abnormal). There can be little doubt that Dasyurus, like various other Marsupials (e. g. Perameles, Macropus, etc.), has suffered a progressive reduction in the number of young reared, but even making due allowance for that, the excess in production of ova over requirements would still be remarkable enough. Whether this over-production is to be correlated in any way with the occurrence of abnormalities during early development or not, the fact remains that cleavage abnormalities are quite frequently met with in Dasyurus.

Technique. As fixatives, I have employed for ovaries the fluids of Hermann, Flemming, Ohlmacher, and Zenker; for ova and early blastocysts, Hermann, Flemming, Perenyi, and especially picro-nitro-osmic acid (picro-nitric acid [Mayer] 96 c.c., 1 per cent. osmic acid 2 c.c., glac. acetic acid 2 c.c.); for later blastocysts, the last-named fluid especially, also picro-corrosive-acetic and corrosive-acetic.

To facilitate the handling of ova and early blastocysts during embedding, I found it convenient to attach each specimen separately to a small square of pig's fœtal membrane by means of a dilute solution of photoxylin (1 to 2 per cent.). Orientation of the specimen was then easily effected during final embedding, under the low power of the microscope. The larger blastocysts were double-embedded in photoxylin and paraffin, the cavity of the blastocyst being tensely filled with the photoxylin solution by means of a hypodermic syringe fitted with a fine needle.

For the staining of sections, Heidenhain's iron-hæmatoxylin method proved the most satisfactory, and was almost exclusively employed. Entire portions of the blastocyst wall were stained either with Ehrlich's or Delafield's hæmatoxylin.

I am much indebted to Mr. L. Schaeffer, of the Anatomical Department of the University of Sydney, and to Mr. F. Pittock, of the Zoological Department, University College, for invaluable assistance in the preparation of the photomicrographs reproduced on Plates 1-5, and also to Mr. A. Cronin, of Sydney, and Miss M. Rhodes, for the drawings from their respective pencils reproduced on Plates 6 and 7.

To Miss V. Sheffield I am indebted for the original of fig. 63. Το my friend Dr. F. P. Sandes, Sydney, I am indebted for kind help in the revision of certain parts of the manuscript.

CHAPTER I.-CRITICAL REVIEW OF PREVIOUS OBSERVATIONS ON THE EARLY DEVELOPMENT OF THE MARSUPIALIA.

Apart from the very brief abstract of a short paper on the development of Dasyurus, which I read before Section D of the British Association in 1908 (included in Dr. Ashworth's Report, 'Nature,' vol. lxxviii), our knowledge of the processes of cleavage and germ-layer formation in the Marsupialia is based (1) on the well-known observations of the late Emil Selenka ('86) on the development of the Virginian opossum (Didelphys marsupialis), published in 1886 as Heft 4 of his classicalStudien'; and (2) on those of W. H. Caldwell ('87) on the uterine ovum, and cleavage process in the native bear (Phascolarctus cinereus).

Selenka's account of the mode of origin of the germ-layers in Didelphys differs widely, as the sequel will show, from my description of the same in Dasyurus. Now Didelphys and Dasyurus are two marsupials, admittedly allied by the closest structural ties, and we should therefore not expect à priori that they would differ fundamentally in the details of their early ontogeny, however much they might diverge in respect of the details of their embryonal nutritional arrangements.

Furthermore, we might reasonably hope, in view of the generally admitted relationships of the Marsupialia, that a knowledge of their early development would aid us in the interpretation of that of Eutheria, or, at least, that their early developmental phenomena would be readily comparable with those of Eutheria. It cannot be said that Selenka's observations realise either of these expectations. "Whichever view is taken of Selenka's description of the opossum," writes Assheton ('98, p. 254), "many obvious difficulties remain for the solution of which no satisfactory suggestion can as yet be offered."

From

As concerns my own observations, I venture to think it is possible to bring them into line with what we know of the early ontogeny in the other two mammalian sub-classes, and I have attempted to do so in the concluding chapter of this paper, with what success the reader can judge, whilst as regards the divergence between Selenka's results and my own, I am perfectly convinced that the explanation thereof is to be found in the fact that the whole of Selenka's early material was derived from but two pregnant females, and that much of it consequently consisted of eggs which had failed to develop normally. From the one female, killed 5 days after coition, he obtained one egg in the 2-celled stage, one with about twenty cells and nine unfertilised ova. the second, killed 5 days 8 hours after coition, he obtained "ausser zwei tauben, 14 befruchtete Eier nämlich je ein Ei mit 4, 8, 42, 68 Zellen, eine junge und eine ältere Gastrula mit noch dicker Eiweisschicht und endlich acht auch gleichen Entwickelungsstufe stehende weit grössere Keimblasen, deren Wand noch grösstentheils einschichtig war" ('86, p. 112). Selenka recognised that the last-mentioned blastocyst "die normale Entwickelungsphäse repräsentiren," since he found as a rule that all the embryos from one uterus were in the same developmental stage. Nevertheless he proceeded to describe the segmenting eggs and the two "gastrula" which lagged so far behind the blastocysts, as if they were perfectly normal developmental stages. He does, indeed, question whether or not the 42-celled stage is normal, but decides in the affirmative, “denn wenn ich von zwei Zweifelhaften Fällen absehe, so habe ich niemals Eier aus den ersten Tag aufgefunden, welche auf irgend welche Anomalie der Entwickelung hinwiesen." This, however, can hardly be accepted as a satisfactory reason for his conclusion, since apart from the other eggs of the same batch, he had but the two eggs from the other female for comparison, viz. the 2-celled egg (and even that is, in my view, not quite normal), and the 20-celled egg, which is stated to have suffered in preparation. With the exception of the two eggs just mentioned, all the crucial

early stages (ranging from the 4-celled stage to the completed blastocyst), on whose examination Selenka based his account of germ-layer formation in Didelphys, would thus appear to have been derived from a single female. No wonder it is impossible to reconcile his description either with what we know of germ-layer formation in the Prototheria and Eutheria or with my account of the same in Dasyurus.

My own experience with the latter has shown me that no reliance whatever is to be placed on segmenting eggs or blastocysts which exhibit marked retardation in their stage of development as compared with others from the same uterus, and also that batches of eggs or blastocysts in which there is marked variation in the stage of development attained should likewise be rejected. Abnormalities in the process of cleavage and of blastocyst formation are by no means uncommon in Dasyurus, and during the earlier stages of my own work I spent much time and labour on the investigation of just such abnormal material as that on which Selenka, no doubt unwittingly, but I feel bound to add, with an utter disregard for caution, based his account of the early development of Didelphys.

I propose now, before passing to my own observations, to give a short critical account of Selenka's observations, my comments being enclosed in square brackets.

The uterine ovum of Didelphys is enclosed by (1) a relatively thin "granulosamembran," formed by the transformation of a layer of follicular cells [really the shell-membrane, first correctly interpreted by Caldwell ('87) and formed in the Fallopian tube]; (2) a laminated layer of albumen, semitransparent; (3) a zona radiata, not always recognisable [in my experience invariably distinct].

Cleavage begins in the uterus, is holoblastic, and at first equal. A 2-celled stage is figured (Taf. xvii, fig. 3) [not quite normal as regards the relations of the blastomeres], and also a 4-celled stage [normal in appearance except for the

1 The collection of my own early material of Dasyurus has involved the slaughter of over seven dozen females.

enormous thickness of the albumen layer], in which the four equal-sized blastomeres are radially arranged round a cleavage cavity and are conical in form, their upper ends being more pointed, their lower ends thicker and richer in yolk-material. The nucleus of each is excentric, being situated nearer the upper pole. [This description is applicable word for word to the 4-celled stage of Dasyurus.]

An 8-celled stage (fig. 6) is next described, seven of the blastomeres being equal in size and one being smaller. They are arranged somewhat irregularly in two circles. [This stage I regard as abnormal both in respect of the arrangement of the blastomeres and the occurrence of irregularity amongst them.] Selenka (p. 119) thought it probable that the third cleavage planes cut the first two at right angles and divided each of the first four blastomeres into a smaller ectodermal cell and a larger more granular entodermal, but states that he was unable to establish this owing to the opacity of the albumen-layer. [My observations show that it is the fourth cleavage in Dasyurus, not the third, which is equatorial, unequal, and qualitative, and that even then the cells formed are not ectodermal and entodermal in significance. The albumen is normally never opaque.]

A 20-celled stage is mentioned, but not described, since it suffered in preparation. It is said to have a large entoderm cell in the cleavage cavity. [A statement of very doubtful value, since the blastomeres were admittedly pressed together and probably displaced by the shrunken egg-membranes.]

The next stage described is a spherical "gastrula" (Taf. xvii, figs. 7, 8), composed of forty-two cells with an open "blastopore" at the vegetative pole, a smaller opening at the animal pole, and a large "ur-entoderm " cell in the cleavage-cavity, just inside the "blastopore." The wall of the "gastrula" consists of cells graduated in size; those in the region of the blastopore are the largest and richest in deutoplasm, those at the opposite pole are the smallest and most transparent. [This is a very characteristic stage in the formation of the blastocyst, with which I am quite familar in Dasyurus. Selenka's speci