16 MASS. EXPERIMENT STATION BULLETIN 281 



is brought out by the figures of table 5. The highest percentage of protein was 

 in plot 5E which received lime, potassium, and phosphorus. This plot also had 

 the most white clover. But, largely on account of the greater yield, plot 4E, 

 which received lime, potassium, phosphorus, and nitrogen, furnished the most 

 pounds of protein per acre. 



Experiment B. — This experiment was laid out in the fall of 1921 and the 

 spring of 1922. As shown in table 6, each original treatment was replicated 

 three times (six times not considering spring and fall applications), and to that 

 extent the experiment was better planned than was experiment A. After the 

 first application of fertilizer, all subsequent applications were made in the spring. 



During the first year of this experiment only slight effects chargeable to 

 fertilizers were observed. In the spring of 1923, a white-clover year, the effects 

 of the fertilizers were marked. After two or three weeks of grazing the cattle 

 were excluded from the field by means of a temporary fence. The effects of the 

 fertilizer were measured first by giving each plot a rank according to the propor- 

 tion of white clover and grass present, the best being ranked at 100 and the 

 poorest at 0. There were five independent sets of observations by diiferent per- 

 sons. Means of these estimates are given in table 7. According to this evidence, 

 there was a striking response first to lime, second to potassium, and third to the 

 combination of lime, potassium and phosphorus. Nitrogen added nothing to the 

 appearance of the plots which received this element in addition to minerals. 

 Later in the same season square-yard areas were harvested from certain plots, 

 dried, weighed and analyzed. The fence was then opened and the cattle given 

 access to the pasture. The data are given in table 8. Unfortunately, yields 

 were not taken from the plots which received nitrogen, and consequently no 

 quantitative comparison between minerals alone and minerals plus nitrogen can 

 be made for the first period of this experiment. Part of the data of table 8 were 

 reported in 1926 (7), but for the sake of completeness these as w-ell as additional 

 data are given here. The high percentage of protein which resulted from the use 

 of potash and lime was no doubt due to the effect of potassium on the propor- 

 tion of clover. However, potassium in combination with phosphorus and lime 

 was more effective than any other comliination or single treatment. 



Experiment B was continued under pasture conditions with fertilizer treat- 

 ments as shown in table 6 until 1929, when an attempt was made to secure addi- 

 tional quantitative data. It happened that the summer of 1929 was extremely 

 dry, with a total rainfall of 2.24 inches for July and August against a normal of 

 8.57. Yield data were obtained, but since they were doubtless seriously affected 

 by the abnormal weather conditions, they are not reported here. However, 

 since the character of the vegetation was fairly well determined before the 

 effects of the drought became serious, it is thought that the data on botanical 

 composition fairly show the effects of fertilizer treatment up to and including 

 the spring of 1929. These data are summarized in table 9. They show, first, 

 the advantage of IItih" in combination with other materials, except for the nitro- 

 gen combination, in inducing white clover to come in; second, the superiority of 

 treatments containing potassium for encouraging clover; third, th(> effect of 

 gypsum, both alone and with lime, in the encouragement of clover; and finally, 

 a pronounced j>ositivo effect of nitrogen on the presence of white clover. 



In table 10 are given data on yields obtained from plots of Experiment B in 

 1930. Clippings were taken from 3 areas of 3 square feet each in each plot of 1 

 squar(> rod. A groujjing of the figures according to series luiving a total of 4, 5, 



