Agricultural Experiment. 



457 



AGRICULTURAL EXPERIMENT. 



By E. Parish, li.Sc 



Vice-Principal, School of Agriculture, 

 Glen. 



Earmers are not ordinarily much concerned with the iechnique of 

 agricultural experiment — the results only affect them. Mainly for 

 this reason u recently prepared scientific bulletin, entitled " Ag^ricul- 

 iural l<]xj)erintent, its Design and Interpretation," is not heing" 

 l)uhlislicd in the Journal, but is being printed as a separate biilh'tni. 

 It is well, however, thai farmers should realize the difficulties 

 atteuch'iii on agricultuj'al exiierinient, of the care necessaiy in the 

 design and execution of it, and in iiiterpretating- the results obtained. 



()]i ihe face of it it a})pears a simple thing to test the effect of 

 the application of fertilizer to a ci'oj). Mnormous variation from 

 the noi'mal or true result, however, ordinarily occurs in the figures 

 obtained from the plots under test and unless the officer in charg-e 

 of the experiment is aware of the extent of this probable variation, 

 he is very liable to be misled. Most farmers during their life-time 

 get yields from crops either lower or higher than their expectation, 

 and for which they are unable to account, but few are aware liow 

 frequent this variation from the normal is, or how large it may be. 

 In field trials in adjoining plots treated similarly in all respects 

 the yield of one may be twice that of the other, and for no reason 

 that can be discovered. 



Eor example, in eight one-acre plots of wheat arranged on the 

 following plan at an experimental station in South Africa, with all 

 plots treated alike in every respect the yield in lb. of combined grain 

 and straw was as follows: — 



Similarly with experiments with animals, the individual varia- 

 tion is so great as to necessitate repeated trials with comparatively 

 large numbers of animals before any reliable conclusion can be 

 drawn. Often under identical conditions, as far as such is humanly 

 possible, one animal will give twice the gain in live weight of its 



