1 88 JOHN S. DEXTER. 



when compared with the normal females. The figures in either 

 of these experiments show an average of about eighty-five indi- 

 viduals in which the coupling persists to one in which the factors 



are interchanged. 



EXPERIMENT III. 



After carrying the above experiments to near their conclusion 

 I suspected that perhaps there was some reason unknown to me 

 which made N naturally couple with R, no matter what the 

 original combination might have been. That my suspicions 

 were ill-founded was demonstrated by the following experiments. 



In the first of these, I selected from the classes of NW females 

 and YR males of the F 2 generation (Table III.) obtained in 

 Experiment II., virgin individuals and mated them. The 

 gametic constitution of these and the nature of their offspring 

 according to the theory are shown in Table V. 



TABLE V. 



THE Pi GENERATION OF THIS TABLE CONSISTS OF MEMBERS OF THE Ft GENERATION 



OF EXPERIMENT II. 



Pi. NWXYWX = NW9 YRXYW = YR<? 



Gametes NWX, YWX YRX, YW 



Fi (i) NWX YRX = NR 9 (3) NWX YW = NWtf 



(2) YWX YRX = YR9 (4) YWXYW = YWtf 



This table shows that equal numbers of NR females, YR fe- 

 males, NW males, and YW males should be produced. I ob- 

 tained : 



NR females, 45 NW males, 52 



YR females, 50 YW males, 47 



I now took some of the virgin females from the NR group and 

 inbred them to the YW males. It should be recalled that the 

 parents of the FZ generation in Experiment II. were also of these 

 classes, and therefore, as shown by Table VI., if body color and 

 eye color were not to couple, w r e should expect the same classes 

 of individuals in the p2 generation in each case. If coupling 

 occurred here (i. e., if having " broken " once, should not give 

 any tendency to do so again), we should expect here, not the 

 same results as those obtained in the previous experiment, but 

 rather that the classes that were then small should now be the 

 large classes, while the large ones should be now small. 



