

EFFECTS PRODUCED BY PINK BOLLWORM ON THE YIELD OF COTTON SEED. 297 



In the first place, it must be stated that there is considerable difficulty in sampling. 

 The true percentage of lint for any given quantity of seed and lint can only be found 

 by examining the whole quantity, and even then it will be correct only for the 

 conditions of atmospheric humidity and temperature under which the examination 

 was made. The smaller the unit taken as a sample, the greater the probable difference 

 of the percentage lint found from the true percentage for the whole. The smallest 

 unit possible for a " percentage lint " is a single seed with its lint. Table XIII 

 shows how enormously the percentages found for single seeds vary ; the extreme found 

 for any one boll being from 26 per cent, to 62 per cent. As against this, the single 

 locks of any one of the seven bolls examined have not varied amongst themselves 

 more than 6 per cent., and the bolls only 4 per cent. 



It is of interest to remark that the examples with " freak " percentages recorded in 

 this table are all seeds which have not been injured by pink bollworm, but whose 

 growth has been arrested at an early stage from some other cause. The seeds giving 

 57 per cent., 62 per cent., and 58 per cent., weighed, without lint, 26, 44, and 38 

 milligrammes respectively, the average weight of a seed from the three bolls being 

 106 milligrammes. Stunted seeds do not however invariably give extreme 

 percentages of lint, as can be seen from the same table. Their influence on the average 

 for the whole boll is necessarily small, and of little practical importance. 



Incidentally it may be mentioned that there appears to be no correlation between 

 seed-weight and percentage lint ; as worked out from 80 samples of Sakellarides the 

 correlation was only 0'009 + 0-074. This is a much lower figure than that found 

 by Craig (Notes on Cotton Statistics in Egypt, L'Egypte Contemporaine, 1911, p. 188) 

 for seed-weight and percentage lint. Nevertheless the figure given by that author, 

 - 220 + 0'094, was already a very low one. It appears, moreover, very unlikely 

 that there should be any correlation between seed-weight and high percentage of 

 lint. The percentage of lint must rise (1) if the lint- weight per seed remains constant 

 and the seed-weight falls, or (2) if the seed-weight remains constant and the lint- 

 weight rises. A rise of both factors simultaneously would only produce a rise or 

 fall in percentage lint if the two rises were not proportionate to each other. That 

 this happens constantly in all samples of cotton, humidity being the driving force, 

 will be shown later. 



The correlation between lint produced per seed and percentage lint is very high, 

 working out at - 906 + 0D13. This high correlation was to be expected, seeing 

 that seed-weight and percentage lint are not correlated and remembering that seed- 

 weight and lint-weight are the only factors used for finding the percentage lint. 



The correlation between the lint produced per boll and percentage lint is much 

 lower, working out for the same samples as used for the above at only - 367 + - 064. 

 Craig found for the same two factors r=0 - 316, which is fairly close to the present 

 figures. The reason why the correlation is so much lower when made per boll instead 

 of per seed is that the number of seeds per boll enters as a disturbing factor. The 

 very high correlation (r = 0'810 + - 035) found by Craig for seed-weight and lint- 

 weight was to be expected, owing to the relatively small limits within which ginning 

 outputs or percentages of lint vary in Egyptian cotton. 



To test the amount of variation in the percentage lint of a single sample of seed 

 cotton, 8,000 bolls of Sakellarides cotton were collected in 1916 at Giza. They all 



