302 BOARD OF AGRICULTURE. 



control what we can, and thon must compare observations for a 

 long- time, to get at the probable elements of error. Where we 

 are drawing deductions as to the results of experiments, we must 

 make allowances for the elements < f error which underlie all our 

 experiments. The distinction between the two lines of inquiry is 

 very clear. In art we want to get at certain rules of practice. 

 An experiment that would be satisfactory, so far as that is con- 

 cerned, would amount to nothing, so far as increasing our scientific 

 knowledge is concerned. Scientific investigations must be made 

 with more care; one great reason why such a change has taken 

 place in our agricultural chemistry has arisen from the fact that For- 

 mer experimenters in chemistry did not control all the conditions; 

 they had not apparatus sufficiently delicate to detect all the slight 

 changes which took place. Just as long as we work in this direc- 

 tion we are going to meet with disappointment. 



In regard to the quality of corn in Maine and St. Louis : I do 

 not know there is such a difference in the feeding quality of this 

 corn. 1 am not aware that the matter has been tested experi- 

 mentally. I am aware there is a difference under certain circum- 

 stances ; an analysis will show a difference in its composition, but 

 it does not follow there is a difference in the feeding quality of the 

 grain. We have been running along for a number of years with 

 the theory that the composition of the grain was an index of its 

 nutritive value. No one now will pretend to advance that doc- 

 trine who has examined the latest researches in physiology and 

 chemistry. But, admitting there is a difference in the corn of 

 Maine and St. Louis, what we want to get at is this : the result 

 of feeding in Louisiana or Missouri under the same conditions pre- 

 cisely, and the same care taken to secure accuracy. If the corn 

 from Maine was taken to St. Louis, there would be a difference. 

 There is an element of error underlying that we can not get at. 

 We may not reach it in our life-time, but we must determine it 

 before we can get at principles that are safe for us to follow in 

 practice. If experiments must agree exactly in order to be of 

 value, we may as well stop experimenting, for you will never get 

 any two experiments to agree precisely, because you can not con- 

 trol all the conditions. But we can get at the general principle 

 after elminating the error. For instance, 1 wish to know what 

 the effect or value of a certain commercial fertilizer is. Should I 

 apply it to one plat, I should have to examine a large number of 

 unmanured plats and find their variation, and then I may compare 



