No.im. BREEDING HABITS AND EGG OF PIPEinSH-GUDGER. 479 



formed by each of the cells in tigs. T, 8, or 9, Plate V, dividing into 

 two. In tig. If) all the cells save one are practically on the same level, 

 or at most with a gentle curve across the upper surface. In tig. 17, 

 the blastomeres are arranged more irregularly. Fig. -10, Plate VII, is 

 a section in the plane a — h of a l)lastoderm like tig. 1(3, Plate VI, pre- 

 paring to divide into thirty-two cells. Th(> two central cells will 

 divide to form two surface and two interior cells, while the outer cells 

 will each divide into two cells, both on the surface. This is shown 

 l)y the position of the centrosomes. The cells form a gentle arch 

 roofing over a considerable segmentation cavity. The planes of seg- 

 mentation are dilated at their outer ends into vesicles which are cov- 

 ered ])y thin protoplasmic sheets or bridges. Fig. 44, Plate VIII, is a 

 section of some such structure as tig. 17, Plate VI, in the plane a — h. 

 Some blastoderms of this stage have been found in which the four or 

 five cells were not cut off from the basal periblast. l)ut these are too 

 infrequent and too little understood to be reproduced here. Fusari 

 (1890) has figured a section like this for Cri-ttieeps, a goby. 



In fig. 45, Plate VIII, there is shown a section of a fiat-topped abrupt- 

 edged sixteen-celled blastoderm of a type which persists till the prep- 

 aration for invagination })egins. What the appearance of such a 

 blastoderm in surface view would be I can not say; probably it would 

 in no wise differ from fig. 16, Plate VI. The essential difference between 

 figs. 44 and 45, Plate VIII, is the circular groove sharply marking off' 

 the outer periblast (f^. j>.) in the latter. Possibly these figures are 

 derivatives of the one-celled stages shown in figs. 28 and 32, Plate 

 VII. In tig. 45, Plate VIII, there is a large segmentation cavity 

 and a volk-laden periblast. The dotted lines show where the outer 

 periblast has been torn away. Note the large dilatation at the outer 

 end of the right furrow and the protoplasmic bridge covering it. 

 Fig. 18, Plate VI, is a derivative of some such forms as figs. 10, 

 11, 13, and 14. It is arched, but the crest of the arch is not in the 

 center but to one side, and the cells lie in two if not three levels. A 

 section through an almost identical form (in the plane a — b) is shown in 

 fig. 40, Plate VIII, and makes clear its sloping outline and its two 

 excentrically placed high cells. It has one interior cell, which in the 

 next section is clear of the central periblast (c. y^.), and has probably 

 originated by the horizontal division of an outer cell. 



Fig. 19, Plate VI, shows a modification of the arched type. Its six- 

 teen cells are in two layers and the seven upper ones are on an approxi- 

 mate level. Fig. 47, Plate VIII. is a section through some such blasto- 

 derm as the aliove. Its surface slopes gently and the left peripheral 

 cell projects over the outer periblast (o. p.). This latter phenomenon 

 will be found frequently in later stages. Vacuoles are found in two 

 of the division walls. 



The high-arched type of sixteen-celled blastoderm is shown in 



