FOOD FOR PLANTS 175 



and it is desired to calculate the percentage of the 

 applied nitrogen that was recovered, it is necessary 

 first to deduct from the total amount of nitrogen 

 recovered in the crop, the amount recovered from the 

 check cylinder, and thus account for the soil nitrogen 

 that the crop used. 



It is at once obvious that this cannot be an abso- 

 lutely correct method of determining the percentage 

 recovered, since in those cylinders to which nitro- 

 genous fertilizers have been applied, the plant will 

 make a quicker start and the roots go farther in 

 search of the nitrogenous materials of the soil than 

 in the check cylinders where there is a pronounced 

 deficiency of available nitrogen, and thus the check 

 fails to be a true check. In this way it happens that 

 the recovery may apparently be more than 100 per 

 cent, as shown in Series 8B, for the years 1901 and 

 1910. However, there appears to be no way of over- 

 coming this error so long as the work is carried out 

 in the natural soil and if one starts with an artificial 

 soil, other and more serious difficulties arise. 



In this work no effort has been made to analyze the 

 roots, since it would be well-nigh impossible to do 

 this correctly, and even if it could be done the same 

 error would be introduced. The roots and stubble 

 are left just as under field conditions so that the 

 residual effects of these may be observed. 



To draw conclusions from 1 to 2 years of such 

 work would be manifestly unfair, but when it is car- 

 ried on for a period of 10 or 20 years, seasonal dif- 

 ferences, differences due to the unequal decomposi- 

 tion of organic matter and differences due to slight 

 errors, which are sure to creep in now and then, are 



