708 PROCEEDINGS OF THE ACADEMY OF [Oct., 



reaching the cleavage cavity, there degenerate. Torrey (1902, 1903) 

 suggests that among these small cells are those two mentioned above, 

 viz., id^-i-2-i and Ic^-^-^-^-^ In Chjmenella the nephroblasts are also 

 small and form part of the dorsal ectoderm of the head, as they do in 

 Arenicola. What the fate of these cells is in the case of Dinophilus 

 I do not know. 



The significance of the annehd and molliiscan crosses and of bilateral 

 cleavage in general has been most thoroughly discussed by recent 

 writers on cell-lineage, and on this particular point I have nothing to 

 add. Since, however, it is generally agreed that bilateral cleavages 

 are not directly referable to purely mechanical causes, but are the 

 result of unknown factors, which cause the throwing back of the 

 bilaterality of the adult upon the embryo, it is remarkable that this 

 appearance of the adult bilaterality should have occurred in the same 

 cell generation, in the same direction, and resulting in the same 

 cell pattern as in the Polychseta. This resemblance is still further 

 continued in the direction of division of the posterior cross arms. 



The rosette cells at the time of their formation lie in the furrows 

 between the stem cells, but almost immediately afterward they move 

 in a clockwise direction (cf. la^-^-^ and la*-^-^ in fig. 20), so that the cells 

 now lie interradially, instead of radialh^, as in the annelid cross (text 

 fig. V, B). 



At a stage numbering between one hundred and one hundred and 

 fifty cells, the rosette divides dexiotropically and equall3^ Its cells, 

 which possibly later undergo another division, form a cluster of cells 

 much smaller than those surrounding it, and can be thus distinguished 

 nearly to the time of the closure of the blastopore. As far as my 

 observations on both fixed and living material extend, the rosette 

 never bears cilia. In Capitella also an apical tuft of cilia is never 

 present, though the rosette is formed as in other Annelida. 



The further history of the first quartette is comparatively simple. 

 The forward movement of the cells of the first quartette has already 

 been described (see text fig. IV, A-E). The limits of the first quar- 

 tette have been determined by two landmarks, the cells 2a^--^ and 

 2(,2.2.i Q^ ^YiG right and left sides respectively, and that chain of small 

 cells which I have identified as the prototroch. These latter extend 

 over the sides, passing just anterior to the cells 2a--' and 2c---^, and 

 around the ventral side as a continuous row, but on the dorsal side 

 there is the wide gap so often found in the early stages of the trocho- 

 phorc larva, and due to the same cause, viz., the non-participation of 

 the products of 2d in the formation of the prototroch. Owing to this 



