125 



showed results "almost wholly nejjative," the field apparently contain- 

 ing all the fertilizing elements necessary for a full crop. The unfer- 

 tilized plats yielded from 36.7 to 50.5 bushels of grain per acre. The 

 yield with stable manure was only 33.1 bushels. 



Of two CO operative farm tests with wheat, one was vitiated by rust, 

 which practically destroyed the crop, the other served to indicate that 

 nitrate of soda and superphosphate were of some benefit. In both 

 cases nitrate was applied in the fall, " probably a mistake," as careful 

 investigators believe its liability to be washed away will be lessened 

 by delaying the application till April or even May. 



At the request of the station a few farmers in different parts of the 

 State made the experiment of applying in the spring of 1889 a small 

 amount of nitrate of soda to wheat, to which boue meal or a similar 

 phosphatic fertilizer had been applied in the fall, the object being to 

 find whether nitrate may be used profitably as a fertilizer on wheat 

 which has received a fall dressing of fertilizer rich in phosphoric acid 

 and poor in nitrogen. The plan was to leave one or two strips without 

 fertilizer in the fall, apply barn-yard manure to one strip and boue 

 meal to two, one of which latter should have in addition an applica- 

 tion of nitrate of soda and muriate of potash in the spring at the rate 

 of 160 pounds each per acre. In three tests iu which the plan was 

 fully carried out, there was a decided increase of crop from the spring 

 application of nitrate and potash. Directious are given for conducting 

 the experiments on a somewhat different plan, so as to determine the 

 proportion of the two substances needed by the plant. 



Experiments on plants grotvn in boxes (pp. 56-62). — Inequalities of soil 

 constitute a serious difficulty in field experiments. One means of meet- 

 ing this difficulty is by trials on a small scale, in boxes or otherwise. 



Whoever has carefully looked over the tables giveu in the bulletin must be struck 

 with the great diversity iu natural productiveness of soils selected because of their 

 apparent uniformity, as shown by the differences in yield of the unfertilized plats. 

 These differences, however, are not exceptional, but are quite within the limits of 

 variation which practical field experimenters have learned to expect in such work. 



By systenuitic and continuous cultivation of a series of plats, under uniform treat- 

 ment, such ine(iualities as have arisen from temporary causes will gradually disap- 

 pear, and in the course of years such results may be arrived at as those shown in the 

 experiments of Messrs. Law^s and Gilbert at Rothamsted, England, where it is evi- 

 dent that all minor inequalities in natural fertility have been obliterated iu the grand 

 aggregate results of a half century of persistent work. 



But we can not afford to wait half a century for results. Something must be definitehj 

 indicated from iiear to year, if field experimentation is to hold a permanent place as a 

 pruclicable method of research. 



By repeating the same experiment on various soils, after some such method as that 

 explained on previous pages of this bulletin, some of the errors indicated may be in 

 a measure corrected. But this method also leaves much to be desired. It does not 

 remove the source of error, but is only useful because the inequalities of one soil are 

 likely to be offset by different inequalities in another soil, and thus one error is made 

 to eountera(;t another. 



The only method by which a soil oi' absolute uniformity can be secured is to deal 



