J. S. KINGSLEY 'i . f «^r • ' ;>. « / s <^THE EMBRYOLOGY Blatta Germanica and Doryphora Decemlineata. BY WM. M. WHEELER. Reprinted from The Journal of Morphology, Vol. III., No. 2, Sept., 1SS9. BOSTON: GINN AND COMPANY. I i THE "embryology Blatta Germanica and Doryphora Decemlineata. BY WM. M. WHEELER. r B. K:ingBl9V. I Reprinted from The Journal of Morphology, Vol. III., No. 2, Sept., 18S9. BOSTON: GINN AND COMPANY. t. ’rOx^ RWCV S9l.3i V/Sb i THE EMBRYOLOGY OF BLATTA GERMANICA AND DORYPHORA DECEMLINEATA. WILLIAM M. WHEELER. The following study of the development of the cockroach and potato-beetle was taken up during the summer of 1887. On the suggestion of Dr. W. Patten, without whose stimulating friendship and assistance the work would not have been under¬ taken, I began with Blatta as a form calculated to help me to a knowledge of the fecundative changes in the Hexapod egg. Dr. Patten kindly placed at my disposal the results of his own work on Blatta^ in the form of much carefully prepared material and some figures, which I have incorporated in Plate III. (Figs. 43, 45 to 47, 52). Later I concentrated my attention on Doryphoray which I found to be a much more profitable object of study than Blatta, especially as far as the more advanced stages were concerned. Thus it happens that my remarks on oogenesis and fecundation are more complete in Blatta, while my account of the germ-layers and subsequent stages is carried into greater detail in Doryphora. I have seen fit to treat of both insects as nearly as possible under single headings, instead of describing them independently in two chapters, because they differ strikingly in all the details of development, while their main ontogenetic features are as strikingly similar. By running both descriptions as nearly as possible in parallel lines, the contrasting details are made more salient, while the general remarks may be taken up at intervals and not reserved en masse till the end of the paper. Preparation. There are three common species of Blattidce in Southeastern Wisconsin : Periplaneta orientalis (Linn.), Platamodes tmicolor (Scud.), and Blatta germanica (Linn.). The first and the last occur, as is well known, about houses; the second is abundant under the bark of decaying logs and stumps in open woods. 542337 WHEELER. [VoL. nr. 292 Periplaneta ormitalis oviposits from April to August; Blatta ge7'fnanica, at all times during the year. The oothecae of Plata- modes may be collected in great numbers where the insects abound. After many futile attempts to open the oothecae of Periplaneta and Platamodes without injuring the ova, I limited my study to Blatta^ the egg-capsules of which may be easily opened by the method given below. By careful treatment of the thick-walled capsules of Periplaneta and Platamodes with a sufficiently strong solution of sodium hypochlorite, it may be found possible to isolate the ova in an uninjured condition. Specimens of Blatta germaiiica can be obtained at all times of the year from places which they haunt, and with very little attention will live long in confinement. The male is long and narrow, tapering anteriorly and posteriorly ; the female is much broader and flatter and uses her wings much less than the male. Males seem to endure the unfavorable condition of captivity much better than females. When it is desired to time the eggs, the capsule cannot be detached from the female without damage till it has been rotated, and during winter must be kept under a bell jar with plenty of moist blotting-paper to prevent the embryos shrivelling from the dryness of the air. The ovarian ova in all stages up to maturity were dissected out in normal salt solution and hardened for fifteen minutes in Perenyi’s fluid. They were then transferred to 70 per cent alcohol, which was changed several times at intervals of an hour, and were finally preserved in alcohol of the same strength. When stained with borax carmine and sectioned, the yolk retained none of the red stain, while the chromatin of the nucleus shone out as a glistening deep red spot. Perenyi’s fluid rendered the chorion of the mature ovarian egg pervious to borax carmine. Hardening in a saturated aqueous solution of corrosive subli¬ mate gave good results with young ovarian eggs. Oviposited eggs were killed by placing the capsules in water slowly heated to 8o°-90° C. The two lips of the crista of the capsule were then separated by the aid of fine forceps, and pieces of the walls torn away, till the eggs could be easily pushed out of the compartments formed by their choria. The ova thus isolated were either transferred directly through 35 per cent (10 min.) to 70 per cent alcohol, or they were left for 15 minutes in Kleinenberg’s picrosulphuric acid, and after re- No. 2.] BLATTA AND DORYPHORA. 293 peated washing in 70 per cent alcohol, preserved in alcohol of the same strength. Both methods gave equally good results. Though I have succeeded in dissolving the chitin of the ootheca with sodium hypochlorite, the method of tearing off the walls after heating to 80° C. gave such satisfactory results that I adhered to it throughout my work. I have found Grenacher’s borax carmine in every way the most expedient and reliable staining fluid. Eggs and embryos up to the time when the cuticle develops were stained before imbedding in paraffine; the sections of other embryos were stained on the slide after attaching them with Mayer’s albumen fixative. The clusters of bright yellow eggs of the potato-beetle (Dory- phora decemlineata, Say) may be found on the under surfaces of the leaves of the potato-plant during the whole summer, as the insect is polygoneutic. The beetles, frequently found copulating, may readily be kept in confinement, and will deposit their eggs in the typical flat clusters on the walls of any box or vessel in which they are kept. As I commenced collecting material late in the season, I did not keep the insects in confinement till they oviposited, but collected the eggs from the plant. It was found convenient to cut out the piece of the leaf to which the egg-cluster was attached and to keep it by itself during the process of preparation, as all the eggs of a cluster are in almost exactly the same stage of de¬ velopment. Beautiful results in preparation were obtained by heating the eggs to 80° C. for 10 minutes in Kleinenberg’s picrosulphuric acid (with 3 volumes of water) and preserving in 70 per cent alcohol. By this process the envelopes, which in the fresh egg adhere closely to the yolk, dilate and stand off from the surface of the egg, and except in the very youngest stages can be rapidly and easily removed with the dissecting needles. A great number of eggs, heated to 65° C. only, or hardened in cold Perenyi’s fluid, corrosive sublimate or simple alcohol, proved to be useless, as the envelopes adhered firmly to the sur¬ face of the yolk. The hot picrosulphuric acid fixes the cells of the embryo in a most satisfactory manner; enough details of the karyokinetic 294 WHEELER. [VOL. III. figures being preserved to enable one to recognize dividing nuclei at a glance. All the eggs were imbedded after treating with clove oil, in paraffine melting at about 55° C. The somewhat gummy yolk cut without any tendency to crumble, and perfect series of sections were obtained without the slightest difficulty. Staining on the slide with borax carmine gave beautiful results in all stages, but was resorted to only in young eggs, the vitel¬ line membranes of which did not stand off from the surface of the yolk, and in advanced embryos which had developed the larval cuticle. To save time, embryos in other stages were stained before imbedding, after removing the egg-envelopes. As much as possible of the borax carmine was extracted with acidulated 35 per cent alcohol. Historical. The cockroaches have long been favorite objects of morpho¬ logical study. Easily obtained at all seasons of the year, of convenient size for dissection, and being but slightly modified descendants of the oldest insects of geological time, they com¬ bine qualifications which make them especially interesting and valuable to the morphologist. Thus we find that no less than twenty investigators have sought material for anatomical and embryological study in the common species of Blattidce. I will mention only those who have treated of the oogenesis and ontog¬ eny of Blatta and Periplaneta. Rathke (42) was the first to publish an account of the devel¬ opment of Blatta germanica. Brandt (6) made a study of the ovarioles of Periplafieta. Huxley’s (22) account of the general anatomy of the same insect contains some valuable remarks on the ovaries. Kadyi (23) has given us a condensed account of the oviposition and micropyles of Pe 7 'ipla 7 teta orientalis. Patten (38) in 1884 published a preliminary note on the development of Blatta. He observed that the first and second maxillae are at first triramous, and made some remarks on the heart and on the peculiar organs developed from the appendages of the first abdominal somite. Stuhlmann (45) treated of the degeneration of the germinal vesicle in Periplaneta. During the same year (1886) also appeared Miall and Denny’s work (32) on the anat¬ omy of Periplanetay containing Nusbaum’s brief embryological No. 2 .] BLATTA AND DORYPHORA. 295 description of Blatta^ carelessly written and with figures often grossly inaccurate. Blochmann’s important paper (5), announcing the discovery of polar globules in Blatta germanica and two other insects, appeared in 1887. The description of the eggs of Blatta is suc¬ cinct and perfectly accurate. Of late, Cholodkowsky (lo) has published a preliminary paper on the formation of the entoderm in Blatta. Doryphora decemlineata has not been investigated heretofore from an ontogenetic standpoint. It is surprising that so com¬ mon an insect, and one whose eggs present such advantages for embryological study, should have been overlooked. The favored Coleopteron of embryologists has always been Hydrophilus^ and it is certain that the water-beetles {Hydrophilidce and Dytis- cidce) are much less modified forms than the leaf-beetles (Ckry- somelidcB)y to which Doryphora belongs. Nevertheless the development of several Chrysomelids has been studied more or less incompletely. Packard (35) made a brief study of Gastropkysa coeruleipennis^ and Melnikow and Kowalevsky (26) studied Donacia. More exhaustive was the attention bestowed on Lina by Graber (i 5), who published his account in the second volume of his text-book on insects (1877). As would be expected from their close systematic affinities, Doryphora and Lina differ but slightly in their development. Ovaries and Oviposition. Blatta. The ovaries of Blatta are flattened, broadly spindle-shaped masses slung in trabecular connective tissue, continuous with the peritoneum. Each ovary consists of from 14 to 26 ovarioles, or egg-tubes opening into the oviduct. The latter extends backwards towards the median longitudinal axis of the body, and after joining the oviduct from the opposite ovary opens into the broad and short vagina. Besides the tubular colleterial glands the vagina carries on its dorsal face near the proximal ends of the oviduct a thick-walled sac, the spermatheca. The ovariole has the structure typical in insects. The fol¬ licles in all stages of formation are inclosed by the membrana 296 WHEELER. [VOL. III. propria in the form of a tube tapering to capillary caliber at its upper extremity, which is attached to the pericardium. The lumen of the capillary portion is filled with protoplasm in which numerous small nuclei are imbedded. This portion of the ova- riole constitutes the germarium. The small nuclei differentiate at the lower end of the germarium, on the one hand into ova, which fill the widening lumen of the tube; and on the other, into flattened epithelial cells, which line the inner surface of the tube, and form the follicles inclosing the ova. There are about ten distinct ova in an ovariole, the lowest being the largest, and the most apical the smallest and most indistinct, those interme¬ diate regularly diminishing in size towards the apex. In Peri- plafieta there are about three times as many ova in an ovariole; but there are only eight ovarioles in an ovary. The lower ova in both species are oval, and are surrounded on all sides by the epithelium, which has grown in between the separate eggs to complete the follicles. The follicular epithelium (Fig. 5) is composed of large, flat, polygonal cells, with lenticular nuclei which present an intri¬ cately coiled chromatin filament and a nucleolus of unusual struc¬ ture. The latter consists of an irregular mass, not stainable in carmine or methylgreen, and is regarded as plastin by Car- noy (9), who describes and figures very similar nucleoli in the egg-follicles of Gryllotalpa. The mass of plastin incloses a smaller mass of chromatin, or at least of a substance which does not differ in its reactions from the chromatin of the coiled fila¬ ment in the same nuclei. In eggs taken from the ovaries just be¬ fore maturity, when the epithelium is still firmly attached to the underlying chorion,- almost all of the nuclei will be found rapidly dividing. Pieces of the epithelium from eggs of differ¬ ent ages were examined in normal salt solution, in methylgreen acetate held for a moment in the fumes of osmic acid, in RabPs chromformic acid, in Zaccharias’ acetic osmic acid; but no traces of an achromatic spindle, or of a regular arrangement of the nuclear filament, so characteristic of karyokinesis could be observed. I therefore conclude that we have here a case of akinesis or direct division. This conclusion is further strength¬ ened by the observation that the nucleolus divides first (Fig. 5 ^), and the nuclear wall is constricted during division, an occur¬ rence exceedingly rare in kinetic nuclei, where the nuclear wall No. 2 .] BLATTA AND DORYPHORA. 297 disappears in all but a few of the recorded cases both in plants and animals. Moreover, the two daughter nuclei are frequently very unequal in size. The chorion (Fig. i) is a thin, chitinous membrane smoothly covering the surface of the egg. In surface view it appears to be finely granular, the finest granules being arranged in large, more or less regularly hexagonal areas, which are bounded by narrow, dark spaces containing somewhat larger though less dense granules. Each of the hexagonal areas is secreted by one of the polygonal epithelial cells described above. It is only in cross-section that the true structure of the chorion becomes apparent. According to Blochmann (5),—and my observations coincide with his, —the chorion consists of two chitinous laminae kept in close apposition by means of numerous minute trabe¬ culae, or pillars. It is the ends of these pillars seen in surface view that look like granules. In the spaces between the hex¬ agonal areas, the trabeculae are more scattered and individually thicker than those of the hexagons. Hence these lines on the chorion seem covered with larger and more scattered granules. When pieces of the dry chorion are immersed in glycerine and immediately examined under the microscope, the thick liquid may be seen entering the spaces between the hexagonal areas, passing along them in obedience to the laws of capillarity, and then slowly creeping from them on both sides into the adjacent hexagonal areas between their denser trabeculae. I have also observed that the structure of the chorion of the ripe egg is most distinct in cross-section at the pole directed towards the germarium. Here the two laminae seen at 00 in Fig. 4 separate somewhat, and the connecting trabeculae become longer and more distinct. I have not been able to trace the formation of the micropyles in Blatta gerrnanica. Their structure is easily demonstrated. They are scattered over a quadrant of the upper hemisphere where the beautiful hexagonal pattern of the chorion gives away to an even trabeculation (Fig. 2). The micropyles are wide-mouthed, very oblique, funnel-shaped canals perforating the chorion (Fig. 2 a, b). The apertures of the funnels appear under a low power as clear, oval spots, the long axes of which are parallel to the long axis of the egg. These perfora¬ tions are scattered over the micropylar area,, sometimes in clus- 298 WHEELER. [VOL. III. ters, sometimes singly. With a higher power the tube of each funnel is clearly visible as a thin canal which dilates rapidly into the large oval aperture on the outer face of the chorion. The narrow tube is sometimes fully as long as the large orifice. The micropylar perforations are all directed from the germa- rium to the vaginal pole of the egg. Hence a line, the hypo¬ thetical path of the spermatozoon, drawn through one of these oblique micropyles, and continued into the egg, would strike the equatorial plane. The female pronucleus, as we shall see further on, moves in this plane. The micropyles of Periplaneta, first described by Kadyi (23), do not differ essentially from those of Blatta. In Periplaneta the hexagonal pattern is continuous over the micropylar area. The large micropyles, which are more yellowish than the sur¬ rounding chorion, are thick walled and not regularly oval, as in Blatta, but oblong or subpentagonal. The tube is shorter and terminates on an hexagonal area. Sometimes the micropyles are very close together and seem to overlap. I have repeatedly sought in vain for a vitelline membrane in the eggs of Blatta. Blochmann (5) had no better success. It may exist, but it must be exceedingly thin and inseparably glued to the inner lamina of the chorion. The colleterial glands of Blatta are like those which Huxley (22) and Kadyi (23) have described for Periplaneta, a number of long, blind tubes opening into the vagina. They furnish the ma¬ terial for the capsule, viz. : chitin and large crystals of calcium oxalate. In Blatta these glands are glistening white till the time of oviposition approaches, when they assume a yellow tint, and the octahedral crystals are seen imbedded in a viscid sub¬ stance which fills their lumina. This viscid substance is sol¬ uble in potassium hydrate, and is consequently not chitin. When excreted to form the ootheca, it slowly hardens, deepens in color, and becomes insoluble in potassium hydrate. Light has nothing to do with this change, which is possibly produced by the oxygen in the air. It is the same change which is undergone by the cuticula of the insect itself immediately after ecdysis. I have made a few observations on the oviposition of Blatta germanica, similar to those published by Kadyi (23) on Peri¬ planeta orientalis. No. 2.J BLATTA ABtD DORYPHORA. When about to form the capsule, the female Blatta closes the genital armature, and the two folds of the white membrane which lines the oothecal cavity close vertically in the middle line. Then some of the contents of the colleterial glands are poured into the chamber and bathe the inner surface of the posterior wall. The first egg glides down the vagina from the left ovary, describes an arc, still keeping its germarium pole uppermost, after having pressed the micropylar area against the mouth of the spermatheca, passes to the right side of the back of the chamber, and is placed perpendicularly two-thirds to the right of the longitudinal axis of the insect’s body. The next egg comes from the right ovary, describes an arc to the oppo¬ site side of the body, decussating with the path of the .first egg, and is placed completely on the left side of the median line. The third egg comes from the left ovary, and is made to lie completely on the right side of the median line : and so the process continues; the ovaries discharging the eggs alternately, and each egg describing an arc to the opposite side of the cap¬ sule. In females killed during oviposition each oviduct will be found distended with eggs, often two or three end to end, in¬ creasing the length and breadth of the lumen to an, abnormal degree. Gentle pressure of the female’s abdomen between the thumb and finger will sometimes cause the insect to oviposit a few eggs, the paths of which can be seen to decussate. The oothecal chamber soon becomes too small to contain all the constantly accumulating eggs, so the anal armature opens and allows the end of the capsule to project. A raised line, the 0 impression of the edges of the white membrane, runs down the end of the capsule. The last egg deposited comes from the right ovary and lies two-thirds on the left and one-third to the right of the median line. Thus the first and last eggs laid lie with their greater bulk on opposite sides of the median vertical plane of the capsule, and serve to commence and close the series and round off both ends of the capsule. Owing to their being crowded up against the walls of the capsule, they acquire quite a different shape from the remaining symmetrically and alter¬ nately deposited ova. They develop normally, however, the embryo appearing on the inner obtuse edge. As soon as the last egg is,laid, a further discharge from the colleterial glands spreads over the vaginal or anterior wall of the cavity, 300 WHEELER. [VoL. iir. and becomes evenly continuous with the secretion which has before been spread over the back and the sides of the capsule by the white membrane. When the anterior end of the capsule is examined, the escutcheon-shaped vaginal opening is found to have left its impression even to the delicate wrinkles into which the surrounding cuticula was thrown by the closing of the ori¬ fice. This end of the capsule is white, gradually shading into the brown of the opposite end. The crista,, a cord-like ridge running the full length of the dorsal surface of the capsule, is a thick-walled tube, either half of which is formed by the edge of the side walls of the capsule split into two laminae (Fig. 3 0^, 0^). The rhythmical clasping of the three pairs of palpi, which guard the vaginal opening, is registered in an exquisite pattern on the inner face of either half of the crista. The canal is filled with a vacuolated substance (Fig. 3 ep) which at first sight resembles the yolk of the egg, but when examined more closely is seen to have quite a different struc¬ ture and origin. In the egg ready to leave its follicle the epi¬ thelium is much thickened at the germarium pole into a bicon¬ vex-lens-shaped cap (Fig. 4), the cells of which are not flat like those on the other portions of the egg, but long, columnar, and more or less curved. The two laminae of the chorion spread apart beneath this cap and dilate into a pear-shaped sac divided up into numerous polygonal chambers by delicate chitinous partitions (Fig. 4 b). While the egg is leaving the follicle, the epithelium at the lower pole is loosened from the chorion, and * the egg protrudes into the oviduct. As it advances, the epithe¬ lium is rolled back and doubled up in folds till it is freed from the chorion as far as the cap. Then it breaks, letting the egg pass into the oviduct with the thick cap of cells firmly attached. The egg is placed in the capsule, and the cap comes to lie in the crista, filling its lumen. The large nuclei degenerate, and soon entirely disappear, the protoplasm becomes dry and vacuo¬ lated, and finally transformed into the yolk-like mass described above. This substance probably serves as a cement to keep the lips of the crista in contact till separated by the emerging larvae. Thus a small portion of the follicular epithelium, that portion which corresponds to the nourishing cells in other insect ovaries, is deposited with the egg. To my knowledge No. 2.] BLATTA AND DORYPHORA. 301 this has not been observed in any other arthropod eggs hereto¬ fore described, excepting Miisca (Bruce, 7). The pear-shaped dilatation of the chorion is directly over the head of the future embryo and hatching insect, and is possibly more easily rup¬ tured or dissolved than the surrounding chorion. Of 40 capsules examined for the purpose of noting which ovary sent out the first and which the last egg, 32 com¬ menced with the right egg (from the left ovary) and closed with the left egg (from the right ovary), six capsules commenced and closed with the right, and one commenced and closed with the left egg. Evidently the 32 were normal; in the insects which deposited the six, one of the ovarioles was probably either atrophied or wanting, though the perfectly alternate arrangement of the eggs in the capsule was in nowise inter¬ rupted on this account. The two remaining egg-capsules were small and abnormal. The number of eggs in a capsule, far from being constant in Pei'iplajieta orientalis, is even more fluctuating in Blatta ger- manica. In 34 capsules counted the average number was about 40, the least number '28, and the greatest 58. The number varies in different localities and is doubtless dependent on the food of the female insect. In several capsules obtained where amylaceous food was abundant the average was much higher than in a much greater number of capsules obtained from a place where fatty food was the only diet. The above description of the oviposition of Blatta gernia 7 iica probably applies to most species of the Blattidce. But this species differs from Periplaneta and probably many other forms, in rotating the capsule, a process now to be described. As soon as the last egg has passed the vagina and been placed in the capsule, the latter begins to rotate on its longitudinal axis, till the crista has described one-fourth of a cylinder to the right. The capsule is now in a horizontal position having its greater transverse diameter parallel with the corresponding transverse diameter of the insect’s body. The abdomen contracts during oviposition, and its end comes to lie anterior to the tips of the wings, so that the broad ends of the latter hide and protect the protruding end of the capsule. The rotation requires about a day. In one case a female kept the capsule in a vertical position for two weeks, apparently from some inability to revolve it. In WHEELER. [VoL. riT. 362 all other cases the capsule was regularly turned to the right, never to the left. The female Peripla 7 ieta orientalis drops the capsule soon after its completion, with predilection in some food supply, such as flour or meal. Blatta gennanica carries it for about a month, then drops it shortly before the hatching of the larvae. Some writers claim that the parent assists the young in escaping from the capsule, but I have proved by many experiments that the expanding and struggling of the young insects are amply sufficient to separate the feebly united lips of the crista. Such an instinct in the female would be of use only if the lips of the capsule could not be opened by the young. Taschenberg (46) claims that the female regularly lays only one capsule and dies soon after its deposition. My observations on fifty females, whose wings were clipped as soon as they had formed their first capsule, have convinced me that they certainly lay two perfect capsules as a rule, and possibly more, in the course of the year. As will be seen from the preceding account, it is a very easy matter to orient the eggs of the capsule, to tell just what position any oothecal egg held in the ovary, or just what position any egg in the ovary will hold in the capsule. The germarium pole of the ovarian egg lies just beneath the crista after oviposi- tion. The concave side of the curved sausage-shaped ovarian egg is turned to the wall of the capsule, and its convex face, on which the embryo will appear with its head towards the crista, is turned to the interior of the capsule and faces the corresponding surface of the opposite egg. The micropyles which pointed away from the germarium are on the convex face of the ova¬ rian, and on the inner face of the oothecal egg, and point down¬ wards away from the crista. The Development of the Egg to the Formation of the Blastoderm. Blatta. The oothecal egg of Blatta^ like the ovarian egg, is glistening white. The latter has been described as sausage-shaped, but the pressure exerted by the eggs on one another in the capsule alters their original form very considerably. They assume the No. 2.] BLATTA A.VD DORYPHORA. 303 shape of half an elliptical disc, the short axis of which is to the long axis as 2 is to 3, with a thickness one-sixth of the short axis (Figs. 36 and 37). The egg is about 3 mm. long, i mm. broad and J mm. thick. Its volume is therefore almost a cubic millimeter. The cephalic end (Fig. 36 c), recognized by its evenly rounded contour, is immediately beneath the crista of the capsule. The opposite or caudal end (Fig. 36 s) is dis¬ tinguished by a slight sinus. The ventral face is traversed by a keel which runs from the cephalic to the caudal extremity and is most pronounced a short distance below the middle of the egg. The dorsal surface is flat and evenly curved antero- posteriorly. Cross-sections' of the ovum are consequently pen¬ tagonal (Fig. 40). Thus the eggs can be easily oriented, and they have a great advantage over spherical or even oval ones, in that all the earliest developmental changes can be traced directly to their relationship with the parts of the future embryo. This is of the highest importance in the early stages. It will have been observed that a complete reflexion of the egg takes place during oviposition as the concave face of the curved ovarian egg becomes the convex back of the oothecal egg, and the convex micropylar face of the former becomes the straight, carinated, ventral face of the latter. Yolk. — I have not been able to observe a passage of fol¬ licular epithelial cells into the egg to form yolk by their dis¬ integration, as has been described by Will (51) in Nepa and Notonecta, and by Ayers (i) in Qicanthus. There is only one layer of cells in the follicular epithelium of Blatta^ and as this persists till after the chorion is completed, no migration of nu¬ clei into the yolk, or even disintegration of nuclei at the surface of the egg, is observable. All of the yolk in Blatta (excepting those portions derived from the germinal vesicle i*) is secreted by the protoplasm of the epithelial cells, not as yolk, but as sub¬ stances which are taken up by the growing ovule, and again secreted in the form of the bodies to be described presently. During this process of yolk-secretion, the epithelial cells re¬ main intact, their slow disintegration not taking place till after oviposition, when their compacted and yellowish remains have assumed the appearance so suggestive of the corpora lutea of the Mammalia. The yolk of the eggs of Blatta has been studied by Patten 304 WHEELER. [VOL. III. {38) and Blochmann (5). The former has described the physical structure; the latter, the peculiar bilateral distribution of the yolk elements. Though my researches have revealed only a few new facts concerning the yolk, I will give them in their entirety. In the young ovarian egg, 0.5 mm. long, the nucleus is surrounded by granular protoplasm in which no yolk has developed. In eggs i to 2 mm. long the yolk consists of two kinds of bodies : transparent fat globules of various sizes, and a translucent albuminous substance broken up into distinctly outlined masses which are polygonal in form from mutual pressure. Beneath the follicular epithelium is a layer of small albuminous masses enveloping the interior yolk. This is at a time when the large nucleus is disintegrating. By the time the egg has reached its full size, the yolk has assumed the highly differentiated structure best studied in the oothecal egg. The yolk of a fresh mature egg crushed between the slide and the cover glass in its own liquids, or in normal salt solution, shows an abundance of different sized highly refractive oil globules, and a greater number of distinctly outlined albumen spheres, which are polygonal in the intact egg where mutual pressure prevents them from taking on the spherical shape which they seem to be continually striving to assume. These albumen spheres are thin-walled sacs full of a thin liquid in which float multitudes of small, irregular granules. Sometimes small oil glob¬ ules are enclosed. That the contents of these sacs is a thin liquid is proved by the exceedingly active Brownian movement of the irregular granules, a movement which is certainly constant in almost all these vitelline bodies in the living egg. But these bodies are far from being alike in structure. In some few the granules are very minute, closely packed, and exhibit no Brownian movement; in others the granules are large and dis¬ tinctly irregular, few in number, and possessed of a tendency to unite in flakes which hang suspended in the thin liquid filling the sacs. In the great majority of these bodies, however, the mean between these two extremes in structure is maintained. It is probable that the granular forms arise from the dense homo¬ geneous bodies by chemical decomposition. This decomposi¬ tion, which is accomplished, as I said above, in the ripe ovarian egg, takes place in different parts of the egg in different degrees, No. 2.] BLATTA AND DORYPHORA, 305 thus bringing about the peculiar bilateral structure first noticed by Patten and subsequently described by Blochmann. In hardened eggs the granules have the appearance of nets with larger or smaller meshes (Figs. 23, 24, etc.). This structure is best seen in cross-sections of eggs hardened in alcohol or picrosulphuric acid. For the distribution of the above-described yolk-elements in the egg I cannot do better than quote Blochmann’s succinct and accurate account. He says : “ Die Hauptmasse des Dotters ercheint auf dem Querschnitt [see my Fig. 40] etwa als gleich- schenkeliges Dreieck, das seine unpaare (kurze) Seite nach aussen, seine Spitze der Mitte zukehrt. Diese Dottermasse besteht aus durch gegenseitigen Druck polygonal abgeplatteten Korpern, deren Umrisse besonders an den peripheren Par- tien deutlich sind, wahrend sie in der Mitte des Eies mehr Oder weniger zusammenfliessen. In den peripheren Regionen zeigen diese Korper eine feine Granulirung ihrer Sub- stanz, wahrend sie im Innern des Eies vollstandig homogen erscheinen. Sie farben sich bei der gewohnlichen Behand- lung mit Boraxkarmin ziemlich intensiv. Umgeben wird diese Dottermasse von einer Zone [my Eigs. 39 and 40] von eben- falls polygonalen Elementen, die jedoch iiberall deutlich unter sich und von den Elementen (der Hauptmasse des Dotters) sich abgrenzen. Sie zeigen eine grobnetzige Struktur und far¬ ben sich mit Boraxkarmin nur ganz wenig. Diese Zone des Dotters hat ihre grosste Dicke liber der Spitze des von dem anderen gebildeten Dreiecks und nimmt nach den Seiten zu ab, um in der Mitte der Riickseite wieder etwas an Machtigkeit zu gewinnen. Dieser blasse Dotter bildet also einen kontinu- irlichen Mantel um den anderen. (Es) finden sich langs der ganzen Eioberflache zwischen die blassen Dotterkorper eingelagert einzelne starker farbbare Elemente des Dotters (der Hauptmasse) oder auch kleine Gruppen von solchen. Unregelmassig durch beide Dotterarten finden sich zahlreiche grossere und kleinere Fetttropfen, die an den Praparaten als Hohlraume erscheinen, da das Eett durch die Behandlung mit Chloroform, Terpentinol, etc. aufgelost wird” (my Eigs. 23, 24, etc.). The ventral thickening of the “continuous mantel ” of granular yolk polyhedra, which surrounds the wedge-shaped central mass 3o6 WHEELER. [VOL. III. of translucent polyhedra, is a pre-arrangement for the reception of the blastoderm, and more particularly the future embryo. The vital activities, which, in forming the embryo, finally be¬ come centred on the ventral face of the egg, find the yolk here already sufficiently decomposed to admit of easy metabolism into protoplasm. Later, in eggs with advanced embryos on their ventral faces, the granular yolk has again become transparent and homogeneous in extensive masses which arise from the con¬ fluence of many of the polyhedral bodies. I differ from Bloch- mann in maintaining that the granular yolk is originally derived from the homogeneous variety. He was led to infer the oppo¬ site process from his study of the yolk in ants’ eggs. Protoplasm, —In young ovarian eggs, 0.25 to 0.5 mm long, the cytoplasm is finely granular. The deutoplasm begins to accumulate, and by the time the egg has become i mm long, the above-described vitelline bodies and oil globules have de¬ veloped in such numbers as to reduce the protoplasm to an exceedingly delicate net. In the mature egg the KeiinhatU described for so many insects’ eggs is a layer so thin that it is just perceptible on the centre of the dorsal surface and at the cephalic pole (Fig. 4pr). This protoplasmic layer is full of what I shall call Blochmann’s corpuscles. They are minute rod¬ shaped bodies so numerous in the surface protoplasm as to make it appear reticulate. They look like bacillar micro-organisms, stain deeply, and according to Blochmann, who is probably right in thinking that they play an important role in the development of the egg, multiply by transverse division (like Bacteria). Weismann was the first to note these bodies in Diptera in 1863. Blochmann (3, 4,) called attention to them in 1884 and 1886, in the eggs of Camponotiis and Formica, and in 1887 in Blatta, Peri- planeta, Pieris, Musca, and Vespa. In the three last genera the bodies are spherical. In several of Stuhlmann’s (45) figures, these bodies are prominent. In Blatta I have found them wher¬ ever the peripheral layer of protoplasm is perceptibly thickened, especially surrounding the polar globules on the middle of the dorsal surface (Figs. 14, 15, and 16), and at the cephalic end of the egg (Fig 4 p'T). They seem to be made of more rigid material than the protoplasm in which they are embedded, for they pro¬ trude as very minute prickles from the surfaces of eggs hardened in Perenyi’s fluid. I have not been able to trace out their der¬ ivation and ultimate destiny. No. 2.] BLATTA AND DORYPHORA. 307 As the granular yolk polyhedra of the ventral portion of the egg are much smaller than the homogeneous ones of the in¬ terior, there is more protoplasm in the ventral part of the egg. The internal portions crumble very easily in sectioning, and I am hence inclined to think that very little or no protoplasm extends in between these bodies from the periphery. Thus we see that the distribution of the protoplasm, as well as the yolk, is in accordance with the position of the future embyro. Later, when cells appear in the egg, their amoeboid cytoplasm con¬ sists of evenly granular protoplasm, faintly stainable in borax carmine. Karyog 7 'apJiy of tJie Egg. — In studying the changes in the nucleus, it has proved to be impossible to preserve the finest cytological details, since the ovarian eggs of BlattUy like the eggs of the Orthoptera in general, are not easily sectioned un¬ less hardened in Perenyi’s fluid, as the yolk is exceedingly friable. Perenyi’s and Kleinenberg’s fluids do not preserve the karyokinetic figures perfectly, but cause the loops of chromatin to fuse in masses. On the other hand, the achromatic spindles are often beautifully distinct. Any discrepancy between Bloch- mann’s figures of the polar globules and mine, is probably to be attributed to a difference in the use of reagents. In young and transparent ova taken from the middle of the ovariole, the nucleus is seen to have reached the acme of its development in volume. It is a large spherical body, more highly refractive than the surrounding cytoplasm. Its fluid contents, the karyoplasm, is distinctly separable into two sub¬ stances, a liquid karyochylema and an achromatic (plastin T) reticulum. In the meshes of dhe latter is suspended a third substance, the deeply stainable chromatin, in one large and sev¬ eral smaller masses, or in several small irregular particles de¬ rived from the former by disintegration. The nuclear wall is very distinct. Whether it is a membrane or merely a peripheral inspissation of the karyoplasm is uncertain. When ovarioles are treated with P'ol’s picrochromic acid or Merkel’s chromplat- inum solution, the whole nucleus contracts and leaves a menis- coid cavity between itself and the cytoplasm, though it still preserves its distinct and evenly spherical contour. In all the following stages the egg, having become opaque, must be studied in sections, and the history of the nucleus con- 308 WHEELER. [VoL. iir. structed from a series of isolated observations. While the yolk is accumulating, the nucleus becomes amoeboid, leaves, its posi¬ tion at the centre of the egg, and travels to the surface. The egg is at this time very slightly curved, and the nucleus always passes to the centre of the concave side. This amoeboid nucleus is seen in Fig. 6 just after reaching the surface. It is still large, and does not stain more deeply than the surrounding yolk bodies. Its long axis is parallel to the long axis of the egg. The chromatin, in large masses in the younger egg, has been reduced to numerous granules of varying size, still recog¬ nizable as chromatin because staining deeply in borax carmine. The pseudopodia are now drawn in, and the nucleus becomes spheroidal. Soon the face in contact with the surface becomes cup-shaped, and round masses of a homogeneous substance indis¬ tinguishable from the surrounding yolk bodies fill the cavity (Figs. 7 and 8 b). In this stage the nucleus is probably giving off the “maturation spheres” (Reifungsballen), which Stuhlmann saw given off from the nucleus of so many insects’ eggs (notably Lepidoptera) ; Will (51) also describes this process in Hemiptera. In many insects Stuhlmann found that these spheres differed from the surrounding yolk bodies in power of refraction, etc. The concavity (Einbuchtung) of the nucleus in Blatta often contains several of these spheres. Figure 8 represents such a nucleus seen from the surface, the plane of section being at right angles to the plane of section in Figs. 6 and 7. Here the concavity is composed of separate cavities which have fused. The maturation spheres, after their escape from the nucleus, mingle with the yolk bodies. Above the orifice in Fig. 8, at pn, is a small body, denser and more refractive than the surrounding plasma. I think it corresponds to Stuhlmann’s paranucleolus. In its centre is a deep red body, probably a granule of chromatin. In Fig. 9 is seen a nucleus containing two paranucleoli, /;/, destitute of the central granule. I cannot say what becomes of these bodies. They do not appear in all nuclei {confer Fig. 7), and are probably evanescent structures formed and again disintegrated during the mysterious process of nuclear degeneration. In Fig. 9 granules of chromatin, most numerous near the point «, are scattered through the karyoplasm. The disintegration of the nucleus by giving off the maturation spheres progresses till, when the egg is about 2 mm. long, it is No. 2.] BLATTA AND' DORVPffORA. 5(3g reduced to a small, somewhat crescentic body, which, unless its position has been marked, is easily mistaken for a yolk body, as it in no way differs from the surrounding yolk in ability to take up coloring-matters. Hence it happens that many investi¬ gators have supposed the insect egg to become enucleate at this time. The resemblance between a yolk polyhedron and the remains of the germinal vesicle is greatly increased by the gran¬ ular contents of both. In the germinal vesicle these granules are the comminuted remains of the large masses of chromatin so conspicuous in the young egg; in the yolk bodies they are albu¬ minous substances destined, like the other yolk materials, to become food for the protoplasm. At the very spot where the nucleus degenerated, viz. at the middle of the concave side of the egg, there appears in eggs almost mature (2.5-2.8 mm. long), a cluster of numerous chromatin granules, which I believe to be the same as those in the germinal vesicle, grown more conspic¬ uous by aggregation. Stuhlmann has figured several bodies like my Fig. 10, representing this aggregation which has pro¬ gressed considerably in the centre. These chromatin granules are probably uniting to form filaments preparatory to karyo- kinesis. The aggregation takes place in such a way that in eggs treated with Perenyi’s fluid a narrowly oblong mass of chromatin is formed, an appearance undoubtedly due to the fusion of the separate filamentous loops. The oblong mass is represented in Fig. ii. When the outlying granules have been added to it, and the achromatic fibrillae have made their appear¬ ance, the first polar spindle in the metakinetic stage is completed (P'ig. 12). The axis of the spindle is directed at right angles to the surface of the egg, which is now mature though still enveloped in the follicular epithelium. In Fig. 13 the equatorial plate has divided transversely, whether by longitudinal fissure of the individual loops or not, I am unable to say, and the two masses of chromatin are on their way to the poles of the spindle. Arrived at the poles the masses become spherical, and the achromatic spindle fades away, while the outer sphere of chro¬ matin surrounded by a mass of protoplasm full of Blochmann’s corpuscles is almost constricted off from the egg to form the first polar globule. While this is taking place the egg is being freed from all its epithelium, except the cap at the cephalic pole, im¬ pregnated and placed in the capsule. The daughter nucleus of the 510 WHEELER. [VoL. nr. first polar spindle remaining within the egg now forms another spindle directed like the first. This is seen in Fig. 14. The blade of the microtome has somewhat raised the loosely at¬ tached first polar globule, the protoplasm of which is seen to contain many of Blochmann’s corpuscles. The karyokinesis" of the second spindle progresses in the same manner as the first and the second polar globule is given off, also surrounded by bacillar protoplasm. The portion of the nucleus which becomes the female pronucleus, is not seen in Fig. 15, where only the two polar globules are represented, as it appears in the next section. Figure 16 is from a surface view of the polar glob¬ ules. By comparison with Fig. i^pgl i it will be seen that they are lenticular. Here the female pronucleus appears as a more indistinct body (because out of focus) lying between the polar globules. The polar globules which I have been able to find in many eggs taken from the capsules while they were still vertical in the genital armature (about 6 to 12 hours after the begin¬ ning of oviposition) lie in the middle of the convex dorsal wall of the oothecal egg. They do not divide subsequently to form what Weismann (48) calls secondary polar globules, but soon disintegrate. In eggs about one day old their remains may be found as an amorphous granular mass, lying just beneath the chorion and entirely separated from the egg. The female pronucleus increases considerably in size before leaving the surface of the egg (Fig. 179 p^t)- This increase is gradual but constant as it makes its way through the dense yolk of the interior of the egg to the apex of the isosceles tri¬ angle of homogeneous yolk abutting on the back part of the granular ventral yolk. It is on this journey that the female pronucleus meets the male pronucleus formed from a spermato¬ zoon which has entered the egg by one of the funnel-shaped micropyles on the upper ventral face. Though I have succeeded in throwing a little more light on the copulation of the pronuclei than Blochmann, I cannot regard my observations as completely satisfactory. The process must be studied in Arthropod eggs with more evenly compact yolk than the eggs of Orthoptera, the numerous cracks and fissures in which render the observation of delicate internal processes exceedingly difficult if not impossible. Moreover, the copula¬ tion of the pronuclei in Blntta is hurried through very rapidly. No. 2.] BLATTA AND DORYPHORA. 311 In more than a hundred eggs which I sectioned from capsules 6 to 24 hours old I found the greater number in the polar globule stage and all the remainder, with two exceptions, in the stages just before, during and after the division of the cleavage nu¬ cleus. In the two exceptions which I describe and figure I found what I take to be the pronuclei. The arrows in the figures point in the direction of the long axis of the egg, only the circumnuclear portions of which are represented. In Fig. 19 the female pronucleus is about one-third the dis¬ tance from the cephalic end instead of being in the middle of the egg where the polar globules are formed. It is no further from the dorsal surface than the female pronucleus in Fig. 17, but is much larger. Hence I believe it has travelled up along the dorsal surface to meet the male pronucleus (Fig. 19 pn)^ which has advanced through almost the whole dorsoventral diameter of the egg. The female pronucleus exhibits the usual coiled chromatin filament. The male pronucleus is granular, of somewhat irregular outline, and surrounded by vacuolated pro¬ toplasm. It is rather deeply stainable in borax carmine. In this Fig. 19 the pronuclei are of very different size, and not¬ withstanding the male pronucleus has passed through three- fourths of the dorsoventral diameter, a distance of f mm., it has not increased much in volume when compared with the mass of the long though attenuated head of the spermatozoon from which it originated. No astral radiation could be seen surround¬ ing these pronuclei. In Fig. 20 we have what I take to be the two pronuclei conjugating The smaller presumably male portion of the compound or cleavage nucleus is larger than the male pro¬ nucleus in Fig. 19. The place of union is about the middle of the homogeneous yolk, i.e. about one-third of the dorsoventral diameter from the back of the egg. The nuclei of Fig. 19 would probably have fused at a point near the dorsal surface one-third the distance from the cephalic to the caudal- pole, but the paths of these two nuclei were undoubtedly aberrant. My observations on single female pronuclei and cleavage nuclei found at various points along the median dorsoventral diameter lead me to conclude that the middle of the homogeneous yolk is, the normal point of conjugation. I have represented (Fig. 18) another female pronucleus from near the point of fusion because its structure is different from that of the other female pronuclei 312 WHEELER. [VOL. III. figured. Its chromatin is limited to a few filaments ; whereas in Fig. ig 9 pi we have a long and intricately wound coil. The cleavage- nucleus continues the course begun by the female pronucleus. Figure 21 represents it when it has reached the middle of the dorsoventral axis. The cytoplasm surround¬ ing this nucleus, which is drawn from an egg hardened in picrosulphuric acid, shows an astral radiation. The cytoplasm passes by insensible degrees into the surrounding homogeneous yolk substance. It is as if the female pronucleus when it left the polar globules took a little mass of the protoplasm abun¬ dant at that part of the egg’s surface and travelled with it, making it convert the yolk into protoplasm during the journey. As soon as the cleavage nucleus reaches the front edge of the mass of homogeneous yolk, or has advanced a very short distance further into the granular yolk, it stops and begins to increase in size till it becomes a clear vesicle (Fig. 22) in which the chromatin, broken into irregular fragments, lies scat¬ tered through the finely granular contents. The nuclear wall grows fainter and disappears. The large spindle now appears, and the typical process of karyokinesis is carried on (Fig. 23). The polar axis of the spindle was in all but one of the many cases in which I observed the division, parallel to the long axis of the egg. In the exceptional case (possibly an abnormal egg) it was parallel to the dorsoventral diameter. Even before the division of the nucleus commences, the polar axis of the future spindle is foreshadowed, as it were, in the shape of the granular (amoeboid) cytoplasm, which, as may be seen in Fig. 23 cp^ is elongated in a direction parallel to the cephalocaudal axis of the egg. The karyokinetic process employed in this and the subsequent divisions is typical. Soon after the nuclear wall again makes it appearance the two nuclei with the cytoplasm which surrounds them separate about one-fifth the longitudinal diameter of the egg, and then prepare for the next division. The polar axes of the two daughter spindles now formed are very nearly, at right angles to the polar axis of the mother nucleus. In all the first divisions the perfect isochronism of the differ¬ ent stages in the different nuclei is striking. Fig. 24 represents two spindles from an egg containing four nuclei in exactly the same (metakinetic) stage. In this figure, taken from a longitu¬ dinal section, one of the spindles (c) has its axis at right angles No. 2.] BLATTA AATD DORVPHORA. 313 to the longitudinal and also to the dorsoventral axis of the egg. The loops forming the equatorial plate are readily seen and counted. Ten in number, each probably attached to an achromatic fibril, they are arranged in such a manner that seven form a circle inclosing the remaining three. The divisions of these four products of the cleavage nucleus continue till about 60 to 80 cells have been formed, scattered irregularly through the granular ventral yolk. The axes of the spindles are inclined in various directions, and nothing in¬ dicates an early differentiation into cells destined on the one hand to remain in the yolk, and on the other to form the blastoderm. The numerous amoeboid cells next migrate to the surface of the egg. In Fig. 25, representing the ventral third of a median cross- section, two of these cells have just reached the surface, while one is still on its way. On reaching the surface the cells first become somewhat conical, and then gradually flatten out. The tension of the cytoplasm is so great that the inclosed nucleus is forced to become lenticular (Fig. 26 a). The cells which have reached the surface, and are much scattered over the roof-shaped ventral face and the adjacent portions of the lateral faces, com¬ mence dividing tangentially, not by karyokinesis, as heretofore, but by akinesis. Figure 36 represents the lateral surface view, and Fig. 26 part of a transverse section of an egg in this stage. The division of the nuclei which have reached the surface is very rapid, and compact colonies of cells of different sizes and in different stages of the unequally constricting process charac¬ teristic of akinesis may be seen embedded in amoeboid masses of protoplasm. I have given such a syncytium (enlarged from Fig. 36) in Fig. 34, and two of the dividing nuclei from other parts of the same egg in Fig. 35 b, c. The method of division is exactly like what was described above for the cells of the fob licular epithelium (Fig. 5), omitting the peculiar nucleoli which I have not been able to detect in these nuclei. My observa-- tions tend to show that all the future divisions in the formation of the blastoderm and those subsequently undergone by the serosa are akinetic, the densely coiled chromatin filament re¬ maining inert, and the division taking place by a constriction which often produces two daughter nuclei of very unequal size. I emphasize the fact that these forms of division could not have 314 WHEELER. [VOL. III. been produced by reagents, as these eggs were hardened in picrosulphuric acid or simple alcohol, which in younger and older eggs preserve the karyokinetic figures of the cleavage nucleus and its immediate descendants with great clearness. The nuclei of the small syncytia spread apart evenly over the surface of the egg, which now presents the appearance of Fig. 37. The pseudopodia of the amoeboid cytoplasmic masses run together to form a net. The egg is now in the blastema stage of Patten (38). The cells at the surface are being continually reinforced by cells migrating from the yolk. Ever since the first division of the cleavage nucleus the nuclei have undergone a gradual and steady diminution in size, and this progresses till the formation of the blastoderm which takes place by the division of the blastema cells. In this stage the yolk is covered with a layer of protoplasm, imperfectly divided into small cells, each containing a lenticular nucleus, which in turn contains two very deeply stainable nucleoli (Phg. 33). Fig. 27 is a trans¬ verse section through the front of an egg in the blastoderm stage. All the protoplasm at the surface of the egg is carried there by the migrating cells or formed from the surface yolk through their influence, as the Keimhaiity so highly developed in Doryphora, as will be seen further on, is undeveloped on the ventral, lateral, and all but a very small portion of the dorsal surface of the egg of Blatta germartica. All the nuclei, formerly in the yolk, probably 7 'ise to the stir- face to fortn the blastema and reinforce it in its formation of the blastoderm. Before the blastoderm is completed, cells sep¬ arate from it and pass inwards to form the yolk cells, or vitellophags. The following are my reasons for believing that all the products of the cleavage nucleus go to the surface. The cleavage nucleus cells have large pale nuclei ' and distinctly amoebiform cytoplasm, like those in the yolk of Figs. 25 and 26. Subsequently none of these nuclei are to be found in the yolk, but in their stead occur at greater or lesser distances from the ventral and lateral faces small deeply stainable nuclei of exactly the same size as the blastoderm nuclei,, not surrounded with amoebiform cytoplasm, but apparently melting their way through the yolk, often in the middle of a dense yolk body, and, above all, exhibiting the same intimate structure as the blastoderm nuclei (Fig. 28). When treated with picrosulphuric acid, these centrip- No. 2.J BLATTA AND DORYPHORA. 315 etal nuclei show the two highly refractive nucleoli of the surface cells; when treated with alcohol only, both the surface and yolk nuclei exhibit the same closely wound, deeply stainable chromatin filament. Such exact similarity in size, shape, and minute structure is very strong evidence in favor of community of origin. Still I have not been able to find an egg without nuclei in the yolk, notwithstanding I sectioned many eggs in the blastoderm stage. I am inclined to think that such a stage may not occur, but that the last of the cleavage nucleus products go to the surface simultaneously with the passage in the opposite direction of the first blastoderm products. Thus the enucleate yolk stage would be slurred over. The time required for the development so far described is approximately as follows : The first polar spindle is formed in the ovaries ; the second polar spindle during oviposition. Both polar globules have been constricted off by about the sixth hour from the commencement of oviposition. By the end of the first day the female pronucleus has fused with the male pronucleus, and the cleavage nucleus thus formed has reached the back of the granular ventral yolk. The products of the cleavage nucleus are formed and reach the surface by the end of the third day. By the end of the fourth day the blastema is completed. During the fifth day the blastema nuclei proliferate and complete the blastoderm by the close of the sixth day. The development is, of course, accelerated by a rise and retarded by a fall in temperature, though not to the extent observed in many other animals. Before passing over to a description of the early stages of the egg of Doryphora, I will recapitulate the movements of the nuclei by means of diagrams of a longitudinal and equatorial cross-section of the egg (Figs. 39 and 40). In these diagrams a is the cephalic, b the caudal end, c the ventral, d the dorsal, 0 the lateral surface. The shaded body p is the homogeneous yolk, the dotted portions r the granular yolk. The germinal vesicle starting from the central point e goes to the surface, describing the path represented by the line ef. Here it gives off by two successive divisions the two polar globules and and then as the female pronucleus turns back to go in the opposite direction. The line representing the pas¬ sage of the germinal vesicle to the surface is really too long, 3 i6 WHEELER. [VOL. III. as this path is travelled over when the egg is much smaller, and the distance from the centre to the periphery much less than in , the mature egg. The nucleus of the sperma¬ tozoon entering the egg at some point on the ventral surface between and r, probably at or near h, advances through the egg as the male pronucleus i and fuses with the female pronucleus, coming from the opposite direction, at a point near k. The segmentation nucleus passes on to the point / and divides in the direction of the anteroposterior axis of the egg, giving rise to the daughter nuclei nun. By sub¬ sequent divisions these give rise to the blastema nuclei, which migrate to the ventral and ventrolateral surfaces of the egg (;/). It is by a tangential division of the blastema cells, forming a layer of much smaller cells, which creep around the sides of the egg and close on the dorsal surface, that the blastoderm is completed. Doryphora. Turning now to Doryphora^ we find that the ovariole differs from that of Blatta in one particular: The terminal thread di¬ lates below into a large oval chamber {Endkamnier), the mem¬ branous wall of which is closely packed with cells of various sizes, containing nuclei which vary in size as the inclosing cytoplasm varies in volume. The nuclei contain delicate, much convoluted chromatin filaments. Besides the difference in size, no other differences are perceptible between the different cells. At the lower end of the Endkammer the differentiation of the cells into ova and follicular epithelium takes place. Careful examination of many sections has convinced me that none of the peculiar phenomena described by Will (51) in the oogenesis of Nepa are to be observed in Doryphora. What I have seen is in perfect accord with Leydig’s observations (29). The large cells of the Endkammer become the ova, and the smaller cells, after undergoing a further reduction in size by division, become the follicular epithelium. In the two upper follicles the ova resemble in every respect the large cells of the terminal chamber, the nuclei retaining exactly the same perfectly spherical form, and the same distribution of their chromatin. In the ovum of the third follicle the chromatin No. 2.] BLATTA AND DORYPHORA. 317 has assumed a different appearance. It is no longer distributed in the long, even, much convoluted filament, but has broken up into several spherules which stain more deeply. One or more large vacuoles are to be found in each one of these nucleoli, or masses of chromatin, which under a high power seem to hang suspended in the meshes of the nuclear reticulum in the same manner as the homologous bodies of Blatta. From the time of its first formation till the ovum has attained a considerable size, the fol¬ licular epithelium is columnar with its elongate nuclei directed at right angles to the long axis of the egg. Later the epithelium flattens, and the nuclei become kidney-shaped, with their long axes tangentially directed to the surface of the egg and their hili directed inwards. The yolk first makes its appearance in the form of numerous granules. In no case have I seen a degeneration of the follic¬ ular epithelium to form yolk. The large granular yolk spheres soon make their appearance. As these bodies are much smaller than, but in other respects very similar to, those in Blatta^ I have given little attention to their study. Though the proto¬ plasm is reduced to a very delicate reticulum by the great ac¬ cumulation of yolk spheres, there remains till the formation of the blastoderm, contrary to what I have observed in Blatta, a thick layer of finely and evenly granular protoplasm, which en¬ velops the whole egg and is equivalent to Weismann’s Keimhaiit, though it is present from the first appearance of the yolk in the form of spheres, and does not originate just before the forma¬ tion of the blastoderm, as in several of. the insects studied by Weismann (47). The thick surface layer stains pale pink in borax carmine, and is quite distinctly marked off from the retic¬ ulate yolk-charged protoplasm of the interior of the egg. After the nuclear filament has become metamorphosed into the spherical vacuolated masses above described, the nucleus moves from the centre of the egg to the surface, travelling along a line at right angles to the polar diameter of the egg. It thus reaches a point midway between the poles, taking the same position as the germinal vesicle of Blatta. During its migra¬ tion the karyoplasm becomes amoeboid, and except at its outer surface retains its irregular form even after taking its position right under the follicular epithelium. The outward directed face becomes excavated, the rest of its surface remaining con- 3 i8 WHEELER. [VoL. iir. nected with intervitelline protoplasmic trabeculae (Fig. 56 tr). The karyoplasm of the nucleus is coarsely granular. In the nucleus figured only two of the vacuolated masses of chromatin (or nucleoli, as most writers call them) are present. The larger contains six vacuoles, one of which incloses a bar-shaped mass of chromatin (Fig. 56 nl). In the cavity of the nucleus a number of more or less oval hyaline masses are seen. They are doubt¬ less the equivalents of the “ maturation spheres,” noted above in Blatta. Stuhlmann (45) has found these same spheres in the degenerating germinal vesicle of LinUy a Chrysomelid allied to Doryphora. The next stage found in the decomposition of the germinal vesicle is represented in Fig. 57. The karyoplasm has become confluent with the intervitelline protoplasm, and only the chro¬ matin portion marks the spot where the nucleus reached the sur¬ face of the egg. The larger yolk spheres, formerly present even in the surface protoplasm, have passed inwards, and only the smaller spheres still remain in what is to become the blastema. Soon these, too, retire further into the egg, and the surface pro¬ toplasm is marked off from the vitelliferous portion. The re¬ mains of the nucleoli are worthy of attention. The vacuoles have disappeared, and the glistening chromatin has grown denser, and stains very deeply. In the case figured, one or two nucleoli have evidently broken into the nine fragments of different sizes and shapes. The larger and more peripheral mass (Fig. 57 «) is surrounded by a pale aureole. Its size, position, and the clear protoplasm surrounding it, seem to point it out as the im¬ portant and permanent mass of chromatin soon to be converted into the first polar spindle. The remaining eight fragments seem to be leaving the surface and passing into the yolk, where they probably disappear, as no traces of them can be found in succeeding stages. In Fig. 58, which represents the next stage in the nuclear metamorphosis, we find that the mass n of Fig. 57, which con¬ tained five spherical masses of dense chromatin, has become a perfect oval nucleus in the resting stage. The chromatin has again passed into the filamentous state. The surface layer of protoplasm is clearly developed and has secreted the vitelline membrane (Fig. 58 v). The chorion, too (Fig, 58 ch)y has ap¬ peared. The follicular epithelium which secreted it is omitted in this and the next figures. No. 2.] BLATTA AND DORYPHORA. 319 The resting nucleus just described soon begins to divide. Its small size prevents an accurate understanding of the peculiari¬ ties of its mode of division. Enough can be gleaned from Fig. 59, however, to show that the karyokinesis is not typical like that described by Flemming (12, 13) and Rabl (41) for the epithelial nuclei of Salamandra. No loops seem to be present in the metakinetic stage, but the chromatin is arranged in mo- niliform strings, each of which seems to be applied full length to one of the achromatic fibres of the spindle. I have not seen all the stages in the metakinetic process. In the next stage observed (Fig. 60) the two masses of chro¬ matin resulting from metakinesis have reached the poles of the spindle, where the fibres before apparent have become invisible. In the equatorial plane, however, the achromatic filaments are very distinct, being noticeably thickened. This ^ thickening of the spindle fibres to form the dividing plate {b) between the two cells is of universal occurrence in plants, as may be seen from an examination of Strasburger’s (44) delicate figures. It is, how¬ ever, not infrequent in Arthropods. Carnoy (9) figures many instances in his “ Cytodierese.” Unfortunately I have been unable, through lack of material, to trace the changes of the nucleus immediately following those just described. The ovum is deposited by the female Doryphora with its nucleus is the stage represented in Fig. 60 (compare Blattd). The outer mass of chromatin {p^) must be regarded as the first polar globule. Probably the process of forming the second polar globe is essentially the same as in Blatta. Though incomplete, my observations prove, I think, that polar globules are also formed in the Coleoptera, thus adding another order of insects to the three in which these interesting bodies were found by Blochmann (5). As in Blatta, the nuclei represented in Figs. 57 to 60 are very small and difficult to find in the enormously larger eggs. Long before oviposition the eggs of Doryphora have acquired the dull orange color which makes them so conspicuous on the under surfaces of the leaves to which they are glued by the females. The coloring-matter is seated in the yolk bodies, and ultimately disappears in eggs preserved in 70 per cent alcohol. The eggs are most deeply colored in living specimens in the earlier stages, and a gradual fading of the color accompanies the 320 WHEELER. [VOL. III. gradual metabolism of the deeply colored yolk into pale yellow- protoplasm. As the yolk spheres are much smaller in Doryphora than in Blatta, mutual pressure does not make them so clearly polygonal. Though granular, the protoplasm, which is spread in such a thick layer over the surface of the yolk, contains none of the bacillar bodies so easily demonstrated in Blatta. It is of course possible that in the beetle’s eggs they may be present, but of much smaller size and of spherical shape like those found in the Lepidopteron Pieris. The chorion in Doryphora is thick and somewhat leathery, though easily torn with the dissecting-needles. It seems to re¬ semble in every way that of Lma as described by Graber (15). The surface layer of protoplasm secretes a very delicate and structureless vitelline membrane, which in the younger stages is closely applied to the surface of the egg. Soon after the formation of the ventral plate it is loosened and stands off from the surface. Besides some clear, irregular patches on the sur¬ face of the chorion I have seen no structures which could be interpreted as micropyles. As there are strong reasons for supposing that the cleavage nucleus is situated in the very centre of the egg while dividing, the copulation of the pronuclei must take place along the radial line joining the centre of the egg to the point where the polar globules are formed. As I possessed no eggs immediately after oviposition, the phenomena of pronuclear conjugation, which could probably be more favorably studied in the eggs of Dory- phoi'a than in Blatta, on account of the perfect and even sec¬ tions obtainable, were not observed. The first stages after the one given in Fig. 60 which I have been able to find in my material showed the division imme¬ diately following the first and second divisions of the cleavage nucleus. As the few nuclei were all near the centre of the egg, and as Graber has found the cleavage nucleus is the centre of the very similar eggs of the allied Lina, I feel justified in believ¬ ing that this is the point at which the first division occurs. The products of the cleavage nucleus go through karyokinesis, but owing to their much smaller size the process is much less dis¬ tinct than in Blatta. In eggs stained with borax carmine, the cytoplasm of each cell appears as a delicate pink cloud among No. 2.] BLATTA AND DORVPHORA. 321 the grayish yolk spheres, and very high powers are necessary to detect the delicate rays of the spindle or even the minute gran¬ ules of chromatin in the metakinetic and succeeding stages. The isochronism among all the nuclei in the different stages of development up to the formation of the blastoderm is quite as apparent as in Blatta. Judging from the great number of eggs with resting nuclei and the very few eggs with kinetic figures, I conclude that division takes place very rapidly and is followed by comparatively long periods of quiescence. Fig. 61 represents one-half a cross-section through the equa¬ torial region of an egg containing a number of nuclei in its yolk. All the nuclei are resting and are surrounded by amoeboid masses of protoplasm. These cells often have the appearance of being in motion, most frequently in concentric paths. The nucleus is in the broader portion of the comet-shaped cell, which seems to be advancing head foremost. These cells are surrounded by numerous minute vacuoles that under the low power, with which Fig. 61 was drawn, appear like coarse granules. These cells divide rapidly and give rise to many smaller cells scattered through the whole yolk. A few enter the blastema layer and begin to proliferate rapidly. As soon as this mi¬ gration to the surface has taken place, the cells which have remained in the interior, and zvJiich do 7 iot go to the surface, stop dividing and take up positions at short but nearly equal distan¬ ces apart, through the whole yolk. Before assuming their defi¬ nite positions they have multiplied so rapidly that one may frequently see strings of three or four cells (Fig. 62 j/;d). Often, too, in this stage most of the cells are in pairs, or, more accurately speaking, the egg contains many binucleate cells. The nuclei which have entered the surface layer of pro¬ toplasm divide tangentially. Sometimes the axis of the spindle is inclined at an angle less than 90"" to the radius of the circular cross-section, but in no case have I seen a spindle with its axis directed radially. The first divisions of the nuclei, which have entered the blastema, give rise to an even layer of cubical cells. By one more division of its constituent elements this blastema is converted into the blastoderm, which consists of smaller and more columnar cells. Sections taken in all directions through the egg show the blastoderm to be of even thickness over the whole surface (Fig. 63). 322 WHEELER, [VOL. III. General Remarks. a. Nuclear Continuity. It will be seen from the above descriptive paragraphs, that I maintain that a portion of the chromatin of the insect egg visibly survives in the decomposition of the germinal vesicle and can be traced through the divisions resulting in the formation of the two polar globules into the cleavage nucleus and its descendants. This conclusion, which has always been held by careful investigators of the transparent ova of lower forms, has been seriously questioned of late by two workers, Stuhlmann (45) and Henking (20). The former, after investigating a num¬ ber of Arthropod eggs in a superficial manner, comes to the conclusion that a stage exists in the ontogeny of the ovum when no traces of a nucleus can be demonstrated. Henking not only indorses this view, but describes in Opilio what are cer¬ tainly the products of division of the cleavage nucleus as arising dc novo in different parts of the egg. As Blochmann (5) has pointed out the errors into which both investigators have fallen with far greater force than I can bring to bear on the subject, I will not increase the length of my paper by entering into a detailed account of their observations. Blochmann’s beautiful researches on the early stages of the egg have proved beyond a doubt that the Hexapoda conform to the fecundative processes and method of polar-globule formation observed in other animals. • That there is no 7 noment.when the nucleus ceases to exist as a nucleus seems to me to be proved by my Fig. ii, where the remains of the nuclear wall are still present while the spindle is forming. True, the wall is absent in the younger stage. Fig. 10, and Fig. ii may represent an exceptional case in which the wall has persisted longer than is usual, but it proves, neverthe¬ less, that the matter composing the nucleus does not diffuse through the protoplasm and ultimately recombine to form the nuclei which give rise to the blastoderm, as Henking (20) would have us believe. The nuclear wall is known to persist in another Arthropod till after the commencement of spindle forma¬ tion. I quote the following from the recent work of Weis- mann and Ischikawa (49): “ Auch in den kleinen und dotterlosen No. 2.] BLATTA AND DORYPHORA. 323 Eiern von Bythostrephes steigt das Keimblaschen bei Eintritt der Eireife in die Hohe, erblasst allmahlich und zeigt zugleich an Eiern, die frisch in den Brutraum iibergetreten sind, den Beginn der Spindelbildung innerJialb des dann noch scharf her- vortretenden Umrisses des Kleimblaschens.” The interesting researches of Weismann and Ischikawa (49), Leichmann’s (28) recent discovery of two polar globules in Asel- hts and Pereyaslawzewa’s (40) discovery of polar globules in Gam^nariis poecilurus prove that the Crustacea^ too, must be included under the general law. b. The Laiv of Orientation. Though considerable attention has been given to ookinesis in Echinoderms and Amphibia, no extended observations of these phenomena in the eggs of Arthropods have been published. In view of this fact, I have taken considerable pains to deter¬ mine the paths of the pronuclei and cleavage nucleus in Blatta^ and have devoted considerable space to their description. A perusal of O. Hertwig’s paper entitled, “ Welchen Einfluss iibt die Schwerkraft auf die Theilung thierischer Zellen V (21), led me to try experiments to prove whether gravitation has any appreciable effects in determining the position of the embryo in the egg, or .whether the position of the embryo with refer¬ ence to the yolk is predetermined long before, during the de¬ velopment of the egg in the follicle. It has been shown that the eggs of Blatta are carried in a hori¬ zontal position after the first day, so that the ventral or germinal faces of the upper row of eggs are directed downwards, and those of the lower row upwards. As the crista is continuous with the right side of the parent’s body, the heads of the em¬ bryos all point to the right. It occurred to me that by keeping capsules in various positions any effects of gravitation on the development of the inclosed eggs could be easily determined. I accordingly placed capsules in five different positions on a block of paraffine provided with holes and grooves to keep them firmly in place. To prevent desiccation, the block was kept in a camera humida. Capsules were kept from fourteen to twenty days in the fol¬ lowing positions : — 324 WHEELER. [VOL. III. 1. Resting with the lateral faces perpendicular and crista uppermost. 2. Resting on the crista with the lateral faces perpendicular. 3. Resting on the left lateral face. 4. Resting perpendicularly on the anterior end. 5. Resting perpendicularly on the posterior end. In all these cases the eggs developed normally, without the slightest indication of displacement in position or alteration of shape in the embryos; whether they were forced to develop with their heads pointing up or down. The development was slow in all of the above cases, but this was not due to the unnatural positions of the capsules, but rather to the low temperature produced by the evaporation of the water under the bell-jar. That this was the true cause was shown by capsules kept under the same bell-jar in the normal position, on the right side. Their development was likewise retarded. We may conclude from these few experiments that the force of gravitation has no perceptible effect on the development of the eggs of Blatta, but that these highly differentiated eggs, utterly unable to revolve in their envelopes like the eggs of birds and frogs, have their constituents prearranged, and the paths of their nuclei predetermined with reference to the parts of the embryo. As the only difference between the mature ovarian and the odthecal egg is a difference in shape, we con¬ clude that the predetermination is effected before fecundation, and even before the formation of the first polar globule. There is nothing in the structure of the newly laid egg of Doryphora to prove that it possesses a dorso-ventral differentia¬ tion like the egg of Blatta ; nor are my facts sufficient to war¬ rant the assertion that the ventral plate develops on the side opposite the point at which the polar globules arise. Still the possibility of such a condition is in nowise precluded, and the observation of Blochmann and myself on Blatta probably apply to the Hexapoda in general. The spherical form of the Crusta¬ cean egg as opposed to the oval shape of the great majority of insect eggs, will be a great obstacle in the way of proving any similar conditions in this lower group. It can be proved, however, that we have as true an anteropos¬ terior differentiation in the eggs of Doryphora as in Blatta. No. 2.] BLATTA AND DORYPHORA. 525 The eggs are deposited by the females in such a way that the pole which leaves the vagina first is glued to the surface of the leaf by a semifluid secretion, which at this point spreads out into a flat disc on which the egg rests and by which it is attached. As the hatching larva always leaves the egg head foremost at the opposite free pole of the egg, and as there is no revolution of the embryo as in Hemiptera, there can be no doubt as far as the cephalic and caudal ends are concerned that the relations of the embryo to the parts of the egg are the same as those described for Blatta. Moreover, the position of the eggs on the leaf and the position of the embryo in the egg, are sufficient evidence that the eggs are oriented in the ovaries of Doryphora with the cephalic pole directed towards the head of the mother insect. Thus I have found that Blatta completely and Doryphora cer¬ tainly in part conforms to the “ loi de I’orientation de Toeuf ” of Hallez (17), who found that the ova of Hydrophihis and Locusta lie in the ovaries with their cephalic ends directed towards the head of the mother insect, and that the dorsal and ventral surfaces of the egg are predetermined in the ovaries. Kadyi’s (23) remarks make it certain that Periplaneta conforms to the law. In viviparous Aphides the same condition obtains, as may be gleaned from the plates of Metchnikow (30) and Will (52). When micropyles are developed at the cephalic pole of the egg, they form a fixed point which is of great assistance in observing whether the “law of orientation” obtains in a particu¬ lar instance. We have such a case in Corixa as described by Metchnikow. In this insect (as also in Aphis), the entoblastic growth of the embryo somewhat obscures the process ; but it can be readily seen that when the ventral plate forms, the por¬ tion of it which will subsequently grow out into the procephalic lobes is situated at the micropylar pole, which is anteriorly di¬ rected while in the body of the mother insect. The growing into the yolk of the embryo tail first brings the head to the opposite end of the egg, but during revolution the embryo regains the position which it held before the formation of the amnion and serosa (see Metchnikow (30), Plates XXVT. and XXVII. Figs. 6, II, 20, 25, 27). 326 WHEELER. [VOL. III. The Formation of the Germ Layers and Embryonic Envelopes. DorypJio 7 'a. As my observations on the formation of the germ layers in the potato-beetle are both more copious and more satisfactory than in Blatta, and as Doryphora probably represents more nearly the typical process of germ layer formation, I will begin my remarks on this subject with DorypJiora, and append what I have to say on Blatta. The first change visible in the blastoderm from the surface is the appearance of a pair of folds which arise on the middle of what is to be the ventral surface of the egg (Fig. 66). The two folds may best be described as resembling a pair of but slightly bent parentheses close together. In cross-sections these folds are scarcely perceptible, and the difference in thickness between the ventral and dorsal blastoderm is very slight. Soon the folds became more decided (Figs. 67 and 68), their anterior ends are continued around and meet so as to inclose a spade-shaped space, while their posterior ends diverge and are continued in the opposite direction to the caudal end of the egg, where the depression inclosed between the two folds turns inward and ends abruptly. The depression inclosed by the fold is to become the groove-shaped gastrula. The nuclei of the blastoderm now present a very different appearance from that represented in Fig. 66. The fold-surrounded depression at the anterior end (Fig. 67 d), lying where the stomodaeum is subsequently invaginated, is apparently a differentiated por¬ tion of the gastrula, to judge from its peculiar shape. In lateral view (Fig. 68 d) this portion of the blastoderm is seen to be concave, the nuclei are closely aggregated, whereas an¬ teriorly and posteriorly they are much scattered and have increased in size. The true distribution of the cells, as shown by their nuclei, is best seen in the lateral view of the same egg (Fig. 68). The aggregated mass of cells, or the ventral plate as it may now be called, is clearly marked off from the serosa or remaining blastoderm, which . is recognized by its larger and more scattered nuclei. The ventral plate is seen to be constricted toward the middle of the egg to form two No. 2.] BLATTA AND DORYPHORA. 327 lobes ; the anterior broader and shorter is the procephalic, and the posterior and longer the abdomino-thoracic lobe. In both views of the egg (Figs. 67 and 68) a number of lines are seen crossing the ventral plate at right angles to the longer diam¬ eter. These lines make the plate appear segmented, and at first reminded me of Kowalevsky’s figure (Plate VIII., Fig. 2), which represents an embryo HydropJiiliis in exactly the same stage as that which I have figured. More careful examination, however, convinced me that the lines were not due to segmen¬ tation, nor, in fact, depressed at all, but were the wrinkles into which the ventral plate was thrown, probably by a contraction away from the anterior pole. The simple method by which the ventral plate is formed is easily seen in a median cross-section of an egg in the stage fig¬ ured in Fig. 67 (Fig. 64). The small projection r is the ridge which separates the thick and sinking gastrular portion of the blastoderm {g) from the remainder of the layer. As we have seen, all the cells of the completed blastoderm are columnar. The thickened ventral plate is formed merely by the cells on the ventral surface of the egg lengthening in a radial direction, and those of the dorsum (P'ig. 647) changing their shape so as to have their long axes tangentially directed. With further development the gastrular groove deepens and the ridges come close together in the median line (Fig. 70). The oral end of the gastrula {ci) is oval, and is no longer marked off anteriorly by the ridge seen in Fig. 68. On each side of the oral widening is seen a small fold {ani) between which and the gastrula the ventral plate cells are much thickened. This fold is the beginning of that portion of the amnion which sub-. sequently envelopes the head, and the thickenings are probably the first traces of the brain. The egg (P'ig. 70) viewed from the posterior end discloses some interesting facts (Fig. 69), The ventral plate has become pushed in at this point, and the thickened lip thus formed grows forward over it. This lip is the beginning of the caudal fold of the amnion and serosa. The gastrular invagination is considerably deeper at its posterior than at its oral end. The caudal fold of the amnion and serosa grows much more rapidly than the cephalic folds. The tail end of the ventral plate advances dorsally till it is one-third the length of the egg 328 WHEELER. [VOL. III. from the anterior pole. It pushes its way through the yolk in such a manner as not to bring the amnion or serosa in close apposition, but to inclose a greater or less amount of yolk between the two membranes. This dorsal growth of the caudal end shows considerable variation, however. In some eggs the end of the tail comes to lie at such a distance from the dorsal surface that it is almost in the centre of the yolk. In other instances the amnion and serosa become closely apposed and extrude the yolk from between them soon after the two enve¬ lopes have closed over the mouth, and the embryo has attained its maximum length. In Fig. 71 I have represented the much curved embryo in this stage straightened out. The gastrular invagination has not yet closed, but it is much narrowed and more lengthened on account of the greater growth in length of the embryo. The oral end is still wider than the remaining portion. At the anal end (Fig. 71 x) the groove seems to bifurcate. At the point x the procto- deeum is subsequently invaginated. Usually in embryos as old as that represented the anterior half of the gastrula has closed completely. In the embryo figured the cephalic, maxillary, thoracic, and abdominal portions are already marked out. The first begin¬ nings of the three pairs of legs are apparent in the undulating edge of the thoracic portion of the ventral plate A median cross-section of the egg in the stage just described cuts the embryo at right angles to its long diameter in two places (Fig. 65). At the posterior end the thick amnion {am) is separated from the serosa by a layer of yolk {y), which will shortly be pushed out from either side to join the great central yolk mass. In this half of the section which is near the caudal end the gastrula is still open, though its cells have ceased to be columnar, and are dividing rapidly to form the thick lump of cells so very conspicuous at this point in slightly later stages. In the anterior portion of the egg the gastrula has closed, and the ectoderm and mesoderm are clearly separated. The tubular walls of the gastrula have broken down to form an irregular mass of polygonal cells which lie in the median line closely applied to the outwardly convex ectoderm. Though the clos¬ ing of the gastrula progresses from before backwards, the closure of the oral portion is retarded and is marked by a No. 2.] BLATTA AND DORYPHORA. 329 narrow oval pit which is visible till the mouth begins to in- vaginate (Fig. 73 d). While the amnion and serosa are closing over the oral portion of the embryo the yolk begins to segment. The first traces of segmentation are visible on the dorsal surface between the end of the tail and the procephalic lobes. There the surface of the yolk assumes a scalloped appearance, and radially directed lines soon mark the divisions between the yolk balls. The segmen¬ tation progresses thence in a ventral direction, both from either side of the dorsum and directly inwards, so that finally the whole yolk is reduced to rounded masses, each of which contains from one to three vitellophags. Each yolk segment is properly a cell with its protoplasm radiating in all directions as a delicate reticulum which holds in its meshes the unequal-sized yolk spheres. By the time the amnion and serosa have completely formed, all the yolk has been converted into distinct subspher- ical segments, except the portions immediately under the anal and oral ends of the gastrula. Here the segmentation remains for a short time indistinct till the entoderm is established at these points. In the embryo with completely closed envelopes (Fig. 73, PL V.) the procephalic lobes {pci) have grown in size, and when attached to the egg clasp its upper pole. The three pairs of thoracic limbs are distinctly formed, while of the cephalic appendages only the 2d maxillae {mx 2) are beginning to appear. The base of the attenuated ribbon-shaped abdomen shows traces of commencing segmentation. Posteriorly it suddenly widens out into a flat, transversely oval body, which I shall call the caudal plate. The gastrular invagination is closed except at its anal end {x 73), and the mouth will soon form at the shallow oval depression which marks the anterior end of the groove. Fig. 82 represents a longitudinal section through an embryo a very little older than the one just described (Fig. 73). The serosa {sr) now covers the entire egg, and is separated from the embryo proper and the amnion {ani). Both mem¬ branes present the same appearance in section, being nodulated with nuclei. The anterior end of the embryo lies on the anterior end of the yolk in such a position that its mouth {st) is almost exactly at the pole; the tail ends somewhat anterior to the mid¬ dle of the dorsal surface W- The ectoderm, which at its cephalic 330 WHEELER. [VoL. III. and caudal ends {p and p) passes imperceptibly into the much thinner amnion, is considerably thickened and has its crowded nuclei in several rows, though but one row of cells is present. The depressions which mark off the incipient appendages are deep and narrow. The mesoderm {msd) is spread out under the whole of the ectoderm and has begun to thicken under each somite preparatory to segmentation. It is very noticeably thickened in two places : under the stomodaeal depression (//^) and under the caudal plate (//^), where it forms a large mass of cells projecting into the as yet unsegmented yolk just beneath it. These two masses of cells are the mdependent sources of the ento¬ derm, which grows backwards as two strmgs f'om the anterior mass {pi ^), a}id forward as two strings from the posterior mass {pl‘^). As we shall see further on, these two strings nnite near the middle of the body and then begin to groiv at their lateral edges till the mesenteron thus formed incloses the yolk. The points from which the chords grow are plainly seen in the figure {ent^ and ent‘^). Under both points of proliferation there are a number of nuclei which at first sight under a low power seem to be dividing karyokinetically. The chromatin is all aggregated in one or two dense masses in the hyaline karyo- chylema, and thus resembles the similar aggregations seen in kinetic nuclei. These nuclei, however, are not dividing, but undergoing decomposition, as we shall see when we come to examine a more highly magnified section through the caudal plate. Before leaving Fig. 82 I would call attention to the three cells at c which are on the surface of the embryo in the amniotic cavity. They are very large and clear, and the more anterior is apparently creeping in the manner of an Amoeba along the surface of the abdominal ectoderm. These cells, the ultimate fate of which I have been unable to determine, probably escape from the anal orifice of the gastrula before it closes. I have in several cases seen such cells issuing from or still in connec¬ tion with the infolded pocket of ectoderm, which is called meso¬ derm as soon as the outer layer has closed over it (Fig. 87 c). These peculiar cells may be the homologues of the ‘ Polzellen ’ long ago observed in certain Diptera. A much clearer understanding of the method of formation of the entoderm may be obtained from Figs. 87 and 88, both rep- No. 2.] BLATTA AND DORYPHORA. 33r resenting cross-sections through the middle of the caudal plate. In Fig. 87 , from a younger embryo, the gastrula has not yet closed. Its walls are seen to be much thickened, and the kary- okinetic figures show that its component cells are still prolifer¬ ating. At the lower surface of the bag-shaped mass the cells are somewhat less compact and form a layer {ent) which at some points is separated from the superjacent cell-mass. This mass will give rise to the entoderm, and that above it to the meso¬ derm, as soon as the orifice .r is definitely closed. We thus have a mass of cells in which all three germ layers hlend^ and to no part of which ca7i %e assigned the 7iame of a germ layer. Not till the groove is closed have we mesoderm, and not till the lower cells of the 7nass have beco77ie clea7'ly differentiatedfro7n those above them ca7i we speak of e7itoder7n. The section Fig. 88 is from an embryo in which three germ layers are definitely formed, shortly after the closing of the gas¬ trula. A depression (x) marks the point where the proctodaeal invagination is to occur. The polygonal mesoderm cells are spread out in a mass {77isd), which is separated by a more or less distinct line {/) from the entoderm beneath {e7it). The differ¬ ences between the cells of the last layer and the superjacent mesoderm are difficult to represent. Their nuclei are somewhat larger and clearer. They gradually merge into the mesoderm cells, the boundary being exceptionally clear in the section fig¬ ured. Heider ( 19 ) says of these same entoderm cells in Hydro- philus that they are more “succulent” than the mesodermic elements. This adjective conveys the idea more clearly than paragraphs of description. The peculiar nuclei which under a low power seemed to be dividing are now seen to be in a process of dissolution. They origiTiate m the e7itoder7nic 7nass a7tdpass mto the adjace7it yolk, wlio'e they disappear, sections through slightly later stages showing no traces of them. From what I have seen I believe these nuclei to pass through the following, stages, examples of all of which may be found in a single embryo. The karyochy- lema becomes vacuolated, probably with substances absorbed from without, to judge from the large size of some of these nuclei (Fig. 88 v), while the chromatin ceases to present the threadlike coil and becomes compacted into irregular masses be¬ tween the vacuoles. Finally, the vacuoles fuse and the masses of 332 WHEELER. [VoL. iir. chromatin, formerly numerous, agglomerate to form one or two large irregular masses which usually apply themselves to the wall of the clearly vesicular nucleus (Fig. 88 t). The wall of the nucleus then ceases to be evenly spherical, and becomes irregular apparently because the karyochylema is escaping through a rent (Fig. 88 6 ). In the last stages seen the masses of chromatin lie between the yolk bodies, all the other portions of the nucleus having disappeared. They still take the char¬ acteristic deep red stain, but finally become comminuted and disappear in the intervitelline protoplasm. The dissolution of these nuclei and their migration into the yolk is brought to a close soon after the entodermic mass begins to grow forward. The oral mass of proliferating cells is essentially the same as the caudal mass just described; but being smaller, I have not seen fit to represent it in the plates by enlarged figures. Some of the entoderm nuclei degenerate in exactly the same manner as those described in the caudal thickening, but the whole mass of cells being smaller, the number of these evanescent nuclei is much less. I am at a loss to assign a meaning to this migration of de¬ generating entoderm nuclei into the yolk unless it be supposed that originally all the nuclei of the egg went to the surface and that a portion of the entoderm passed into the yolk to form vitel- lophags while another portion proliferated forward in compact sheets to form the walls of the mesenteron. Later, when the ontogeny was abbreviated in the blastoderm stage by cells being left in the yolk, this migration of entoderm cells became unnec¬ essary, as the yolk, which is already segmented, is copiously sup¬ plied with vitellophags. The lack of distinct yolk segmentation just beneath the two proliferating points may lend some proba¬ bility to this view. I am aware that my explanation halts, but it will have to stand, for the want of a better one, till more facts are forthcoming on these degenerating nuclei in other forms. In examining the literature the only observation which I can find similar to the one just recorded is in Hatschek’s paper on Bonibyx cJirysorrJioea (i8). He observed a mass of nuclei, which in his figures have all the appearance of undergoing degenera¬ tion, just anterior to the large mass of entoderm cells attached to the oral ectoderm. This mass of nuclei, designated by him as a “gland,” soon disappears in the yolk. No. 2.] BLATTA AND DORYPHORA. 333 Blatta. My observations on the formation of the germ layers in Blatta are less satisfactory than those on the same process in DorypJiora^ because the eggs of the former are difficult to section and have small, indistinct cells in the later stages. I have, however, given much attention to the subject, sufficient, I believe, to be able to assert that the method of germ layer formation departs from the type observed in HydropJiiliis and DorypJiora. As soon as the blastoderm is completed by the rapid prolifer¬ ation of the blastema cells, the whole layer of protoplasm with its embedded nuclei contracts from the lateral faces towards the front of the egg. The blastoderm thus becomes exceedingly thin on the lateral and dorsal surfaces of the egg, and the nuclei of these surfaces become much scattered and flattened, while the protoplasm is thickened on the whole ven¬ tral face, where the nuclei are crowded together and have again become spherical. A slight further contraction away from the cephalic and caudal ends towards the centre of the ventral face shortens this mass of thickened cells into the ventral plate. While the blastoderm is thickening, nuclei are being given off centripetally to form the yolk cells. A few of these nuclei go deep into the yolk, but the great majority remain at or very near the surface. They are not given off in a continuous sheet, nor are they produced from the blastoderm by any invagination. They are simply nuclei which have been sent into the yolk from different and often widely separated points of the contracting blastoderm. The few nuclei which descend into the yolk remain for a long time small and indifferent. Sometimes the number of these nuclei is very limited so that they occur in only a few of a great number of complete longitudinal sections passed through an egg. The nuclei of the surface yolk undergo considerable differen¬ tiation, and are soon easily distinguished from the superjacent blastoderm. They surround themselves with stellate cytoplasm, retain their spherical or spheroidal shape, and often present one or more large nucleoli (Figs. 29 and 32 v). Their function for many days is the conversion of the yolk into soluble compounds to be absorbed by the rapidly dividing cells of the embryo. During this process they grow rapidly, and soon become the 3’34 WHEELER. [VoL. iir. largest cells with the largest nuclei in the egg (Fig. 30 v). The long pseudopodial continuations of the finely and evenly gran¬ ular cytoplasm can often be traced for a considerable distance between the yolk polyhedra. The yolk cells are never seen in process of division, and as their number in eggs of widely different stages is approximately constant, I conclude that they rarely or never divide. The thickening process which formed the ventral plate still continues in a spot about one-fourth the length of the egg from the caudal end, and gives rise to a rounded mass of cells which are much thicker than the surrounding portion of the ventral plate though but slightly raised above the general surface (Fig. 41). A faint depression appears in the centre of the rounded mass {bp). While this thickening is forming, the nuclei of the ventral plate are also proliferating very rapidly at two points on the ventrolateral edges about one-fourth the length of the egg from the cephalic end (Fig. 41 pci). These cells also increase in depth, but do not rise above the general surface of the egg. The two patches of thickened cells are the precursors of the procephalic lobes. The deceptive appearance of a groove is presented by the keel {cfi), which runs the whole length of the ventral face, but soon disappears as it is absorbed by the young embryo. The formation of the mesoderm can be traced in eggs sec¬ tioned during or shortly before the stage figured in Fig. 41. Figure 38 is a longitudinal section through the posterior portion of the egg (Fig. 41) through the middle of the thickened mass of cells and the depression bp. In this egg the mesoderm has been forming for some time. The ventral-plate cells are col¬ umnar at bp, and their nuclei are elongated in just the opposite direction to their former longest axis, which is still the longest axis of the serosa nuclei at sr. The depth of the ventral-plate cells gradually decreases anteriorly. The stellate yolk cells are scattered at various distances from one another under the ven¬ tral plate. The mesoderm {msd), as is clearly seen from the section, arises partly from proliferation of the ventral-plate cells under bp and passes forward as a single but incomplete layer of cells. Towards the head this layer splits, and each of the two bands thus formed continues forwards under one of the procephalic thickenings. This is seen in Fig. 31 from a sec- No. 2.] BLATTA AND DORYPHORA. 335 tion through the point pci of the egg represented in Fig. 41. ecd is the thickened ectoderm, msd the mesoderm, which is absent in the middle of the ventral face at c. That the mesoderm grows forward from the rounded caudal thickening may be proved by sections through a number of eggs taken from capsules six and one-half to eight days old. The eggs will be found in various stages more or less close to that given in Fig. 41. In some a very short row of mesoderm cells is found just in front of the thickening; in others the row will be longer as it has advanced further to the cephalic end. Only part of the mesoderm is formed at the thickening. As can be concluded from the even arrangement of the two layers anterior to bp in Fig. 38, each ectoderm cell has a mesoderm cell beneath it, showing that the mesoderm is derived from the ectoderm by centripetal division. The impulse to this division, however, seems to originate in the incomplete invagination at bp and to travel towards the head of the embryo. After the mesoderm is formed, the depression bp disappears, and the amnion and serosa begin to develop. They rise as a crescentic fold from the rounded posterior edge of the area of proliferation (Fig. 42 as). The cells of the procephalic lobes become more prominent, and while the caudal fold of the amnion and serosa is growing in length and continuing up the edge of the ventral plate, a fold also arises from the outer edge of each procephalic area and bends inward. This stage in the development of these membranes is seen in Fig. 43. The embryo is now slipper-shaped, the toe of the slipper being the caudal and the heel the cephalic end. The growth of the membranes continues, the toe of the slipper completing itself more rapidly than the heel. Soon the two procephalic folds are connected around the anterior tip of the ventral plate, which is undergoing a change in outline. Figure 44 shows the amnion and serosa almost closed. Over the spot where the stomodaeal invagination will soon appear, there is still a small, slit-shaped opening in the membranes, but this soon closes; not, however, till after the appendages, both cephalic and tho¬ racic, have begun to appear. The structure and formation of the amnion and serosa, as well as their relations to each other, can be made out from the sec¬ tions (Figs. 29, 30, and 32). Figure 29 is a transverse section 336 WHEELER. [VOL. III. through the point c of Fig. 43, a region at which the two folds have not yet arisen. The ectoderm {ecd) is very thick, espe¬ cially in the median ventral line. The mesoderm i^nsd) is incompletely separated from the ectoderm, and is seen only in the median portion of the section. The yolk cells are large and distinct. Their chromatin forms large nucleoli. The yolk under the embryo is becoming much vacuolated. The section (Fig. 32) passes through a point a little in front of as in Fig. 43. This embryo was older than the one represented in Fig. 42 ; hence the yolk cell {v) is much larger and the meso¬ derm more distinctly separated from the ectoderm. The infold¬ ing of the ectoderm to form the amnion and serosa is seen on either side. The amnion {avi) is thicker than the serosa isi). Its nuclei are small, close together, and spherical, while the nuclei of the serosa are large, flat, and scattered. Figure 30 is a section through a point near (Fig. 43), of a slightly older em¬ bryo. The amnion and serosa are completed and in close con¬ tact with each other, so that the yolk cannot pass in between them. The difference between the nuclei of the amnion and serosa is very pronounced. The mesoderm {insd), which is several cells deep, and is now distinctly separated from the ectoderm, has become an independent layer. Its nuclei are more spherical than the ectoderm nuclei, many of which are considerably elongated and flattened. The entoderm is formed very late in Blatta (about the 15th or 16th day), and in cross-sections of embryos of that age it may be seen as a thin layer of cells on either side of the ventral yolk closely applied to the inner face of the splanchnic meso¬ derm (Fig. 54 eiii). Cholodkovsky (10) claims that these bands of entoderm are derived from the splanchnic mesoderm by de¬ lamination, but an examination of a number of sections has convinced me that it is next to impossible to come to any defi¬ nite conclusion as to the mode of origin of the entoderm in Blatta. By the time it has begun to form the cells have become very minute, and different tissues of very different origins have become closely united, so that the proximity of splanchnic meso¬ derm and entoderm is no proof of the derivation of the latter from the former. I have been able to satisfy myself, however, that the entoderm appears in two thin layers, one on each side of the median ventral yolk, and that these two layers converge No. 2.] BLATTA AND DORYPHORA. 'y n y 00 / anteriorly and posteriorly, and become united to the inner ends of the stomodaeal and proctodaeal pockets, in a manner which differs in no respect from what is observed in Doryphora after the entoderm has become definitely established. I can there¬ fore see no reason for adopting Cholodkovsky’s view as proving that the entoderm arises in a manner differing from that de¬ scribed above for Doryphora. As far as Blatta is concerned the question of the formation of the inmost germ layer must still be regarded as an open one. The embryo is hammer-shaped when the amnion and serosa have closed. The cephalic lobes extend around on to the lateral surfaces of the yolk, and each soon becomes divided into two lobes, a larger anterior and a smaller posterior one (Fig. 44). The posterior lobe is the commencement of the antenna {at). In many embryos one, or, more rarely, both antennary lobes are temporarily bilobed (Fig. 44). This may be a slight rever¬ sion, tending to show that the antennary lobes originally gave rise to two pairs of appendages which were perhaps homologous with the two pairs of antennae in the Crustacea. The mandi¬ bles and first and second maxillae are just visible as faint out¬ growths of the ectoderm, while the more pronounced three pairs of thoracic legs have produced a widening of the embryo in the middle of its length (Fig. 44/). The time required for the development described in the pre¬ ceding paragraphs is about four days. The blastoderm begins to contract and thicken on the seventh day from the commence¬ ment of development, and by the end of this day the area of proliferation is formed. Early on the eighth day the amnion and serosa begin to appear. They develop rapidly, so that the embryo is in the slipper stage by the close of the eighth day. The amnion and serosa are almost or quite closed by the end of the ninth day. During the tenth day the mesoderm is seg¬ mented, and the cephalic and thoracic appendages begin to appear. The embryonal envelopes are completely formed by the end of the tenth day, and the embryo has assumed the shape of a hammer or mallet. 338 WHEELER. [VOL. III. General Remarks. a. Germ layer's. Our knowledge of the formation of the germ layers in the Orthoptera is still less satisfactory than in other orders of insects, as in most of the species studied even the formation of the mesoderm has not been clearly determined. Korotneff (25) failed to find the typical groove-shaped gas- trula in Gryllotalpa, and Ayers (i) had no better success with CEcantJms. On the other hand, Bruce (7) has observed the typical process of mesoderm formation in Mantis, and Graber (i 5) in Mantis and Stenobothrus. Thus only representatives of the Orthopteran families Mantidce and Acridiidce are known to form their mesoderm in the typical manner. Nusbaum (33) concludes, on apparently very little evidence, that the mesoderm of Blatta is formed in the same manner as in Mnsca, and he gives a figure of a cross-section with an immense invagination. Cholodkovsky (10), in his account of the entoderm formation in Blatta, says that the gastrula which gives rise to the inner layer is very easily observed. As I spent much time on a full series of stages between the formation of the ventral plate and the appearance of the appendages, without being able to find a trace of the elongate gastrula for which I was searching, I conclude that both Nus¬ baum and Cholodkovsky may have been deceived by the carina on the ventral surface, which, in reflected light, looks like a narrow groove. The method of germ-layer formation in Blatta, at first sight so different from the method observed in DorypJiora, may, how¬ ever, be traced to the typical process. My studies on Dory¬ pJiora make it probable that the entoderm of Blatta originates in the mass of cells found under the area of proliferation, the more superficial cells of which grow forward as the mesoderm in a continuous median sheet bifurcating under the cephalic portion of the ventral plate. We should thus have a spot differ¬ ing in no essential particular from the caudal plate of Dory- phora. If we suppose that in Blatta the tendency to form the median groove has become very weak, and is now confined to No. 2.] BLATTA AND DORYPHORA. 339 the posterior end of the ventral plate, we may regard the de¬ pression bp as the remains of the blastopore. I have remarked that the gastrula in Dorypliora is much deeper posteriorly than anteriorly, and providing a tendency to obliterate the invagina¬ tion should become apparent in an equal degree throughout its whole length, a short posterior depression like that in Blatta would be the result. But the opposite possibility, viz. : the derivation of a gastrula like that of DorypJiora from a circular form like that of Blatta, is likewise worthy of attention. Ac¬ cording to Sedgwick (43), the gastrula of Peripatus elongates with a concomitant closure of the median portion of its orifice. Of the two openings thus formed the anterior becomes the mouth while the posterior becomes the anus of the embryo. Providing a similar stretching of the gastrula has taken place in the ancestors of DorypJiora and HydropJiiliis, it would be easy to see how the cells of the original single Entodermanlage might be separated to form two masses, which now arise beneath the stomodaeal and proctodaeal area, and how the formation of the mesoderm from the edge of what was once the gastrula lips might continue throughout the whole portion of the embryo between the mouth and anus. If this has been the true evolutionary process in the development of the elongate gas¬ trula of insects, it seems probable that Blatta may represent Fig. I. Fig. 2. Figure i. — Diagram of germ-layer formation in DorypJiora. vp. ventral plate; og. oral end of gastrula; ag. anal end of gastrula; g. central portion of gastrula; en. entoderm; e. prolongations of entoderm; aa. plane of cross-section A\ bb. of cross- section B', cc. of cross-section C; ec. ectoderm of ventral plate; ms. mesoderm. Figure 2. — Diagram of germ layer formation in Blatta (somewhat hypothetical). Letters the same as in Figure i. 340 WHEELER. [VoL. nr. the more primitive and Doryphora the more modified method of germ-layer formation. I would note in this connection that the mesoderm of Blatta is formed in a manner strikingly sim¬ ilar to that observed in PeripaUis by Kennel (24). Reference to the diagrams, Figs, i and 2, will make further remarks on the relation of the two modes of germ-layer formation in Blatta and Doryphora unnecessary. Kowalevsky, Hatschek, Patten, Heider, and Biitschli have published observations which have a decided bearing on the method of entoderm formation in Doryphora. Kowalevsky (26) claimed in his epoch-making work that in Hydrophiliis the two longitudinal bands of entoderm are derived by delamination from the splanchnic mesoderm, which he erro¬ neously supposed to originate from the primitive mesodermic layer by an incurling of its lateral edges. As we have seen, Cholodkovsky also claims that the entoderm of Blatta is derived by delamination from the splanchnic mesodern. Hatschek (18) figures in Bombyx chrysorrhoea a large mass of entoderm cells immediately beneath the stomodaeum. This mass may be compared with either the anterior or posterior cell masses of a similar nature in Doryphora. Patten (38) figures a cluster of several huge entoderm cells attached to the stomodaeum of Neophalax (PL XXIV. C. Fig. 36), and in a more recent paper (39) he has figured similar cells in the same position in, the embryo Aciliiis. Two recent papers on Mtisca, one by Kowalevsky (27) and one by Biitschli (8), contain accounts of a method of entoderm formation very similar to that observed by me in Doryphora. Kowalevsky finds that the entoderm originates in two widely different points, anteriorly at the inner end of the stomodaeum and posteriorly at the inner end of the proctodaeum, from some of the cells of the gastrular invagination. The mass of entoderm at either of these places forms a watch-glass-shaped body with its concavity applied to the yolk. From the lateral edges of each mass the entoderm cells proliferate to form two bands, each of which unites with the one on the same side growing from the opposite direction. By a dorsal and ventral growth of the edges of the two bands the entoderm envelops the yolk and thus completes the mesenteron. Kowalevsky also regards the entoderminic invagination in insects as a greatly elongated gas- No. 2.J BLATTA AND DORVPHORA. 34r trula, which has retained its ability to form entoderm only at the oral and anal ends. Putting this construction on the gastrula, it is, of course, easy to reduce the germ layers of insects to the Sagitta pattern. Kowalevsky’s results on Mtisca have been corroborated as far as the posterior Entodernianlage is concerned by Biitschli (8) ; while his main results on Hydrophihis have been con¬ firmed by Heider (19). According to Heider, the tube formed by the closing gastrula flattens out, and the half of it immediately below the ectoderm becomes mesoderm, afterwards splitting into the somatic and splanchnic layers; while the other (inner) half becomes the entoderm which spreads apart to form two bands, one on each side applied to the part of the mesoderm inclosing the coelomic cavity. Subsequently the proliferating edges of the bands unite ventrally and dorsally to complete the mesenteron. A comparison of my account of DorypJiora with Heider’s account of Hydrophihis will show that I differ from Heider on one point only. I claim that all the entoderm between the oral and caudal widenings of the blastopore is not derived from the inner cells of the gastrular depression, but grows in from the ends of the body. I do not deny that the entoderm may arise in Hydrophihis (and possibly in Doryphord) in the manner described by Heider, i.e. from the inner layer of cells of the gastrular tube when it flattens out and breaks down, but I would regard the process as confined to two very small areas, one being stomodaeal, the other protodaeal. Heider’s figures are undoubtedly correct and correspond in every way to sections through Doryphora embryos in corre¬ sponding stages. Unfortunately, he has not noted with any precision the plane of section of the different preparations figured. Starting with his Plate II., his figures may be inter¬ preted in harmony with Doryphora, thus : — Figure 23 is exactly like my Fig. 88, omitting the degenerating nuclei, and I should interpret it in the same way as Heider. My section passes through the caudal plate of the embryo; Heider’s passes “durch den Abdominaltheil,” which is indefi¬ nite. In Figs. 24 and 25 I can see no entoderm, but merely the apposed splanchnic and somatic layers of mesoderm between the lateral ends of which the coelomic cavities are about to 342 WHEELER. [VOL. III. form, these cavities being simply enlargements of the at first very limited space between the two layers at their outer extremi ¬ ties. Figure 25 is a section through the “ Abdominaltheil.” Both figures accurately represent cross-sections through the middle of the ventral half of the egg of Doryphora before the proliferating bands of entoderm have reached the plane of section (compare my Fig. 78). Heider’s Fig. 26 just skims the ends of the proliferating bands, showing four entoderm cells at i beneath the left coelomic cavity in the figure. No portion of the band was cut on the opposite side, as either the section was slightly oblique or one of the bands had grown somewhat more rapidly than the other, a condition which I have often observed in DorypJiora. The most definite proof that Hydrophihis does not differ from DorypJiora is to be gained from Heider’s own words. He says : “Wir miissen aber nun auf ein hbchst merkwiirdiges Verhalten eingehen, welches uns beweist, bis zu welchem Grade coenoge- netische Veranderungen den urspriinglichen Typus der Insecten- entwicklung entstellen. Wahrend namlich die geschilderte Abtrennung des Entoderms von dem unteren Blatte Kowalev- sky’s im vorderen Theil des Hydrophilus-Embryos (den Kopf- und Thoraxsegmenten) deutlich zu beobachten ist und ebenso klar in den letzten Abdominalsegmenten zur Ausbildung kommt, treffen wir entsprechend den vorderen Segmenten des Abdomens eine Ouerzone des Embryos, in welcher keine Ento- dermchicht zur Anlage kommt — mit anderen Worten: die Entodermanlage entwickelt sich im Vordertheil und nahe dem Hinterende des Embryos in zwei gesonderten Stiicken, welch e erst in spateren Stadien gegeneinander wachsen und mit einan- der verschmelzen. Diese gesonderte Ausbildung des Entoderms vom Vorder und Hinterende des Embryos ist ein Seitenstiick zu dem von uns geschilderten und (Taf. L. Fig. 4) abgebildeten, selbstandigen Auftreten des Vorder-und Hinterendes der rin- nenformigen Einstiilpung. Wie ich aus'den Angaben Kowalev- sky’s and Grassi’s ersehe, weist das Darmdriisenblatt der Biene hinsichtlich seiner ersten Anlage ahnliche Verhaltnisse auf.” b. Embryonic Envelopes. No phenomenon in the development of the insect embryo is better suited to call forth conjecture than the embryonic enve- No. 2.] BLATTA AND DORYPHORA. 343 lopes and the dorsal organ formed soon after their rupture. For the sake of clearness I shall here consider only the envel¬ opes and relegate the discussion of the dorsal organ to my remarks on the revolution of Blatta and Doryphora at the end of the next descriptive division of my subject. Various theories, all more or less vague and intangible, have been advanced by different investigators to account for the amnion and serosa. Balfour (2) regarded these membranes as possibly derived from an early ecdysis. Ayers (i) refuted Balfour’s suggestion; but as he started out in his own explana¬ tion with incorrect suppositions regarding the homologies of the different germ-layers in insects with those of other animals, he could not fail to involve the amnion and serosa in the gen¬ eral error. Kennel (24) regards the embryonic membranes of insects as homologous with the so-called “amnion” in Peripatiis^ and both structures as the remains of the trochosphere of the annelid ancestor. Emery (ii) suggests that the envelopes may be homologous with the shell of the Entomostraca. The question as to the meaning of the envelopes in insects has been greatly confused by drawing in the widely different envelopes of Peripatus, ScorjDions and Myriopods and the Crus¬ tacean dorsal organ, presenting all the different forms observed in Oniscus, Asellus, Cyinothoa^ Mysis, etc. Some authors agree with Kennel in regarding the embryonic envelopes throughout the Arthropoda as homologues. According to others the dorsal organs of the Crustacea are the homologues of the amnion and serosa of hexapods. Still others maintain that the Crustacean dorsal organ is to be brought into connection with the occasion¬ ally similar dorsal organ of insects. Will (52) has of late advanced a theory to account for the formation of the embryonic envelopes of insects only. His theory has the advantage over its precursors in that it replaces, such indefinite terms as “early ecdysis,” “shell of the Entomo-. straca,” and “Trochosphere” by facts derived from the com¬ parative morphology of the membranes themselves. As I came to essentially the same conclusions as Will long before reading his article, I may be pardoned for presenting the subject in my own words, though they repeat in great measure what has appeared in Will’s paper. The problem as to the meaning of the amnion and serosa is restricted to the Hexapoda by postulating the following : — 344 WHEELER. [VOL. III. 1. There are no sufficient reasons for homologizing the embryonic envelopes of insects with the homonymous but dissimilar structures in Myriopods, Scorpions, and Peripatus. 2 . There is no more than a superficial resemblance to speak for an homology between the dorsal organs of the Crustacea and the embryonic envelopes of Insects or between the dorsal organs of Crustacea and the homonymous structures in Insects. 3. The dorsal organ of insects may be regarded as the neces¬ sary result of the rupture and absorption of the embryonic envelopes, and consequently as in no way related to such structures as the dorsal organs of Cymotlwa, Liniuhts, etc. The process of envelope formation has been observed in numerous insects of all orders with sufficient accuracy to war¬ rant the assertion that all the pronounced types are connected by intermediate forms in fine gradation ; a fact which was long ago expressed by Kowalevsky (26). Fig. 6 . Fig. 7. Fig. 8 . Diagrammatic longitudinal sections of embryos just after the completion of the envelopes. Figure g, Geophilus; Figure 4, Calopteryx; Figure g, Aphis; Figure 6, Doryphora; Figure 7, Blatta; Figure 8, Bombyx. a. amnion; s. serosa; as. amnion and serosa apposed; vp. ventral plate; pd. procephalic lobes; y. yolk. If we suppose such a form as Calopteryx to present the origi¬ nal mode of embryo formation in the Hexapoda (and Paleontol- No. 2 .] BLATTA AND DORYPHORA. 345 ogy makes such a supposition probable), we have a form which will unite readily with an embryo Myriopod in a corresponding stage of development. In Fig. 3 I have given a diagrammatic longitudinal section of Geophilus about the time the appendages appear. Evidently either the excessive growth in length of the annelid-like body has necessitated the complete invagination of the embryo into the yolk, bringing the caudal and cephalic ends together, or this process has been adopted as a means of bring¬ ing the surface of the embryo into more complete contact with the yolk. It is only necessary to reduce the posterior half of the infolded embryo (included between x and m in the figure) to a thin membrane to reach the condition of Calopteryx (Fig. 4). The membrane resulting from the attenuation would be the amnion. Will has emphasized the fact that the stage with the amnion almost as thick as the ventral plate with which it is continuous, still occurs in the ontogeny of insects (compare Blatta and NeopJialax). He suggests that the attenuation of the posterior half of an embryo like Geophilus to form the amnion may account for the great disparity in the number of segments between the Hexapoda and Myriopoda. A further difference is observable between Calopteryx and Geophilus. The end of the original caudal extremity in the former is joined to the anterior end of the ventral plate, thus closing the sack whose anterior wall is the ventral plate and whose posterior wall is the amnion. This sack is attached at one point to the serosa enveloping the egg. The union of the membranes to close completely the amniotic cavity is the hinge about which the further explanation turns. It seems essential in all the insects so far studied excepting Musca, where the envelopes are rudimentary, that the amniotic cavity should be shut off from the space between the vitelline membrane and the surface of the yolk. The reason for the closure is apparent if we regard the amniotic cavity as a place for the temporary deposition of excreted matters, as an organ functionally analogous to the allantois of higher animals. It has often been observed that the amniotic cavity of insects soon after its formation becomes filled with a clear liquid which during and after revolution is found as a much vacuolated coagu- lum about the feet of hardened embryos. It seems probable that while the inner ends of the ventral-plate cells are absorbing and 346 WHEELER. [VOL. III. metabolizing the yolk, their outer ends are at the same time giving off into the amniotic cavity a less amount of liquid waste products. Providing this supposition is true, we should have a sufficient reason for the constant closure of the amniotic cavity. We have a complete series of finely graduated forms of enve¬ lope formation from the method observed in Calopteryx (Fig. 4) to Blatta (Fig. 7). Aphis represents the first step in the tran¬ sition of an e 7 itoblastic embryo like Calopteryx to the decidedly ectoblastic form seen in Blatta. This transition consists in leaving more and more of the anterior end of the embryo on the surface of the yolk. In Aphis (Fig. 5) the whole head is left outside the invagination ; in Doryphora, the head and the anterior half of the body. When a portion of the embryo is left on the surface, the closure of the amniotic cavity necessitates a backward growth of the angle formed by the fore end of the head and the abutting serosa {aSy Fig. 5) to form a fold which unites with a similar fold formed at the opposite end of the embryo. In Doryphoray where much of the embryo lies on the surface of the yolk, the posterior or caudal fold of the amnion and serosa has to grow forward a considerable distance to meet the cephalic fold. Blatta has advanced still further than Dory- phora. The embryo no longer grows into the yolk, but the formation and ultimate closing of the membranes continues. At first sight it would seem more natural to suppose that the result attained in Blatta was brought about simply by an extru¬ sion of the yolk between the amnion and serosa of such a form as Aphis or Calopteryxy but the law of orientation, as explained in a preceding paragraph, forbids such an interpretation. The head of the Aphis embryo is at the time of the completion of the membranes close to the spot before occupied by the caudal end of the ventral plate, and after revolution the caudal end of the embryo will again be located at this end of the egg. Hence the typical ectoblastic originated from the typical entoblastic embryo, not by an extrusion of the yolk from between the amnion and serosa, but by a gradual weakening of the invagi- native process. The weakening, of course, results in more and more of the anterior portion of the ventral plate remaining inert, though the growth of the membranes to shut off the amniotic cavity continues. The peculiar free entoblastic embryo observed in Lepidoptera INTO. 2.] BLATTA AJVD DORYPHORA. 347 (Fig. 8) may have originated in two ways : either from a sus¬ pended entoblastic embryo like Calopteryx by a separation of the amnion from the serosa at the point of suspension (Fig. /\as) and the consequent passage of yolk between the two membranes, or from the ectoblastic type of Blatta by a separation of the amnion from the serosa throughout their area of contact, accom¬ panied by an intrusion of yolk. All that we at present know concerning the formation of the envelopes in Lepidoptera tends to prove that the latter method is the more probable. The hypothesis of Will and myself as set forth in the above paragraphs might be called a mechanical explanation as opposed to the views of those who see in the embryonic membranes rudimental structures like the remains of the trocosphere, larval skins, etc. Our hypothesis has at least the virtue of utilizing the facts near at hand. Description of the External Changes in the Embryo Blatta and Doryphora up to the Time of Hatching. Blatta. Soon after their completion, the amnion and serosa become more attenuated on account of the flattening of their cells and the consequent diastasis of their nuclei. This thinning out of the envelopes permits a better view of the embryo and its form¬ ing appendages. On about the tenth or eleventh day from the beginning of development, the embryo presents the appearance represented in Fig. 45. The broadly rounded procephalic lobes are separated by a deep incision in the median line, and the antennae {at) growing from the posterior lateral corners of the lobes have become prominent, while the backward direction of their growth is apparent. The labrum {lb) has appeared as a thick, crescen¬ tic and slightly divided fold in front of a faint depression which is the commencement of the stomodaeal invagination. Of the three pairs of oral appendages, the second and third {mx^, mx^) are clearly rounded and directed backwards ; the mandibles are still small. Each segment of the abdomen presents a pair of indistinct appendages. These subsequently disappear, with the exception of the pairs on the basal and terminal segments, which undergo a differentiation peculiar to themselves. As I 348 wheeler. [VOL. III. have given elsewhere (50) a minute account of the pair of appen¬ dages belonging to the first abdominal segment, I shall not con¬ sider them in the present paper. The pair of appendages belong¬ ing to the terminal segment persist and become the anal stylets. Unlike most Arthropod embryos, the caudal end of the embryo Blatta {cpl) is never bent dorsally, but from the very early stage in which it is in a line with the long axis of the abdomen, shows only a ventral flexure. Owing to this flexure, which soon be¬ comes very pronounced, the formation of the protodoeum cannot be as easily observed as in Doryphora. The changes which are apparent in surface views by the four¬ teenth or fifteenth day have been carefully represented in Figs. 46 and 47. In the former the embryo is in situ on the yolk, in the latter it is isolated and seen from the ventral surface. As may be indistinctly seen in Fig. 47, the first and second maxillary appendages have each become split up into three divis¬ ions. In QLcantJiuSy according to Ayers, “ the three oral appen¬ dages are trilobed ; the lobation is most prominent in the second maxillary, and least in the mandibular appendages. The primi¬ tive appendage is first divided into two lobes, and the inner of these becomes secondarily divided into two.” There are appar¬ ently no traces of lobation in the mandibles of Blatta. The outer of the three lobes of each maxilla becomes the palp, while the inner two become the galea and lacinia of the adult. Fig. g. Fig. 10. Fig. ii. Fig. 12. Fig. 13. Figure g. — Embryo of Blatta, 15 days old; revolution about to begin. The stages in revolution are represented, after the rupture of the amnion and serosa, in Figures 10-13, which are from embryos 16, 16^, i6|, and 17 days old respectively. as. amnion and serosa; s. edge of serosa; b. dorsad growing body wall; d. 0. dorsal organ; x. clear zone covered with scattered amniotic nuclei. No. 2.]' BLATTA AND DORYPHORA. 349 In Blatta the formation of the nervous system in its earlier stages cannot be clearly seen from the exterior. The same holds true of the small tracheal invaginations, though several pairs, especially those of the thorax and basal abdominal rings, may be seen on the pleurae in good preparations before revolu¬ tion. Still they are so much less distinct than in Doryphora that I have given them little attention. The peculiar phenomena of revolution are hurried through by the embryo from the beginning of the sixteenth to the end of the seventeenth day. Several successive stages in the process are represented in the woodcuts (Figs. 9 to 13). When fifteen days old (Fig. 9) the embryo still occupies the middle of the ventral surface of the egg, the distance from the head to the cephalic end of the yolk being almost equal to the distance of the tail from the caudal end of the yolk. The amnion and serosa {as) still envelop the embryo, though they have become much attenuated. By the end of the fifteenth or the beginning of the sixteenth day, the envelopes rupture, an irregular slit being formed down the median ventral line. The amnion now appears to undergo degeneration, at least in part, while the serosa is drawn back from both sides by a contraction of the protoplasm of its cells, the large nuclei of which make it easy to trace all the steps in the formation of the dorsal organ. Soon after the rupture of the envelopes the embryo and egg, when seen from the side, resemble Fig. 10. The embryo stands out free from its envelopes on the yolk; the edges of its dorsad growing walls {b) are distinctly marked. Near these, on the sur¬ face of the egg, are seen a number of scattered nuclei, which are of the same size as the nuclei of the cells forming the embryo. These I take to belong to the portion of the amnion which has become folded back on the yolk and forms a zone w extending the whole length of the yolk in contact with the dorsad growing body wall. Next to this zone lies another zone, which is bounded by the distinct edge of the serosa {s), and which I regard as a portion of the yolk left bare, as no nuclei are to be found on its surface. Besides the ventrodorsal contraction in the substance of its anterior edge {s), the serosa contracts in an antero-poste- rior direction, thus producing the constriction seen at j/ in Fig. 10. The rounded and projecting lump formed at the caudal pole is the beginning of the dorsal organ (Fig. 10 d. 0). The 350 WHEELER. [VoL. nr. contraction continues towards the median dorsal line and towards the cephalic pole. In Fig. ii the dorsal organ has moved half¬ way up the dorsal surface. Its darker color in stained embryos is due to the fact that the serosal cells have become deeply columnar with a consequent approximation of their large nuclei to one another. The embryo, besides increasing in size, has undergone a change in position. Its tail now lies at the caudal end of the egg. Notwithstanding the embryo’s growth in length, its head lies much lower than in the preceding stage (Fig. lo). The body wall {b) is still distinct, the zone of sporadic amni- otic (i’) nuclei {x) has increased in breadth. As in the preceding stages, these nuclei are most closely aggregated near the edge of the advancing body wall. In the next stage (Fig. 12) the dorsal organ has reached the cephalic end of the yolk and bulges out like a large hood. The body walls of the embryo have nearly enveloped the yolk at the caudal end. The next change takes place very rapidly. The stage repre¬ sented in Fig. 12 is attained towards the end of the sixteenth day. By the seventeenth day the walls have closed in the median dorsal line, and the embryo has grown in length to such an extent as to bring its head to the cephalic pole. The dorsal organ has been shut in by, and lies immediately below, that por¬ tion of the body wall, which will form the tergum of the pro¬ thorax. On entering the yolk the cells of the dorsal organ begin to disintegrate. Two of the stages in the formation and disso¬ lution of the dorsal organ are represented in Figs. 50 and 51, both from longitudinal sections, the former being sagittal, the latter frontal. Figure 50 represents a section through the centre of the thick¬ ened mass of serosal cells. The deeply stained nuclei are seen crowded together in the inner ends of the cells, the contours of which are rendered papillose apparently by the pressure of the nuclei against the cell walls. The outer ends of the elongated columnar cells form a thick layer of granular protoplasm con¬ siderably depressed at 0. This depression is equivalent to the tubular cavity in the dorsal organ of Hydrophilus. In Fig. 51 the large lump of cells has become engulfed in the yolk. The body wall has closed over it, and the heart {cc) has formed between it and the ectoderm. The large deeply staining nuclei {7171) are seen in the various stages of active degeneration, No. 2.] BLATTA AND DORYPHORA. 351 many of which recall the degenerating nuclei in the entoderm of DorypJiora. We have the same vacuolization of the karyo- plasm and agglomeration of the ehromatin. In the centre of the mass a pale, oval spot surrounds an elongated cavity {p), which is almost obliterated. This cavity results from the de¬ pression 0 of Fig. 50 by a closing in over it of the peripheral edges of the dorsal organ. Figure 48, drawn from an advanced embryo saturated with clove oil, shows the condition of the different organs shortly before hatching. The embryo preserves the shape of the egg, being much flattened laterally. The segments of the body are all distinctly defined. The mouth parts have become closely approximated, and have assumed their definite relations to one another. The long antennae {at) extend as far as the two anal stylets {ast) in which the ventrally bent tip of the abdomen terminates. The different divisions of the alimentary canal, oesophagus {pe), ingluvies (c), proventriculus {gz), stomach {st)^ still containing the remains of the yolk with its degenerating nuclei, and rectum {ret), ending in the anus posteriorly, and surrounded by a wreath of Malpighian vessels {nipg) anteriorly, may be readily traced in the figure. The heart is seen as a delicate tube just beneath the dorsal integument. The large supraoesophageal ganglion {cgl) connected with the large lateral compound eyes, in which the pigment is being deposited, fills the greater portion of the brain-box. One of the commissures is seen connecting it with the inf race sophageal ganglion (^/^). The three thoracic ganglia (^/^, gl‘^, gl^) are much larger than the six abdominal ganglia. A large, granular, fat body {ad) is applied to the inner surface of pleural wall of the abdomen. The refractive granules imbedded in it form a chevron in each of the first five or six somites. Patten (38) in his preliminary note on Blatta thus describes the physical and chemical nature of these bodies: “ In the embryos of Blatta, as well as in those of most if not all other insects, there appears in each of the segments at a certain time a great number of clear, highly refractive particles that at first might be taken for oil globules, and which have always been regarded as such. On more care¬ ful examination, however, it will readily be seen that this suppo¬ sition is incorrect. A number of tests have been made in order to ascertain the nature of these bodies, and the results show 352 WHEELER. [VOL. III. that there are some salts of uric acid. That they are not of a fatty nature is indicated by the fact • that treatment of the embryos with hot benzole, chloroform, or clove oil has not the slightest effect upon the bodies in question. Further examina¬ tion with a high magnifying power shows that they consist of small spheres of an extremely refractive substance, from the centre of which dark lines radiate in an irregular manner, pro¬ ducing the same appearance seen in the crystals of urea from the Malpighian vessels. It was this similarity which first sug¬ gested the true nature of these bodies; and further tests con¬ firmed this view, for, after heating an embryo with nitric acid upon a glass slide, and then adding a little ammonia, the char¬ acteristic red color of Murexid was formed. A still further test was formed by dissolving the granules in dilute caustic potash, and then precipitating the urea by adding acetic acid, although this method did not give such definite results as the first.” The embryonic development of Blatta is completed by about the thirtieth day from oviposition. Figure 49 shows the embryo soon after hatching. Shortly after leaving its narrow place in the capsule, the insect under¬ goes a peculiar change in shape. While confined by the cho¬ rion the diameter from one pleural wall to the other is about one-third the dorsoventral diameter of the insect. Soon after hatching, its dorsoventral diameter is only about one-third as great as its greatest breadth. The tip of the abdomen, ventrally flexed in the egg, bends dorsally as indicated by the position of the anal stylets, which now point directly upwards and out¬ wards. The spines and onychia, most abundant on the legs, are developed shortly before hatching. Doryphora. Doryphora embryos, when carefully prepared, reveal much more in surface views than Blatta embryos prepared according to the same methods. The last stage described is represented by Fig. 73. I shall pass over a few of the succeeding stages, and stop to describe the embryo represented in Fig. 72, which shows, with great clearness, all that has taken place in the omitted stages, and makes a description of them unnecessary. The figure is slightly No. 2 .] BLATTA AND DORVPHORA. 353 diagrammatic, being drawn from a number of different embryos, each of which contributed some of the details in a clearer and more pronounced manner. The mouth {o) and anus {a), both triangular depressions, have become clearly established. The former has in front of it a heart-shaped prominence, the bilobed labrum. The lateral half of the head presents some interesting facts, first elucidated in Patten’s paper on the eyes of Acilms (39). Each half of the head is divided by longitudinal constric¬ tions into three parallel rounded ridges, each of which is further divided by two transverse depressions into three subquadran- gular thickenings. The three inner b^) on each side, rep¬ resenting the three segments of the brain, are directly continuous with the ventral ganglion chain, extending to the protodseum. The row of prominences {og^y og‘^y og^) on the outer side of the three brain segments are the optic ganglion, the further three {op^y op^^y op^)y somewhut indistinctly seen because situated on the very edge of the head, are the divisions of the optic plate, each of which, in AciliuSy according to Patten, bears a pair of ocelli. The only appendages of the head proper in this stage are the antennae {at)y which are directed backwards, and the heart-shaped labrum. The head also presents three pairs of small invaginations somewhat less distinct than in the figure {t^y t'^y t^). These lie near the longitudinal constriction, sepa¬ rating the brain thickenings from the thickenings of the optic ganglion. This is best seen in the third segment, where the invaginations lie at the bases of the antennae. Following the third segment of the brain is distinctly seen in some embryos a short segment inserted between the antennary and mandibular segments. Its short ganglionic swellings {gl'^) are far apart, and connected by a broad commissure. This somite may also have a pair of small invaginations, but I have been unable to find them. Hereupon follow the mandibular, and the first and second maxillary segments, each with a pair of invaginations. Those of the second maxillary segment are concealed behind the bases of the elongated appendages, but are readily seen in sections. The three broad thoracic segments are provided with the three pairs of legs, all of which are of the same length. The division into femur, tibia, and tarsus is indistinctly marked. In none of the preceding stages have I observed what is so prom- 354 WHEELER. [VOL. III. inent in Blatta, viz. the appearance of appendages on the abdominal somites. There are not the slightest traces of even the pair of appendages of the first abdominal somite, which in Blatta develop into the large glandular organ of which I have treated elsewhere (50). The tracheal invaginations are situated at the bases of the legs. Those of the first thoracic segment are small, and soon close over and disappear. The second pair, which are almond-shaped, are the largest in the whole embryo, and so remain. They are situated near the constriction divid¬ ing the first from the second thoracic segment, and in later stages often have the appearance of belonging to the first seg¬ ment. The metathoracic invaginations are somewhat smaller, and are also placed near the edges of the somite to which they belong. In the succeeding abdominal segments there is a tracheal invagination in the middle of each lateral half. These invaginations become successively smaller till they can be de¬ tected only with great difficulty on the loth and nth somites {t 19, t 20). The nth somite is followed by the broad subhex- agonal caudal plate with its large protodaeal invagination. The corners of the plate are formed by rounded lobes containing apparently spherical bodies, the ends of the three pairs of Mal¬ pighian vessels. These grow off from the protodaeum at an unusually early period in Doryphora, and turn back till their rounded blind ends terminate just beneath the surface ecto¬ derm. The paired ganglionic thickenings are seen in the embryo figured to be slightly kidney-shaped with their hili directed laterally. The Mittelstrang is apparent in the small and nodular intersegmental thickenings {insf), which appear from the surface as small masses of cells of a somewhat dif¬ ferent nature from those of other portions of the median line. The surface changes which the embryo undergoes in the stages immediately following that represented in Fig. 72 may be briefly summarized. The embryo just described lies like a band on the ventral yolk, the caudal portion still extending round the hind pole of the egg, and up the dorsal surface a short distance. The isochronous changes which ensue are, (i) a shortening of the embryo, bringing the tail to the pole of the j (2) a broadening of the embryo, the sides of which now bend dorsally and clasp the yolk; (3) a greater concentration of the cephalic, mandibular, and maxillary somites to form the No. 2.] BLATTA AND DORYPHORA. 35 S head of the larva, and (4) an increase in the length of all the appendages except the antennae. The shortening of the embryo stops as soon as its tail reaches the caudal pole of the egg, the lateral growth of the body continues, and we reach the stage figured (Fig. 74). The yolk is not yet covered by the dorsad growing walls of the embryo. The head is distinctly marked off from the thorax, the wide-spread mouth parts of Fig. 72 have converged, and are assuming the relations which they bear to one another in the larva. The ventral nerve chord has devel¬ oped considerably,, and it is now possible to recognize near the centre of each ganglion the mass of ‘‘Punktsubstanz"’ definitely marked off from the cellular portion and united with the corre¬ sponding mass of its fellow-ganglion by two-cross commissures. Figure 75 represents the larva ready to hatch. The dorsal body wall has closed, the six ocelli have become pigmented, the cuticle has developed spines, the meso- and metathoracic and first abdominal segments have each developed a short, sharp, black spine in a line with the abdominal spiracles. These spines are used by the larva in rupturing the chorion, the vitelline membrane, and the various cuticles which it has shed before reaching this stage. The movements of the hatch¬ ing insect at first produce a rent in the chorion extending from the first to the third spine, by further struggling the two rents from opposite sides are made to meet over the head, and the insect emerges from between the two lips thus formed. The embryonic development requires about six days. Before passing on to a description of the internal changes of Doryphora and Blatta, it is necessary to consider the fate of the embryonic membranes of the former insect. This is very differ¬ ent from what was observed in Blatta. The serosa, instead of rupturing when the amnion ruptures, separates from it and also from the entire surface of the yolk, and forms a third egg enve¬ lope, beneath the vitelline membrane, to which it applies itself. It remains clearly recognizable by its large and deeply staining nuclei till the insect is almost ready to hatch (Fig. 86 sr)^ when it disappears, probably by absorption. The fate of the amnion is peculiar. On rupturing, its two ventrally bent folds turn back and become in part applied to the yolk. A few of its cells are loosened from the bulk of the membrane, and are often seen sticking to the serosa at different points. They are probably 356 WHEELER. [VoL. irr. endowed with amoeboid tendencies, for when the ectodermic wall is about to be completed in the median dorsal line these cells are seen to have accumulated at the place of closure. The amniotic cells, which have become applied to the surface of the yolk by the bending back of the two folds resulting from rup¬ ture, have closed in the yolk while the serosa is separating from it. The advancing body walls of the embryo, however, soon make the amniotic covering unnecessary, and it contracts in the median dorsal line to form what may be called an amniotic dorsal organ, to distinguish it from the serosal dorsal organ of Blatta. Figure 90 is a part of a section through the median dorsal por¬ tion of an embryo in the stage represented in Fig. 74. At do the protoplasm of the amnion has thickened, and the nuclei are seen passing in between the yolk bodies. At m are a number of nuclei undergoing degeneration. These resemble the degener¬ ating entoderm nuclei to which I have called attention in a much younger stage. The amnion cells, which have become applied to the yolk when the membrane ruptures, enter the yolk after the formation of the dorsal organ by the very narrow slit left in the closing ectoderm in the median dorsal line. This is seen somewhat indistinctly in Fig. 85. Nuclei are observed at b passing in between the two cardioblasts {cby cb), which are about to meet and form the heart. The splanchnic mesoderm islni), with the underlying entoderm, still leaves a wide gap through which the migration into the yolk takes place. The granular matter surrounding the nuclei is probably the remains of the cytoplasm of the amnion cells. In the figure a number of entire Fig. 14. Fig. 16. Big IS- Figures 14-16. — Three diagrammatic median cross-sections through the egg of DoryphorUy before and during revolution, ch. chorion; v. vitelline membrane, ap¬ plied tp the inner face of the chorion; a. amnion; s. serosa; em. embryo; d.o. dor¬ sal organ (amniotic) ; y. yolk. No. 3.] BLATTA AND DORYPHORA. ssr amnion cells ici) are still seen just beneath the serosa, and one is seen right at the narrow space between the cardioblasts. The last steps in the process are represented in Fig. 93. The dorsal ectoderm has become continuous at ect in the median line. The two cardioblasts are still in the same stage. One of the last amnion nuclei is passing in surrounded by a mass of granules.. A clear idea of the revolution of the embryo Doryphora may be obtained from the three stages in the diagrams. The method of revolution just described is very similar to that observed by Graber (15) in Lma. Though he did not give a description of the complete process, he made the important observation that the serosa remains nnchanged till after the chiti- nous cuticle is formed. General Remarks. Dorsal Organ. The term “dorsal organ” has been applied to the peculiar thick lump of cells resulting from the concentration on the dor¬ sal yolk of the remains of either the amnion or serosa, or of both, preparatory to their sinking into the yolk and being absorbed. A similarity in form and position has led many investigators to look for an homology between the dorsal organ of insects and the homonymous organ of the Crustacea. My observations on the dorsal organ of Cymothoa have con¬ vinced me that there are fundamental differences between the Crustacean and Hexapod dorsal organ. First, the dorsal organ of this form, and probably other Isopoda, is a structure which persists from an early stage almost to hatching, and may persist throughout life in some Branchiopoda, whereas the dorsal organ of insects is a very transitory structure. Secondly, the dorsal organ of Cymothoa seems to be a secretory organ, as was deemed probable by Balfour (2). I have observed that the elongated cells which form the organ secrete a reniform sack of chitin, which is joined by means of a corrugated chitinous tube to the cuticle shed from the surface of the embryo at a very early stage. Nusbaum (34) has observed that the cavities of the dorsal organs of My sis are filled with a clear substance, probably a secretion. The presence of the so-called dorsal organ in insects is prob- 35^ WHEELER. [VOL. III. ably due to the fact that the embryonic envelopes are to be absorbed. As these membranes consist of assimilable matter, it is obviously an advantage to the embryo to be able to add them to the stock of food represented by the yolk. The simplest conceivable method of effecting the resolution of the envelopes into food material, considering their position when fully devel¬ oped, would, of course, be to engulf them in the yolk, where, under the influence of the yolk cells, metabolism is being actively carried on. There are two methods of inclosing the membranes in the yolk. According to one, they might undergo dissolution in siUi; according to the other, they might be brought together in a mass and swallowed up by the yolk some¬ where in the median dorsal line. Obviously the latter method is the more advantageous, as the body walls, continually grow¬ ing towards the median dorsal line, might be impeded in their advance if the membranes were absorbed at all points on the surface of the yolk. Probably the inconvenience which would thus result from a diffuse absorption accounts for the fact that it does not occur, though a modification of the method occurs in Doryphora, where the serosa is absorbed very late in develop¬ ment after the larva has secreted its second cuticle and is almost ready to leave the egg. Given a thickening, somewhat flattened mass of cells, des¬ tined to be swallowed up in the yolk, and it is most natural to suppose that the organ, in passing into the yolk, would become cup-shaped as in Blatta, or form a thick-walled tube, if the organ extended the full length of the dorsum, as in Hydrophilns. In either case the outer ends would be made to converge, by the lateral pressure of the yolk and the sinking of the median por¬ tion of the organ, and we should get a closed tube or sack. This would not, of course, hold true of an organ formed like the amniotic dorsal organ of Doryphora^ for the reason that in this case the decomposition begins as soon as the organ is formed, and not after it has passed into the yolk, as is the case with the serosal dorsal organs of other forms {Hydrophilns, BlattOy Neophalax). Hence the cavity of the Hexapod dorsal organ would resemble the cavity or so-called micropyle of the Isopod dorsal organ, though the two cavities would not be homologous. No. 2.] BLATTA AND DORYPHORA. 359 The Fate of the Different Germ Layers. A. Entoderm. Recent writers on insect embryology have recognized two forms of entoderm, — one called primary, and represented by the yolk cells; and one secondary, represented by the epithe¬ lial wall of the mesenteron. In Blatta the yolk nuclei steadily increase in size from their first appearance at a time when they are no larger than other nuclei. At the time of the formation of the dorsal organ they are by far the largest nuclei in the egg (Fig. v). Their chromatin is distributed through the karyo- plasm in the form of a fine, convoluted thread, and as two or more nucleoli (Fig. 54 v). Soon after the closing-in of the yolk, they lose their rounded outline, and become irregular and more homogeneous (Fig. 55 v). In the last stages of their dissolution they may be seen as stellate spots in the remains of the yolk aggregated in the stomach of the advanced embryo (Fig. 48). Yolk segmentation, though occurring in Blatta, takes place after the appendages are formed, at a much later period than in DorypJiora. The segments, usually very obscurely de¬ fined, become confluent again as development continues. In DorypJiora the yoke cells undergo no increase in size from the time of their first appearance; but soon after the yolk has become segmented their cytoplasm is reduced to a scarcely perceptible layer surrounding the nucleus, which has become irregularly polygonal (Fig. 83). As all the eggs I studied were killed and preserved in the same manner, this difference in form between the yolk nuclei in the stages during yolk seg¬ mentation and after this process till the setting-in of degenera¬ tion, must be regarded either as a normal change in the living nuclei, or as indicating that their chemical nature is changed, and their resistance to the altering effects of reagents lessened. After the completion of the mesenteron at a time when the larva is almost ready to hatch, the remains of the yolk nuclei are pushed back into the stomach in the same way as in Blatta. Here they degenerate in a manner very similar to that observed in the secondary entoderm nuclei, and the nuclei of the amniotic dorsal organ. They become swollen and vesicular, and their chromatin is reduced to irregular masses. The yolk becomes a compact, granular mass, staining pink in borax carmine. 36 o WHEELER. [VOL. III. The origin of the secondary entoderm in Doryphora has been treated of at length in preceding paragraphs. We have now to trace its fate, from the condition in which we left it, as two masses, — one under the blind end of the oesophagus, and the other under the blind end of the proctodaeum. In an embryo of the stage of Fig. 72, each of the isolated masses has begun to send out the two bands of entoderm. These diverge from their point of origin, and apply themselves to what is to be the splanchnic mesoderm (Fig. 78 ent). Their divarication is so slight that a thick longitudinal section will sometimes include a whole band, if still short. Thus, in Fig. 92, from a section through an embryo in the stage of Fig. 72 ; passed to one side of the median line, we have the entoderm {ent) still attached to the mesoderm {msd), which is now distinctly marked off from the ectoderm of the stomodaeal invagination {st). The entoderm is attached to the inner end of the oesophagus, and extends along the yolk as a band two or three cells thick. It is clearly distinguishable from the mesoderm by the greater clearness of its cells, and by its paler nuclei. The process at the posterior end of the embryo is similar, as may be concluded from Fig. 76, from the same embryo. Like the anterior thickening, the posterior mass does not remain stationary, but grows out in two bands. Thus it happens that little entoderm is found right under the proctodaeum. The knife has not passed through the proctodaeal invagination in the figure, but has cut to one side of the median line, through the Malpighian vessels mp^ and 7 np^ of Fig. 72. The true relations of these vessels may be understood from an examination of Fig. 77, and the present section. Fig. 76. The strand of entoderm, very similar to the anterior strand described above, is attached to the proctodaeal pocket near the point at which one of the first pair of Mal¬ pighian tubes {mpg^) turns out toward the surface of the embryo, pushing aside the mesoderm, which elsewhere forms a continuous sheet under the ectoderm. In the same figure may be noticed a second mass of entoderm {ent) attached to the proctodaeum near the point at which the third pair of Malpighian tubes branches off from the common proctodaeal pocket. This mass ends with a sharp point between two yolk segments. I have seen similar masses of entoderm in a few other embryos, apparently independent of the two main No. 2.] BLATTA AND DORYPHORA. 361 forward-growing strands, and terminating in the same acute point which seems to force its way between the yolk segments. These points of entoderm do not grow far, as I have concluded from an examination of slightly older stages, but soon fuse with the bases of the two main strands, and form a meniscoid mass in every way comparable with the watch-glass-shaped mass in Miisca, as described by Kowalevsky (27). A median cross-section of an embryo, with the entoderm bands fully established, but not confluent, is shown in Fig. 78. The embryo is cut in two places. The upper half passes through one of the basal abdominal segments, while the lower half passes through the abdomen, a short distance from the tail. In the upper half no entoderm cells are to be found, as the two bands have not yet reached in their forward growth the basal abdominal segments in which they fuse with the two bands growing back from the stomodeeal invagination. In the lower half a heap of succulent entoderm cells is seen on each side, separated from the coelomic cavity (cl) by the splanchnic mesoderm {slm). Four stages in the formation of the mesenteron after the establishment of the entoderm as two long bands, are repre¬ sented in Figs. 83 to 86 e7tt. The entoderm remains throughout embryonic development perfectly distinct from the splanchnic mesoderm, to which it is nevertheless very closely applied. At first the cells are irregularly arranged in the band which is deepest in the middle, but gradually flattens out to a single cell in thickness at its dorsal and ventral edges (Fig. 83). In a later stage, however, the nuclei have their long axes directed at right angles to the long axes of the splanchnic mesoderm cells (Fig. 84), and thus indicate that the cells of the entoderm are begin¬ ning to assume a definite columnar arrangement, though they still lie, in some places, in two or more rows, one above the other. By the time the body walls are about to close, the cells of the entoderm have formed an even layer of columnar ele¬ ments (Fig. 85), an arrangement which is retained in all the subsequent stages till the embryo hatches. The growth of the entoderm, accompanying the adjacent mesoderm and ectoderm in their dorsad movement, is at first largely along the dorsal edges of the bands, as may be seen by comparing Figs. 84 and 85, where the distance between the ventral edges of the two bands is nearly the same, while the 362 WHEELER. [VoL. IIL distance between the dorsal edges is greatly lessened. The transformation of the original ribbon of several superimposed rows of cells into the simple epithelium of columnar cells is not entirely due to cell division. As may be seen from Figs. 84 and 85, either the wandering of the inner rows of cells over the outer towards the dorsal edge of the ribbon, or a stretching of the whole band, so as to permit an intercalation of the cells of the inner rows between those of the outer row, are the more probable factors in the thinning out of the entoderm. The latter method is more probable, though the former method is certainly in keeping with the gliding and mobile movements of the entoderm. The nuclei of the entoderm have their chromatin distributed in the typical filament, which is more attenuated than in either mesoderm or ectoderm nuclei. Shortly before hatching the chromatin of the mesenteron nuclei appears to have dissolved, as they seem to have become perfectly homo¬ geneous, though they still stain deeply. In the hatching larva the cells have become more deeply columnar, on account of a diminution in calibre of the mesenteron. The nuclei cease to absorb more of the staining fluid from the surrounding eyto- plasm, though the walls retain their evenly rounded contour. Such a fundamental change in the nuclei would seem to indicate that some important change is about to take place in the mes¬ enteric layer of cells ; but whether this change is dissolution I am unable to say, as I have not studied the insect in the stages beyond hatching. The process of mesenteron formation is essentially the same in Blatta as that just described for Doryphora. From the first, the entoderm cells of Blatta are as small and indistinct as the yolk cells are large and prominent. They form, as stated above, two bands of very flat cells bearing the same relations to the mesoderm as the corresponding bands of Doryphora (Fig. 54 ent). The edges of the two ribbons continue their growth, and meet ventrally and dorsally, to complete the mesenteron (Fig. 55 Besides the lining of the mesenteron the corpus adiposajn^ represented during the embryonic life of Doryphora by a number of granular cells which constantly increase in size up to the time of hatching, probably originates from the entoderm. I have observed in several cases that before the two posterior bands of No. 2.] BLATTA AATD DORYPHORA. 363 entoderm have reached the middle of the embryo a number of granular and somewhat larger cells are to be found mingled with the cells of the bands. I conclude that these cells are of ento- dermic origin because when first seen they are associated with the entoderm cells and resemble them more closely than they resemble the adjacent mesodermic elements. At first small (Fig. 85 ad)^ these fat cells gradually but constantly increase in size, their cytoplasm and nuclei increasing in about the same ratio. They wander about in the body cavity, but finally attach themselves to the ectodermic body walls, especially in the pos¬ terior two-thirds of the embryo on each side of the heart (Fig. 86 ad). They remain more or less globular or oval, the side in contact with the wall hollowing out a concavity in the cells of the ectoderm. The granulation of the cytoplasm which first distinguishes the fat cells from the true entodermic elements becomes coarser with the increase in volume. In the embryo ready to hatch the adipose cells have acquired gigantic dimen¬ sions, being many times the size of those represented in Fig. 86. Both nuclei and cytoplasm stain deeply, so that these fat cells are rendered among the most conspicuous objects in a section. B. Mesoderm. In Doryphora as soon as the gastrular tube has collapsed, the polygonal mesodermic elements form a layer several cells in thickness, applied to the inner surface of the median ventral ectoderm (Fig. 65 msd). This layer of cells thins out at its lateral edges. With the first traces of segmentation in the outer layer the mesodermic layer also divides, though incompletely, at the same places of constriction (Fig. 82). Soon the single intersegmentally divided band of mesoderm splits in the median line so that each segment contains two subquadrangular flat¬ tened masses. The coelomic cavity is formed at the outer edge of each of the masses by a separation of the cells of the two layers (Fig. 78 cl). The inner constitutes the splanchnic mesoderm, while all the remainder of the mesoderm constitutes the somatic layer (Fig. 78 shn). When the appendages appear, it is the latter layer of cells which supplies their cavities with muscle-forming cells; the portions inclosing the coelomic cavity accompany the adjacent ectoderm in its dorsad growth. As 3^4 WHEELER. [vbL. rrr. soon as the growth in this direction is fairly started the ecto¬ derm bulges out (Fig. 83 ect) and, drawing with it the somatic mesoderm, leaves a cavity between the yolk and the embryo which soon communicates with the coelomic cavity and assumes large dimensions (Fig. 84). Through the body cavity thus formed a thin plasma found as coagulated masses in hardened embryos probably circulates. The cell cb (Fig. 83), which is recognized even at a much earlier stage by its peculiar form, and which is destined to take part in the formation of the heart, is the only element still uniting the splanchnic and somatic layers. This cell, as may be clearly seen in Figs. 83 and 84 cb^ is triangular in cross-section, and inserts one of its acute angles between the yolk and the ectoderm. As this cell with those of exactly the same shape anterior and posterior to it form the heaft, I shall call it a cardio- blast. The true shape of the cardioblasts may be seen in a thick frontal section (Fig. 89) through the embryo of which Fig. 84 is a cross-section. Here the compact row of crowded but regular heart-forming elements {cb) is seen running between the actively proliferating entoderm cells {e 7 it) on the one hand, and the meso¬ derm cells {msd) on the other. These last are somewhat scat¬ tered in the space between the cardioblasts and the thickened ectoderm. In the more advanced embryo (Fig. 85) the cardioblasts {cb) from either side are near together. They have retained their characteristic form and position, while the somatic mesoderm has been converted into muscles {^nsl) and connective tissue, and the splanchnic layer {shn) has applied itself closely to and is coextensive with the single-celled layer of entoderm {ent). In Fig. 86 the heart is completed. A glance at this figure and Fig. 85 shows the manner in which the two cells from opposite sides unite. Though forming the two halves of the tube in Fig. 86, they still show the three angles which were apparent just after the formation of the coelomic cavity. The heart remains in the condition shown in Fig. 86 till the embryo hatches. I have not studied the formation of the blood in Doryphora. My observation on the formation of the sexual organs, though more complete than in Blatta, are still very fragmentary. These organs originate as two elongate thickenings of splanchnic No. 2.J B'LATTA AND DORYPHORA. mesoderm, one on each side projecting into the body cavity. Later (Fig. 84 they become rounded and are attached by a thin band of splanchnic mesoderm only. I have seen the much attenuated duct leading from each organ to the exterior, but have made no observations on its origin. The ducts converge posteriorly, but end by separate openings on the i ith abdominal somite; thus presenting a condition which in Ephemerids is permanent throughout life, according to Palmen (37). There is probably some connection between the two pairs of very indistinct tracheal openings in the lOth and nth somites and the openings of the efferent ducts, but I was unable to deter¬ mine whether the large sexual openings originate by enlarge¬ ment from a single pair of these tracheal openings, or from the confluence of all four to form two orifices. The cross-section (Fig. 80) includes the openings of the efferent ducts {go go), the knife having taken away a very thin layer of surface cells. Up to the formation of the coelomic cavities the mesoderm of Blatta closely resembles the same layer in Doryphora. With the evagination of the appendages from the entoderm a decided difference is, however, observable. Each coelomic segment, if situated in an appendage-bearing segment, instead of retreating dorsally as in Doryphora, sends a diverticulum into the appen¬ dage. This is clearly seen in Fig. 53 cl, a cross-section from an embryo twelve days old. The cells of the diverticulum develop into the muscles of the appendage, and together with a portion of the mesodermic layer still remaining in the body cavity are shut off from what probably represents the true coelomic cavity (Fig. 54 cl). The further changes again resemble those in Doryphora. Long before the heart is formed and the lateral walls have met , in the median line, the body walls of the embryo are observed to pulsate regularly like the body walls of Gryllotalpa, as described by Korotneff (25). As in Doryphora a plasma is at this time found in the body cavity which is divided by films of connective tissue into a great number of small intercommunicat¬ ing lacunae (Fig. 54). In regard to the primitive blood sinus, my observations confirm Patten’s (38). He says : “The primitive blood sinus is the space between the somatic and splanchnic mesoderm, divided into a number of smaller and irregular sinuses by meshes of connective tissue, some cells of which, in the earlier stages, become free and form the blood capsules. By WHEELER. [VOL. III. 366 the pulsation of the mesodermic folds, long before a special heart is formed, a circulation through the body cavity is brought about like the circulation in many of the lower worms.” One of the stages in the formation of the heart is seen in Fig. 52. The cardioblasts {cb cb) are both more numerous and more irregular than in Doryphora. They unite to form a tube, the lumen of which is at first oblong in cross-section (Fig. 52 Ji). In the cardiac walls amoeboid cells are occasionally seen ibl), which loosen themselves from the mesodermic elements and pass into the lumen of the tube, probably to form blood cor¬ puscles. C. Ectoderm. My observations on the organs derived from the ectoderm are limited almost exclusively to Do7ypho7'a. The hypodermis of the embryo secretes two cuticles, the second of which covers the larva, while the first is cast off at the time of hatching. Shortly before hatching the embryo is confined by four loose envelopes, — the chorion, the vitelline membrane, the serosa, and the first cutiele. Graber (15) has made a similar observation on the embryo Lina. The three broad-based chitinous spines used by the insect in rupturing its envelopes, and which are analogous to the frontal spine observed in Strojigylosoma by Metschnikow (31) and the deciduous claw on the beaks of birds, are secreted by pyramidal ' thickenings of the hypodermis (Fig. 86 hsp), the cells of which are much lengthened, though forming a single layer. In the surface views of the embryo (Fig. 72) it is possible to trace the origin of all the ganglia as paired thickenings of the outer layer. At first these thickenings differ histologically in no particular from the surrounding ectoderm (Fig. 78 ;/r). Gradually, however, the cells in the centre of each thickening enlarge, while their cytoplasm becomes drawn out into fine threads. At the same time all the ganglion cells thus formed arrange themselves in such a way as to have their threads inter¬ mingle. This mass of hiteiLzvined threads becomes the P2inktsub- stanz (Fig. 94 to 104 pet). The outer layer of cells if) contin¬ uous with the hypodermis {ecd) stands off somewhat from the ganglionic thickenings, leaving a space which is in early stages occupied by several large, clear, oval cells (gbl), which divide No. 2.] BLATTA AND DORYPHORA. 367 rapidly by karyokinesis, and might be called£‘ang’/wd/asts, as the products of their divisions reinforce the mass of ganglion cells. In a series of sections through the mesothoracic pair of ganglia (Figs. 94 to 104) the Mittlestrang may be readily traced. The shape of its cross-section varies with the plane of section through the ganglion and its cross or longitudinal commis¬ sures. At the two points in the ganglion where the two pairs of Punktsubstanz masses fuse to form the commissures c cnty the Mittlestrang is in great part obliterated. The median strand is largest where the two longitudinal commissures are passing into the anterior ends of the three thoracic ganglia. Here they persist in the larva, while completely disappearing elsewhere, and become converted into the three chitinons fiircce, each one of which is just in front of a thoracic ganglioji. Figure 105, from a section through the fore end of the meta- thoracic ganglion, shows the Mittlestrang portion (inst) contin¬ uous with the hypodermis {ect) and broadening out into the furca (/) after passing between the two halves of the ganglion {gl). Muscles {7?tsl) are attached to the two divergent ends. The mesothoracic furca, which is formed in exactly the same manner, is seen in Fig. 86 f, where it passes between the commissures {cm)', its connection with the muscles is seen at r. A frontal section (Fig. 91) shows the structure of the meso- and meta-thoracic ganglia after they have become loosened from the surface ectoderm. The longitudinal {cm) and cross-com¬ missures {c cm) are clearly seen as white Punktsubstanz sepa¬ rated from the ganglionic cells by the inner neurilemma {inrl). The outer neurilemma {onrl) is also developed, as are also the two main nerve trunks {fd-, ft), the anterior of which bifurcates {n) while leaving the ganglion. The separation of the nervous system from the integumen¬ tary ectoderm progresses from before backwards. The two brain masses separate first. The first segment becomes very small and possibly disappears. The three segments of the optic ganglion are invaginated and pushed under the optic plate in a manner which I believe to be similar to that de¬ scribed by Patten (39) in Acilius, though I have not followed the details of the process. The frontal ganglion is formed as an unpaired thickening of the dorsal imll of the oesophageal ecta- 368 WHEELER. [VOL. III. derm Jiear the base of the labriim. Punktsubstanz is formed in this ganglion in the same manner as in the brain and ventral ganglia. It is ultimately loosened from the stomodaeum, and becomes surrounded by mesodermic elements. Later the four pairs of ganglia of the intercalary, mandibular, ist and 2d max¬ illary segments fuse and form the infraoesophageal ganglion. That the outer neurilemma is of ectodermic and not of meso¬ dermic origin seems to be proved by the fact that shortly after the separation of the nerve-chord from the integnmentary ectoderm^ it sheds from its surface a delicate chitinous cuticle simultaneotisly zvith the shedding of the first integumentary ciUicle, This cuticle, which is separated from the surface of the outer neurilemma, and even from the surfaces of the main neural trunks, is afterwards absorbed. The six ocelli are formed as apple-shaped thickenings of the optic plate. Their small size has hindered me from studying their structure in detail. I have represented in Fig. 8i, soon after their first appearance, two of the ocelli corresponding to the eyes of Acilius numbered V. and VI. by Patten. Each forms a slight depression somewhat paler than the surrounding ectoderm. The nucleus {ft) of one of the central cells is seen to be much larger than the nuclei of the surrounding cells. Patten has described and figured this same large nucleus in the eyes of Acilius (PI. XL, Figs. 63, 64, 65, etc.). The five pairs of invaginations anterior to those of the second maxillary segment form the tentorium of the larval head. These invaginations grow inwards as slender tubes, which anastomose in some places. Their lumina- are ultimately filled with chitin. Palmen (36) found that the tentorium of Ephemerids breaks across the middle during eedysis, and that each half is drawn out of the head, like the chitinous lining of a tracheal tube. This fact, together with my observations on the tentorium of Doryphora, makes it highly probable that, as Palmen suggests, the tentorium is formed from tracheae, which have become modified for muscular attachment. Of the true tracheal invaginations, those of the pro- and meta-thorax disappear. The mesothoracic spiracle comes to lie between the pro- and meso-thoracic segments near the base of the legs and ventral to the line of abdominal spiracles. This first spiracle is the largest, and in cross-section appears as a No. 2.] BLATTA AND DORYPHORA. 369 short, chitinous cylinder, projecting somewhat from the general surface of the segment. Its inner walls are lined with spines which are directed outward. The abdominal spiracles, though smaller, are also lined with similar spines (Fig. 85 tr). The first pair of tracheae send large branches to the head. The abdominal tracheae of each side of the body anastomose to form a longi¬ tudinal trunk, from which the branches ramify to the different organs. The invaginated ectodern, at first very thick (Fig. 84 tr)y gradually thins out as the respiratory tubes lengthen and ramify. The thin epithelium thus formed secretes the chitinous lining provided at the time of hatching with the spiral thicken¬ ings so characteristic of insect tracheae. The proctodaeum and stomodaeum when first formed are tri¬ angular in cross-section. Later both become hexagonal (Figs. 79 and 80). A like pronounced similarity in form between the stomodaeum and proctodaeum of Gammariis has been observed by Pereyaslawzewa (40). She says : “Fait tres interessant, qui merite d’etre note, c’est qu’a mesure de I’acroissement du rectum et de I’oesophage, leur partie interieure affecte absolument la meme forme carree, dont les parois sont concaves. Ce qui concerne la configuration des cavites, elles n’en different aucune- ment et se dessinent sous forme d’une croix oblique; la dis¬ semblance consiste en ce que dans I’oesophange ce sont les parois qui s’enfoncent, tandis que les parois du rectum sont tapisses d’un epithelium cylindrique, dont les cellules s’aplatis- sent graduellement vers les angles.” The three pairs of Malpighian vessels appear at a very early period, while the proctodaeal invagination is still very shallow. They are from the first hollow diverticula, and have their blind ends pushed back by the forward growth of the proctodaeum. Thus it happens that a transverse section through the tail-end of an embryo in the stage of Fig. 74 passes through both the proctodaeum and the Malpighian vessels (Fig. 79, ;r, mpg, When the blind ends of the six tubes have struck the body wall, their continued growth forces them to turn and grow forward. After the formation of the dorsal body wall, they may be found lying as thin undulating tubes, surrounding the mesenteron at approximately equal distances from one another (Fig. 86 mp^. The very early appearance of the Malpighian vessels and their paired arrangement in Doryphora^ Fig. 72, would seem to 370 WHEELER. [VOL. III. indicate that at one time they opened on the surface of the body, and that their orifices were subsequently carried in by a deepening of the proctodaeal invagination. Possibly these tubes in insects are homologous with the anal tubes of the Echiuriis larva, which are modified segmental organs. Gegenbauer (14) has intimated the possible derivation of the Malpighian vessels from paired tubes opening on the outer surface of the body. He says: “As they [the Malpighian vessels] are formed at the same time as that portion of the hind-gut, which in tfie embryo is developed from the ectoderm, it is not improbable that they primitively opened on the surface of the body, or were derived from organs which did so. In all divisions there are two chief canals, as is often seen at the point where a large number of canals open and unite. This number may therefore be regarded as a primitive character.” Milwaukee, Sept. 12, 1888. LIST OF WORKS REFERRED TO. \/h. Ayers, H. On the development of OEcanthus niveus and its Parasite Teleas. Memoir. Boston Soc. Nat. Hist. Vol. III., No. 8. 1884. 2. Balfour, F. Treatise on Comparative Embryology. Vol. II. 1881. 3. Blochmann, F. Ueber eine Metamorphose der Kerne in den Ovarial- eiern und iiber den Beginn der Blastodermbildung bei den Ameisen. Verhand d. naturh. medicin. Vereins zu Heidelberg. Bd. III. Heft 3. 1884. 4. Blochmann, F. Ueber die Reifung der Eier bei Ameisen und Wespen. Festschrift zur Feier des 100 jMir. Bestehens der Ruperto-Carola. Heidelberg. 1886. 5. Blochmann, F. Ueber die Richtungskdrperchen bei Insekteneiern. Morph. Jahrb. Bd. XII. Heft 4. 1887. 6. Brandt, A. Ueber die Eirdhren der Blatta (Periplaneta) orientalis. Mem. Acad. St. Petersbourg. Ser. 7. Tom. XXL 1874. 7. Bruce, A. T. Observations on the Embryology of Insects and Arachnids. Baltimore, 1887. 8. Biitschli, O. Bemerkungen iiber die Entwicklungsgeschichte von Musca. Morph. Jahrb. Bd. XIV. Heft i. 1888. 9. Carnoy, J. B. La Cytodierese chez les Arthopodes. “ La Cellule.” Tom. 1 . Fasc. II. 1885. vXio. Cholodkovsky, N. Ueber die Bildung des Entoderms bei Blatta ger- manica. Zoolog. Anzeig. No. 275. 1888. Emery, C. (Critical note on Korotneff and Grass!) Biolog. Centralbl. Bd. V. 1887. II. No. 2.1 t -* BLATTA AND DORYPHORA. 371 12. Flemming, W. Zellsubstanz, Kern u. Zelltheilung. Leipzig, 1882. 13. Flemming, W. Neue Beitriige zur Kenntniss der Zelle. Arch. f. mikr. Anat. Bd. 29. Heft 3. 1887. 14. Gegenbauer, C. Elements of Comparative Anatomy. Trans, by F. Jeffrey Bell. London, 1878. /I5. Graber, V. Die Insekten. II. Theil. Miinchen, 1877. 16. Graber, V. Ueber die primare Segmentirung des Keimstreifs der In¬ sekten. Morph. Jahrb. Bd. XIV. Heft 2. 1888. ^^17. Hallez, P. Loi de I’orientation de I’embryon chez les Insects. Compt. Rend. Tome 103. 1886. yi 8 . Hatschek, B. Beitrage zur Entwicklungsgeschichte der Lepidopteren. Naumburg a/S., 1877. V^iQ. Heider, K. Ueber die Anlage der Keimblatter von Hydrophilus piceus. Abh. der K. Preuss. Akad. d. Wiss. z. Berlin v. Jahre 1885. y/' 20. Henking, H. Untersuchungen fiber die Entwicklung der Phalangiden. Theil 1 . Zeitschr. f. wiss. Zool. Bd. 45. 1886. 21. Hertwig, O. Welchen Einfluss iibt die Schwerkraft auf die Theilung derZellen? Jena, 1884. , 22. Huxley, T. A. Manual of the Anatomy of Invertebrated Animals. 1878. 23. Kadyi, H. Beitrag zur Kenntniss der Vorgange beim Eierlegen der Blatta orientalis. Zoolog. Anzeig. No. 44. 1879. 24. Kennel, J. Entwicklungsgeschichte v. Peripatus Edwardsii, Blanch. and P. torquatus, n. sp. Arbeit, aus d. zoolog. zootom. Instit. in Wiirz- burg. Bd. VII. Heft 8, and Bd. VIII. Heft i. 1886. 25. Korotneff, A. Die Embryologie der Gryllotalpa. Zeitsch. f. wiss. Zool. Bd. 41. Heft 4. 1885. ■y/26. Kowalevski, A. Embryologische Studien an Wiirmern u. Arthropoden. Mem. de PAcad. imp. d. Sciences de St. Petersbourg. Ser. VII. Tom. XVI. No. 12. 1871. 27. Kowalevski, A. Zur Embryonalen Entwicklung der Musciden. Biolog. Centralbl. Bd. VI. No. 2. 1886. ,^28. Leichmann, G. Ueber Bildung von Richtungskorpern bei Isopoden. Zoolog. Anzeig. No. 262. 1887. v^^9. Leydig, F. Beitrage zur Kenntniss des thierischen Eies im unbefruchte- ten Zustande. Zoolog. Jahrbiicher. Bd. III. Heft 2. Jena, 1888. 30. Metschnikow, E. Embryologische Studien an Insecten. Leipzig, 1866. 31. Metschnikow, E. Embryologie der doppeltfiissigen Myriapoden (Chil- ognatha). Zeit. f. wiss. Zool. Bd. XXIV. 1874. 32. Miall, L. C., and Denny, A. The Structure and Life History of the Cockroach (Periplaneta orientalis). London, 1886. ■^33. Nusbaum, J. The Embryonic Development of the Cockroach (Blatta germanica) (in the work of Miall & Denny). 34. Nusbaum, J. L’embryologie de Mysis chameleo, Thomp. Arch. Zool. Exper. TomeV. 1887. ^ 35. Packard, A. S. The Embryology of Chrysopa and its Bearings on the Classification of the Neuroptera. Am. Naturalist. Vol. V., p. 564. 1871. 372 WHEELER. 36. Palm^n, J. A. Morphologie des Tracheensystems. 1877. 37. Palm^n, J. A. Ueber paarige Ausfuhrungsgange der Geschlechtsorgane bei Insekten. Helsingfors, 1884. 38. Patten, W. The Development of Phryganids, with a preliminary note on the Development of Blatta germanica. Quart. Journ. Mic. Science. Vol. XXIV. 1884. ^39. Patten, W. Studies on the Eyes of Arthropods. II., Eyes of Acilius. Journal of Morph. Vol. II. No. i. 1888. 40. Pereyaslawzewa, Sophie. Le developpement de Gammarus poecilurus, Rthk. Bulletin d. la Soc. Imp. d. Naturalistes d. Moscou. Annde 1888. No. 2. 41. Rabl, C. Ueber Zelltheilung. Morph. Jahrb. Bd. X. Heft 2. 1884. 42. Ratlike, H. Zur Entwicklungsgeschichte der Blatta germanica. Meckel’s Archiv f. Anat. u. Physiol. Bd. VI. 1882. 43. Sedgwick, A. On the Origin of Metameric Segmentation, and some other Morphological Questions. Stud, from the Morph. Lab. of Cam¬ bridge. Vol. II. Pt. I. 1884. Strasburger, E. Zellbildung und Zelltheilung. 3 Auflage. Jena, 1880. 45. Stuhlmann, F. Die Reifung des Arthropodeneies nach Beobachtungen an Insekten, Spinnen, Myriapoden u. Peripatus. Bericht d. naturf. Gesellsch. z. Freiburg i/Br. Bd. i. Hefte 5, 12. 1886. \/ 46. Taschenberg, E, L. Blatta germanica. In Brehm’s Thierleben. Vol. IX., p. 536. Leipzig, 1884. 47. Weismann, A. Beitrage zur Kenntniss der ersten Entwicklungsvorgange im Insektenei. Bonn, 1882. V 48. Weismann, A. Ueber die Zahl der Richtungskorper und ihre Bedeutung fiir die Vererbung. Jena, 1887. >^ 49- Weismann, A., and Ischikawa, C. Ueber die Bildung der Richtungs- kbrper bei thierischen Eiern. Berichte d. naturf. Gesell. z. Freiburg i/Br. Bd. III. Heft i. 1887. ^ 50. Wheeler,W. M. On the Appendages of the First Abdominal Segment of the Embryo Cockroach (Blatta germanica). Proceed. Wis. Acad. Science, Arts and Letters. Vol. VIII. 1888. 51. Will, L. Bildungsgeschichte und morphologischer Werth des Eies v. Nepa cinerea, L. u. Notonecta glauca, L. Zeitschr. f. wiss. Zool. Bd. 41. Heft 3. 1885. 52. Will, L. Entwicklungsgeschichte der viviparen Aphiden. Zoolog. Jahrb. Bd. III. Heft 2. 1888. r'' ‘ • » ■ ♦ y» ’! * ' . , >• k r~ I I : v> % f 1 !(*'■ ■ I ' • ''rJ' 'U . • . ' ^ 'I " .. ■ ■ * 'i..; / •» . I —I 1 >. ■. ‘r^ ■'' f.- i 374 WHEELER. DESCRIPTION OF PLATE XV. Blatta ger 7 nanica. Fig. I. Piece of chorion from lateral face of oothecai egg. Fig. 2. Piece of chorion from micropylar area with micropyles. a, aperture of micropyle; b, tubule. Fig. 3. Transverse section through the crista of a capsule. 0, oothecai wall split¬ ting into two laminae, 6 ^ and cr, crystals of calcium oxalate; ep, remains of fol¬ licular epithelium. Fig. 4. Longitudinal section of epithelial cap on the germarium pole of the egg. ep, follicular epithelium; ch, chorion secreted by the same; a, the point at which the chorion widens, showing its lengthened trabeculae; b, pyriform dilatation of the chorion; Pr, protoplasm of the Keinhaut, containing numerous bacteria-like corpuscles. Fig. 5. A number of cells from the follicular epithelium. «, resting nucleus; b and c, nuclei in akinetic division. Fig. 6. Longitudinal section of immature ovum i mm. long, being the circum- nuclear portion in the centre of the dorsal concavity of the egg; nucleus amoeboid just after reaching the surface, n, chromatin particles; x, layer of small yolk bodies just beneath the epithelium. Fig. 7. Same section of an egg 2 mm. long. Nucleus emarginate on outer surface, giving off the maturation sphere b-, y, small yolk bodies under the epithelium.. Fig. 8. Section at right angles to that in Fig. 7, of a nucleus in the same stage. b, cavity into which the maturation spheres fitted; pn, paranucleolus. Fig. 9. Nucleus in the same stage; pn, paranucleolus. Egg 2 mm. long; section in the same plane as Fig. 7. Fig. 10. Longitudinal section of median dorsal surface of an egg 2.5 mm. long. Particles of chromatin aggregating into a somewhat stellate central mass. Fig. II. Longitudinal section corresponding to Fig. 10 of an egg somewhat older, 2.8 mm. long. A delicate wall encloses the particles of chromatin. Fig. 12. Pdrst polar spindle in the metakinetic stage. Egg mature. Fig. 13. First polar spindle in anaphasis. Egg mature. Fig. 14. Median longitudinal section through the dorsal surface of an oothecai egg 4 to 6 hours old. pgl, first polar globule; x, second polar spindle in metakinesis. Fig. 15. Longitudinal section through botl^ polar globules, 3 to 4 hours older than egg represented in Fig. 14. Fig. 16. Surface view of the median dorsal portion of an oothecai egg about 10 hours old. pgl pgl, polar globules; 9 P^h female pronucleus (out of focus). Fig. 17. Median longitudinal section of dorsal portion of an oothecai egg. a, dorsal contour; J p 7 t, female pronucleus just after leaving the polar globules. Fig. 18. Longitudinal section of circumnuclear region, from near the middle of the homogeneous yolk, with female pronucleus. The arrow in this and the following figures points in the direction of the cephalo-caudal axis of the egg. Fig. 19. Longitudinal section of circumnuclear portion of egg, one-third the length of the egg from the cephalic pole, a, dorsal contour; 9 female ^ronucleus; ^ pn, male pronucleus. Fig. 20. Longitudinal section of circumnuclear area from middle of homogeneous yolk. $ pn and 9 male and female pronuclei conjugating. Fig. 21. Longitudinal section of circumnuclear area from the middle of the front portion of the homogeneous yolk. Cleavage nucleus. Fig. 22. Cleavage nucleus preparing to divide, cp, cytoplasm. f .Joiirn. Morjih. Vol.lll PL xp: . '''>.'' ^oc..*.','^- o9 ‘ rojOtA^y .%o V i*''/>0®o‘’o c®. j;#, OoftO ®®*a.S^*oo^ ■ ' O^' ^ i\ a ■. *' rtJ o > 1,0 ®c ^ <•,.'■'* r-*^ ' ' • iOi«\» ^ ‘’ 1 . V ^ ^ Q o - o c® 0 r-^ ^0 ® J"' ' C - %'’o^’V'’*0'^cn^yo<5.. “„' . “ •’^o 'i'>^6^“.»'> V ,"<.o»" - ■’ ° Q--0 J^o •.'* ' o ' -’ 0 3 0 Jo S» 0 « ® C ® ' Oo^* 4 =Scvk V-" VS<^V^C ^61 V c;- / ‘ r.o ^-> 55 'i: W.M.Wheeler.del. B.PfetseLJilh.Boston. / 3/6 WHEELER. DESCRIPTION OF PLATE XVI. Blatta germanica. Fig. 23. Cleavage nucleus dividing, a, points to the ventral contour. The nucleus in this egg had moved out into the granular yolk some distance. Fig. 24. Two nuclei; a, seen in lateral, and b, in polar view, from an egg contain¬ ing 4 nuclei in the metakinetic stage of division. Fig. 25. Ventral third of a transverse section of an odthecal egg; a and b, cells which have just reached the surface. The amoeboid cell in the interior is on its way to the surface. Fig. 26. Corresponding section of the somewhat older egg (Fig. 36). a, a group of cells formed by tangential akinesis from one of the nuclei, like a, Fig. 25. Fig. 27. Ventral third of a transverse section of an egg in the blastoderm stage; 6 days old. Fig. 28. Ventral third of a transverse section. Blastoderm contracting. At 3r', yolk nucleus being given off; at x, one which has sunk deep into the yolk. Fig. 29. Transverse section through e, of an embryo like the one represented in Fig. 43, V, yolk-cell; msd, mesoderm; ecd., ectoderm. Fig. 30. Transverse section through d of the embryo 43. am, amnion; sr, serosa; ecd, ectoderm; msd, mesoderm; v, yolk-cell. Fig. 31. Transverse section through the point pd of Fig. 41. ecd, ectoderm; v, yolk-cell; i7isd, mesoderm. Fig. 32. Transverse section through a point a little in front of as of Fig. 43. am, amnion; sr, serosa; ecd, ectoderm; 7 nsd, mesoderm; v, yolk-cell. Fig. 33. Surface view of blastoderm, showing the binucleolate nuclei. Fig. 34, Enlarged view of syncytium x, in Fig. 36; showing the unequal size of the nuclei. Fig. 35. a and c, surface nuclei in akinetic division from an egg in the same stage as Fig. 36; b, serosa nucleus in akinesis from a much older egg. Fig. 36. Surface view of an egg shortly after the nuclei have begun to appear on its surface, c, cephalic end; s, sinus; v, ventral surface; dd, dorsal surface; x, a group of nuclei like that seen in section at a, in Fig. 26. Fig. 37. Surface view of an egg in the blastema stage; 4 days old. Fig. 38. Long. sect, along the carina cn, of Fig. 41; through the depression bp. sr, serosa; bp, depression; ecd, ectoderm; msd, mesoderm; v, yolk-cells. r* « ;* rj f- ■ ’ \ • * I f J •jJaf-w*'.- uBxm ' .. OF.THE r f - .*'1 • ■> “ t \ ■■■. tr » I . r f; r I* ’‘l' r'-4 i ‘ >« jM ' . PL XVL Joani. Lfoijili. VoL.lII. r?zscL am T > t * WM. Wheeler, del. B.MeiselJith Boston. 3/8 WHEELER. DESCRIPTION OF PLATE XVII. Blatta gernianica. Fig. 39. Diagram of a median longitudinal section of an egg to show paths of nuclei. Explanation in the text. Fig. 40. Diagram of a transverse equatorial section of an egg to show paths of nuclei. Explanation in the text. Fig. 41. Surface view of the ventral face of an egg *j\ days old. cn, carina; pd^ beginnings of procephalic lobes; hp, depression (blastopore?) in the middle of the rounded area of proliferation. Fig. 42. Surface view of the ventral face of an egg 8 days old. as, amnion and serosa beginning; pd, beginning of procephalic lobes. Fig. 43. Surface view of an embryo in “ slipper ” stage, as, amnion and serosa. Fig. 44. Surface view of an embryo in the “ hammer” stage, 9| days old. 0, open¬ ing in amnion and serosa over the oral region; at, antennary lobe; p, beginning of the second thoracic appendage. Fig. 45. Embryo about ii days old, isolated from the yolk, lb, labrum; st, stomo- dseum; brain; antennary lobe; mandible; ist and 2d maxillae; p^ to p^, thoracic appendages; as, amnion and serosa; cpl, caudal plate. Fig. 46. Embryo just after rupture of amnion and serosa; about 14 days old. b, cephalic end of yolk; oc, eye; ad, fat body, ast, anal stylets; appendages of ist abdominal segment; remaining references same as in Fig. 45. Fig. 47, Front view of same embryo. References the same as in Fig. 45. Fig. 48. Embryo almost ready to hatch, oc, eye; cgl, brain; gl'^, infraoesophageal ganglion; gl^ to the three thoracic ganglia; gl^, last of the abdominal ganglia; oe, oesophagus; c, crop; gz, gizzard; ht, heart; st, stomach; ret, rectum; mpg, Mal¬ pighian vessels; ast, anal stylets; y, yolk; lb, labrum; nid, mandibles; inx^ and vix^, 1st and 2d maxillae; p^ to p^, thoracic appendages; at, antennae. Fig. 49. Blatta germanica about a day after hatching. Fig. 50. Sagittal longitudinal section through the middle of the dorsal organ from an embryo 16 days old. a, anterior; b, posterior end of dorsal organ; 0, depression; V, a yolk nucleus. Fig. 51. Frontal section through the anterior end of dorsal organ in an embryo 18 days old. cc, heart; nn, nuclei of serosa cells; 0, cavity formed from the depression 0, of Fig. 50; V, yolk nucleus. Fig. 52. Cross-section through the dorsum of an embryo to show formation of the heart. /^, cavity of heart; cardioblasts; ectoderm; somatic mesoderm; slm, splanchnic mesoderm; ent, entoderm; bl, blood corpuscle becoming loosened from the wall of the heart; y, yolk. Fig. 53. Cross-section through the thorax of an embryo 12 days old. nr, neural thickening; am, amnion; sr, serosa; v, yolk nucleus; nisd, mesoderm; cl, coelomic cavity; p^^, second pair of legs. Fig. 54. Cross-section through 2d maxillcC of an embryo 16 days old. mx^, second maxilla; g, duct of salivary gland; y, yolk; ent, entoderm; cb, cardioblast cells; remainder of references as in Fig. 53. Fig. 55. Cross-section through the basal abdominal somite of an embryo 26 days old. gl, ganglion; ad, corpus adiposum; v, yolk nucleus; tr, tracheal opening; yolk; cc, heart. I > J THE Of m nf; t» »#!«<»I *• it \ \ Jouni. Mor/ih. Vol.lll. PL XVII XM.V.'heelet.id. B.MeiselJilh.Boston. 38 o WHEELER. DESCRIPTION OF PLATE XVIII. Doryphora decemlineata. Fig. 56. Portion of a longitudinal section through a young egg shortly after the germinal vesicle has reached the periphery, pi, karyoplasm; tr, protoplasmic trabe¬ culae connecting the karyoplasm with the intervitelline protoplasm (cytoplasm of the large cell which the egg represents); y, yolk bodies; ep, follicular epithelium; nl, germinal spot (nucleolus); mb, maturation balls (Reifungsballen of Stuhlmann) Zeiss oc. III. Fig. 57. Section corresponding to that of Fig. 56 from an egg almost mature. n, large mass of chromatin destined to enter into the first polar spindle; bl, layer of surface protoplasm; y, yolk bodies; ep, follicular epithelium. Zeiss oc. III. Fig. 58. Longitudinal section of a mature egg. resting nucleus originating from the mass of nuclein n of Fig. 57; bl, differentiated peripheral layer of protoplasm; y, yolk bodies; v, vitelline membrane; ch, chorion. Zeiss Fig. 59. Portion of median transverse section of a mature egg. n, first polar spindle in metakinetic stage; remaining references the same as in Fig. 58. Zeiss oc. III. Fig. 60. Same section as in Fig. 59 of an egg about to be deposited, nucleus in last stages of ahaphasis; the peripheral mass the first polar globule. Remaining references the same as in Fig. 58. Zeiss jLj oc. III. Fig. 61. Half of the median transverse section of an egg containing several nuclei n^, n^, none of which have as yet entered the surface layer of protoplasm bl. y, yolk. Fig. 62. Half of the median transverse section of an egg shortly before the blasto¬ derm stage. V, yolk-cells; a, blastema cell resting; b, blastema cells in metakinesis; c, blastema cells in anaphasis; y, yolk. Fig. 63. Half of the median transverse section of an egg in the blastoderm stage. V, yolk-cells; bid, blastoderm. Fig. 64. Half of a median transverse section of the egg of Fig. 67. g, half of the gastrular groove forming on the ventral face; r, ridge separating the groove from the remainder of the blastoderm; p, flat blastoderm cells on the dorsal surface of the egg. Fig. 65. Median transverse section of an egg carrying the embryo represented in Fig. 71. The upper half of the section passes through p^ of Fig. 71, the lower half a little in front of x. sr, serosa; am, amnion; ecd, ectoderm; msd, mesoderm; ent, cells still forming part of the induplicated ectoderm at pi. which give rise to the ento¬ derm; y, yolk. Fig. 66. Ventral view of egg shortly after formation of blastoderm; the two paren¬ thesis-shaped ridges inclose the portion which will sink in to form the gastrula. Fig. 67. Ventral view of egg with gastrula more advanced, a, oral spade-shaped broadening of the gastrula; b, the point at which the groove turns abruptly inwards; sr, serosa. Fig. 68. Lateral view of same egg. pci, procephalic lobes; remaining references the same as in Fig. 67. Fig. 69. Caudal end of egg represented in Fig. 70. g, gastrula; am, caudal fold of the amnion; sr, serosa. Fig. 70. Ventral view of an egg with gastrula fully formed, a, oral widening of the gastrula; am, cephalic fold of amnion; br, brain thickening. Fig. 71. Slipper-shaped embryo removed from egg and unrolled, p^ to p^, indica¬ tions of the 3 pairs of thoracic appendages; as, amnion and serosa; am, cephalic fold of amnion and serosa; a, oral and x, anal end of gastrula; b, maxillary region. I 1 : V Ji (pCywrr ‘ , V- A ^/ / / ^ -y/ >;‘tv - = ‘ ; ■.' f f f , ♦ I*" ' ' p # > ►' ; • _ .» If ✓ I PL XVIIl : (.'' 'V?»« ^ ^ • am. yo 0 Q »| 1 « 0*^ f ®v am cun k-si Joiini. Mor/ih . Vo 1.III. ft'l/.IiSee/a-. Je/. B.Meisel, Jilh. Boston. 382 WHEELER. DESCRIPTION OF PLATE XIX. Doryphora decemlineata. Fig. 72. Embryo shortly after the appearance of the appendages, unrolled and iso¬ lated. c, stomodaeum; a, proctodseum; lb, labrum; to three brain segments; og^ to og^, three segments of the optic ganglion; op^ to op^, three segments of the optic plate; to five pairs of invaginations which form the tentorium; f to tracheal invaginations; the two last pairs and either disappear or form the openings of the sexual ducts; at, antennce; nid, mandibles; inx^ and mx^, first and second maxillae; to p^, three pairs of thoracic appendages; c, commissure con¬ necting the two ganglionic thickenings g'^ of the intercalary segment; gl, ganglionic thickening; mst, Mittelstrang thickening; mpg^ to nipg'^, three Malpighian vessels. Fig. 73. Embryo just after the closing of the amnion and serosa (jr) unrolled and isolated, a, oral, x, anal end of the gastrula; pci, procephalic lobe; nix^^, second maxilla; p^ to/^, three thoracic appendages; cp, caudal plate. Fig. 74. Ventral view of embryo, the lateral walls of which have embraced half the yolk; references same as in Fig. 72. Fig. 75. Lateral view of embryo ready to hatch, at, antenna; mesothoracic spiracle; to abdominal spiracles; hsp, first hatching-spine. Fig. 76. Sagittal section to one side of the median line through the tail of embryo. Fig. 72. a?n, amnion; sr, serosa; ? 7 ipg'^ and mpg^, first and third Malpighian vessels; ecd, ectoderm; 7 Jisd, mesoderm; oit, forward growing band of entoderm; e 7 if^, mass of entoderm left under the end of the proctodseum. Fig. 77. Transverse section through the second pair of Malpighian vessels ( 77 ipg‘^) of the same embryo, x, proctodaeum; other references same as in Fig. 76. Fig. 78. Median cross-section through an egg containing an embryo a little older than that represented in Fig. 72. The section cuts the embryo at two places. That with the entoderm belongs to one of the last abdominal segments, that without entoderm to one of the basal abdominal segments, wr, neural thickenings; ecd and eiit, ectoderm and entoderm; sh 7 i, S 7 n, splanchnic and somatic mesoderm, inclosing the coelomic cavity; cb, cb, cardioblast cells. Fig. 79. Cross-section through the middle of the proctodaeum of an embryo some¬ what younger than that represented in Fig. 74. x, proctodaeal cavity; cl, coelomic cavity; gl, ganglion; 771st, Mittelstrang; 77 tp^ to 7 npg^, Malpighian vessels sur¬ rounded by mesodermic elements. Fig. 80. Cross-sections through the sexual orifices of an embryo in the stage of Fig. 74. ^(9, ^(3, openings of efferent ducts; remaining references as in preceding figures. Fig. 81. Surface view of two eyes (corresponding to eyes V. and VI. of Acilius as described by Patten). 71, large nucleus in the centre of each eye. Fig. 82. Median sagittal section through an embryo in about the stage represented in Fig. 73. aTTi and sr, amnion and serosa; 0, point at which the anterior end of the embryo passes into the amnion; p, point at which the posterior end passes into the amnion; st, stomodaeal depression; x, point near which the proctodseum will appear; pl'^, pl'^, anterior and posterior thickenings of the inner layer; from which the entoderm, eTit^, eTit^, originate. Tnsd, mesoderm; 77 ix^ and mx‘^, first and second maxillae; p^ to p^, legs; c, cells which entered the amniotic cavity. Under the two thickenings pL, pl'^, are seen the peculiar degenerating nuclei passing into the yolk, which has become segmented. f ^ ? l!' / S'!- i f 1 j: ■ ^ •4 f • * 1 , t ' V ••A >r ’ ^ f r*-.. '•> ri , r I I ■| I ♦ i V t jf: Joiini. Mor/i/i Voi.III. PL XIX ■s*^# ^ €C(i. SR') or .^, 4^.t/\ ■<^$- \ m^d \ ‘vf \ • *Xi» msct ant. L WyMli'eelei.dtl. B.MeiselJith.BostQn. WHEELER, 0 84 DESCRIPTION OF PLATE XX. Doryphoj'a decemlineata. Fig. 83. Portion of cross-section through a basal abdominal segment of an em¬ bryo somewhat younger than that represented in Fig. 74. am, amnion; ent, entoderm; ect, ectoderm; tr, tracheal invagination; gl, ganglion; nist, Mittelstrang; sni, somatic mesoderm; slm, splanchnic mesoderm; cb, cardioblasts. Fig. 84. Transverse section through middle of abdomen of embryo, Fig. 74. gn^ sexual organ; mpg, Malpighian vessel; remaining references same as in Fig.*83. Fig. 85. Cross-section through metathoracic segment of an embryo considerably older than Fig. 74. In, lateral nerve trunk; ap, appendage; sr, serosa; ad, fat cells; a, amnion cells entering the yolk at b ; remaining letters the same as in Fig. 83. Fig. 86. Cross-section through mesothorax of an embryo after the formation of the heart, ht. hsp, thickening of ectoderm which secretes the metathoracic hatching- spine; cP, first cuticle shed by the embryo; cm, longitudinal commissures between two thoracic ganglia; f, furca; r, point of attachment of muscles to the furca; msl, muscular tissue; remaining letters as in preceding figures. Fig. 87. Cross-section through caudal plate of embryo before the closure of the anal end of the gastrula, x. c, cell about to wander through the blastopore into the amniotic cavity; msd, mass of cells which will form the mesoderm as soon as the blas¬ topore closes; ent, cells imperfectly differentiated from those of the mass, msd, which are to give rise to the entoderm; v, yolk nucleus; am, amnion; sr, serosa. Fig. 88. Cross-section, more highly magnified than the preceding, through the caudal plate of an embryo somewhat younger than that of Fig. 73. The blastopore has closed, and the depression x marks the beginning of the proctod^al invagina¬ tion; / is the line, unusually distinct in the section figured, separating the mass of mesoderm cells (msd) above it from the smaller mass of entoderm cells (ent) below it; t, 0, V, degenerating nuclei which pass from the entoderm into the yolk; y, yolk nucleus. • Fig. 89. Frontal section through the cardioblasts (cb) of an embryo in the same stage as Fig. 84; cb, row of cardioblasts; ecd, msd, ent, germ layers; ad, fat cells. Fig. 90. Portion of a cross-section through the lower part of the dorsal yolk of an embryo in the stage of Fig. 74. ch, chorion; sr, serosa; am, amnion; do, amniotic dorsal organ; m, amniotic nuclei degenerating; v, yolk nucleus. Fig. 91. Frontal section through the meso- and meta-thoracic ganglia of the em¬ bryo, Fig. 74. cm, longitudinal commissures; ccm, cross-commissures; fP, anterior nerve trunk; fp, posterior nerve trunk; onrl, outer neurilemma; inrl, inner neu¬ rilemma. r rc .1 I 0 r?5N‘' / r r /• 4 ' I. <• •• I • ■■* ► i \ V»-' Uh • Joiun. Mor/i/z . Vo 1. Ill PL XX sc^' Z' PiPil am msol WM.Wheaiec. dtl. B-Meisellilh.Boston. 386 WHEELER. DESCRIPTION OF PLATE XXI. Doryphora decenilineata. Fig. 92. Sagittal section through stomodaeum {st) of an embryo in the same stage as Fig. 72. ecd, ent, msd, germ layers; lb, labrum. Fig. 93. Cross-section of median dorsal portion of an embryo, just after the union of the ectoderm {ecd') from either side of the body; cb, cardioblasts; ad, fat cells; am, one of the last amnion cells passing into the yolk through the opening between the two lateral halves of the mesenteron. msl, muscular tissue; ent, entoderm; slm, splanchnic mesoderm. Figs. 94 to 104. Sections i, 3 to 8, 10, ii, 13, and 14, through the mesothoracic ganglion of an embryo somewhat younger than that of Fig. 74. cm, longitudinal commissure; cross-commissures; /j, Punktsubstanz; Mittelstrang; ecd e, hypodermis; gbl, ganglioblasts. Fig. 105. Section through fore part of metathoracic ganglion. The Mittelstrang, mst, persists as the furca (subsequently chitinous). gl, ganglion; ps, Punktsubstanz; msl, muscular tissue attached to the furca; ecd, hypodermis. Joiirn. Miu/ih . Vol.JII. PL XXI crrt. JOS. Ur.V.H7ree/cr. dti. B^MeiseJ.lUh. Boston. A ('ONTRIl-iUTlON '!'( 1 0 Insect Embryology ■V AX IXAlXiURAL IMSSERTATIOX FOR THE DEGREE OF DOCTOR OF PHILOSOPHY, PRESENTED TO THE FACULTY OF CLARK UNIVERSITY, A/ay 10^ i8g2. BY WILLIAM MORTON WHEELER. Reprinted from JOURNAL OF Morphology, VoI. VIII., No/ l. BOSTON : GINN <& CONIF»ANY. 1893 . ? 'k ■i ' A CONTRIBUTION Insect Embryology. AN INAUGURAL UISSERTATI(3N FOR THE DEGREE OE DOCTOR OE PHILOSOPHY, PRESENTED TO THE FACULTY OF CLARK UNIVERSITY, May lo, i8g2. BY WILLIAM MORTON WHEELER. Reprinted from Journal of Morphology, Vol. VIIL, No. i. BOSTON: OINN & CONlF*ANY. 1893 . Volume VIIL April, l8gj. Number /. JOURNAL OF MORPHOLOGY. A CONTRIBUTION TO INSECT EMBRYOLOGY. WILLIAM MORTON WHEELER. TABLE OF CONTENTS. Page Introduction . 2 I. The embryonic development of the Locustidas. 3 1. The oviposition of Xiphidium ensiferum, Scud. . 3 2. The formation of the Xiphidium embryo and its backward passage through the yolk . 3 (a) Description of surface changes beginning with the completed blastoderm . 3 (b) The indusium {prceoral organ) in section . 12 3. The development of the embryo from the time of its reaching the dorsal stirface of the yolk to revolution . 18 4. Variations in the development of the indusium . 23 3. The revolution of the embryo . 27 6. The stages intervening between revolution and hatching. . 29 7. The develop 7 nent of Orchelimu 7 n vulgare . 35 II. Gastrulation in the Orthoptera. 36 III. The indusium and its homologues in the Arthropoda. 33 IV. General consideration of the embryonic envelopes and revolution of the insect embryo.-. 59 1. The amnion and serosa . 59 2. The yolk . 64 3. Blastokinesis . 67 4. The elimination of the embryonic envelopes .. 78 2 WHEELER. [VoL. VIIL V. Neurogenesis in the Insecta. 82 1. The nerve-cord . 82 2. The bram . 99 3. General remarks on the nervous system . 108 VI. The development of the reproductive organs in the Insecta. 113 1. The go 7 iads . 113 2. The male ducts . 116 3. The female ducts . 119 4. General considerations . 126 VII. The subcesophageal body in Xiphidium and Blatta . 136 VIII. Technique. 139 IX. Bibliography. 143 Description of the plates. 150-160 The very primitive and synthetic character of the Orthoptera has long been recognized by systematists and comparative anatomists, but the full importance of the group from an embryological standpoint has been but little appreciated, owing to the meagre and fragmentary nature of the observations hitherto published. For this reason I have made the Orthop¬ tera the starting point of my studies, with a view to determining their relations, on the one hand to the Apterygota and on the other to the higher Pterygote orders. Only a portion of the evidence bearing on these relationships is presented in the following paper ; a number of observations on the Malpighian vessels, corpus adiposum, oenocyte-clusters and abdominal ap¬ pendages will be published as separate papers. I have devoted more attention to Xiphidittm than to other Orthoptera, partly because the Locustidae occupy a somewhat central position in the order, and partly because this curious form exhibits in its embryogeny better than any other insect hitherto studied, the co-existence of certain very ancient with very modern characters. My German co-workers in the field of insect development will probably regard my treatment of the literature as rather perfunctory ; but Prof. Graber, Dr. Heider and others have given from time to time such complete resumes of past and current literature that I feel justified in departing from the general custom. If I have failed to give credit where it is due, No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 3 I beg that this may be regarded as a fault of omission and not as a fault of commission. I would express my sincere gratitude to Prof. C. O. Whit¬ man for his kindly guidance and friendly counsel throughout the progress of my work in his laboratory at Clark University during the autumn and winter, and at the Marine Biological Laboratory during the summer months, of 1891 and 1892. I am also indebted to Mr. S. H. Scudder for the identification of several Orthoptera. I. The Embryonic Development of the Locustid^. I. The Oviposition of Xiphidiiim ensiferimt, Scud. Xiphidmm ensiferum, Scudder, a very common Locustid in Wisconsin and the neighboring states, deposits its eggs in the silvery napiform galls produced by Cecidoinyia gitaphaloides (and perhaps allied species) on the low willows that abound in the marshy lands and along small water courses. I have found the insect ovipositing from the middle of August to the middle of September. It thrusts its ensate ovipositor in between the imbricated scales of the gall and places its eggs singly or in a more or less even row with their long axes directed like the long axis of the gall. The eggs are completely concealed by the scales, the overlapping edges of which spring back to their original positions as soon as the ovipositor is withdrawn. The number of eggs deposited in a gall varies greatly: some¬ times but two or three will be found; more frequently from fifty to one hundred; in one small gall I counted 170 and I have opened a few which contained more. Sometimes as many as ten eggs will be found under a single scale; when this is the case, the eggs adhere to one another and are more or less irregularly arranged, as if two or three insects had in succession oviposited in the same place. The Cecidomyia galls vary considerably in shape: some are long and more or less fusiform, others are spheroidal. In the former variety the scales are pointed and flat, while in the latter they are rounded and have their median concave por¬ tions less closely applied to the convex surfaces of the scales 4 WHEELER. [VOL. VIII. which they overlap. These differences materially affect the eggs, for many of those thrust in between the closely appressed scales of the spindle-shaped galls are so much flat¬ tened as to be incapable of developing; on the other hand the eggs deposited in the more spacious interstices of the globular galls are usually in no wise injured. The two forms of gall do not always occur in the same locality and may be the produc¬ tions of two distinct species of Cecidomyia or of one species on different willows. The Locustids, however, seem to show no preference for the globular galls. The galls of Cecidomyia, being essentially stem-galls, do not drop to the ground in the autumn like the various leaf- galls on the willows, but persist through several seasons. Al¬ though the insects are not averse to ovipositing in the fresh galls, they nevertheless seem to prefer these blackened and weather-beaten specimens, probably because their scales are more easily forced apart. I have called attention to the fact (’ 90 ^) that X. ensifenim departs widely in its habits of oviposition from its congeners, several of which are known to lay their eggs in the pith of easily penetrated twigs, like the species of the allied genus Orcheli- immi. X. eiisiferum has evidently found it of great advantage to make use of the galls so abundant in its native haunts. So recent may be the acquisition of this habit, that on further investigation some females may, perhaps, even now be found to have a tendency to oviposit, like Conocephalns ensiger, be¬ tween the root-leaves and stems of plants, or even in the plant tissues. It still occasionally happens that the eggs are run through or into the tissues of the gall-scales, and not loosely deposited. The fact that the insects have not yet learned to distinguish the kind of gall best adapted to their purposes, lends some support to the view that it is not so very long since X. e 7 isiferum agreed with its congeners in habits of ovi¬ position.^ 1 In the vicinity of Worcester, Mass., I found galls very similar to those formed on the Wisconsin willows. They contained a few slender yellow eggs, smaller than those of X. ensiferum. As this species does not occur in New England I conclude that these eggs were probably deposited by the very common X, fascia- turn, De Geer. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 5 I 2. The Formation of the Embryo and its Backward Passage Through the Yolk. a. SURFACE CHANGES. The sub-opaque, cream-colored egg of Xiphidium is elongate - oval, 3-5 mm. long and i mm. broad through its middle. One of its poles is distinctly more attenuate than the other, and there is a faint curvature in the polar axis which causes one side of the egg to be distinctly convex and the other distinctly concave. The broader pole is the posterior, and is the first to leave the vagina during oviposition; the attenuate pole is, therefore, the anterior. In the galls the eggs stand with their attenuate poles pointing upwards. The convex face of the egg is the ventral, the concave face the dorsal region. Inasmuch as the egg undergoes no change in shape during development, it is easy to orient the embryo in its different stages. This is of considerable importance, as will appear from the sequel. The yolk is pale yellow and very similar in constitution to the yolk of other Orthopteran eggs. It is enclosed by a thin leathery chorion which suddenly becomes transparent on immersion in alcohol. When dry it is white, and the creamy color of the egg is due to the yellow yolk shining through. As in Blatta, the chorion is the only envelope of the freshly laid egg ; what I described in a former paper (’90t>) as the vitelline membrane is in reality comparable to a ‘‘Blastodermhaut ” as I shall point out. The chorion varies somewhat in thickness at different points in the egg, being ii/x towards the middle and igfx at the poles. It is quite elastic and when cut curls in at the edges. Its inner surface is very smooth, while outwardly it is covered with round or oval projections which measure about 3.7 /a in diameter. They are flattened at their summits and are placed so closely together that only narrow channels run between them and give the chorion the appearance of being covered with a fine net of nearly uniform meshes. On closer examina¬ tion it is seen that the projections are arranged in hexagonal groups. These are very distinct at either pole but fade away 6 WHEELER. [VOL. VIII. on the median portions of the egg till they become very diffi¬ cult to resolve. They evidently coincide with the areas covered by the polygonal cells of the follicular epithelium. No traces of micropyles could be found. Their absence in Xiphidium is of interest, since Leuckart (’ 55 ) long since described and figured them in several European Locustidae {Meco 7 ievia, Decticiis^ Loctista, Ephippigera). In these genera they consist of funnel-like perforations on the ventral surface of the chorion either near the anterior pole or nearer the middle of the egg. The preblastodermic stages were not studied. They prob¬ ably resemble the corresponding stages of Blatta^ of which I have given a detailed account in a former paper (’ 89 ). When fully formed the Xiphidium blastoderm, like that of Blatta^ consists of a thin sheet of cells, that have in part reached the surface from the interior of the egg, and are in part derived from these centrifugal cells by tangential division after their arrival at the surface. Numerous cells — the future vitellophags — are to be found at different points in the yolk. Whether they are derived from the incompleted blastoderm by centripetal division, or are inhibited before reaching the surface, my limited observations will not permit me to decide. The cells forming the blastoderm are polygonal, much flat¬ tened and of uniform size and distribution. Those on the center of the convex, or ventral face of the egg soon begin to change their dimensions; from being broad and flat, they become more nearly cubical, their lenticular nuclei again assuming the spherical or oval shape which they had in preblastodermic stages. These changes take place over a limited and somewhat oval area and result in the formation of the ventral plate. The few eggs that I have been able to find in the very first stages after the completion of the blastoderm leave me in some doubt as to the exact process whereby the embryo is established. I am satisfied, however, that the thickening and narrowing of the individual blastodermic cells does not take place simultaneously o’^er the whole ventral plate area, but that there appear, as in the crustacean egg {eg. Astacus, Homarus), several discrete centres about which the No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 7 cells are at first more closely aggregated. The spaces between these centres are subsequently filled in by tangential cell- divisions. Of such centres I can distinguish four: two of them, the precursors of the procephalic lobes, are paired, while the other two form respectively the growing caudal end of the ventral plate and what I shall call the indusium.^ The indusial centre, which does not make its appearance till a short time after the other centres are formed, does not join the body of the embryo till after the spaces between the procephalic and caudal centres are filled in. This is distinctly seen in Fig. I (Stage A) where the somewhat T-shaped embryo is already established and distinctly marked off, at least posteriorly, from the undifferentiated blastoderm. The nuclei of the blastoderm are as yet no larger than the nuclei of the ventral plate. Numerous caryokinetic figures in all parts of the embryo bear witness to active cell proliferation. No such figures were to be seen in the extra-embryonal blastoderm during and after this stage. The ventral plate including the indusium is scarcely a fifth as long as the egg, being much smaller in pro¬ portion to the size of the yolk than in some other Orthoptera {Blatta, Gryllotalpa). The blastopore is seen in the stage figured as a very narrow but distinct groove extending from the oral region to the caudal end of the embryo, where it bifurcates before its termination. The infolded cells give rise to the mesoderm and also, I believe, to the entoderm. In Xiphidhim the three folds that form the amnion and serosa arise like their homologues in Blatta. The first appears as a crescentic duplication surrounding the caudal end; thence it grows forward and after enveloping the whole postoral portion of the embryo coalesces with the two head-folds, each of which arises from the edge of a procephalic lobe. The pro¬ gress of the anal fold is shown in Fig. 2 (Stage B) PI. I. Although agreeing in its main features with what has been described for most insect embryos, the process of envelope- ^ In a preliminary note (’ 90 °) this structure was called the praeoral plate (Praoralplatte). Many reasons have led me to abandon this term together with others referring to the parts of the organ in its subsequent development. 8 WHEELER. [VOL. VIII. formation in Xiphidmm^ is, nevertheless, peculiar in two respects: first, the envelopes are so closely applied to the germ-band that in surface view their advancing edges can be detected only with difficulty, though they may be distinctly seen in sections; second, the point of closure of the envelopes is situated further forward on the head than in Blatta^ Hydro- pJiiltis, DorypJiora, etc. This I infer from an embryo, which I figure (Fig. 15. PI. II.) Here the cells and nuclei of the amnion and serosa have become much larger than the cells and nuclei of the embryo. The edges of the folds are unusually distinct and enclose a circular space through which the oral and praeoral regions are clearly visible. On the median anterior edge of the head the amnion and serosa are completely interrupted. In no other insects have I found the envelopes lacking on the anterior edge of the head in so late a stage. This fact is probably significant when taken in con¬ nection with changes about to occur in front of the head. The wide procephalic lobes are succeeded by the strap-shaped body In this a number of segments have made their appear¬ ance. These are in order from before backwards: the mandib¬ ular ((ind. s), the first maxillary {inx. ji), the second maxillary, (p 7 ix. s'^) the three thoracic (/. s^-p. sP), and the first abdominal {a. j*!). Further back lies a small segment which is incom¬ pletely constricted off from the first abdominal and which I take to be the proliferating terminal segment, or telson. The seven segments depicted in the figure are undoubtedly de¬ finitive segments. The manner of their appearance will be clear from a glance at Fig. I. In A the ligulate part of the germ-band is seen to be faintly constricted at its base into two segments with indications of a third. In B, a slightly later stage, four definitive postoral segments are present, but a portion of the germ-band still remains unsegmented. This is, however, soon broken up into segments and we reach the stage in Fig. 15, PI. II. It will be observed that the embryos in Fig. I are in many respects older than that in Fig. 15, PI. II. The antennae have made their appearance and the amniose- rosal fold has closed completely. These embryos prove several points: — first, that the wave of metameric segmen- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 9 tation passes from before backwards dividing the germ-band into 7 or 8 segments; second, that these segments are the definitive segments and not macrosomites, or complexes of definitive segments; and third, that there is considerable varia¬ tion in the time when segmentation sets in. To these points I may add a fourth; segmentation appears first in the ectoderm and only somewhat later in the mesoderm. A and B. Isolated embryos of Xiphidiuni in successive stages of metameriza- tion. ind., indusium ; pel., procephalic lobe ; st., stomodaeum ; at., antenna ; md. s., mandibular segment; nix. s'-, first ; mx. second, maxillary segment; p. prothoracic segment. The indusium (Fig. 15,/. t?.) is still only a rounded thickening of the blastoderm. Its small deep cells are continuous through a zone of larger cells with the relatively very large and flat elements of the primitive cell-layer. Two broad and flat commissures appear to connect the organ with the procephalic lobes. Thus a small space containing a few larger cells is enclosed between the indusium and the head of the embryo. This space (j), seen as a clear spot in surface view, lies at the breach in the envelopes. In many embryos the indusium is lO WHEELER. [VOL. VIII. united with the head of the embryo (Fig. I, A and B) before the stage of Fig. 15 and soon after this stage is, I believe, normally united with it. This union is probably purely mechanical — the organ remaining at its place of origin on the surface of the yolk, while the embryo lengthens till its head unites with the posterior end of the organ. This union is of brief duration as is seen in Fig. 3 (Stage C). During this stage the caudal tip of the embryo shows a tendency to bury itself in the yolk. The amnion and serosa, hitherto closely applied to each other, now separate at the caudal end, where, as I have said, they first arose as a crescentic fold. Soon the tendency to enter the yolk becomes more pronounced so that the tail curls back and leaves the ventral face of the egg. Meanwhile the remainder of the embryo moves down the ventral face a short distance, thus pushing its tail still further into the yolk and causing the separation of the envelopes to advance still further headwards. The indusium does not accompany the embryo in this move¬ ment, but remains nearly or quite stationary ; consequently the head gradually separates from the organ till it is connected only by means of a slender band of cells in the median line. (Fig. 3 and Fig. 16.) This link soon ruptures and the indusium is set adrift from the embryo, or, more precisely, the embryo is set adrift from the indusium. (Fig. 4, Stage D.) In profile the embryo now resembles the small letter j,—the dot being supplied by the isolated indusium. Important changes begin to affect the indusium during or more frequently just after its separation from the embryo. The closely packed cells at the periphery, as indicated by their nuclei, begin to arrange themselves radially (Fig. 16). Some of the large nuclei of the serosa may be seen encroaching on the edges of the disk from all sides, leaving only the median portion free. Sections show that the organ is now forming an amnion like that of an embryo. In the middle of the disk appear several shrunken but distinctly defined nuclei which are proved by focusing to be confined to the surface of the organ.^ 1 Only four of these peculiar bodies are represented in the figure {n7i ); there were several others in the same preparation, but for the sake of clearness I have omitted them in the drawing. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. II The serosal fold continues to advance from all sides till the organ is entirely covered. Viewed from its ventral surface the egg now has the appearance of Fig. 4 (Stage D). Here the indusium is cordate in outline and somewhat larger than usual. Of the abdomen only the two basal segments still remain on the ventral face of the egg ; the remaining seg¬ ments curl back into the yolk. During this and the two preceding stages the cephalic and thoracic appendages have become distinctly established as rounded lateral outgrowth of their respective segments. The antennae (at) originate as lobular outgrowth from the posterior edges of the procephalic lobes. They are distinctly postoral in origin. The margins of the triangular oral orifice are some¬ what swollen; the anterior edge, where the labrum is about to appear, is cleft in the median line (Fig. 16). The three thoracic segments are very slightly or no broader than the two maxillary segments. The appendages of these five segments are also alike in size, shape, and position. In very early stages of other insect embryos, even before the amnion and serosa are fully formed, the thoracic become broader than the maxil¬ lary segments, and the legs, as soon as they appear, may be readily distinguished from the two pairs of maxillae by their greater size and prominence. The Locustid embryo, there¬ fore, has even a stronger tendency to revert to annelid-like or myriopod-like ancestors than is apparent in any of the other insects whose ontogenies have been investigated. The mandibular segment of XipJiiditim like that of other insects, is somewhat retarded in its development. Between this and the antennary segment careful study of sections and surface preparations reveals the presence of another segment, shown very distinctly in outline in Fig. \6 (^tc. s}j. This is no other than what I have called the intercalary segment in Dorypliora. It is the tritocerebrum of Viallanes (’ 90 ^’ ’ 90 ^). The embryo continues to move back into the yolk, fol¬ lowing the curved path established by the inflexion of the posterior segments till its tail is finally arrested by striking the serosa on the dorsal surface. At this time the embryo has the form of an arc subtending the dorsoventral diameter of the egg. 12 WHEELER. [VOL. VIII. Returning to consider the indusium, we find that it begins to increase in size before the embryo’s head leaves the ventral face. The organ stains much less deeply, and even in surface view its expansion may be seen to be due to a flatten¬ ing of its component cells. In Fig. 5 (Stage E) is represented an embryo merged in the yolk up to the first maxillary seg¬ ment. The indusium extends around on either side nearly to the middle of the lateral face of the egg. Either the tran¬ sition of the embryo takes place rapidly or the organ changes very gradually, for the latter is in about the same stage after the embryo has become established on the dorsal surface. The manner in which the expansion of the indusium is brought about will be apparent when I come to describe its structure in sections. b. THE INDUSIUM IN SECTION. As will be seen from the preceding account, the indusium is simply a circular thickening of the blastoderm, situated in the median line, between and a little in front of the procephalic lobes. It does not arise as a part of the ventral plate but as a separate centre which is at first merely a cluster of blastoderm cells that have changed from the pave¬ ment to the cubical or columnar type. This centre is further increased in breadth and thickness by caryokinesis. In the earliest stages examined, sections of the organ show the same cell-structure as sections of the procephalic lobes. Median longitudinal sections of the embryo in Stage C are interesting as showing the relations of the indusium to the embryo and its envelopes. I reproduce such a section in Fig. 21, PI. III. Here the organ (/. 0) appears as a large flattened cell-aggregate somewhat thinner in the centre than nearer its periphery. Owing to the shape of the mass, the median cells, as indicated by their nuclei, are arranged with their long axes perpendicular to the flat outer surface of the organ, while the cells of the thickened lateral portions become gradually more oblique till those on the extreme periphery assume the same position as the serosa cells {s). The nuclei are most frequently No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 13 situated at the inner ends of the cells so that masses of enucleate protoplasm are left at the surface. Posteriorly the organ is linked to the embryo by means of a few flattened cells. In the section two of these cells are seen at z differing in no wise from the serosal elements (i-.) in front and on either side of the organ ; the upper cell passes directly into the serosa covering the embryo, while the lower abuts on the cells that form the transition from the ectoderm to the amnion. The ectodermal layer of the embryo {ec^j is nearly as thick as the indusium and of similar cytological structure. The begin¬ ning of the stomodaeal invagination is shown at 0. The next section figured (Fig. 17 PI. II) is from an indusium in a somewhat younger stage than that represented in surface view in Fig. 2. Being transverse the section shows an evenly convex outer surface, continuous with the surface of the serosa (i-.) enveloping the yolk. The cell-contours are still visible and show that the cells constituting the median portion of the organ are polygonal. The nuclei of these elements are spherical or oval and contain one, or more rarely, two nucleoli besides the usual chromosomes. In the peripheral ring-shaped thickening the cells (<^.) are larger and pyramidal or fusiform in outline, while their nuclei differ in no wise from the nuclei of the median cells. The serosal cells stain more deeply than the cells of the organ, as may be seen at where a single cell overlaps the edge of the disk. This depth of color is appar¬ ently purely optical, being due to the greater size and flatness of the serosal nuclei. The walls of both the small polygonal and larger pyramidal elements fade away towards the surface, where the bodies of the different cells become confluent to form a homogeneous mass. In this surface-mass of protoplasm which takes the normal pink stain in borax carmine, are to be found several of the peculiar nuclei, mentioned above as distinctly discernible from the surface (Fig. 16). They differ markedly in structure and appearance from the normal nuclei in the inner portions of the indusium as will be seen by comparing the cells of Fig. 24 with those in Fig. 23, both of which figures were drawn with a high power. The normal cells (Fig. 23) have spherical or 14 WHEELER. [VOL. VIII. oval, evenly rounded nuclei with one or two nucleoli and their chromatin is distributed in what I take to be the typical resting reticulum. The caryolymph, or Kernsaft, is faintly stainable. On the other hand, in the nuclei of Fig. 24 the nuclear wall is very irregular, the caryolymph much more limpid and refrac¬ tive and the chromatic reticulum has coarser meshes. The chromatic nodes of the reticulum are larger than in Fig. 23 and seem to be applied to the indentations of the nuclear wall. Nucleoli appear to be absent. These specialized nuclei also vary greatly in size. In a series of sections it is easy to find nuclei intermediate between the two extremes here described, being evenly rounded but with colorless caryolymph and coarse chromatic reticulum. A cluster of four such nuclei is shown at mi'^ Fig. 17. These intermediate forms, occurring as they usually do, between the normal and the modified nuclei may be taken to indicate that the nuclei of the extreme types are genetically connected. Some of the normal nuclei probably leave their respective cells in the median portions of the organ and move up into the syncytial protoplasmic layer, undergoing the modification in structure during their emigra¬ tion. When they have reached their destination they are perhaps broken down and converted into protoplasm. Certain it is that later no traces of them are to be found in the indusium. I do not believe that I am here considering collapsed and distorted caryokinetic figures, as these delicate structures are quite faithfully preserved in eggs killed by means of heat. The distorted nuclei are not confined to the indusium but occur also in the ectoderm of the embryo itself. When the organ has reached the state just described it usually separates from the head of the embryo; it may, how¬ ever, remain attached for some time longer. Like the embryo it is now an isolated body lying on the yolk; but unlike the embryo it is still only a part of the serosal envelope (which is itself only the extra-embryonal portion of the blastoderm). The serosa is a closed sack enveloping the whole yolk and the indusium is simply a swelling at one point on its inner face. (Fig. II, A.) The process of envelope formation which now begins in the indusium is much less clear than the cor- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 15 responding process already completed in the embryo. From among.the numerous preparations which I have made I select for illustration one (Fig. 18) which seems to show the process clearly. In surface view the organ would appear as in Fig. 3. The spreading of the serosal cells over the edges of the disk from all sides is now seen to be due to a process of induplication, or folding. The circular fold is, of course, cut in two places in the median transverse section figured. It advances in such a manner as to leave the outer face of the indusium evenly rounded and undisturbed, the upper sur¬ face of the fold usually forming a continuous line both with the outer surface of the serosa and with the median still uncovered portion of the organ. The fold continues to advance from all sides till the layers of which it consists meet and become confluent in essentially the same manner as the folds that form the amniotic and serosal layers over the embryo proper. We now have three layers of cells. (Fig. 19.) The outermost layer, j, is the serosa which has everywhere the same structure and evenly envelops the whole egg, having been separated first from the embryo and now by a similar process also from the indusium (Fig. II, B). The innermost layer consists of the unchanged greater portion of the organ. The median layer, to judge from its component cells, seems to be derived exclusively from cells of the original body of the organ and not from the serosa. This layer is, therefore, like the amnion of the embryo proper, structurally more closely related to the body it envelops than to the serosa. Fig. 18 favors this conclusion, which presupposes that only the outer half of the circular fold is derived from the serosa, for in this section the lower and thicker layer of the fold on either side certainly consists of cells derived from the body of the organ. Even before the layers are fully formed the edges of the two¬ layered organ are sharp and somewhat irregular (Fig. 18), not rounded like the edges of the embryo when its amnion is com¬ pleted. The whole organ still has essentially the same form that it had in the stage represented in Fig. 17. It will be convenient to name the different layers of cells, thus far distinguished. For the amnion of the embryo proper i6 WHEELER. [VOL. VIII. I shall retain the old name; the corresponding envelope of the indusium and the body of the organ will be designated as the outer and inner indusium respectively. In by far the greater number of cases the process of Fig. II. Diagrams illustrating the movements and envelopes of the Xiphidium embryo. A, after the closure of the amnioserosal folds ; B, during the embryo’s passage to the dorsal surface ; C, just after the straightening of the embryo on the dorsal surface, ind., indusium — afterwards forming the inner, and ind'^, the outer indusium ; ch., chorion ; sr., serosa ; am., amnion ; gb., germ-band ; v., yolk ; bl. c., Blastodermhaut. envelope formation over the indusium is much obscured by rapid slurring. In fact the whole process has frequently the appearance of being due rather to a shifting and migration of cells than to the formation of true folds. The cells of the serosa seem to creep over the disk while the cells forming the No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 17 edge of the organ itself appear to creep along under and a little in the rear of the advancing serosal elements. I cannot here go into greater detail without unduly increasing the number of my figures. Nor is it necessary, since it will, I believe, be Fig. II. Diagrams illustrating the movements and envelopes of the Xiphidmm embryo. Z?, the stage of the shortened embryo on the dorsal yolk ; E, embryo returning to the ventral surface ; Z% embryo nearly ready to hatch, ch., chorion ; bl. c., Blasto- dermhaut ; sr., serosa ; ind^, outer indusium ; ind'^, inner indusium ; md^ am., inner indusium and amnion fused ; am., amnion ; ind'^ c., cuticle of the inner indusium ; ind'^ s., granular secretion of inner indusium ; am. s., amniotic secretion ; v., yolk ; cl., columella ; gb., germ-band. acceded that the process briefly described in the above paragraph, though now occurring in comparatively few embryos, is very probably the more primitive process, whereas the slurring observed in so many cases is to be attributed to an unquestionably rudimental condition of the organ. i8 WHEELER. [VoL. VIII. By the time the folds have closed over the indusium the abdomen of the embryo has sunk into^ the yolk to a con¬ siderable extent, presenting in surface view the appearance of Fig. 4. The organ seems to undergo no further change till the embryo has almost left the ventral face of the egg. Then, as we have seen, it begins to increase by spreading. An early stage in this process is shown in section in Fig. 20. No change is perceptible in the serosa, which is now independent of the organ ; the outer indusium (am'^') is much attenuated, as may be seen by comparison with Fig. 19. Its cells have assumed the same shape and dimensions as those of the superjacent serosa ; only along the edges of the disk, where the outer becomes continous with the inner indusium, or body of the organ, do the cells still retain their original shapes. In the body of the organ the cells are arranged in two irreg¬ ular rows, whereas in the previous stage (Fig. 20) there were three. This diminution in the number of cell-rows is the result of horizontal spreading, a process which also accounts for the stretching of the outer indusium as indicated by the flatness of its cells. At nn is seen one of the large modified nuclei, which has persisted unusually late. In Fig. 22 I give a section of the indusium seen in sur¬ face view in Fig. 5. The spreading of the cells has pro¬ gressed till the organ lies like a saddle on the ventral face of the egg, covering nearly half of its circumference. The serosal layer {s) is, of course, unaffected. The outer indusium {ani^) is stretched to such an extent that its cells are united only by an exceedingly thin and in many places, almost imperceptible layer of protoplasm. The inner indusium now consists of a single row of cells, instead of two rows as in the preceding stage. It is in about the same state of tension as the outer layer in Fig. 19. 3. The Development of the Embryo from the Time of its Reaching the Dorsal Yolk to RevoliUion. In the foregoing paragraphs the development of the embryo was traced to Stage E, when the germ-band hangs festoon-like No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 19 in the yolk with its cephalic amnion applied to the ventral serosa and the amnion overlying its terminal abdominal seg¬ ments applied to the serosa covering the dorsal yolk. No sooner has the caudal end become fixed than the head is detached from the ventral face of the egg and the embryo swings back, straightens out, and becomes applied full length to the dorsal serosa. The movements whereby this condition is attained resemble the movements of a leech in passing from one side of a test-tube to the opposite surface; holding fast to the glass by means of the oral sucker, the tail is stretched out till it reaches the opposite surface, when the anterior end is loosened and the body drawn over. There is, however, a marked difference between the embryo and the leech since the body of the former is not contracted during its transition. Fig. 5 represents a rather rare condition in that the pro- cephalic lobes lie at the same level and are symmetrically dis¬ posed with respect to the long axis of the egg. More frequently the germ-band is twisted during its transition so that one of the procephalic lobes reaches further forward than the other on the surface of the yolk. Sometimes it is the left lobe which extends further forward but more frequently it is the right. The twist in the germ-band occurs in the thoracic or abdominal region, more often in the former, the abdomen being nearly straight. I take this twisting of the embryonic axis to indicate that the germ-band executes a screw-like movement while penetrating the yolk, and I believe it to be perfectly normal, having observed it in the majority of embryos. Traces of this twisting are clearly discernible even in embryos which have almost straightened on the dorsal surface. As a consequence of the passage of the embryo through the yolk in the manner above described, the germ-band has shifted its position from the median convex ventral to the median concave dorsal surface of the yolk, so that it is now reversed: originally its head pointed to the tapering anterior pole, now it lies with its head directed towards the blunt posterior pole of the egg. The amnion, of course, accompanies and remains in close contact with the ventral surface of the embryo during all this time. 20 WHEELER. [VoL. VIIL During or more frequently at the close of the embryo’s migra¬ tion the primary serosa secretes from its whole outer surface a thin chitinous cuticle. In my preliminary notes (’ 90 ^’ ’ 90 °) I wrongly designated this cuticle as the vitelline membrane, an error which is, to a certain extent, pardonable, inasmuch as the layer in question is structurally exactly like the vitelline membranes of other insects. But it certainly cannot be homologized with these membranes since it is secreted during a comparatively advanced stage by an embryonic cell-layer, the serosa, and not by the surface protoplasm of the un¬ segmented egg. As soon as the embryo has taken up its position on the dorsal surface, the yolk segments ; each vitellophag appro¬ priating as many of the yolk-bodies as the radiating filaments of its cytoplasm can hold together and fashion into a rounded mass. Apparently the process is delayed in order that the passage of the embryo through the yolk may be facilitated, for obviously the embryo will move more easily over a prescribed path through a mass of small mobile particles than between large masses formed by the aggregation of such particles. The yolk-masses, at first very distinctly marked, soon fuse with one another so that their boundaries can be traced only by reference to their centres, which coincide with the nuclei of the vitellophags. After leaving the ventral face of the egg the embryo in¬ creases greatly in length. Just before burying its tail in the yolk and while still completely on the ventral surface it measured only .7 mm.; now it measures 1.7 mm. This in¬ crease in length, as will be inferred from the foregoing descrip¬ tion, is due to two causes : an intercalation of new segments in front of the anal plate to complete the abdomen, and a stretching of the segments thus arising. A glance at Fig. 6, which represents an embryo in the stage of its greatest elongation on the dorsal surface, shows that many important changes have taken place since it left the ventral surface. The cephalic and thoracic appendages have assumed a more definite character. The labrum {lb}) has sud¬ denly appeared, the first and second maxillae mx^') have No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 21 each become trilobed, while the metathoracic leg (/ 3 ) already exhibits unmistakable traces of its characteristic thickening in the larva and imago. The pleuropodia (//. stand out clearly from the edges of the first abdominal segment. Shining through the stretched ectodermal layer of the abdominal seg¬ ments may be seen the paired mesodermal somites {coe}j^ or mesomeres. The anal plate with its pair of cerci (<:<:. > and the anus are definitely established. A faint neural furrow runs from the mouth to the anus, and in the thoracic region faint metameric indications of the ganglia are apparent. All these important changes have taken place within the yolk during the transition of the embryo. This renders their study on hardened material very difficult, for although the embryo may be dissected away from the yolk, it is so much curved that it can be mounted only in pieces, and the yolk is at this period so difficult to cut that only fragmentary series of sections can be obtained. One of the most interesting changes undergone while the embryo is still in the yolk is the appearance of the labrum. In Fig. 6 (Stage F) the labrum is a distinctly unpaired circular appendage. But that it has a paired origin I infer from a transverse section, part of which is represented in Fig. 35. This passes just in front of the mouth of an embryo but little older than Stage E. The appendage {Ib^j is here seen to be distinctly bilobed although it does not yet project beyond the general level of the head. This bilateral condition is speedily slurred over and the organ grows into an unpaired and in most embryos perfectly circular disk overhanging the mouth. Very rarely, as in Fig. 7 it may show traces of its paired origin even during later stages. Let us return to the indusium which we left as a thin round plate gradually spreading over the yolk just beneath the ventral serosa. The outlines of this plate are not always cir¬ cular but exhibit traces of lobulation (Fig. 5). The spreading is at first uniform along its whole circumference so that the organ soon assumes the shape of a circular scroll clasping the egg. Its lateral edges approximate on the dorsal surface just over the ventral face of the embryo but are temporarily arrested 22 WHEELER. [VOL. VIII. in their growth before they unite. The anterior and posterior edges, however, continue to advance without interruption, so that the disk if spread out on a plane surface would in its suc¬ cessive stages represent a series of ellipses with constant short axis but continually increasing longitudinal axis. In this manner the disk grows towards either pole while envelop¬ ing the egg laterally. The edges of the organ continue to approximate on the dorsal surface but stop growing just before they meet. Hence, when the egg is viewed from the dorsal surface a long, narrow slit is seen extending nearly its entire length and separating the dorsal edges of the organ. It is not till the anterior and posterior edges have nearly or quite reached their respective poles that this slit closes with the fusion of the edges of the organ. The raphe is at first so weak that the edges may be broken apart by slight pressure with the needles, but it soon becomes permanent and the egg is now completely enveloped by two further membranes — the' inner and outer indusia. Before the fusion of these two mem¬ branes the amnion of the embryo was in contact with the serosa but now that the edges of the indusia have worked their way in between the serosa and amnion, the latter comes to lie in contact with the inner indusium. Henceforth the serosa is excluded from taking any part in the development of the embryo; both its position and function are now usurped by the inner indusium. One is enabled to follow the different stages in the progress of the indusium, from its disk-like condition on the ventral yolk to the complete union of its dorsad-growing edges, by means of a peculiar secretion of its inner layer. This is a brownish or blackish granular substance, probably some urate, which appears to be secreted by all the cells of the inner indusium and which gives the organ the appearance of a large brown blotch in a stage a little older than E. At first pale and hardly perceptible, this spot gradually deepens in color till its advancing edges become distinctly outlined on the underly¬ ing yolk. A clear idea of the closure of the edges may be obtained from Fig. HI, A-C. The dark granular secretion is shown in Fig. 6 at envl. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 23 Soon after the union of the edges of the outer and inner indusial layers a chitinious cuticle is secreted by the outer surface of the latter. This cuticle is thicker and seems to be of a deeper hue than the cuticle secreted by the serosa. It Fig. III. Two stages in the spreading of the indusium. A, lateral view of egg just after the arrival of the embryo on the dorsal yolk ; lateral view of the egg with the indusium nearly reaching the poles ; C, same egg seen from the dorsal surface. definitely excludes the outer indusium from any further share in the development of the embryo. Even in Stage E, this cell- layer was reduced to an exceedingly thin membrane. (PI. Ill, Fig. 22, am.^) It seems to fuse with the serosa and to retain a connection with the inner indusium only at the ex- 24 WHEELER. [VOL. VIII. treme anterior pole of the egg. I confess, however, that my observations on this envelope are rather unsatisfactory. After the completion of the processes described in the preceding paragraphs we may distinguish several envelopes in a median transverse section of the egg. Passing from with¬ out inwards we have (i) the chorion, (2) the Blastodermhaut- like cuticle secreted by the serosa, (3) the serosa, (4) the outer indusium, (5) the layer of dark, granular secretion, (6) the cuticle secreted by the inner indusium, (7) the inner indusium and (8) the amnion. While envelopes 1-7 invest the whole egg, layer 8, the amnion, covers only the embryo. The general development of the embryo has been traced to Stage F, when it lies as a straight and attenuated body on the dorsal yolk with its head directed towards the caudal and its tail towards the cephalic pole of the egg. Like all other insects that have a stage during which the body is greatly elongated (Coleoptera, Diptera, Lepidoptera) Xiphidium passes into a series of stages during which the germ-band is gradually shortened. The shortening is accom¬ panied by a broadening of all the segments, a growth of the appendages, and very important internal changes. The com¬ pletion of this process is reached in Stage G (Fig. 7). Besides a greater development of the appendages seen in Stage F, Fig. 7 also shows that the abdominal appendages have appeared. Of these there are nine pairs, exclusive of the pleuropodia and cerci, so that in Xiphidium, just as in Blatta and many other insects, every segment of the abdomen bears a pair of appendages. Starting with the basal segment there are eight pairs of stigmata. These are not all seen in the figure. Just back of each pair of tracheal invaginations appears a second pair of ingrowths — the metastigmatic depressions — seen as small white spots just outside the appendages, near the pos¬ terior edges of their respective segments. They are in line (homostichous) with the tracheal invaginations which occupy corresponding positions near the anterior edges of their respective segments. The ventral flexure of the abdomen constitutes another very important difference between Stages G and F. In Xiphidmm No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 25 this flexure always takes place between the 7th and 8th seg¬ ments and is brought about during the shortening of the embryo. It is essentially the same flexure which is found in Blatta and in Decapod Crustacea. In Stage G the antennae have increased to nearly one-third the length of the embryo. The procephalic lobes on which the segmentation of the brain is plainly visible, have developed greatly. The appendages, instead of projecting laterally, as they do in the younger embryo, are folded over the ventral surface of the germ-band. The nerve cord is distinctly marked out. (See abdominal region. Fig. 7.) It is in this stage, or one but slightly more advanced, that the embryo passes the winter. Cleavage and the succeeding stages up to F are passed within a month after oviposition — during the warm days of August and September. But even should October and November be mild and sunny, development seems to have come to a temporary standstill on reaching Stage G. Among the hundreds of embryos which I collected during three succeeding autumns, I did not find one that had passed far beyond this stage. Nevertheless if kept in a warm, moist atmosphere during winter, a certain number of eggs will continue their development almost to hatching. Before passing on to later stages in the development I will here give a brief account of some anomalies in the development of the indusium. 4. Variations in the Development of the Indusium. In the preceding pages I have described what I take to be the normal development of the indusium of Xiphidium. A considerable number of embryos (about 100), being nearly one half of the total number examined for the stages thus far described, deviated more or less widely in so far as the in¬ dusium was concerned from what I consider the normal type of development. Unfortunately I did not discover the organ till it was too late in the season to obtain a large supply of material in the requisite stages, so that the variations here briefly noticed probably represent only a small fraction of those 26 WHEELER. [VOL. VIII. observable in a large number of eggs. The variations may be tabulated thus : — 1. Variations in size. Normally the indusium is of the same size as one of the procephalic lobes (.2 mm. in diameter) so that the head of the embryo resembles a clover leaf as long as the organ is attached to it. When the chorion is removed the organ may be distinctly seen with the unaided eye as a milk-white spot on the translucent yolk. Occasionally, how¬ ever, embryos will be found in which it is less than .1 mm. in diameter, and all variations between this and the normal size ipay be observed. 2. Variations from the typical circular form. These varia¬ tions are very numerous and may be regarded as belonging to two classes. In one class the indusium is rounded in outline, while in the other it is ragged and more or less irregular. To the first class may be assigned the oval, cordate and multilobulate varieties not infrequently observed; to the second belong a number of irregularly stellate and rhizopod-like forms. In one of my preparations, midway between the two classes, the indusium is evenly rounded anteriorly and ragged poste¬ riorly along that portion of its periphery which has just broken away from the head of the embryo. 3. There is a variation in the time at which the organ is set free from the head. This cannot be proved directly by observation of the organ itself, for it usually does not begin to form the circular fold till after its isolation, but differences in the embryo, especially in the prominence of the segments and appendages, show that the organ remains at¬ tached to the head in some cases longer than in others. 4. Variations in the development of the circular fold. These variations, alluded to above, are characterized by a greater or less distinctness in the folds that give rise to the inner and outer layers. All shades in the process may be found between the distinct and comparatively rare method described and figured (Fig. 3), and the more frequent and obscurer method whereby the three layers are formed by a shifting of the individual cells. 5. Variations in number. I have twice observed two indusia No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 27 in the same egg. In the first case the embryo itself was in every way normal, and the first indusium of the normal size and shape, and in the usual position. The second, somewhat smaller, though regularly circular organ, was situated in front of the first and a little to the right of the median line. The distance between the two organs was about double the distance between the first organ and the head of the embryo. The outlines of the second or more anterior organ were less definite than those of the first. The amnion and serosa had formed over the embryo, but neither of the indusia showed as yet any tendency to form envelopes. Whether these two organs were derived from the division of one original praeoral cluster of cells, or were originally established as two separate centres on the blastoderm, I am unable to decide. The latter method would seem to be the more probable. The other case is somewhat singular. The first indusium was normal in size and position, but irregularly heptagonal in outline. The second, situated a short distance to the side of the right procephalic lobe, was not more than a third the size of the first Organ and quite regularly quadrangular. The embryo itself was normal and covered with the amnion and serosa. The envelopes had also formed over the two organs, which in this case also probably originated from two discrete centres in the blastoderm. The smaller organ had probably never been attached to the head of the embryo. 5. The Revohttion of the Embryo During the first warm days of spring the Xiphidiufu embryo resumes its development. This is characterized for some time by a growth of the germ-band in breadth and length and a lengthening of the appendages. The body of the embryo, which in Stages F and G was much narrower than the egg now becomes almost as broad so that its pleural edges embrace the yolk. This increase in size brings the head somewhat nearer the lower pole, and there soon sets in a decided move¬ ment of the whole body in this direction. When the head has almost reached the lower pole, the amnion covering the face 28 WHEELER. [VOL. VIII. of the cephalic end fuses with the overlying inner indusium. A rent appears in this fused portion of the envelopes and through it the head is soon seen protruding. Gradually more of the body is pushed through the orifice, first the mouth parts, then the thoracic legs and finally the abdominal segments, till the whole embryo comes to lie free on the surface of the yolk in the space between the inner indusium and its cuticle. The amnion and inner indusium, which during the evagination of the embryo have remained united at the edges of the rent are folded over the pleural region of the embryo onto the yolk. The two envelopes now form but a single layer enclosing the yolk like a bag. The inner indusium is united to the edges of the amnion and these in turn are united to the pleural edges of the embryo, with the ectoderm of which the amniotic cells are continuous. The small size of the amniotic cells as compared with the huge flattened elements of the inner indusium enables one readily to distinguish the limits of the two envelopes. During its evagination from the cavity of the amnion the embryo gradually passes around the lower pole of the egg head first and begins to ascend the convex ventral surface. An embryo freed from all its envelopes except the two that take part in revolution is represented in Fig. 8, in the very act of turning the lower pole. The amnion and inner indusium are folded back over the yolk, the former {ant) characterized by its small rounded nuclei, the latter {sr^j by its large flat elements. The line of juncture of the amnion with the body of the embryo is marked by a denser aggregation of nuclei. The ventral flexure still persists on the dorsal surface. The cavity of the amnion contains a quantity of serum-like liquid, which during the evagination of the embryo is poured into the space separating the inner indusium from its cuticle. This liquid collecting at the lower pole, may function as a lubricant and cushion, and thus facilitate the movements of the germ-band. In hardened specimens it is found as a gran¬ ular magma enveloping the appendages. It is not shown in Fig. 8. In many respects the embryo in Stage H has advanced con¬ siderably beyond that represented in Fig. 7. In the head, the No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 29 eye is distinctly marked out and its cells are arranging them¬ selves to form the ommatidia, as is evident from the regular series of pale dots. The labrum, now considerably enlarged, is spade-shaped in ventral aspect. The antennae have grown in length, and the saltatory legs (/ 3 ) are assuming their defin¬ itive characters. The large tapering pleuropodia stand out prominently on the first abdominal segment. Near the bases of the legs the thoracic stigmata are distinctly seen. They had made their appearance in Stage G, but for obvious reasons could not be shown in the figure. The anterior end of the embryo continues to move up the ventral surface of the egg, straightening out as it rises. Finally the flexed terminal segments of the abdomen are again bent back to their original position in line with the rest of the body. Since their flexure these segments (the 8th- iith) have been the only portion of the body provided with a completed dorsal wall {vide Fig. 7). After the bending back of the abdominal tip its segments still retain a certain inde¬ pendence and make no attempt to embrace the yolk of the posterior pole as do the segments in front of them. It is for this reason that the abdomen presents a constriction just in front of the eighth segment. This constriction is especially noticeable in profile view. The turning of the lower pole of the egg seems to take place very rapidly compared with other equally important processes of development, such as the passage of the embryo through the yolk. I infer this from the relative scarcity of embryos in the act of returning to the ventral surface. I have, however, succeeded in finding all the stages in the process of revolution, and feel quite as confident of having correctly interpreted my preparations as if I had studied the living egg. 6. The Stages Intervening between Revolution and Hatching. Fig. 9 represents an embryo that has just straightened out on the ventral surface of the yolk, which the reader may imagine as extending up beyond the head to nearly twice the 30 WHEELER. [VoL. VIIL length of the embryo and terminating in the pointed anterior pole. A comparison of Figs. 8 and 9 shows that, although the former embryo has completed its revolution, it is neverthe¬ less in an earlier stage so far as the development of its organs is concerned. This is particularly noticeable in the labrum, antennae and mouth parts, the eyes and the saltatory legs. Hence we may infer that the time for turning the lower pole is subject to considerable variation. In Fig. 9 it will be observed that many of the abdominal appendages have disappeared. Pairs are, however, retained on the 8th to nth segments (ap^-cc, The pleuropodia are also still present though concealed behind the bases of the metathoracic legs. The disappearance of the appendages on the 2d-8th segments probably has its immediate mechanical cause in the lateral stretching which characterizes these seg¬ ments in their attempts to embrace the yolk. The embryo continues its growth as before in two directions — the body constantly lengthening and thus bringing the head nearer the pointed anterior pole, while its lateral walls, envelop¬ ing more and more of the yolk, gradually grow towards each other and finally unite in the median dorsal line. The union begins with the 7th abdominal segment, just in front of the seg¬ ments which have for some time been provided with a dorsal wall, and continues headward. I am not certain as to what becomes of the amnion during this process. Its cells appear to take no part in the formation of the dorsal wall, but very probably degenerate and become supplanted by the cells of the advancing ectoderm. It must be remembered that a hard and fast line cannot be drawn between the amnion and the pleural ectoderm ; the cells of both structures passing into one another by insensible gradations. My reasons for supposing that the amnion proper takes no part in building up the embryo are mainly of a theoretical nature and will be given in the latter part of this paper. Concerning the fate of the inner indusium there can be little doubt. While the embryo is continually advancing towards the cephalic pole and enclosing more and more of the yolk — this envelope, which, as above stated, is characterized by No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 31 huge flat cells and nuclei, is being as gradually restricted to a more and more limited yolk surface. In consequence of this restriction its component cells become broader radially and narrower tangentially. In this stage the envelope functionally corresponds to the “dorsal organ” of other insects. It cannot, however, be thus designated without still further increasing the number of heterogeneous structures included under that unfortunate term, since the “dorsal organ” of other insects is a thickening of an envelope represented in Xiphidium by the serosa. The thickened inner indusium is soon reduced to a cap of cells on the anterior pointed pole of the egg. As the head of the embryo advances to cover more of this pole, the envelope is pushed further forward and finally stripped from the yolk altogether. The anterior cranial walls then close over the pole and thus effectually separate the yolk from the inner indusium. The latter is reduced to a small conical mass, the cells of which soon show unmistakable signs of degeneration. Soon after the embryo has thus rid itself of its envelopes and has taken into its mesenteron the whole mass of yolk not utilized in the processes of development hitherto undergone, a chitinous cuticle is shed from its entire surface. This may be designated as the first larval cuticle. It appears first on the ventral abdominal surface and spreads thence headward and dorsad. The progress of cuticularization is readily traceable by staining embryos in this stage, for the parts over which the cuticle is formed will not take the color ; where it is being deposited the stain takes faintly and where it has not yet appeared, the stain, of course, penetrates easily. Ayers (’ 84 ) observed in CEcanthus that the secretion of the cuticle began on the ventral surface of the embryo and extended dorsad. This is just what we should expect from the fact that the dorsal hypodermis is ontogenetically a more recent forma¬ tion than that of the ventral surface. The first larval cuticle is about 5 /x thick and consists of three layers. The innermost is apparently homogeneous and stains deeply in Orth’s lithium carmine while the middle layer remains clear and vitreous. The outer layer is radially striated 32 WHEELER. [VOL. VIII. and has the distinctly yellow tint of old chitin. Its outer sur¬ face is minutely papillate. On the appendages the cuticle is much thinner than it is on the trunk and though it stains it does not show a differentiation into three layers. Before shedding the first cuticle the hypodermis secretes a second larval skin which persists till after hatching. In Fig. IV, I have attempted to represent semi-diagrammatic- ally the condition of the envelopes at a time when the eyes begin to acquire pigment. The chorion {ch^i is much distended and the egg larger and more resistent to the touch then it was during the autumn. Passing from without inward we first meet with the cuticle secreted by the serosa {sr. c). Then follows the serosa itself (sr^ to the inner face of which the remains of the outer indusium {ind.^) are applied. At the ex¬ treme anterior end of the egg both these cellular envelopes appear to be much thickened and pass into a cylindrical pedicel of granular plasma which I shall call the columella {cl). This in turn is continuous with a conical mass of cells the re¬ mains of the inner indusium which was stripped from the head in a preceding stage. Its cells, as shown in the figure, are in an advanced stage of disintegration. The cytoplasm of the different elements is reduced to a mass of granules and the chromosomes have become agglomerated into little spheres floating in the clear nuclear plasma. The process of degeneration is similar to that which I have described as occurring in the ‘‘dorsal organ” of Blatta. Between the mass of degenerating cells and the head of the embryo lies a granular coagulum {am. s). This I take to be the amniotic serum which is forced up into the anterior pole by the enlarging of the embryo and the consequent decrease in the space between the body walls and the chorion. The columella and the remains of the inner indu¬ sium are held together and thus temporarily prevented from complete disintegration by the thick cuticle of the latter. This cuticle still envelops the embryo and extends forward to the anterior pole where it seems to be attached to the inner face of the outer indusium. Passing further inward we next meet with the first larval cuticle {Iv. which has been shed, and the second larval cuticle {Iv. c^)y which is still in organic No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 33 connection with the hypodermis. In a little later stage than the one here described the columella and the conical lump of inner indusial elements have disintegrated, and can no longer be distinguished from the granular amniotic serum. The changes in the configuration of the embryo since its arrival on the ventral yolk, relate mostly to the appendages. At first the antennae are of about the same thickness as the Sagittal section through the anterior pole of a Xiphidium embryo, with pig¬ mented eyes, ch., chorion ; cL, columella ; sr. c., Blastodermhaut; sr., serosa ; ind^ + remains of the inner indusium and amnion ; md^, outer indusium ; ind'^ j-., secretion of the inner indusium ; am. s., amniotic secretion ; Iv. d, first larval cuticle ; Iv. d, second larval cuticle ; br., brain ; e., eye. legs. The dark line running parallel with their inner edges, and distinctly marked in Fig. 9, is in section seen to be a meso¬ dermal partition dividing the cavity of the appendage into two tubular sinuses. The antennae grow directly tailward till their tips reach the femorotibial joint of the hind legs, when they di¬ verge laterally, describe an arc, and then grow forward. When the tips have reached the head further progress is arrested 34 WHEELER. [VOL. VIII. by the envelopes, but as the growth of the appendages does not cease, the arcs surrounding the hind legs gradually move tailward. This movement is arrested just before the time for hatching, when the antennae have grown to nearly twice the length of the embryo. The mouth-parts and thoracic appendages have been gradu¬ ally assuming their adult characters in the meantime. The pleuropodia, as described in a former paper (’ 90 ®), are shed during hatching and just previous to that event may be found attached to the pleural cuticle by means of very slender pedicels. In the male the appendages of the 9th and nth abdom¬ inal segments persist, the former as the stylets, the latter as the cerci. In the female the cerci also persist but together with them also the pairs on the 8th, 9th and loth segments (Figs. 9 and 10 — op'^ (ap^) — op 3 (ap^°). These are converted into the gonapophyses. Apart from the eyes little pigment is developed in the hypo- dermis during embryonic life, unless we regard as such the brown granular secretion of the inner indusium. A number of eggs kept in the house the greater part of the winter hatched May I5th-i8th, but I am inclined to believe that out of doors the regular time for hatching is later, prob¬ ably not till the end of May. Xiphidium fas datum apparently does not hatch till early in June, since I found larvae of this species on Naushon Island June 9, which could not have been more than a few days old. Inasmuch as the imagines of Xiphi- ditim ensiferum oviposit on the average about Sept, ist, the whole postembryonic development cannot occupy more than three months. As this Locustid is monogoneutic, nine months is therefore required for embryonic development. Even if we deduct the period of quiescence due to cold weather, it will still be apparent that the embryonic stages must succeed one another very slowly in Xiphidium as com¬ pared with those of other Ametabola {e.g. Blatta), not to mention the Metabola. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 7. The Development of Orchelimnm vulgare. This Locustid oviposits like many of the smaller members of the family in the pith of dead plants. I found the eggs in Ohio during the last days of September in the stems of the wild lettuce {Lactuca canadensis), so common along the edges of fields and thickets, and in the petioles of the common elder {SambuctLS canadensis). Oviposition probably takes place in the beginning or towards the middle of September. In the case of Lactuca and a few other plants which I did not identify, the insects had invariably selected for oviposition the main stem of the flower-panicles. From base to apex this portion of the stem was punctured at intervals, and a single egg thrust into the pith a short distance above each orifice. It is an easy matter to recognize the punctures by the little tufts which the insect evidently gnaws from the woody fibre, before inserting its scimeter-shaped ovipositor. Great care must be taken in splitting the stem, so as not to tear or cut the eggs which adhere very firmly to the pith. The eggs are larger than those of Xiphidiimi ensifencm, being fully 6.-6.2 5 mm. long. In shape they are very similar to Xiphidiimi eggs except that the sides are compressed. In the fresh state they are smooth and opaque, and of a pale drab or bluish tint. In this respect, as also in the flattening of their lateral faces, they form a transition to the eggs of our larger Locustidae, e. g. Cyrtophyllus concavus, Amblycorypha uhlerii and Microcentrum retinei'vis} The chorion is not readily wetted with water, but like that of the Xiphidium egg, immediately becomes transparent when immersed in alcohol. The outer envelope is then seen to have a yellow tint, deepening into brown at the poles. As would be expected from its close systematic affinity the embryonic development of Orchelimum does not differ much from that of Xiphidium. I have not seen all the stages, nor have I, as yet, sectioned any of my material, but the stages which I have examined are essentially the same as those 1 For a description of the eggs of these species see an article on Orthoptera, by Prof.'C. V. Riley, in the Standard NaUiral History, Vol. II. pp. 188-189. I 36 WHEELER, [VoL. VIII. described in XipJiidmm. The embryo of Orchelimnm passes through the yolk in the same manner as the Xiphiditim embryo, shortens on the dorsal yolk, then grows apace, moves around the lower pole and finally begins the yolk-enveloping process on the ventral surface of the egg in the same way as the XipJiidmm embryo. It also develops an indusium which is set free from the head and spreads over the yolk while the embryo is passing through it backwards. In OrcJiel- imum the inner indusial layer also secretes a brownish pigment¬ like substance which enables one to follow its movements as it gradually covers more and more of the yolk. A clear slit is likewise left on the dorsal surface between the folds of the organ. But in the time of closure of this slit OrcJielimuin differs from XipJiidhiin. In the latter insect we found that the slit closed soon after the embryo had straightened on the dorsal yolk, before it had shortened very decidedly. In Ordielimum the closure is considerably delayed. The embryo shortens, then grows in length and breadth, passing beyond Stage G of Xiphidium and its head nearly reaches the lower pole before the two folds of the indusium meet and fuse. Frequently in this stage, when the embryo is about to revolve, the polar ends of the slit are still open, the membranes having fused over the embryo. In a little later stage, however, the indusium has completely enveloped the yolk. II. Remarks on Gastrulation in the Orthoptera. Although many important observations have of late been contributed to the embryology of the Insecta, our knowledge of the formation of the germ-layers in the Orthoptera cannot be said to have made any signal advance. As late as 1889 so few forms of this order had been studied that I felt justified in expressing some doubt as to whether their mesentoderm was formed in the same manner as in the higher Metabola (Cole- optera, Diptera, Lepidoptera). My doubts were confirmed by a study of Blatta, when I failed to find the oral formative centre of the entoderm (’ 89 ^).^ 1 We need not go far to seek the reasons for this gap in our comparative studies. The eggs of the Orthoptera are almost without exception extremely No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. Bruce (’ 86 ) appears to have been the first to describe the origin of the mesentoderm from a median ingrowth of the germ-band in the Orthoptera. The species which he studied, is, I have every reason to believe, Stagniomantis Carolina. His description is very meagre and his figures are unsatis¬ factory. More convincing are Graber’s figures and description of mesentoderm formation in Stenobothriis ‘variabilis (’88, PI. XIV, Fig. II ; PI. XV, Fig. 13). His Fig. ii shows that there is in the median line a distinct infolding of the ventral plate cells — a true invagination. In a more recent paper (’9o), the account is briefly repeated without any important additions. In his recent study of the embryogeny of Blatta germanicay Cholodkowsky (’ 91 ®) gives an account of the formation of the germ-layers more in harmony with what we know of the process in the Coleoptera than the account which I gave. But he has not come to any definite conclusion respecting the formation of the entoderm, and although he maintains that there is a distinct blastoporic groove running the length of the germ-band, he does not figure it in surface view, and most of his sections betray such an amount of distortion in his preparations that one may hesitate to regard the slight depressions in his figures (Figs. 7, 8, 10, etc.) as indicating invagination. Nevertheless I believe from renewed study of the Orthoptera, that Cholod¬ kowsky is correct in deriving the mesoderm from a median proliferation of the primitively one-layered germ-band, and the entoderm from two formative centres — one in the oral and one in the anal region. In Xiphiditmiy soon after its first appearance, the blastoporic depression, when seen from the surface (Fig. i), is a straight refractory from a technical point of view. The cells of the embryo are often smaller and less distinct than they are in the Metabola. Moreover, the great quantity of yolk and its singular brittleness in hardened specimens renders paraffin sectioning most unsatisfactory, and rather than incur the great expend¬ iture of time which working with celloidin involves, the student gladly selects some Coleopteran or Dipteran egg which is all that can be demanded from a purely technical point of view. Nevertheless the Orthoptera constitute, by com¬ mon consent, one of the most primitive orders of the Insecta ; their eggs are large and may be readily procured in great numbers; their development is so gradual that all the requisite stages may be obtained without the least difficulty. 38 WHEELER. [VoL. VIIL groove extending nearly the entire length of the germ-band and dividing it into two symmetrical halves. Anteriorly the groove is rounded and seems to end rather abruptly, but posteriorly it bifurcates, each of the two grooves thus arising being continued for a short distance to either side till they gradually fade away. There can be no doubt, it seems to me, that the bifurcated termination of the blastopore is the homo- logue of the similar structure first figured by me in DorypJiora (^ 89 , PI. XVIII, Fig. 71; PI. XIX, Fig. 73) and subsequently seen by Graber (’ 9 o) in the corresponding stages of Lina tremidcE (PI. II, Figs. 25 and 27). More recently Cholodkowsky has observed a similar widening of the blastopore in Blatta. Fie attempts to identify it with the posterior depressions of Graber’s ‘Gateral gastrulation.” In Stage B (Fig. 2) when the caudal amnio-serosal fold has covered the ligulate portion of the germ-band, the blasto¬ pore presents a widening of its anterior end at a point which probably lies just in front of the definitive mouth. This widen¬ ing was observed in several embryos, and I therefore take it to be a normal occurrence. It also has its homologue in the Doryphora embryo (see my Fig. 70, PI. XVIII, ’ 89 ). In the stage under consideration (Fig. 2) the anal bifurcation has grown more indistinct and is apparently about to disappear. The closure of the blastopore proceeds simultaneously in two directions: from its anterior end backwards, and from its posterior end forwards, so that the last portion of the groove to disappear lies in that part of the germ-band which is to become the thoracic or baso-abdominal region. In sections the groove is seen to be much shallower than it appears in surface view. Along its whole extent its floor is somewhat thickened and in this portion — destined to form the mesentoderm — the cells soon lose their columnar shapes and become more polygonal in outline and more irregular in their arrangement. The groove closes in such a way that no tubular cavity results as in the Coleoptera and Diptera; the cells at the edges of the depression appear to glide over the median elements, so that after the fusion of the edges in the median line the mesentoderm has the form of a solid cord No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 39 applied to the inner surface of the germ-band. The process whereby the inner layers are formed is, therefore, a slurred invagination. In this respect Xiphidium resembles Blatta. The further differentiation of the mesentoderm is quite as difficult to follow in Xiphidium as in other Orthoptera. In these stages the embryo cannot be satisfactorily isolated from the yolk and sectioned by itself, and so friable is the yolk that it is almost impossible to obtain thin sections through the entire egg by the ordinary methods. After studying a few series of sections obtained by means of the celloidin method I can, how¬ ever, affirm that the invaginated cells give rise to both entoderm and mesoderm. The former has a bipolar origin, as has been made out in the higher forms; in Apis by Grassi (’ 84 ); in Hydrophilus by Heider (’ 89 ); in Dojyphora by myself (’ 89 ); in Musca by Voeltzkow (’ 89 ) and by Graber (’ 89 ); and in Chalicodovia by Carriere (’ 90 ). The anal is considerably larger than the oral formative centre and its elements seem to arise in part from the bifurcation and in part from the deeper portion of the blastopore just in front of the bifurcation. In Xiphidium, just as in the higher Metabola, a pair of entoderm-bands grows towards the baso-abdominal region from either entoderm-pole. Each band, consisting of only one layer of much-flattened cells, meets that of its respective side and then begins to envelop the yolk by proliferation at its ventral and dorsal edges. Transverse sections show that at first the bands are only two or three cells in breadth and that these are closely applied to the dorsal faces of the mesomeres which are formed by this time. I have made no observations on the relations of the procto- daeum to the posterior end of the blastopore, but in regard to the anterior end and its relation to the stomodaeum my results are more definite. Figs. 32-34 represent three successive sections through the head of an embryo in Stage D. The last section (Fig. 34) passes through the stomodaeum {sti) which is just forming as a rounded depression in the cephalic ectoderm. Its large columnar cells are regularly arranged and have their nuclei in the inner ends. The next section (Fig. 33) passes just in front of the stomodaeiim and cuts two masses of cells in 40 WHEELER. [VOL. VIII. the median line. The upper of these masses is a thickening of the ectoderm distinctly separated on either side from the elements of the same layer by the peculiar character of its cells. These are much smaller than those of the remaining ectoderm and stain more deeply, especially in the inner portions of the layer. The lower mass of cells is entirely cut off from the ectodermal thickening, though its elements are very similar in size and staining qualities. The ectodermal thickening marks the point where the paired labrum is about to appear {cf. Fig. 35). In the next section (Fig. 32), which also passes through the labral region, we again meet with the thickening of the ectoderm. Unlike its portion in the pre¬ ceding section, it is not bounded below by a curved line, but juts in as a ragged mass of cells, in which it is possible to distinguish a pair of lateral wings and a median projection. The median portion thus proliferated beyond the limits of the ectoderm, is the anterior or oral entoderm centre — the lateral wings I regard as mesodermal. By combining Figs. 32 and 33 the flattened mass of cells underlying the ectoderm in the latter section is seen to be the backward continuation of the mesentoderm. Section Fig. 34 shows that this median unpaired mass splits into two masses, one on either side of the mouth. In this paired condition the bands run backwards through the trunk of the embryo. Essentially the same condition of the germ-layers in front of the mouth persists till the labrum is definitely formed, as I have observed in a few series of sections. It is difficult to account for the late and intimate union of the mesentoderm with the ectoderm in the labral region, unless we suppose that the blastopore originally extended as far forward as this region and here closed in such a manner that the three layers were not at once separated into ectoderm on the one hand and mesentoderm on the other. It is mainly on this supposition that I take the labral region to coincide with the anterior widening of the blastopore seen in Fig. 2. This widening probably does not coincide with the stomodaeum, but lies in front of it, and the definitive mouth is a later formation arising independently from the ectoderm alone. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 41 I would here insert a few observations on gastrulation in Stag 7 iionia 7 itis carolma, Gryllus hccticosics, and CEca 7 itJius 7 live Its. In Fig. 12 the egg of Stag 77 ioi 7 ia 7 itis is represented in out¬ line for the purpose of showing the relatively small size of the germ-band which arises as in other forms from a thickening of the blastoderm on the ventral face of the yolk. It is seen to lie somewhat nearer the broad posterior than the pointed anterior pole. It is but slightly longer than broad, and its wider anterior end, which is directed towards the upper pole of the egg, foreshadows the procephalic lobes. Fig. 11 shows that the germ-band of the Mantid, unlike that of Xiphidiimi, is raised above the niveau of the yolk and has its marginal cells sharply separated from the extra-embryonal blastoderm — or serosa — as it is now called. This much flattened layer is, nevertheless, encroaching on the edges of the germ-band to form the amnio-serosal fold (aiiis.). At the anterior edge lies a small cluster of cells (/. 0) but little larger than those of the germ-band. I believe that these may represent all that remains of an indusium in StagiTtoinantis. The narrowly pear-shaped blastopore is very short. Sections show it to be a deep groove, which like the median infolding of other forms {po 7 ypJiora, Hyd 7 'op]iihLS^ Musca, etc.) is deep¬ est posteriorly and grows shallower headward. As I failed to find any of the stages immediately following the one figured, I could not trace out the formation of the germ-layers. According to Bruce (’86, p. 17), who studied the same species of Stag 77 i 077 ta 7 itis, “ When the union of the folds (of the amnion and serosa) is effected and the embryo is separated from the sur¬ face and covered ventrally by the amnion, the under layer is formed, as in Meloe and TJiyridopteiyx as an ingrowth from the median line of the embryo.” This remark, together with his Figs. XLII-XLIV, PI. IV, shows that he could not have ob¬ served the formation of the layers from a groove and that he must have based his inference on a stage later than the one I have figured. In Gryllus luctuosus the blastopore is more like that of Xiphidiu 77 i. The outline of the egg is shown in Fig. 14. The 42 WHEELER. [VOL. VIII. germ-band is relatively much larger when compared with the yolk-mass than the germ-band of Stagmo^nantis. It arises on the ventral surface very near the lower pole. That such is the correct position of the embryo may be easily ascertained, since the mother-insect thrusts her eggs into the ground with their long axes perpendicular to the surface. In a glass jar con¬ taining a few inches of earth, many eggs were deposited between the surface of the glass and the earth, so that the exact position of the apical pole could be noted, and the egg removed and hardened with this pole constantly in sight. Thus it was possible to determine the exact topographical relations of the embryo to the yolk throughout the important stages of early development. During gastrulation the germ-band of Grylltcs (Fig. 13) is more elliptical and somewhat narrower than the germ-band of Stagmomantis. Its edges are also distinctly marked off from the blastoderm and here, too, the amnio-serosal fold (ams.) arises along the entire periphery. The blastopore (d/.) is much narrower than the corresponding depression in Stagmomantis. It is deepest posteriorly. The discovery of an invaginate gastrula in Grylhis made it extremely probable that this stage had been overlooked in the other members of the same family which have been studied from an embryological standpoint. Neither Korotneff in his study of Gryllotalpa (’ 85 ), nor Ayers in his study of CEcantJms (’ 84 ), succeeded in finding an invagination. I was unable to secure the eggs of any of our native Gryllotalpce, but I col¬ lected a great number of CEcantIms eggs in Ohio during the last days of September. An examination of these soon convinced me that Ayers had not seen the youngest stages in the develop¬ ment of the germ-band. The youngest germ-band that he figures (Figs. 1-18) lies near the posterior end of the egg with its tail pointing towards the micropylar pole. According to Ayers “ A tract of the blastoderm along the median line of the ventral (concave) side, lying nearest the deep or primitively head-end of the egg, becomes thickened into a germinal band, which is the first trace of the body of the embryo.” But this is not the first trace of the body of the embryo, nor does it No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. arise on the concave face of the egg. The germ-band of CEcantJmSy like that of Grylltts., first makes its appearance as a thickening of the blastoderm on the convex surface near the lower pole of the egg. This convex surface is, therefore, the ventral surface and the micropyle marks the “ primitively head- end ” of the egg as is shown by the fact that the procephaleum is originally directed towards this and not towards the opposite pole, which Ayers incorrectly calls the “primitively head-end.” The germ-band, however, soon leaves its position on the convex ventral surface and, moving around the lower pole tail first, comes to lie on the concave dorsal yolk. It is clear that Ayers could not have seen the stages preceding the arrival of the germ-band on the dorsal surface, and it is during these very stages that the blastopore forms and closes. Before turning the lower pole the germ-band of CEcantJms resembles that of Stagmomantis. Its anterior is distinctly wider than its posterior end and represents the future pro- cephalic region. A narrow, but distinct groove runs from the oral to the anal end as in the forms we have been considering. At the posterior end the groove bifurcates much as in Xiphidiiim. That this median groove gives rise to the mesen- toderm admits of little doubt after what has been said of other Orthoptera. The amnio-serosal fold appears to arise simul¬ taneously along the entire margin of the germ-band as in Gryllus. It follows from the observations here recorded, fragmentary as they are in many respects, together with Graber’s observa¬ tions on Stenobothrus, that the Orthoptera can no longer be regarded as hors de ligne so far as the formation of their germ-layers is concerned. In all the families of the order, save the Phasmidae, an invaginate gastrula has been found, and there can be little doubt that the investigator who is so fortunate as to study embryos of this family will find in them essentially the same process of germ-layer formation. 44 WHEELER. [VOL. VIII. The view is now pretty generally held that in the Insecta both mesoderm and entoderm arise from a median longitudinal furrow — the former layer throughout nearly the entire length, the latter only in the oral and anal regions of the germ-band — and that the vitellophags, or cells left in the yolk at a time when the remaining cleavage products are traveling to the sur¬ face to form the blastoderm, take no part whatsoever in the formation of the mesenteron, but degenerate in sitn and finally undergo dissolution. Discussions of the literature on this subject are to be found in the papers of Heider (’ 89 ) and Graber (’ 89 , ’ 90 ), and so few are the facts accumulated since these resumes were penned that I may dispense with an historical consideration of the insect germ-layers in the present paper. In the interpretation of the insect gastrula the entoderm has always played an important role. The origin of the mesoderm has long been known and has been duly provided for in the various germ-layer hypotheses which have from time to time been advanced. But the true origin of the lining of the mid-gut has been ascertained only within the last few years, so that we cannot expect to find an adequate treatment of this layer in the older theories. Led astray by what had been observed in Crustacea and Arachnida, some writers chose to regard the vitellophags as forming the mesenteron and shaped their theories accordingly (Oscar and Richard Hertwig, ’8i). But now that it has been shown that the vitellophags take no part in forming the lining of the mid-gut, their morphological position is rendered even more obscure, and we are brought face to face with the question: Are the vitellophags a portion of the entoderm which has been set apart very early in develop¬ ment for the purpose of yolk-liquefaction or are they an entirely new segregation of cells belonging to none of the conventional germ-layers ? Those who defend the former alternative main¬ tain that the vitellophags of insects are entodermal in function inasmuch as they digest yolk and closely resemble the amoeboid Crustacean yolk-cells which are actually converted into the lin¬ ing of the mesenteron. On the other hand it is urged, that as the yolk-cells arise and function before the blastoderm is com- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. pleted and hence some time before the germ-layers are formed, they cannot properly be assigned to the entoderm.^ It is probably best to await the results of further investiga¬ tion before deciding on the phylogenetic relations of the vitel- lophags to the entoderm. Heider (’ 89 ) has also expressed himself to this effect and I fully endorse his opinion when he says: ‘‘Immerhin wird man vorlaufig liber vage Vermuthungen nach dieser Richtung nicht hinauskommen, und ist die Frage nach der Auffassung der Dotterzellen bei dem Nachweise, dass sie an dem Aufbau des Embryos keinen Antheil nehmen, meiner Ansicht nach von geringerer Wichtigkeit.” 1 Besides these vitellophags which with Cholodkowsky (’ 91 ^) we may call the primary yolk-cells — there are other cells which detach themselves from the blas¬ toderm or embryo and enter the yolk. These Cholodkowsky calls secondary yolk-cells. While the origin of the primary yolk-cells has been quite satisfactorily demonstrated, this cannot be said of those of the second class. They appear to descend into the yolk at different times in different species. Thus, according to Patten (’ 84 ), all the cleavage products in Neophylax ascend to the surface, the yolk-cells subsequently descending from the blastoderm. I claimed a similar total migration of the cleavage products to the surface in Blatta (’ 89 ) ; Cholodkowsky, however, claims that some of the cells never reach the surface, but remain in the yolk. Be this as it may, in later stages I believe it can be shown that cells do migrate into the yolk from the embryo and especially from the entoderm-centres. This was shown by me to be the case in Doryphora, where many cells pass into the yolk from either entoderm pole (PI. XIX, Fig. 82; PI. XX, Fig. 88). I have since observed an exactly similar phenomenon in Telea polyphemus in a correspond¬ ing stage of development. Graber, (’ 89 , p. ii) too, has made a similar observa¬ tion on Melolo 7 itha, w^here he saw “ vom invaginirten Blastodermwulst aus unter lebhaften Theilungserscheinungen ganze Strbme von Zellen in den Dotter hinein- wandern, Zellen die freilich von den primaren, gleichzeitig vorkommenden und auffallend grosskernigen Centroblastelementen ganz enorm verschieden sind, und die sich iiberhaupt durch ihre ganze Beschaffenheit als unzweideutige Abkommlinge, man konnte sagen Auswiirflinge eines wahren Keimblattes, erweisen.” So far as the migrant cells described in Doryphora are concerned, I am sure they come from the entoderm. They occur only at or near the entodermal Anlagen and may be traced from this germ-layer into the yolk. These cells are not actively dividing like those described by Graber, but actively disintegrating. (May not Graber have mistaken disintegration-figures for caryokinetic figures ?) In somewhat later stages no traces of these migrant cells are to be found. The yolk is segmented at the time of their leaving the entoderm and their invasion appears not to disturb in the least the activities of the vitellophags. Whether there is any relation between these evanescent entoderm cells and the “secondary mesoderm” of Reichenbach (’ 86 ), the “spores” of F. H. Herrick (’ 86 ), or the “chromatin nebulee” of Bumpus (’ 91 ) is a question which cannot be answered at present. 46 WHEELER. [VOL. VIII. Among those who take a decided stand on the relations of the vitellophags to the definitive entoderm, Graber and Cholod- kowsky may be mentioned. Graber (’ 89 , p. lo), after intro¬ ducing the superfluous and inapplicable term “ centroblast,” ^ says: Dabei nehme ich zugleich, was indessen kaum misbilligt werden diirfte, stillschweigend auch an, dass dieses gegenwartig, wie es scheint, von der Darm- and Gewebsbildung ausgeschlos- sene Zellenlager auch friiher niemals eine dem echten Ento¬ derm anderer Thiere entsprechende Rolle inne gehabt habe, sondern vielmehr dem letzteren gegeniiber ein neues, wahr- scheinlich mit der starkeren Entwicklung des Dotters im Zusammenhang stehendes Differenzirungsproduct ist.” Cholodkowsky (’9i®) does not dismiss the matter so briefly. Like Graber he draws a hard and fast line between primary and secondary yolk-cells, and admits no phylogenetic continuity between the vitellophags and the definitive entoderm. The vitellophags belong to none of the germ-layers. His reasons for not regarding them as a precociously segregated portion of the entoderm are neither new nor conclusive. Like other recent investigators he admits that the vitellophags are in part digested or discharged from the alimentary tract along with the remains of the yolk after hatching. But he is not satisfied that the yolk-cells should play a humble role in the insect economy. Some of them were predestined to a higher function than yolk-liquefaction — viz: to give rise to the blood, the fat- body and even to the germ-cells. He therefore supposes that the vitellophags are undifferentiated cells. But this supposition is not supported by the facts. That they are on the contrary, considerably specialized is shown by their limited function and mobility, their gradual and prolonged growth (especially in some Orthoptera), their inability to undergo caryokinesis or even akinesis, and their suspicious relations to the bacteria¬ like corpuscles of Blochmann. On a priori grounds we should not expect to derive whole sets of tissues from such specialized elements. 1 Superfluous because we have enough names for these cells already, inap¬ plicable because the termination “blast” is properly applied only to cells or tissues of a germinal character — not to decrepit elements like the yolk-cells. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. But a more weighty objection may be adduced. It has been shown by Heider (’ 89 ) and Heymons (’ 90 ), not to mention many previous investigators, that the fat-body and sexual-cells arise from the mesoderm, and my own studies fully confirm this view. Concerning the origin of the blood there is some doubt, but it should be stated that Cholodkowsky has made no satisfactory observations of his own on this point and that, although some facts point to a derivation of the blood from the yolk-cells, others as definitely point to its origin in the mesodermal layer. After taking for granted that the vitellophags are undiffer¬ entiated cells, that they have nothing and, what is more, never have had anything to do with the entoderm, and that they give rise to blood-corpuscles, adipose-tissue and germ-cells, Cholod¬ kowsky ushers in the parablast theory. It was to have been hoped that this theory might have been permitted to end its days in peace within the confines of vertebrate embryology where it originated. Fortunately, however, it has grown too old and decrepit, even under the skillful medical treatment which it has received from time to time, to be of any service in invertebrate morphology. There is something almost ludicrous in Cholodkowsky’s ap¬ plication of the parablast theory to the Insecta when we con¬ sider the methods which he employed in preparing the yolk of the Blatta egg. The capsules opened at both ends were sub¬ jected to the action of undiluted Perenyi’s fluid for 12 hours and the eggs after treatment with the customary grades of alcohol, cleared in green cedar oil 24 hours. Thence they were transferred to paraffine (55—60° C.) and left 3-5 hours. The result of this heroic method is apparent enough in the dis¬ tortion of the tissues, but its effect on the yolk is quite remarkable. Both Blochmann (’ 87 ) and myself (’ 89 ) described the yolk of the Blatta egg as consisting of a mass of homogeneous and granular albuminoid bodies sharply polygonal from mutual pressure and interspersed with spherical oil-globules. We also described a peculiar distribution of the polygonal bodies; those of a homogeneous nature constituting an oval central core in¬ vested with the granular bodies. I further claimed that the 48 WHEELER. [VOL. VIII. Blatta egg exhibited a yolk-segmentation which though faint and appearing late was, nevertheless, comparable to the yolk- segmentation in such forms as Doryphora. Cholodkowsky (’ 91 ®) thus describes the yolk: ‘‘ So kann ich, z. B., nicht bestatigen, dass der Dotter aus einzelnen polygonalen Dotterkorpern bestehe, wie derselbe von Blochmann (und Wheeler) beschrieben und abgebildet wird. Der ganze Dotter besteht aus einer continuirlichen plasmatischen Substanz, deren Vacuolen grossere und kleinere Fetttropfen enthalten. Die Continuirlichkeit der Dottermasse tritt nun um so deutlicher hervor, je besser die Objekte conservirt sind. Das Bild (ich mochte sagen, das Trugbild) der polygonalen Dotterkorper ent- steht durch Bersten des Dotters nach der Bearbeitung mit nicht ganz passenden Reactiven.” .... ^‘Auch kann ich die Blochmann’sche Unterscheidung des ‘inneren’ und ^ausseren’ Dotters nicht annehmen; der ganze Unterschied in den Farbungsverhaltnissen der beiden angeblichen Theile des Dotters lasst sich einfach dadurch erklaren, dass die Farbe aus den peripherischen Theilen des Eies leichter als aus den inneren mit Saure ausgezogen wird.” And at p. 58 he remarks: “ Es ist bemerkenswerth, dass bei Blatta germanica eine Dotterzerkliiftung vollkommen fehlt. Ich kann also mit Wheeler nicht iibereinstimmen, wenn er sagt (p. 359), dass bei Blatta der Dotter, wenn auch sehr spat (nach Bildung der Extremitaten) sich furchen soil; hochst wahrscheinlich war Wheeler zu dieser irrigen Annahme durch die ausserordentliche Briichigkeit des Dotters verleitet.” On reading these criticisms I re-examined my preparations and must emphatically re-assert what I claimed in my descrip¬ tion of the yolk of the Blatta egg. Among my preparations I find several mature ovarian eggs hardened in Perenyi’s fluid — not, however, treated with that vigorous reagent for 12 con¬ secutive hours — and these show the yolk-bodies very distinctly as polygonal masses. There are no traces of a ‘‘Bersten des.i Dotters.” Eggs killed in ordinary alcohol and mounted in toto\^ show the polygonal yolk-bodies distinctly and in these same specimens the distribution of the different yolk-elements may be followed by carefully focusing. That Cholodkowsky should No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 49 be unable to detect the outlines of the segments in the yolk of eggs treated for half a day with Perenyi’s fluid is not surpris¬ ing, especially as this segmentation is of very short duration in Blatta as in other Orthoptera. It is present, however, as I have convinced myself from eggs mounted in toto and from sections. If prolonged immersion in Perenyi’s fluid can bring about a complete fusion of the yolk-bodies and an obliteration of their true structure, what must be its effect on the vitellophags scattered through the yolk ? And how much importance are we to attach to Cholodkowsky’s assertion that the fat-body, blood-corpuscles and sexual-cells arise from the vitellophags, and to the parablast theory as applied to the Blatta-owwm.} Let us return from this digression to the germ-layers. The curious fact that the definitive entoderm of the Insecta arises from two separate centres — one oral and the other anal — is too recent to have given rise to much speculation. Since the entoderm of other animals arises from a single centre it is tacitly assumed that such must originally have been the case with the Insecta, and that the present bipolar condition must be due to secondary modification. Starting with this postulate, there are, of course, many ways in which the bipartition of the original unipolar entoderm may be supposed to have taken place. Two of these possibilities are worked out in the hypo¬ theses of Kowalevsky (’86) and Cholodkowsky (’ 9 i^). Kowalevsky has expressed his views so clearly and con¬ cisely that I cannot do better than quote his own words: “Wenn wir jetzt versuchen, diese Bildung des Ento- und Mesoderms bei den Musciden mit der Bildung dieser Blatter bei anderen Thieren zu vergleichen, so sehen wir erstens, dass hier auch eine Art sehr in die Lange ausgezogener Gastrula entsteht, und dass aus dem eingestiilpten Teil das Ento- und Mesoderm sich bildet. Also in diesen allgemeinen Ziigen finden wir eine Uebereinstimmung. Es scheint mir aber, dass die Parallele noch weiter gezogen werden kann. Namentlich wenn wir der Bildung des Ento-Mesoderms bei Sagitta uns erinnern, so finden wir bei derselben dass der eingestiilpte Teil des Blastoderms in drei parallele Sacke zerfallt, von 50 WHEELER. [VOL. VIII. denen der mittlere das Entoderm liefert, die seitlichen aber das Mesoderm. Bei den Musciden entsteht auch eine solche Ein- stiilpung wie bei Sagitta, und auch der mittlere Teil — aller- dings nur an beiden Enden vorhanden — liefert das Entoderm, die seitlichen Teile liefern das Mesoderm: also ahnlich dem, was wir bei der Sagitta beobachten. Um die Aehnlichkeit weiter zu fiihren, kann vorausgesetzt werden, dass bei der so in die Lange gezogenen Gastrula der Insekten der mittlere, das Entoderm liefernde Sack so ausgezogen ist, dass er in der Mitte ganz verschwindet und nur an seinem vorderen und hinteren Ende bestehen bleibt. Bei dieser Auffassung wird es von selbst schon folgen, dass die sich schliessende Rinne fast auf ihrer ganzen Lange nur das Mesoderm liefert. Jetzt bleibt noch die Frage iibrig: wie verhalten sich die Flachen der Gastrula zu den Flachen des sich bildenden Ento¬ derms. Bei der Sagitta wird die aussere Oberflache der Blastula nach der Einstiilpung zur inneren Oberflache des Darmkanals, d. h. die Seiten der Zellen, welche bei der Blastula nach aussen gerichtet waren, werden im Darmkanal nach seinem Lumen gerichtet. Bei den Insekten kann dasselbe auch vorausgesetzt werden. Wenn wir uns die eingestulpte Rinne vorstellen, so sind deren Oberflachen ganz ahnlich gelagert wie bei der Gastrula; wenn wir weiter die Bildung der beiden Entoderm- anlagen dem mittleren Sacke der Sagitta vergleichen, so bleibt die Lagerung der Zellenfiachen noch ganz dieselbe. Wenn wir dann voraussetzen, dass der mittlere Sack durch die weite Aus- breitung und durch das Eindringen der Masse des Dotters gewissermassen in seinen vordern und hintern Teil zersprengt ist, so kommt der Dotter ins innere des hypothetischen Sackes, und die Zellen, die den Dotter bedecken, werden zu dem Dotter in derselben Beziehung stehen, wie bei der Sagitta zu der eingestiilpten Flache.” A few years after these remarks were written Heider (’ 89 ) and myself (’ 89 ) at about the same time published observations on the Coleopteran germ-layers which seemed to support the hypothesis of the celebrated Russian embryologist. As further support to Kowalevsky’s view I believe we may point to such gastrulas as that of Stagmomantis described above. In this No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. gastrula, which is so very short and broad, we may suppose that the oral and anal entoderm-centres are really continuous, covering the floor of the blastopore from end to end. In sec¬ tions, it is true, I failed to detect any differentiation of the cells forming the walls of the furrow, into entodermal and mesodermal elements, but this would also be the case in the elongated gastrulas of other insects in a correspondingly early stage {XipJndium^ Doryphora). As favoring a purely mechan¬ ical separation of an originally single entoderm Anlage, it may be noted that the most rapid elongation of the germ-band occurs at a time when the entoderm is differentiating from the mesentoderm Anlage. There is probably more than an acci¬ dental correlation of these two processes. During this time some germ-bands {Doryphora, Xiphidiufr) double their length.^ Inasmuch as the lengthening of the superficial layers of the embryo is much more rapid than the differentiation of the entoderm, this germ-layer must lag behind. In most insects the embryo is at the time of its greatest elongation much longer than the yolk-mass and must again shorten to the length of this mass, so that a rapid proliferation of the entoderm may be superfluous, since this layer would have to readjust itself to the yolk with the contraction of the embryo. It may, therefore, be an advantage for the entoderm to be somewhat retarded in its growth. In the Orthoptera, where the embryo lengthens rapidly, shortens, and then lengthens again to envelope the yolk we may suppose, for reasons to be given in the sequel, that yolk has been acquired. This seems also to be suggested by the histological structure of the embryonic entoderm; this layer consisting of large polygonal cells in the Coleoptera, which have only a medium amount of yolk, while in the Orthoptera the attenuate entoderm-bands consist of a very few flat cells.^ It is probable that when more forms have been carefully 1 Blatta forms a very rare exception in this respect. 2 A very similar condition may be observed in the case of the blastoderm. In the Coleoptera and Diptera, which have a medium or small amount of yolk, the newly formed blastoderm is a deeply columnar epithelium ; in the Orthoptera it is a true pavement epithelium. 52 WHEELER, [VOL. VIII. studied a method of entoderm formation midway between the unipolar and bipolar methods will be found to obtain in some insects. We must admit that a contribution of elements to the entoderm from the interpolar region of the furrow is not with certainty precluded in several of the species which have been studied. Thus Heider, (’89) while inclined to believe that such a contribution does not take place in the anterior portion of the germ-band, believes that it may take place in the posterior abdominal segments. Cholodkowsky (’9i^) is inclined to accept a still more diffuse origin for the entoderm. “ Untersuchungen zahlreicher Schnittserien machen es wahrscheinlich, dass an verschiedenen Stellen des Keimstreifens sich einzelne Zellen vom ausseren Blatte abspalten und an der Bildung des inneren Blattes betheiligen, so dass die Entstehungsart des letzteren sehr complicirt erscheint.” Starting with the same postulate as Kowalevsky, viz: that the bipolar is derivable from a unipolar condition of the ento¬ derm— Cholodkowsky (’91®- ’91^) proceeds to account for this phenomenon in a very different way from his compatriot. He takes the small, round blastopore of As tarns, stretches it till it equals the insect blastopore in length, introduces a number of modifications — such as the median groove and the pairs of lateral depressions and believes that he has found an explana¬ tion ‘‘ sehr klar und ungezwungen ” for all the different blasto¬ pores, not only in the Insecta, but also in the meroblastic eggs of vertebrates. It was Kleinenberg who said: “ Gewagte Hypo- thesen, kiihne Schliisse nutzen der Wissenschaft fast immer, die Schemata schaden ihr, wenn sie die vorhandenen Kenntnisse in eine leere und dazu noch schiefe Form bringen und bean- spruchen tiefere Einsicht zu geben.” The latter part of this aphorism seems to be particularly applicable to Cholodkowsky’s exposition. As Graber (’9i) has briefly pointed out, there are no grounds for comparing the Astactis blastopore with the entire insect blastopore. In the Decapods this orifice is confined to the anal region and if comparable at all to the median furrow in insects, must be compared with the caudal entoderm pole. This is all that is admissible, since the mesoderm of the Decapoda arises from the anterior lip of the blastopore and No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. proliferates headward. That such is its origin has been shown by Bobretzky and Reichenbach for Astactts (’86), by Paul Mayer for Ettpagurits igii), and by Bumpus for HomartLs (’ 9 i). To Cholodkowsky both the extent and position of the blastopore are of little consequence as is abundantly evident from his reply to Graber’s well-founded objection. It is this very neglect of what are generally and, I believe, rightly considered two of the most important matters in the discussion of the germ- layers, which stamps Cholodkowsky’s hypothesis as superficial and inadequate. There is, however, one redeeming suggestion in his hypo¬ thesis, viz: that the diverging grooves at the posterior end of the blastopore in insects may correspond to the “ Sichelrinne ” of vertebrates. Certainly the relations of the grooves to the median furrow \x\. XipJiidium (see Fig. i.) closely resemble in surface view the relations of the “ Sichelrinne ” to the primitive streak in the chick as figured by Roller (’ 8 i) and in the triton as figured by Oscar Hertwig (’90, p. 99 ). While most investigators probably agree with Kowalevsky and Cholodkowsky in deriving the bipolar from a unipolar con¬ dition of the entoderm, Patten does not share this view (’90). In his opinion, which is based on Kleinenberg’s interpretation of the gastrula, the blastopore is restricted to the oral region, and such depressions as occur at the posterior end of the germ- band, as well as the formation of teloblasts in that region, are supposed by him to have no connection with the blastopore, but to be merely the instruments of unipolar growth. ‘‘The Arthropod body represents an outgrowth from the trochosphere, but the trochosphere itself, the coelenterate stage, has disap¬ peared. Hence there is no such thing as a gastrula in Arthropods and strictly speaking, no germ-layers.” It is clear that this view must stand or fall with Kleinenberg’s theoretical conclusions on which it is based, and we may venture to say that E. B. Wilson’s recent work (’90) has rendered this founda¬ tion very insecure, notwithstanding Patten’s rather confidant assertion that Lopadorhyiicims it is certain that the greater part of the mesoderm arises from the ectoderm at the growing tip of the tail, and has nothing to do with primitive mesoderm.” 54 WHEELER. [VOL. VIII. But it would be out of place to consider the widest bearings of Patten’s hypothesis in this paper since I am concerned with it only in so far as it bears on the germ-layers of insects. Starting with the assumption that the blastopore is confined to the mouth, he attempts to show that the median furrow is a purely secondary structure. “That the median furrow of in¬ sects is merely an ontogenetic adaptation is sufficiently evident from the fact that it may be present or absent in closely related forms.” This, however, is not the case. On the contrary the furrow or a slight modification of it is, we have every rea¬ son to suppose, universally present in the Insecta, at least in the Pterygota, and this wide occurrence of the structure is one of the surest indications of its high antiquity and phylogenetic importance. In the latter part of his discussion Patten admits that there are “ structures in Arthropods which may represent remnants of gastrulas. For example, if the mouth and oesophagus of Arthropods is primitive — and there is no reason to suppose it is secondarily acquired — then we must look for primitive entoderm at its inner end. I have figured in ‘ Eyes of Acilius,’ at the very anterior end of the embryo, a great sac of entoderm cells which probably arise by invagination, although the process was not directly observed. The sac, which soon opens outward by the oesophagus, afterwards becomes solid, and finally is converted into two longitudinal bands, one on either side, extending backwards to the middle of the body, where they become continuous with similar bands extending forwards from the posterior end of the embryo.” Patten admits that true entoderm is formed at two widely separated regions of the body, but he implies that only the anterior centre is comparable to the entoderm of other animals, the posterior centre being a new and purely adaptive formation. It is just here that his theory appears to me to fail, since it does not explain why the oral and anal centres should resemble each other so very closely in origin, method of growth and histological structure. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 55 III. The Indusium and its Homologues. In none of the Pterygota hitherto studied has there been found any trace of a structure comparable to the indusium of XipJiidiiim and OrcJielimum. The organ appears to have been retained by the Locustidae and completely lost by the embryos of other winged insects. In some of the Apterygota, however, there is an embryonic organ which gives a clue to the possible homologues of the indusium. I allude to the so-called “micropyle” of the Poduridae. During the summer of ’91 I was so fortunate as to secure the eggs of Anurida maritima in great numbers. They are much larger than any of the Poduran eggs hitherto studied — so large that they may be removed from their choria by means of dissecting-needles and partially stained for surface views. It is also an easy matter to obtain good sections.^ When first deposited the eggs are provided with a thin transparent chorion and vitelline membrane, but after cleavage, which is total, is completed and the blastoderm formed, a yellow, peculiarly striated chitinous membrane is secreted from the surface. The egg then enlarges till the chorion and vitelline membrane are burst. The striated membrane was described by Ryder (’86), but he failed to observe that it is attached to a large circular ring — the micropyle.” In section (Fig. V) this organ is seen to be a very decided thickening of the blastoderm which at this time covers the whole yolk-mass as a single layer of minute columnar cells. In the “micropyle” the cells and nuclei are much enlarged and often considerably vacuolated. Surface views prepared according to the partial staining method show that the embryo is already faintly outlined on the yolk and that the ring-shaped organ lies just in front of its head (Fig. VI). The egg being spherical, the embryo is curled in a semicircle and the ^‘micropyle” thus comes to lie on the dorsal surface nearer the head than the tail of the germ-band. In the figure a more advanced embryo is represented as spread out on a flat 1 I mention this because the few fragmentary accounts that have been published on the development of the Poduridse are based on the study of the embryo viewed through the chorion and other envelopes. This has given rise to some errors which I hope to point out in a future paper. 56 WHEELER. [VOL. VIII. surface. The resemblance of the “micropyle” to the indusium is apparent at a glance {cf. Fig. 2, PI. I). I have followed the organ in Amirida through the later stages by means of sections and find that it persists for some time as a simple thickening of the blastoderm, still connected with the peculiar striated membrane which stands away from the surface of the blasto¬ derm at all other points. Finally, when the embryo has become flexed dorsoventrally and the body-walls are closed, it sinks into the yolk and is absorbed. Fig. V. Median section of the egg of Anurida niaritima. d.o., “micropyle”; bid., blastoderm. Although much simpler in its structure, I do not hesitate to homologize this “micropylar” organ in Amirida and the Poduridae in general with the indusium of Xiphidium. A pos¬ sible objection to this homology, on the ground that the indu¬ sium arises on the ventral face of the egg, while the Podurid “ micropyle ” is dorsal, has little weight, since the organ bears in either case the same relation to the head of the embryo. Provided, therefore, the egg of Amirida were to acquire yolk No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. j?; #% t K.Ji: and become greatly elongated, like the Xiphidium egg, the micropylar organ must come to lie on the same surface of the yolk as the germ-band. It has been repeatedly suggested, and, I believe, on very good grounds, that the Podurid “micropyle” is the homologue of the crustacean “dorsal organ.” In both groups the organ arises soon after the germ-band is mapped out on the yolk, and in both groups it is a circular or oval thickening of the blas¬ toderm lying in the median dorsal line distinctly nearer the head than the telson. In the Crustacea its centre often shows a depression to the walls of which the Blastoderm-haut is attached, while standing away from the surface of the egg at other points. An exactly similar condition obtains in the Poduridae; a slight depression marks the centre of the organ in Amirida, while in Amirophorus (Le- moine, ’87) there appears to be a deep pit at the attachment of the chitinous envelope. This depression is com¬ parable to the depression seen in Fig. 3 in Xiphidmm, where the cir¬ cular fold is encroaching on the disk. Up to the stage represented in the figure just referred to, the indu- sium will bear close comparison with the crustacean “ dorsal organ.” In the first stages of its spread¬ ing it also resembles to some extent the saddle-shaped “dorsal organ” of OitiscuSf Porcelho, and Ligia. VI ^ V. 5.W- .....aYV/. Fig. VI. Embryo of Anurida mari- tima spread out on a flat surface, d.o., “ micropyle ”/(5., labrum; yc/.,procephalic lobe; at, antenna; tc. ap., minute ap¬ pendage of the tritocerebral segment; mandible; mx^, first and second maxillae; P^-p^^ first to third thoracic appendages; first abdom¬ inal appendage (= left half of collophore); ap"^, second ab¬ dominal appendage; an., anus. But beyond this point it differs widely from its homologues, and it is difficult to see why it WHEELER. [VOL. VIII. 5S should persist, and instead of sinking into the yolk, envelop the whole egg, secrete a granular and thereupon a chitinous layer, and finally, during revolution take on the function of a true serosa. That the organ is rudimental is shown by its tendency to vary, especially during the earlier stages of its development; that it still performs some function is indicated by its somewhat complicated later development and by its sur¬ vival in but very few forms out of the vast group of Ptery- gotous insects. This seeming paradox may be explained, if we suppose that the indusium was on the verge of disappear¬ ing, being the last rudiment of some very ancient structure. As such a rudiment it no longer fell under the influence of natural selection, and for this reason began to vary consider¬ ably like other rudimental organs. Some of these fortuitous variations may have come to be advantageous to the embryo, and were perhaps again seized upon by natural selection; the nearly extinct organ being thus resuscitated and again forced to take an active part in the processes of development. Pursuing the homologies of the indusium still further we come to the Arachnida, where we find in the primitive cumulus of spiders a structure comparable in many ways to the Podurid “micropyle,” as v. Kennel (’85,’88) and Lemoine (’87) have suggested. There is, however, so much difference of opinion regarding the position and signification of the primitive cumulus that I should hardly be willing to agree with these authors, were it not for two of ClaparMe’s figures of the Pholms embryo (’62, Figs. 6 and 7, PI. I). These show in the median dorsal line a thickening which forcibly recalls the “micropyle” of Anurida. Still it must be admitted that Clapar^e has failed to prove the identity of this thickening with the primitive cumulus. In Pentastomids the ‘Tacette ” or “cervical cross” described by Leuckart (’ 60 ) and Stiles (’9i) is very probably the homo- logue of the crustacean dorsal organ and the insect indusium. Although no homologous structure has yet been detected in the Myriopoda, the occurrence of a dorsal-organ-like structure in such widely separated groups as the Hexapoda, Araneina, Pentastomidae, and Crustacea is sufficient reason for regarding No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 59 it as exceedingly ancient and as well-developed before the existing subdivisions of the Arthropoda were established. To seek a homologue of the ‘‘dorsal organ” among existing annelids may be regarded by some as a hopeless undertaking. Still I would call attention to Apathy’s observation (’88) on Clepsiiie biocidata. The adult of this species has long been known to possess a chitinous plate in the median line between the head and the praeclitellum. Apathy finds that this plate is the remnant of an embryonic sucking-disk, the glandular cells of which secrete a bundle of byssus-like threads that harden on contact with the water and serve to anchor the undeveloped young to the ventral concavity of the mother-leech. A similar organ is also found in the young of Clepsine heteroclita. It is certainly no great step from this embryonic sucking-disk of the Hirudinea to the Phyllopod “cervical gland” which is also used as a sucker, and which Fritz Muller (’ 64 ) and Grobben (’ 79 ) regard as homologous with the “dorsal organ” of the . Amphipoda. IV. The Envelopes and Revolution of the Insect Embryo. I. The Amnion and Serosa. The formation of two cellular envelopes, the amnion and se¬ rosa, by a folding of the primitive extra-embryonal blastoderm, is rightly considered one of the most characteristic features of the Hexapod embryo. The envelopes are not, however, com¬ mon to all insects. An amnion is completely lacking in the Poduridae,^ and consequently the extra-embryonal blastoderm in these forms is strictly comparable to the corresponding por¬ tion of the blastoderm in Crustacea, Myriopoda, and Arachnida. This is proved by the fact that it ultimately forms the definitive dorsal body-wall. So far as our present knowledge extends, the Apterygota may be regarded as Hexapoda Anamniota, and 1 Lemoine (’ 87 ) describes a cellular “membrane amniotique ” in Anurophorus, but he does not represent it in his figures and did not study it in section. I therefore incline to doubt the correctness of his observation, especially as I can find no traces of a cellular envelope in the Anurida egg, which on account of its size is a far more favorable egg for study than that of Anurophorus. 6o WHEELER. [VOL. VIII. placed over against the Pterygota, which are characterized by the possession of an amnion (Ilexapoda Amniota). There is a gap between these two groups of insects similar to the gap between the amniote and anamniote vertebrates. Whether it will be filled by the future study of such orthopteroid forms as Machilis, Lepisma and Forjictila remains to be seen. For the present I am inclined to believe that the amnion first made its appearance in the ancestral Pterygota. Even if it be contended that the amnion was once present in the Apterygota and subse¬ quently lost, its origin could not consistently be pushed further back than the Hexapoda, since this envelope is lacking in the Myriopoda, which, there is reason to believe, lie in the direct line of descent. The proof that the so-called amnions of Peri- patiis, Scorpions and Pseudoscorpions are the homologues of the insect amnion is not forthcoming. Judging from the few descriptions of their formation, they appear to have arisen in¬ dependently within their respective groups. Just as many of the Pterygota develop only rudiments of wings or have altogether ceased to develop these organs in the adult state, so the embryos of the Pterygota in some cases develop only rudimental envelopes or none at all. It is reported that the amnion is lacking in the Proctotrupid Hymenoptera (Ayers, ’84) and rudimental in Muscidae (Kowalevsky, ’86 ; Graber, ’89) and viviparous Cecidomyidae (Metschnikoff, ’66). Certain ants of Madeira are incidentally mentioned by Met¬ schnikoff as having the envelopes represented only by a small mass of cells in the dorsal region. The absence or abortion of the amnion is almost certainly a secondary condition. The Proctotrupidae are egg-parasites and undergo an extremely ab¬ errant embryonic and larval development. Both these and the other insects mentioned belong to groups characterized by high specialization. This is notably the case with the ants and with the Muscidae which show considerable aberration in their embryonic and larval stages. The paedogenesis of the Cecidomyids studied by Metschnikoff stamps them also as ab¬ errant. Moreover the embryos of other Orthorrhaphous Dip- tera (Simulidae, Chironomidae, Tabanidae) have perfectly normal envelopes. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 61 Many attempts have been made to explain the origin of the amnion in insects. It first appears abruptly and fully devel¬ oped in the Orthoptera just as the vertebrate amnion appears abruptly in the Reptilia. One school, represented by Nus- baum (’ 87 ) and v. Kennel (’ 85 , ’88), regards the insect amnion as a structure of high phylogenetic value and would trace it to some organ in the lower Arthropods or in the worms. Ac¬ cording to another view advocated by Will (’88) and myself (’ 89 ), the amnion has had no such remote phylogenetic history, but has arisen more recently in response to certain purely mechanical conditions of development. Nusbaum advances the opinion that the cellular envelopes of the insect embryo are homologous with the “dorsal organ” of Crustacea. The saddle-shaped “dorsal organ” of Ligia and Onisctts is regarded as the key to this homology, the two flaps which clasp the sides of the Isopod embryo being equivalent to undeveloped amnioserosal folds. But I have shown in the present paper that the indusium of Xiphidium is very probably the homologue of the crustacean “dorsal organ,” and as there is besides a well developed amnion and serosa in Xiphidium, Nusbaum’s hypothesis must fall to the ground. His assertion was certainly premature that the “ deux series des organes aussi characteristiques que le sont I’organe dorsal et les enveloppes embryonnaires, s’excluent reciproquement dans les deux groupes des Arthropodes, c’est-a-dire chez les Tracheates et les Crustaces.” So far as the insect envelopes are concerned v. Kennel’s views do not differ essentially from Nusbaum’s. He likewise homologizes the crustacean “dorsal organ” and the Poduran “micropyle” with the Hexapod amnion and serosa. But he goes further and includes under the same homology the amnion of Peripatus, Scorpions and Chelifer and the chitinous envelopes of Myriopods. He supposes all these structures to represent remnants of the annelid trochophore. I feel con¬ fident that he has jumbled together at least three categories of organs which cannot be regarded as homologous inter se, viz.: (i) the series of structures typically represented by the Crus¬ tacean “dorsal organ”; (2) the cellular envelopes of insects; 62 WHEELER. [VOL. VIII. ( 3 ) the chitinous cuticles. As stated above, the amnions of Peripatus and Scorpions probably also represent structures of independent origin and no wise homologous with the envelopes of insects. It is perhaps unnecessary to add that the reduction of all these structures to the annelid trochophore is in the present state of our knowledge little more than a wild guess. Graber (’90) has criticized the view advanced by Will and myself, that the insect amnion arose by an invagination of the germ-band like that of some Myriopods (Geophilus), His contention is certainly in great measure well-founded. Still I believe that it does not affect the essential point of the hypothesis which implies that the amnioserosal fold is the mechanical result of a local induplication of the blastoderm due to rapid proliferation in a single layer of cells. Ryder (’ 86 ) has sought a mechanical explanation for the amnion, and although his paper treats mainly of the vertebrate amnion, he evidently implies that the homonymous envelope of the Insecta had a similar origin. According to him “the amnion in all forms has arisen in consequence of the forces of growth resident in the embryo, encountering peripheral and external resistance either in the form of a rigid outer egg-shell (zona radiata) or decidua reflexa, or even the walls of the uterine cavity itself, supposing of course that a large vesicular blastoderm containing yolk has been formed by epiboly.” This view applies with little alteration to the Insecta. There is the vesicular one-layered blastoderm filled with yolk and the germ-band arising by rapid proliferation at one point. The resistance of the yolk being less than the external resistance of the tightly fitting chorion and vitelline membrane on the one hand combined with the peripheral resistance of the extra- embryonal blastoderm on the other, the germ-band is forced to invaginate. This invaginative process is favored by the dis¬ placement of yolk during its liquefaction and absorption by the growing embryo. We may suppose that this invagination which results in the formation of the amnioserosal fold, assumed a definite and specific character in different groups of insects. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 63 Conditions similar to those to which the insect germ-band is subjected during its younger stages are often present in the ova and young of other animals, and would be expected to lead to the formation of structures similar to the insect amnion. And this is found to be the case. A hasty glance through the animal kingdom at once suggests a number of parallel instances: the invagination from which the Cestode head develops in the Cysticercus ; the similar invagination in the larval Gordiid; the origin of the Nemertine in the Pilidiitm; the formation of the definitive trunk in Aulastoma, according to Bergh (’85); the development of the trunk and Scheitelplatte in Siptinctdusy according to Hatschek (’84); the formation of the young Spatangid in the Pluteus, according to Metschnikoff, and the somewhat similar conditions in the development of the An- tedon, according to Barrois (’88); the formation of the trunk in the Actinotrocha of Phoronis (E. B. Wilson ’8l); the de¬ velopment of the Polyzoan within the statoblast (Oka ’91 ; Davenport ’9i). I need hardly say that the development of the amnion and serosa in vertebrates is a strictly analogous case. A case still more to the point, because occurring in the Insecta, is the formation of the imaginal disks. In this process we have all gradations till we reach the extreme in Miisca, where the hollow disks whose inner walls bud forth the imaginal appendages are almost completely abstricted from the original hypodermis. The resistance of the chitinous cuti¬ cle of the larva in causing the invagination of the disks admits of easy observation. It certainly cannot be claimed that in all the different forms here enumerated genetic relationship lies at the bottom of the mutual agreement in the methods of form¬ ing the trunk or certain organs. On the contrary, everything goes to show that these similar methods in widely separated groups have been independently acquired under the stress of similar developmental conditions. Perhaps the most difficult point to explain in the view here advanced, is the complete abstriction of the amnion from the serosa in nearly all insects. It is more natural to suppose that the inner envelope would remain continuous with the outer, so that the embryo could the more readily be everted 64 WHEELER, [VOL. VIII. during revolution. The only explanation I have to offer, will be given in connection with a discussion of the movements of the germ-band. In that connection the variations in the devel¬ opment and amputation of the envelopes in the different groups of insects may also be treated to greater advantage. 2 The Yolk. To my knowledge, the quantity of yolk in the insect egg has not been made the subject of comparative study. It has long been vaguely stated {vide Brauer, ’69 and ’70) that the eggs of Ametabolous insects contain relatively more yolk than the eggs of the Metabola. In other groups of animals (Crustacea, Annelida, Mollusca, Vertebrata) it is often observed that ab¬ sence of yolk is correlated with free larval development, while in eggs provided with an abundance of yolk the larval stages are either lacking or considerably modified. This same law obtains also in the Hexapoda, though it can hardly be formulated so concisely as in other groups of animals. And this is not sur¬ prising when we stop to consider that, as regards complexity of organization, the difference between the simplest insect larvae, such as those of the Muscidae and their highly special¬ ized imagines, is far from being as great as the differences between the trochophore and the Annelid, or the Nauplius and the crustacean. Beginning with the Orthoptera we find that the egg is pro¬ vided with an abundance of yolk,—the germ-band when first formed in most cases covers only a very small portion of its surface, and when it reaches its maximum length before revolu¬ tion is no longer than, and usually not so long as, the egg.^ The period of embryonic development is greatly prolonged; most of the species are monogoneutic and oviposit in the fall, the larvae not hatching till the following spring or summer. There is practically no metamorphosis. In the most highly metabolic insects (Muscidae) on the other hand, the quantity of yolk is comparatively limited. The germ- band before revolution is nearly double the length of the egg, 1 To this rule Gryllotalpa seems to be a noteworthy exception. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 65 SO that the head and tail ends nearly meet. Embryonic de¬ velopment is completed in a day, and the larva must pass through a complex metamorphosis to reach the imaginal state. The chasm between these two extremes is bridged by the less metabolic insects (Coleoptera, Neuropotera, Lepidoptera, Hymenoptera, etc.). The quantity of yolk is intermediate be¬ tween that of the Orthopteran and Dipteran egg. The germ- band, like that of the Muscidae, is longer than the egg when it reaches its full length. But it is at this time much narrower than the yolk-mass, whereas in the Muscidae it embraces nearly half the circumference of the yolk. The larvae usually hatch after a period of ten to thirty days in a relatively more ad¬ vanced stage of organization than Dipteran larvae. It is probable that the quality of the yolk is also an impor¬ tant factor in development. The yolk of the Orthoptera and Rhynchota is dense and resembles that of the crustacean and Arachnid egg, while the yolk of the Metabola seems to have a much looser molecular structure. Hence, bulk alone is no criterion of the amount of yolk in an insect’s egg. The view here advocated, that the eggs of the Ametabola contain more yolk than those of the Metabola, admits of some exceptions. Thus the 17 -year locust {Cicada septejtdeciiri) is a large insect with incomplete metamorphosis, but it never¬ theless produces a great number of very small eggs. This is, however, seen to be a greater advantage to the insect than the production of a few large eggs, when we consider the extremely long period of larval life and the vicissitudes to which the larvae may be subjected during all this time. Similarly, Meloe angicsticollis produces a great number of very small eggs, while the eggs of the smaller beetles {Doryphora, e.gi) are much larger. But Meloe is a parasite form, and probably only a few of its many offspring ever succeed in gaining access to the eggs of the bee. The larvae, as shown by their hypermetamorpho¬ sis, are subjected to very varied conditions, and this would still further tend to reduce the number of successful individuals. As in anemophilous plants many germs are produced, but very few are destined ever to prosper. Many other exceptions to the general rule, like these two, are probably due to habits 66 WHEELER. [VOL. VIII. which necessitate the production of a great number of ova at the expense of their size. The opposite exception occurs in the parasitic Pupipara, where the nourishment of the single larva within the parent is equivalent to the production of a large yolk-laden egg.^ The question naturally arises: Were the eggs of the prim¬ itive Insecta poor or rich in yolk } As all the evidence of com¬ parative anatomy, embryology and paleontology goes to show that the Metabola are the more recent, the Ametabola the more ancient forms, we are justified in maintaining that prim¬ itive insects, or at any rate the primitive Pterygota supplied . their eggs with a considerable quantity of yolk. At first sight the Apterygota, which have holoblastic eggs, would seem to constitute a serious obstacle to this view, but it must be remembered that total cleavage is not necessarily a criterion of paucity of yolk (witness Arachnida, Crustacea, and Myriopoda). Furthermore, the eggs of some Thysanura, Amirida, e.g. are provided with an abundance of yolk. Holoblastic cleavage in this group is probably a Myriopod trait, as was long ago sug¬ gested by Metschnikoff (’74). We might perhaps conclude that the superficial type of cleavage, like the embryonic en- 1 The differences between the eggs of different insects with respect to the amount of yolk is systematically disregarded by Graber (’90). This is shown by his classification of germ-bands as microblastic and macroblastic, brachyblastic and tanyblastic. These distinctions are readily shown to be distinctions in the amount of yolk and not in the germ-band. Thus the just-established germ-bands of the Saltatory Orthoptera appear to be very small because the eggs contain an enor¬ mous quantity of yolk; while the germ-band of the Muscidae appears correspond¬ ingly large on account of the small quantity of yolk. The amount of yolk fluctuates even within the limits of the single orders so that the newly-formed germ- bands appear to differ in length more than they really do. In the Orthoptera we have the following series in which the amount of yolk decreases, the germ-band in consequence appearing to increase: Melanophis, Mantis, QLcanthns, GrylhiSy Xiphidium, Blatta, (?) Gryllotalpa. Graber’s further classification of germ-bands as orthoblastic and ankyloblastic, or straight and curved, is equally artificiaL In the great majority of cases the shape of the germ-band depends upon the yolk surface on which it arises, or over which it happens to grow. The uselessness of such a classification is also shown in the case of Xiphidium and Orchelimum, where the just-established germ-band is straight, but becomes curved in passing to the dorsal surface, and thereupon again becomes straight. To which of Graber’s classes does this germ-band belong t No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. velopes and the wings, originated in the ancestral Pterygota. But Lemoine (’ 87 ) claims that the segmentation of the Poduran AimrophortLS laricis approaches the superficial type, so that this latter may have had a still more remote origin. It is, however, hopeless to speculate on this subject till the eggs of many more Thysanura and Myriopoda, including the Symphyla, have been studied. The relations of yolk-quantity to the movements of the embryo will be considered in the following paragraphs. 3. Blastokinesis. According to Hallez (’85 and ’86) ‘‘ La cellule-oeuf possMe la meme orientation que I’organisme maternel qui Ta produite: elle a un pole cephalique et un pole caudal, un cote droit et un cote gauche, une face dorsale et une face ventrale; et ces differentes face de la cellule-oeuf coincident aux faces cor- respondantes de Tembryon.” This law was founded on a study of the eggs of Periplajieta^ Hydrophilus and Locicsta, but it finds full support in the descriptions and figures of all investi¬ gators of insect development.^ My own observations, based on some thirty different insects, accord perfectly with those of Hallez. In most eggs the cephalic and caudal poles are readily dis¬ tinguishable, the micropyle being usually located at or near the former. In exceptional cases, however, it is located at the caudal pole. There is frequently a slight flexure in the longi¬ tudinal axis of the egg, foreshadowing the dorsal and ventral, and consequently also the lateral regions of the mature em¬ bryo. The more nearly the egg approaches the spherical form, as in certain Lepidoptera and Coleoptera and in the Tricho- ptera, the more obscure become the relations of the egg-sur¬ faces to the body-surfaces of the mature embryo. There is, however, every reason to suppose that these relations still exist. The practical value of Hallez’ law was shown in studying the Xiphidiinn egg ; all the movements of the germ-band could 1 The only exception is Ayers, who was undoubtedly mistaken in regard to the orientation of the young CEcanthus embryo. 68 WHEELER. [VOL. VIII. at once be referred to the axis of the mature embryo. When the eggs of other insects are oriented in the same manner, it is seen that the germ-band invariably arises on the ventral sur¬ face of the yolk with its procephaleum directed towards the cephalic, and its tail towards the caudal pole. No matter what positions it may subsequently assume, it always returns to its original position before hatching. Frequently the germ-band, when newly formed, lies nearer the lower than the upper pole {Calopteryx^ OEca^ttJms^ Siag7nomantzs, Hydrop/iiltis, etc.). The usual movements are very simple ; from a position of rest on the ventral surface of the egg, the germ-band moves through an arc till its body is completely inverted. Then it rests and again passes back through the same arc to its original position on the ventral yolk. These movements may be compared to the single vibration of a pendulum. The ascending movement I shall designate as anatrepsis^ the descending as katatrepsis, the intervening resting stage as the diapause. The general term blastokinesis may be used to include all the oscillatory movements of the germ-band. Inasmuch as the germ-bands in other Arthropods (Crustacea, Myriopoda, Arachnida, and Thysanura) exhibit no movements comparable to those of the lower Pterygota, and since, more¬ over, the insect germ-band is formed in exactly the same manner as that of other Arthropods and ultimately returns to its original position, no matter what oscillations may intervene, it is safe to infer that blastokinesis has been acquired within the Hexapod and probably even within the Pterygote group. We may also infer from the intimate relations of envelope- formation to blastokinesis in most forms, that both of these processes arose at about the same time. No attempt has been made to account for the origin of blasto¬ kinesis. It has occurred to me that it may be due to causes of a purely physiological nature. The eggs of the primitive Pter¬ ygota were, as I have attempted to show, provided with a con¬ siderable amount of food yolk. Like their modern descendants they were probably also invested with dense chitinous envelopes. These must render the respiration of the embryo difficult as compared with embryonic respiration in annelids, mollusks and No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. vertebrates, or even as compared with _the Crustacea, which usually have much thinner envelopes than insect eggs. Special provision is also made in many of the Crustacea for aerating the eggs. Now the cells of the rapidly growing insect embryo not only absorb and metabolize the yolk but also give off a certain amount of waste matter. That this is not wholly of a gaseous nature is seen in older embryos which have considerable accumulations of uric salts in the blood corpuscles and fat- body. Waste products are undoubtedly given off during the stages preceding anatrepsis, and probably permeate the yolk in the immediate neighborhood of the germ-band. As the oxida¬ tion of these waste products is very probably retarded by slug¬ gish transpiration, and as growth under such conditions would be seriously impeded, we may suppose that the embryo has acquired the habit of moving to another part of the egg where the yolk is as yet unpolluted. Here it grows apace till the surrounding yolk is again charged with excreta. Growth is then temporarily suspended and the embryo moves back to the ventral surface. The embryo reaches a considerable size be¬ fore katatrepsis, so that its rotation must cause a considerable circulation in the yolk bodies. This would also serve to aerate the yolk and to bring fresh pabulum in contact with the assimilating cells of the embryo. It may also be noted that in many insects the movements set in at critical periods of growth. Thus in Xiphidium anatrepsis occurs during the addition of new segments, and in many other forms it im¬ mediately precedes the formation of new segments. In the Orthoptera, katatrepsis usually occurs in the spring and is the signal for a decisive advance in the development of the heart, sexual organs, compound eyes, etc. During this period, also, the abortion of such rudimental structures as the pleuropodia, abdominal appendages and envelopes seems to be hastened. In short, the whole process of katatrepsis, at least in Xiphidium^ has the aspect of rejuvenescence. It will be remembered that the amnion is formed just before or during anatrepsis. It is probable that the complete abstriction of this envelope from the serosa is a device for favoring the movements of the embryo. The germ-band is thereby set adrift on the yolk and enabled to 70 WHEELER. [VOL. VIII. migrate to some other surface. This, of course, necessitates a secondary union of the envelopes previous to katatrepsis. The hypothesis set forth in the preceding paragraphs is also supported indirectly by the fact that in the eggs of the Meta- bola which are less abundantly provided with yolk than the eggs of the Ametabola, blastokinesis is either faint or wanting. Aeration would be much less necessary in such small eggs. The lengthening and shortening movements seen in the embryos of the Metabola as well as in those of the Ametabola may suffice to keep the yolk circulating. The Lepidopteran germ-band, it is true, exhibits movements, but the eggs of these insects are laid in exposed situations and provided with unusually thick envelopes, so that the movements of the em¬ bryo, though differing widely from the typical blastokinesis of lower forms, have perhaps been independently acquired for a similar purpose. I had intended to give a comparative description of blasto¬ kinesis in the different orders of insects but as the known facts have been recently summarized in a masterly manner by Korschelt and Heider (’92) I shall confine my remarks mainly to the Orthoptera. Although Graber, Ayers and others have studied representatives of this very important group, they have given but fragmentary and often inaccurate accounts of the relations of the embryo to the yolk-mass at different periods of development. I may begin my account with the Saltatoria which comprise the three families Gryllidae, Locustidse and Acrididae. As representatives of the first, Gryllus luctuosus and CEcmithus niveiLS were studied. In both of these insects as was pointed out at p. 42 the germ-band arises on the ventral surface of the yolk near the caudal pole. During the formation of the envelopes anatrepsis sets in and carries the germ-band to the dorsal surface where it rests through the winter in an inverted position with its head directed to the caudal and its tail to the cephalic pole. In the spring the envelopes over the head end first fuse and then rupture ; the embryo is thereupon everted and during katatrepsis passes around the caudal pole to regain its upright position on the ventral yolk. The envelopes during No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 71 this process are stripped back over, and finally drawn into the yolk, where they undergo dissolution when the body walls have met in the median dorsal line. The defects in Ayers’ descrip¬ tion of CEcantJms (’84) were pointed out at p. 43 . Gryllotalpa., the only other Gryllid, which has been studied, seems to differ considerably from Grylhts and QLcantJms. Ex¬ amination of Korotneff’s figures (’85) shows that this difference is probably more apparent than real. In his surface views, there is a wide gap between his Fig. 2 , representing the egg in a preblastodermic stage, and his Fig. 3 , representing quite an advanced embryo. One is thus left without any guide to the exact relation of the just-established germ-band to the yolk- surfaces. Korotneff’s defective account of the formation of the germ-layers would seem to show that he did not study these early stages closely. It is obvious that Gryllotalpa is blasto- kinetic both from Korotneff’s statement that the embryo moves during revolution and from his figures 5 , 7 , and 8 , but the exact nature of the process is not clear. The possibility of the embryo’s passing to the opposite surface of the egg is not precluded by the conditions seen in Figs. 7 and 8 . Judg¬ ing from Grylhis and CEcanthus I am inclined to think that the embryo exhibits both ana- and katatrepsis, but that Korotneff has overlooked the former and misinterpreted the latter movement. In the Locustidae, as represented by Xiphidium and Orcheli- miim, we find a modification of the blastokinetic process ob¬ served in Grylhis. Instead, however, of arising near the caudal pole, the germ-band is formed on the middle of the ventral surface, and instead of passing around the caudal pole during anatrepsis it passes through the yolk as if to reach the dorsal surface by a shorter path. Katatrepsis is essentially the same as in the Gryllidae, the embryo passing around the caudal pole. This lack of coincidence in the anatreptic and katatreptic paths is one of the most striking peculiarities of Locustid develop¬ ment; since it is known to occur in no other insect. It is probable that the anatreptic embryo originally passed around the lower pole, but that owing to the formation of the embryo higher up on the ventral surface, and perhaps also to an acqui- 72 WHEELER. [VOL. VIII. sition of yolk at the lower pole, this movement has been deflected. Melanophis femiir-rtibriLin was studied as a representative of the Acrididae. The germ-band is formed very near the caudal pole of the egg, but still on the convex ventral surface. During the formation of the envelopes the posterior end of the body grows around the pole onto the dorsal surface, while its head remains fixed at the pole. It is not until the germ-band has reached a stage corresponding to Stage F. in Xiphidium that its head leaves the pole and the whole body moves upward on the dorsal surface. It soon comes to a standstill and passes the winter in this inverted position. In the spring it moves back around the lower pole and, like the Gryllid and Locustid embryo in a corresponding stage, proceeds to lengthen and en¬ velop the yolk till its head reaches the cephalic pole. Packard (’83) seems to have been the first to study the devel¬ opment of Acridians {Melanophis spretns and M. atlanis). But he had no conception of the true relations of the embryo to the yolk, as is shown by his Fig. i, PI. XVII, where the egg is depicted with the micropylar end uppermost. Leuckart (’ 55 ) long ago showed that the Acridian micropyle is located at the caudal pole. If the egg figured by Packard be inverted, it will represent the embryo on the point of undergoing katatrepsis. The same error is committed by Graber in his accounts of Stenobothrns variabilis (’88, ’90). Misled, like Packard, by the position of the micropyle, he has mistaken the caudal for the cephalic pole. To mean anything his figures must be in¬ verted. As I have not yet studied the later stages of Melan- opliis in section, I will not attempt to describe the details of katatrepsis. Graber claims to have observed that the pleural ectoderm, where it passes into the amnion proliferates a thin cell-lamella to form the dorsal wall, while the amnion remains intact and still covers the ventral face of the embryo. This account is not substantiated by his figures (i and 2 , PI. I). The two thin cell-lamellae extending over the dorsal surface have every appearance of being the walls of the heart, and therefore mesodermal, although it is difficult to see how this organ could be so completely formed in so early a stage. As No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. Graber has paid no attention to the movements of the Steno- bcthriis embryo, and as he most assuredly has not demonstrated from a careful study of the later stages that the lamella in question is really converted into the dorsal wall, I cannot attribute much value to his observation. The foregoing observations go to show that the blastokinetic processes are essentially the same throughout the suborder Saltatoria. Each family presents certain deviations from the type, which is probably most closely adhered to in the Gryllidae. Anatrepsis is aberrant in the Locustidae, while the Acrididae are aberrant in the tardy separation of the procephaleum from the lower pole. Notwithstanding these deviations the Salta¬ toria form a clearly circumscribed group embryologically as well as anatomically, and were it not for Gryllotalpa would be separated by a wide gap from all other Orthoptera. Gryllo¬ talpa is a generalized form, as Brauer has pointed out from a study of its anatomical peculiarities (’86), and his conclusions are to some extent substantiated by the large size of the germ- band as compared with the yolk mass. In the Cursoria, as represented by Blatta germanica, move¬ ments of the embryo are far less apparent. The germ-band never leaves the ventral surface, on the middle of which it first appears. I have shown, nevertheless (’ 89 , text-figures, p. 348), that it moves down the yolk after the rupture of the envelopes till its tail reaches the lower pole. The tail then remains sta¬ tionary, while the head gradually rises to the cephalic pole as the body walls develop and invest more and more of the yolk. Slight as are these movements, they nevertheless recall the blastokinesis of the Saltatoria. I would regard the movement of the whole Blatta embryo towards the caudal pole as ana- treptic ; katatrepsis is probably represented only by the up¬ ward growth of the embryo. The very late occurrence of the former movement may be due to its rudimental character, since it is too weak to carry the germ-band around the caudal pole. Few observations have been published on the relations of ♦ the embryo to the yolk in the Gressoria. In Mantis, as I have shown, the germ-band when first formed lies somewhat nearer the posterior than the anterior pole. The embryo never leaves 74 WHEELER, [VOL. VIII. the ventral surface of the egg, but whether or not it exhibits any traces of blastokinesis my limited material will not enable me to decide, and Graber (’ 77 ), Bruce (’86), and Viallanes (’ 90 ®, ’ 90 ^^), have contributed no observations bearing on this point. It is clear, nevertheless, that in its development. Mantis resembles Blatta more closely than either of these forms re¬ semble the Saltatoria. This merely confirms the view which has long been held respecting the affinities of the Blattidae and Mantidae. From the structural similarity of the Phas- midae and Mantidae we may venture to infer a similarity of embryonic development. It thus appears that the Orthoptera are clearly separable into two groups — the Saltatoria on the one hand and the Gressoria and Cursoria on the other. The Saltatoria are decidedly blastokinetic whereas the non-saltatory forms re¬ tain only faint reminiscences of blastokinesis {Blatta). I am inclined to believe that primitive embryological features have been preserved more faithfully in the Saltatoria than in other Orthoptera. That the habits of oviposition are more primitive in this group is shown by Brongniart’s discovery of a fossil Blattid provided with an ovipositor (’ 89 ). Moreover, several features in the development of the Saltatoria show great con¬ servatism, e.g. the retention of the indusium in the Locust- idae, the order in which the metameres arise, and the myriopod- like habitus of the XipJiidiiim embryo in Stage D. Not only does a study of the Saltatoria throw light on the development of other Orthoptera, but it brings the order into closer union with the Odonata and Rhynchota. The blasto¬ kinesis of the Gryllidae agrees closely with that of the Hydro- corisa among the Hemiptera — eg. Corixa, as described by Metschnikoff (’66). Ra^iatra and ZaitJia will bear even a closer comparison with the Gryllidae. In the much elongated egg of the former, which has the cephalic pole marked by the pair of diverging pneumatic threads, the germ-band arises as usual on the ventral surface with its head directed upwards. As the envelopes develop it passes around the lower pole and finally assumes an inverted position on the dorsal surface. During katatrepsis it returns over the same path. The inclusion ob- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. servable in Corixa and probably also in Ranatra^ of a small quantity of yolk between the caudal amnion and the overlying serosa when the embryo first passes to the dorsal surface, is often observed in the Saltatoria. It is no great step to pass from the conditions seen in the Hydrocorisa to the “ entoblastic ” condition of other Hemiptera {Pediculus, Aphis, Cicada) and the Odonata {Calopteryx), where the germ-band instead of passing to the dorsal yolk during anatrepsis, comes to lie in the middle of the yolk, or even near the ventral surface (Pj/j'-rkocoris). The Thysanoptera, as may be in¬ ferred from Jordan’s brief statement (’88), the Corrodentia (Mallophaga) according to Melnikow (’69), and the Psocidae, according to Packard (’84), are also referable to the ‘‘ento- blastic” type. Concerning the embryonic development of the Plecoptera and Dermaptera nothing is known. So far only the Homomorpha have been considered. The eggs of the Heteromorpha, as I have attempted to show, con¬ tain less yolk. Blastokinesis is nearly or quite lost in this more recent group, a fact that perhaps indirectly tells in favor of my view that the movements of the embryo have been ac¬ quired for the purpose of ventilating the yolk and supplying the growing embryo from time to time with fresh pabulum. The transition to the Heteromorpha is probably represented by the Ephemeridea. According to Burmeister’s account of the development of Palingenia horaria (I quote from Zaddach, ’54): ^‘am dritten Tage, nachdem das Ei gelegt war, hatte sich der Keimstreif gebildet, der zungenformig war, und sich liber zwei drittel der Eilange erstreckte, also in Form und Ausdehnung ganz dieselben Verhaltnisse zeigte, wie im Phryganidenei.” This may, perhaps, be taken to indicate that the Ephemeridea exhibit no blastokinesis ; but the subject requires urgent investigation. Among the Heteromorpha it is especially the Coleoptera which still show distinct though abortive movements of the germ-band. Hydrophihis may be taken as an example. As may be seen from Holder’s figures, the germ-band forms on the lower ventral surface of the egg. As it grows in length, and the amnion is formed, the tail curls around the caudal pole on- 76 WHEELER, [VOL. VIII. to the dorsal surface, but soon separates from the serosa so that a small amount of yolk is enclosed between the two en¬ velopes. Later the yolk is expelled from this region and the envelopes become applied to each other. A true movement then sets in and carries the anterior portion of the germ-band forward up the ventral surface till the procephaleum overlaps the cephalic pole {Cf. Heider’s figures, 4 6 J a and 9, PI. II. (’ 89 ). A certain similarity of these movements to those exhibited in Blatta leads me to believe that they represent a weakened blastokinesis. Whether or not similar movements occur in the other so- called “ ectoblastic ” forms (Diptera, Hymenoptera, Siphonap- tera, Neuroptera, Trichoptera) cannot be decided at present. If such movements occur at all they are probably exceedingly weak. As stated above, the Lepidoptera have developed embryonic movements peculiar to themselves. In all the members of the order hitherto studied, the germ-band arises on the ventral sur¬ face of the egg, and its envelopes are formed while it is still in this position. As development proceeds the convex ventral surface of the germ-band, with its adherent amnion, moves back from the ventral serosa and soon comes to lie in the middle of the yolk. Hereupon the ventral surface of the embryo be¬ comes concave, and its dorsal surface is applied to the dorsal serosa. I have already remarked that this movement of the embryo may have been independently developed for the same purely physiological purposes as blastokinesis in the Homo- morpha. The fact that the movement is represented in the Trichoptera only by the change in flexure of the longitudinal embryonic axis, would seem to indicate that it has been acquired since the Lepidoptera diverged from the Trichopteroid ancestor. Graber (’ 9 o) has recently made the interesting discovery that the Phytophagous Hymenoptera closely resemble the Lepidop¬ tera in the movements of the embryo and in the amputation of the envelopes. This, taken together with the striking resem¬ blance between the eruciform larvae of the two groups, appears to point to a closer relationship than has usually been claimed. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. While studying the movements of the embryo and the for¬ mation of the envelopes in the different orders and families of insects, with a view to testing the current classification, which is the outcome of a great amount of comparative anatomical and paleontological work, I have been especially impressed with two facts: First, the embryological data in no wise conflict with the generally accepted classification of Brauer. The de¬ velopmental variations within limited groups are never greater than the post-embryonic differences in the members of the same groups. Usually there is great uniformity in embryo- logical development between systematically allied insects of the same order ; the wide gaps usually occur between the orders just where gaps have long been pointed out by com¬ parative anatomy and paleontology. Second, developmental differences between members of different allied families of Orthoptera are greater than the differences between remotely related families in more recent orders. For example, the dif¬ ferences between a Locustid and an Acridian or a Locustid and a Gryllid embryo, or between any of the Saltatoria and the Blattidae, or Mantidae, are greater than the differences be¬ tween an embryo Hydrophilid and a Chrysomelid, a Tabanid and a Chironomid, or a Bombycid and a Shingid. Frequently, it is true, the differences between the extremes in the higher orders are considerable, as between the Tenthredinidae and the Proctotrupidae among Hymenoptera, or the Chironomidae and Muscidae among Diptera. If any conclusions bearing on classi¬ fication can be drawn from the few embryological data which I have collected, they refer to the ordinal value of the vari¬ ous Orthopteran families. It would appear that these groups have really more than family value. They are older than the families of more recent groups, and therefore exhibit greater divergence. The Rhynchota will probably be found to present conditions similar to the Orthoptera. There are certainly more considerable differences between the embryos of such forms as Pyrrhocoris and Ranatra than there are between the embryos of widely separated families among the Coleoptera. 78 WHEELER. [VOL. VIII. 4. The Elimination of the Embryonic Envelopes. Anatrepsis and katatrepsis in the lower insect orders, or the completion of the envelopes and their rupture in the higher orders, are separated by a distinct interval, during which the germ-band undergoes a considerable development. But during this interval, the diapause, no change is noticeable in the envelopes themselves beyond a thinning of the amnion with the increased growth of the embryo. The elimination of the envelopes is preceded by katatrepsis just as their formation was preceded or accompanied by anatrepsis. This elimination is immediately followed by the completion of the dorsal body- wall and may take place in a variety of ways. Korschelt and Heider (’92) distinguish the following types in this process: 1. The amnion and serosa become continuous and, after the eversion of the embryo, are drawn back over the yolk to form a single layer of cells. As the dorsad growth of the body-walls proceeds, both envelopes are drawn together and pushed into the yolk to form a sack or longitudinal tube which is ultimately enclosed by the walls of the mesenteron and absorbed. To this type belong the Odonata, Rhynchota, some Orthoptera {Blatta, QEcanthns, Gryllotalpd) and some Coleoptera {Hy- drophihis ). 2. The serosa is shed from the yolk and the amnion alone contracts on the dorsal surface preparatory to being drawn into the yolk and absorbed. (Certain Coleoptera, e.g. Doryphoral) 3. The serosa alone is agglomerated and drawn into the dorsal yolk, the amnion being cast off. (Certain Diptera \ChironomiLS~\ and Trichoptera.) 4. Both envelopes are shed. (Lepidoptera and certain Hymenoptera.) In Xiphidium we may perhaps recognize a fifth type, in which as in the fourth, both amnion and serosa are shed. But while the serosa is in great part shed as a simple membrane, the indusium which is a modified portion of the serosa, together with the amnion is drawn together in a mass and cut off from the embryo. It is more than probable that other types of envelope elimination will be discovered when more forms have No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. been studied. Mtisca may perhaps be regarded as representing a distinct type, since in this highly modified form the rudi- mental amnion and the serosa are neither shed nor agglomer¬ ated and engulfed in the yolk, but are supposed to form the definitive body-wall. (Kowalewsky, ’86; Graber, ’ 89 .) It is clear that the revolution of the insect embryo includes three distinct processes: first, the eversion and katatrepsis of the germ-band; second, the formation of the dorsal walls; and third, the elimination of the envelopes. The mechanical cause of eversion and katatrepsis is probably a contraction on the part of the envelopes after their fusion and rupture over the ventral surface of the embryo. After the embryo is everted from the amniotic cavity, or exposed after the rupture of the amnion and serosa, these envelopes temporarily form the dorsal covering of the yolk. Do they ever form the definitive dorsal body-wall } For both envelopes this is claimed to be the case only in Mtisca. In all other insects the serosa, at least, takes no part in forming the permanent body-wall, as it is either shed or engulfed in the yolk. The question is, therefore, restricted to the fate of the amnion. In many insects (Lepidoptera, Hymenoptera Phyto- phaga, some Diptera and Coleoptera), it has been shown that the amnion takes no part in the formation of the definitive body- wall, although a decision on this point is rendered difficult by the fact that no hard and fast line can be drawn between the ectoderm of the germ-band and the cells of the amnion. In other insects the decision is even more difficult. Still, I may say that I have seen nothing in the insects I have studied, to convince me that the amnion is converted into a portion of the permanent body-wall. Even in Musca it seems probable that the amnion and serosa only temporarily function as the body-wall, and that their cells are ultimately replaced by true ectodermal elements from the germ-band. In Blatta and Xiphiditim I have seen appearances which lead me to believe that at least a part of the amnion may be eliminated by such a process of cell-substitution. I incline, therefore, to the views of Korschelt and Heider (’ 92 ), who hold that the envelopes are probably completely eliminated, and that the entire body-wall is derived from the ectoderm of the germ-band. 8o WHEELER. [VOL. VIII. If this be the correct view, it follows that the dorsal body- wall is formed in essentially the same manner in all insects —■ by a growth and meeting of the germ-band edges. This pro¬ cess is, therefore, remarkably simple and uniform compared with the processes whereby the envelopes are eliminated. The great variability in the latter case has been dwelt on by Graber (’88) in a paper devoted to dorsal-wall formation in the Insecta. After reviewing all the literature on the subject and contrib¬ uting many new facts, he proceeds to base a classification of the insects hitherto studied, on the “Keimhiillenzustande.” He finds some fault with the current classification on the ground that insects which systematists regard as closely re¬ lated often present great differences in their respective methods of dorsal-wall formation, whereas remotely related insects often agree very closely in this respect. Thus Lina and Hydro- philus differ more than HydropJiilus and CEcantJms in the processes whereby the dorsal-wall is formed. In considering Graber’s views, I may pass over the awkward and kakophonous nomenclature which he has introduced, to what I regard as his main error, viz. the superficial analysis of his subject. Graber’s term Keimhiillenzustande,” I take it, includes the formation of the envelopes as well as their condition preceding and during their elimination. Now I have attempted to show that there is nothing in the formation of the envelopes nor in the concomitant anatrepsis of the germ-band in the different insect orders to conflict with the current classification. Nor is there anything in the closure of the dorsal-wall in different groups — restricting this term to the confluence of the pleural edges of the germ-band — to support Graber’s conclusion. His state¬ ment must therefore be restricted to the elimination processes. That these are highly variable must be admitted, but they are probably of very little taxonomic value, as Graber would prob¬ ably have observed, had he attempted to account for the wide differences in allied forms and the agreement of remotely re¬ lated species. It is my ophiion that this high degree of varia¬ bility in the elimination process is to be traced to the same causes as the variability of the indusium, viz., the rudimental character of the envelopes. Up to the close of the diapause No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 81 the envelopes vSubserve a distinct function, but as soon as the germ-band has invested the yolk with its own ectoderm, they have become functionless, or rudimental. Long before this time, in fact ever since their completion, the envelopes show no traces of cell-division. Moreover, their involution into the yolk or complete shedding shows conclusively that their mor¬ phological value is at this time reduced to 7iiL Whether both envelopes are shed instead of being drawn into the yolk, or whether one is shed and the other drawn into the yolk, may depend to some extent on the ease with which the pleural folds can close without their temporary assistance. But which of these processes shall occur in a given insect is probably a matter of no vital importance to the embryo, and has prob¬ ably played no role in the struggle for existence. The involu¬ tion of the envelopes, it is true, may add assimilable matter to the embryo, but enough energy to counterbalance this addition is probably consumed in metabolizing the dead cells. Hence the adoption of this process may be of no greater advantage to the embryo than the complete sloughing of the useless en¬ velopes. The insect envelopes, therefore, present only another case of an organ which has become specialized for a particular function at the expense of its formative power. This same phenomenon recurs in insect ontogeny. During cleavage certain cells are segregated for the express function of yolk-metabolization (vi- tellophags), while the remaining cells go to form the blastoderm. Later the cells of the blastoderm separate into those of the germ-band proper and those of the specialized envelopes. Still later, if the insect be metabolic, another splitting occurs, a por¬ tion of the hypodermis being set aside in the form of the imag- inal disks to supplant the specialized primitive larval hypoder¬ mis. The formative material of the insect, like that of other organisms, thus undergoes a successive splitting into a special¬ ized and a comparatively non-specialized portion. The former, being incapable of metamorphosis, is cast off or broken down, while the latter persists until a new segregation takes place. The analogy of this process to that occurring in rhizomatous plants, Polyzoa, etc., need not be pointed out in detail. 82 WHEELER. [VOL. VIII. V. Neurogenesis in the Insecta. I. The Nerve-cord. The first traces of the central nervous system of Xiphidium make their appearance at a very early stage, before the blasto¬ pore is closed and while the envelopes are still incomplete. In this stage (Fig. 2) surface preparations made according to the methods given in the latter part of this paper, show a number of pale spots scattered over the procephalic lobes. They fre¬ quently occur also in the maxillary region, and, were it possible to remove the amnio-serosal fold without injuring the surface of the germ-band, would probably be found to extend still further caudad. The meaning of these spots is apparent when sections of embryos in Stages B-D are examined. In a transverse sec¬ tion (Fig. 25) through the middle of the abdomen of an embryo in Stage D, the ectoderm, which bulges out somewhat on either side of the median line, is seen to consist of two kinds of ele¬ ments. First, there are a few large, clear, polygonal cells with spherical nuclei (nb^, lying in the deeper portion of the layer; and second, a much greater number of small and more deeply stainable cells (^^.), differing in no essential respect from the cells forming the remainder of the ectodermal layer, such as the appendages. The latter cells have smaller, oval or cuneate nuclei, which appear to contain more chromatin than the large inner cells. While the small cells form a continuous layer, the large elements make their appearance singly or in small clusters, as seen in the figure. It is these pale clusters underlying the darker cells which produce the pale spots seen from the surface. The large pale cells may be called neuroblasts — since it is they that give rise to the purely nervous elements of the cord.^ I The term “neuroblast” was originally used by Whitman (’78 and ’ 87 ) to designate' the two offspring of the large posterior macromere of the Clepsine egg, which give rise by a process of budding to two rows of cells — the “ neural rows.” From these rows the nerve cord arises. His (’ 89 ) subsequently employed the same term “ neuroblast ” to designate such of the offspring of the “ Keimzellen ” as give rise directly by differentiation and not by further divisions to the ganglion- cells, or, to use Waldeyer’s term, to the neurons of the vertebrate central nervous system. More properly the term would have been applied to the “ Keimzellen ” themselves, and by mistake it has been thus used by at least one recent writer (C. L. Herrick, ’ 92 , p. 430, Fig. 10). Haeckel (Anthropogenie, 4th ed. p. 268^ ’ 91 ) uses ‘ neuroblast ’ in the sense of ectoderm in general. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. The remaining cells which cover the neuroblasts and extend down between them in the median line, give rise to purely integumental structures and may therefore be called dermato- blasts. The two thickenings of the ectoderm are to become the lateral cords (Seitenstrange). They extend from the anterior edge of the eleventh abdominal segment, just in front of the anus, to the mouth, where they diverge and pass without interruption into the brain. The groove which separates the lateral cords and which is very faint in Fig. 25, is the neural furrow (Primitivrinne). It appears soon after the closure of the blastopore and takes the place of this depression. It is deepest anteriorly. All the neural structures develop in an anteroposterior direc¬ tion, beginning with the brain; hence different stages in the development of the lateral cords may be studied in the same embryo. Fig. 26 shows a section passing through the first abdominal segment of the embryo from which the section in Fig. 25 was taken. Here we see a distinct advance in structure. The neural furrow is more clearly marked and the neuroblasts four in either lateral cord, have arranged themselves side by side in a regular layer in the deepest portion of the ectoderm. Over them the dermatoblasts {db}) also form a single regular layer, while the cells lying in the median line on either side of the neural furrow have grown more elongate. Sections further forward show essentially the same conditions — the neuroblasts which were at first differ¬ entiated as small clusters or as isolated cells, have arranged themselves throughout the anterior portion of the embryo as an even layer entad to the dermatoblasts. The further changes in the development of the nerve-cord, are brought about — first, by a proliferation of the neuroblasts ; second, by a proliferation of the dermatoblasts and a deepening of the neural furrow; third, by the development of the median cord; fourth, by the formation of the connectives and com¬ missures, and fifth, by the development of the neurilemmata. These changes which occur simultaneously may be described singly for the sake of convenience. As cross-sections show, the neuroblasts are arranged in from 84 WHEELER, [VOL. VIII. 3-5 longitudinal rows in either lateral cord. In surface view these rows may often be followed through one or two seg¬ ments as continuous strings of cells. I assume that there were originally four of these rows, but that owing to the pressure exerted by the developing appendages on the lateral edges of the cords and to a more rapid growth of the neuro¬ blasts than of the germ-band, the primitive regular arrange¬ ment has been considerably obscured. The neuroblasts are polygonal in outline from mutual pressure. When they divide, as they very soon do, their spindle axes are directed at right- angles to the surface of the body. As soon as one cell has been given off, the nucleus rests for a short time and then again divides in the same direction. This process continuing, a column of cells is budded off from each neuroblast and stands at right angles to the surface of the germ-band. The divisions do not take place simultaneously in all the cells although cor¬ responding neuroblasts in either cord will frequently be found in the same phase of caryokinesis, especially in the earlier stages of their proliferation. A section (Fig. 27) through the first maxillary segment of an embryo in Stage F shows that each of the eight neuroblasts has produced a row of daughter- cells. The large succulent mother-cells are evenly rounded on their outer surfaces which are overlaid by the dermatoblasts. Their inner faces are fiat or concave and in every case closely applied to the latest daughter-cell. The nuclei of the mother- cells are spheroidal and take no deeper stain than the pale suc¬ culent cytoplasm which surrounds them. The neuroblasts are in all essential respects typical proliferating cells like the ter¬ minal cells in plant-shoots and the teloblasts of annelids. The daughter-cells are at first characterized by their small size, cuneate outline and deep stain. Their nuclei are con¬ siderably flattened, probably from mutual pressure. These characters are retained by the daughter-cells till they have been pushed some distance from the neuroblast by later offspring, when they become larger and considerably paler and assume the appearance of definitive ganglion cells {g-). Turning now to a somewhat older embryo (Stage G, Fig. 28) we see that the columns of daughter-cells have greatly increased No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 85 in length, while the neuroblasts remain to all appearances un¬ altered. The increase in the number of daughter-cells is so great that they are forced to arrange themselves in several rows. In the figure this is best shown in the progeny of the innermost neuroblasts, and the tapering columns there formed may be regarded as typical. In my preliminary note (’9i®) I held that the daughter-cells themselves divide to form the multiple rows in each pillar. I in¬ cline to think that I was mistaken on this point. The daughter- cells probably never divide but are directly converted into ganglion cells. All reproductive powers seem to be confined to the neuroblasts. Some of the nuclei of the daughter-cells exhibit peculiar chromatic structures which I may have mistaken for caryokinetic figures ; this being an easy error to make in the case of small cells killed by means of heat, since the achromatic portions of the spindles are obliterated by this method. The last stages in the proliferation of the neuroblasts are shown in Fig. 31, which is taken from an advanced embryo (Stage J). The columnar arrangement is no longer visible since the individual cells are now converted into the de¬ finitive ganglionic elements. On the outer periphery of the ganglia, however, neuroblasts are still to be found, and ex¬ tending from them short series of small flattened cells their latest progeny, still distinguishable from the ganglion cells by their deeper stain. It will be noted that these cells, which like their precursors will become ganglion cells, are no longer budded off at right angles to the surface of the nerve- cord but parallel to it, a condition undoubtedly due to a lack of space. Finally the neuroblasts stop proliferating and shrink to the size of their progeny. Their chromatin then shows signs of senility. Beyond this point I have been unable to trace them satisfactorily. They are probably broken down and absorbed by the growing ganglion cells. Some of them may persist as ganglion cells of a particular character and function, though I deem this improbable. The dermatoblasts play an important part in the develop¬ ment of the ventral nerve-cord, as will be seen by returning to the younger stages. We left these cells as a layer covering 86 WHEELER. [VOL. VI11. the neuroblasts and continuous laterally with the general ecto¬ derm. In the median line they extend to the deepest portion of that germ-layer in the form of a few compressed cells (Fig. 26, db). These compressed cells form the walls and bottom of the neural furrow. The proliferation of the neuroblasts has caused the lateral cords to bulge out enormously (Fig. 27), so that the dermatoblastic layer becomes stretched and attenuated. Such divisions as occur in the cells of this layer seem to be confined to the outer surface and do not extend into the furrow. The spindle axes lie parallel to the surface, as shown at nh. The bulging of the lateral cords naturally brings about a deep¬ ening of the neural furrow, since the cells at its bottom have a fixed attachment. At this point in Fig. 27 there is seen a triangular cell-mass, capped by a single large element (innb.), a true neuroblast which resembles in nearly all respects the neuroblasts of the lateral cords. Its more pyramidal outline is obviously the result of its position between the converging walls of the furrow. To the same mechanical cause is due the shape of the cell-mass, which consists of the heaped up daughter-cells of the neuroblast. Inasmuch as the proliferating cell occurs in the median line, and together with its offspring and the dermatoblastic cells of the median furrow, is equivalent to the “ Mittelstrang ” of authors, I shall call it the median- cord neuroblast. Its exact origin I have not been able to determine. To judge from the number of cells which it has given off it must have begun to proliferate at about the same time as the lateral-cord neuroblasts. There can be no doubt, it seems to me, that it originated as a polygonal ectoderm cell like the lateral cells seen in Figs. 25 and 26, but whether it was originally median in position or arose unilaterally I am unable to decide. The pale surface spots of embryos in Stage B show that neuroblasts are arising at a time when the blasto¬ pore occupies the position of the neural furrow and hence, if the median cells are median in position from the first, they must arise somewhat later than their sister neuroblasts. There is one important difference in the arrangement of the mother-cells of the lateral and median cord. Whereas the for- mer, as has been stated, form continuous though irregular rows No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 87 from mouth to anus, the latter constitute an interrupted series between the same two points. They are single, isolated cells, which occur only intersegmentally. That such is their distribu¬ tion may be distinctly seen in frontal sections like the one rep¬ resented in Fig. 30. This section passes through the first to fifth abdominal segments at the level of the median cord neuroblasts {innb), which are seen to lie distinctly between the segments, where the walls of the neural furrow dilate at intervals for their accommodation. At first the daughter- cells are given off in the same direction as those of the lateral cords, but soon the triangular space to which they are confined will no longer contain the older cells of the series and these are pushed along the floor of the neural furrow. This produces an angular flexure in the cell-column, but later the whole mass, including the neuroblast, assumes a horizontal position. This change in the position of the median cell-mass is seen to have taken place in the median sagittal section from an embryo in Stage G (Fig. 29). The neuroblast i^nmb) is in each segment directed caudad, while the mass of small and deeply stainable daughter-cells {mg) is wedged in under the commissures. The section passes through the sub-oesophageal ganglion, which consists of the fused ganglia of the mandibular and both maxil¬ lary segments iind. g; mx. g ^; mx. g-), and through the pro- and mesothoracic ganglia Transverse furrows (f. ; i.g^)j which I shall consider later, separate the unfused ganglia from one another, and as the median cord cells lie in front of these furrows, they must be regarded as belonging not to the intersegmental region of the ectoderm, but to the posterior portions of the separate ganglia. Each ganglion possesses a median cord neuroblast, so that, beginning with the mandibular, which is the first ganglion in the nerve-cord proper, and ending with the tenth abdominal, there are in all sixteen median mother-cells. Each of these, after producing its quota of ganglionic elements, deteriorates in the same way as the mother-cells of the lateral cords. The development of the Punktsubstanz may be readily fol¬ lowed in XipJiidium. It arises in each ganglion as two separate masses. Each of the daughter-cells of the lateral neuroblasts 88 WHEELER. [VOL. VIII. sends out a cytoplasmic process which soon ramifies. The mass of fibres thus formed increases in size very probably by the addi¬ tion of further ramifications till the Punktsubstanz is definitely established as a scarcely stainable body, lying on either side of the median line in the deepest portion of the lateral cord (Fig. 27,/.j.) In its earliest stages the formation of the substance is easily followed, but very soon the felted fibres become too dense for analysis by ordinary methods of investigation. It is only after a distinct mass is formed in either half of a ganglion that the longitudinal commissures, or connectives as they are best called, make their appearance, and unite the hitherto iso¬ lated centres in two longitudinal series. Very soon the trans¬ verse commissures, or commissures proper, of which Xiphidiumy like all other insects, has two in each segment, make their appearance. The daughter-cells of the median cord neuroblasts take no part in the formation of the anterior commissure. Whether they contribute fibres to the posterior commissure or not, I must for the present leave undecided. I have seen no evidence in the median cord of a distinct and isolated Punkt¬ substanz centre, such as is described and figured by Graber for some Coleoptera ((’ 90 ) PI. V, Fig. 66). I deem it more probable that in Xiphidium the commissures arise wholly from the Punktsubstanz masses of the lateral cords. Both commissures are distinctly seen in cross section in Fig. 29. The connectives and commissures incompletely divide the cellular portion of each ganglion into five parts, — two lying laterad to the connectives and a median series of three smaller portions separated from one another by the two commissures. The former may be called lateral gangliomeres, and the three median portions respectively the anterior, central, and poste¬ rior gangliomere.^ Of the median divisions the posterior is distinctly the largest from the first. This is due to its being formed in great part by the progeny of the median neuroblast, whereas the anterior and central gangliomeres consist of a comparatively small number of cells, contributed by the lateral neuroblasts. 1 These are equivalent to Graber’s laterale Zellenlager, vorderes, centrales, and hinteres Medianlager. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 89 It is not till after the commissures and connectives are formed that the inter-ganglionic regions become clearly marked out. Throughout the early stages, in fact till the embryo reaches the ventral surface of the egg (Stage J), the ganglia are as long as their respective segments and are separated from one another only by the intersegmental constrictions. These have grown very deep in Stage G, especially in the thoracic and abdominal regions. In the median line, as shown in sagittal section (Fig. 29, i.g^^ they form deep tubular ingrowths which may be called furcal pits. Since these pits are median in position they are to be regarded as differentiated interganglionic portions of the neural furrow. They therefore belong to the median cord. They are not found between the mandibular and first maxillary, nor between this and the second maxillary ganglion, and are also wanting between the eighth and ninth, and ninth and tenth abdominal ganglia. Evidently their absence in these cases is due to early fusion to form the infraoesophageal and last abdominal ganglion. In the thoracic segments the furcal pits are converted into chitinous apode- matous structures which give attachment to some of the leg- muscles. It is interesting to note that in the abdomen also furcal pits are distinctly developed as late as Stage K. Here, too, they serve for the attachment of a few weak muscle-like structures, which run from their tips to points in the adjacent abdominal wall, perhaps corresponding to the insertions of the rudimental appendages. Later both muscle-like cords and ab¬ dominal furcae disappear, —the latter by a very simple process. It will be remembered that at this time the embryo is growing in length and continually covering more and more of the yolk. The tail end is practically fixed at the lower pole of the egg, while the head slowly moves upwards. The body-wall is thus stretched in both a longitudinal and lateral direction. Hence the intersegmental constrictions, so deep in Stage J, must gradually become shallower, and the furcal pits, which are nothing but portions of these constrictions, are drawn out from between the connectives to form part of the sternal integu¬ ment. The stretching not only draws out the folds in the em¬ bryonic body-wall, but also reduces it to a much thinner layer 90 WHEELER. [VOL. VIII. of cells. The length of the individual segments is thereby greatly increased and the nerve cord, which is firmly attached in the infraoesophageal region and more loosely in the terminal abdominal segments, is compelled to lengthen. The separate ganglia, besides assuming a somewhat fusiform outline, are scarcely affected by this traction, whereas the connectives are drawn out into thin threads denuded of all ganglionic cells and covered only by the neurilemma. The presence in the abdomen of temporary furcal pits cor¬ responding to the persistent furcas of the thorax admits of an easy explanation, if we take these structures to be correlated with the development of ambulatory appendages. The tem¬ porary abdominal appendages have usually been regarded as the rudiments of once functional walking-legs, and they are still so well preserved in the Orthoptera that it need not surprise us to find traces of correlated structures which served for the attachment of some of their muscles. The progeny of the median neuroblast together with the interganglionic portion of the neural furrow have been ac¬ counted for; the former becoming the posterior gangliomere, the latter a portion of the sternal integument ; but I have not yet accounted for the remaining portion of the median cord — viz. the intraganglionic walls of the neural furrow. This portion of the groove is crossed by the two commissures and separates those portions of the lateral cord which will ulti¬ mately constitute the anterior and central gangliomeres. Its cells are of an epithelial nature. Those of the opposite walls of the furrow become applied to one another by the swelling of the lateral cords. The lumen is thereby oblit¬ erated though its walls are still continuous on the outer surface of the ganglion with the integumental ectoderm. The two lips of the furrow finally fuse and the ganglion together with the portion of the furrow included between its two halves is liberated from the ectoderm. It is these epithelial walls thus set free from the integument which appear to give rise to the outer and inner neurilemmata. Both these neural envelopes are ectodermal ; there are no traces of mesodermal structures taking any part in their forma- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 91 tion and it seems to me that they can have only two possible sources — they either arise from some of the progeny of the neuroblasts or from the intraganglionic portion of the median cord. I deem it highly improbable that they should arise from the former source, since the daughter-cells of the neuroblasts have every appearance of being early specialized as ganglion cells. Furthermore, the cells of the neurilemmata when they definitely appear, closely resemble the cells of the neural fur¬ row both in size and in the great avidity with which they take the stain. The outer neurilemma covers first the inner surface of the ganglion — then the outer or neuroblastic sur¬ face;— the thin cellular membrane apparently progressing ing laterad in either case and meeting near the origin of the nerve-trunks. The inner neurilemma, which envelops the Punktsubstanz is completed before the outer envelope. Histo¬ logically both envelopes resemble each other in every respect. The fusions of ganglia in the nerve-cord take place gradually and may be easily followed in Xiphidiuni. Several stages of these fusions are represented in Fig. VII, A-D. In Fig. A, the nerve-cord is shown much as it appears in Stage F. The ganglia form an unbroken series from mouth to anus. The connectives are very short and not as yet distinguishable from the surface. Fig. B, is taken from an embryo just turning the lower pole (Stage H). Here the mandibular and two maxillary ganglia, and also the three terminal abdominal ganglia still remain as in the preceding stage, while the other ganglia are being drawn apart by the stretching of the embryo, so as to show their short connectives. In Fig. C, the suboesophageal and last abdominal ganglia are established as two fused masses. The number of ganglia comprising each of these masses may still be easily determined by counting the commissures. It will be noticed that in this stage the first abdominal is closely approximated to the metathoracic ganglion and that the second and third abdominal also lie close together. Between the other ganglia the connectives have lengthened. In the later stage represented in Fig. D, the connectives are still longer; the first abdominal ganglion has fused with the metathoracic, and the second and third abdominal form a single mass. 92 WHEELER. [VOL. VIII. These fusions become more intimate as the time for hatching approaches, so that the ventral cord finally consists of only ten ganglia instead of sixteen, the original number. A-D. Diagrams of four consecutive stages in the development of the brain and nerve-chain of the Xiphidium embryo. I, cephalic; II, thoracic; III, abdominal region ; st., stomodeeum ; an., anus ; e, optic plate ; pd{o.g), first protocerebral lobe, or optic ganglion ; pc^, pc^, second and third protocerebral lobes ; dc, deutocerebrum ; tc., tritocerebrum ; i-i6, the sixteen postoral ganglia ; po.c., postoral commissure ; fp., furcal pit ; ac., anterior; pc., posterior ganglionic commissure ; ag., anterior ; pg., posterior ; eg., central; Ig., lateral gangliomeres. The description of the ventral nerve-cord of Xiphidmm here given applies equally well to the other Orthoptera which I have studied {Blatta gennanica, MelariopliLS feimtr-riibritDt). The No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY, 93 points in which the Blattid and Acridian nerve-cord differ from that of the Locustid are so insignificant that I need not burden the reader with their enumeration. I will stop to mention only two peculiarities in Blatta. Here I fail to detect the pale spots in the “slipper” stage of the germ-band, and sections show that the neuroblasts do not differentiate as early as they do in Xiphidiiim. They are, however, readily detected in late stages, when they stand out with even greater distinct¬ ness than in the Locustid. The median cord neuroblasts, ' though present and occupying positions corresponding to their homologues in Xiphidmm, are more difficult to trace, probably on account of the smaller size of the embryo. Neuroblasts, or cells of a similar character have been de¬ scribed and figured by a number of investigators of Arthropod development. Perhaps the earliest mention of these cells is to be found in Reichenbach’s beautiful Astacits monograph (’86), where the nerve-cord is described as consisting in an early stage of two kinds of cells — a few large pale elements arranged in a single layer and confined to the periphery, and a much greater number of small and more deeply stainable cells forming the bulk of the ganglia. The developing ganglia of the cray-fish resemble the ganglia of the Orthoptera in many particulars. The number of large cells in the lateral cords in Reichenbach’s figures (notably his figures 114-133) is 3-6, the average being 4 or 5, the same as in XipJiidiiim^ Blatta^ etc. Furthermore the ganglia of Astacus show a foliated arrange¬ ment of. the smaller cells, which is not unlike the condition seen in the older ganglia of the Orthoptera. Some of the figures (188 and 189 for example) show a single neuroblast¬ like cell surmounting the median cord cell-mass. There are, however, two points in Reichenbach’s work, which throw some doubt on the homology of his large cells with the neuroblasts of the Orthoptera. First, Reichenbach neither figures nor deseribes these cells as dividing to form ganglion cells. This negative observation, however, loses much of its force when we consider that caryokinetic figures are singularly absent from all of Reichenbach’s figures, excepting his surface views of young embryos, and when we recall the fact that amitotic 94 WHEELER. [VOL. VIII. division is a very general phenomenon in the Crustacea (ac¬ cording to Carnoy, ’ 85 ). A more serious objection to the homology under consideration is Reichenbach’s statement that the large cells ‘gehen schliesslich in die von Leydig, Dietl, Krieger und anderen beschriebenen, grossen Ganglienzellen iiber.’ This, too, is an objection only if the neuroblasts really degenerate, a point on which I am still doubtful. Nusbaum found huge succulent cells in the young nerve cord of the embryo My sis (’ 87 ). He compares them with the large cells of Reichenbach and believes that they have a similar fate. Similar cells were also observed in the brain of Onisctis ^niLrariiis ‘‘pendant les stades relativement jeunes.” On one point only does he add to Reichenbach’s observations: he depicts (Fig. 78) a caryokinetic figure, which from its size and position must be referred to one of the large cells. Its spindle axis is directed at right angles to the surface of the body. This observation, small and incidental as it is, would tend to show that the large cells proliferate as in the Orthop- tera. I am inclined to think that a renewed study of the Crustacean nerve cord will show that the ganglion cells are budded forth from the large cells and that these are equivalent to the insect neuroblasts. Korotneff (’ 85 ) was the first to find gangliogenic cells in the Insecta. At p. 589 in his description of the Gryllotalpa embryo he says: “ Einige der Ektodermzellen, welche die Nervenauftreibung bedecken, fangen an zu w'-achsen, ihre Kerne vergrossern sich bedeutend und zeigen dabei eine karyo- kinetische Figur. Grosstentheils sind diese Zellen (Ganglien) so angeordnet, dass einer einfachen platten Ektodermzelle eine wachsende Neuroektodermzelle folgt. Hat sie eine be- stimmte Grosse erreicht, so sinkt jede wachsende Zelle in die Tiefe des Ektoderms und wird von den benachbarten, unver- anderten Zellen bedeckt. Jede Ganglienzelle theilt sich dabei, eine ganze Folge von neuentstandenen Zellen bildend, nur an der Fig. 60 ist leicht zu unterscheiden, welche Gruppe von Zellen der oben gelegenen Ganglienzelle entspricht. Durch eine solche Vermehrung von Zellen wird der Nervenstrang mehr und mehr in die Hohe getrieben.” The Fig. 60 referred No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 95 to in this description is taken from a stage corresponding to my Fig. 28. Both in this figure and in Fig. 61 he represents four neuroblasts in one of the lateral cords. Korotneff seems not to have seen the early stages of proliferation. In the developing nerve-cord of Doryphora I observed (’ 89 , p. 366) that “the outer layer of cells continuous with the hypo- dermis stands off somewhat from the ganglionic thickenings, leaving a space which is in early stages occupied by several large, clear, oval cells which divide rapidly by caryokinesis, and might be called ganglioblastSy as the products of their divisions reinforce the mass of ganglion cells.” In my figures the polar- axes of the neuroblast spindles lie parallel to the surface of the ganglia. Re-examination of my preparations has convinced me that this observation is essentially correct. I find also that the newly-formed daughter-cells of the neuroblasts occasionally divide caryokinetically and thus give rise to further generations of daughter-cells. The daughter-cells are not budded forth in regular rows, but very irregularly. I am not sure that I can distinguish the median cord neuroblasts in Doryphora, though I believe that I have detected homologous structures. In my figure 72 I represented circular intersegmental patches in the median line between the lateral cords. Closer examination shows these to be clusters of cells of the same appearance and dimensions as the lateral-cord neuroblasts. They are very clearly brought out by Graber in his figures of Hydrophilus (’ 89 , Figs. 40, 41, and 43, PI. Ill) and are described as taking part in* the formation of the posterior gangliomere of each ganglion. I doubt whether the large cells constituting the posterior gangliomere of Periplaneta in Miall and Denny’s Fig. 43 (’86) to which Graber refers, are to be regarded as the equivalents of the median-cord clusters in Doryphora and Hydrophilus. Peripla^teta very probably has in each segment only one median-cord neuroblast, which atrophies before the close of embryonic life, and the large cells in Miall and Denny’s figure probably arise from the daughter-cells and are therefore merely large ganglionic elements. Graber figures and describes (’ 89 , p. 47, PI. X, Fig. 130) a cross-section through an abdominal ganglion of a Melolontha 96 WHEELER. [VOL. VIII. embryo in which he finds a neuroblast in either lateral cord and three symmetrically arranged cells of the same character in the median cord. Similiar median cells were seen in LiLcilia. He refers to Korotneff’s observations on Gryllotalpa and states that he has found the lateral ‘‘ganglionare Grosszellen ” in Lma. He is inclined to regard them as a widely occurring differen¬ tiation of the ectoderm. In a subsequent paper (’ 9 o) Graber describes and figures a foliated condition of the ganglia in the nerve-cord of Steno- bothnis. In Fig. 52 the neuroblasts are distinctly seen, and in one lateral cord five, in the other four pillars of cells may be distinguished. So far as the neuroblasts are concerned, he cannot be said to have added anything to Korotneff’s account. Nusbaum (’ 9 i), in a recent Polish paper on the development of Meloe\ figures neuroblasts in the lateral cords. They are frequently represented in mitosis — the spindle-axes being in some cases perpendicular to the surface of the ganglia (Figs. 94, 95) while in others (Fig. 107) they are parallel to the surface, as in Do 7 yphora. Such portions of the text as were translated for me contained nothing new on these structures. Viallanes, in two recent papers (’ 90 ® ’ 90 ^) on the structure of the nervous system in the embryo Mantis^ comes to con¬ clusions agreeing with my own, which were arrived at inde¬ pendently. His observations on the neuroblasts may be briefly summarized in his own words (’ 90 ^, p. 293): “A Tori- gine le bourrelet primitif n’est qu’un simple epaississement de rexoderme, c’est-a-dire une region de ce feuillet dont les cellules sont devenues columnaires et ont augmente de volume. Bientot ces cellules se multiplient et se divisent en deux couches. Tune superficielle {dermatoghie), Tautre profonde ga7iglioghie). A une periode plus ou moins tardive, suivant la region consideree, la couche des cellules dermatogenes se separe de la couche des cellules gangliogenes et devient I’hypo- derme. Les cellules gangliogenes en se multipliant donnent naissance aux cellules ga 7 zgli 07 i 7 iatres.'' Viallanes’ figures do not show a regular arrangement of the cells budded forth from the neuroblasts, and he has not de¬ scribed the neuroblasts of the median cord, probably because No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 97 his attention was concentrated on the structure of the brain. He has observed the degeneration of the gangliogenic cells, or neuroblasts. In a late stage (’ 90 ^, p. 301), he says Ils montrent des signes evidents de decrepitude; beaucoup des cellules gang- liogenes ont deja disparu, les autres sont en voie d’atrophie.” Our knowledge of the median cord cannot be said to have made much advance since this structure was first described by Hatschek (’ 77 ). While all writers agree that it originally ex¬ tends as an uninterrupted structure from the mouth to the anus, there is wide difference of opinion respecting the ultimate fate of its inter- and intraganglionic portions. Hatschek (’77), Tichomiroff (’ 82 ), and Korotneff (’ 85 ) maintain that the inter- ganglionic portions remain attached to the integument when the nerve-cord is liberated and that they ultimately disappear. Ayers (’ 84 ) on the other hand holds that the whole median cord is liberated from the ectoderm, but does not affirm that the interganglionic portions form a constituent part of the ganglia. Graber (’ 90 ) has very recently come to a conclusion which differs from the views hitherto advanced. With Ayers he holds that the interganglionic portions of the median cord are delaminated from the ectoderm along with the intraganglionic portions, but he goes further and claims (p. 103) that das Zellenmaterial des interganglionalen Mittelstranges, zum Theil wenigstens mit den Ganglien vereinigt wird, oder mit anderen Worten, dass eine Vergrosserung des ganglionalen Mitteltheiles auf Kosten des interganglionalen erfolgt.” As will be inferred from the above descriptive paragraphs, I hold to Hatschek’s view that the interganglionic portions of the median cord take no part in the formation of the ganglia but are drawn out from between the connections and constitute a portion of the sternal integument. Graber’s re¬ searches on this portion of the nerve cord are limited to the Coleoptera and as the insects of this order certainly differ to some extent from the Orthoptera in the formation of the nervous system, I have no grounds for doubting the correctness of his observations. I believe, however, that Ayers’ account of the median cord in CEcantJms is open to criticism. After 98 WHEELER. [VOL. VIII. clearly implying that the median cord is set free from the ectoderm along its whole extent he remarks (p. 252): “Be¬ tween the successive pairs of ganglia the median ingrowth atrophies, and at the time of the closure of the dorsal wall of the body there is seen between the connecting cords of two adjacent pairs of ganglia, a small triangular or cylindrical mass of cells, concerning the fate of which I am not absolutely certain. I believe, however, that they go to form a part of the internal skeleton. The chitinous rods in the thoracic region to which the muscles of the legs and wings are attached probably arise from the remnants of the median invagination, but in the abdominal region they may disappear entirely with¬ out giving rise to such structures.” If I understand this passage correctly, Ayers implies that the chitinous rods are originally interganglionic portions of the median cord. But if this is the case, how can the median cord separate completely from the ectoderm unless we are to suppose that there is a reunion of the interganglionic portions with the integument to form the endoskeletal structures } The chitinous rods are directly continuous with the chitin of the integument so that until observations are forthcoming to show that portions of the integument can loosen and pass into the body-cavity and subse¬ quently reunite with the integument, I must regard Ayers’ account as inadequate. I am still in some doubt as to the exact origin of the commis¬ sures. Grassi (’ 84 ), Ayers (’ 84 ), Heider (’ 89 ), and Graber (’ 90 ), all maintain that the commissural fibres arise from the median cord cells. A priori, there are no reasons why the daughter- cells of the median neuroblast should not send out processes to form Punktsubstanz and thus form a commissure. From the position of these cells, however, I regard it as highly improbable that anything but the posterior commissure could be formed in this way. The isolated Punktsubstanz masses in the Coleopte- ran median cord in Graber’s and Heider’s figures may arise from cells equivalent to the daughter-cells of the median neuro¬ blasts of the Orthoptera. It is very improbable that the der- matoblastic cells which form the walls of the median cord in the region of the anterior commissure, and which I regard as No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 99 non-nervous, should take part in forming the fibres of that structure. Regarding the origin of the neurilemmata in insects, there is still considerable doubt. The inadequacy and inconsistency of Nusbaum’s observations on Blatta (’ 83 ) have been sufficiently pointed out by Eisig (’8?) and Korotneff (’ 85 ). Nusbaum derived the median cord (which, by the way, he did not recognize as the median cord) from the entoderm, and compared it with the vertebrate chorda. So far his observations and conclusions were erroneous, but he derived the inner and outer neurilemma from the cells of this “chorda”—an observation which agrees essentially with my own. Korotneff’s view that the neurilemmata arise from migrant mesoderm cells has not been confirmed by recent writers, who are inclined to derive these envelopes from the ectoderm (Heider, ’89 ; Graber, ’ 9 o). Though I venture to say that my own observations are somewhat more definite than those hith¬ erto published, I cannot regard them as in any way final. 2. The Bram. In the following account of the Xiphidium brain, I shall use the nomenclature employed by Viallanes in his recent papers (’ 90 ®, ’ 90 ^), since his studies on the brain-development of Mantis religiosa agree very closely with my own. Before passing to a description of my sections I would refer the reader to* the diagrammatic figure (VII) which represents the main points in the structure of the embryonic brain. Here it is seen that the ventral nerve-cord bifurcates just in front of the mandibular segment and passes on either side of the mouth, where it forms two successive pairs of ganglia. The posterior of these {tci) is the tritocerebriLm, or third brain seg¬ ment. Its two halves are united by the infraoesophageal com¬ missure^ shown in the figure as a broad white band connecting the Punktsubstanz-masses of the ganglia. The anterior pair of swellings {dci) constitute the deutocerebrum, or second brain segment. From this portion the antennae are innervated. Further forward the deutocerebrum passes into a large paired lOO WHEELER. [VOL. VIII. supraoesophageal mass, the protocerebnim^ or first brain seg¬ ment, which constitutes the greater portion of the brain. Each of its halves may be separated into three lobes ; the first, or outermost lobe i^pc^ forms the optic ganglion of the larva and imago, while the second and third lobes {pc~y pc 3 ) ultimately form the bulk of the brain proper. The third lobe is united with the contralateral lobe by the broad siipra- (ssophageal commissure. Such is the structure of the Orthopteran brain reduced to its simplest terms. It may now be considered a little more in detail. Like the nerve-cord, of which it is simply a modified portion, the brain arises from neuroblastic cells. These first make their appearance in clusters (the spots seen on the procephalic lobes in Fig. 2). Later they form a single layer of proliferat¬ ing centres continuous with and in every way comparable to the neuroblasts of the ventral nerve-cord. Like the latter they are covered externally by a layer of dermatoblastic cells. That the deuto- and tritocerebral ganglia are strictly homodynamous with the ganglia of the. nerve-cord is clearly shown in Xiphidium. In the first place these brain segments are directly continuous with the segments of the cord; second, they have at first the same size and shape as the latter, and third, they present on the average four neuroblasts in cross- section on either side. The suppression of the median cord in the deutocerebrum (if it be not drawn forward into the proto¬ cerebrum), is perhaps sufficiently explained by the presence of the stomodaeal invagination. A partial suppression of the median cord in the tritocerebrum may be due to the same cause. The infraoesophageal commissure is perhaps the morphological equivalent of both the commissures of a ventral ganglion. The early clustered condition of the neuroblasts is seen in Figs. 32-34 at 7 tb. At the edges of these cross-sections a rounded mass of pale cells (^pc^ distinctly marked off from a more deeply stainable layer which encloses it on nearly all sides. This mass, the future optic ganglion (first proto¬ cerebral lobe), is delaminated from the ectoderm at a very early stage. The cells of the mass agree with the neuroblasts No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. lOI in their slight affinity for stains; they differ in the more elongate shape of their nuclei and cytoplasm. The dark layer enclosing the optic ganglion on all except its innermost face is the optic plate and will give rise to the compound eye. Passing towards the median line in these sections (especially in Fig. 33) two other thickenings may be distinguished pc^) — the second and third protocerebral lobes. The cross-section, Fig. 36, runs through the labrum of an older embryo (Stage F) and shows a considerable advance in the structure of the brain. The three protocerebral lobes are distinctly marked out. In the second and third, the neuro¬ blasts have arranged themselves in a row and have budded forth strings of ganglion cells. In the first lobe (/<;i [o.^.]) no teloblastic arrangement is ever present. The cells are small and narrow and early assume a radial arrangement around the Punktsubstanz core at the base of the mass. The cells of the optic plate, which stand away from the surface of the ganglion, already show a tendency to differentiate in that they have become smaller and narrower than, the dermatoblasts covering the two other lobes of the protocerebrum. At the juncture of the second with the third lobe, several large dermatoblastic cells are intercalated (Ir'I.). They are continuous with the dermatoblasts covering the second protocerebral lobe. This intercalated mass is called by Viallanes the bourrelet ectoderm- ique intraganglionnaire. I shall call it the intraganglionic thickening. A still more advanced stage in the development of the brain is seen in Fig. 37 (Stage G). This section passes above the base of the labrum. Owing to the active proliferation of the neuroblasts, the mass of the protocerebrum is greatly aug¬ mented. The Punktsubstanz has made its appearance as a confluent mass.^ The optic plate is much thickened and its small cells are about to arrange themselves to form the ommatidia. The intraganglionic thickening {igll) presents an interesting appearance. The edge of the optic plate is united with the inner edge of the optic ganglia, but between it 1 1 have seen nothing to corroborate Cholodkowsky’s view (’91) that there are at first three distinct and separate pairs of Punktsubstanz masses in the brain. 102 WHEELER. [VoL. VIIL and the second protocerebral lobe there is a fissure {w) which extends in some distance. Examination of a number of succes¬ sive sections and stages has convinced me that this fissure is not the result of artificial rupture during sectioning but that it is brought about by an infolding of the intraganglionic thicken¬ ing. The shape and position of the involuted mass may be clearly seen from the surface in Stages H and I (see Figs. 8 and 9, igl). In the frontal section (Fig. 39) are shown the relations of the protocerebrum to the outer brain-segments and to the ventral cord. Only a small portion of the optic plate (e) is cut. Beneath it lies the optic ganglion (/<;i [<^g-])) the small cells of which contrast with the large cells of the brain proper. The second protocerebral lobe (pc^) still contains many neuro¬ blasts which are budding forth their last progeny. The older daughter-cells have already assumed an irregular arrangement. The brain is separated from the attenuated dermatoblastic, or integumental layer (dd.) beneath which the outer neurilemma (enl.) is forming. The inner neurilemma (ml.) envelops the Punktsubstanz portions of the brain. The broad supraoesoph- ageal commissure (s. cm) connects the third protocerebral lobes of the two sides. As shown in the figure, the deutocerebrum is distinctly praeoral. At an is shown the point where the antennary nerve leaves the fibrous portion of this brain seg¬ ment. Caudad to the deutocerebrum lies the tritocerebrum, a pair of somewhat smaller ganglia united by the infraoesoph- ageal commissure. It is this segment which according to Viallanes innervates the labrum and the frontal ganglion. Be¬ sides the supra- and infraoesophageal commissures and the connectives which arise in the third protocerebral lobes and traverse the deuto- and tritocerebrum to pass into the man¬ dibular ganglion and thence through the nerve-cord, I may call attention to two other masses of Punktsubstanz which lie in front of the supraoesophageal commissure. These are shown at/, in the figure. They appear to be connected by a small band running parallel to the robust supraoesophageal commis¬ sure. I did not succeed in finding these connected Punktsub¬ stanz masses in all embryos of this stage, and as they were not No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 103 seen in later stages, their morphological value as indicating the presence of another segment in front of the protocerebrum is somewhat doubtful. The three divisions of the protocerebrum may still be rec¬ ognized in the transverse section (Fig. 40) of an embryo which has passed beyond Stage J. In the median line at me lies a rounded and compact mass of cells which I regard as the prae- oral representative of the median-cord cell-masses of the ventral cord. A large cell, which I take to be a neuroblast, lies at the outer periphery of this median cell-mass at a younger stage. Structures at a corresponding position in the brains of other insects {Hydrophilus^ Mused) and likewise comparable to the median-cord have been described by Heider (’89) and Graber (’89, p. 49). It is not improbable that the brain neurilemmata may have their origin near this median cerebral mass, just as the neurilemmata of the ventral cord probably arise from the non-ganglionic portions of the neural furrow. Two other interesting structures are shown in Fig. 40: the intraganglionic thickening {igl)^ now completely separated from the integument and lying as a deeply stainable mass wedged in between the optic ganglion and the second protocerebral lobe, and a peculiar pale thickening at the edge of the optic plate {tJi). In later stages it is very difficult to locate the intraganglionic mass, so that I am unable to decide whether it atrophies or persists in a modified form as a portion of the brain. The re¬ searches of Viallanes on Mantis would seem to lend great probability to the former alternative. The thickening at the lateral edge of the optic plate is con¬ stant in Stages G to J and somewhat later. It soon entirely disappears, without taking any part in the formation of the eye, so far as I have been able to observe. Does it represent an abortive ocellus ? The fully established optic nerve is shown in Fig. 41 ((?. n). In a much earlier stage it may be found as a delicate band of cells connecting the posterior edge of the optic ganglion with the optic plate (Fig. 38, 0. n). I agree with Viallanes, that it seems to arise from the ganglion and to grow outwards into the ommatidial layer; for there is from the first a sharply defined 104 WHEELER. [VOL. VIII. intercepting membrane between the nerve and the plate, whereas the nerve passes without interruption into the ganglion. But I am led to lay little stress on these appearances by the researches of Watase (’ 90 ) and Parker (’ 90 ) on the adult ommatidia of a number of Arthropods. They have shown in a very convinc- ino" manner that each retinula-cell is the termination of an optic nerve fibre. The retinulae are undoubtedly modified optic plate cells; and judging from recent observations on per¬ cipient cells in other forms (vertebrate eye, ear, olfactory nerves, v. Lenhossek’s observations on Liimb 7 'icus (’ 92 ), etc.), we must suppose that the nerve fibres grow out from the retinula-cells,' pass through the optic nerve and enter the ganglion. Such prolongations from all the retinulae would be amply sufficient to form the optic nerve, although it is probable that some of its fibres are centrifugal prolongations from optic ganglion cells. It has been suggested that the optic nerve may be established at a very early stage, when the optic plate and optic ganglion are still closely applied to each other {vide Figs. 32-34), and that the nerve may not become visible till the two Anlagen separate with further development. But I do not think this is the case in Xiphidhim. Sections like Fig. 36 show a distinct separation of the optic plate and ganglion in the region of the future optic nerve; and Viallanes has made an exactly similar observation on Mantis. The sympathetic nervous system arises in part at least from the dorsal median wall of the oesophagus. At three separate points (Fig. 61) the ectoderm becomes thickened and its outer cells enlarge and assume the character of ganglion cells. The most anterior of these thickenings (^f. g.) is the frontal gang¬ lion. It arises just behind the base of the labrum. The two other thickenings which are placed further back {rg^y rg^) are the second and third visceral ganglia. I have not followed the development of the nerves which unite these ganglia and ramify from them. Concerning the origin of the peripheral nervous system I have no positive data. In a few cases I have seen appearances which led .me to believe that they arise as outgrowths from their respective ganglia. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 105 The development of the brain of Blatta germajiica and Melanoplus feimtr-riibrimi agrees in all essential respects with the development of the Xiphidium brain. Certain Hemiptera, e.g. Ranatra fusca^ conform very closely to the type of brain structure seen in these Orthoptera. I may mention in this connection that the brain of Anurida niaritima shows the typical division into proto-, deuto- and tritocerebral segments with great distinctness. The last segment especially is re¬ markably distinct. Until very recently the detailed study of the embryonic Hexapod brain has been limited to the Coleoptera and the results obtained have been naturally enough extended to include not only other insects but other Arthropods as well. The Coleoptera, however, are far from being primitive forms and the role which they play in contemporary embryological literature is largely attributable to the unusual technical ad¬ vantages presented by their eggs. As far as development is concerned, the simpler brain of the Orthoptera and Ametabola in general offers many points of resemblance to the Crustacea and Myriopoda,^ whereas the brain of the Metabola, like so many other points in their organization bears witness to a considerable amount of modification. It is therefore more consistent with our general views of phylogeny to reduce the Coleopteran brain to the Orthopteran type than to proceed vice versa. We owe the most important contributions to the subject of Orthopteran brain development to Viallanes. After a decade of study devoted to the histological structure of the adult Arthropod brain he has selected Mantis as a subject for embryological investigation. His previous careful study of the adult brain of other Orthoptera {CEdipoda coernlescens and Caloptenus italicus, ’ 87 ^) has enabled him to avoid the con¬ fusion with which the inexperienced investigator is overwhelmed when attempting to follow the rapidly increasing complication of neural structures. With his usual skill and patience he has traced the development not only of the main structural features but of many details, so that we have a well-established point 1 See the papers of St. Remy (’90) and Viallanes (’87®, ’87^). io6 WHEELER. [VOL. VIII. of departure for further comparative studies. So far as my own observations are concerned I am able to corroborate Viallanes’ results on nearly all important points. I must state, however, that I have not followed the development into such detail. In the light of these researches a reconsideration of the Coleopteran brain must be undertaken. Patten’s description of the Acilius brain (’88) and my description of the brain of Doryphora (’ 89 ) need revision and alteration. We described the organ (see my Fig. 72, PL XIX) as consisting of three segments, each of which was subdivided into a brain portion, continuous with the ganglia of the ventral cord, an optic ganglion portion and an optic plate portion. Between the third brain and the mandibular segment, a segment was found which I designated as intercalary. This segment is also clearly seen in some of Patten’s figures (Figs. 2 and 2" s 4 PI. VII). Thus according to our account there were four preman- dibular segments or seven segments in the entire head. Our results were obtained almost exclusively from surface views — by itself a defective method. But greater error was incurred, it seems to me, in ascribing segmental values to the various prominences of the optic ganglion and optic plate. In order to bring our observations into harmony with the results obtained from a study of the Orthopteran brain, our figures must be interpreted in a very different way from that in which we chose to interpret them. Our first brain-segment is probably no segment at all, but merely a slight elevation often seen near the median line at the extreme anterior end of the germ-band. Our second, third and intercalary segments are equivalent to Viallanes’ third protocerebral lobe, deuto- cerebrum and tritocerebrum. The three divisions of the optic ganglion are not parts of three segments, but the whole struc¬ ture belongs to the protocerebrum, of which it forms the first lobe. In the same way we cannot regard the optic plate as trisegmental since it has no connection with the deuto- and tritocerebrum but only with the optic ganglion. It follows that the ocelli of Coleoptera are not originally formed on different segments as Patten would have us believe, but No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 107 belong to one segment — the protocerebrum. That such is the case I am convinced from a study of the eyes in embryos of Dytiscus verticalis, a form closely related to Acilius. As will be seen in the profile view Fig. 8, the optic ganglion and optic plate of Xiphiditini are at first folded back so as to lie along side the deuto- and tritocerebrum. The antennal furrow runs forward, separating the optic ganglion from the brain but stops when it reaches the protocerebrum. The value of this furrow as completely separating the second and third brain-segments from the optic ganglion was overlooked by Patten and myself: hence our false interpretation of the structures lying laterad to it. I believe that I am justified in putting this new interpreta¬ tion on the Coleopteran brain, because it harmonizes with Heider’s careful study of Hydrophihcs (’ 89 ). He has failed to find indications of segmental constrictions in the optic plate and optic ganglion and his figure 4 A. B. at p. 37 agrees closely with Viallanes’ description of the Mantis brain. It should be observed that the embryos of Hydrophihis are much larger than those of Aciliits and Doiyphora and therefore much more favorable for surface study. On the other hand it may be urged that Heider evidently did not employ so good a method of surface preparation as Patten. The distinct invagination associated with the formation of the optic ganglion in Coleoptera and described by Patten, Heider and myself, is probably homologous, as Viallanes sug¬ gests, with the intraganglionic thickening of the Orthop- teran embryo. This structure in Ma7ttis, and probably also in Xiphidmin^ takes no part in the formation of the optic ganglion, which arises — at least in great part — by delamina¬ tion as in the Crustacea (see Parker (’ 90 )). Only the outer or lateral portion of the optic plate becomes the compound eye, so that in a later stage the intraganglionic thickening is separ¬ ated from the edge ©f the eye by a considerable space. The thickening then lies just laterad to the antennal furrow as shown in Fig. 8. Whether or not the invagination in the Coleoptera really plays any part in forming ganglionic tissue as has been claimed, must be decided by renewed investi¬ gations. io8 WHEELER. [VOL. VIII. Concerning the researches of Cholodkowsky on the brain of Blatta germa7iica, I must say a few words, since I have de¬ scribed the brain and nerve-cord of this form as agreeing in all essential respects with those of Xiphiditmi. Cholodkowsky lays great stress on the existence of three distinct pairs of Punkt- substanz masses in the supraoesophageal ganglion as indicat¬ ing the presence of three segments. When we come to exam¬ ine his figures we find that he takes a very unusual view of brain-segmentation, for the three pairs of Punktsubstanz masses are seen to belong (Fig. 46 PI. IV; Fig. 67 PI. VI) to the protocerebrum and correspond to the centres of its three lobes. He did not distinguish the deuto- and tritocerebral segments ! Such of his remarks on the development of the brain and ventral nerve-cord as are at all comprehensible show similar glaring defects in observation. Thus he has failed to detect the small dermatoblastic cells which from the first cover the brain and nerve-cord. He asserts that these organs are at first naked and are only subsequently covered by an overgrowth of the integument from the sides of the body. The antennal and neural furrows do not play the part in development that he ascribes to them. The last abdominal ganglion of the mature embryo does not consist of four but of three fused ganglia; the fusion of the second and third abdom’inal ganglia was com¬ pletely overlooked. 3. Genteral Remarks on the Nervoiis System. The nervous system of Arthropods is by common consent derived from the nervous system of annelid-like forms, and it is to this group that we naturally turn in seeking an explanation for certain structures in the Hexapod brain and nerve-cord. In a brief preliminary paper Patten (’88) made the statement, that “ the ventral cord and brain of Arthropods is at first composed entirely of minute sense-organs, which in scorpions have the same structure as the segmental ones at the base of the legs.” This would seem to indicate that the Arthropod nervous system can be traced back to the condition seen in Polychseta — Lopadorhynchus, according to Kleinenberg (’86) — where both brain and nerve-cord arise in connection with and No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. ultimately supplant certain larval sense-organs. So far, how¬ ever, as the Hexapoda are concerned, Patten’s statement is, to say the least, inapposite, since, as I have pointed out, both brain and nerve-cord arise from peculiar ectodermal cells — the neuroblasts — which under no circumstances can be re¬ garded as primitively sensory. They are simply generalized cells, like the teloblasts of worms and the meristem of plants. The development of the nerve-cord in the Hirudinea and Oligochaeta agrees more closely with the conditions seen in in¬ sects. As Whitman has shown for Clepsine (’ 87 ), and E. B. Wil¬ son for Lumbricus (’ 83 ), the nerve-cord is proliferated forward from a pair of neuroteloblasts situated at the posterior end of the germ-band. Hence, in these worms, the whole of the nerve- cord is condensed, as it were, into two huge mother-cells, whereas in the Insecta it is condensed into a single layer of huge cells. There are reasons, however, for believing that this layer is, in part at least, derived from a few large cells situated just in front of the anus, and therefore corre¬ sponding to the Annelid neuroteloblasts.^ That there are only two rows of these cells in Annelids, while there are eight in insects, can form no very serious objection to their homology, as I pointed out in my preliminary note (’ 90 °). Certain conditions in the Crustacea also lend probability to the view that the Hexapod neuroblasts may be budded forth from a prae-anal row of teloblasts.^ Patten (’ 9 o) has pointed out in Cyniothoa a row of proliferating cells which form ectoderm, and Nusbaum (’9i) has described a very similar condition in Ligia. I have observed the same phenomenon in Porcellio, and believe it to be of general occurrence throughout the Isopoda. The cells are budded forth so as to form regular transverse and longitudinal rows. Reichenbach describes and 1 What I have called the neuroblasts in insects therefore correspond to the “ neural cell-rows ” in Annelids (the cells np. c. in E. B. Wilson’s Fig. 59 , PI. XIX, and the cells nc. in Whitman’s Figs. 9 and ii, PI. V). 2 The neuroblasts of the last row {tb., Fig. 56 , PI. VI) in the nerve-cord of Xipkidium are always distinctly larger and clearer than the neuroblasts of the remainder of the cord. They may be true neuroteloblasts and give rise to the neuroblasts, but as I have never found unmistakable caryokinetic figures in them, I am still in doubt as to their homology with the neuroteloblasts of worms. I lO WHEELER. [VOL. VIII. figures a very similar budding-zone of ectoderm cells in the Decapod Astacus (’86). That a portion of the ectoderm cells thus budded forth in these forms goes to form the nervous system, admits, it seems to me, of very little doubt. The clustered condition of the neuroblasts in the young germ-band of Xiphidium may be due to the rapidity with which the germ-band grows in length and breadth; the original regular arrangement of the cells being thereby disturbed and not re-established till a somewhat later stage. Although I have maintained a phylogenetic connection of the insect neuroblasts with the neural cell-rows of Annelids, I admit that they may be regarded from an entirely different standpoint, viz. as having arisen independently in insects by a process of precocious segregation ; but it should be noted in this connection that it is just the oldest Pterygota, the Orthoptera, which show this segregation most clearly, while in the more recent forms (Coleoptera, etci) the process is more obscure. A comparison of the histogenesis of the insectean with the histogenesis of the vertebrate nervous system brings out some interesting analogies. The neuroblasts may be compared with His’ Keimzellen which divide by caryokinesis to form his neuro¬ blasts. These are directly converted into ganglion-cells by sending out axis-cylinder processes. They correspond, there¬ fore, to the daughter-cells of the insect neuroblasts, which are likewise converted into ganglion-cells. The “Keimzellen ” of vertebrates differentiate close to the central canal, which is, of course, morphologically the outer surface of the cord, just as the Arthropod neuroblasts lie at the surface of the cord. In both groups the daughter-cells appear to be budded off in the same direction morphologically; though in vertebrates the gan¬ glion-cells migrate, while in insects they are pushed inwards by their newly proliferated sister-cells. The early separation of the neural ectoderm in vertebrates into Keimzellen and susten- tacular tissue (spongioblasts of His) is paralleled in insects by the precocious splitting of the same germ-layer into neuro¬ blasts and dermatoblasts, the latter giving rise to supporting tissue in the form of neurilemmata. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. m The majority of authors hold that the Arthropod brain is either wholly or in part homologous with the Annelid brain. Patten (’ 90 ) alone takes the view that the Annelid prostomium is absent in Arthropods, and that the brain of the latter is formed by the moving forward of three segments which are postoral in the Annelids. Apart from the fact that we have as yet no means of deciding whether what we call the first seg¬ ment of the Arthropod head (protocerebrum) is really a single segment or a complex of several, it is extremely improbable that so highly important a structure as the Annelid brain should have completely disappeared in the Arthropods. So great is the resemblance between the Arthropods and Annelids in all the more important morphological features and even in the de¬ tailed structure of the ventral nerve-cord, that the complete elimination of the brain certainly makes strong demands on one’s credulity. Will (’88) goes to the opposite extreme and regards the praeoral portion of the Arthropod brain as the exact homologue of the Annelid brain. He goes so far as to call the procephalic lobes of Aphis the “ Scheitelplatte.” He finds that they lie at the pole of the egg opposite the blastopore, or rather what he takes to be the blastopore, and that they arise independently of the nerve-cord. Now the “ Scheitelplatte ” of Aphis must include at least the proto- and deutocerebral segments — prob¬ ably also the tritocerebrum. The deutocerebrum in all the Orthoptera which I have examined is provided with a pair of true mesodermic somites and with a pair of appendages, the antennae. Each mesodermal somite sends a hollow diverticu¬ lum into an antenna, which is thus shown to be homodynamous with the other appendages of the embryo. The tritocerebral segment also contains a pair of abortive somites and in Anurida maritima, as I have lately ascertained, bears a pair of minute but distinct appendages (see Fig. VI, tc. ap.). Viallanes (’ 87 ®) and St. Remy (’ 9 o) have found that the second pair of antennae in Crustacea belong to the tritocerebral segment. These facts go to show that the deuto- and tritocerebral segments are homo¬ dynamous with the postoral segments and, as the “ Scheitel- platte ” of Aphis must extend at least as far back as the I 12 WHEELER. [VOL. VIII. tritocerebral segment, it cannot be homologized with the Annelid Scheitelplatte, a structure which is not segmented. A view midway between Will’s and Patten’s probably accords best with the facts at our disposal. The Arthropod proto¬ cerebrum probably represents the Annelid supraoesophageal ganglion, while the deuto- and tritocerebral segments, originally postoral, have moved forward to join the primitive brain. This is essentially Lankester’s view (’8i), according to which in Arthropods ^-'the prseoesophageal ganglion is a syn- cerebrum consisting of the archicerebrum and of the ganglion masses appropriate to the first and second pair of appendages which were originally postoral, but have assumed a praeoral position whilst carrying their ganglion-masses up to the archi¬ cerebrum to fuse with it.” In comparing the Arthropod with the Annelid brain much stress has been laid on the fact so clearly brought out by Kleinenberg (’86) — that the Annelid supraoesophageal ganglion originates independently of the ventral nerve-cord. Several investigators — Balfour (’ 80 ), Schimkewitch (’ 87 ), Will (’88) and others — have fancied that they could detect a similar onto¬ genetic discontinuity of the brain and nerve-cord in Arthropods. But more recent observations all tend to prove that there is a direct continuity of the central nervous system from the time when the ganglia first make their appearS.nce. So far as the insects are concerned I may note that Will’s conclusions were based on the defective surface observation of a form {Aphis) ill adapted to the study of the central nervous system.^ Even granting that the Annelid brain arises independently of the nerve-cord — and this is not yet settled — at least so far as the Oligochaeta are concerned (see E. B. Wilson, ’ 89 ) — Lankes¬ ter’s view of the Arthropod brain is in no way invalidated. The line of separation corresponding to the Annelid prototroch must fall in front of the deutocerebral segment, since it has been shown that this segment in some insects contains a pair of well- 1 Little value can be attached to Cholodkowsky’s assertion that in Blatta the supraoesophageal ganglion originates independently of the nerve-cord, since he has failed to see the deuto- and tritocerebral segments which are quite as well developed in Blatta as in other Orthoptera. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 113 developed mesodermic somites. When we stop to consider the intimate union of the proto- and deutocerebral ganglia from the time of their first appearance, we need entertain little hope of finding traces of a separation which existed, if indeed it existed at all, in a very remotely ancestral period. VI. The Development of the Reproductive Organs IN THE InSECTA. I. The Gonads. The following account of the development of the sexual organs is based almost exclusively on Xiphidhmi. Some attention was devoted to the study of Blatta^ but this form proved to be so much less satisfactory and to depart so little from the Xiphidmm type, that it was abandoned. Before passing to a description of the sexual organs and their ducts, it will be necessary briefly to consider the meso¬ dermal somites, since the history of the organs under consid¬ eration is intimately bound up with the history of the middle germ-layer. The mesoderm of Xiphidium, like that of other insects, is coextensive with the blastopore and hence reaches from the region of the definitive mouth to the region of the definitive anus. At first a continuous cell-layer, it soon splits up into segmental masses as metamerism sets in. These are further divided in the median ventral line so that each segment has a .pair of mesoderm blocks. Each of these acquires a cavity and the somites are established. The appendages are from the time of their first appearance intimately connected with the somites, since each of the latter sends a hollow diverticulum into the appendage of the corre¬ sponding half of the body. All the somites are fully formed when the embryo has reached Stage F. There are then eight¬ een pairs in all. The most anterior pair occupies the deuto¬ cerebral segment and sends hollow diverticula into the antennae. The walls of these somites are much thinner than those of succeeding pairs and, curiously enough, persist much longer. The pair in the tritocerebral segment is very small WHEELER. [VOL. VIII. II4 and indistinct and disappears very early. Each of the succeed¬ ing segments, with the exception of the eleventh abdominal has a distinct pair of somites. In the last abdominal, meso¬ derm is present, but I have been unable to find a trace of a true coelomic cavity.^ The youngest embryo in which I was able to detect repro¬ ductive cells had almost reached Stage F. In still earlier stages careful scrutiny failed to reveal any differentiation of the mesoderm cells. These cells, it is true, vary considerably in size and appearance but I have found it impossible to fix on any one set of elements which might be brought into con¬ nection with the germ-cells of older embryos. It is not, therefore, till the somites are established as distinct sacs that unmistakable primitive germ-cells make their appearance. In frontal sections of embryos in Stage F (Fig. 52 gd b gd 3 ), the primitive germ-cells are seen to lie in the inner wall of the somite. They are considerably larger than any of the sur¬ rounding mesoderm cells, and much paler. The chromatin of their nuclei is arranged in a more delicate skein. Like the neuroblasts they stain very deeply in picric acid. Normally, they occur only in the first to the sixth abdominal segments, each cluster being confined to the inner wall of a somite. The reproductive organs of Xiphidiimi are therefore truly meta- meric in their origin. There is nothing to show that they arise from vitellophags which have migrated into the somitic wall ; nor can they arise from the entoderm, since they are differentiated before the entoderm-bands have reached the basal abdominal segments in their growth backward from the oral and forward from the anal formative centre. I conclude, therefore, that the primitive germ-cells are enlarged and modi¬ fied mesoderm-cells. In explanation of Fig. 52 it may be noted that the plane of section is somewhat oblique so that it 1 I mention this because Graber (’90) has recently described a coelomic cavity in the anal segment of Hydrophilus (Fig. 29 , p. 62 ). Cholodkowsky also describes and figures (’91, Fig. 49 PI. IV) such a cavity in the eleventh abdominal seg¬ ment of Blatta. Every little slit in the mesoderm is not a coelomic cavity, and the figures referred to show only small spaces between the mesoderm cells of the telson. This may have been produced artificially, for aught the figures show to the contrary. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 115 strikes only the lowermost germ-cell of the cluster in the third abdominal segment (s'd 3 ) and in the fourth segment passes completely under the cluster. Even at this lime certain meso¬ derm-cells, the future epithelial elements (ep.) begin to flatten out‘and apply themselves to the surfaces of the germ-cells. The exact relations of the primitive germ-cells to the walls of the somite are readily seen in transverse section (Fig. 53). The inner face of the triangular somite is applied to the surface of the yolk and besides giving rise to the germ-cells will ulti¬ mately form the splanchnic, or visceral layer. The remainder of the coelomic wall is somatic, or parietal, and is converted into fat-body and musculature. The heart arises where the outer edge of the splanchnic passes into the somatic layer. In the section figured the entoderm is still wanting on the left side, whereas on the right side a single cell (en) of the right posterior band has already reached the segment. Similar ine¬ qualities in the rate of growth of the entoderm-bands are by no means infrequent. In this stage some of the primitive germ-cells show a tendency to leave the wall of the somite and to drop into the coelomic cavity. This is distinctly seen in Fig. 53. These cells sometimes enlarge considerably, become vacuolated and take on the appearance of young ova. A cell of this kind, nearly filling the coelomic cavity, is shown in Fig. 55. I do not believe that the cells are loosened from the coelomic wall during the process of sectioning. Although the clusters of germ-cells normally occur only in the first to the sixth abdominal segments, in one somewhat older embryo (Stage G) a well developed pair of clusters was found in the tenth segment. One of these is shown in sagittal section in Fig. 56 It resembles the normal clusters in every particular. The same section shows the diverticulum of the tenth somite (m. d). In the next section laterad to the one figured, the hollow tip of the diverticulum is seen to terminate in the right appendage of the segment. A similar relation of the coelomic diverticula to the append¬ ages obtains in all the abdominal segments in front of the tenth. WHEELER. [VOL. VIII. 116 The primitive germ-cells, which at first occupy only a limited portion of the splanchnic wall, increase in number during a stage midway between F and G. Beautiful caryokinetic figures may then be found in some specimens — showing that the primitive germ-cells themselves proliferate. New germ- cells may be added to the clusters by a differentiation of elements in the splanchnic wall but I have seen nothing to convince me that this occurs. The epithelial cells become much flattened and stain more deeply so that they stand out distinctly among the pale rounded germ-cells which they invest. The inner wall of the somite soon becomes too small to contain all the rapidly accumulating cells and is forced to send out a solid diverticulum. This is directed anteriorly, and in a little later stage fuses with the wall of the antecedent somite. This fusion is probably preceded by the shortening of the embryo which takes place during a stage immediately succeeding Stage F. The result of the fusion is the formation of a continuous strand of germ-cells with their accompanying epithelial cells. For some time the typical hexa- metameric arrangement is still visible in the strand, but later the whole mass shortens very decidedly to form the definitive ovary and testis and all traces of metamerism are lost. In the present paper I shall not follow the development of these organs further, but will pass on to a description of the sexual ducts. This will enable me to supplement the recent work of Heymons (’ 9 i) who has given us an extended and valuable account of the development of the sexual organs in Blatta^ but has contributed only a few observations on the development of the ducts. 2. The Male Ducts. The sexual ducts like the germ-cells are modified portions of the mesodermal somites. While considering the excep¬ tional embryo in Fig. 56, attention was directed to the diverticulum ((in. d) of the tenth abdominal somite. This diverticulum, which is quite normal, is destined to form the terminal portion of the deferent duct (spermaduct) and the seminal vesicle of the adult insect. At the base of the divert- No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 117 iculum a constriction is formed which converts the proximal portion into a thin cord but leaves the distal end expanded as a hollow sac, which I shall call the terminal ampulla. The remainder of the deferent duct — viz. the portion extending from the sexual gland in the sixth to the anterior end of the tenth segment is formed by a cord-like thickening in the splanchnic wall of the three intervening somites. Anteriorly the cells of the duct pass into the epithelium enveloping the germ-cells. There is no lumen in the duct proper except towards its end where it widens into the terminal ampulla. Thus only the ap¬ pendage diverticula of the tenth segment go to form the ends of the ducts ; in all the other abdominal segments the diverti¬ cula break down and disappear, together with their respective appendages, before the embryo reaches Stage H. In the thoracic, oral and antennary segments, however, the diverti¬ cula are converted into the muscles of the persistent ap¬ pendages. The further history of the male ducts is readily followed in partially stained embryos mounted in toto. Sex is determined, so far as I have been able to make out, during or soon after katatrepsis, at which time the appendages of the second to the seventh abdominal segments disappear. Fig. 42 represents the caudal end of an embryo just after katatrepsis (Stage J). Appendages still persist on the eighth to eleventh segments while the pleuropodia, not seen in the figure, have begun to degenerate. The testes {tsi) and the spermaducts {m. d.) are represented in blue. The former have shortened considerably and moved caudad so that they now lie in the sixth and seventh segments. The long terminal threads run forwards from the anterior ends of the testes, while the spermaducts run back¬ wards and end in the terminal ampullae {ta. m.) which still fit into the cavities of the tenth pair of abdominal appendages {ap. 10). A section through these appendages is seen in Fig. 57, showing very clearly the connection of the ampullae {ta. m) with the ducts {7n. d.). In front of this section the ducts have no distinct lumen. In Fig. 43, taken from a slightly older embryo, the append¬ ages of the eighth segment have completely disappeared, ii8 WHEELER, [VOL. VIII. while those of the tenth have grown smaller and approached the median ventral line. They have, in fact, grown too small to contain the ampullae which are drawn away from them and lie between the ninth and tenth segments a little in front of the abortive appendages. At the same time the inner faces of the ampullae have become flattened and applied to each other in the median line. Each of these sacs has a pointed tip directed caudad. The more arcuate course described by the ducts in this stage is undoubtedly due to the mutual approxi¬ mation of their terminal ampullae. The appendages of the eleventh segment, the cerci {cc. [<^/“]), which in the pre¬ ceding stage were rounded like the other abdominal append¬ ages, have become oval. A more advanced embryo is represented in Fig. 44 (some¬ what younger than Stage K). The appendages of the tenth segment {ap^^) have almost completely disappeared. Those of the ninth segment, the future styli {ap 9 ) have lengthened and now point outwards and forwards. The cerci have grown more pointed. The terminal ampullae lie completely in the ninth segment, having shifted their position headwards. The move¬ ment takes place in such a way that what were the posterior faces of the ampullae in the younger stage (Fig. 43) are applied to each other, while the pointed tips are directed forwards. The ducts are thereby rendered still more arcuate towards their terminations. An intermediate stage in this singular movement is shown in Fig. 45, where only small portions of the posterior faces of the ampullae are as yet applied to each other. Comparison of Figs. 43, 44 and 45 shows that the pointed tips remain united and move forward while the sur¬ faces of mutual contact are being shifted. Finally in Fig. 46, which represents the abdominal end of an embryo ready to hatch, we see that the terminal ampullae have increased con¬ siderably in size at the expense of the thickness of their walls. They have also lengthened, and brought still more of their surfaces in contact in the median line. The pointed tips of the ampullae extend into the eighth segment. It may also be noted that the points where the spermaducts meet the ampullae have moved forward. The appendages of the tenth segment No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. II9 have long since disappeared and the pleuropodia have lost their organic connection with the embryo, so that only two pairs of abdominal appendages persist, the stylets {st. \ap 9]) and the cerci {cc. [<^/^^]), both provided with setae. The pointed fusiform cerci are now folded back so as to bring their insertions on the anal segment into view. Beyond this stage the development of the male ducts was not followed in Xiphidium ensifertmt, but several larval stages of an allied species, X. fasciatum, were studied for the purpose of connecting the embryonic with the adult condition. It will be noticed that in Xiphidiu^n ensiferum there exists at the time of hatching no external opening to the sperma- ducts; the ampullae are completely closed sacs applied to the ventral hypodermis of the ninth abdominal segment, and the ducts connecting them with the testes have no lumen. In the X. fasciatitm larva 10 mm. long the ejaculatory duct has made its appearance as an unpaired invagination of the hypodermis in the median line between the ninth and tenth segments. Fig. 47 shows the sexual organs of such a larva seen from within, the ventral scutes of the tenth and anal seg¬ ments having been entirely removed. The prominent terminal ampullae, which become the seminal vesicles of the adult, are considerably enlarged and their walls have increased in thick¬ ness. The short spermaducts, now provided with a small lumen, run from the under surface of the sacs to the prom¬ inent testes. Only the outer opening of the invagination which'is to form the ejaculatory duct is seen at m.o. It runs forward as a flattened chitin-lined depression beneath the seminal vesicles. Sagittal sections show that there is as yet no communication between the lumina of the mesodermal and ectodermal portions ; it is not till a later stage that such a communication is established. 3. The Female Ducts. The oviducts, like the vasa deferentia, are derived from a pair of coelomic appendage-diverticula, but in the female the diverticula belong to the seventh abdominal segment. The diverticula of the female embryo also become constricted I 20 WHEELER. [VOL. VIII. proximally and end in terminal ampullm, which are from the first somewhat smaller and more elongate than the homo- dynamous structures of the male. The appendages to which the ampullae belong are also less prominent than the tenth pair of appendages in the male. Examination of several series of cross-sections from embryos in Stage J — this being the stage in which the sexes differentiate — reveals the curious fact that in the female, besides the pair of ampullae in the sev¬ enth, a pair is also retained in the tenth segment. Figs. 58 and 59 represent two sections taken from such an embryo — the former passing through the tenth, the latter through the seventh abdominal segment. In Fig. 58, the two terminal ampullae {ta. m.), and small portions of the ducts leading to them, are still preserved, but the cells and nuclei, especially in the ducts, are being broken down. The ampullae soon share the same fate. In the seventh segment (Fig. 59,/*.<’80 Balfour, F. M. Notes on the Development of the Araneina. Quart. Jou?' 7 i. Micr. Sci. Vol. 20. 1880. '88 Barrois, J. Recherches sur le Ddveloppement de la Comatule (Coma- tula mediterranea). Recueil zool. Stiisse. Tom. 4. 1888. ’85 Bergh, R. S. Die Metamorphose von Aulastoma gulo. Arb. zool. Inst. Wurzburg. 7. Bd. 1885. ’87 Blochmann, F. Ueberdie Richtungskdrper bei Insecteneiern. Morph. Jahrb. 12. Bd. 1887. ’69 &’70 Brauer, F. Verwandlung der Insecten im Sinne der Descendenz- Theorie. Verhandl. zool. bot. Gesell. Wien. I, 1869. II, 1870. ’85 Brauer, F. Systematisch - zoologische Studien. Sitz.-Ber. Akad. Wiss. Wien. 91. Bd. 1885. ’86 Brauer, F. Ansichten iiber die palaozoischen Insecten und deren Deutung. Annal. d. k. k. fiaticrhist. Hofinuseums. Wien. i. Bd. 1886. ’89 Brongniart, C. Les Blattes de I’dpoque houilRre. Cotfipt. Re 7 td. Tom. 108. 1889. KlQl Bruce, A. T. Observations on the Embryology of Insects and Arach¬ nids. A memorial volume. Baltimore, 1887. ^91 Bumpus, H. C. The Embryology of the American Lobster. Journ. of Morph. Vol. 5. No. 2. 1891. ’85 Carnoy, J. B. La Cytodierese chez les Arthropodes. La Cellule. Tom. I, Ease. 2. 1885. ’90 Carriere, j. Die Entwicklung der Mauerbiene (Chalicodoma mu- raria, Fabr.) im Ei. Arch. f. 77 iikr. A 7 iat. 35. Bd. 1890. ’91a Cholodkowsky, N. Die Embryonalentwicklung von Phyllodromia (Blatta) germanica. Me 77 i. de VAcad. I 77 tp. des Sciences de St. Petersb. 7 ser. Tom. 38. No. 5. 1891. K’91t> Cholodkowsky, N. Ueber einige Formen des Blastopors bei mero- blastischen Eiern. Zool. Anzeig. Jahrg. 14. 1891. ’62 Claparede, E. Recherches sur I’Evolution des Araignees. Naturk. Verhandel. Deel i, Stuk i. Utrecht. 1862. ^’91 Davenport, C. B. Observations on Budding in Paludicella and some other Bryozoa. Btdl. Mus. Co 77 ip. Zool. Harvard College. Vol. XXII. No. I. 1891. ’75 Dewitz, H. Ueber Bau und Entwicklung des Stachels und der Legescheide einiger Hymenopteren und der griinen Heuschrecke. Zeitschr. f, wiss, Zool, 25, Bd. 1875. WHEELER, [VOL. VIII. 144 ’87 Eisig, H. Die Capitelliden des Golfes von Neapel. Fauna undFlora des Golfes v. Neapel. Berlin. 1887. '77 Graber, V. . Die Insecten. Naturkrafte. 22. Bd. Miinchen. 1877. ’88a Graber, V. Vergleichende Studien iiber die Keimhiillen und die Riickenbildung der Insecten. Denkschr. d. niath. naturw. Classe d. k. Akad. d. Wzss. Wien. 55. Bd. 1888. ’88tJ Graber, V. Ueber die primare Segmentirung des Keimstreifs der Insecten. Morph. Jahrb. 14. Bd. 1888. ’89 Graber, V. Vergleichende Studien iiber die Embryologie der Insecten und insbesondere der Musciden. Denkschr. d. znath. naturw. Classe d. k. Akad. d. Wiss. Wien. 56. Bd. 1889. ’90 Graber, V. Vergleichende Studien am Keimstreif der Insecten. Denkschr. d, znath. ziatzirw. Classe d. k. Akad. d. Wiss. Wien. 57. Bd. 1890. ’84 Grassi, B. Intorno alio sviluppo delle Api nell’uovo. AttidelVAcad. Gioeziia di Scienze Nat. izi Catania. 3. ser. Vol. 18. 1884. ’89 Grassi, B. Les Ancetres des Myriapodes et des Insectes. Arch. Ital. Biol. Tom. II. 1889. ’79 Grobben, C. Die Entwicklungsgeschichte der Moina rectirostris, etc. Arb. zool. Inst. d. Univ. Wiezi. 2. Bd. 1879. ’89 Haase, E. Ueber die Zusammensetzung des Schabenkorpers. Sitz- ungsb. Gesell. naturf. Freuzide. Berlin. 1889. ^ ’89 Haase, E. Die Abdominalanhange der Insecten, mit Beriicksichtigung der Myriapoden. Morph. Jahrb. 15. Bd. 1889. ’85 Hallez, P. Orientation de I’Embryon et Formation du Cocon chez la Periplaneta orientalis. Coznpt. Reztd. Tom. loi. 1885. ’86 Hallez, P. Sur la Loi de I’Orientation de I’Embryon chez les In¬ sectes. Cozzzpt. Rezzd. Tom. 103. 1886. ’77 Hatschek, B. Beitrage zur Entwicklungsgeschichte der Lepidopteren. Jena. Zeitschr. f. Natzirw. ii. Bd. 1877. ’84 Hatschek, B. Ueber Entwicklung von Sipunculus nudus. Arb. zool. Izzst. Wiezz. 5. Bd. 1884. K ’89 Heider, K. Die Embryonalentwicklung von Hydrophilus piceus, L. I. Theil. Jena. 1889. A ’92 Herrick, C. L. Additional Notes on the Teleost Brain. Azzat. Azzzeig. 7. Jahrg. June, 1892. < ’86 Herrick, F. H. Notes on the Embryology of Alpheus and other Crustacea and on the Development of the Compound Eye. Johzzs Hopkizzs Uzziv. Circ. No. 6. 1886. K ’90 Heymons, R. Ueber die hermaphroditische Anlage der Sexualdriisen beim Mannchen von Phyllodromia (Blatta, L.) germanica. Zool. Azzzeig. 13. Jahrg. 1890. ’91 Heymons, R. Die Entwicklung der weiblichen Geschlechtsorgane von Phyllodromia (Blatta) germanica. Zeitschr. f. wiss. Zool. 53. Bd. 1891. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 145 ’89 His, W. Die Neurobiasten und deren Entstehung im embryonaien Mark. Abhandl. d. k. sacks. Gesell. Bd. XV. 1889. ’^'11 Huxley, T. H. A manual of the Anatomy of the Invertebrated Animals. London, 1877. ’88 Jordan, K. Anatomie und Biologie der Physopoda. Zeitschr. f. wiss. Zool. 47. Bd. 1888. ’85 & ’88 Kennel, J. v. Entwicklungsgeschichte von Peripatus Edwardsii, Blanch, und Peripatus torquatus, n. sp. i. u. 2. Theil. Arb. zool. zoot. Inst. Wurzburg. 7. u. 8. Bd. 1885 and 1888. ’86 Kleinenberg, N. Die Entstehung des Annelids aus der Larve von Lopadorhynchus. Zeitschr. f. wiss. Zool. 44 Bd. 1886. ’81 Roller, C. Untersuchungen iiber die Blatterbildung im Hiihnerei. Arch.f. 7 nikr. Anat. 20. Bd. 1881. ’85 Korotneff, A. Die Embryologie der Gryllotalpa. Zeitschr. f. wiss. Zool. 41. Bd. 1885. K ’92 Korschelt, E. und Heider, K. Lehrbuch der vergleichenden Ent¬ wicklungsgeschichte der wirbellosen Thiere. Spez.-Theil. 2. Heft. 1892. ’72 Kowalevsky, a. (Note on Uljanin’s researches on the sting of the bee.) Sitz.-Ber. d. zool. Abth. d. 3. Versamml. Naturf. in Kiew. Zeitschr. f. wiss. Zool. 22. Bd. 1872. V’86 Kowalevsky, A. Zur embryonaien Entwicklung der Musciden. Bio¬ log. Centralbl. 6. Bd. 1886. ’73 Kraepelin, C. Untersuchungen iiber den Bau, Mechanismus und Entwicklungsgeschichte des Stachels der bienenartigen Thiere. Zeitschr. f. wiss. Zool. 23. Bd. 1873. ’49-’53 Lacaze- Duthiers. Recherches sur I’armure gdnitale femelles des Insectes. Ann. d. Sci. Natur. 1849-1853. ' ’81 Lankester, E. Ray. Appendages and Nervous System of Apus can- criformis. Quart. Journ. Micr. Sci. Vol. 21. 1881. ’87 Lemoine, V. Recherches sur le Ddveloppement des Podurelles. Con- grls de la Rochelle (1882), Ass. Frang. pour VAvancement des Sciences. Paris. 1883. (1887.) ’92 Lenhossek, M. V. Ursprung, Verlauf und Endigung der sensibeln Nervenfasern bei Lumbricus. Arch. f. inikr. Anat. 39 Bd. 1892. ’55 Leuckart, R. Ueber die Micropyle und den feineren Bau der Scha- lenhaut bei den Insecteneiern. Miill. Arch. f. Anat. u. Phys. 1855. 60 Leuckart, R. Bau und Entwicklungsgeschichte der Pentastomen. Leipzig und Heidelberg, i860. '^’77 Mayer, P. Zur Entwicklungsgeschichte der Decapoden. II. Bd. 1877. ’69 Melnikow, N. Beitrage zur Embryonalentwicklung der Insecten. Arch.f. Naturgesch. 35. Bd. 1869. f^’66 Metschnikoff, E. Embryologische Studien an Insecten. Zeitschr. f. wiss. Zool. 16. Bd. 1866. [VOL. VIII. 146 WHEELER. ’74 Metschnikoff, E. Embryologie der doppelfiissigen Myriapoden (Chilognatha). Zeitschr. f. wiss. Zool. 24. Bd. 1874. ’86 Miall, L. C., and Denny, A. The Structure and Life-History of the Cockroach (Periplaneta orientalis). London. 1886. ’64 Muller, F. Fiir Darwin. 1864. tA ’83 Nusbaum, J. Vorlaufige Mittheilung iiber die Chorda der Arthro- poden. Zool. Anzeig. 6 . Jahrg. 1883. ’84 Nusbaum, J. Rozwdj przewodow organdw plciowych u owaddw. (Ueber die Entwicklungsgeschichte der Ausfiihrungsgange der Sexual- driisen bei den Insecten.) Lemberg. 1884. ’87 Nusbaum, J. L’Embryologie de Mysis Chameleo, Thomps. Arch, zool. Exper. Tom. 5. 1887. K’ 91 a Nusbaum, J. Beitrage zur Embryologie der Isopoden. Biol. Cen- tralbl. II. Bd. 1891. ’ 91 t> Nusbaum, J. Studya nad Morfologia Zwierzat. 1 . Przyczynek do Embryologii Maika (Meloe proscarabaeus, Marsham). 1891. ’91 Oka, A. Observations on Fresh-water Polyzoa. Journ. Coll. Sci. I 7 np. Univ. Japan. Vol. 4., Pt. i. 1891. ’87 Oudemans, j. T. Bijdrage to de Kennis der Thysanura en Collembola. Academische Proefschrift. Amsterdam. 1887. K ’83 Packard, A. S. The Embryological Development of the Locust. Third Report U. S. E 7 it. Comm. 1883. A ’84 Packard, A. S. Article on Pseudoneuroptera. The Standard Natural History. Vol. 2. 1884. ’84 Palm^n, j. a. Ueber paarige Ausfiihrungsgange der Geschlechts- organe bei Insecten. Helsingfors. 1884. K ’90 Parker, G. H. The History and Development of the Eye in the Lobster. Bull. Mus. Co 7 np. Zool. Cambridge. 1890. M ’84 Patten, W. The Development of Phryganids with a preliminary note on the Development of Blatta germanica. Quart. Journ. Micr. Sci. Vol. 24. 1884. H ’88a Patten, W. Studies on the Eyes of Arthropods. 11 . Eyes of Aci- lius. Jour, of Morph. Vol. 2. 1888. /<’88t» Patten, W. Segmental Sense-Organs of Arthropods. Journ. of Morph. Vol. 2. 1888. K ’90 Patten, W. On the Origin of Vertebrates from Arachnids. Quart. Journ. Micr. Sci. Vol. 31. 1890. ’86 Reichenbach, H. Studien zur Entwicklungsgeschichte des Fluss- krebses. Abhandl. Sencke 7 tb. naturf. Gesell. 14. Bd. 1886. ’86a Ryder, J. A. The Origin of the Amnion. A 7 n. Naturalist. Vol. 20. 1886. K’86t> Ryder, J. A. The Development of Anurida maritima, Gudrin. Am. Naturalist. Vol. 20. 1886. ’90 St. Remy, G. Contribution a I’etude du cerveau chez les Arthropodes tracheates. Arch. zool. ExJSr. (2.) Tom 5, and Supl. (1887). 1890. No. I.] CONTRIBUTION TO INSECT EMBRYOLOGY. 147 ’87 SCHIMKEWITSCH, W. Etude sur le Developpement des Araigndes. Arch, de Biol. Tom. 6. 1887. K ’ 85-88 Sedgwick, A. The Development of the Cape Species of Peripatus. Part I-VI. Quart. Journ. Micr. Sci. Vol. 25-28. 1885-1888. K ’91 Stiles, C. W. Bau und Entwicklungsgeschichte von Pentastomum proboscideum und Pentastomum subcylindricum. Zeitschr. f. wiss. Zool. 52. Bd. 1891. ’82 Tichomiroff, a. The Development of the Silkworm (Bombyx mori) in the Egg. (Russian.) Works Lab. zool. Mus. Moskow. Vol. I. 1882. ’ 87 a ViALLANES, H. Sur la morphologie du cerveau des Insectes et des Crustaces. Co 7 npt. Rend. Tom. 104. 1887. ’ 87 t> ViALLANES, H. Etudes Histologiques et Organologiques sur les centres nerveux et les organes des sens des animaux articules. V. Memoire. Le cerveau du Criquet (Oedipoda coerulescens et Caloptenus italicus), etc. Ami. d. Sci. Natur. Zool. Tom. 4. 1887. ’90 ViALLANES, H. Sur quelques Points de I’Histoire du Developpement embryonnaire de la Mante religieuse (Mantis religiosa). Revue Biol, du Nord de la France. Tom. 2. No. 12. 1890. ’91 ViALLANES, H. Sur quelques Points de I’Histoire du Developpement embryonnaire de la Mante religieuse (Mantis religiosa). Annal. des Sci. Natur. Tom. ii. 1891. ’89 VOELTZKOW, A. Entwicklung im Eivon Musca vomitoria. Arb.zooL zoot. Inst. Wurzburg. 9. Bd. 1889. t\ ’90 Watase, S. On the Morphology of the Compound Eyes of Arthro¬ pods. Stud. Biol. Lab. Johns Hopkins Univ. Vol. 4. 1890. ’66 Weismann, a. Die Metamorphose der Corethra plmtiicornis., ein weiterer Beitrag zur Entwicklungsgeschichte der Insecten. Zeitschr. f. wiss. Zool. 16. Bd. 1866. fC’OOa Wheeler, W. M. Ueber driisenartige Gebilde im ersten Abdominal- segment der Hemipterenembryonen. Zool. Anz. 12. Jahrg. 1889. < ’89t) Wheeler, W. M. The Embryology of Blatta germanica and Dory- phora decemlineata. Journ. of Morph. Vol. 3. 1889. K ’ 90 a Wheeler, W. M. On the Appendages of the First Abdominal Seg¬ ment of Embryo Insects. Trans. Wis. Acad. Sci.., Arts and Letters. Vol. 8. 1890. ^ ’ 90 t> Wheeler, W. M. Note on the Oviposition and Embryonic Devel¬ opment of Xiphidium ensiferum. Scud. Insect Life. Vol. 2. Nos. 7 and 8. 1890. K" ’ 90 c Wheeler, W. M. Ueber ein eigenthiimliches Organ im Locustiden- embryo. Zool. Anz. 13. Jahrg. 1890. ’ 91 a Wheeler, W. M. The Embryology of a Common Fly. Psyche. June, 1891. ’911) Wheeler, W, M, The Germ-band of Insects. Psyche. July, 1891. 148 WHEELER, K’9ic Wheeler, W. M. Neuroblasts in the Arthropod Embryo. Journ. of Morph. Vol. 4. 1891. f ’92 Wheeler, W. M. Concerning the “ Blood-tissue ” of the Insecta. Psyche. Feb.-April, 1892. K ’78 Whitman, C. O. The Embryology of Clepsine. Quar. Journ. Micr. Sci. Vol. 18. 1878. A' ’87 Whitman, C. O. A Contribution to the History of the Germ-layers in Clepsine. Journ. oj MorJh. Vol. i. 1887. ’88 Will, L. Entwicklungsgeschichte der viviparen Aphiden. SpengeVs zool. Jahrb. Abth. J. Anat. u. Ontog. 3. Bd. 1888. f\ ’81 Wilson, E. B. The Origin and Significance of the Metamorphosis of Actinotrocha. Quart. Journ. Micr. Sci. Vol. 21. 1881. K ’89 Wilson, E. B. The Embryology of the Earth-worm. Journ. oj Morph. Vol. 3. 1889. K ’90 . Wilson, E. B. The Origin of the Mesoblast-Bands in Annelids. Journ. oJ Morph. Vol. 4. 1890. ’54 Zaddach, G. Untersuchungen fiber die Entwicklung und den Bau der Gliederthiere, I. Die Entwicklung des Phryganideneies. Berlin. 1854. I y t WHEELER. 150 EXPLANATION OF PLATE 1 . LRiphidium ensiferum, Scud.) Fig. I {A). Surface view of embryo during gastrulation. p.o.^ indusium ; pcd, procephalic lobe; < 5 /., blastopore; a., anal bifurcation of the blastopore; ams., amnioserosal fold. Fig. 2 {B). Surface view of embryo with amnioserosal fold closed over trunk- region. pcl.n., neuroblast-centres on the procephalic lobes. 0., anterior widening of the blastopore. Remaining letters as in Fig. i. Fig. 3 {C). Surface view of embryo with the amnioserosal fold encroaching on the indusial thickening; po.am.., amnioserosal fold of the indusium ; 2;., pedicel temporarily uniting the indusium with the head; at., antenna; md.s., mandibular segment; first maxillary segment; mx.s.p second maxillary segment; p.s.,^-p.s.p first to third thoracic segments ; a.s.,^ first abdominal segment. Re¬ maining letters as in Fig. i. Fig, 4 {B>). Surface view of embryo just after the separation of the indusium from the head. Letters as in Figs. 3 and i. Fig. 5 (E). Indusium spreading over the yolk. View of embryo nearly completely submerged in the yolk, on its way to the dorsal surface, tc.s., tritocerebral segment. Other letters as in preceding figures. Fig. 6 (B). Surface view of elongate embryo on dorsal surface of yolk, lb., labrum ; md., mandible ; mx.^ first maxilla ; mx.^ second maxilla ; p.^ — p.,^ the three thoracic appendages (legs); coe., coelomic sac of first abdominal segment showing through the body wall; pi. (a/.^), pleuropodium (appendage of the first abdominal segment); v., yolk ; envl., cellular envelopes torn away from the ventral face of the embryo; cc. (^?/.”), cerci (appendages of the eleventh abdominal segment). Fig. 7 {G). Surface view of shortened embryo on dorsal yolk, pc.,^ second protocerebral lobe ; pc.,'^ third protocerebral lobe ; dc., deutocerebrum ; tc., trito- cerebrum ; e., eye; x., metastigmatic, or oenocytic invagination ; ap..,* fourth ab¬ dominal appendage. Remaining letters as in Fig. 6 (R). Fig. 8 (H). Surface view of embryo turning the lower pole of the egg. sr., inner indusium functioning as the serosa; am., amnion reflected back over the yolk and continuous with the membrane sr. ; at., antenna; pi., right pleuro¬ podium ; igl., intraganglionic thickening. Fig. 9 {/). Surface view of embryo just after returning to the ventral face of the egg. e., eye ; other letters as in Fig. 8 {H). Fig. 10 {K). Surface view of advanced 9 embryo just before the secretion of the larval cuticle, d.o., “ dorsal organ ” ; f.d., oviduct; ta., terminal ampulla of oviduct; op.^ {op.^), op.^ {ap.'^), first and second pairs of gonapophyses (append¬ ages of the 8th and 9th abdominal segments), cc. (a/.”), cercus. THE imm Of THE tSsv-'Rsjrv OF ! Jk'D’S Journal .of Morphologij. Vol. OT. W. M. Wheeler del. PL I ..p.o. .o.ain. pci 0 at md.s. -inx.s- p.s- ps? p.s^ a.s- sr cmi -igl al 4 -. (D) 5 'E) . i ’ • i. • '1 . i ■■ . , -—-T- - i ^ ',iV ‘ I . , « • • r • t • • ‘ I "" i * •* I - I • r * ! •■"1 ' • • • .♦ . pel . o . at — md.s. ■ mx. s- ..inx.s- r-P-S- . -P-s- .-—p.s^ ' t' •I D. (J) cc(ap ) ■p.o . pel - -at -md.s "'■4c=s '"'tnx.s- 10. (k: Mk.Anst. V Werner A Wmtei; FrojilefurOM. 152 WHEELER, EXPLANATION OF PLATE 11 . (Figs. II and 12, Stagmomantis Carolina; Figs. 13 and 14, Gryllus luctuosus; Figs. 15-20, Xiphidium ensiferum.) Fig. II. Surface view of gastrula of Stagmomantis. X 150. /. 0^''->'^^ ' * U o2 B?§|^'^’- A2 u «#«;p »;SSp»?:;v ■-^ -'_; i " ^ ;| ^-, ^ 0 :il! ' :-.^i!. if> ■ i"’-v r®'%i Sil.'W ^ aiJis ) -'^>y ,>'’7x »^^ ''^ 0 %V‘ 4 M feS.! 2 i-;,'i C// 7/.9 Journal of Morphologij. I oI \lll. 15. 16 . 17. WM.V/heeler del. lith. Anst. VWerner & WCnter, Frankfnrt^M. I 154 WHEELER, EXPLANATION OF PLATE III. {Xiphidhim ensiferum.) Fig. 21. Median longitudinal section through the head of an embryo over which the envelopes have just closed. X 230. o., stomodaeum ; am., amnion ; s,, serosa; p.o., indusium ; z., pedicel uniting the indusium to the head of tho embryo ; ec., ectoderm ; ms., mesoderm ; vph., vitellophag ; v., yolk. Fig. 22. Median tranverse section of the indusium when it has reached half way round the egg. X 230. p.o., inner indusial layer ; am.,^ outer indusial layer ; s., serosa. Fig. 23. Three normal cells from the indusium. X 700. Fig. 24. Three cells with shrivelled nuclei from the indusium. X 700. Fig. 25. Transverse section through basal abdominal region of an embryo passing to the dorsal surface of the yolk. X 230. nb., neuroblasts ; db., derma- toblasts ; am., amnion ; ms., mesoderm ; ec., ectoderm. Fig. 26. Transverse section through first maxillary segment of an embryo passing to the dorsal surface of the yolk. X 230. ng., neural groove ; en., ento¬ derm. Other letters as in Fig. 25. Fig. 27. Transverse section through the second maxillary segment of an embryo in the stage of Fig. 6, (E) Plate I. X 230. mx.p second maxilla (trilobed) ; mnb., median-cord neuroblast; ps., Punktsubstanz ; vph., vitellophag. Other letters as in Figs. 25 and 26. Fig. 28. Transverse section through the mesothoracic ganglion of an embryo in a stage somewhat younger than G (Fig. 7, PI. I). X 230. g.p younger offspring of the neuroblasts; g.,^ older offspring of the neuroblasts (ganglion cells); ecd., ectoderm ; me., median cord (“ mittelstrang ”). Remaining letters as in preceding figures. Fig. 29. Sagittal section through nerve-cord a little to one side of the median line. Embryo in Stage G (Fig. 7, PL I). X 175. md.g., mandibular ganglion ; mx.g.,^ first maxillary ganglion ; mx.g.p second maxillary ganglion ; ig.,^ ig.,^ first and second interganglionic depressions; p.g->^ first thoracic ganglion; p-g-? second thoracic ganglion ; mnb., median cord neuroblasts ; mg., their progeny; ini., inner neurilemma. Fig. 30. Frontal section through the base of the abdomen of an embryo somewhat older than Fig. 6, (F), PI. I. X 175. nb., neuroblasts; mnb., median cord neuroblasts in the intersegmental regions ; p.,^ metathoracic leg; pL, pleuro- podium ; ap.p-ap.,^ second to fifth abdominal appendages grazed by the knife. Fig. 31. Transverse section through the mesothoracic ganglion of an embryo in a stage between Fig. 9 (_/) and 10 {K) PI. I. X 230. hy., hypodermis (product of dermatoblasts) ; nb., neuroblasts ; g.^, latest progeny of the neuroblasts ; g., ganglion-cells; ini., inner neurilemma; enl., outer neurilemma; ps., Punkt¬ substanz. J * , ■ i ■ «' \ i- t \ I i -. i ) Pi.m can (un inn 7 . '(0 Jl/rSiUCACN ^Wi$WW’ W. M. Wheeler del. Journal of Morphologij JolMII. i lith.Anst V Werner & Winter Frankfurt‘^M. WHEELER, EXPLANATION OF PLATE IV. {Xiphidium ensiferum^ Fig. 32. Transverse section through the head of an embryo in the stage of Fig. 5 (E) PL I. X 230. /( 5 ., labrum; pc} {o.g.), pc.p pc.p first, second and third protocerebal lobes ; op., optic plate ; en. ms., mesentoderm. Fig. 33. Next following section to that represented in Fig. 32. X 230. n.b., neuroblast; am., amnion. Remaining letters same as in Fig. 32. Fig. 34. Next following section to that represented in Fig. 33. X 230. st., stomodaeum. Remaining figures the same as in Figs. 32 and 33. Fig. 35. Transverse section through the labrum, in a stage intermediate be¬ tween E and F (Figs. 5 and 6, PI. I.) X 230. pc.,'^ third protocerebal lobe; lb., labrum ; ejt. ms., mesentoderm. Fig. 36. Transverse section through labrum and brain of an embryo in Stage F (Fig. 6, PI. I.) X 230. Ib.y labrum ; pc.p pc.ppc.,^ ^xst, second and third proto¬ cerebral lobes ; op., optic plate ; nb., neuroblast; db., dermatoblasts ; coe., head- coelom ; ms., mesoderm cells ; igl., intraganglionic thickening ; st., stomodaeum ; am., amnion. Fig. 37. Transverse section through praelabral region of an embryo in stage somewhat later than F (Fig. 6, PI. I). X 145. me., median cord; pc} (o.g.), first protocerebal lobe (optic ganglion) ; zo., orifice of involution of the intraganglionic thickening. Remaining letters as in Fig. 36. Fig. 38. Transverse section through optic ganglion and optic plate of an em¬ bryo in Stage G (Fig. 7, PI. I). X 145. th., clear thickening in the optic plate; e., eye ; on., optic nerve ; pc} {o.g.), optic ganglion ; am., amnion. Fig. 39. Frontal section through the brain of an embryo somewhat older than I (Fig. 9, PI. I). X 175. p., problematical brain segment; pc} {o.g.), optic gan¬ glion; pc,.^ pc.,^ second and third protocerebral lobes; dc., deutocerebrum; tc., tritocerebrum ; e., eye ; md., mandible ; md.g., mandibular ganglion ; r.g., recurrent ganglion ; st., stomodaeum. Fig. 40. Transverse section of brain through the supracesophageal commis¬ sure of an embryo in Stage K (Fig. 10, PI. I). X 175, nb., neuroblasts; igl., interganglionic thickening; th., clear thickening in the optic plate; coe., head- coelom ; ps., Punktsubstanz ; on., optic nerve. Remaining letters as in Fig. 39. Fig. 41. Transverse section from same series as that represented in Fig. 40^ but passing through the frontal ganglion. X 175. dc., deutocerebrum; f.g., frontal ganglion; sb., suboesophageal body. Remaining letters as in Figs. 39 and 40. THE tJBBARY OF THE OHiVgRSITY OF 11 IINOiS PL IV. s.cm l^-fcUo.g.) w . ..pc^Co.g.) '&my “ r e> S 'i $ mm ■0^---- . -pc (O.CJ.) «iftfi' - 'S pc (O.CJ.) in "i'«.''■ V pcUo.j.) Journal of Morphologij Vol. MR. P.<-' W. M. Wheeler del. Ittk. Anst. vYkrner & Mnter, Amnkfurt 158 WHEELER. EXPLANATION OF PLATE V. {Xiphidium ensiferuni, Figs. 42-46, 48-50. X. fasciatum. Figs. 47 and 51.) Fig. 42. Tip of abdomen in surface view from a $ embryo which has just passed the lower pole. (Stage J, Fig. 9, PI. I.) fifth to eighth abdominal stigmata ; testis ; m.d., vas deferens ; ta.m., terminal ampulla; ap.P appendages of eighth abdominal segment; st. {ap?), stylets (appendages of ninth abd. seg.; ap.,^° appendages of tenth abdominal segment, in the cavities of which the ampullae lie in this stage ; cc. {ap.,'^^) cerci.; prd., proctodaeum ; an., anus. Fig. 43. Tip of abdomen in surface view from a $ embryo somewhat older than the one shown in Fig. 42. Letters same as in Fig. 42. Fig. 44. Tip of abdomen in surface view from a $ embryo somewhat older than the one shown in Fig. 43. v., yolk; ag., last abdominal ganglion. Re¬ maining letters as in Fig. 42. Fig. 45. Vasa deferentia {m.d.), with their terminal ampullae {ta.tn.), from an embryo just before the development of the larval cuticle. Fig. 46. Tip of abdomen of a ^ embryo ready to hatch. Letters same as in Figs. 42 - 44. Fig. 47. Tip of abdomen opened and seen from within from a ^ larva i cm. long, m.o., sexual orifice, c.n., connectives ; remaining letters as in Figs. 42-44. Fig. 48. Tip of abdomen of 9 embryo in a stage corresponding to that represented in Fig. 42. ov., ovary ; f.d., oviduct; ta.f., terminal ampulla; m.d., vas deferens and terminal ampulla of the male type, still persisting; sixth to eighth abdominal stigmata; ap.^ persisting appendage of the seventh abdominal segment; op.^ {(ip‘^)i op.^ i^p-^), op.^ {ap.'^°), three pairs of abdominal appendages which become the gonapophyses (ovipositor) ; an., anus ; cc. (ap.^^), cerci; z/., yolk. Fig. 49. Tip of abdomen of 9 embryo seen in full surface view in Fig. 10. (X), PI. I. mnb., location of median cord neuroblast; cm., posterior commissure ; cn., connective ; ag., last abdominal ganglion. Other letters as in Fig. 48. Fig. 50. Tip of abdomen of 9 embryo ready to hatch. Letters same as in Figs. 48 and 49. Fig. 51. Tip of abdomen of 9 larva i cm. long, opened and seen from within; all the parts being dissected away except the reproductive organs, vg., vagina; op.'^ {ap.^), op.^ {ap‘^), op.^ three pairs of gonapophyses. Other letters as in Fig. 48. int OF THE umvERsiiy OF ii imts Journal of Morpliologij \olJTIL PL V. F i I t ta.ni. st(ap^) ofr(ap^) opUap^) op"lap J I ''op"(ap^) op '^ lap‘°) 47 ts ag. c.n. ia st.(ap'^j ni.o. I I W PI Wheeler del. lith.Arstv. Werner SiWinieTj FrankfurtW. CONTRIBUTION TO INSECT EMBRYOLOGY. 159 EXPLANATION OF PLATE VI. {Xiphidium ensifertwi. Figs. 52-63 ; X. fas datum, Figs. 64 and 65). Fig. 52. Frontal section through first to fourth abdominal segments, show¬ ing segmental arrangement of the gonads. Embryo in Stage F (Fig. 6, PL I). X 230. ml., longitudinal ventral muscle ; gd.^gd.^gd.,^ gonads of first, second and third abdominal segments ; coe., coelomic cavity; ecd., ectoderm ; ep., epithelium. Fig. 53. Transverse section through the third abdominal segment of an embryo in Stage F, (Fig. 6, PI. I). X 230. nc., nerve cord ; am., amnion, sms., somatic wall of somite ; spms., splanchnic wall of somite ; v., yolk ; en., ento¬ derm. Remaining letters as in Fig. 52. Fig. 54. Frontal section through the fourth abdominal segment of an embryo in Stage F (Fig. 6, PI. I). X 230. The diverticula point towards the head, v., yolk ; gd.,^ gonad of fourth abdominal segment; coe., coelomic cavity ; ep., epithe¬ lium. Fig. 55. Section through a somite from the third thoracic segment showing a single enlarged germ-cell protruding into the coelomic cavity. X 230. Fig. 56. Sagittal section through the end of the abdomen of an embryo in Stage G (Fig. 7, PI. I), tb., neuroteloblast ?; nb., neuroblasts; coe.'^-coe.,^° coel¬ omic cavities of the seventh to tenth abdominal segments ; 7 n.d., diverticulum of the tenth abdominal somite which becomes the vas deferens and its terminal ampulla; gd.,^° gonad in tenth abdominal segment (abnormal and atavistic); n.c., nerve cord in unflexed portion of abdomen. Fig. 57. Transverse section through ninth abdominal segment of embryo represented in Fig. 42, PI. V, cutting the tenth pair of appendages. X 175. msc., muscular tissue ; prd., proctodaeum ; n.c., nerve cord ; h., heart; ecd., ectoderm ; m.d., vas deferens ; ta.m., terminal ampulla; ap.,^° appendage of the tenth ab¬ dominal segment. Fig. 58. Transverse section of the abdomen,of an embryo in the stage repre¬ sented in Fig. 48,.PL V. The greater portion of the section passes through the ninth, its anterior portion through the tenth abdominal segment. X 175. coe.,^° coelom of tenth abdominal segment; bl., blood corpuscle ; ta.m., terminal ampulla of vas deferens ; m.d., vas deferens disintegrating ; op,^ {ap.^°), third pair of gona- pophyses ; n.c., nerve cord ; prd., proctodaeum ; ec., ectoderm ; k., heart. Fig. 59. Transverse section through the seventh abdominal segment, taken from the same embryo as the section in Fig. 58. X 175. v., yolk; en., ento¬ derm ; bl., blood-corpuscle dividing; h., heart; coe.P coelom of the seventh abdominal segment; f.d., oviduct; ta., terminal ampulla; oe., oenocytes; ec., ectoderm ; nc., nerve-cord ; ap.p appendage of seventh abdominal segment. Fig. 60. Transverse section through the seventh abdominal segment of an embryo somewhat older than that .in Fig. 48, PL V. X 175. ov., ovary; me., median cord ; ad., fat-body; spms., splanchnic mesoderm. Other letters as in Fig. 59 - Fig. 61. Sagittal section through the head of an embryo in Stage F (Fig. 6, PL I). X 175. frontal ganglion ; first recurrent ganglion ; second recurrent ganglion ; v,, yolk ; pc.,^ third protocerebral lobe ; ms., mesoderm ; lb., i6o WHEELER. labrum; j/., stomodseum; tc.^ tritocerebrum; md.g., mandibular ganglion; en.^ entoderm ; s.b.y subcesophageal body ; coe-d" coelom of mandibular segment. Fig. 62. Frontal section through the head of an embryo in Stage F (Fig. 6, PI. I). X 175. pc.'‘- {o.g), optic ganglion ; md.^ mandible. Remaining letters as in Fig. 61. Fig. 63. Transverse section through the subcesophageal body from an em¬ bryo in Stage G (Fig. 7, PI. I.). X 500. Fig. 64. Section through subcesophageal body of larva 7 mm. long. X 500. Fig. 65. Section through the subcesophageal body of a nymph i cm. long. X 500. f ■ ^ >_Y>. sof' y I 1 I 1 i msc ) f r. C4: ‘ ®;, : - 1 ^ Journ al of Morphology VolMll Pl.W. IV. M. IVheeler del. IiiJt.lTist. V. Werner & Winter, FrarO^fuit V€ c f 4 ’ A'" \ p,-- • W .•> 1. V / >2£j I •.' V^-« llj V., r 'i \ ' •< a ■t ‘ • • f ^ 'A 75*1 a< ■/^ -d ' i university of illinois-urbana 0112 070130668