A. H. Trow 263 



Let us now examine the variability of the hair character in the six 

 remaining groups of Table XII, in which the parents were all homo- 

 zygous for hair (HH). Table XIV gives the result for those groups 

 whose parents were R7\ Table XV for those whose parents were MR 

 or rr. 



In Exp. 61 the hairiness varies in degree, shewing the three grades 

 H', H" and H\ and the RR, Rr and rr types are irregularly distributed 

 amongst these. There are so few H^ types and these are so loosely 

 distinguished from H- types that the colony may be arranged thus : — 



92 plants 



I — 



I I 



34H3 58H2 



1 I 1 I — I 1 



III ill 



2'21tR WRr 2rr 4RR 31Rr 23n- 



The numbers, although 58 : 34 might represent 9 : 7, shew that there 

 is apparently no real segregation of hair factors in this case. The 

 difiSculty of arranging the plants in definite groups according to their 

 hairiness supports this view. We have therefore to explain as best we 

 can the remarkable association of RR and W, and rr and H-. We are 

 perhaps justified in assuming that the greater type of hairiness — H' — 

 is due to the association of the hair factor H with R. On this 

 assumption 



HHRR plants should be very hairy and of the type H^, 



HHRr „ „ „ less hairy „ „ „ ,, H", 



and HHrr „ „ „ slightly hairy „ „ „ „ HK 



The facts revealed by Exp. 61 agree very well with this assumption. 



Applying the hypothesis to the explanation of the results of Exps. 63 

 and 68, we are at once met with a difficulty, the standard of hairiness is 

 constant at the grade H" and the segregation of the ray character does 

 not affect it. 



This difficulty may be removed by amplifying the original hypo- 

 thesis. Let Z be an unknown factor which when present nullifies the 

 stimulating effect oi R on H. Then, since the presence o{ R promotes 

 hairiness in Exp. 61, praecox is probably a sz plant, and since the hair 

 character is constant in Exps. 63 and 68, erectus and multicaulis are 

 probably ZZ plants. 



Such a hypothesis is somewhat involved and perhaps unnecessary ; 

 but as it explains the results hitherto obtained and can readily be tested 



