TESTICLE-CELLS OF FUNDULUS 139 



existence of nucleocytoplasmic exchanges, as the nucleus is cer- 

 tainly not a sort of impermeable rubber-vesicle enclosed in the 

 cell. But it would be rash to base on the mere existence of such 

 appearances as described above any definite conclusion. The ar- 

 guments for the cytoplasmic nature of the chondriosomes I do 

 not want to repeat here and refer the reader to former papers, 

 limiting myself to state that no indications of a nuclear origin 

 can be found in the staining reactions.^ 



During the mitotic division of the primary spermatogonia the 

 shape of the chondriosomes changes somewhat: they round up 

 and become more regular (figs. 4 and 5). Their location in the 

 cell is also modified: at the stage of metaphase they surround 

 the spindle (fig. 4) and later are found between the daughter- 

 nuclei (fig. 5).'^ 



Next to these cells are others differing but slightly from them. 

 They are somewhat smaller in size and their chondriosomes are 

 not quite so coarse. These cells are assembled in rosettes of 



the rabbit and quite lately on Ciona, where the chondriosomes form the mate- 

 rial of the yellow crescent, the continuity of which has been demonstrated by 

 Conklin. 



^ The original colors of the preparations could not be reproduced in the plates; 

 as is well known, they are, in acid fuchsin-methylgreen preparations, red for the 

 chondriosomes and green for the chromatin; in Benda's preparations, dark pur- 

 ple for the chondriosomes and pale brown for the chromatin. 



' Concerning the fate of the chondriosomes during the mitotic division of the 

 spern.atogonia of Passalus, Shaffer expresses himself as follows (p. 410): "the 

 spermatogonial cysts which are in mitotic activity, stand out very clearly in con- 

 trast with the resting cysts. This is because of their lighter staining capacity; 

 whether this in turn is due to the partial disappearance of the mitochondria 

 could not be ascertained." Shaffer quotes Buchner as having found that in 

 Gryllotalpa vulgaris, the chondriosomes disappear during or just before cell- 

 division and gives three possible explanations "for the partial loss of mitochon- 

 drial structure dating mitotic activity." Interesting though they may be, these 

 explanations appear to me for the present useless, as, after my own experience, 

 chondriosomes do not disappear during mitosis, no more in Gryllotalpa, as 

 I have shown ('10), than in any other case I know of. 



Payne ('17) quotes both Buchner and me and sees no reason why we should 

 differ so much in our observations: "In this case, one or the other has certainly 

 made a mistake." Between a negative result, however, and a positive one, there 

 is, in my opinion, no room for hesitation. It must be added that since, Buch- 

 ner has considerably modified his attitude towards the chromidial theory, as 

 appears from a text-book he recently published. 



