PROCEEDINGS OF THE THIKD EXTOlfOLOGICAL MEETING 489 



Gelechia attack can produce results ranging from total suppression 

 of the production of Jint to an apparent rise in the percentage of lint 

 produced. 



Table XIII has been prepared to show the limits within which 

 fluctuations may under the present conditions (crop 1917) be most 

 frequently expected to lie. All the samples came from the same crop, 

 being Sakellaridis grown on one feddan of land at Gemmaiza in 1917. 

 For the purposes of this examination 30 sets of 100 sound seeds were 

 selected as control, with 30 sets of 100 damaged seeds and 30 sets of 

 50 " double " seeds (=100 seeds per set) for comparison. As additional 

 control 30 random samples each of about 150 grammes weight were 

 taken from the same material. 



It may be stated at once, that all of these samples with the exception 

 of the random ones may be seriously biassed on account of sampling. 

 In picking out the sound seeds, the operator may have unintentionally 

 selected the largest, and, in selecting the damaged seeds, a rather more 

 than the average damage may be expected as the standard. The 

 " double " seeds are on the other hand probably quite representative ; 

 as being rare every one found would be retained, and as they are fairly 

 easy to recognise by feeling before ginning, all those encountered would 

 be kept. The only test we can apply for unbiassed sampling is by 

 comparing the total weight of the 3,000 seeds in question, with 

 the average weight of 3,000 seeds as calculated from the " Gemmaiza 

 crop " figures, which may be accepted as being quite free from bias, 

 as will be seen later. The " Gemmaiza crop " samples came from the 

 same field as the others, but do not include them. 



From the comparison it would appear that the sound seeds selected 

 were below the average for " Gemmaiza crop " sound bolls. This was 

 to be expected, if the sampling was unbiassed, as in the seed cotton 

 from which the seeds were selected sound and attacked bolls were 

 inextricably mixed together. In any case it may be maintained, that 

 the good seeds selected were not above the average in weight. Similarly 

 it will be seen that the random samples varied very little (about 1-5 per 

 cent, only) in weight from corresponding quantities of " Gemmaiza 

 crop," of the same quahty. This was expected in this case, as there 

 was no reason for any bias to be introduced. 



The " Gemmaiza crop " figures have a greater claim to be accepted 

 as accurate. There is no possibility of errors in sampling ; the entire 

 yield has been taken under circumstances which exclude error as far as 

 is possible ; the sound and the damaged bolls were separated from each 

 other carefully, and although a small percentage of attacked bolls was 

 included under the sound ones, the proportion of attacked seeds wrongly 



