Experiniental Studies on Germinal Localization. 247 



the case of only a few kinds of cells. It is evident that the limi- 

 tations of potency vary in different cells — the i^ cells, for exam- 

 ple, contain more complex potencies than the 1= — and It is quite 

 possible that dependent or correlative differentiation may play a 

 larger part in the development than my experiments have thus 

 far shown. But the proof by experiment of definite specification 

 and self-differentiation in only a few categories of the early cleav- 

 age-cells establishes a principle that is to be reckoned with as a 

 most important factor in the whole problem of embryonic differ- 

 entiation. If, in considering some aspects of this problem, I 

 again take up a discussion that has been so prolonged, it is because 

 I believe that the importance of the principle of mosaic-develop- 

 ment, and of the nearly related one of specific formative or deter- 

 mining stuffs, has received insufficient recognition by many embry- 

 ologists, and by some has been prematurely discredited. That a 

 reaction is well under way will be evident to everv reader of Fis- 

 chel's ('03) excellent recent discussion of development and dif- 

 ferentiation, the essential conclusions of which agree closely with 

 those earlier stated in a brief form in my paper on cleavage and 

 mosaic-work ('96, appended to Crampton's paper), and more 

 fully considered in the discussion of my nemertine results ('"03) ; 

 the agreement with the conclusion reached in this and the preced- 

 ing paper is still closer. In full harmony with the same general 

 conception are the important cytological results of Lillie ('99, 

 '01) and Conklin ('98, '99, '02) cited in the two preceding 

 papers. 



The long continued discussion of the mosaic-theory of develop- 

 ment that followed its first definite formulation by Roux in 1888, 

 in the course of which Roux so ably defended his position, has 

 been greatly prejudiced by the fact that the experimental analysis 

 of cleavage was at first confined to the so-called "indeterminate" 

 types of cleavage, such as those of the echinoderm, the medusa 

 and Amphioxus ( it may for the time be left an open question 

 whether that of the frog should be placed in the same class) . The 



directed towards the role played in development by the segregation and locali- 

 zation of cytoplasmic materials. Roux himself ('03) has now abandoned the 

 second of these "Annahmen" (qualitative nuclear division). 



