R. C. PUNNETT 81 



they consisted of the three classes erect long, erect round, and hooded 

 long approximately in the ratio 2:1:1. It is probable that the rela- 

 tion between E and L is similar to that existing between B and L, and 

 between B and E, but at present the exact nature of the series must 

 be left undecided. 



(4) The relations hetiueen B, E, and L in plants lulncli are 

 heterozygous in all three factors. 



As previously pointed out ((4), p. 7) there are four ways of mating 

 to produce such plants, viz. : 



(a) BEL X bel, 



(/3) BeL X bEI, 



(7) BEI X beU 



(3) bEL X Bel. 



Owing to misadventure an F^ generation has been raised from the 

 first two crosses only and we may now proceed to consider the results 

 separately. 



(a) Nature of mating BEL x bel. 



The appearance of a single red hooded round plant in 1910 (cf (6), 

 p. 297) made this mating possible and an F.^ generation was raised in 

 1912. 'Details of 13 such families are given in Table V. The results 

 are evidently complex but some attempt at analysis may be made by 

 considering separately each pair of the three factors concerned. The 

 figures then become 



(«) BL : Bl : bL : bl :: 479 : 58 : 66 : 143, 



(6) BE : Be : bE : be :: 532 : 5 : 6 : 203, 



(c) EL : El : eL : el :: 479 : 59 : 66 : 142. 



In each case the nature of the gametic series is evidently /i : 1 : 1 : n 

 where n > 1. Since for any two of the factors the nature of the 

 original cross was AB x ab the "coupling" form of the series in each 

 case was to be expected. But it is equally evident that the value of n 

 in cases (a) and (6) is different from what it would have been had the 

 plants been heterozygous for the two factors in question alone. The 

 experimental numbers in (a) are not on the 7 : 1 system bub the results 

 are given far more closely by the gametic series 5BL : IBI : IbL : 5bl. 

 On this supposition expectation would be 501 blue long : 57 blue red : 



