May 11, 1900.] 



SGIENC:2. 



729 



be a stock of eggplaut or tomato plants far 

 in excess of the number to be set, that the 

 selection may be made in such a waj' as to 

 give a uniform stand, fulfilling the require- 

 ments of the experiment. Such plants 

 should be raised from the seed by, or under 

 the eye, of the experimenter that he may 

 feel sure of the kind and character of his 

 plants. One would scarcely trust to ordi- 

 nary commercial plants for use in a plot 

 experiment. The seed might be mixed at 

 the outset and an over stimulus might make 

 the plants unusually tender. 



In cultivation and all those processes 

 which find a place under the term ' care of 

 crop' one cannot be too serious. By a little 

 carelessness here ah element may be in- 

 troduced that will spoil all the previous 

 work. It is as essential, for example, that 

 the whole plot be treated alike in the culti- 

 vation as in the sowing or planting. It is 

 a good rule to never cultivate one row or 

 plant unless the watch, and the clouds as 

 well, indicate that there will be time that 

 day for the same treatment of all in the ex- 

 periment. If a half of a plot is hoed one 

 day and the job finished the next, there are 

 substantial reasons for inferring that the 

 harvest of results will be scattered and 

 scarcely worth the gathering together. 

 There is a chance in all this work of being 

 faithless to the ideals of. the genuine experi- 

 menter who is constantly striving to elimi- 

 nate all but the one point of variation in the 

 test. For example, the wheel cultivator 

 may be used twice in the row for a half of 

 the plot and only once for the remaining 

 portion and, while saving time it is robbing 

 the experiment of its full value. A person 

 who could permit such a thing is entirely 

 unfit to occupy, much less fill, a position in 

 a center of research, which we call in this 

 country an Experiment Station. Far better 

 to clean up the paths in the late afternoon 

 than to make a beginning only upon a plot 

 with hoe or cultivator. 



It seems almost unnecessary to add any 

 remark here concerning the necessity of 

 keeping experiment grounds clean of weeds 

 until the time is past when they can inter- 

 fere with the results. Each plant in the 

 crop should have its full soil and air space, 

 and if weeds are permitted to contend for 

 these essentials, the test, whether it be of 

 fertilizer or distance or depth or any other 

 thing, fails. Pigweeds and purslain and 

 the whole list of weeds must be kept out of 

 a plot experiment, and all results should 

 read between the lines — these are with 

 clean culture always understood. This is 

 the cheapest as well as the only way. 



There is a watchfulness demanded by the 

 superintendent of field experiments that 

 savors of the wisdom of the owl. He must 

 also love truth beyond all else. Pardon 

 me if I mention one of the elements in plot 

 experiments that has been a great source 

 of annoyance. It is the end plants and the 

 outside rows. There ought not to be any, 

 but it is not easy to get rid of them. 



Their terrors came to me very forcibly 

 last year in the old plot of eggplants, where 

 the end plants, that is the outside row, all 

 around the plot, was vastly better than those 

 within, and demanded a separate record ; 

 but it was not made. With beans, an ex- 

 tra row has sometimes been planted upon 

 the outside and rejected in the final results. 



This pi'oblem of the end of side row only 

 emphasizes what has gone before, namely 

 the importance of placing each and every 

 plant under precisely the same conditions 

 as all the others. This is not done in case 

 of the end plants and the results may be 

 considered accordingly. 



Troubles not to he Avoided. — There are 

 dangers which all crop plants run. Stray 

 animals may break in and destroy. Pigeons 

 may get the first sowing of peas before 

 germination. Moles may burrow under a 

 plot and introduce a source of error. The 

 hole thus made will turn the flow of an ir- 



