TESTICLE-CELLS OF' FUNDULUS 145 



such a role, as Cowdry ('16, p. 437) seems to believe, I fail entirely 

 to see. Concerning the hypothesis of their motile function 

 which, first formulated by Benda, reappears occasionally in the 

 literature, I do not see that any arguments have been brought 

 forward in its favor, nor is there any clear expression of how we 

 should imagine this function. Benda considered his 'mitochon- 

 dria' as contractile bodies : how can this conception be applied to 

 the spherical chondriosomes of the spermatozoa of so many inverte- 

 brates and of Fundulus? Furthermore, those who advocate this 

 hypothesis entirely overlook two groups of observations, which 

 we have to accept as long as their inexactitude has not been 

 demonstrated : first, Meves' experiments on the spermatozoon of 

 Salamandra and second, the observations of a number of au- 

 thors, lately Koltzoff's, on the spermatozoa of decapods (see 

 Duesberg, '12, p. 687). 



Finally a few words concerning the occurrence of a constant 

 number of chondriosomes in male germ-cells. 



The first indication of this was given by Meves ('00) who 

 found that the small spermatocytes (i.e., as the apyrene genera- 

 tion) of Paludina vivipara contain on the average eight loop- 

 shaped chondriosomes. Numerations made on spermatids of the 

 same generation a short time after the second division likewise 

 revealed an almost unvarying number of chondriosomes,. this 

 time four. 



Two other cases, much more striking, have been described 

 lately, both in arachnoids, the first one by Sokolov ('13), the 

 other by Wilson ('16). 



In the spermatogonia and in the young spermatocytes of Eus- 

 corpius carpathicus Sokolov describes mitochondria which soon by 

 confluence form filaments. Later rings appear, which are proba- 

 bly formed by fusion of the free ends of the filaments of the pre- 

 ceding stages. The average number of these rings is twenty- 

 four. During mitosis they are not divided as is the case in the 

 small spermatocytes of Paludina, but are segregated into two 

 equal groups between the daughter-cells. Thus each spermatid 

 contains one quarter of the number of rings, on the average six. 



