THE ROTHAMSTED EXPERIMENTS 395 



1 1-2 produced an increase of 845 pounds of hay over plot u-i, 

 as a 3o-year average. When we remember that the sulfates of mag- 

 nesium and sodium contain large amounts of water of crystalliza- 

 tion, and that potassium sulfate is an anhydrous salt, the value of 

 potassium for its own sake is still more questionable. 



Attention is called to the fact that the total weight of salts 

 applied to the best-yielding plot (11-2) is greater than the total 

 weight of field-cured hay produced on the unfertilized land, as an 

 average of the last zo-year period. 



It seems very probable that the benefit of the alkali salts is due 

 in part at least to their power to increase or maintain the solu- 

 bility of the phosphorus, and thus provide a means by which that 

 element is carried deeper into the soil, where it may be taken up 

 by the plant roots. Even then it is probable that a very consider- 

 able part of the phosphorus applied to The Park plots during the 

 past half-century still remains within an inch or two of the surface. 



The botanical composition of the herbage (first crops only) is 

 given in the last four columns of Table 70; first for the average of 

 nearly fifty years, and second for the season of 1902. It is espe- 

 cially interesting to note the large percentages of legumes on plots 

 6, 7, and 15, which receive the minerals alone and consequently 

 must depend upon legumes for a supply of nitrogen. Plot 8 (miner- 

 als, except potassium) shows the next highest percentage of leg- 

 umes in 1902; and, in proportion to the actual application of 

 anhydrous alkali salts, this is relatively higher than the figures 

 indicate. 



Plot 16, which receives the minerals and the smaller application 

 of nitrate, shows about the same percentage of legumes as the 

 unfertilized plots and the acid-phosphate plot. Where heavy 

 applications of nitrogen are used, the legumes are almost lacking, 

 and entirely so in a few cases. 



On some plots the herbage is largely weeds. Thus, the 1902 crop 

 of plot 2 (unfertilized since 1864) consisted of 30 per cent of 

 grasses and legumes and 70 per cent of weeds, so that the produce 

 is deteriorating in quality as well as in yield. The following state- 

 ment by Lawes and Gilbert was published in 1900: 



"The total number of species that have been observed on the plots is 89, com- 

 prised in 63 genera, and 22 orders; whilst, to take some of the more important 



