EFFECTS PRODUCED BY PINK BOLLWORM ON THE YIELD OF COTTON SEED. 309 



That the percentage of attack is no index of the numbers of Gelechia larvae present 

 at any moment, was indicated in a previous paper. From Table XVIII, it can 

 be seen that in 1917 at Gemmaiza the highest figure for population (580 attacked 

 bolls per 100 plants) was obtained at a time when only 64 per cent, of the bolls 

 were attacked. The highest percentage (94 per cent.) was found after the population 

 had been reduced to one-quarter of the previous figure (142). It may also be 

 noted that there were comparatively few green bolls at the time when the highest 

 percentages of infestation were observed. This can readily be gathered from the 

 columns showing the items dealing with the vegetative characters. 



A factor of considerable importance in forecasting the damage is to know how 

 evenly the attack is distributed amongst the green bolls in a field. Apparently 

 the moths do not exert much trouble in selecting bolls, and do not hesitate to lay 

 on such as have already been injured by earlier broods. This in itself ought to 

 lead to quite a simple and fairly even distribution of the damage ; and in fact we 

 find that, as was expected, the standard deviation is low when most of the bolls 

 are sound (4 per cent.) or damaged (7 per cent.), that is, at the beginning and the 

 end* of the season, but that it rises to a quite considerable figure (11 per cent.) 

 when the infestation has reached about 50 per cent. (The figures given are for 

 samples of 100 bolls each). However, the problem is not merely a simple one of 

 binominal distribution, as it is disturbed to some extent by the fact that the supply 

 of bolls to infest is subject to considerable fluctuation, and that the quantities 

 of young bolls produced after the middle of the season are very restricted. Thus 

 the average age of the boll population is also undergoing considerable variation. 

 It seems fairly evident, in the light of the examination of large numbers of sets 

 of 100 green bolls during 1917, that it is not safe to rely on percentages obtained 

 from 100 bolls ; ten such samples are required to produce a sufficiently low probable 

 error. 



The sets of 100 green bolls examined during 1917 were derived from two sources. 

 One part was collected by the Inspectors of Agriculture, who sent in 120 bolls 

 weekly from each Markaz in their province ; these bolls were supposed to be taken 

 a few here and a few there. No selection was to be exercised, but the collectors 

 may have been biased to take the older bolls ; if this happened, too high percentages 

 might be expected. The other part consisted of all the green bolls found on the 

 100 cotton-plants which were examined weekly at Gemmaiza. This lot may be 

 considered to have been taken entirely without bias. 



In addition to these two lots, special sets of samples were taken in the Wadi 

 Tumulat (Tel el Kebir) and at Armant ; 900 bolls were received weekly from the 

 former, 1000 from the latter, place. These figures and the Gemmaiza figures have 

 not been used in obtaining the provincial figures. 



In analysing all these sets of samples, we find considerable agreement in the 

 proportion of samples whose percentages of sound bolls lie between 100-76 per 

 cent., 75-51 per cent., 50-25 per cent, and 24-0 per cent. The period required 

 for the infestation to spread from less than 25 per cent, to over 75 per cent, appears 

 to be a relatively short one. The ultimate damage to the crop depends greatly 



* At the end of the season the numbers of samples available are insufficient for the 

 purpose of calculating the standard deviation with accuracy. 



