GENERAL DISCUSSION. 17 



The spermatogeuesis of Stenopelmatus, therefore, differs from that 

 of the other Orthoptera which have been described in having (i) a 

 larger number of chromosomes (46), (2) an even number in the sper- 

 matogonia, (3) an accessory chromatin structure in the spermatocytes 

 of the first order, which disappears before the second maturation 

 division. 



\nBlattellavfe. have a typical "accessory chromosome," accord- 

 ing to McClung. It appears (i) in all resting spermatogonia closely 

 associated with a nucleolus, (2) in the spermatogonial mitoses as an 

 odd chromatin element, making 23 in all, (3) in the growth stage of 

 the spermatocytes connected with an end of the spireme and also with 

 the nucleolus. It becomes separated from the other chromatin in the 

 tetrad-stage, remains nucleolus-like in form, and later appears in the 

 first maturation division either among the chromosomes or in a more 

 or less aberrant position. It passes into one of each pair of sperma- 

 tocytes of the second order, persists during the rest stage, appears in 

 the second mitosis as a dyad and then divides, going into one-half of 

 the spermatids. The spermatids, however, as in Stenopelmatus, all 

 have the same appearance : each has in the center not against the 

 nuclear membrane a small element that stains like chromatin. Occa- 

 sionally a mass of chromatin is found outside the nucleus, but this is 

 not constant enough to support the contention of Moore and Robinson 

 ('05) that the ' ' nucleolus " of the related form, Periplaneta americana, 

 is fragmented and cast out into the cytoplasm. The spermatids all 

 appear to develop equally well for some time, but as they approach 

 maturity a varying proportion of them become degenerate. This can 

 not, however, be due to absence of the accessory chromosome, as Miss 

 Wallace supposes, in the spider ; for in some follicles no degenerate 

 spermatozoa are found, and in others more than half may be degen- 

 erate. All attempts to study fertilization stages of the egg have so 

 far failed, and the chromosomes in the female somatic cells have not 

 proved favorable for counting. Twenty-three have been counted in 

 several cases, but there was always some chance of error. If 23 is 

 the somatic number in both sexes, it must be maintained by union of 

 sex-cells containing n and 12 chromosomes, respectively, the same 

 unequal number occurring in the maturated eggs as in the sperm. 

 Under such conditions it is difficult to see how the odd chromatin 

 element of the spermatozoa can determine sex. 



The brief description of the chromatin element x in Sagitta, intro- 

 duced here because it behaves like the accessory chromosome in 

 many particulars, serves as an example of the occurrence of such an 

 element in the spermatogenesis of a hermaphrodite form, where it can 



