Experimental Studies on Germinal Localization. 217 



of cilia rarely coincide in direction the group does not, as a rule, 

 rotate in a constant direction, but irregularly. 



If now the two products (i^-^ and i"-) of the first division of 

 the primary trochoblasts be separated, each divides once, and only 

 once, thus giving a pair of cells that become ciliated and swim 

 together like the above-described group of four (Fig. 43). If, 

 finally, these two cells be separated at the time of ciliation — i. e., 

 at a period corresponding with the 64-cell stage, no further divi- 

 sion occurs, but In due time each trochoblast develops its row 

 of cilia (Figs. 44-45) and swims singly with the fullest vigor 

 and activity. Such single trochoblasts often rotate steadily in a 

 nearly constant direction, proving that the action of the cilia 

 is normally coordinated. They may live for two days or more 

 when the action gradually ceases and disintegration occurs. 



The history of these cells gives indubitable evidence that they 

 possess within themselves all the factors that determine the form 

 and rhythm of cleavage, and the characteristic and complex dif- 

 ferentiation that they undergo, wholly independently of their re- 

 lation to the remainder of the embryo. Roux's "self-differentia- 

 tion" here appears in the clearest and most unmistakable form. 



Similar results were obtained in Dentalium, but I did not in this 

 case attempt to isolate the trochoblasts individually, but merely 

 allowed entire eggs, or isolated ^ or ^-blastomeres to continue 

 their development In the calcium-free water. As in Patella, the 

 result at the end of 24 hours is a chaos of more or less com- 

 pletely separated cells of different forms and sizes, among which 

 are trochoblasts, actively swimming, singly or In groups. These 

 trochoblasts fall roughly into three groups, large (Figs. 49-51), 

 medium (Figs. 52-53) and small (Figs. 54-55). The large 

 trochoblasts, which are considerably larger than in Patella, are 

 probably primary ones, the medium and small forms secondary 

 ones; and the difference in size among the latter suggest that as 

 in Patella they may arise from different quartets. It is worthy 

 of note that the cilia in all the trochoblasts are considerably longer 

 than in Patella, and are also relatively less numerous and 

 crowded. The small trochoblasts sometimes have as few as six 

 cilia (Fig. 55). Sections of the entire larvae show that the cilia 



