NO. 1431. BREEDING HABITS AND EGG OF PIPEFLSII—GUDGER. 483 



arranged in three laj^er.s, lis highest cell is slightl}' eccentrically 

 placed, and one of the axes of the blastoderm is somewhat longer 

 than the other. Fig. 52, Plate VIII, is a central section through a 

 similar but slighth' older blastoderm. The marginal cells are sharply 

 marked ofl' from the outer periblast {<>. ]>.). The arch is high and 

 round. On the left, two cells ai-e imperfectly separated, and a tongue 

 of protoplasm, from Avhich a cell has been cut off, projects into the 

 large segmentation cavit}'. The periblast, torn otf at the right, is in 

 the center reduced to a mere film of protoplasm with much 3'olk 

 adiierent })elow, thus giving it the breadth as drawn. 



Fig. 53 shows a structure by no means uncommon in the egg of 

 Slphontonid. It is a thirty-two-celled stage in which no periblast has 

 yet been formed. The cells are in two layers, the long cell on the 

 upper right is nearly ready to divide, and underneath the whole is a 

 thick layer of protoplasm in which three vertical cell walls extend 

 downward and are lost. Later transverse walls will appear and cut 

 these cells out of the syncytium, finally leaving a periblast layer 

 below. There is a very small segmentation cavity (.s. c.) and the large 

 cell to the right has a vacuole (v.). Ziegler (1882, fig. 2) figures an 

 almost identical structure for the salmon. Kowalewski (1886, figs. 1 

 and 2) portra3"S essentially the same conditions in the goldfish. Hoff- 

 mann (1888, figs. 6 and *J, especiall}') describes a similar structure in 

 the salmon germ. And latest of all His (1898, figs. 7 and 10) confirms 

 the figures and descriptions of the earlier workers on the Salmonoids. 



Fig. 26, Plate VI, is a very interesting divided blastodei'm of this 

 stage with eighteen cells in one division and fourteen in the other. 

 Such structures have been met wnth occasionally in stages of from 

 sixteen to sixty-four cells, but especially a])Ound in the eggs from one 

 fish. Out of twenty of these eggs killed in picro-acetic, five were like 

 the one figured. That these wei'e not artefacts is shown ))y the fact 

 that eggs of the same lot killed in formalin also contained divided 

 blastoderms, the numbers of which were unfortunately not noted. 

 In each division a segmentation cavity exists, and the line of separa- 

 tion is broad and definite down to the periblast. These points are 

 brought out very definitely in fig. 5-1, Plate VIII, a section through a 

 similar but older blastoderm. In the left half there is a small segmen- 

 tation cavity (.v. c. ); on the right, however, there is none. There is 

 no pei-iblast. Cells have been cut out of the mass of protoplasm, 

 leaving a thick germ basis in which are found vertical cell walls and 

 a number of vacuoles (/'.), and which is filled below with fragments of 

 yolk. Fig. 55, Plate IX, is a divided sixty-four-celled stage of the 

 thick-ended type. The furrow between the two parts is here not so 

 wide. In other blastoderms this may swell out to a vesicle at the 

 bottom or be reduced to a mere line, as in the two-celled stages above. 

 There is a segmentation cavity in each portion^ but there is no distinct 



