484 PROCEEDINGS OF THE NATTONAL MUSEUM. vol. xxix. 



periblast, the basal layer of protoplasm being thick with a large vacu- 

 ole and full of ,yolk in its lower part. In some cases, where the plane 

 of separation is reduced to a line, the cells are drawn out into long 

 points toward the base as if a fine thread, used to separate the parts, 

 had elongated the cells downward. 



The only reference to such peculiar conditions as shown in these 

 fig-ures is found in a short section on Coregonus in Ej'cleshymer's 

 paper on Amblystoma (1895, fig. 35 and others). This writer thinks, 

 however, that these divided blastoderms do not result in double 

 embryos. The same seems to hold true for the pipefishes of Beaufort, 

 for although thousands of eggs and larva? and hundreds of adults, 

 alive or preserved, have been examined, only two apparent cases of 

 deformation have been found b}' the writer. The literature of these 

 fishes contains but few references to abnormalities. M. Malm (1862) 

 describes a SyngnathuK with two caudals. Ryder (1884) reports a 

 >Si/ngnat/tu« with two anals. However, Rathke (1837) reports in the 

 Syngnathida? of the Black Sea many abnormalities of the snout, eyes, 

 and tail, duo. he thinks, to retardation of development. 



A fair example of the late stages of segmentation is shown in 

 fig. 27, Plate VI. Here the thirty-eight cells are in three tiers, with 

 one cell high alcove all. There is an elongation in one axis, possibly 

 a derivation of the condition found in the eight-celled stage, and a 

 curiousl}' regular arrangement of certain cells. On the whole, how- 

 ever, the segmentation is very irregular, and it becomes more so later; 

 finally all trace of symmetr}^ is lost, and the blastoderms become 

 almost circular in outline. No surface views of later stages will be 

 given, since, as the cells grow smaller, the blastoderms approach more 

 and more the ordinary teleostean form. 



STAGE OF SIXTY-FOUR CELLS. 



Artificial fertilization being impossible in Slphodonta^ one can not 

 divide late material into stages by hours, and the greatly var3ang 

 shapes of the blastoderms make it impracticable to classif}' sections 

 by the number of rows of cells in each, as some writers do, so it 

 becomes necessarj- to devise an arbitrary scheme. This scheme is to 

 count the peripheral cells in the central section of a blastoderm, then, 

 assuming a like number in a section at right angles to this, by squai'ing 

 this number the approximate number of surface cells is found. The 

 size of the cells serves as a check to this. 



Fig. 5«), Plate IX, with eight peripheral cells, is from a normal type 

 of the sixty-four celled stage. The central periblast (c'.y>.) is thick 

 and yolky, and at the right is a cell not yet cut off from it. The 

 segmentation cavity (.s-. c.) is filled with cells, some of which are ready 

 to divide. 



Fig. 57 is derived from a flat ])lastoderm of the preceding stage, 

 and, by comparison with figs. 45 and 47, Plate VIII, is seen to have 



1 



