326 Bulletin 1Y9. 



either nitrogen or superphosphate. Potash, when used with nitrate 

 of soda, gave better i-esults than wlien used with superphosphate. 



The results of tliese experiments would seem to indicate that a 

 complete fertilizer would give the best results when used on tliis 

 field, and that the greater portion of it should be potash with only 

 a moderate amount of nitrogen and but little superphosphate. 



In 1S97 Mr. Ijonnell experimented with potatoes on another part 

 of the farm ; the results indicated that potash gave rather the best 

 results. In 1898 oats were grown on this piece of ground and the 

 superphosphate plats gave the l)est jields. Again in 1899, two years 

 after the fertilizers had been applied, wheat was grown on it. The 

 plats that had received superphosphate in 1897 still gave the largest 

 crops. Tliese results indicated one of two conditions: either that 

 the cereal plants, oats and wheat, could not find enough phosphoric 

 acid in that soil, unless supplied in the fertilizers; or that the cal- 

 cium sulphate (gypsum or land-plaster), of which all superplios- 

 j^hates are largely composed, gradually made available some of the 

 tightly locked potash that existed in the soil, and that it was this 

 liberated potash and not i\\Q phosphoric acid that gave such marked 

 results the second and third year following the application of the 

 fertilizers. 



It was formei-ly a common ])ractice to use calcium sulphate (plas- 

 ter) upoji land for the purpose of making available some of the 

 tightly locked plant-food, especially j>o^(i.sA. 



Experiments of Mr. A. 0. Stewart, Mariposa, N. Y.— Mr. 

 Stewart has experimented for the past three years, in 1897 and 1898 

 on potatoes and in 1899 on corn for the silo. On Sept. 21, '99, one 

 s(piai-e rod of each plat was cut, shocked and photographed. (See 

 cuts, next page.) Tiien each shock was weighed and the yield per 

 acre estimated. Also after a week of warm weather the remaining 

 crops on each plat were cut and weighed in order to determine 

 whether the estimated yield per acre would vary much whether 

 based on the yield of one srpiare rod, or of eight square rods. In 

 general, the smaller the area taken for estimating the crop per acre, 

 the greater the probable en-ors in the calculation. These results 

 are tabulated on page 328. 



