1 86 Edmund B. Wilson 



probably due to an antecedent nuclear division without cell divi- 

 sion, or to multipolar mitoses such as now and then occur in both 

 spermatogonia and spermatocytes. 



As regards the chromosome-nucleoli of the growth-period, the 

 contrast between those of the 21 and 22-chromosome forms, or 

 between either of these forms and those with higher numbers is 

 usually at once apparent; but in very many cases where more than 

 one supernumerary is present the number of components can only 

 here and there be clearly seen. Contrary to what might be expected 

 from their compact form, the compound chromosome nucleoli 

 seem to be rather difficult of proper fixation, their components 

 often clumping together or breaking up more or less when they 

 coagulate. I infer this from the fact that different slides differ 

 materially in the clearness with which these bodies are shown. 



Two discrepancies, apparent or real, should be especially men- 

 tioned. One is the difficulty of recognizing the larger supernu- 

 meraries in the somatic groups. As already explained, these chro- 

 mosomes, like the idiochromosomes, appear relatively much 

 larger in the somatic groups than in the first maturation division 

 (owing to their univalence in the latter case) ; but we should expect 

 to recognize them more clearly, at least in the female groups, than 

 is actually the case. This is perhaps due to their undergoing a 

 greater degree of condensation than the others during the growth- 

 period; but I am not sure that this explanation will suffice. A 

 second discrepancy, which may involve an important conclusion, 

 is that in perfectly clear views of the first division, the number of 

 supernumeraries is often less than would be expected from the 

 spermatogonial groups. This is notably the case with femoratus, 

 No. 40 (Fig. 9, h-f), which has clearly 26 spermatogonial chro- 

 mosomes, but very rarely shows 16 in the first division, the usual 

 number being 15. A similar discrepancy has been noted in other 

 individuals, and in several of the types. Since the typical num- 

 ber in all these cases appears in some or many of the first sperma- 

 tocytes, I long supposed the occasional deficiency to result from an 

 accident of sectioning. I now incline to believe, however, that in 

 some cases one (or possibly more) of the supernumeraries may 

 really disappear (by degeneration ?) during the growth-period, 



