PROCEEDINGS OF THE "THIRD ENTOMOLOGICAL MEETING 487 



Intensity of attack is always calculated above by the proportion of 

 sound to damaged seed. Calculating the number of damaged seed in 

 a sample is always a difficult and tedious task, but can be done if trouble 

 is taken. None of the samples used for the correlations above men- 

 tioned were made for that purpose, and using the data obtained from 

 them for correlations was not considered until a whole year after the 

 records had been made. They were originally intended for use for a 

 totally different purpose. 



That seed samples which have been damaged will be deficient in 

 oil is evident, but has not been investigated by us. 



The effect of Pink Bollworm attack on the lint has also been studied, 

 Tjut from the quantitative side only. We know that the quality suffers, 

 but have not yet examined this aspect of the question. 



Gimied lint can only be studied in bulk, and can only be understood 

 in connection with the seed from which it was derived. 



The first question to examine was the effect of Pink Bollworm attack 

 •on the percentage lint. 



The percentage hnt is obtained by dividing the weight of lint, after 

 multiplying the lint-weight by one hundred, by the total weight of 

 seed cotton. 



Before we could attempt any examination along these lines, we 

 found it necessary to study the percentage hnt in normal samples in 

 order to understand to some extent the problems involved. Very 

 little literature was available beyond the writings of Balls, which did 

 not throw much light on this subject. 



The percentage Hnt of any lot of cotton can only be known by 

 examining the whole lot. The errors of sampUng are great. They 

 consist of two different kinds, («) an error due to the portion of the 

 sample examined not being truly representative of the whole lot, and 

 (b) an error introduced by the atmospheric humidity at the time of 

 weighing. 



To make, this clear, eighty samples of about 100 bolls each, together 

 forming the whole lot of cotton collected in an uniform field, varied 

 in the percentage lint from 29-5 per cent, to 32-2 per cent. ; the whole 

 lot gave 31-1 per cent.iO-Oi, and the standard deviation was 0-6 per 

 cent. 



The error introduced by the atmospheric humidity can be quite 

 great. Lintweight rises and falls pari passu with the atmospheric 

 humidity, as can be seen from Plates 89, 90. In this case the 

 weight was recorded by a spot of light thrown on a mirror fixed close 

 to the knife-edge of a chemical balance and reflected from thence on 

 to a sheet of gas-light paper revolving on the drum of the hygrometer ' 



