Ill DEVELOPMENT AND ENVIRONMENT 



57 



experiments. If further application of Wood and 

 Stratton's analysis can be made to enhance the precision 

 of cotton experiments, the attempt is worth making. 



The method of examining single rows and computing 

 to single plants is capable of high precision. The 

 examination can be carried out daily, weekly, or casually, 

 and may be extended to any observable character, or 

 restricted to mere pickings of the ripe cotton. In the 

 latter case it is actually less laborious than the ordinary 



Week 

 ending 



A ugust 

 20 27 



September 

 3 10 



October 

 1 8 



17 24 



Fig. 42. — Comparison op Varieties. 



Yield expressed as boUing-ourve per mean plant. 

 Six scattered groups of 30-40 plants per variety. 



15 



29 



method, since smaller quantities have to be dealt with. 

 Confining attention to total yield only, the author 

 obtained a series of data from double rows of five varieties, 

 badly cultivated on an extremely irregular strip of land 

 (Fig. 42). Comparing two adjacent rows of the same 

 variety, each comprising from 25 to 35 plants in the row, 

 and expressing the difierences in terms of the mean yield 

 per plant for the two rows, the final figure for nineteen 

 such pairs gave a P.E. of ± 5'5 per cent, for number of 

 bolls, or 6 "8 per cent, for weight of seed cotton. Two 

 little rows growing side by side, of 30 plants each, may 

 thus be made to give as good a comparison for the 



