246 Edmund B. JFilson. 



entire first quartet of Patella, when isolated, produces a mass of 

 ectoblast-cells, which, though It closes, does not gastrulate, but 

 undergoes essentially the same differentiation as if It formed the 

 upper hemisphere of a complete larva. The isolated quadrant 

 of a 4-cell stage gastrulates, produces a group of trochoblasts and 

 an apical organ, the latter structure appearing apparently in any 

 of the quadrants in Patella, while in Dentaliuin It Is restricted to 

 the D-quadrant. In Dentaliiim, further, only the D-quadrant 

 produces a post-trochal region, which Is due to the fact that this 

 quadrant alone contains the material of the lower polar area from 

 which arises the somatoblasts. Finally, the two halves of the 2- 

 cell stage gastrulate, but (at least in Dentalhim) differ widely in 

 their later development. Both in Dentalium and in Patella the 

 half-embryo forms a prototroch, which In the former seems al- 

 ways to close to form a complete ring, but In Patella frequently 

 remains open at one side, forming a half ring. In Patella both 

 halves form an apical organ; in Dentalium only the CD-half. 

 In Dentalium, finally, only the CD-half forms a post-trochal 

 region, for the same reason as in case of the D-quadrant. It is 

 probable, further, that only the CD-half and the D-quadrant pro- 

 duce coelomesoblast. This conclusion has not thus far been sat- 

 isfactorily established by direct examination of the half-embryos, 

 but is Indirectly rendered very probable through the observations 

 on the lobeless larvae recorded in my preceding paper. 



The foregoing facts constitute a strong body of prima facie 

 evidence that the entire cleavage-pattern in the molluscan egg 

 represents (with certain reservations considered beyond) a mosaic- 

 work of self-differentiating cells, exactly in the sense of Roux's 

 general conception.^ The proof is indeed entirely complete In 



^Here, and in all that follows, I exclude from that conception the hypothe- 

 sis of qualitative nuclear division. It should be borne in mind that Roux 

 himself expressly stated as early as 1893 that this hypothesis did not form a 

 necessary part of his conception. "Die beiden AmuiJimefi" (nuclear idioplasm, 

 distribution by qualitative division) "sind jedoch nicht unerlasslich noth- 

 wendige Glieder mciner in iliren wesentlicJten Theilen experimentell erwles- 

 enen Auifassung;" ('93.2, p. 874, Italics in the original). This fact has been 

 ignored by many of Roux's critics, in spite of the fact that some of his most 

 important contributions to experimental embryology have been specifically 



