 .; \/ y ' | f ,/ 1 ft i « 1 a n T /\ n 1 H 1- - r |\ / \ / \ P 1 V \/ V 1 A ,. | LL H ,. 1 | \ \ r 1 \ i J J f. f 1 / ^ \ y ,, / V S , I M , \/ »6p. f/1. Fie;. Hi.— Curve showing distribution of all nuclei, exclusive of primary yolk nuclei, iu the same egg as represented by Fig. 9. (See u, '•. Table i, and compare with Fig. 8.) A, Anterior; P, Posterior; Abp, Ventral plate; EA. Embryonic area, 436 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. TABLE II. Stage. Nuclei ;it siiil:if.' <>l Al. P. Nuclei in Ab. P.. be- low sur- face. Total num- ber i.rnurb'i in Ab. P. T'riin;n \ yolk nuclei. Remaining t-iiibi yonic nuclei. Total num- lier of nuclei of egg. ji. I'egion nl' \ eutral [date; O 1>, liegion of optic ilis. s. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 4M7 Table i shows the remarkable fact that the percentage of increase of wandering cells, which in Stage in was twice as great as that of other embryonic nuclei, has now dropped until it is actually less than the latter. This means that the wandering cells are multiplying less rapidly or that their numbers are being depleted. The, relative increase of yolk nuclei from Stage II to in, and from Stage in to IV, namely, 81 per cent and 17 per cent (showing a marked falling off), is not specially significant, since these numbers depend largely upon the ago of the embryo or time which elapses between successive stages. But this element does not enter into the relation which exists between the relative Increase of wandering and embryonic cells in the same egg, expressed by SI per cent and 41 per cent, respectively, in Stage in and by 17 per cent and 20 per cent in Stage IV. The curve constructed from Stage iv (Fig. 11) shows the wandering cells much more widely diffused through the egg than at any earlier period. The thoracic-abdominal plate is no longer sharply marked off from adjacent parts, and a considerable number of nuclei underlie the optic disks. In this egg there are thirty-seven nuclei in various phases of karyokinesis (Figs. 70, 75); two belong to the yolk, three to the deeper cells of the thoracic-abdominal plate, and the rest to the surface cells of the embryo. In the optic-disk region a very few cells appear to be delainiiiating; the rest are cases of radial division. Cell disintegration has not yet become a disturbing factor in the enumeration of cells. There are but two nuclei in the egg, one in the yolk, the other in the deeper part of the ventral plate, which show any traces of dissolution. The above facts lead to the conclusion that the wandering cells are being rapidly depleted at this time, that they leave the yolk in considerable numbers and attach themselves to the growing embryo and to the extra-embryonic surface of the egg. They thus warrant our interpre- tation of a cell like yc, Fig. 70, or yc1, Fig. 71, as a migrant from the yolk. While the wandering cells are being rapidly depleted, it does not of 'course follow that they never receive any recruits from any part of the egg without the limils of the ventral plate, and it requires no very elaborate calcu- lation to show that wandering cells are making early contributions to the mesoblast, but the study of individual sections and the facts brought out in Table i, prove beyond a doubt that the only cells which enter the yolk, up to this time, arise from one of the three sources already named, the blastoderm, the invagiuation, and the ventral plate. STAGE V. — Et/f/-Xauj>liiin. — The early egg-nauplius stage is represented by two individuals, one (V. rt, Table i) cut in transverse and the other (v. b) in longitudinal vertical planes. Respect- ing the wandering cells we now notice: (1) that their numbers have markedly decreased; (2) that they are far more widely and evenly distributed; and (3) that many are close upon or in contact with the embryo or with the general surface of the egg. In egg v, b (Table i) the number of yolk cells is only one hundred and twenty-eight, consid- erably less than are present in Stage in, a decrease of S7 per cent, while in the other egg the decrease is 21 per cent. On the other hand, the rate of increase of embryonic nuclei is greater than at any previous stage, 59 per cent in one egg and 60 per cent in the other. That this is not explained by a large interval of time existing between Stages iv and v is shown by the fact that during the period (Stage IV and Stage v, l>, Table I) the total number of nuclei of the egg has scarcely more than doubled, while the percentage of increase of embryonic nuclei has more than trebled. Between Stages n, a, and in, on the other hand, the number of nuclei has nearly doubled, while the percentage of increase of embryonic nuclei has risen from 32i to only 42. How is this very rapid increase iu embryonic nuclei and coordinate decrease in wandering cells explained in the egg-uauplius stage? The conclusion reached iu Stage IV applies here also, with a certain restriction. The problem is now not a simple one, since perturbations caused by the disintegration of nuclei appear to some extent in this stage. The diminution in the. number of wandering cells is now due to two causes, to cell disintegration and to the gradual subtraction of cells from the yolk by emigration. Disintegration of cells occurs both iu the yolk and in parts of the embryo. It is perhaps most marked near the line of contact of the yolk with the embryo. Yet the embryonic cells are meantime making a rapid net gain. We are thus warranted in our conclusion that the wandering cells, which spread far and wide through the egg, play a formative role in development, to a large extent at least. This conclusion is rendered certain by the changes which ensue between Stages n and iv, already noticed. The 438 MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. percentage ot increase of wandering cells between Stages n and in is double that of the embry- onic cells. Between Stages m and iv the increase per cent of wandering cells is less than that of embryonic cells, and up to this time cell disintegration is ruled out as a disturbiu- factor aiv Fii;. 11, uinl fur details, see Table The second fact which was pointed out as characteristic of this stage, the distribution of the yolk nuclei commensurate with that of the yolk itself, also points to the conclusion already r^hcd This is well shown by two curves (Figs. 12, 13). In curve 13, which is constructed from a series ... , b 6: F,,, 13._Curve abowins tl,«. dlriritatfcn ,,f wan,lerin!t cells in Stage v, a. (v. Table ,., A, Anterior; P. Posierior. old reon * abundant in the thoracic-abdomiual. fold egion and in hat which answers to the future heart. We also notice the forward extension o he fuHry CellS.'ey°m ^ anttrior ed*e of the oP«e ^bes. In cut 12 the lateral extension the nuclei on e.ther side the embijo is well shown. The mouth is involved in section No 28 MEMOIRS OK THE NATIONAL ACADEMY OF SCIENCES. 439 and the embryonic area is included between sections 14 and 41. The marked peripheral distri- bution of the migratory cells is very significant. There seems to be a general movement of these bodies to all parts of the supertic.ies. What is the ultimate fate of those cells which wander out to the surface of the egg? Fig. 34, PI. XXIX, represents part of a. sect ion of the extra-embryonic surface at this stage. Here is undoubted ectoblast (Ep.). and cells (Y. C.), which have undoubtedly come from the yolk, are pressing against the surface. In a single egg one may meet with twenty or more such cells, as well as with cells in various degrees of proximity to the surface. Do these cells (Fig. 34, Y. C.) eventually contrib- ute to the mesoblast or ectoblast of the embryo? This question can not be answered by direct observation, since, it is clearly impossible to follow the fate of the individual cells. From all the evidence which I have gathered I conclude that the wandering cells give rise to mesoblast — muscles of body wall, connective tissue, and blood corpuscles, and, later, to the definitive entoblast liniug of the midgnt. The part which the primary yolk cells play can not be decided, nor can it be deter- mined whether degeneration is more characteristic of these elements than of the invaginate wander- ing cells. In a lobster's egg at the delamination stage, equivalent to Stage II, Table I, I find 213 nuclei present. Eleven of these belong to the yolk. Upwards of twenty, two of which are yolk cells, are in process of division. Of the eighteen superficial or peripheral cells which are in karyokinesis, two are. dividing horizontally or delaminating. In the lobster the primary yolk cells degenerate, in part if not wholly, at a very early stage, as already stated. In the egg-nauplius phase it was noticed thaf wandering cells settle down upon the ectoblastic bands out of which the nervous system is subsequently developed (Fig. 127, Y. C.). In this case also it is impossible to ascertain with certainty the fate of such cells. They might form ectoblast or mesoblast, but it is more probable that they contribute to the latter layer only. It is evident that the history of the wandering- cells is largely the history of the raesoblast and entoblast. The mesoblast,* where it has been. studied in Decapods, as in Astacns, is found (54) to originate in certain swollen cells in or near the anterior margin of the "blastopore" or pit. From this primary mesoderm cells are budded off, which extend forward in a more or less continuous sheet over the ectoderm. It is possible that in Alpheus the wandering cells may serve as a means of a precocious development of the mesoblast and entoblast. In connection with this idea, it is interesting to recall the fact that in high temperatures of the tropics the developmental stages are passed very rapidly. At Woods Holl, Mass., the late egg-nauplius of the lobster, Homar-ux americantis, is from fourteen to sixteen days old, while a similar stage is reached by Alpheus saitlcyi at Nassau, N. P., in about seven days. VIII. — THE DEVELOPMENT OF THE NERVOUS SYSTEM. The nervous system can be referred back in the embryo to an early stage (Figs. 58, 62, 68), when V-shaped ectoblastic thickenings unite the optic disks to the thoracic abdominal plate. The intervening space is gradually encroached upon until the optic disks are completely bridged by a dense sheet of ectoderm. There is an apparent concrescence of the limbs of the V, and in the egg- uauplius (Pis. XLI, XLII) these thickenings form a pair of more or less closely united cords, which are separated on the middle line'by a median longitudinal furrow. The shallow furrow is formed by the swellings of ectoderm which correspond to the future ganglia, and extends from the suprace- sophageal ganglia to the segment of the first maxilhc. The nervous system of the egg-nauplius is not differentiated from the general integument, and the ectoderm is still a single layer on the middle line in the maxillary region (Fig. 121), while at the base of the first pair of auteuiue (Fig. 116) it has the appearance of an elliptical plate in trans- verse section. * Weldou, whose paper 011 the germinal layers in Crangon has been referred to, Bays truly that the difference between iuvaginated cells is not Buflieieut to enable one, to say that certain cells are endoderm and that others are mesoderni, but be designates as eudoderui all cells which are derived from the iuvagiuatiou, and restricts the origin of the mesoderni to the lower layer cells of the ventral plate. Judging from the evidence which has thus far been presented, the cells which be bas marked endoderm, lying against the embryo and near the folds of the appendages, are in my opinion to be interpreted as mesoblast. The thoracic-abdominal thickening is composed of a pair of concave "neuro-muscular" or ventral plates, which correspond to tbe single plate described in Alpheus. 440 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. The aiiteunular ganglion is in close union with the optic ganglion and unites also with the auteuual ganglion which lies along the sides of the stotuodamin, extending slightly behind it. The histological differentiation of the nervous elements is not very considerable at this stage. In the ectoblastic thickenings, out of which the nervous system is formed, we can distinguish three kinds of cells: (1) the superficial cells, (2) the central cells, and (3) the accessory cells which come from the yolk. These are best seen in a section of the antenuular ganglion (Fig. 110). The outer cells (1), which form the integument, possess very large granular nuclei. Some of these, on either side of the middle line, can be distinguished beyond doubt as the nuclei of those large ganglion cells so characteristic of later stages (see Figs. 146, 147, 169, 170, 191, g. c.). They possess a more or less definite cell body of a round or oval contour. In preparations this is fine grained and, like the nucleus, stains but feebly. The weak stain of the nucleus is due to its very fine and loose chromatiu reticulum. Karyokiuetic figures attest to the multiplication of these cells (Fig. 191), and it is highly probable that they give rise to similar cells which occur in both larva and adult. But what is remarkable in the earlier stages is their enormous size and their peripheral position. Keichenbach calls attention to similar cells in the embryo crayfish, which also arise from the outer layer of ectoderm cells and for a long time help to form the outer wall of the body. It is hardly probable, however, that these cells are relatively more highly differentiated in respect to their ultimate function than any of the surrounding cells which take part iu the nervous system. The central cells (2) are the ordinary ganglion cells of the cortex which inclose the fibrous masses. They have smaller and less regular nuclei, which stain very heavily. Cell boundaries are entirely effaced, and the cell protoplasm is reduced to a minimum. The accessory cells (3) rest on the dorsal surface of the thickening, and represent indifferent or wandering cells derived from the yolk (Fig. 116, cts., PI. XLI, mes.). I am not prepared to say that any of the ultimate nervous cells are derived from this source, but I am certain that cells migrate from the yolk and attach themselves to the ectoblastic thickening out of which the nervous system is formed, and that they multiply by indirect division. It is probable. that the connective tissue sheaths of the nervous system may be due, to some extent at least, to such cells. The ectoblastic thickening is increased by the radial division of superficial cells and by the horizontal division of the deeper cells. In the larval and adult stages the large balls of fibrous substance, particularly those of the brain, are surrounded by a delicate cell layer or internal envelope. The nuclei are small and spindle-shaped and form an exceedingly thin sheet. It is possible that this represents intrusive mesoblast, derived from the yolk. Keichenbach states very positively that in Astacus connective tissue cells are squeezed into the fiber balls and eventually surround them. I have no positive evidence to show that the cells in question are not of ectodermic origin, but the behavior of the wandering yolk cells renders it probable that they should rather be referred to the mesoblast of the embryo. The Puuct-substanz of Leydig or fibrous substance is not present at this time, uuless it is represented by a very delicate reticulum in the midst of the nerve cells of the ganglia of the first antenna? on their dorsal surface next to the yolk. Degenerating cells (Fig. 114 s1) occur in abun- dance close upon the optic ganglia and the ventral ectotlerrnal thickening. It may be interesting to notice that the structure of the antemiular ganglion (Fig. 116) is similar to that of the optic lobe. In either case there is a peripheral tier of cells possessing large granular nuclei, an inner layer with smaller nuclei, and an imperfect layer of investing cells. Passing to Stage vn (PI. XLIV) we find the nervous system still very rudimentary. The super- ficial cells, particularly in the region of the optic lobes and the antennte, have large nuclei, which can be seen in the act of division, dividing both longitudinally, thus increasing the superficial area of the plate, and also tangentially, in this way adding to its thickness. The very intimate union of the optic ganglion (Fig. 132 0. L.) with the antentiular ganglion (S. 0. 0.) is still very noticeable, and the delicate investment of these parts on the side of the yolk (mt<4.) is more marked. Punct-substanz has definitely appeared in the supra-oesophageal ganglion where there is a marked transverse commissure, and can even be distinguished in the (esophageal commissures. It forms a very delicate protoplasmic reticulum, and there can be no doubt that the fibrous sub- stance of this part of the nervous system arises as an outgrowth from the protoplasm of ectoderm cells. MEMOIRS OF TIIK NATIONAL ACADEMY OP SCIENCKS. 441 The paired structure of the eetodermal plate is well shown in the antennnlar ganglion on a level with the tran.sver.se commissures, or even in front of this, where paired masses, \\illi small, deeply dyed nuclei, are separated by a median sheet of much larger and clearer cells. This ma\ possibly correspond to the mittelstrang, referred to again. Shortly after this (Fig'. 130) the ganglia are blocked off by a series of superficial constrictions. At least seventeen such gauglionic segments can be counted, beginning with the optic; and supra- oesophageal ganglia and passing to the last abdominal segments. The ganglionic blocks are formed rapidly from the front backward. The ganglia of the first antenna- are now the most con- spicuous part of the nervous system, unless we accept the large optic ganglia. There is a broad, transverse, fibrous commissure in the antennular segment, which is still more prominent at a little later period (PI. XLVI), when eye pigment is forming. From this commissure longitudinal rods extend forwards and unite the brain with the optic ganglia, while similar rods grow backward and form the fibrous axis of the circum-oesophageal commissures. The plane of section in Fig. 148 passes just in front of the oesophagus and through the roots of the first pair of antennae (A 1), which should appear in the drawing as cunt in nous with the integument. The ganglionic cells, which are directed toward the appendages, represent tiie anten- nular nerves, and are more apparent in the following section. The antennular ganglion is both preoral and preanteuual, lying in front of the first pair of auteuure, which it supplies with nerves. The brain and ventral nerve cord are now plainly separated from the hypodermis, and are bathed throughout their extent with blood plasma, in which numerous blood corpuscles are seen floating at every point. Giant ganglion cells have become most conspicuous in the- optic region and at the periphery of the brain next the hypodermis. The brain is partially divided in front next the optic ganglion by a delicate membrane, which forms' a median superficial partition between its two halves. This is continuous, with a delicate envelope, which in some cases can be detected about the brain, and is like that which covers the optic ganglion and nervous system generally. A similar non-cellular membrane at this time divides the retina from the optic ganglion, and is continuous with the cuticular sheath of the latter. The intercepting retinal membrane is directly continuous with the delicate basement membrane of the hypodermis. The cuticular sheaths of the nervous system are present in the embryo (Figs. 157, 168 p>:), the larva (Figs. 175, 17G), and the adult. It may not seem easy to harmonize this account with the view already taken that the wandering cells attach themselves to the nervous rudiments and form a delicate investment to them (Figs. 129, 131 mes.}. Such is plainly the fate of some of the wandering cells, but the number of cells is probably too small to form a continuous structure, and it is possible that the delicate membrane secreted by the ectoblast may serve as an accession to that formed by mesoblast. With respect to insects, Wheeler (07) concludes that in Doryphora the "outer neurilemma" is ectodermic rather than mesoderniic in origin, since — Shortly after the separation ot'thc IKTVO cord from the integumentary ectoderm, it sheds from its surface a deli- cate chiteuous cuticle simultaneously with the shedding of the first integumentary cuticle. This r.ntirlr, which is separated from the surface of the outer neurilemma, and even from the surfaces of the main neural trunks, is after- wards absorhed. At the time when the nervous system has completely separated from the integument there is a slight ingrowth of ectoderm cells along the in id ventral line, most pronounced between the ganglia, and the appearance of a corresponding constriction on the side next the yolk. In transverse sec- tion the jierve cord is somewhat hourglass-shaped. This may be due to a mechanical necessity, arising from a more rapid development of the nerve cells in the lateral masses than in the other parts. Ectoblast cells derived from the integument appear to be infolded between ganglia (sec Fig. 157— a, thin sheet of cells, with spindle shaped nuclei bending in between the last thoracic and first abdominal ganglion, in the lower right, hand portion of the figure), but these infolilings may be somewhat deceitful, since they are straightened, to some extent at least, with the growth of the abdomen (Fig. 168, abg. I). On the ventral surface of the thoracic region (Fig. IC.s) spindle- shaped nuclei are seen wedged between thi> nerve cords on the middle line. It is not, however, certain that these cells are ectoblastie, since the sternal blood sinus is already formed, to which 442 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. blood cells have penetrated, aud here eventually the sternal artery is developed. While the evi- dence is not conclusive, we have only to decide between the former conclusion — that the intrusive tissue is derived from the wandering cells, aud is to be referred to inesoblast, or the view that it represents differentiated ectoblast. A general account of the structure of the nervous system of the larva is given in the first section. Further than this the details of development have not been followed. In comparing this account with that given by lieichenbach for the crayfish, Astacusjliiritttilix, there are numerous particulars in which there is no agreement, while in some important matters ;ve are in accord. I find in Alpheus the oral iuvagiuation occurring on a line drawn between the bases of the auteunular buds, aud I have a great many preparations of the eggs of the lobster. Homarus ameri- cnniiii, which show the earliest traces of the stomoda-um. Before th;' first aiitemi;« are folded, wheu they are distinguished as dense patches of cells, some eggs show the primitive mouth as a minute circular pit, lying nearly on a line drawn between the centers of these proliferating cell areas, but, so far as my observation goes, never distinctly iu front of them. The relative positions of the mouth and first pair of auteuntB shift very rapidly during the early period of their growth, which precedes the fully developed egg-nauplius condition. The pit elongates and becomes a transverse furrow, aud by the time the first pair of antenna? are clearly marked off as rounded buds, and before the second pair are raised into folds, the mouth is still on a line with the first of these appendages. When the second antenna- are elevated iuto folds the month is behind the buds of the first pair, or on a Hue between their posterior edges. Iteichenbach (I at'. II, Fig. 7!' the mouth in Craiigon as distinctly postoral, but an inspec- tion of figure 11 of his paper, leaves some doubt as to whether he is not mistaken in this particular. A single rudimentary appendage, marked as the first antenna, is represented as occupying nearly the entire space between the optic disk and thoracic abdominal fold. This does not agree with my own preparations,* and since iu Alpheus, Homarus, and Astacus the mouth does not appear until a dense stratum of cells carpets the intermediate space between the optic disks and lateral cords, there is some difficulty in interpreting the cluster of cells marked m, in Kingsley's paper, as the invagination of the mouth. t In Alpheus, Elomarus, Astacus, and probably in I )ec;ipods generally, the ganglion of the first antenna cannot be said to be postoral, but its development begins nearly on a line with the invagination of the primitive mouth. The ganglion of the second antenna is developed behind the primitive mouth, but gradually shifts forward with its appendage until it comes to lie, iu the larva, considerably in front of the mouth. In this movement, the ganglion however outstrips the appendage. The ganglia of these two appendages unite to form the brain or 'supra (esophageal ganglion. The ganglion of the first pair of antenna.', is constricted into two portions marked by an oblique, transverse line at the surface. The anterior of these parts lieichenbach calls the * 1 have preparations of the eggs of Grant/on ralgarit, iu various stages of development, from the segmentation onward. Iu one egg, which i.s somewhat, more advanced than that of figure 10 (:il), or than the Alpheus iu Fig. 58 11!' tliis paper, the- optic disks aud ventral plate are dense patches of cells. On either side of the ventral plate aud in close relation to it there is a marked area of cell proliferation which represents the mandible. Iu the space between this aud the antenna the nuclei are more scattered, but the karyokiuetie figures show the activity of cell division. In a late nauplins stage the storuodoeum is on the middle line between the first aud second antenna1, and the auteu- uular ganglion is segmented into two parts on each side, as shown for Alpheus ill Fig. 110. t Tins criticism is supported by Weldon's obierv.-iti'Mis on Oaugou, who, with reference to this subject, says: "The first uiiteniiiu are evidently prawral from the very c.urlirst period at which the mouth is visible." Op. cit. MEM01US OK THE NATIONAL ACADEMY OP SCIENCES. 443 "vordere Hirnanschwellung" and tin- posterior the, "Seitenanschwellung," using tbe terms of Krieger tind Dietl. These latter particulars accord with BiMchenbach's dese.ription of the crayfish. 1 have not, however, found that in Alpheus, behind the level of the first pair of antenna-, the lat- eral parts (Seitenstrange) divided up into three sections. Beichenbach fnrtlier states that in each segment a middle-strand invagination is found, while the ganglia of the fifth (first maxillary) seg- ment has a prominent median string. In Alpheus I find an undoubted median ingrowth of surface ectoblast between the two nerve cords. This in all probability corresponds to the mittelstrang, and, as already stated, it is most marked at a stage just before or at the time of the first appearance of the eye pigment, when the ganglia are separating from the integument. It is seen between the successive segments of the ventral nerve cord, where a single cell or string of several cells appears to be wedged between the two cords which are entirely separated (compare Figs. 151, 160). The nuclei of these cells are elongate and perpendicular with the surface. They are derived by delaminatiou from the super- ficial ectoblast, as the karyokinetic figures of dividing cells clearly show. These ingrowths are most noticeable between the segments, and whether they form any part of the nervous system or not, I have been unable to determine. In insects the mittelstrang appears to take no part in the nervous system of the adult, but in the thoracic region it is converted into the chiteuous furcaj. According to Beichenbach the lower rr driiugt sicli bei tieferer Betrachtuug inimer winder di-r < ii-danke nuf, dass man sohou voni ersten Stadium au eiiicm uutreunliarrn Ganzen gegeuiibersteht. (o4, \>. *y Margaret Robinson. The meilian eye was observed in some rijjlit ilillerent species of the Candida-, including the genera Palmmon, Hippolyte, Virbius, C'raugon, and Pan- diilus. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 445 Alpbeus in the only genus in which I have found glands in the eyestalk. These arc most notice- able at the peripheral parts of the stalk between the basement membrane and the ganglia. They are in reality parts of the green gland, which scuds outgrowths from the bases of the second antenna? into the antennules, the eyestalks. the labrum, and the whole front of the head, so as to completely envelope the brain. The histology of the glandular cocca is the .same in all parts. They consist of a cubical epithelium, composed of very large cells, supported by a basement mem- brane. Near the coeca of the autennal gland comes a layer of very loose connective tissue. This is specially abundant below the basement membrane of the retina. It forms a continuous sheath for the optic ganglion, and is reflected over the optic peduncle and brain. By far the greater mass of tissues of the eyestalk belongs to the optic ganglion. This is composed of ganglion cells (Fig. 187), nerve fibers, and the peculiar fibrous tissue variously called " Punct-substanz" or " Ball-substanz," and "substance ponctuee." Viallaues, who has made very careful and detailed studies of the optic ganglia of Arthropods, uses the following terminology (61). He divides the optic ganglion into two parts, an external and an internal portion. The surface of the external region is covered by the limiting membrane of the eye. The external is united to the internal parts by a bundle of crossing fibers, the external clii- asma (Fig. 178, Cb. Ex.). This, according to Viallaues, corresponds to the center of the corneal surface, and consequently to that of the limiting membrane of the eye. The internal portion between the external chiasma and the optic peduncle is composed of three principal masses of punct-substauz : (1) la masse medullaire externe, (2) la masse inedullaire interne, (3) la masse inedullaire terininale. The external medullary mass is united to the masse medullaire interne by an internal chiasma, while a fibrous peduncle joins the internal medullary mass to the masse inedullaire terininale. The nerve fibers which pass between retina and ganglion, he calls the post, retinal fibers, and designates as " optic nerve" the peduncle by which the optic ganglion is united to the brain. The distal mass of puuct-substanz is styled lame ganglionnaire,* which he divides into a nuclear layer (couche a uoyaux), a molecular layer (couche moleculaire), and a cellular layer (couche a cellules ganglion naires). The punct-substanz of the optic ganglion is thus divided into four principal masses (Figs. 178, 209). Without adhering closely to the rather cumbrous terminology of Viallanes, the parts of the ' nervous system contained in the eyestalk between the brain and retina may be designated as follows: Optic peduncle (optic nerve of Viallanes and others) ; proximal segment (masse medullaire terminate) ; internal middle segment (masse medullaire interne); e.rtenttil middle segment (masse me'dullaire exterue) ; distal segment (lame gaugliouuaire) ; optic nerve (couche des fibers post- r6tinienues). GENERAL STRUCTURE OF THE COMPOUND EVK. The transparent cornea is the secreted product of a specialized layer of the hypodermis, which was designated by Patten as the "corueal hypodermis" and later as the "corneagen" (50, 51). Beneath this lie several strata of dioptic and sensory cells, separated from the ganglion by a base- ment membrane, which is continuous with that of the general hypodermis of the eyestalk. The ouimateum, or eye proper (including those parts which intervene between the cornea and basal membrane), is formed by the repetition of a highly specialized'unit, the eyelet or ommatidium. The size, number, and arrangement of the ommatidia is characteristic of species or genera, but is subject to considerable variation in different individuals, and the shape and arrangement of the ommatidia may be very irregular in different parts of the same eye. The ommatidia are differentiated clusters of ectoderm cells. There is a single ommatidium for each lens or corneal facet. The number of cells composing theommatidinm is very uniform in Decapods, Stomatopods, and Schizopods, so far at least, as the most essential cells are concerned. They are as follows: Cells of corneal hypo- dermis, 2; crystalline-cone cells, 4; outer pigmeuted retinular cells, 2 ; inner pigmented retinular cells, 7 (functional) ; Accessory pigmented cells — irregularly distributed, both above and below the basement membrane, probably of ectodermic origin. * The lame ganglionn:iire is called " Retina ganglion" by Clans, who regards it as tho trur ivt.ina ; '- monetes. The seven proximal retinular cells secrete on their inner sides the. rhabdom or rhab- domeres. A transverse section of the rhabdom gives the peculiar seven-pronged figure shown in the drawing (Fig. 205). The cells appear as fused in section, but possibly they would separate with readiness if macerated. Unfortunately I had no fresh, material to experiment with. The proximal retinular cells appear to penetrate the basement membrane, and they are continuous below it with nerve fibrils. As to their distal ends, I have seen no evidence that they extend out to meet the cornea. The retinular cells abound in dark pigment. The accessory pigment cells secrete a peculiar pigment which is glistening white in rellected light and is amber color in transmitted light. This may be similar to the pigment of certain cells which occur beneath the cuticle of the larva of Decapods in various parts of the body. It is not decolorized when subject to the prolonged action of weak solutions of nitric acid, while the black pigment is completely removed. What I once regarded as chitiuous bodies (20-21) were fused masses of this pigment which had been treated with nitric acid. These cells penetrate the base- ment membrane, beneath which there is a considerable mass of both yellow and black pigment. The trains of cells which accompany nerve fibrils into the fibrous portions of the optic ganglia also contain granules of black pigment. The number of accessory pigment cells belonging to each ommatidiiim is indeterminate. They have the power of free movement or migration outward from the basement membrane and the power of retraction like the retinular cells. In Fig. 200 they are seen widely diffused, while there is a zone of black pigment cells (the, pigment withdrawn by acid) enveloping nearly the entire, cone and the distal end of the rhabdom. In an eye taken from a prawn which had just moulted, the yellow pigment was restricted to a narrow zone next the basement membrane. Outside of this belt the retinular cells were colorless nearly up to the proximal ends of the coues, while the cones themselves were draped in black. ARRANGEMENT OF THE OMMATIDIA. In Alpheus saulcyi the omraatidia are arranged in a hexagonal system, subject to variations in different parts of the eye.* In the central parts of the cornea the facets are symmetrical hexa- gons. On the lower side of the, eye the rows are more irregular and individual facets tend to become square and rounded. Toward the outer side of the eye the facets are very nearly square, next to these they become irregular and rounded, and on the extreme outer edge the facets are sometimes hexagonal. There is probably considerable individual variation. 1 have examined the cornea in four other species of Alpheus, namely in Alpheus heterochclis, A. minor, A. normani, and a West Indian species closely allied to A. heterochelis. These cases afford some very interesting * For ;i stmly of the coruea, adults of the largest size were selected and the cuticle was cleaned by boiling in a concentrated solution of potassic hydrate. MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 447 facts in connection witli the arrangement of omniatidia. In Al/i/icus nonnmi-i the facets arc gen- erally symmetrical hexagons, two of the sides being ([iiile sliorl. These, lend to run into squares or rounded areas at parts of the periphery. Alpln-im minor lias the facets in the lower part of the. eye either of the hexagonal or some polygonal form, while in the upper and outer parts the faeet .; are perfect squares. These blend on all sides into symmetrical hexagons. The lirst transition from the hexagon to the square is seen on the outer edges <>f this area where the individual facets become more, and more rhomboidal, as the two opposite sides of the hexagon are, more and more reduced. Finally these sides disappear and the facets become rectangular. The square facets of any given row now lie opposite to those of the adjoining rows. Four facets belonging to any two rows meet at a common point. In passing, however, from the area of square facets to the periph- eral parts, the facets, though square, are out of line. Starting from a point where the facets of a given row are both tetragonal and opposite those of adjoining rows, and following the line over the hemispherical surface toward the periphery, the facets of this row soon tend to become alter- nate with those of the row next to it. A facet of the second row lies slightly in front of the cor- responding facet of the first row. The facet of the third row lies a little in advance, of this and so on until the, hexagonal shape is gradually assumed. In the, Bahamau variety of Al/i/if/ix ln'tci-ochclix the facets in the larval eye are markedly hex- agonal. In the adult there is the same curious transition from the hexagon to the square as we have noticed in Alphexn minor, only it is here much more, striking. In the peripheral parts of the eye, especially in the lower and inner portions, the facets are generally hexagonal. In the, upper half of the eye between the center and periphery there is a small area of square facets. I n . I 1/iJicnft heterochelix the facets are characterized by much greater looseness of arrangement, there being large interlenticular spaces. The facets are squares with rounded corners. In some parts of the eye, as on the inner lower side, the facets of the adjoining rows lie opposite each other. In other parts they regularly alternate and show a tendency to become hexagonal. At the periphery the facets are widely separated and circular.* The markings on the conical lenses generally consist of a small central spot which is slightly elongated and strongly refracts the light. It appears in section to represent a slight depression. In the lobster, as first shown by Parker, there is a faint diagonal band which intersects a central hazy spot and divides the square into equal triangles. These diagonal bauds are all parallel in adjoining rows. In Alplieus saulcyi the elongated spots, if continued across the lens, would form a series of similar diagonal lines, but none such as these could be detected. In Alplieus hetcrochelis the markings are very constant and peculiar. In the center of the lens there is a large irregular impression from which numerous rays extend on all sides. The latter do not always reach the periphery. They appear like grooves in the substance of the lens. In the older embryo (Fig. 194) the apex of the cone cells seems to touch the under side of the lens in certain parts of the eye. It is therefore possible that these cells remain in contact with the cornea and the interference thus caused gives rise to the spot. A similar explanation is offered by Parker to account for the con- ditions found in the lobster. The significance of the radiating lines seen in Alphem heterochelis I have not determined. In Homarus the facets are -square and grouped with remarkable regularity up to the very edge of the cornea, excepting at a point on the upper side, where a peninsula of tough cuticle juts in from the surface of the stalk and interrupts the elliptical jet black area of the surface of the eye. About this process, particularly in front of it, the facets are hexagonal or irregular and very much smaller. Parker (48) states that in the lobster " the omniatidia rearrange themselves between the times when the young animal is 1 inch and 8 inches long. During this period the omniatidia increase about ten times in length and about five times in breadth." I find that this rearrangement begins at a much earlier period, in fact in the older larval stages. My examination comprised the follow- ing stages : (1) Length 8-9""" (first larval stage) ; (2) length 11""" (fourth larva) ; (3) length 15.3""" (sixth larval stage) : (4) length 49""" (lobster 1 year old). * A similar transition of the square into the hexagonal facet in the same eye occurs in AstacuH. See Howes' Biological Atlas, Fig. 111. 44S MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. Ill the first larva of both Ho mar us and Alpheits saitlci/i I iiiul that the facets are not ouly hex- agonal but tend also to be slightly rounded. In the larva 11""" long the lenses tend to become square toward the center of the cornea, while at the periphery they are smaller and generally hex- agonal. Occasionally, however (just in what part I did not ascertain), the peripheral facets tend strongly to the tetragonal arrangement. Ill the next case (larva 1.5.3""" long) the facets in the outer parts of the eye are small and symmetrical hexagons. Toward the center they become larger, nearly square, and there is considerable iuterlenticular space. lu the yearling lobster the readjustment of the corueal lenses from the hexagonal to the tetrag- onal system has been effected over the greater portion of the eye, but the transition is still illus- trated iu a very beautiful manner. The cuticular ingrowth or peninsula, already referred to, seen at the upper surface, is hook-shaped, bending backward. In the open angle in front the retina is least differentiated. The corneal facets in this region are small and mostly hexagonal. Following the lines of facets as they curve outward and backward from this point over the convex surface of the eye, we see illustrated in a very striking way the passage of the hexagon into the square by the gradual reduction of the opposite sides. The sides which are sacrificed are the third and sixth, couuting from the side which lies between two adjoining lenses of the same row. Over a consid- erable area the facets are not quite square and thus tend to alternate with those of adjoining rows. The gradual transition to the square is attended by a gradual increase iu size. In the narrow angle behind the peninsula, the area of the hexagonal facets is smaller. In all other parts of the periphery the facets are square up to the very edge, as in the adult. The corueal lens in an 8-inch lobster has about twice the area of that of the lobster one year old (length nearly 2 inches). After reviewing these details the difficult question arises: What is the significance of this remarkable change, and how is it effected ! Parker (48), who has made a careful study of the arrangement of the ommatidia in different Crustaceans, recognizes two plans on which these organs are grouped, the hexagonal and the tetrag- onal. He .says that " in the Brachyura, as well as in three families of the Macrura, the Hippiihe, Paguridaj, and Thalassinidae, the arrangement of the ommatidia is invariably hexagonal. In the remaining macrurous Decapods the ommatidia are grouped on the tetragonal plan." There are. however, exceptions to the latter statement, some of which are mentioned. He regards the change from the hexagonal to the tetragonal system as apparently due to the increase in size anil the con- sequent crowding of the ommatidia, and reaches the conclusion from the various facts presented, that the hexagonal arrangement is phylogeuetically the oldest. Upon this view we should expect to find the eyes of the most highly differentiated of the Crustacea arranged on the tetragonal sys- tem, whereas in points of fact the crabs, who are notorious for their great activity and keen powers of vision, permanently adopt the hexagonal arrangement. The lower Macrura adopt both methods, and in certain species of Alpheus, as Alpheus minor, which are active in habit and show no trace of degeneracy, the change is begun but not completed in the same retina. In order of time the hex- agonal prism precedes the square prism and the conditions which determine the permanency of each of these systems in the adult life of the individual are undoubtedly inherited, but they do not appear to have a phylogenetic significance, at least I do not see the way clear to an explanation upon this ground. Taking, for example, the lobster and the crab, in each case the larval eye represents, we must believe, the more generalized type, the adult eye the more specialized type. The larval eye of both Macrouran and Brachyouraii has the hexagonal facet. We may safely conclude that this is a primitive arrangement. The adult crab, which has a more highly organized nervous system and keener senses, retains this primitive arrangement, while the lobster, whose senses are without doubt duller, departs from the type, and in the adult the facets of the cornea take on a permanent tetragonal shape. The conditions which we would naturally look for are thus reversed in these two forms. We must assume that the change in the case of the lobster is a useful variation, that it is in some way subservient to keen vision, otherwise it could never, upon the theory of natural selection, have secured such a permanent characteristic footing. On the other hand we must believe that the hexagonal facet puts the eye of the crab into better harmony with its environment than the square facet would do, for if the eye had not been in harmony with external conditions it must MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 449 have varied and attained to a new structure, during the course of the evolution of the Brachyoura from Macrouran ancestors. Many crabs like the sand crab, Oct/poda arenaria, spend a good portion of time out of the water and their eyes are, admirably adapted for vision in the air, as in the- case of this species, which will detect any moving object, such as a man or a dog, a long distance off. We might thus hope to find a way of escape troin the. difficulty in the diversity of habits in the two forms, the lobster being exclusively an aquatic animal. But the primitive Crustacea must have all been aquatic, and the hermit crabs, some of which spend all or nearly all of their time in the water, agree with the Brachyoura in having a corneal membrane composed of hexagonal facets, not to speak of other Macrouran forms which are invariably aquatic, like Alpheus, in which the hexagonal system is retained or there is a transitional condition between the hexagon and the square in the eye of the adult. We do not yet know the physical or physiological significance of these two kinds of facets, and possibly in this fact lies the root of the difficulty pointed out above. The hexagonal arrangement is the natural one for tubes with elastic walls to assume when there is sufficient mutual pressure, and it is also the most economical arrangement, so far as wall space is concerned, for regular prisms of equal capacity. Next to this iu point of economy of wall space comes the square prism, and next to this the triangular prism, where, as before, all available space is occupied. This last arrangement, however, is not regularly assumed. If a given number of bexagonal prisms occupying a given space tend to increase iu size, then they would tend also to assume another form less economical of wall space, such as the square prism. If the wall area of the hexagonal prism remains the same and the number of prisms increases iu a giceu space, then each prism must be less economiciil of wall area and assume another form like that of the square prism. If we may apply the same principles to the growing ommatidia, it is possible that crowding may have something to do with the change. In the lobster the change from the hexagonal to the tetragonal system is attended by a growth of the omuiatidium iu all directions and by the addition of new ommatidia. If an outline drawing of the eye of the first larva of the lobster be compared with a similar one of the fifth larva, it will be seen that the, convexity of the outer surface, that is, the area of the eye, has very greatly increased, whereas the diameter of the eye stalk has remained very nearly the same. During this period the change from the hexagon to the square has been begun. During this time the ommatidia have increased in length with the increasing convexity of the cornea, and they have also increased in sectional area and in actual number. It is a very noticeable fact also that in the lobster the eye stalk is compressed dorso-ventrally, so that the horizontal radius of curvature of the retinal surface is considerably greater than the vertical radius of curvature. Furthermore, the sides of the hexagonal facets which suffer reduction lie iu a vertical plane, indicating a strain or pressure iu a dorso-ventral direction. In Alpheus saulcyi the relative increase iu the convexity of the corneal cuticula is very slight in passing from the first larva (length, about 4""") to later stages, and the eye of an adult (i:> long) is only about one-fourth larger than that of the larva at the time of hatching. Moreover, in Alpheus the convex surface of the eye is nearly a perfect hemisphere, the curvatures being the same in every plane. If we examine the eyes of a crab in a similar way we find that the area of the surface of the retina increases less rapidly in passing from the zoiia to the megalops stage than in the case of the lobster in going from the first to the fourth larval forms. The comparison, however, is of little value since the lobster has an abbreviated development. The eye stalk of the adult crab presents a large retinal surface, but it preserves a nearly cylindrical form, although the radii of curvature of the retinal surface are very unequal. In the hermit crab (iu a single species examined) the hexag- onal arrangement is preserved, not, however, without indications of a tendency to become tetrag- onal, the hexagons becoming asymmetrical iu certain parts of the eye. There is, however, the same compression of the eye stalk in a dorso-ventral plane as we see in the lobster. It has seemed to me worth while to point out these facts as offering some suggestions to the problems under discussion, although 1 make no attempt at a mechanical explanation. S. Mis. 94 29 450 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. m The primitive arrangement of the ommatidia was probably in the form of simple tubes or cylinders, with spaces between them and with rounded or indefinite facets. Mutual pressun- among these tubular eyelets, arising from any cause, produces the hexagonal arrangement, the most economical method so far as wall space is concerned. Interferences such as have been sug- gested, as growth of individual ommatidia or increase in the number of ommatidia in the same area, thus admitting a method of arrangement less economical of wall space, or the great increase in length of theoinmati'liaand a relatively less increase iu width, attended by a progressive change of the hexagons into squares (the apparent slipping of the rows of facets on one another), may enter as factors into this change, but they do not suffice to explain all the conditions. It will be understood, of course, that there are no changes in the individual facets, these remaining in the same shape until they are cast off iu the moult. The changes which the individual ommatidia undergo are very gradual, and since the number of cells for each omiuatidium is constant and deter- mined at a very early period, excepting the accessory pigmeut cells, they must be attributed to the change in the size and relative positions of the cells themselves rather than to intussusception. It is possible, that the change from the hexagon to the square is not produced in the same way in all cases and that the conditions of growth which bring about this result are far more compli- cated than would appear from the suggestions which have been made. A careful study of the arrangement of cells in the ommatidia of the eye of the young lobster during the period of transi- tion would possibly throw some light upon this interesting subject. THE DEVKI.OPMKNT OF THE COMPOUND KYK. Five years ago (20) 1 stated my conviction that the compound eye of Alpheus, and probably also of Palaemontes and of the large Isopod, Ligea oceanica, originated I rom a thickening of the superficial ecloblast. The development of the eye iu Alpheus was more fully described in a pre- liminary notice (131). 1 will now recapitulate the main results, at the same time correcting such errors as I have detected. In studying the development of the eye the following are some of the subjects which present themselves for investigation: The origin and structure of the optic disks; the separation of the optic disks into a ganglionic and retinal portion by an intercepting basement membrane; the dif- ferentiation of the retina into ommatidia or eyelets; the differentiation of the optic ganglion and the development of the optic nerve, by means of which the sensory end organs of the retina come in direct relation with the ganglion. (1) Origin of the Optic Disk.— The optic disks (Fig. 58, PI. xxxn) consist of large ectodermic areas or patches on either side of the middle line. They are centers of rapid cell division, united by means of the lateral cords, which are bauds of proliferating cells, with the thoracic-abdominal plate. So far as 1 am aware we have no account of the origin of the optic disk in any Decapod except- ing Alpheus, Astacus (54), Crangou (30), and Homarus (47\. Parker, in his careful studies on the eye of the lobster, was unable to obtain the earliest traces of the developing optic disk, and the accounts of Reichenbach and Kingsley differ very materially. I will therefore describe somewhat in detail the process by which the optic disk is produced in Alpheus. The optic disks at the time when they consist of a single stratum of cells are shown in Figs. 58, 08, and 69. A series of four transverse sections through the central portion of the left optic disk is represented iu -Figs. 64-67. The posterior face of each section is presented, the series pass- ing from the front backward. Since it is from the optic disk that the eye and its ganglion are developed, the important inquiry which arises at this stage is, how is the change effected by which the disk passes from this single- layered to a many-layered condition like that seen iu the egg nauplius (Figs. 107, 114)? If the eye represents a series of hypodermal pits, it would be reasonable to look for some trace of these infoldings in the embryo. If, on the other hand, the compound eye of the higher Crustacea repre- sents a closed vesicle produced by a single iuvagiuatiou of the hypodermis, of the type seen in the pi -ototracheate Peripatus, we should expect to find some trace of an involution at this stage. So the answer to this question may have an important bearing upon the phylogeny of the compound eye. Cell boundaries are easily discernible at the surface, but it is evident that the nuclei do not MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 451 all lie at the same level. Passing now to Stage IV (Fig. 72), we notice several important changes in the external appearance of the optic disks. They have approached nearer the middle line; they have increased in area; their nuclei are more crowded ; near the center of the disk and on the side of it toward the middle line (Fig. 72. ('. M.) the nuclei are distinctly larger. In inter- preting these changes we must resort lo sections and a careful study of dividing nuclei. A com- plete series of consecutive transverse sections through the left optic disk at this critical stage is given in PI. xxxv. The last of the series cuts the rudiment of the first antenna. Karyokinetic figures come to our aid in pointing out the way in which the growth of the disk is ellectcd. The nuclei divide either radially, the plane of division, clearly marked by the, equatorial plate, being perpendicular to the surface, or horizontally, the plate being in this case parallel with the surface. In a single disk at this stage there were eight cells undergoing radial division. Of these, six were near the periphery, where the cells formed a single stratum, while two were near the center, where the disk was slightly thickened (Fig. SO, C. J/.j. It is evident from this and similar cases that the increase in area of the disk is accomplished by radial cell division. The same is true of all proliferating areas in the lateral cords, where the appendages are soon budded. In the area marked C. .!/. (Fig. 80) the optic disk is no longer a single layer. This thickening- is due either to horizontal cell division, that is, delamiuatiou, or to emigration. The appearances of emigration are often very deceitful, but I think we may safely conclude that the initial thick- ening of the optic disk in the proliferating area, marked C. J\f. (Figs. SO, 90), is due to emigration, that a solid ingrowth akin to invaginatiou takes place at this point. Thus the cell marked ec in Fig. 80 is distinctly below the surface. The boundaries of the cell can be clearly seen. The cell ec in Fig. 90 (dotted line should be extended), on the other hand, is clearly in contact with the surface by a slender protoplasmic process, while the nucleus lies at a much lower level. I inter- pret the latter as a cell at the point of breaking all connection with the surface and migrating to a lower position. In the first instance this has already been accomplished. In the stages under discussion there are one or two cases out of a large number of sections involving the optic disks of several individuals, which probably indicate delamination in the peripheral parts. At a later period (Figs. 102-107) the nuclear figures are conclusive. In one instance (Fig. 102, ec.) two cells are seen delaminating side by side. The thickening of the optic disk is thus due in part to delani-ination, and this process is probably supplemented by emigration, at least in the central area. The central area represents in all probability the " optic invaginatiou" of the crayfish, and is concerned solely with the production of the optic ganglion. In Alpheus there is a proliferating area simply, but no superficial depression or invaginatiou in the strict sense. It is noticeable that in Stage HI (PI. xxxin) wandering cells, or cells which travel through the yolk, have not appeared in the neighborhood of the optic disks. In Stage iv (Fig. 70, Y. C.) they are not far away, and in later stages (Fig. 91) these cells are close upon the disks. Some of them which enter this region, coming near to or uniting with the disks, undoubtedly degenerate (compare Y. C., Fig. 94, *-«', Figs. 95, 96, 99, 100). Some of the cells of the disks next the yolk elongate, and appear to form a somewhat transitory covering. As already suggested, these bodies are probably derived from the wandering cells. The various stages by which the disk, already described, is converted into a conspicuous lobular mass of cells in, closest relation with the autennular ganglion such as we have in the egg- nauplius (Fig. Ill, 0. L), can be seen by reference to the plates (Pis. XXXVI-XL). (2) The Development of the Retina and Optic Ganglion. — The next event of importance is the differentiation of the optic disk into gauglionic and retinal portions. This is already begun in Stage VII. A deeper layer from which the ganglion is developed (Figs. 129, 132, G. L.) is gradually separated from a superficial tier of cells with very large nuclei (OE). This layer is the retinogeu. The differentiation begins in the lateral lower halves of the optic disks and extends upwards and toward the middle line. As the disk spreads outwards, and at the same time increases in thick- ness, it tends to overgrow the hypodermis and becomes raised into a lobe or fold. The optic lobe (Figs. 136, 138) thus represents a thickening of the hypodermis. It is covered next the yolk by a delicate basement membrane (Bm.), which is continuous with that of the surrounding hypodermis. 452 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. Wandering cells are seen in contact with this membrane (Fig. 136), but tbey probably do not share in its secretion, although tbey occur in the closest relations with it. At a little later period (Fig. 138) the retinal portion is several cells thick ou the outer edges of the lobe, while it is a single stratum in the. sagittal section, shown in Fig. 138. The plane of section is near the center of the lobe. The deeper nuclei of the ganglion are large and clear, the outer are smaller and stain more intensely. This section can be clearly understood if compared with the transverse section, Fig. 146. We see that the optic ganglion is here, divided into an external or distal part and an internal or proximal portion by a thin sheet of very large and clear ganglion cells. Parker (47) describes and figures an exactly similar structure in the lobster, aud 1 fully agree with liitu in regarding this band of nuclei as representing similar bands, which Eeichenbach (Taf xn, Figs. 173,174,71. \\'. I. IV.) describes in the rraytish. In Iteichenbaeh's plates these nuclei appeal1 as a narrow fold, forming the lining of what is described as a secondary optic invagination. Three, puuct-substauz masses have already appeared in the inner half of the optic ganglion the external middle segment, which lies next to the band of large nuclei, and the internal middle and proximal segments. The proximal medullary mass is much the largest and is the first to be differentiated, although the others follow close upon it. The dividing unclear band lasts but a short time, and in Stage 10 (Fig. 167) lias disappeared. In front of it is developed the lame gang- lionaire or distal segment of the optic ganglion. This has an outer convex surface which is con- centric with the basal membrane and with the outer surface of the retina, and it is carpeted by a special layer of ectoderm cells. These appear in section as a single row of elongated nuclei. In an early communication (20) I stated my belief that the punct-substanz arose from a metamorphosis of ganglion cells. This view was suggested by certain appearances presented by the large clear nuclei, more particularly by those of the dividing band in the optic glanglion (see Fig. 180). Kingsley (3-) came to the same conclusion in regard to certain large clear nuclei in or near the distal segment of the optic ganglion of Craugon. In reviewing this subject more carefully I am convinced that this interpretation is erroneous. These, large clear cells are in reality undergoing indirect cell division, as proved by the karyokiuetic figures which are occa- sionally seen. Both tue chromatin network and the chromosomes are exceedingly delicate, and when the section is in the plane of the equatorial plate an appearance is presented which under certain conditions of staining and preparation might easily be interpreted in favor of retrogressive metamorphosis. I conclude that the punct-substauz of the nervous centers is in all cases derived from the protoplasm of cells, not from cell nuclei. In Stage IX when eye pigment first appears, the structure of the retina is very simple. By the transverse section (Fig. 146), we see that the retina consists of a thickened ectoderm plate, thickest in its deeper portions, thinning out toward the middle line at the surface. It has the shape of the half section of a concavo-convex lens. In Paheuiouetes the structure of the eye is precisely similar at this stage. Fig. 189 gives a section of the eye. of this prawn, on a line with the (esophagus behind the optic ganglion, where it is seen to rest against the yolk. The black pigment, though appearing to arise in connection with certain mesodermic cells (wandering cells from the yolk), it in reality belongs to deep ectoderm, and marks the retiuular cells. The cell protoplasm bearing the pigment bodies grows outward (Fig. 146), and also pierces and extends some distance below the basement membrane (Figs. 191, 19U). The latter is a delicate cuticular structure secreted by the ectoderm cells which lie along the line of division of retina and ganglion, and continuous with the basement membrane of the hypodermis. In some sections it appears to be duplex, a condition described for the eye of the lobster by Parker (47), in which the inner layer enfolds the optic ganglion. The wide open fissure which now exists between retina and ganglion (seen in transverse section at «», Fig. 136) is partially filled wilh yolk. There is not the slightest doubt that cells enter this fissure from the yolk (Mes. Figs. 14A l:KX!:s.-. Ill June, 1890, while enjoying the facilities for biological research afforded by the labora- tory of the U. S. Fish Commission at Woods Holl, Mass., it occurred to me that some valuable experiments could be made by testing the effects of direct sunlight and total darkness upon the growth aud behavior of the pigment cells of the compound eye of Crustacea. After finishing my experiments upon one form I learned of the experimental work of Exuer* upon the eyes of the glowworm, Lampyris npUndidula , of Hydrophilus, Dysticus, and Colymbetes, in which he records the same phenomenon in insects which I have observed in a Crustacean. Later a paper has also appeared, by Mademoiselle M. Stephauowska, on the histological arrangement of pigment in the eye of Arthropods. I have seen only an abstract of this work, from which I gather that it deals either largely or entirely with the eyes of insects. My experiments were made upon the common prawn, Palcvmonetes rulgaris. A dark chamber was constructed aud rendered as absolutely light-proof as possible. Inside of this a small glass aquarium was so arranged that a stream of sea water could be kept run- ning through it for any length of time. Three egg-bearing females were then placed in the aquarium and the chamber was sealed. The egg embryos were early nauplius stages. Females with eggs iu a similar stage were also kept under observation iu an aquarium exposed to the light. The general cast of color of the prawn taken in the light is some shade of light brown or brownish green. After spending eighteen days in the dark, the prawns were taken out aud exposed to tln> moderately bright light of the laboratory. The eyes were jet black aud appeared to nave greatly Since these notes were written I have received the completed work of Exner, Die Physiologic (let- FacMirten ran lirel>*tii iniil Iimtcleii. in which the field of experiment is greatly enlarged. MEMOIES OF THE NATIONAL ACADEMY OF SCIENCES. 455 swelled in size, and the body was bleached nearly white. The peculiar appearance of the eyes ICHS caused !>// the for/card extension of the distal retinular cells, of which there is a single pair in ench ommatidium. The eggs of some of the prawus were hatching, and the pigment of the /ora was carefully compared with that of the first larva of Palremouetes hatched in the light. Roth the black pigment of the retinnlar cells and the yellowish green pigment of the accessory pigment cells of the eye and the large brown chromatophores in different parts of the body were of the same character, whether the embryo had developed in darkness or light. Another prawn was kept in the dark thirty-eight days, and on exposure to the light it pre- sented the same appearance. As in the other cases, as soon as light reached the eye the distal retinula? began to retreat to a deeper level. At first the black pigment which characterizes these cells extends out to the cornea. After an exposure of two minutes to direct sunlight a slight transparent band is seen below the cornea. This light zone increases as the pigment continues its retreat until, in the course of three-quarters of an hour, the distal retinula? eusheath only the lower ends of the cones. In another experiment a prawn was left only about twenty-four hours in darkness. The same effects were produced in the eye, which assumed its former condition after being in the diffused light of the room twenty-five minutes. The distal retinular cells thus respond very promptly to the action of the light, and in the course of a few hours (the exact time needed was not determined), if excluded from the light, completely enshroud the proximal ends of the cone cells. In the eye of Palsemonetes, taken in ordinary daylight, there are three distinctly marked strata of pigment between the basement membrane and the cornea, a proximal narrow stratum of yellow- ish brown pigment belonging to the accessory pigment cells; a wider and much lighter area peppered with dark granules, pertaining to the proximal retinular cells, the nuclei of which form a conspicuous belt or layer on a level with the distal extremity of the rhabdoms. Lying close upon the tier of retinnlar nuclei is a thin stratum of intensely black pigment, composed of the distal retiuular cells. As stated above, there are two of these cells to each ommatidium, and they each send out a slender thread-like process, which extends in some cases as far forward as the corueal cuticula, where it is possibly attached. I have not detected any similar prolongations in the direc- tion of the basement membrane. Below the level of the cone, which terminates abruptly in a convex proximal surface, the cone cells are prolonged into a long slender stalk consisting of proto- plasm or of a refractive substance of a different nature from the cone. The cone cells are not apparently prolonged below the level of the retinular nuclei. The distal retiuular cells thus sur- round the proximal ends of the cone cells. In the eye exposed for thirty-eight days in the dark the distal retinular cells form a stratum about midway between the corueal cuticula and the layer of nuclei of the proximal retinular cells. The nuclei occupy a central position in this layer. Piginented pseudopodia extend forward to the cornea, and occasionally a cell shows a slight inward prolongation. Had the oyes been preserved without bringing them into the light, even for a moment, the distal retinular cells would undoubt- edly have occupied a still more peripheral position. In the eye kept in the darkness for the same length of time and afterwards exposed to the light for five hours the distal retiuular cells have retreated until they lie around the proximal ends of the cones. The nuclei of these cells lie immediately upon the nuclei of the proximal retiu= ular cells, and it is interesting to notice the pigiuented body of each cell folded on itself. In section the pigment takes the form of plaited black ribbons. When the eye is again stimulated by light the ribbon unfolds as the cell travels forward. These cells are called by Exiier the iris pigment, since they regulate the brightness of the retinal image in much the same way as the vertebrate iris does. x.— SUMMARY. In the review, including Sections v-ix of Part Second of this memoir, the principal embryo- logical facts have been summarized, and it will now suffice to recapitulate only some of the more interesting results. 456 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. ix. — (1) Tbe majority of the Alphei luitcli as zoea-ltke larvse, while two species are known, A. heterochelis and A. saulcyi, in which the metamorphosis is abbreviated. This shortening of tiie metamorphosis appears to be directly related to the habits ami environment of the species. A. heteroclielis has one metamorphosis at- Beaufort, North Carolina, a more abbreviated develop- ment at Key West, Florida,, and, it we are right in considering the Bahamau form as a member of this species, at Nassau, New Providence, the metamorphosis is complete or unabbreviated. The Nassau form of Alpheua snnlri/i either has the metamorphosis greatly abridged or it hatches with all the external characters and the instincts of the, adult. When we inquire into the modes of life of these species we. tiud the remarkable fact that the Nassau Alphcus sK SCIENCES. 457 is differentiated from a distinct epithelial cell, the nucleus of the cell becoming the .nucleus or germinal vesicle of the egg. Some of the epithelial cells enwrap I he developing ovum and form the follicle, or pocket in which it is lodged. The chorion or inner egg membrane is the. direct secretion product of the follicnlar cells. (7) -In Homarus and Palinurus the character <>f the germinal epithelium is somewhat different from that of Alpheus. The outline of individual cells is obscured and the germinal epithelium extends inward from the wall, in the form of radial sheets or folds, between which are reentrant blood vessels. There are a number of gerniogenal areas corresponding to the folds in which the ova originate. During growth the eggs gradually pass from the center toward the, periphery. In the germogeu the cell outlines are obscured.* (8) The yolk arises within the cell protoplasm, and in Ilomarns degenerating nuclei occur in the ovarian stronui, and it is probable that a certain number of nuclei degenerate and enter into the food yolk of the egg. In the lobster also some of the follicle cells develop into gland-like struc- tures which eharacterix.e the mature ovary. They appear to have a direct relation to the growing- eggs, but their true significance has not yet been ascertained. In about two weeks after the eggs have been extruded these structures have almost wholly disappeared. (9) 1 have observed a single polar body in a section of the egg of Stenopus, in which the male and female pronuclei were present, and two polar bodies in the ripe unextrnded egg of the lob- ster. In lobster's eggs also which failed of extrusion at the proper time, and which eventually degenerate in the ovary, I find that the nucleus is at the surface. It has the appearance of a female pronncletis. It is thus probable that the polar bodies are often, if not always, given oil before the eggs are laid, t Segmentation in Al/iln'/i* ininor. — (10) The segmentation in AlpltrH* minor is in some respects anomalous, and the conclusion seems to be warranted that we have here a cast.- of a mitosis, unlike anything which has been hitherto described in Crustacea. Unfortunately my material is not at present sutlicient to enable me to say in exactly what way the usual process of cell division has here been modified. Delamin&tion. — (11) The segmentation has been thoroughly reviewed in Section v, and it is unnecessary to repeat the details. 1 wish to call attention, however, to the fact that at 'the close of segmentation in the lobster some, of the blastoderinic cells delaininate and their products pass into the yolk. In Alfil/i'iifi mtiilci/i a similar migration of cells from the superficial to the deeper parts of the egg occurs, but in this case it was not determined whether this migration was preceded by delamination or not. These cells appear to originate in greatest number over that side of the egg which corresponds in position to the embryonic area. It seems possible that these cells may represent a primitive endoderin, the function of which has been usurped. In the lobster they speedily degenerate. Imagination Stage. — (12) The invaginatiou stage, which soon follows, results in the admis- sion of more cells into the yolk and in the formation of an organ called in this memoir the ventral or thoracic-abdominal plate. Cells also continue to pass into the yolk from the ventral plate While cells are constantly being subtracted from the plate, it is constantly receiving new recruits from the surface, owing to the activity of cell division in this region. \Ve thus have in Alpheus a multitude of migrating cells, derived originally from three sources: from the blastoderm, from the cells which are first invaginated, and from those which originate later from the ventral plate, after all trace of the superficial pit has disappeared. Germ-layers. — (13) These migrating cells, which are collectively called >• the wandering cells" in Section VII, spread to all parts of the egg. While it is perfectly obvious that these bodies represent rnesodermic and endodermic tissues, it is not so easy to determine what particular cells give rise to this or that layer, nor is it easy to decide in many cases, from a superficial study, whether migrating cells may not be derived from the ectoderm in various parts of the embryo. This sub- 'The structure of the ovary of the lobster lias been recently described by JJumpn.M in a .kt.-iilr.il paper upon thu embryology of this species. He lias called attention i.. tin- folded character of the ovarian epithelium, which is so marked in the youn'g or immature, ovary. (The Embryology of the American Lobster, by Ili-nuon Carry Humpns, Jovrn. of Morpholoyy, Vol. v, No. 2, 1891.) t Polar bodies have been recently described in the . xierual eggs of the lobster by Bumpus. <>],. cit. 458 MEMOIES OF THE NATIONAL ACADEMY OF SCIENCES. ject is fully considered in Section vii, the general results being : (1) That it is not possible to decide what part the primary yolk cells play in Alpheus, for reasons which have been already considered; (2) that the great bulk of the cells which migrate forward from the area of invagiua- tion and attach themselves to the embryo, or proceed to the peripheral parts of the egg arid take up a position at the surface, are undoubted mesoblastic elements ; (3) that those cells which give rise to the endoderinal epithelium in the egg nauplius are derived largely from cells which migrate in a posterior direction from the area of in vagiuation ; (4) that degeneration, followed by the death and dissolution of the chromatin and cell protoplasm, is characteristic of the wandering cells at about the beginning of the egg-nauplius period. The mesoblast has become a well-recognized layer before the endodermal epithelium has appeared. (14) The egg, with centrally moving cells which have budded from the blastoderm, may be compared with tlie_p/««K?« stage of Coeienterates, and the internal cells may represent the primi- tive endoderm. According to this view, the invagiuatiou stage has no reference to an adult gastrula-like ancestor, but is a purely secondary condition, which became so impressed upon the ancestors of the present Decapods that it has remained in their ontogeny. In the majority of Decapods which have been studied theiuvagiuatiou has no direct relation to the mouth or anus, or to the alimentary tract. The conditions which are present in the cray- fish cannot be regarded as typical or primitive. (15) In Alpheus and Homarus the primitive mouth arises on a line between the rudiments of the first pair of antenna?, but these appendages are never post-oral. The hind gut originates as a nearly solid ingrowth, apparently at a point considerably behind the position of the pit due to the first invagiuation, and is formed one or two days later than the mouth. Cell dissolution. — (16) The degeneration of embryonic cells is treated at length in Section vi. It is remarkable that the early segmentation stages of Alplmun minor are attended with the degen- eration of protoplasm. The chromatin residues remain for some time in the yolk, and eagerly react upon dyes, but gradually lose this power and eventually enter into the general nutrition. (17) Degenerating cells appear in greatest force in Alpheus, Astacus, and Homarns at about the egg-nauplius stage, and from that time their numbers begin to wane. They appear in one instance before the differentiation of the germinal layers, and are not confined to any one Liyer at a later period, but in Alpheus saidcyi they are most characteristic of the wandering cells, which represent mesoderm and endoderm. The " secondary mesoderm cells " and " white-yolk elements " which have been described by Eeichenbach, are to be regarded as degenerating cells. Degenera- ting cells also occur in connection with the " dorsal plate." The Eyes.—(\&} The details of the structure and development of the eyes and nervous system are fully reviewed in Sections Tin and ix. The eyes and optic glanglia are derived from the optic disk, in the formation of which there is in Alpheus no proper invagiuation. The thickening of the disk is accomplished by emigration from the surface and by the delaminatiou of superficial cells. An area of active cell division can be distinguished, which corresponds to the invaginate area of the optic disk of the crayfish. The cells which migrate from the center give rise to the rudiment of the optic ganglion. The disk grows out in the form of a lobe, and becomes differentiated into an outer retinal layer and an inner gaugliomc layer. The eye proper is differentiated from the retinogen, which is primitively a single layer of eotoderrnic cells. (19) I am inclined to regard the "compound eye" not as an aggregate of simple eyes, as its name implies, each of which is due to a hypodermal infolding, but rather as a collection of differ- entiated clusters of ectoderm cells, originating in a single epithelial layer. (20) The absence of light has no appreciable effect on the development of the eye pigment, but in PaUvmonctes the distal retinular cells respond very promptly to the action of light. If the light is excluded from the eye, these cells migrate outward and enshroud the proximal ends of the cones, sending out pseudopodal prolongations to the cornea. When the eye is again stimulated by light the pigment immediately retreats from the surface, and the cell takes the form of a plaited black ribbon, leaving the cones free. ADELBERT COLLEGE, MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 459 XI. — REFERENCES. 1. BATE, C. SPENCE. On the Development of the Crustacean Embryo, aiM the variations of form exhibited in the Larvai of thirty-eight genera of Poilopthalmia. Abstract. Proo. Roy. Ho<\, Vol. \.\iv, p. 375. 1876. 2. . Report of the Present State of our Knowledge of the Crustacea. Sept. of 50//i. Meet, nf Jlrit. .las., pp. 230-241. 1880. 3. . Report on the Scientific Results of the Voyage of H. M. S. Challenger during the years 1873-1876. Zoology, Vol. xxiv. Text and separate volume of Plates. Report on the Crustacea Macrnra dredged by H. M. S. Challenger during tlie years 1873-1876, pp. i-xc, 1-942. 76 figures and 157 plates. 1888. 4. BOAS, J. E. V. Kleinere carcinologische Mittheilungeu. Ueber den ungleicheu Entwicklungsgang der Salzwasser- uud der Siisswasser-Form von Pala'monetes various. Zool. Jahrb., Bd. 4, H. 4, pp. 7113-805. Taf. xxm, Holzschn. 1-4. 1889. ft. BOBRETZKY, N. Zur Embryologie des Oniscus mnrarius. Zeit.f. m'ss. Zool. Bd. xxiv. pp. 179-203. Taf. xxl, xxii. April, 1874. 6. — — . Development of Astacus and Pahemon. Kirw, ls?:j. Abst. by Hoyer in Hoffmann and Sehwalbe'a Jahres- berichte. Bd. II, pp. 31^-318. 1875. 7. BROOKS, W. K. The Metamorposis of Alpheus. Johns Hopkins University Circulars, No. 17. 1882. 8. - — . Lucifer, a Study in Morphology. Phil. Trans. lioy. Soc., Pt. I, pp. 57-137, Pis. 1-11. 18»2. 9. . A preliminary abstract of researches by W. K. Brooks and F. H. Herrick on the Life History of Stenopus. Johns Hopkins Unirersity Circulars, Vol. vill, No. 70. 10. BRUCE, ADAM TODD. Observations on the Embryology of Insects and Arachnids, pp. 1-31+10, Pis. i-vi. .Memo- rial Folumc, Johns Hopkins 1'itiversily. Baltimore, 1887. 11. BULLAR, J. F. On the Development of Parasitic Isopoda. Phil. Trans. Roy. Soc., Vol. CLXIX, pp. 505-5(J1, Pis. XLV-XLVII. 1878. 12. CLAUS, C. Uutersuchungen ilber die Organisation u. Eatwickelung von Brauchipus und Artemia, nebst ver- gleicheuden Bemerknngen iiber andere Phyllopoden. Arbeit. Zool. Insl. JFien, Tom. VI, Heft 3, pp. 207-370, Taf. i-xn. 1886. 13. DIETI,, M. J. Untersuehungeu uber die Organisation des Gehirns wirbelloser Thiere. n Abth. (Crustaceeu), ]ip. 1-20, I Taf. Siteb. der k. Akad. der Wissenscli., Bd LXXVII, i Abth. 1 878. 14. DOHRN1, ANTOX. Untersuehungeu iiber Ban uud Eutwicklung der Arthropodeu. Zweites Heft, Zur Eutwick- luugsgeschichte der Pauzerkrebse (Decapoda Loricata), pp. 105-128, Taf. x. Leipzig, 1870. 15. ERDL, M. P. Eutwicklung des Hummereies von den ersten Veriinderungeu im Dotter an bis zur Reife ues Embryo, pp. i-x, 11-40, Taf. l-iv. Miincheu, 1843. 10. FAXON*, WALTER. On some Young Stages in the Development of Hippa, Porcellana, and Piniiisa. Bull. J/«s. Comu. Zool., Vol. v, No. 11, pp. 253-268, Pis. i-v. 1879. 17. - — . On the Development of Paln-monetes rulgaris. Bull. .!/«*. Contp. /no/.. Vol. v, pp. 303-330, Pis. i-iv. 1879. 18. GUERIN, M. F. E. MEJTEVILLE. Auimaux articules a Pieds articnle's, in Histuirt I'liy«i«'. n-nl Geog. Survey of the Territories, Vol. iv, Xo. 1, Art. vn, pp. 189-199. 30. . The Development of the Compound Eye of Crangon. Journ. of Morphology, Vol. I, No. 1, pp. 49-64, PI. II. September, 1887. See also Zool. Anz., Jahrg. IX, No. 234, pp. 597-600, 1886, and Amer. Naturalist, Vol. xx, pp. 862-867. October, 1886. 31. . The Development of Crangon Vulgaris. Second paper. Bull. Essex Inxt., Vol. xvni, Nos. 7-9, pp. 99-153, Pis. I-H. Salem, July, 1886. 460 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 32. - The Development of Craugon Vulgaris. Third paper. Hull. Essex lust., Vol. xxi, Nos. 1-3, pp. 1-42, Pis. i-m. 1889. 33. KKIEGKR, K. R. Ueber das Ceutraluerveusysteui des Flusskrebses. Ztit. f. wias. Zool., B. xxxni, pp. 1-70, Taf. xxxi-xxxm. 1679. 34. LEI!EI>IXSKI, J. Einige Untersnchnngeu iiber die Eutwickluiigsgcschichte der Stekrabbeu. ' Iliol. i , i,i'lill., B. X, Nos. ."> ;in«l »!, pp. 178-18o. 1890. :;.">. LEHEBOULLET, A. Keclierches sur le Mode tie Fixation des CEufs aux t'ansM-s put tes abdominales dans les Ecrevisses. Ann, de Sti. \'' Recherchos d'Embryologie Compared sur le DeVeloppemeut dn Broehet, de la Perche rt de I'F.crevisse. Mem. /ii/:!<. /KI i- ilirrrx Siirant* ti r Acml. des Sci. de I'lnst. Impcr. de France, T. 17, pp. It; -sn5, 1'ls. 1-G. Paris, 1862. 37. LUCY, W. A. Observations on the Development of Agelena uaevia. Bull. Mns. Comp. Zool., Vol. xn, No. 3, pp. 63-I03. 1'ls. 1-1-2. 1886. 38. Lui>WHr, H. Ueber die Eibilduug im Thie.rreiche. Arbeit, aus dem zoolog. -zoo torn. lust, iu \Viiivlinig, B. i, pp •JS7-.MO, Tat'. XIII-XV. 1874. 39. MAYER, PACI,. /nr Entwieklungsgeschichte der Dekapodeu. Jeiiuische Zeil. Xntiirn-ix«, R. xi, pp. I ->--jr,:i, Tat. xia-xv. 1877. in \li,i:i:s. iiKox\ M,I. i '. \,,N. Eine neue Art vou. Blastodermbildung bei den Dekapodeu. /on/. An-:., No. 101, pp. v!l-23, Fi-s. 1-,-i, v Jahrg. 188'i. II. MOLC.AX, 1'. II. A Coiiti-iltutioii to the Embryology ami Phylogeuy of the Pyeuogouids. Miulics from Hiol. Lal'ij nf tin' .full nx ll<>iikinx Ciiim-silii, }>]>. 1-7G, Pis. i-vill. Baltimore, Jauuary, 1891. 4^. Mi'n. LEI:, FIIIT/.. I'.ila-mon Potiuna. Em Beispeil abgekiirzter Verwaudlung. Zool. Anz., No. f.vi, pp. l.VJ-157. L880. -13. Beriehtigiing, die Verwaudlung .l.-s Pal.i-.inon Potiuuu lietrottoud. Zool. Anz., No. '>:>, p. 'j::3. 1880. II. Ni su.U'M, .lusEi'ii. L'Embryologii' L'Enibry.il.i^i,. ,lc< Mysis Channel. 'o. .In-li. ii,' Znol . Kjcper. rt Ci:ii<;;ili; •>(( Ser., T. 15, Nos. l->, |>|>. l-r.\-20i, Pis. i-xn. l.s-7. 4C. PACKARD, \ - Notes on tin- Early Larval Stagt-N of the Fidi ll.-r Cral. anct of Alplieita. Imer. .\ ~tit«rali>it. Vol. XV, pp. 7.-I-7KI, Figs. 1--I IT. I'AUKKK, (i. II. Tin' Histology and Iirvt- lupim-nt of the Eye in the Lobster. Bull. MII.I. I'mn/i. Zool., Vol. XX, No. 1, pp. I i;-.', I I'ls. I -!MI. 4^. Tin- t'omponinl KV 111 Crustaceans. Hull. Man. (',1,11/1. /uul., Vol. xxi, No. :.', ]ip. 4f.-140, 10 Pis. 1891_ 4'.t. I'ATI IC.N. WILLIAM. Thr Development of Phryganids, with a Preliminary Note on tin- 1 1. '\.-lop ntof Blatta Germanica. >>unrt. Jmtrn. Micros. Sci, Vol. xxiv, New Sili»ltiiiii, Vol. I, pp. 193-^27, 1 PI. 18.-7. t>2. KATHKE, HEINKICII. Untersnchungen iiber die Bildnng nud EntwicUelung des Flusskrebses. pp. 1-97, 5 Pis. Leipzig, 1SJ9. 53. - — . Zur Eutwickelungsgeschichte der Dekapoden. Ai-chiv. f. Naturgeseh., B. vi, pp. 241-241). H40. 54. EEICHENBACH, HEINRICII. Stadien zur Eutwicklnugsgeschichte des Flusskrebses. Abhandl. Sencktnbvri/: Xatur- forsch. (if.ii -llxcli.. !',. 14, pp. l-137+'2, Taf. i-xiv, i«-iv«, ivh. Frankfort, l-^H. fif). KYI>I:I:. .1. A. The Metamorphosis of the American Lobster, Homarus americauns, H. Milne Edwards. Amer. Xdlin-aHsl, Vol. XX, pp. 739-742. :">r,. Scin.MKE\viTsi'ii, WLADIMIK. Einige Bemei-knngen iiber die EntwicklungsgeBchiehte des Flnsskielises. y.ovl. Anz., No. 19."., pp. 303-304. 1885. .".7. SIIEI.IMIX. LILIAN. Ovum in the Cape and New Zealand P.-ripat us. IJiKirl. ./num. Mil-rut., .s'ci., Vol. xxx. New Ser., pp. !-•£>, Pis. I-HI. 5s. SMITH, SIDNEY I. The Early Stages of the American Lobster (Humanm americanns— Edwards). Trans. C'ouu. Acail., Vol. II, I't. -2, pp. 3."l-3Sl, 1'ls. xiv-xvm, +Figs. 1-4. 1873. Earlier papers in .tin. .Imini. .svi. and. li -la, 3d Ser., Vol. in, pp. 401-401;, 1 PI., June, 1872, and in Kept. U.S. Fi*]i ('mum. t>f J-'ixh inn/ I<'inhi'rii'K on the Condition of the Sea Fisheries of the Southern Coast of New England iu 1871 and 1872, pp. 522-:i37. Washington, 1873. 59. - Keportim the I )i-capod Crustacea of the Albatross Dredgiiigs off the East Coast, of the I'liiteil States dnriug the Summer and Autumn of ISs.l. Ann. /,'./~lirrii'« for Irs.'i. pp. 1-101, Pis. l-xx. Washington, 1886. 60. VAX BEXEDEN, EDOUARD, and BESSELS, fiiiiLE. Memoire sur la Formation du Blastoderm chez les Amphinodes, lea Lerneens et les Cop<5podes. M6m.Cour.et M6m. de Sav. I5tr. publ. parl'Acad. Roy de Belgiqiio, T. xxxiv, pp. 1-59, Pis. i-v. 1870. b'l. VIALLAXES, II. Etudes histologiques et organologiques 8ur les centres uerveux et les orgaues des sens des auiniaux articul^s. Premier ruetnoire. Le ganglion optique de la laugouste (PaUnurus vulgaris). Ann. des Sci. Nat. Zool. et Paleoutol. MEMOIRS OF THE NATIONAL ACAMKMY OF SCIENCES. 4(11 G"2. — — . Etudes histologiques et orgauologiijues sin li\s rrntres JHTVIMIX rl !<•» m-jrum's di-s srn-i dr« ;IJIIIH,MI arth-iili:.s. OiiH|in. mr M. nmn-e. I. Le Cerveau du Criquet (ffidipoda coertilescens et tul,,^!,,,,,, ii,ili<-H*). II. l''iiii|i:i.riiiK()ii ilu rrrviMii ilrs crusl:u'i:s rt drs insri-tes. III. l,n rnrvr:m H l.i morpliolo^ie >ln si|in'li-tte i c i,h.-ilii|iia. Ann. ilcs Sri. .Nilt.. Zm>l. ot r'alfontnl., 'I', iv., NIIH. 1-:'., vn.Si-r., pp. I -I1.'!), |'|H. i-t;. i-,x7. ('.:'.. WAIASK. S. (In (ho Morphology id' thr Compound Kyes of Arthropods. Htuitirx Jr. Iliiil. l.tili'ij <>f lln .li'lnis Hop- kins I'll in mil I/, Vol. IV, pp. ->7-:',::i, 7 I'ls. IS'.IO. (54. WALDEYER, w. Kn-rstnrU mid Ki. (if). WiT.sux, IlKxity V. On Ilif I'lvoiling SeasoiiH nf Marine Animals in tin- liali:im;i,s. .Inlm Iliijiiiii:: I iiinrsili/ I'ii-i-nl.ii:*, Vol. vni, No. 70, p. 38. 6(i. WOOD-MASOX. Stridulating Crustacea. Keniarks of Mr. Wood-Mason at the November meeting of the Kntiniio- lni;ii-;i] Siifirty of London, \nlii i\, Vol. XVIII, p. 53. 1878. l'.7. \VHKELKR, WILLIAM M. The Einlii'vology of Blatta Germa.nica and Doryphora I>ecriiilinr:H:i. Am. .Inurn. J/ni- l>holotjy, Vol. in, No. '^, pp. 291-386, Pis. xv-xxi. 1889. APPENDIX I. THE LITE HISTORY OF STENOPUS. e this paper was written Chun has described (Die pelagische Thierwelt in grosseren M steefeu, Bibliotheca Zoolog'ica, I, 1888) a small transparent crustacean which he calls clavigna. It dccurs at the surface and also at various depths clown to liOO M. A comparison of his description and figure (Taf. iv, Fig. H) with the Stenopus larva shown in I'ls. iv and x of this memoir shows that Chun's Meicrsia flm-ii/na is undoubtedly a Stenopus larva, a little older than the one shown in PI. x. (\V. K. B.) It is suggested at the bottom of page 340 that the cement by which the eggs are fastened to the abdomen may possibly come, from the oviducts. According to recent observations of Cano (Mittheil Zool. Ktat. Neapol., ix, 1891; abstract in Journ. Roy. J/iV. iillti f/reenlandica of Seba, which appears under several names, may be the same as .s'f<'/«>/)».v Itixpifliut. "The genus," says Bate, ''thus appears to inhabit regions so widely apart as Greenland in the north, the Bermudas and Mediterranean in the west, and the southern coasts of India and the Fiji Islands in the east. It has been found in the cold waters of the Arctic regions as well as in the warm shallow waters of the tropics, but, despite this cosmopolitan range, it has not been recorded as having existed in any geological formation." While the conclusion that the Arctic form is a Stenopus may be correct, it seems highly im- probable that it is specifically related to sti'n»/mx liixpidus. There is no evidence at least to show that this is the case. Bate figures a late egg embryo of Stenopus (Fig. 40, p. 212), and erroneously concludes that the animal has a short metamorphosis and that it hatches as a "Megalopa." He also gives a draw- ing (PI. XXIX, Fig. 2, v.) of the first larva of Xpongiola vennata (a prawn which is placed by Kate in the family Stenopidsv). This is clearly not a zoea, but a protozoeii, as is better shown by the sketch of the recently hatched larva (Fig. 42, p. 21C) by von Wille.moes Suhin, and the strong resem- blance which it bears to the protozoea of Stenopnn l/isjiidus is very striking (compare with PI. vii, Fig. 11, of this paper). 462 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. According to Bate the branchial formula of Stenopus was first elucidated by Huxley in his memoir on the classification of crayfishes (Proc. Zool. Soc., London, 1878). There are six pleuro- brauchife; eleven arthrobrauchia?, five of which are anterior and six posterior; one podobrauchia, and nix mastigobranchi*, of which the first is the only efficient appendage. Speuce Bate states that after careful comparisons he failed to find specific differences between specimens from the Eastern and Western Hemispheres. ERRATA. As numerous errors have unavoidably occurred in this paper, I will correct the more important of them. Page 341, line 8, for -'PI. vn " read PL x. Page 341, line 13, for " Lesneur" read Lesuenr. I 'a go 343, line 2, for " cells have spread more rapidly at a given point on the egg" read cells have increased more rapidly over a given area of the egg. Page 343, over table for "Temperature 80° F.," read Temperature of air, 80° F. Page 344, lines 11, 16, 31, 32, and 47, for " Fig. 10" read Fig. 11. Page 344, line IX, for " largely developed " read highly developed. 1'age 345, lino 20, for "Fig. 10" read Fig. 11. Page 345, line 30, for " Fig. 1 1 " read Fig. 10. Page 345, line 38, for " the first and second maxillipeds" read the second and third maxillipeds. 1'age 340, line 31, for " larger than telsou" read longer than telson. Page 347, line 16, for "xn and Fig. 40" read xni and Fig. 39. Page 347, line 20, for "and 38 " read and 39. Page 347, lines 37 and 41, for " PI. xi" read PI. XII. Page 347, line 40, for " Figs. 43, 45'' read Figs. 43, 44. Page 347, liue 47, for '• Fig. 47" read Fig. 46. Page 348, line 15, omit " errinem larve." Page 348, for lines 32-34 read: Body nearly cylindrical; tergal surface, covered with spines. Carapace \viih prominent laterally compressed rostrum and distinct cervical and branch io-cardiac grooves. Outer antenna1 with long bristle-bordered scale bent under the inner antennae toward the middle line. Second inaxillipeds with setig- eroiis lamina, attached to endopodite. Page 348, line 46, for " a marked transverse fossa" read a marked cervical groove. Pa triuiao-:r. The fnugns has no mycelium, Imt is composed of single cells of various sizes. In a section like that shown in Fig. 199 nearly one hundred large cells or cysts can be counted, and it, is seen that the peripheral parts of the embryo are packed with them. These embryos were alive, although the embryonic cells were considerably altered from their normal condition, where they came in contact with the parasitic growths and showed traces of degeneration. The parasitic bodies arc mainly (1) large naked cysts or encysted cells, and (2) very small spore-like bodies. The naked cyst (c. s., Fig. I'M) is a thick shell which has collapsed and curled up with the escape of its contents. It is yellowish and is unaffected by staining reagents. The surface of the cyst is covered with very uniform, short projections or tubercles, which refract the light in a characteristic way. Other encysted cells contain a protoplasmic reticulutn (cs1), and there are very similar but smaller bodies which are either naked or possess but a slight cuticular wall. These encysted bodies just described possibly represent zoosporaugia, and give rise to the myriads of minute spores which occur in close relation with them. The spores (Fig. 199, up., represented by small black dots) are minute, oval, and highly refractive. In the eye and other organs certain nuclei take up the stain very eagerly and refuse to part with it. These are probably the nuclei of em- bryonic cells which have undergone modification. Occasionally- one of the cysts appears black (c.s2), which is due mostly, if not wholly, to refraction. According to Goebel, reproduction in the Chytridieai is effected by means of swarm spores. Besting cells occur, which germinate and become sporangia, producing large numbers of swarm spores. Some forms, like Chytridium, have no mycelium. Its single cells, which live on or within the host plant, after reaching their full size become zoosporangia. These give rise to swarm spores, which are liberated into the water. The Chytridiecc are described as parasites on other aquatic plants, Fungi, Alga?, and Phanerogams. According to De Bary resting spores are known to occur in certain species. These develop directly into sporangia or produce them after a short intermediate stage, and appear to resemble the sporangia in size and in possessing a warty cellulose coat. APPENDIX III. Some early abstracts of this work (Alpheus : A Study in the Development of Crustacea) were iucladed in the Introduction published in the Johns Hopkins University Circulars, No. 97, April, 1892. The part relating to the embryology of Alpheus was here printed in its unrevised form, and differs materially from the results of later studies which are given in this memoir. While this work was in press it was thought best to change the name Alpheus minus of Say to the correct form, Alpheus minor. As some of the pages were stereotyped before this correction was made, both forms of the name appear. ADELBERT COLLEGE, Cleveland, Ohio, May, 1892. 464 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. PL MI; I. Al/>li/'ii>i tuiiinr, drawn from life by \V. Iv. Brooks. (Enlar.ucd oii;lit dinincters.) Dorsal view ol';i N|iccinnMi of tin- ^r;iy variety <>!' (roiiiiiliu-ti/liin <-hin«jnt, twice natural size. Ptatel. W.K.Brodks,del. ALPHEUS MINUS AND GONODACTYLUS CHIRAGRA. S. Mis. 94 30 466 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE II. Alpheus heterochelis, drawii from life by W. K. Brooks. (Four times life-size.) W.K.Bmok,-i,del. ALPHEUS HETEROCHELIS. 468 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. PLATE III. Gonodactylus ohiragra, drawii by W. K. Brooks. Adult female of the green variety, in her burrow, with eggs, twice natural size. Memc.irs Natioual Academy of Sciences, 1889. PLATE II', K. Hl'iinkx. ill I GONODACTYLUS CHIRAGRA. 470 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE IV. Adult male and female of AlpJieus saulcyi, var. brevicarpus, from Nassau, New Providence. FIG. 1. Lateral view of female from green sponge. x3£. FIG. 2. Dorsal view of the same. x3J. Parts only of the ovaries are visible in Fig. 2, while the eggs, which greatly distend the abdomen laterally, show plainly between the bases of the swimmerets. In Fig. 1 the small chela is bent downward, the position in which it is usually carried. In Fig. 2 the chela} are represented in the attitude of defense. The dactyle of the left " hand," or large chela, is raised preparatory to striking. FIG. 3. Small male. L = |gin. x7£. Drawn under slight pressure, owing to which the antennal spines and the antennular exopoditeshave assumed an unnatural position. The pos- terior margin of the carapace is more correctly represented in Fig. 1. /if lie IV /' H. Herri-fir, ALPHEUS SAULCYI,VAR. BREVICARPUS. 472 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE V. Dorsal view of the adult male Stenopits hiapiiiux. Nassau, N. P., June, 1887. L = 1J in. L. first antenna, exopodite— 3J in., endopodite— 3g in. L. second autenna = 4|£ in. xlf. Excepting the brilliant pigtneut bands the body and appendages are nearly white, and could be better represented against a black background. The arching flagella of the antennae are greatly foreshortened, and the spines and setae are of necessity unduly empha- sized in a pen and ink drawing. 474 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE VI. FIG. 1. Part of section of egg, showing the male pronucleus. The female pronucleus lies nearer the center of the egg, is less regular in outline «uid has less perinuclear protoplasm. A single polar body (not represented) is seen in this section. It lies close to the sur- face of the egg, beneath the membranes, not far from the male pronucleus. It appears as a small mass of chromatin, which stains quite as intensely as the nuclei. Egg about 6 hours old. x 276. FIG. 2. Section of egg with four nuclei, none of which are at the surface. x!52. FIG. 3. Part of same section, showing the nucleus and surrounding protoplasm and yolk, x 276. FIG. 4. Lateral section, cutting yolk segment on a level witli the disk-shaped nucleus. Compare Fig. 5 a. Eight-cell stage. Age about 12 hours. x276. FIG. 5. Section through egg in eight-cell stage. Compare Fig. 6. Age about 15 hours. x!52. FIG. 6. Surface view of egg in the third segmentation or eight-cell stage. The egg membranes have been removed. The nuclei lie at a deeper level than they appear in the draw- ing. Compare Fig 5. x 78. FIG. 7. Section of egg in the fourth segmentation stage. Sixteen cells. x!52. FIG. 8. Fifth segmentation stage. Age, 19 hours. Cells not yet at surface. x!52. FIG. 9. Imagination stage. A solid ingrowth of blastodermic cells has taken place at ly, where a slight pit is formed. The section cuts obliquely through the in vagiuate cells, x 152. REFERENCE LETTERS. a, perinnclear protoplasm. Ch, chitinous egg envelopes (removed, except in Fig. r>). Ep, ectoblastic cell. Ig, shallow pit of invagiuation. y. c., yolk spherule. Plate VI Ch. Fig.6. FI&.8. Fig.9. STENOPUS HISPIDUS. 476 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE YTT. FIG. 10. Left second maxilla, of larva at the point of hatching, before the first molt, x 276. Coin- pare with Figs. 25 anil 21. FIG. 11. First swimming larva, after the first molt, seen from below. Pigment cells, brown. L= jJjft, in.- (measured from tip of rostrum to median notch of telson). Length of rostrum = T$ „ in. x 70. FIG. 12. Right first maxilla of first larva, seen from the outer side. Setae rudimentary. Compare Figs, lit, 25. x276. FIG. 13. Telson of larva before first molt, seen from below. Compare Fig. 11. The set* are invaginated and covered with a loose cuticle. x276. FIG. 14. Right first maxilliped of larva on the point of hatching, seen from the outer side. Setre- tnvaginatod. Compare Figs. 22, 25. x276. FIG. 15. Labrum and right mandible of larva, seen from above. x276. FIG. 16. Right third maxilliped of larva on the point of hatching, seen from the outer side. Com- pare Fig. 25. x 276. REFERENCE LETTERS. a, outermost spine in telson of larva at the point of hatching. d, equivalent of a in first locomotory larva. Lb, labrum. f'i'S.10. Plalv VI Fig. 12, STENOPUS HISPIDUS. 478 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE VIII. FIG. 17. Second larva after second molt. L=-1^L0- in. x 70. FIG. 18. Left mandible, outer side, of second larva, x 276. FIG. 19. First maxilla of second larva, x 276. FIG. 20. Telson of second larva, seen from below, x 70. FIG. 21. Left second maxilla of second larva, seen from the outer side. x276. FIG. 22. Eight first maxilliped of second larva, seen from the outer side, x 276. REFERENCE LETTERS. g g, gastric gland. .... the dotted line in Fig. 20 pointt* to the outer Hpine, the equivalent of d, Fig. 11. 1'ldlc VIII Fig. 22. STENOPUS HISPIDUS. 480 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE IX. Advanced 'pelagic larva of Stenopus hixpidus, from Beaufort, North Carolina, drawn from life by W. K. Brooks. Jin ft- IX. STENOPUS S. Mia. 94 31 482 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE X. Dorsal view of a larva like the one shown in Plate IX, drawn from life-by W. K.'Brooks. Pin IP X WK.. Biaaks, del. STENOPUS 484 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XL FIG. 25. Embryo nearly ready to hatch, released from the egg membranes. x70. Some food yolk is still u n absorbed ; swimming hairs very rudimentary; compare Fig. 11. FIG. 26. Eight first antenna of older larva. x70. FIG. 27. Profile view of hinder end of abdomen of same larva. x28. FIG. 28. Second maxilla of same larva. x276. FIG. 29. Mandible of same larva. x27G. FIG. 30. First maxilla of same larva, x 27fi. FIG. 31. Portion of third maxilliped of same larva. x70. FIG. 32. Terminal segment of second pereiopod of same larva. x70. FIG. 33. First pereiopod of same larva, x 70. FlG. 34. Portions of third, fourth, and rudimentary fifth pereiopods of same larva, x 70. REFERENCE LETTERS. A. I, first auteuna. A. II, second antenna. AM., mandible. ifipd. I, in, first and third maxillipeds. If., rostrum. Th., TJi. 1, first maxilliped. Th. 3-Th. 5, third to fifth maxillipeds. 1, first maxilliped. Plate XI H. F.rf.fferrUJr.det Th.5. ' STENOPUS HISPIDUS. 486 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XII. FIG. 35. Older larva, taken iu tbe tow-net outside of Nassau Harbor May 7, 1887. L=9mni. L. of eye-stalk=2'"m. L. between eyes=4.7rai". x 15. The long flagellaof the antenna- are conventionally represented to bring them into the plate. They trail above and behind the animal as it swims through the water. I 'I ate -V//. Figure 35. F.H.Herridc.del. • STENOPUS HISPIDUS. 488 MEMOIRS OP THK NATIONAL ACADEMY OP SCIENCES. PLATE XIII. FIG. 3(5. First maxilla, outer side. Adult male, x 15. FIG. 37. Lateral view of carapace of adult male. x5. FIG. 38. Left mandible, outer side. Adult male, x 14. FIG. 39. Stalk and portion of flagella of left first antenna, seen from above. Adult male. x5. FIG. 40. First right pleopod of male, outer side. x!4. FIG. 41. Left second antenna with flagellum cut off near its base. Adult male". Seen from above, x 5. FIG. 42. Second maxilla of adult male. x!4. FIG. 43. Eight first maxilliped from outer side. Adult male, x 14. FIG. 44. Right second maxilliped, from outer side. Adult male, x 14. FIG. 45. Right third maxilliped, from under side. Adult male. x5. FIG. 4G. Right first pereiopod, under side. Adult male. x5. FIG. 47. Right fifth pereiopod, under side. Adult male. x5. Plate XIII **TT>^; "• - : "K -. -1- A - " , • . - - STENOPUS HISPIDUS. • 490 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. PLATE XIV. Metamorphosis of Gonodactylus chiragra, drawn from life by W. K. Brooks. FIG. 1. Dorsal view of egg just before hatching. FIG. 2. Front view of the same egg. FIG. 3. Side view of tbe larva immediately after hatching. FIG. 4. Side view of the same larva after the first molt. FIG. 5. Side view of the same larva after tlie second molt. FIG. <». Dorsal view of the larva at the beginning of its pelagic life. XIV. 2. GONODACTYLUS CHIRAGRA. • 492 MEMOIK8 OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XV. Metamorphosis of Oonodactylus chiragra, drawn from life by W. K. Brooks. FIG. 7. Dorsal view of the larva shown in PI. xiv, Fig. 3. FIG. 8. Ventral view of the same larva. FIG. 9. Dorsal view of the larva shown in PI. xtv, Fig. 4. FIG. 10. Ventral view of the larva shown in PI. xiv, Fig. 5. FIG. 11. An older larva in dorsal view. FIG. 12. Same in ventral view. FIG. 13. Raptorial claw of a still older larva. I 'I nlc XV GONODACTYLUS CHIRAGRA. . 494 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XVI. Metamorphosis of Alpheun. drawn by W. K. Brooks and F. H. Herrick. FIG. 1. Third larval stage of Alplieus minor from below, drawn by W. K. Brooks. FIG. 2. Second larval stage of Alpliens minor, about one-tenth of an inch long, drawn from below by W. K. Brooks. FIG. 3. Telson of the Nassau form of Alpheux licti >•<>• hiiin during the second larval stage, drawn at 10 a. in.. April 17, 1887, by F. H. Herrick. FIG. 4. Second antenna of Alplieus minor during the first larval stage, from the iuside drawn by W. K. Brooks, May 13, 1881, D. 2. (Zeiss lenses.) FIG. 5. First and second maxilla? of the Nassau form of Alpheus heterochelis during the fourth larval stage, drawn by W. K. Brooks from a sketch by F. H. Herrick. The larva at this stage is shown in PI. xn, Fig. 3. FiG. 6. First maxilla of Alplieus minor during the first larval stage, drawn at Beaufort, June 2, 1SS1, by W. K. Brooks, 1). l>. FIG. 7. Second maxilla of the same larva. FIG. 8. Mandible of the same larva. xvi Brooks, X- Hf.rru-J< ,

Hitri>lu>xin . Km. ~a. Line to .show length of larva. L.=4""". FIG. 3. Head of young from same brood. Four days old. X'r>2. FIG. 4. Right tirst pereiopod of larva of A. xitnlcyi, var. brevicarpus, before the molt preparatory to stage shown in Fig. 1. Seen from inner side. x52. Swimming hairs of exopodites rudimentary. FIG. 5. Egg embryo of .4. suiilcyi, var. lonyicarpus, nearly ready to hatch. The large chela of the left first pereiopod is conspicuous below the, antennae. x46. FIG. 5«. To show natural si/.e of the same. Slightly too large. Dimensions: riroxTcb inch. FIG. <>. First and second maxilla of first larva (Kig. 1) before preparatory molt. The parts are gloved with the embryonic skin, which is usually cast off at the time of hatching. X227. FIG. 7. Left first pereiopod of same, seen from inner side. x52. FIG. 8. Third larva of Aljihrux xuulcyi, var. lin-ricar/Hix. From same brood as second larva, Fig. U. Not over twenty-eight hours old. Food yolk not wholly absorbed. x2G. FIG. !l. Telson and rudimentary uropods, seen from below. x32. I. ALPHEUS SAULCYI 1 506 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXII. FlG. 1. Left second pereiopod of first larva of A. xiiiiiciii, var. />/•<'/•/>« /•/«/?<, from inside. xi. FlG. 7. Right second antenna of tlie same, seen from below. x<>4. FlG. 8. Right first antenna of the same, seen from above. x"4. FIG. 9. Second antenna of young of AlpJifitu miulcyi, var. l>n-ricttri>/if,, six and a half days old. x 64. FIG. 10. First antenna nf the same. x*>l- FIG. 11. Head of male of ./ m< ah-yi... var. loiu/icoriiiix, troni "loggerhead" sponge. Mediau spine of rostrum waufing. Drawn from lite. L.=5.5mm. x31. FK;. 11!. Mandil)le of first larva of A. nault-yi, var. hr<'ri<-u$. Xl!">a. FIG. 13. Left second antenna of male of A . xaulcyl, seeu from below. No, 8 of Table I, p. 385. x 33. FIG. 14. Left second antenna of female of A. *.<;ux, shown iu Fig. 17, at time of hatching. Compare, this with the same appendage of ihe adult. x64. FlG. Hi. First pereioiiod (small chela) of young of .1 . mnilci/i, var. !»•< ricarpus. From.greeu sponge. Compare this with Fig. 3, PI. xxiv. x<>4. FIG. 17. Front of a larva of .1. suulcyi, var. liiiit/ii-tifpint, which was hatched April 25. Drawn uuder pressure ; eyes slightly distorted. Equivalent to the ordinary third larva, Fig. 8, PI. xxi. x 04. FIG. 18. Part of stalk of right first antenna of male of A I /then.? fMulcyi, seen from below, showing the aural scale. The median <•>«• is seen <>n the right, between the basal segments of the auteuuules. From No. 8 of Table I. x'-!6. /'/it if XXII 7. !•' H.Herri rJr,(M. ALPHEUS SAULCYI. •- - . 508 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PT.ATK XXIII. FIG. 1. Eight second pereiopod of male of Alpheux xanlcyi, var. brevicarpws, seen from the outer side, x 33. FIG. 2. Terminal segments of right fifth pereiopod of the same. x33. FIG. 3. Left mandible of the same, seen from the outer side. x64. FIG. 4. Left first antenna, and left compound eye. of the same, seen from above. x33. FIG. 5. Left third maxillipetl of the same, seen from outer side. x33. FIG. 6. Right second maxilliped of the same, seen from the outer side. x64. FIG. 7. Right first maxilliped, seen from the outer side. x64. FIG. 8. Right second autenua of the same, seen from above. x33. I 'Id IL- A'AV/A Fig.,'). ALPHEUS SAULCYI. 510 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXIV. FIG. 1. Right fifth pereiopoil of male of Alphcitx .taulcyi, var. brevicarpus, seen from outer side. X33. FIG. 2. Small chela of malt' of A..i«nli'>ii, var. loHyicii.rpim. From No. 9, Table I. x33. FIG. 3. Small chela of male of A. .in ulci/ i, var. bri-rii-nrjutx. From No. 8, Table I. Compare with the typical form of the other variety, shown in Fig. 2. x33. FIG. 4. First pleopod of male ol'.l. xiixli-i/i, fV;>m "loggerhead" sponge. x33. FIG. 5. Left first pleopod ••>( i'emale of .1. si<>i. /'.. . linr.ii LC ;ilMloliilll:ll phlli-. II. '_'., lilc.oil cell. III. .S. , l>!(irnl Minus. I'll., rhltiuulls i-iijiMhcll. I'll, jr., limiting membrane of blood Minns. <'l. N.. (ivai-ia.ii Hiroma. K. (., c'.nii i'nilirli-. F. (.'.. i-.jrg fdllirlc. Iri-r.. i;c'i iiiii-.'iial area. /. E.. nvariau stronia (undififerentiated). a. n1— o~, nuclei of ovarian stroraa and developing e.ngs. <>. D., optic disk. II. L., optic lobe. r. P., yolk pyramid. T. S., yolk spherule. f'ac. , yolk vucuole. XXV. Fig. 2. J?.E Tig.4 v > '«Jr i.»l.^gp>^.; ,. x , ^/,, *f7$Xl e VMST r5'^. , ;-». ., // •''- --...t OL \ BC- oik BIS. F.H.Herrick,clel. HIPPA, HOMARUS ANDALPHEUS. Mis. 94 33 514 MEMOIRS OF TR2 NATIONAL ACADEMY OF SCIENCES. I'l.ATK XXVI. FIG. 9. Sectiou tbrougb segmenting egg of Alpheus saulcyi. Eight cells present. Yolk unseg- inented. Egg membranes diagramatically represented, x 70. FIG. 10. Surface view of the same. Sixteen cells present. Yolk pyramids formed. The periph- eral nuclei are seen through a thin layer of yolk. x70. FIG. 11. Transverse section through the immature ovaries of Alpheus. Ovary taken in Juno, from female "in berry." Diameter of largest ovarian egg, one one hundred and sixty- eighth of an inch. Diameter of extruded egg, one fiftieth of an inch. Contents of ovarian egg, one thirty-seventh of that of the extruded egg. X-!81. FIG. 12. Sectiou through segmenting egg of Alpheus minor, from Beaufort, North Carolina, show- ing uests of nuclei, x 70. FIG. 13. Swarm or uest of nuclei, like those of preceding figure. x281. FIG. 14. Sectiou through egg of Alnhvm minor, cutting segmentation nucleus. Nucleus elongated, with irregular, indefinite boundary, x 70. KKFKRENOK I.KTTKKS. .41. ('., a)iment:ir\ canal. B. C., blood cell. B. S., blood space (possibly unnaturally distended). i'h., chitinotis egg envelopes. D. A., dorsal aorta. c, e', young ova. F. E., ovarian stroma. F. E.', follicnlar epithelium. /•'. C., follicular epithelium. Her., germogeri. Her.*, position of germogen in ovary, with i>va nearly ripe. '.'. I"., germinal vesicle. '). H". , ovarian wall. SS, swarm of nuclear bodies. Vac., yolk vacnole. I'it.. vitellogen. x, cell shown in Fig. oO. Y. P., yolk pyramid. I tut i- XXVI Fig 10. ch. B.C OW: F.E: Fig.//. Vit FE. GV. Figl2. FiglZ. Fig.14 ac. F.H.tferru-Jr.ffe/.. ALPHEUS SAULCYI AND A MINUS. 516 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXVII. FIG. 15. Section of egg of Baliaimin variety of Alpheun heterochelis in typical yolk pyramid stage Sixty- four cells present, x 70. FIG. 16. Segmentation nucleus of egg of A. sauleyi, nearly central in position. x277. FIG. 17. Section of an egg of A. sauleyi, wbich was normally laid but unfertilized, showing the female pronucleus. x70. FIG. 18. Degenerating nuclei containing spore like bodies, from tlie egg-uauplius embryo, the structure of which is shown in Pis. XLI-XLIII. x610. FIG. 19. Blood cells of adult Alpheus. x 610. FIG. 20. Endodermal cells from the ventral wall of the primitive alimentary cavity of Astacus fluviatilis. After Reicheubnch (54) Taf. Vin, Fig. 67. This is taken from the egg- uauplins stage to show the origin of " secondary mesoderrn." The elements here marked »«', A- are described as cells which have originated from the eudoderm, and completed their metamorphosis into ordinary mesoderm cells. These may be com- pared directly with b, FIG. IS, and s, s~, Fig. 21, from the egg-uauplius of Alpheus sauleyi, and are rather to be regarded as nuclear bodies in the earlier stages of retrogressive metamorphosis. x256. FIG. 21. Part of transverse section through the foregut of the egg-nauplius of Alpheus sauleyi, to show the degenerative cell products. x610. REFERENCE LETTERS. A. T.S., altered food-yolk. Ch., chitiuous egg membranes. ec., ectoblast. i., nuclear body, with vesicular chromatin mass. k, k1, I, m, m1"3, nuclear products in yolk. Mes., mesoblast. JV., JV.1, nuclei of eutoblastic cells, n., uucleolus of cntoblastic cell (not clearly shown). 01., oil drop. Bel., protoplasmic reticulum. S, s, a-., degenerative products. Sep.', cleavage plane. Std., foregut. Vac. , yolk vacuole. r., yolk. T. P., yolk pyramid. Y.S., yolk sphere. I 'It i In .VAT/7. Vac. Fig. 18. d • e... Y.S J.YS. -Mes. Std. ec. Rei ALPHEUS AND ASTACUS. 518 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXVTTI. FIG. 22. Part of section of segmenting egg of Alpheus 'minor from Beaufort, North Carolina, show- ing nuclear body in clear area. x277. FIG. 23. Swollen, probably degenerating, elements, from segmenting egg of A. minor. x277. FIG. 24. Section through base of yolk pyramid of egg of Palatnonetex rulgarin. About sixty-four cells present. x'-'77. FIGS. 25, 2G. Two successive sections through clear area in segmenting egg of Alpheus minor, showing degenerative products and nuclear bodies in process of breaking up. x277. FIG. 27. Part of section through segmenting egg of Pontonia domestica, before cleavage of the yolk. The egg contains three nuclei, one of which is seen to be in karyokiuesis. X277. FIG. 28. Part of section of an Alpheus egg in same stage as that shown in Fig. It. Cell dividing indirectly and in horizontal plane. x277. FIG. 29. Section of egg of Alfiheus minor, probably at close of segmentation. x277. FIG. 30. Enlarged view of cell .*•. and part of section shown in Fig. 9. x277. REFERENCE LETTERS. Ch., egg membrane!!. _V., nucleus. /'. A., protoplasmic area. /'. .iV., perinuclear protoplasm. SC, SC'-5, degenerating cell products. Y, yolk. 1'. B., yolk hall. .1'. 6., yolk sphere. Vac., vacuole. Plate XXVIII. Fig.23. sc • I • sc. PJ. Fig.25. Vac. r -SC. ....CJi. S'(T Fig. SO. Ch. ALPHEUS AND PONTONIA. 520 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXIX. FIG. 31. Part of section of egg of Bahaman variety of Alphem heterochelis, showing two yolk pyramids. Same stage as Fig. 15. Sixty-four cells present. x!477. FIG. 32. Part of transverse section of egg-nauplius of A. saulcyi, showing the fold of one of the antenna3 and the niesoblastic cells and degenerative products contained within it. x610. FIG. 33. Wandering cells in yolk above the same embryo, showing protoplasmic union. xGlO. FIG. 34. Part of section of egg of the BahamaH heterochelis in egg-nauplius stage, showing wander- ing cells, which have left the yolk and have attached themselves to the superficial ectoblast. The nuclei are flattened against the surface, but are clearly distinguished from the epiblast. x610. FIG. 35. Part of transverse section of older embryo, showing blood cells and wandering mesoblast cell (Me*.). Eye-pigment beginning to form. x610. FIG. 36. Part of longitudinal section of embryo shown in Fig. 153, to show the degenerative prod- ucts of the dorsal plate, x 610. REFERENCE LETTERS. App., appendage. A. ¥. S., altered yolk. B. C., blood corpuscle. Ch., egg membranes. C. P., united pseudopodia of two wandering cells. Ect., ectoblast. Ep., spindle-shaped nuclei of surface epiblast. Me*., Mes.1, mesoblast. Mil., muscle cells. Pn., cell protoplasm. PL, coagulated blood. s, «', degenerative cell products. Sep., inner wall of yolk pyramid. S. W., outer wall of yolk pyramid. Y. C., wandering cells. T.S., yolk sphere. Vac., vacuole. xxix 51. Sep. PN. -S.W. Fig.33. Ect jMes. YS. Ecf. ALPHEUS 522 MEMOlltS OF TUE NATIONAL ACADEMY OF SCIENCES. PLATE XXX. FIG. 37. Part of section of egg before iuvaginatioii stage, showing primary yolk cells. All the figures ou this plate, excepting Fig. 46, refer to the Babamau form of Alpheus hetero- cheUn. x277. KM;. 38-44. Consecutive sections of the same egg, showing the progress of the primary yolk cells in their migration from the blastoderm to the central parts of the egg. x70. FIG. 45. Section through the same egg, showing sernidiagramrnatically tjhe structure of the yolk. X70. FIG. 40. Section through the egg of A. .•mi/lci/i at a slightly later stage, but before iuvaginatioii. The blastodermic cells lie at the surface, the primary yolk cells toward the center of the egg. Traces of the primary yolk cleavage are still seen at the surface, and a secondary cleavage has occurred below the surface, x 70. KM;. 47. Surface view of the side of egg, corresponding to the germinal area in nearly the same stage. X70. KM;. IS. Tangential section, showing blastodermic cells of same egg. x277. REFERENCE LETTERS. ii. a' 7. L't'lls migrating from blastoderm into the yolk. I'.il. I'., blastodermic cell. Ch., eggshell. G. D., embryonic area. Sep., yolk cleavage plane. 1'. £., yolk ball. Fig.37. I late XXX Ch Fig.58 Fig.40. ,2 a Fi^.42. Ch. Fig. 45. MC. F. H. Herri fl<, <'"' ALPHEUS 524 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXXI. FIGS. 49-55. Serial transverse sections through the embryo in the iuvagination stage. In the most anterior section the germinal area (O. Z>.) is traversed, and in Fig. 53 the shal- low depression in the middle of the invaginate area is cut through. In Fig. 54 (G. D.) we see the forward extension of the invagiuate cells and the first trace of the thoracic-abdominal plate. The distinction between the primary yolk cells (Figs. 49, 52, 53-55) and the invagiuate wandering cells (b, 62"1'', Figs. 5H-54) and their product, which now begin their migrations, is very plainly shown. In Fig. 50, which cuts the shallow pit at the surface of the invagiuate area, we see the amosboid cells with large granular nuclei making their way from the bottom of the pit into the depths of the yolk. Figs. 49, 52-55, xl!5. -Figs. 50, 51, x291. REFERENCE LETTERS. b, b'^b", in-wanderiug cells derived from the invaginate cells aud their products. Ch., egg capsule or shell. Ep., ectoblast. G. D., embryonic area. /. C., iuvaginate cell. Ig., pit formed by the in vagi nation. Y. B., yolk ball. P. Y. C., primary yolk cell. Y. S., yolk sphere. .V.VA7 Y.S o &.J '*.•• ^ * Ep. Ep. YC. G.D F.ff.Herrick,del,. ALPHEUS . . 526 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATK XXXII. * FIGS. .">(», 5!l, 60. Longitudinal serial sections through the entire embryo in the stage shown in Fig. 58. Fig. ~>9 is median. The primary yolk cells (P. Y. C., Fig. 60) can still be distin- guished from the wandering cells derived from the invagination (*S*. Y. 6'., Fig. <><>)• In Fig. .">!> a primary yolk cell ( /'. Y. r.1) is in the metakinetie stage of division. Traces of the primary segmentation of the yolk are still present, and the secondary yolk segmentation is very marked in the neighborhood of the wandering cells. Where the .shell (Ch.) is not removed it is seen to be considerably distended and to have, epiblastic cells sticking to it. showing the ('lose adherence which normally exists between the egg membranes rind the egg. x 11">. Fid. 57. Section through egg, cutting germinal disk just before invagination. Twenty and one- half hours older than yolk-pyramid stage seen in Fig. 1">. x 73. Fi<;. -"iS. Surface view of embryo after the appearance of the thoracic-abdominal plate and the optic disks. The shallow depression which marked the invaginate area has disappeared. Its approximate position is indicated by ///. Compare /.', eCtoderMl. fr. /)., germinal disk. Iy., pit of invagination. L. I'd., lateral ventral hands. O. !>., optic disK. P. M., wandering rolls, wen In-low surfarr, mining oil' from ventral )>late. P. Y. C., P. T. C.\ primary yolk cells. »j>. , yolk cleavage plane. .S. F. r., ,S. Y. C.1, wandering cells derived from the invaginate cells and tlieir p T. <'ii.. cell area uniting optic disks. T. /,'., yolk hall. Y. C., primary yolk cell. I lute XXXII Fig.56. Jib.P st:c. s.rc \---Ep. --PY.C. Ch.. ALPHEUS • 528 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXXIII. FIGS, 61, 62, 68, 69. Transverse sections of embryo in stage shown in Fig. 58, PI. xxxn. Fig. 61 cuts the thoracic-abdominal plate, and Figs. 68 and 09 involve the optic disks. Pri- mary yolk cells (P. Y. C.. Fig. 69) are still plainly distinguishable. xl!5. FIG. 63. Portion of median longitudinal section of the same stage. The larger and clearer nuclei in the invagiuate area represent the mother cells of both mesoderm and endoderra. The yolk ball or secondary yolk segment is characteristic of this stage, x 291. FIGS. 64-67. Consecutive transverse sections of left optic disk of same stage before any thicken- ing has occurred. The-most anterior section is represented in Fig. 64. x291. REFERENCE LETTERS. Ab. I'., veutral platf. Ch., eggshell. Ep., ectoderm. Ig., invagiuate cavity. L. Cd., lateral ventral cord. O. D., optic disk. P. >'. C., primary yolk cells. Sep., yolk cleavage plane. S. V. ' ., in-wandering cells derived from ventral plate. T. /(., yolk ball. I 'laic XXXI II Fig. 61. CJi., t PY.C. £>. Cft, Fig. 68. QD. rs. * . PY.C. ,; ! Sep. Ep. F:H.Herrick,dei. ALPHEUS . S. Mis. 94 34 530 MEMO1RF OF THE NATIONAL ACADEMY OF SCIENCES. PLAIK XXXIV. (Stage IV.) Fms. 70, 71, 73, 74. Longitudinal serial sections of the entire embryo in the stage shown in Fig. 72. In Fig. 73 the plane of section is nearly median. The primary yolk cells are now generally indistinguishable from the other wandering cells. Compare cut, Fig. 11, which shows the distribution of the wandering cells at this stage, xllo. FIG. 72. Surface view of embryo in Stage iv. Rudiments of the mandibles and first pair of an teuuse are present. An area of cell ingrowth in the optic disks ((!. M.) is characterized by the large si/e of the nuclei. From them and their products the optic ganglion takes its origin. Some of the surface cells on either side of the middle line were acciden- tally cut away, xli'.ll. REFERENCE LETTERS. .I. (I.), proliferating center af firs! antenna. Ah. P., ventral plate. C. jU., proliferating area of optic dink. £j)., ectoderm. /,. Cd., lateral ventral cord. MA., proliferating center of mandible. O. D., optic disk. T. Cd., transverse cord uniting optic disks. Y. A'., yolk ball. Y. C., T. C.', wandering cells. Fi&lO L.Cd. IT?-/* yc- % If. ?»"'•« °9°90 F. H H*wJi ,< ALPHEUS. .' 532 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. I'l.ATK XXXV. (Stage IV.) Fiir. 75. Mediau longitudinal sectioii of the entire embryo. One of the wandering cells, which is approaching the surface of the egg opposite the thoracic abdominal plate, is in the process of division. xl!5. FIGS. 76-83. Consecutive serial transverse sections through the left optic disk of same, to illus- trate the earliest stages in the thickening of the disks. The most posterior section (Fig. 83) cuts the first pair of antenna?. x291. FKI. 84. Transverse section through the middle of optic disks, x 115. FIG. 85. Transverse section through thoracic abdominal plate, showing the multiplication of sur- face cells, by which the plate is increased, and cells below the surface ( Y. C.) which pass into tin- yolk. x-91. REFERENCE LETTERS. • A. ( 1 ), first antenua. .4b. P., ventral plate. Ch., eggshell. C. M., proliferating area of optic disk. «•., /'I-.'--1, fctoderniic rells of ventral plate (Fig. !*). ec., (Fig. 80) ectoderuiic cell of optic disk. KJI., I'rtoderin. O. O., O. D'., optic disks. Ret , protoplasmic reticiiluiii. T. I'd., transverse cord. Y. R., yolk ball. F. ('., Y. C.'~2, wandering cells. T. S., yolk sphere. I'lnlc XXXV ^ p.p. C.M. ec. ' Fig.81. Kp. red. Yf '' YC. O.D. Y.B. Ch. Y.C. Ep. : H. Horru-k , del - , ALPHEUS 534 MEMOlliS OF THE NATIONAL ACADEMY OF SCIENCES. PLATI: XXXVI. (Stage V.) FIGS. 86, 87. Parts of longitudinal sections ol' embryo seven or eight hours older than that shown in Fig. 71.'. The optic disk (0. D.) is cut in Fig. 86, and in Fig. 87 its inner part is involved, with the outer border of the thoracic abdominal plate. There is no sharp demarcation between the protoplasm and tlie yolk, as is indicated by the dotted lines under the embryonic layers. x201. Fi<;s. SS-S9. Longitudinal serial sections through the entire embryo, somewhat younger than the last, and six hours older than that represented in Fig. 72. The optic disk is sectioned in Fig. 90 through its central proliferating area (C. M.), and the rudiments of the three naupliar appendages appear in Fig. 811. x 115. Fie. 91. Transverse section cutting optic disks of embryo about nineteen hours older than that of Fig. 72 and twelve hours older than that represented by Figs. 86, 87. Wandering cells ( Y. C.1) have traveled, to remote parts of the surface, and karyokinetic figures (Y. ('.-', Fig. 89) prove that they are in active division, x 115. s REFERENCE LETTERS. ,1. ( / ), rudiment, of lirst antenna. .1. (// ), rudiment of second antenna. All. I'., ventral plate. .Ijiji., area, of appendages. O/i., eggshell. C. M., proliferating area of optic disk. «•., migrating cctolilast cell. EI>. , ectoderm. Mel., rudiment of mandible. <>. !>., optic disk. £., product of degenerating chromatin. $t. .1., sternal area. 7'. Ctl., transverse cord. )". <.'., F. C.l-'\ wandering cells. F. 6'., yolk sphere. I'/nlc XXXVI s. O.C- Jb.p StJ. Ep. Fig 90 Fig. 89. Jlb.P rc. Fig.9/. T.Cd. C.M. . ALPHEUS 536 MEMOIRS OF THE NATIONAL ACADEMY OP SCIENCES. PLATE XXXYTT. (Stages V-VI.) FIG. 92. Part of transverse sectiou, showing the structure of the keel-shaped ventral plate, aud indicating the origin of mesoblast from the surface of the latter. x291. Fi<;. !>3. Surface view of embryo with buds of naupliar appendages. The intermediate area (St. A.) is covered by a single layer of ectoderm. The invagination of the mouth has not yet appeared. Some nuclei of cells which lie immediately below the surface, especially in the thoracic abdominal plate region, are represented. x291. FIGS. 94-95. Transverse sections of embryo, belonging to the same series as Fig. 92, to illustrate the structure of the thickening optic disks. Degenerating nuclear products («.', .«.-, Fig. 95) are present, and two cells are seen delaminating side by side in Fig. 95. REFERENCE LETTERS. A. (/), rudiment of first antenna. A. (II ), rudiment of second antenna. Ab. /'., ventral plate. ''. M., proliferating area of optic ganglion. Kft., ectoderm. Md., rudiment of mandible. Mes., wandering cells (mesoblast) attached to ectoderm. O. G., rudiment of optic ganglion. O. D., optic disk. «.', s.2, products of degenerating chromatin. St. A., sternal area. T. Cd., transverse cord uniting optic disks. T. C., wandering cell; Y. C.\ wandering cell degenerating. Plate XXXVII. ^hp. Eft. \ Fig. 93. T.Cd. Ect. F.H.fferrick,del. ALPHEUS , |« . . • 538 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XXXVIII. (Stage VI.) FIGS. 9fi-97. Longitudinal sections through the embryo shown in Fig. 03. The more lateral of the two, Fig. 96, cuts the middle of the optic disk, and shows the large cells of the prolif- erating area, one of which is caught in the act of dividing. Fig. 97 cuts the inner portion of the optic disk and the. ventral plate on a level with the budding mandible (Md.). x295. FIGS. 98-100. Longitudinal serial sections through an embryo six hours older, from the same batch of eggs. The mouth (Fig. 9S, 8td.) has already appeared. The mesoblast, formed chiefly from wandering cells, is well established on either side of the middle line of the body, and is well seen under the folds of the appendages (Fig. 100, Meg.) into which it extends. The- mesoblast represented by the lower layers of the ventral plate is still being increased by the migration of cells from the surface of this plate, as is indicated by cell ec., which is interpreted as a superficial cell about to migrate (in Fig. 98). In Fig. 100 a cell at the surface of the optic disk is in the act of delami- nating. Large numbers of degenerating cells and their products arc now encountered (K.C.,x.). X295. REFERENCE LETTERS. A. ([), bud of first antenna. -I. (II), bud of second antenna. Ab., Ab. P., Ad. P., ventral plate. App., area of appendages. ('. ,\l., proliferating area of optic disk. re., ec.[, migrating and dividing cells at- surface of ventral plate. /,'(/., Ep., ectoderm. Md., rudiment of mandible. Men., mesoderm. O. C., optic ganglion. '). I)., optic dink. «., s.'~2, products of degenerating chromatin. •S'. C., »s'. <_'.'"-', cells in various stages of degeneration. xi. A., sternal area. Std., stomodieum. T. Cd. , Transverse sheet of ectoderm uniting optic disks. ¥., T. S., yolk spheres. I'lnle A'.YAT/// JIM) Fig. 96. Md. Fig.98. CM MA. 1 ALPHEUS 540 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PL AH; XXXIX. (Stage, VI.) FIGS. 101-105. Serial longitudinal sections of early nauplius embryo, twelve and one-half hours older than that represented by Figs. 08-100, PI. xxxvm, and eighteen and one half hours older than the, stage represented in Fig. 93, PI. xxxvu. The thoracico-abdotni- nal fold or papilla is now forming, apparently by the ingrowth of the surface ectoblast (Fig. 104, A1>. €.). Fig. 102 is exceptionally favorable in showing the undoubted delamination of two cells standing side by side at the surface of the optic disk («,•.). The common radial division of the ectoderm of the thoracico-abdominal region and other parts is illustrated by the cell ec.} of the same section. Fig. 105 cuts the straight tubular stomodseuui. x295. FIG. loti. Longitudinal median section of embryo several hours older than the last. A deep, narrow, transverse furrow (Ab. G.) now abruptly separates tin- thoracico-abdominal papilla from the. sternal area lying between it and the stomodicum. x-91. FIG-. 107. Transverse section through the optic disks, from same stage. Cell delamination in this region is again met with. x291. FIG. 108. Longitudinal lateral section .through entire egg, showing the distribution of wandering cells, and the relations of the embryo to the ovum. The eggshell is unnaturally dis- tended. An inner molted membrane is present, as is better shown in Fig. 106 and Fig. 104, M/>. In Fig. 108 a large cell is seen at the surface, and below this a large, cell followed by a row of similar cells. The first two cells possibly represent a budding ectoblast and mesoblast, and the rows of cells at the surface and below it are possibly derived from them, x 115. REFERENCE LETTERS. • A. (I), Inul of first antenna. .1. (II), Inul of second antenna. Ah., thoracico-abdominal papilla. Ah. ('., transverse superficial furrow by which folil of Mm thoracico-abdominal process is formc'd. .1. )'. iS., products of degenerating chromatin. />'. /., building zone. ''/(., eggshell. ' '. .I/., proliferating area of optic disk. ri. N., cells on ventral side, of yolk next to optic disks, probably representing mesoblasf derived from \\ andcring cells. re., ?<•.', dividing ectoblastic cells. /•'c/., /•-'/<, ectoderm. ,>/.. wandering cell at surface behind thoracico-abdominal fold (Kig. UiH). Mli., embryonic molt. Mil., rudiment of mandible. Mf*., mesoblast. <>. !>., optic disk. ". '.'.. optic, ganglion. 0. N. G., brain. I'd., proctodicmn. N.. N. , products of degenerating chromatin. A'. C'., ilegenerat.ing cells. •*>''. A., sternal area. *lll., Momod;i'lllll. ).. 1. S., yolk. ATA7.V. Fi£10% OD. Ml). Ect. - Wes. Fig. 103., •-4YS. AW. -ctfes. ~Hd Jib.— Mes. StA M.C.-- Ch. lib Jib. CfL ALPHEUS 542 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XL. (Stage VI.) FIG. lO'.l, Sketch of egg-nauplius. Anus not so clearly .seen 111 surface view, as represented in this anil the following figure. Mouth on a level with auteunules. x 72. FIG. 110. Sketch of older embryo. Appendages all bending backwards and inwards toward middle line, x 72. FlG. 111. Egg-nauplius less developed than shown in Fig. 109, but from same batch of eggs. The position of the mouth, which is post-antennal from the first, is now on the middle line between the antenna* and the antennules. The probable position of the anus is indicated, but it could not be clearly seen. The bud which represents the endop- odite of the antenna, is just appearing on the right side, x 157. FIG. 112. Oblique transverse section, through egg-nauplius of a common shore crab of Beaufort, North Carolina, probably Srxiiniiii. x2M(i. FIG. 113. Median longitudinal section, through a similar embryo. The egg membranes are not naturally shown. The yolk is diagramatically represented. Wandering cells occur in it ( 1". C.), and in Fig. 113 degenerative products (Dey.) are met with. x28G. REFERENCE LETTEliS. A., anus. A. (I), anteuual bud. A. (I), antenuular bud. .1ft., thoracico-al>domiual fold. ' '!>., eggshell. Deg., degenerative cell products. Iff., ectoderm. Gl., ectoblast of neural plate. H., inesoblast cells, forming rudimentary heart. Bg., hind gut. Lb., labrum. J/(J., tiiaridilmlar bud. Men., incKoblast below surface. <). . (,'., riidiinuiitiiry brain. .*>'/<(., stoniodit'Uiii. I'oc., vacuole. )". C., wandering cells. Xuuibers 114-125 mark tue jilaUL'B of the tiausverse and lougitudiiial si'ctions rc]ircseiitcd mi I'ls. XLI-XLIII. /Kite XL Fig.109 .OL. Fig.110 Fig.lll. g ^ ° o -^ ^ *cr<£r-v°-o°^ A € QC^ 122 - J23 . 11*- & "'*°'5->::.i t_&oe-*S*'e ^ e «^"«^P^ ©r o : -^W-<% &w =^ gj^ V'Ttv'f 5- e©*0*^191^ • •• * «* © o ^ « ^o . »tf * 9 ^ & s© ^i ^^ ^ f eo © <& T> r? O F. H. Herri.,*, Jrl . ALPHEUS AND SESARMA' 544 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATK XI, I. (Stage VI.) FIGS. 11 1-1 1!>. Plane ol section indicated iu Fig. Ill, which is from au embryo a trifle less advanced. The lobular condition of the enlarged optic disks is well shown in Figs. 114, 115. In Fig. 1 14 a delaminating cell (<'<:) at the surface <>f the optic lobe is cut, and in Fig. 115 a superficial ectodermic cell next the brain is dividing perpendicularly. The intimate fusion of the braiu and the optic ganglion is seen iu Figs. 115, 116. Fig. 1 17 cuts the stomodit'iiiii passing through the mouth and the antenna1,. Mesoblast is already well established in the. pockets of all the appendages, as indicated at au earlier period. Degenerating cell products (Ull. A. (11), autennular buil. I. )". /?., alteration products of yolk. ''/. X., cells partially covering brain, derivatives from yolk-wandering cells. KC. , surface cell of ectoderm dividing horizontally. Kct., Kp., ectoderm, t/rf., mamlibnlar bud. Mil. . L., optic lobe. AY/., protoplasmic reticulum. .•>'. N.1, products of cell degeueration. N. i>. G., brain. xt., vetoileriu. '»/., gaiigliouic rudiiueut. H., rudiment of heart. IIij., intestiue. f.h., labruin. Mil., I'lnbryouii' mult. .l/c«., mesoderm. Mo., mouth. I'll., ivijioii of i>roctoilii'iil invagiualiuu. «., s1., products of cell degeneration. •s'. ' '., wandering cells, probably in early stages of degeneration. >'. <>. <-.. rudiinrut of brain. tt. ./., sternal area. )'., yolk. F. ''., 1 ".(''.'-••', \vauderiug cells. )'. .S'. , yolk spherules. Vac., vacuole. XL111 J.YS Fig. 125. Kp. SAG. .Y. Sid Fig. 126. \ Pd. .-YC. Fig. 127. Vac Ect. CJv Ab. F.H.Herrick.deL. .. . ALPHEUS SAULCYI 550 MEMOIRS OP THE NATIONAL ACADEMY OF SCIENCES. PLATE XLIV. v (Stage VII.) FIG. 128. Transverse section through embryo, in the region of the first maxilla. Nervous system not yet differentiated from the skin. x234. FIG. 123. Lateral longitudinal section through optic lobe and extremities of antenna;. The differ- entiation between the retinal and ganglionic parts of the eye is very clear. Meso- blastic cells, representing rudimentary muscles, are seen attached to the body wall. x234. FIG. 130. Surface view of embryo of this stage, with buds of four post-ruandibular appendages present. The antenna- are covered with a hairy exuvium, which was probably stripped off from the anteumiles in this preparation. The mouth is concealed by the labriim, which nearly meets the thoracico-abdominal fold. The anus is situated nearly at the extremity of the latter, which is slightly emargiiiated. x!37. FIG. 131. Median longitudinal section in the series from which Fig. 129 was taken. x234. FIGS. 132-135. Serial transverse sections through embryo in similar stage. The germinal layers, definitely established in the egg-nauplius stage, are clearly differentiated at this period. An incomplete layer of elongated cells (probably mesoblastic in origin, com- ing from wandering yolk cells), Mes., Figs. 131, 132, is seen between the yolk and the neural thickening, from which the nervous system is iu process of development. In Fig. 134 rudimentary muscles suspend the stomodseum to the body wall, and in Figs. 133 and 135 the three germinal layers can be clearly seen. The heart is repre- sented by the space filled with mesoblast and serum, between the entoblastic lamella and the ectoblast. x234. REFERENCE LETTERS. A. I, first antenna. A. II, second autenna. Al., thoracico-abdominal fold. A. r. ,$., alteration products of the yolk. Ect., ectoderm. End., endoderui. ';. f.., rudiment of optic ganglion. (it, A. II, autennular ganglion. Lt>., labrnm. Mes., inesoderm. .Vo. , mouth. .)/«., rudimentary muscles. MX. I, first maxillary bud. U. E., retinal portion of optic lobe. S. 0. G., brain. Std., .stoniodremn. /'fate A7/T Fig. 128. ^m^^^M:^W^mw^ ^»4%*^epf-^^^' .,, ,J ./fw Fig.130. Jfe. Mes. S(d. End. .- ''-M^-^-iX-iSTTO.^ J.I. Jfes. End ALPHEUS 552 MEMOIKS OF THK NATIONAL ACADEMY OF SCIENCES. PLAIT, XLV. (Stage VIII.) FIG. 136. Lateral longitudinal section of embryo in stage intermediate between YII and VIII, represented in surface view in Fig. 110. To this phase also belong Figs. 137, 144. and 145. Fig. l.'ili is to be compared with the slightly older embryo in Fig. 129. Blood cells (B. V.) and other wandering cells are here seen .settling down upon the body wall. A wandering cell is also seen nearly in contact witli the optic ganglion, x-41. FIG. 137. Transverse section of embryo in same phase, just behind the lev,el of the first antennae, showing the relations of the wandering colls at this period to the embryo and egg. x 61. FIGS. 1, '58, 139. Serial longitudinal sections through embryo in Stage VIII. Fig. 138 should be compared with Figs. 136 and 129. All the ganglia of the nervous system, at least as far back as the eighteenth segment, are marked at the surface by deep constric- tions. The ganglia of the nineteenth and twentieth segments are less distinct. The ganglia of the eleventh segment lie in the angle made by the thoracico-abdomiual flexure, Wandering cells occur in the yolk, lint are less abundant, and the products of cell degeneration, which enter into the general nutrition, have mostly disappeared. X241. FIGS. 140-143. Parts of sections taken at various points on the surface of the egg (series to which Figs. 136, 137, 144, 145, belong), remote from the embryo, to show the role of certain wandering cells which reach the surface and represent mesoblast. In Fig. 140 two cells (ins., ms.1) are -partially flattened against the surface, but here, as in Fig. 142, the wandering cell ms. is clearly distinguishable from the spindle-shaped ectoderm cell on the left. Compare Fig. 34. x241. FIGS 144-145. Serial longitudinal sections through the embryo and entire egg to show the distri- bution of the wandering cells. Certain wandering cells not yet flush with the surface, enter into an organ — the dorsal plate (TJp.) — which is characteristic of a later stage (Fig. 153, Dp.). The strictly superficial cells of the dorsal plate are probably in all cases ectoblast, and some of the wandering cells degenerate before they reach the surface ectoblast. There seems to be a general dispersal of wandering cells from the vicinity of the thoracico-abdominal fold. The wandering cells which appear in this part were taken from four consecutive sections, including that represented in the drawing. x61. REFERENCE LETTERS. A. I, lirst antenna. A. II, second antenna. ««., lower margin of optic lobe. A.s. a., superior abdominal artery. IS. (.'. , blood corpuscle. b. HI., basement membrane. oh., eggshell. Dp., dorsal plate. End., endoderm. G. IV-SriII, segmental ganglia. Gl., gaugliogen. B., hc:iit. hil., liypodcrniis. ffff., hindgut. »ie«., mesoblast. mo., month. ms., ms.1, wandering cells at surface. 0. L., optic lobe. I{t., retinogen. Std., stouiodit'iiiii. T/i.ab., thoracic-abdominal fold. i/. a., wandering cells. XLY. B.C, 138. M mes 144. !. •mes. mcs ALPHEUS 554 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XLVI. (Stage IX.) FIGS. 1-16-151. Serial trausver.se sections, through the embryo 01 A. saulcyi, at the time when pigmeut is first deposited in the, eye. In Fig. 146 the developmental history of the retinal layer is well shown, x 230. FIG. 152. Nearly median longitudinal section of embryo in similar stage. x58. FIG. 153. Sagittal section of similar embryo, showing degenerating elements in yolk below dorsal plate, x 58. REFERENCE LETTERS. A. 1, first antenna. A. II, second antenna. Ab., abdomen. B. C., blood corpuscle. B. S.. blood sp:ioe. i-p., i';ir:i|>;lcT. Deg., degenerating cells. Dp., dorsal plate. End., endoderm. /(/., tbregnt. fa., fiber mass of nervous system. tj.lI-III, bruin. g. c., ganglion cell. (j.m.a., anterior gastric muscle. H., heart. ltd. , bypodermis. H'h.. eggshell. cp., carapace. J'cy., products of ci-ll ilcgeimration. /.'(•/., ectoderm. l:- nd., endoderiii. /.?., liber-substance nf uerve cord. ''. /'', ganglion of mandible. yc., ganglion eel], ijf., liber ball ot'Nerinid antenna. U., heart. /<((., bypodermia. hi/., liindgut. //., lateral liber-mass of brain. .!/((., mandible. Mrs., meaoblast. ilu., muscle cells. Mu.f., flexor niust-le. Mts., metastoma. ii. c., neural cord. O. G., optic ganglia. «f.. optic eulargenient. p. c.. pillars of carapace. p. r., perineurinm. p. s., pericardial sinus. 7i'/., retinogen. >7if.. Btomodseum. vac,, vacuole. i/. r., wandering cell. !/. "., yolk nest. I'lnlc 154. nc. mts mes ALPHEUS 558 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE XLVI1I. (Stage X.) FIGS. Ifi2-165. Parts of serial sections through the region of the heart and thoracic-abdominal fold to show the extension and relations of the emloderm. x57. FIGS. 166 — 167. Parts of serial transverse sections of the embryo of Alphens saulcyi. x!25. FIG. 168. Median longitudinal section through a slightly older embryo, showing the ventral endodermic ibid (./'), the foregnt still screened from the yolk, and the nervous system separated from the skin. x2l.'7. REFERENCE LETTKRS. Ab , itlxUmiuii. Ali.y. I, first :i Inli mi m. 1 1 ganglion. aor., aorta. B. I'., Mood .-ell. />'. N. , blood space. cc., crystalline cone cells. Deg., products of degeneration. Ec)>., proximal retinular colls. End., eiidodiTiii. /., ventral endoderujic fold. g.m.a., anterior gastric mimcle. H., heart. Hg., hindgut. imb., intercepting mrmliranr. »i. s., pericardial sinus. Ht., retinogen. ^. O. G., snpra-cesophageal ganglion. T., telson. Tli., thorax. Til. y. I, ganglion of tirst amlinlatory limb. y. c., wandering cells. Mate XLY1I1 162. 166. svr^^r* • » *"^ '' a~* !!* '"Ws ~*£SH •1»Xv'*r >, M a**«S * ' 7*4 ' ' t m i- • i" / - r<* . »./• *<;.«^^-sopha- geal commissures. x234. FIGS. 175, 176. Transverse sections through the neural cord of the first larva. In Fig. 170 the transverse commissure of the ganglia of the tentli segment is cut and in Fig. 175 the short longitudinal commissures between the tenth and eleventh segments are sectioned, x !-"•)• REFERENCE LETTERS. A.I, first antouua. A. II, .second autenn:i. I//., abdomen. ay., antenna! gland. H.c., Mood corpuscles. /)'. A'., blood ,ii nils. /.'c., branchia. <•/>., ra.ra|iarr. End., endoderui. /., reproductive organ. "<>i/., brain. I. c., transverse commissure. .//. c., wandering cells. Plate XLfX muf. F.H.tfernrJr.Jel. ALPHEUS S. Mi,s. 94 36 562 MEMOlIia OF THE NATIONAL ACADEMY OF SCIENCES. IY.ATK L. (Shu.;,. XI.) FlGS. 177-170, IS], 182. Serial transverse sections of the embryo of Alplint* JtctcrtH-Iirlis, which is nearly ready to hatch. The shell is somewhat diagramatically represented and appears thickened in Fig. 182, owing to a coagnlable substance beneath it. The cells rep- resented in the yolk in Fig. i *- appear to be endodemi cells, which have become mechanically detached from the walls of the mesenteron. x 74. FlG. 180. Nearly median longitudinal section through a similar embryo. The eiidodermal lining of the mesenteron is not yet nearly completed. x74. DEFERENCE LKTTKRS. .tl>., VI, ^an^lion nl'sixlli ;iliiliiniiii:il :i)i|irinlage. «(/., aiitcnnal uniinlion. aus., mins. ch.er., oxti'rual c-liiasnia. <•<•!., ectoderm. cnii.. eniloderiii. ./(/., forejjnt. ijma., anterior Kast.vii: imiHi'lf. //., heart. /ii/., liindgnt. hy., hypndermis. If., lateral fiber-mass of luain. mil , mesenteron. tin/.1 wiiy.^ of otber figflres, jiosii-iio: lobo of midgut. mii.f., Ili-xor muscles of inid^iil . »»».(•., extensor muscles nf niid^ut. nciii.. irsophageal eDinmissnre. n. jiil.. tic iKNlnnrli-. llt-l., retina. MI;/., brain. T., telson. 1-1, ganglia of eye-stalk. PUtie L. Fig. 119. Fig. in muf. mue mue. muf. ALPHEUS HETEROCHELIS. 564 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE LI. (Stage XI.) FIGS. 183-186. Continuation of series of transverse sections of embryo begun on Plate L. x 74. FIG. 187. Part of sagittal section of similar embryo, cutting eyestalk somewhat obliquely. The specimen was depigmented in nitric acid. The distal retinular cells, occupying the spaces (Pg.c.) between tbe peripheral ends of the cones, are not represented. x305. REFERENCE LETTERS. oo. /'., accessory pigment cells. . i'.. Mood vessel, o. . crystalline cone. eg., corneagen. "•I., octoderm. end., cudoderm. fg., foregut. //., beart; lit/., bypodcTiniN. 6 MKMULKW UK T11K NATIONAL ACADEMY OF SCIENCES. 1'l.ATK LIT. FIGS. 1S8. ISO. Parts of trans verso serial sections through the embryo of Pnhrmonetes vulgnris, at the stage when pigment is just appearing in the eyes. In tlie anterior section (Fin- 1S8) the retiuogen is ., abdomen. Ab. n, sixth abdominal nppcndaignient cells. ay., green gland. «. 4. FIGS. '_>o.">-208. Transverse sections through adjacent ommatidia of the adult eye, taken at various levels. Fig. 205, the deepest section, shows the peculiar seven pronged figure of the rhabdom. The proximal retinular cells appear in sections, as it' fused together, x 294. FIGS. 209-211. Transverse serial sections through the first larva of Alpheus saulvi/i. -x 74. KEKKKK-NCK LKTl'EES. A. I, first antenna. A. II, second auteuua. a. op., accessory pigment, cell. uc.pn., nucleus ot .iiTi's.Movy jii^ini'iit cell. ao., car. lint., intiTccptiiif; or liaNcinciil ninniliniiif . cc., crystalline' c.nnii CC.HH. • 'i-, '-onica-icii. c7., Icllh. Co., no., crystalline- cone. i-nili., cone membrane. ltd., hypodermis. me., membrane of distal retiuular cells. »/., nerve fibers. oc. , ocellus. 017., optic ganglion. a!'., optic enlargement of brain. pap., papilla of ocellus. p.tj. <•., distal retinular cells. /;., rostrum. Rb., >'li., rhabdom. Set., retina. rtl. , proximal retiuular cells. Plate LI\' 'WO. - d. -{LC.p. B.m. acp 209. 201. 102. me ...-co. •Mc .CO. CO,_ .ff.C.p. cc ago. ZOfi. 205. acpn p§.c 201 CO."" rtl. Co. 211. $10. .. -E. .-E .•hd. Ret.... pap a a. ALPHEUS 572 MEMOIKS OF THE NATIONAL ACADEMY OF SCIENCES. PLATE LV. (Stage XII.) PIGS. 212-223. Transverse serial sections of the first larva of Alpheus saulcyl from the same indi- vidual as Figs. 209-211, excepting Figs. 222, 223. x73. REFERENCK LETTERS. A. //, second antenna. ad. HI., adductor of inaudible. ag., green gland. a/., ant.ttnnnlar fiber-mass of brain. ao., ear. a.op., ophthalmic artery. Kg., branchiostegite. B.gl., gland-like body. />'. ,s'., blood sinus. /;/., foregut. fi>., tibor-mass continiiod into cesophageal commissure. gf., anteunal liber-mass. its., lateral poticli of masticatory stomach. Lb., labrum. //., lateral fiber-mass of brain. Md., mandible. Mi/., mulgiit. Afy'., anterior lobe of midgnt. Mg'*.. lateral lobe of midgut. Mp., septum between anterior lobes of midgnt. M. S., masticatory stomach. Mt«., metastoma. Mr. I, first maxilla. Mr/iil. I, first maxilliped. n. ay., antennal nerve. n.nn., antenniilar nerve. or., avm., oesophagesl commissure. of., anterior filier-iiians and transverse commissure of brain. /). ["., pyloric valve of masticatory atomach. at. a., sternal sinus. I '{tile LV 212. ..AH. as. 214. a op a.op . at if. 219. 2/S. aop ms mxl. \mxpd.I. ms-' 222. ocm 225. mp. ,.a°P sts ms. FIRST LARVA OF ALPHEUS SAULCYI. 574 MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. PLATK LVI. (Stage XII.) FIGS. 224-235. Serial transverse sections through the first larvar, continued from Plate LV. x73. REFERENCE LETTERS. Ab. I', fifth abdominal app o. i. a., inferior abdominal aorta. a. op., ophthalmic artery. £• &£ F.H.H<-rridr,