112 



S. IKENO : 



2)riori very probable, I could find no such iudicatiou at the beginning of niy 

 experiment. On studying, however, the IVgeneratiou of the present cross, I 

 have found some reds in the oflfepriug of one magenta i^^'P'irent (Table YI, 

 B: Magenta, No. 6). This points out natiu-ally towards the composite 

 nature of the magenta factor, and I was able to establish the fact liy the 

 Cross-experiment YIII. 



Cross ym. 3Iagentaxom7H/i'. (PL II, fig. 1 and 3). 

 ecu BBS X CCrrhh F^ = CCBvBb. 



The i^,-hybrid beai-s magenta flowers (s. the Table of Coloui's, p. 96}. 

 The i^'.-ofispriug are composed as in the following Table : — 



Table ^TE, A. 



Fj-generation (1017). 



The deviation of the actual results fi'om the theoretical calculated on the 3 : 1 

 basis is 17, and consequently 2'6 times the standard error ( = 6'ü). Thus the 

 actual results are not in very good agi-eement with the expected ; the chief 

 cause of this discrepancy is in all probability to be sought in the jxxjr 

 germination of seeds, and I think that we have here in spite of discrepaucj- 

 a case of segregation into 3 magentas and 1 orange.' The lack of i^eds in 

 this case is evidently due to the absolute Unkage of B and J5, as it was the 

 case in the Cross Y, "\'I and YII, or at least to the linkage of very high 

 intensity where so few reds are es^Dected that they will not appear at all 

 unless an enormous number of plants are in cultivation. 



Now what is very remarkable about the i^.-oflspring derived fi'om the 

 /''.-plants under discussion is the fact that soma of the magentas have uuder- 



1 We may liave 1 such case out of about 112 trials .is the result of mndom sjimpUng. 



