SHAPE AXD SIZE OE PLOTS IX TESTS 01' VARIETIES OF GRAIN. 51 



Table 3 shows the percentage of the total area contained in 

 a strip six inches wide within the border of plot rmits commonly 

 used in plant breeding tests. It will be noted that the part of a 

 square plot lying in this marginal strip ranges from 0.96 per- 

 cent in an acre area to 38.26 per cent in a 2000th acre area. In 

 a plot with a length two times its width this marginal area 

 amounts to i.oi per cent in an acre and 40.86 per cent in a 

 2000th acre area. The percent of the total area lying in the 

 strip six inches wide within the border of a long narrow plot 

 is much greater than that of a square plot of the same area. In 

 the cases considered the' marginal area may be 6 percent to 115 

 percent greater in long plots than in square plots of equal- size. 

 The magnitude of this percentage depends on the relation of the 

 length to the breadth of the plot. 



IMercer and Hall * maintain that there is practically no differ- 

 ence in the accuracy of square plots and long narrow plots. How- 

 ever, it should be borne in mind that the basis of the conclusions 

 of these authors was the yield of an acre plot selected out of a 

 large field. This acre was divided into small units each of which 

 was harvested separately. There were no pathways around the 

 acre plot and likewise, and of much greater importance, none 

 around the small units of the acre. Hence their conclusions 

 are based on conditions where the competition among plants is 

 that ordinarily found in an}^ field. These authors do not con- 

 sider the possible effect of pathways around plots. Neverthe- 

 less the marginal plants play an important part in making up 

 the total yield of plots. 



To overcome the influence of pathways it has often been sug- 

 gested as advisable to discard the outside rows of a plot in order 

 to attain a fairer test of the cropping ability of a variety of 

 grain under so-called normal conditions. However, to throw 

 away the outside rows of each plot is exceedingly bad practise 

 iTi plant breeding work because of the possibility of such odds 

 and ends becoming mixed up with other varieties. Further the 

 d'iscarding of outside rows requires a greater labor expense and 

 also greater land area in attaining results on unit plots. Trim- 

 ming plots to overcome certain effects of environment is ex- 



*Mercer, S. W. and Hall, A. D. The Experimental Error of Field 

 Trials. The Journal of Agricultural Science, Oct. 191 1. 



