144 THE COTTON PLANT IN EGYPT chap. 



An interesting proof of the tangibility of these 

 gradations in fuzziness was afforded by a natural hybrid 

 from Abbassi, bearing normal Egyptian slight fuzz on its 

 seed. When selfed it threw recessive wholly naked seed 

 derived from an unknown pollen parent. When it was 

 crossed with American Upland we raised seven Fj families, 

 four of which contained the typical Egyptian seed as one 

 plant in sixteen, while three contained the naked seed in 

 the same proportion. 



The behaviour of Hindi, the seed of which is devoid of 

 fuzz, when crossed with Egyptian (Charara) has been 

 somewhat remarkable. In the first generation the seed 

 was entirely fuzzy, like an Upland. This was probably a 

 case of reversion due to the meeting of cryptomeres, and it 

 was naturally expected ^^' ^^ that the Fj would show a ratio 

 of 9 : 3 : 3 : 1, both the Egyptian fuzz and the Hindi fuzz 

 reappearing. An unsuccessful sowing gave " 14 entire : 3 

 Egyptian," which seemed to support this view. In the 

 following year a family of 130 Fj plants was raised, but 

 without producing a single Hindi seed ; the result was the 

 same as if Uplands had been used instead of Hindi, 

 namely, " 123 entire : 7 Egyptian." The absence of the 

 Hindi type in a family of this size can scarcely be 

 accidental, and a few more crosses of this kind might 

 throw light on the phylogeny of the Hindi cotton. 



This complex inheritance is made even more interesting 

 by the fact that a cross of the same Hindi strain with a 

 natural hybrid of Hindi and Afifi yielded simple 

 segregation of the naked Hindi seed in Fo. 



Summarising the evidence with respect to this character 

 we have tolerably convincing data as to the increasing 

 complexity of inheritance when passing from varietal to 

 reputedly inter-specific crosses, this complication taking 

 the form of an additional pair of allelomorphs. In one 

 case we have found reversion through the meeting of 

 cryptomeres, and in general it would seem that the less 



