CONDITIONS OF AN EXPERIMENT. 69 



he found almost all sets of experiments worthless except those 

 of Lawes and Gilbert in England. Of course he did not mean 

 to say that no accurate experiments have been made ; but that 

 they have been so intermingled with defective experiments, and 

 vitiated by conditions not reported, that they are almost worth- 

 less as data for general conclusions. 



As an illustration of the extreme difficulty of eliminating 

 disturbing causes so as to reach sound conclusions, I give a few 

 of his experiments and their results as found in the Report of 

 Michigan Board of Agriculture for 1868. 



Two acres of land were selected, with soil of friable loam 

 " of apparently uniform character." From that plat ten small 

 plats were taken, so as to represent fairly every portion of the 

 two acres. All the work except the ploughing on all the pieces 

 was done on the same day. In other words, these ten plats of 

 ground, to all appearance alike in the beginning, were cultivated 

 and treated exactly alike. This uniformity in soil and culture 

 might have been expected to give uniform results at harvest. 

 But we find among these plats a very great diversity of yield. 

 One yielded, of shelled corn, at the rate of 51.28 bushels to the 

 acre. Another 76.14 bushels, an amount more than one-third 

 greater than the former. Of stalks, one yielded 1.28 tons to 

 the acre ; another 2.40 ton, or nearly twice as much as the 

 other. On the other plats of the two acres which were manured 

 exactly alike and treated in the same manner, there was a like 

 variation in the crops. 



Now when we have conditions that apparently ought to give 

 the same results, and which in ordinary experimenting would 

 be assumed as giving the same results — when we find such con- 

 ditions giving us a variation of more than 50 per cent. — the 

 difference between fortune and bankruptcy — we see what care- 

 ful experiments we must have to secure the conditions of ex- 

 periments that can be relied upon in practical agriculture. 



Another field was selected and divided into twenty-four plats 

 (two by four rods each). With a single unimportant exception, 

 he says, " it ivould he difficult to find a piece of ground presenting 

 a greater uniformity in the appearance of its subdivisions. ''"' The 

 plats were all treated in the same way, the same amount of 

 work was spent on each, the corn was all cut the same day, 

 husked the same day, and weighed the same day. One plat 



