56 THE COTTON PLANT IN EGYPT CHAP. 



A natural outcome of these methods of observation has 

 been the employment of some of the simpler tools of the 

 statistician. The author having for some years depended 

 on detailed observation of random rows whereby to obtain 

 expressions for the behaviour of plots in ordinary culti- 

 vation, was particularly interested when Prof. Wood and 

 Mr. Stratton combined data and treatment in a critical 

 discussion of field experiment methods. The appearance 

 of their article implies that science has at last been intro- 

 duced into agricultural field experiments, and an examin- 

 ation of the available data respecting cotton should 

 therefore be profitable. The figures for the yield of 

 nominally identical plots in Egypt are too scanty to 

 provide a definite statement of the probable error ; those 

 available work out to a probable error of 6 per cent, of 

 the mean yield for the total yield of half-acre plots ; the 

 same figure in the case of Wood and Stratton's investi- 

 gations for English crops was db 5 per cent. 



The difficulty of obtaining identical plots of Egyptian 

 cotton, even in the same field, is extreme, since great varia- 

 tions in .the physical texture of the sub-soil are normal, 

 which involve similar differences in the water-supply of 

 the roots, when they have penetrated thereto, though the 

 surface soil may appear to be uniform. Consequently, we 

 are not justified in expecting the Probable Error (P.E.) of 

 Egyptian cotton plots to be any lower than that of plots of 

 English wheat and we may take 6 per cent, as a very 

 liberal estimate of the accuracy of cotton plot experiments 

 in Egypt. Taking odds of 30 : 1 as practical certainty it 

 follows that two plots of different varieties grown 

 for comparison side by side may show yields of four 

 kantars and six kantars respectively, and will yet give us 

 no justification for stating that the greater yield is due to 

 the superiority of one variety over another. 



This is a very unsatisfactory state of things, since much 

 money has obviously been wasted in trying such useless 



