172 SYSTEMATIZED PLANT-BREEDING 



most uniform conditions possible. Otherwise the 

 favoured plants would undoubtedly be chosen as 

 starting points of the new varieties. Fortunately 

 this is easily effected by a system of uniform 

 spacing between the plants. As the germination 

 of the grain is usually from 95 to 100 per cent, 

 blanks, and consequently favoured plants, do not 

 occur frequently. 



The ordinary method of testing the yield per 

 acre is to grow plots of an acre or so alongside 

 similar plots of some standard variety. The actual 

 yields of grain and straw are then compared 

 directly. The method is not satisfactory, for the 

 margin of error even under good experimental 

 conditions may amount to five per cent., and 

 where a real difference of -two or three per cent, 

 is too valuable to be missed it is folly to attempt 

 to use it even if the expense of testing large num- 

 bers of varieties in this way could be faced. The 

 method now in use consists of growing small 

 plots under the most uniform soil conditions 

 obtainable, each variety under trial being repre- 

 sented by a minimum of twenty plots. These are 

 four feet square with the rows six inches apart 

 and the plants two inches apart in the rows. At 

 harvest the central square yard of each plot is 

 gathered plant by plant and notes made of the 

 number of plants surviving and the number of 

 ears they bear. These figures give the number of 

 tillers per plant a feature of great importance 



