TESTICLE-CELLS OF FUNDULUS 143 



exactly six granules while Benda's preparations of the same 

 stage (fig. 19) show an undecipherable heap of mitochondria. 

 Later when the asymmetry of the head has become conspicuous, 

 we find almost invariably four mitochondria (fig. 16), disposed 

 with remarkable regularity upon the posterior facet of the head. 

 Finally in the ripe spermatozoon the number is still more re- 

 duced, usually to three. Here Benda's material is more service- 

 able than Regaud's owing to the change in the staining reactions 

 of the head mentioned above. A comparison of the different 

 stages of this evolution, as they appear after fixation in Regaud 

 fluid, shows that the increase in volume of the mitochondria is 

 not directly proportional to their decrease in number (figs. 13 to 

 18) ; and, as there is no evidence of an elimination of mitochon- 

 dria, one would be led to believe in a strong condensation of the 

 chondriosomal substance. This conclusion is however not sup- 

 ported by Benda's preparations and I am forced to admit that 

 the swelling produced by the formalin-bichromate mixture is 

 greater in the first stages of spermiogenesis than in the later 

 ones. 



As stated above, the average number of mitochondria in the 

 ripe spermatozoon, as counted in Benda's preparations, is three. 

 They are especially conspicuous in face- views (fig. 21), where 

 they are found regularly disposed on the posterior facet of the 

 head. Occasionally spermatozoa are found with four, five or 

 even six granules taking the chondriosomal stain. The majority 

 of these granules are undoubtedly mitochondria and in such cases 

 the fusion has, for some unknown reason, apparently not pro- 

 ceeded normally. Whether it is completed later is difficult to 

 say. It is probable also that occasionally the centrioles are 

 stained, for in certain cases it was possible to recognize a rela- 

 tionship between the proximal extremity of the axile filament 

 and a small granule stained like a chondriosome (fig. 20). I 

 cannot give any definite information about the behavior of the 

 centrioles during the spermiogenesis of Fundulus,^ but there is no 

 doubt that they are located in that region. 



* One thing however is certain: that their behavior is very different from the 

 same in selachians (Suzuki, '98). 



