CROSSINGOVER IN DROSOPHILA 187 



nial division with the follicular envelope just forming up to the 

 large mature eggs ready to pass down the oviduct, were 69 and 

 70. The two females would have in two ovaries of this size 138 

 and 140 oocytes respectively. The correspondence between 

 these actual oocytes present and the numbers given above cal- 

 culated from the genetic evidence is exact, and practically es- 

 tablishes the conclusion, that the change in the percentage of 

 crossingover, and therefore without much doubt the crossing- 

 over process itself, takes place in the very early oocytes just after 

 the last oogonial division. The cytological picture at this stage 

 does not properly belong in this paper, but it will be of interest 

 to say that at this point the nuclei of the nurse cells and the 

 egg proper are indistinguishable, and the chromosomes in all 

 are in what resembles a late leptotene or early diplotene con- 

 dition in other forms. 



Corroborative evidence 



Exceptional tube counts. What remains to be said is simply 

 corroborative of the above conclusion. If the crossingover 

 process is determined at this early stage, and if for any reason 

 the eggs are not laid continuously after the heat has been with- 

 drawn, but after the high point has been registered by previous 

 counts, then the high percentage should be maintained when the 

 process of laying is begun again. Such a case occurred in the 

 series recorded in table 14. Female No. 626 had been exposed to 

 heat with the other pairs in the series and the percentage of 

 crossingover among her offspring laid between February 16 and 

 18 showed, like the others, a slight increase. Between February 

 18 and 28 no eggs hatched in vials 626. On February 28 the 

 record shows that the female was served with a new male, and 

 from that point on offspring appeared as before Because of the 

 interruption, caused perhaps by the sterility of the previous 

 male, the results were not added in with the remainder of the 

 series. When the percentages of crossingover among her off- 

 spring laid between February 28 and March 2 were figured, it 

 was found, table 14, that they showed the high value, although 

 the remainder of the series had been giving the control value for 



