METHODS OF INVESTIGATION 195 



so carefully extracted in such a way as to include fibres 

 from all parts of each of the five lots are very often ex- 

 tremely wild; the probable error for the average of a 

 single bunch is 24 per cent. Had it been practicable to 

 effect the work on ginned lint, this figure would have been 

 very much lower; but it is quite useful to have a detailed 

 example of these sampling difficulties, for the routine 

 method just described was most strictly followed in obtain- 

 ing every bunch tested in the Daily Picking Series of 1912. 



Grading. The only grading data included in the 

 account of the dated samples are those for strength. The 

 determinations were made by Mr. Harold C.Thomas, of the 

 National Bank of Egypt, Alexandria, and they are a strik- 

 ing instance of the accuracy to which the grader's hands 

 can attain; the samples graded were about 8 grammes in 

 weight, and were given to Mr. Thomas in irregular 

 sequence, in three separate batches, marked with dummy 

 reference numbers which bore no relation to their actual 

 daily sequence; Mr. Thomas knew only that he was 

 grading these dated samples from a familiar pure strain. 

 In spite of these precautions against subjectivity, it will 

 be seen that his hands assigned sample after sample to 

 what was obviously its correct place in relation to its 

 neighbours, and that there are only one or two wild 

 points in his strength curve. 



Such results emphasize the futility of attempting 

 to introduce so-called " scientific methods " into the 

 ordinary commercial practice; grading by hand has its 

 limitations, and so have the scientific methods; each has 

 its proper function, and the results of each are of interest 

 to the other. 



