H. Onslow 



285 



closely realised must be due to there being sufficient numbers tu give 

 an average. A glance at Fig. 6 shows that the curve of the com])iiu'd 

 families divides fairly sharply into three classes, the DR class being about 

 twice as numerous as the other two. The individual families, however, 

 which are seen in Fig. 5, show wide divergences. This is probably due 



^ parent 

 of '19 K 



cJ parent 

 of 191 



Colour Units 



Fig. 4 (cf. Fig. 3). Diagram showing the percentage frequency of the black values 

 of the type ^ ^ in the last figui'e. The two t\ families have been combined. 



to the variation in the colour of the parents, i.e. to the modifying genes 

 which shift the range of colour of single families up or down the scale. 

 Family '20 E is, for instance, shifted towards the dark end of the scale, 

 whereas family '20 B is shifted in the other direction, and has, moreover, 

 much the palest male parent. If the curve is considered as a whole, the 

 classes are all a shade darker than the values given on p. 282 ; for in- 

 stance the DD class of " pure rustica " extends as far as the black value 

 1"0. The extracted type males are many of them very dark, some being 

 like No. 2, PI. XXVII, black rather than brown, and darker than the 

 ordinary wild male. The extracted rustica are like the Irish race, but 

 some have a small patch of light grey scales on the milky ground some- 

 what lighter than No. 8, PI. XXVII, and others are more heavily spotted 

 than usual, with a black streak down the costa (No. 10, PI. XXVII). 

 Journ. of Gen. xi 19 



