414 INVESTIGATION BY CULTURE EXPERIMENTS 



the points of chief interest are that from the four or five best 

 yielding plots the annual losses per acre are probably not more 

 than 50 pounds of nitrogen, i^ pounds of phosphorus, and 7 pounds 

 of potassium. Dyer assumes an average drainage of 10 inches per 

 annum for Rothamsted, which would reduce these figures by one 

 fourth, but he also suggests that the losses in drainage are probably 

 greater now than they were in 1866-1869. 



While the drainage certainly accounts for most of the loss of 

 nitrogen, there remains not accounted for an annual loss of about 

 7 pounds of phosphorus and 50 pounds of potassium per acre. 

 Dyer suggests that these losses are to be accounted for by descent 

 into the subsoil. The data for potassium, representing the " acid- 

 soluble" only, are too uncertain to warrant any conclusion. In 

 the author's opinion it is not improbable that some of the potas- 

 sium, applied as soluble potassium sulfate, may have reacted with 

 silicates and formed compounds that are not dissolved by strong 

 acid. This seems less doubtful when we consider the proper- 

 ties of cement, and the changes that occur even in a short time 

 in the " setting " of that material. 



The data afford practically no evidence for the "descent into 

 the subsoil of either phosphorus or potassium. The phosphorus 

 determinations so nearly represent the total amounts that they 

 serve satisfactorily for general computations, and as an average 

 they show less phosphorus in the subsoil of the plots where phos- 

 phorus has been applied, although plots 5 and 14 are exceptions. 



At least most of the unused phosphorus remains in the plowed 

 soil. Thus plots 4 and 5 have produced almost the same average 

 yields, and plot 5 contains 1121 pounds more phosphorus in the 

 first 9 inches, but only n pounds more in the second depth, than 

 plot 4. The third depth shows a different relation, but this is 

 reversed in the case of plots lob and n, whose average yields are 

 not markedly different. 



With 2250 pounds of phosphorus in the surface 9 inches, it 

 would require about one inch of erosion in 35 years to account for 

 an annual loss of 7 pounds of phosphorus. This would also ac- 

 count for 10 pounds additional loss in nitrogen, and it seems the 

 most probable explanation. Land that has sufficient slope to 

 provide any surface drainage will suffer some erosion if such drain- 



