116 BREEDING CROP PLANTS 



of selections of like nature together, the use of single rows 

 and more replications, may be desirable under certain condi- 

 tions. 



6. Computation of a probable error for the method of test. 

 The use of this probable error as a means of determining signifi- 

 cant differences. 



7. Increase of the better selections and either a trial by careful 

 farmers or a further test in field variety plots followed by distri- 

 bution of the better strains. If placed in field variety plots, 

 borders should be removed and each variety tested in repli- 

 cated plots. Probable errors should be obtained and used as in 

 the row trials. 



From five to eight years' time is needed before the new 

 selection is introduced to the farmer. 



Crossing. The improvement of commercial varieties of self 

 fertilized small grains by the head or plant method of selection 

 is a very easy process, although several years are required to do 

 the work. The production of new forms by crossing is not so 

 simple. A standard plan of attack has been developed which 

 is the application of the Mendelian method. 



The first step is the initial cross. Promiscuous crossing is not 

 advised, but each cross should be the result of a determination 

 of parents which most nearly approach the ideals in mind. By 

 recombination of characters there is the possibility of obtaining 

 a sort which is more desirable. 



The FI generation is grown so that each plant has space for 

 individual development. A knowledge of the inheritance of 

 characters allows those plants which are not crosses to be elimin- 

 ated in FI. The F 2 generation plots should be as large as can be 

 studied and each plant grown with enough free space that it may 

 be examined. Numerous selections of plants which have de- 

 sirable field and seed characters should be made. Each of these 

 Fz plants selected should be grown in an individual progeny plot in 

 F 3 and individual plant notes taken. Selection of desirable 

 plants should be continued in later generations. When plots 

 show apparently uniform progeny of a desirable sort, the strain 

 should be included in the rod-row tests and compared with 

 standard varieties. 



Knowledge of the results of continued self-fertilization in 

 generations following a cross, shows the reliability of another 

 method which was first used at Svalof, Sweden (Babcock and 



