METHODS OP INVESTIGATION 179 



fairly definite. More precision can be obtained by label- 

 ling flowers on successive days, and collecting the bolls 

 which ripen from them some seven weeks later. In this 

 case the boll can be left on the plant as long as may be 

 convenient, but the method involves more trouble and 

 risk of mistakes in identification . 



In either case absolute exactitude is unattainable, since, 

 although the maturation period is definite (vide supra) for 

 any given strain, it also has a definite range of fluctuation 

 from accidental circumstances acting on the individual 

 boll or plant . The probable error of the maturation period 

 is 3 per cent, in Egypt ; this on forty-eight days is one and 

 a half days, or, in other words, half the bolls observed will 

 mature between forty-six and a half and forty-nine and 

 a half days, and no ordinary accident can make the boll 

 take more than fifty-three or less than forty-three days. 

 Therefore, even if we define either end of the maturation 

 of the particular boll with complete precision, we cannot 

 be quite sure of the age of the boll at the other end on any 

 given day; we can, however, define exactly what the 

 chances are, and from this we can deduce the size of 

 sample necessary to smooth out these accidents. In 

 practice a ten-boll sample is quite satisfactory, and fifty 

 bolls are ample. 



The seed-cotton so collected is weighed to determine 

 the average weight of the boll-content, combed to deter- 

 mine lint length, and ginned with concurrent determina- 

 tion of the ginning out-turn and the mean seed weight. 

 Subsequently the lint is graded, and tested for break- 

 ing strain. The numerical statements thus obtained 



