viii HEREDITY 137 



excess of the paler parental colour-type. Testing this on 

 the assumption of a 9:3:4 ratio, we obtained consistent 

 results up to a certain point. Whites and some full 

 yellows bred true, some intermediates threw out all colours 

 like the F x ; other intermediates threw whites only, 

 while yet others threw only the full parental yellow, 

 giving approximate 3 : 1 ratios in both cases. It seemed 

 at one stage that the double-pair hypothesis had met all 

 contingencies, until the following test was applied. 



A family raised from an F 3 plant which bore the 

 intermediate colour had given twenty-two intermediates 

 to seven full yellows. Six of these F 4 intermediates 

 were grown into F 6 ; the obvious expectation was that 

 four should throw out yellows, while two should breed 

 true to the intermediate colour. The actual result was 

 that none threw any yellows at all, but all threw whites. 

 These whites were not, moreover, of the same colour as 

 the parent white, but much nearer the intermediate itself. 

 The figures for the six families, in ratio of " intermediate : 

 new white " were (4 : 0), 21:5, 16:9, 26:1, 10:1, 11:3. 

 The families were almost gametically pure in all other 

 known respects, and all vicinists had been eradicated. 



From this evidence it is plain that petal colour 

 in crosses of Upland by Egyptian may be controlled by 

 not less than three pairs of allelomorphs. The presump- 

 tion is that our matching methods are not sufficiently 

 precise, and that some form of colorimetric grading is 

 needed. Even the inter-Egyptian cross shows the same 

 peculiarities. 



We shall see that similar evidence is to hand in the 

 next character to be considered, together with a strong 

 probability for gametic coupling. If the colour and 

 marking of a flower's petal is controlled by at least five 

 allelomorphic pairs, complicated by gametic interaction, it 

 need not surprise us to find that the modes of seed-weight 

 in F 2 , for instance, are not very definite. 



