(d) That the difference between the select population and the 

 total population really represents lost crop. 



There is great probability that the amount of damage done to a 

 boll by the worms infesting it is proportional to the number of worms 

 themselves. Put otherwise, each worm eats about the same amount, 

 and n worms will eat n times as much as one worm does. Whilst 

 allowing for small individual differences, taken in the bulk, over large 

 samples, the rule can be expected to hold good. It has been utilised 

 to find out, as far as it can be used for this purpose, what the maximum 

 number of worms is that can be present in a pickable boll. But it 

 may conveniently be stated here that there is little likelihood of 

 finding a correct answer to the problem by direct observation. By 

 the time a boll is ripe and picked, some of the worms it had harboured 

 will have migrated. They leave traces, but the traces are hard to class 

 as belonging to one or more worms. The same worm may leave 

 traces behind in two locks, and these traces might be ascribed to two 

 worms. 



During the last five years large numbers of samples of ripe 

 "pickable" cotton bolls have been examined in the laboratory of the 

 Entomological Section, for estimation of the amount of damage due 

 to the pink boll worm. The examination requires that the number 

 and weight of all the bolls in the sample be ascertained. The bolls 

 are individually examined, and the sound ones separated. The cotton 

 from the sound bolls is weighed, and the average weight multiplied by 

 the total number of bolls in the sample. The difference between the 

 total weight of the sample and the " ideal " weight it would have had 

 if all the bolls had been sound is the loss. This loss is, for the pur- 

 poses of comparison best expressed as a percentage of the ideal weight. 

 And here it may be mentioned, that the individual variation of the 

 sound bolls amongst themselves in weight has not been overlooked. 

 It has been found that this variation is likely to introduce serious 

 error, unless the samples are of such a size as to eliminate the effect of 

 individual fluctuations. This makes the determination of loss in 

 samples of high percentage attack perplexing, owing to the increasing 

 difficulty in obtaining sufficient sound bolls to base an average on, 

 and to the number of sound bolls used being too small as compared 

 to the total sample. The safest figures are obviously those where the 

 sound bolls are in the majority, or certainly not less than say one- 

 third of the total. 



Two hundred and seventy-seven such samples have been used on 

 Table V to produce the column percentage loss. They have been 

 arranged according to percentage, from three to three per cent. Each 

 figure contains all the results available for itself and the percentage 

 above and the percentage below. Samples containing less than 50 bolls 



** 



