Spermatogenesis in spiders. 545 



remember, however, that the errors are all on one side and that 

 they tend in every case to favor undercounting", this table will be 

 of value. 



First of all, it seems probable that the very low and very high 

 counts are due to such divisions as is shown in Fig-. 76. As pointed 

 out, in the description of this stage, such cells are not common. 

 The rest of the cells show from three to nine supernumeraries ; and 

 when we bear in mind the chances of undercounting-, this table 

 tends to show that the supernumeraries are fairly evenly distributed. 

 At the same time it must be admitted that the irreg-ularities shown 

 in Table I, might be interpreted in another way; namely, that the 

 distribution was more irreg-ular than was supposed. 



The planosomes as a rule do not divide in the second maturation 

 division but they lag- behind in the spindle and ultimately go to 

 one pole or the other. Twenty late telophase stages have been 

 observed in order to determine the ultimate distribution of these 

 bodies. The results are given in Table II. Only cells showing a 

 cell plate are included in this count and it should be remembered 

 that each count gave the distribution for two spermatids. 



Table II. 

 No, of supernumeraries 1 2345678 



No. of cells showing 411851632 



This table brings out two points clearly. First, the planosomes 

 do not divide in the second maturation mitosis, and second, it shows 

 that the spermatids receive a varying number of these bodies, no 

 general law being followed in their distribution. 



After the telophase, a nuclear wall is formed and the autosomes 

 of the spermatids become resolved into a very faintly staining reti- 

 culum. At this time the hétérochromosomes retain their compact 

 form and appear as densely staining nucleoli (Fig. 84). It is possible, 

 by counting the nucleoli at this time, to determine approximately 

 the number of hétérochromosome elements present in the spermatids. 

 Out of eleven counts taken at random, we get the following: 



