342 Trials o* New Varieties of Cereals. [Jm»i 



trials as hitherto earned out, the probable errors* of to weight® 

 TLZ have often been greater than the observed differences. 

 IT ImI lv. the existence of these probable errors has 

 ! nerX b n ignored in published reports. Sufficient evidence 

 Tthe miiversality and extent of probable errors in field tiials 

 1 b^nd i/to publications of (1) Wood and Strain* 

 ra> Mercer and Hallt , (3) of the present write + It may at 

 least be partly due to ineffective methods employed in vaneq 

 S aXoJonsequent misleading reports of them, to pas , 

 that tore has been so little increase in the average yields oi 

 e a s in Great Britain; for according to agricultural statistic 

 Z lX Tylelds of grain in Great Britain are still only about 

 the a^ eia e j iemb e compares badly 



^e Si that of the horticulturists. The -^-^^ 

 in barley yields in Ireland is. however, an example of 

 directly due to systematic yield trials. 



i has sometimes been tacitly assumed that i a se of taal 

 of the same varieties are carried out at a number of Afferent 

 stations on different soils, the average results obtained give figures 

 wh h are more useful ton those obtained at any one station 

 If a rarely happens, one variety gives better results than all 

 he oth at a number of stations tore is no doubt a pr - 

 tbil y that it would have given comparatively good resm 

 under still other sets of conditions in the same season What 

 gene aliv happens is that the order of merit varies at different 

 S ^ in to same season and in different seasons at the sam 

 ton. For this reason the averaging of results obtained a .a 

 nu nber of stations, whilst useful, is at best of 

 example : _lt is not of much use to a grower m Norfolk to be 

 told tot a certain variety of wheat grown in. say, Shropshire .« 

 even at several other stations, has given comparative y high 

 yields if to external conditions at all the stations are different 



^""experiment in variety trials are of two orders 

 (1) Systematic and generally avoidable; (3) Casual, and not 

 generally avoidable. To the first class belong 



^^p^ete^Keid Trials. Jnl. of A,,: Science, Vol. 4. Part 

 ll > t%»£* Cereals for Increased Production. Jnl. Fanners' Olui, Nov., 



1920 



