286 BOARD OF AGRICULTURE. 



ly accurate result can be obtained from any experiment ; that it is 

 only by comparison — only relatively, in their relation one to 

 another — that they become of value. Crops may differ largely in 

 weight, as to the times in which they are taken. It has been 

 shown that after they have been considered dry, there is a differ- 

 ence in weight. This fact, that they do differ in weight at differ- 

 ent times, does not affect the experiment, because if they were 

 weighed green, just as they were cut and taken into the barn, it 

 shows the relation of these plats to one another ; I think this 

 would be found to be true in almost every case. Every part of 

 these experiments is attended with some error, and we wish to 

 eliminate the error as far as possible, and in weighing of plants in 

 that way — all of the plats at the same time, or as soon as may 

 be — we get a result that is almost correct — not absolutely cor- 

 rect, for no result we obtain is absolutely correct, and correct 

 results can only be obtained by carrying on these experiments for 

 a great number of years. The system of experimenting, as it has 

 been carried on, has been a failure, inasmuch as they have been 

 dropped after they have served a certain purpose ; they have not 

 been carried on persistently year after year, thereby developing a 

 principle in agriculture, and not merely satisfying the curiosity of 

 some person who is experimenting. 



Now, in our experimenting at the Agricultural College in Penn- 

 sylvania, we have tried to avoid what the gentleman has clearly 

 shown is an error we are apt to fall into, and become confused — 

 that is, attempting two systems of culture, or putting in the same 

 thing for two objects. It cannot be done. We failed on several 

 plats for that very reason. The earlier experimenting was about 

 a failure, just because we attempted to do too many things on the 

 same grounds, and had more objects than one. There must be 

 confined to each plat but a single object. If you have more, 3 r ou 

 lose control of the experiment, and afterward you have to experi- 

 ment again to find out which of the two it is. 



In regard to the size of these plats, the plan of having them 

 very small is probably more liable to error than having them large, 

 although there are difficulties on the other hand ; if your experi- 

 ments are extended over a great deal of ground it is almost impos- 

 sible to have them all carried along under the same conditions* 

 but you can overcome that by going to extra expense. We try 

 in all experiments to do this in one day, and in a portion of the 

 day after the dew had gone from the earth ; we try to get it as 



