by chemical methods. 



101 



be taken from large numbers of roots, and that roots sampled in 

 this way are still capable of being used as mothers for seed 

 bearing. The authors adopted the method, but found that dupli- 

 cate samples takeD from 50 roots of the same variety grown side 

 by side gave results differing far more than could be accounted 

 for by errors in analysis. 



This might be due, either to such great individual variation in 

 the roots that 50 roots did not give an average sample, or to a 

 defect in the actual method of sampling. The following figures 

 show that a mixed sample taken from 50 roots should give con- 

 cordant results. The percentage of dry matter was determined 

 separately in cores taken from 150 to 200 individual roots of each 

 variety, and entered in the order in which the roots were sampled. 

 These percentages were then averaged in consecutive tens, 

 twenties, and fifties. 



The above figures show that mixed samples of 50 roots give 

 concordant results, and apparently therefore indicate with suffi- 

 cient accuracy the composition of the roots on the plot from 

 which they were taken. It is clear also that when the mixed 

 samples are taken from only 10, or even from 20 roots, they give 

 no approach to an accurate indication of the composition of the 

 crop, and this probably explains the discrepancies in the results of 

 many experimenters. 



Fig. 1. 



