I30 THE COTTON PLANT IN EGYPT chap. 



In rare cases we have beeu able to demonstrate the 

 existence of ]\Iendelian ratios in Fo, confirmed by the 

 behaviour of later generations. In others we have failed 

 to analyse the Fj but have dissected the F3, where the 

 phenomena were simpler. In others, again, such analysis 

 has been utterly impossible, and we have been obliged to 

 rely on the bare fact that a certain character has " bred 

 true " in the end, though we have been unable to trace the 

 steps of its purification. As a last infirmity, and last 

 resort, we have fallen back on comparisons from year to 

 year in massed data : if the graph for seed- weight in F, 

 shows certain modes, and if those modes reappear in the 

 graph for all the F3 plants in the following year, we have 

 a claim to assume that the modes are at least due to 

 a systematic cause and not to accident. 



The data to be quoted are drawn largely from crosses of 

 Egyptian with American Upland, especially from Afifi with 

 Truitt Big Boll (Xo. 252)," and from Charara with King 

 (No. 255). Other crosses which have been made, but only 

 partially examined through lack of space and labour, are 

 Hindi x Charara, King x Russell, Russell x Charara, and 

 Sultani x King. In the early stages of the work many 

 natural hybrids were examined, which had resulted from 

 natural crossing between Egyptian cottons, while full 

 analysis has been made up to F2 in an inter-Egyptian cross 

 between Afifi and another Sultani. The phylogenetic 

 relationships of the various parents is doubtful, to say the 

 least. The author has leaned to the designation " inter- 

 specific," but this has been questioned, in view of the 

 cultivated origin of the parents.* Perhaps a description 

 of them as '" reputedly inter-specific" would best meet the 

 case. 



The cross of Afifi x Sultani was made with the object 

 of studying some simple examples in place of the com- 



* See references "Egypt," in Sir G. Watt's Monograph. 



