EFFECTS PRODUCED BY PINK BOLLWORM ON THE YIELD OF COTTON SEED. 295 



weight. There is no reason to suppose that this reduction in germination is in 

 consequence of any poisoning of the seeds ; it is merely an outcome of starvation 

 during the early stage of development of the seed in the capsule. Any other cause 

 which reduced the weight of the seed produced would probably have the same 

 effect. Balls writing in 1910, at a time when Gelechia was rare, already remarked 

 on the better germination results of heavier seed. 



C. The Number of Seeds, sound or attacked, developed m Attacked Bolls. 



The number of seeds normally set in a sound boll appears to be subject to consider- 

 able variation. Sound bolls of Sakellarides produce on an average 15 to 16 seeds 

 per boll, the limits actually observed by the present writer being from 4 to 24. 

 The three divisions of a boll do not necessarily each produce the same number of 

 seeds ; asymmetry in this respect appears to be very frequent, if not indeed the rule. 



It has been found that errors ha sampling in the field spoilt the material taken 

 in 1916 for the purpose of an investigation into this question. Thus it is doubtful 

 whether the samples on which Tables VII and X are based actually consisted of 

 100 bolls each, as it was intended that they should do. The discussion of this point 

 is consequently based on material collected with greater accuracy in 1917. 



The main difficulty that prevents one (in Egypt) from establishing the normal 

 number of seeds in a sound boll is the probability of including, by an oversight, 

 attacked bolls, when trying to obtain perfect samples. Sampling in the field leads 

 instantly to such inclusion of attacked bolls, as the external appearance of the 

 least damaged ones is very deceptive. For this reason the apparently sound bolls 

 taken in the field for this work were each packed separately at the time of picking, 

 and were examined seed by seed in the laboratory. Even with this precaution it was 

 found impossible to obtain many samples with less than two or three per cent, of 

 attacked seeds. 



Supposing these best samples, which exteriorly showed no sign of attack, which 

 contained no double seeds, and in which very few damaged seeds were ultimately 

 discovered after ginning, to be normal, and contrasting them with samples consist- 

 ing of material taken from damaged bolls, the following conclusions appear to be 

 justified : — 



(1) The ripe boll cannot be taken as the unit when estimating the severity 



of the attack, as ripe bolls vary in the degree of attack from being 

 almost perfect, to being totally destroyed. 



(2) The intensity of the attack is better gauged from the proportion of sound 



to attacked seeds. This proportion has been shown in the two pro- 

 ceeding chapters to be correlated to seed weight and germination. 



(3) The intensity of attack appears further to be correlated to the number 



of seeds produced in a given number of bolls. The correlation is as 

 high as 0-889 + 0-063 (see Table XII). There are very distinct indica- 

 tions that progressive increase of intensity of attack is accompanied by 

 progressive decrease in the number of seeds present at the time of 

 ripening. 



