No. 3.] AMPHIOXUS AND THE MOSAIC THEORY. 589 



6. In three instances gastrulas about yi the normal size 

 were produced from 4-celled stages (two of the three are shown 

 in Fig. 53). Since there is reason to believe (see below) that 

 the yi blastomere never gives rise to a gastrula, I believe the 

 %. blastomere must have been broken by the operation into 

 fragments, from which these minute forms arose. The gas- 

 trulas were nearly typical in form, with nearly the normal ratio 

 of size between the cells of the two layers. 



7. The }i blastomeres are of two sizes (micromeres and 

 macromeres) which, as far as could be determined, do not differ 

 essentially in mode of development. The isolated blastomere 

 segments in a form (Figs. 1 21-129) approaching that of a 

 complete ovum but 7tever ideittical with it. In rare cases a 

 y^ blastula is formed, either closed or with a pore at the lower 

 pole ; but the gastrula-stage is never attained. As a rule, 

 however, the yi blastomeres produces a flat or curved plate of 

 cells, of which those at one margin are at first larger (Figs. 129, 

 130-134). All gradations exist between blastulas having a 

 cleavage-pore and the curved plates. In later stages the curved 

 plates often appear at the same age in two varieties, in one of 

 which the cells are much larger than in the other (Figs. 132, 133). 

 These two forms are similar to fragments of ectoblast or 

 entoblast of the normal gastrula at the corresponding age, and 

 suggest the view that the corresponding blastomeres have 

 undergone a partial development — i.e., as if they still formed 

 part of a complete normal embryo. This subject requires 



ftirther investigation. All the forms of yi embryos acquire 

 cilia and swim actively about. The plates invariably died after 

 12 to 18 hours ; the closed blastulas lived in some cases for 

 two or three days. {Cf. Driesch, No. 8, p. 9, who has observed 

 ^ blastulas of the same type, which likewise lived for several 

 days without forming a gastrula.) 



8. Incomplete separation of the blastomeres of the 8-celled 

 stage gives rise to a great variety of forms which, however, I 

 have not been able to study satisfactorily. Among them were 

 typical Yt, and % gastrulas (Figs. 54, 56) as well as double 

 gastrulas ; but these may have arisen from 4-celled stages 

 mixed with the 8-celled. 



