104 The Rothamsted Barley Experiments. - 



(4) with superphosphate and sulphates of potash, soda, 

 and magnesia ; (5) with superphosphate and sulphate of 

 potash. Taking the average for the twenty years in each 

 case, the quantity of ammonia required to produce one 

 bushel increase of barley, and its proportion of straw, was, 

 on the three plots with superphosphate (plots 2, 4, and 5), 

 2-131b., 2'411b., and 2'lOlb. : on Plot 3, with salts of potash, 

 soda, and magnesia, without superphosphate, 3 \591b. ; and 

 on plot 1, without any mineral manure, 3'681b. Thus, taking 

 the mean of the three experiments with superphosphate, the 

 amount of ammonia required was rather under 2lb. ; but 

 with the mixed alkali-salts without superphosphate, and 

 without any mineral manure at all, it was between 3^1b. and 

 3|lb. That is to say, a given amount of ammonia-salts was 

 more than one and a half times as effective when there was a 

 liberal provision of mineral constituents, but especially of 

 phosphates, within the reach of the roots, than when there 

 was not. Assuming that, with otherwise favourable soil 

 conditions, and with an application of not more than 501b. of 

 ammonia per acre, an increase of one bushel of barley (521b.) 

 and its straw may, on the average of seasons, be obtained 

 for every 21b. to 2|lb. of ammonia applied, still the table 

 indicates that the amount may vary very greatly according to 

 the characters of the seasons. Thus, on Plot 2, with super- 

 phosphate, only about IJlb. was required in the favourable 

 seasons of 1863 and 1869 ; but in the bad seasons of 1853 and 

 1856, 5'361b. and 4'481b. respectively were necessary. 



Before briefly summarising the conclusions established 

 in this section, one or two minor points demand notice. 

 During the last eight years of the twenty, one or two series 

 of plots, which otherwise were receiving the same manure, 

 also received 4001b. of silicate of soda per acre per annum. 

 Almost every year the eye detected a marked effect from the 

 silicate on the plots where no superphosphate was used, but 

 comparatively little, if any, on the plots with superphos- 

 phate. So striking was this result, that the silicate was 



