R. C. PUNNETT 85 



the "repulsion" series, and the low value of n in the present case 

 offered the most favourable opportunity hitherto met with for testing 

 this point. Accordingly large numbers of pcjUinations from normal 

 fertile plants (NF) were made on to sterile cretins (nf). In not a single 

 instance however was the operation successful and it wns found subse- 

 quently that the cretin, whether producing pollen or not, is always 

 sterile on the female side. The pods, and even the seeds, may some- 

 times undergo swelling but in none of the numerous cases examined 

 was a viable seed produced. The (NF x nf) plant had therefore to be 

 sought for in another direction. 



Since the gametic series produced by the plant formed from the 

 two gametes Nf x nF is of the nature INF : 3Nf : 8nF : Inf such 

 plants should give rise to the following zygotic forms : 



NNFF 1 NNff...9 nnFF ... 9 nnff ... 1 



NNFf 6 Nnff ... 6 nnFf ... 6 



NnFF 6 



(Nf xnF) ... 18 

 (NF X nf) ... 2 



normal fertile 33 : 



normal 

 sterile 



15 



cretin] 

 fertile) 



15 



cretin ] 

 sterile f 



Hence out of 33 normal fertiles in such a family there should be 5 

 distinct classes distributed in the following proportions : 



1 (NNFF) giving only normal fertiles. 



6 (NNFf) giving normal fertiles and steriles only. 



6 (NnFF) giving normal and cretin fertiles only. 



18 (Nf X nF) giving all 4 classes with "repulsion" between N 

 and F. 



2 (NF X nf) giving all 4 classes with "coupling" between N 

 and F. 



30 such normal fertiles from F^ families were grown on to give an 

 F3 generation in 1912. The expected 5 classes were all found, and as 

 shewn below (and also in Table IX) the proportions in which they 

 occurred accorded closely with expectation. 



