New York Agricultural Experiment Station. 249 



longer than unsprayed ones. In the present case this advantage 

 of prolonged growth was lost through the killing of the plants by 

 an early frost. It appears that the increase obtained was due to 

 the partial control of tip-burn, the better control of " bugs " (in a 

 few cases) and the imperfectly-understood stimulation effect of 

 the bordeaux. 



In experimental work of this kind the experimental error is cer- 

 tainly large for individual experiments, though for an average of 

 forty-seven experiments it is probably small. That is to say, we 

 believe that the average increase from thorough spraying was, 

 actually, between 17 and 18 bushels per acre as shown by the results 

 of the experiments, notwithstanding the fact that in certain of the 

 experiments there was an unaccountable reduction in yield. In 

 five of the experiments the row very thoroughly sprayed by the 

 Station yielded less than the unsprayed check row adjacent and in 

 three other experiments (in sprayed fields) the extra spraying done 

 by the Station apparently decreased the yield. 



There is no reason, whatever, for believing that spraying was 

 harmful in these eight experiments. Undoubtedly, the true explana- 

 tion is that the check row possessed some advantage over the 

 Station row and would have outyielded it still more if neither row 

 had been sprayed. In one case (J. Darrohn's experiment) an 

 explanation was found in the fact that the check row contained 21 

 more hills than the Station row; but what was the cause of the erratic 

 results in the other seven experiments is unknown. 



If the reduction in yield in these eight experiments is ascribed to 

 original inequalities between the test rows it must be admitted that 

 similar inequalities existed in some of the experiments showing 

 increased yield from spraying. It is doubtless true that, in some of 

 the experiments, the increase in yield apparently due to spraying 

 was, in reality, partly due to other causes. However, by averaging 

 the results of a large number of experiments the probability of error 

 is greatly diminished. 



The average gain from the spraying done by the Station was nearly 

 as large in sprayed fields as in unsprayed ones, being 15.04 bushels 

 per acre in the former and 17.76 bushels per acre in the latter. If 

 these figures are reliable, they indicate that the spraying done by 

 the owners was of small value; but it should be considered that the 

 number of experiments in sprayed fields was rather small — only 



