304 



STATE BOARD OF AGRICULTURE. 



Experience is a good teacher, and it seems that experimenters get their 

 education that way. Especially is this true when science has not offered 

 reliable methods. Thus the years of 1906 to 1913 were used in working 

 out a reliable method of variety testing at the Michigan Station. Until 

 1911 the plant-row series with frequent checks was the most reliable plan 

 that had been devised. In later tests of larger quantities, many ex- 

 perimenters would multiply the number of these rod rows scatter them 

 among the others and depend upon the average of the results obtained to 

 indicate the relative yield of varieties. This is known as a reduplication 

 series. The difficulty that the Michigan Experiment Station has found (in 

 this class of work) is that very small losses of grain in these plats upset 

 the results and that the amount of labor necessary made the testing of 

 large numbers of strains out of the question. 



Fiy;. 1. A portion of the 1914 oat variety series. Each variety is planted in a long 

 narrow strio. Each fifth Dlat (sti'ip) is planted from the checli or standard variety with 

 which all of the other varieties are compared. Each plat is four drills wide and about 400 

 feet long. Two such plats can be planted at one trip of an eleven hoe drill, planting four 

 hoes on either side and leaving three hoes vacant in the center for an alley. A thirty inch 

 alley is allowed between the drill widths. The length of the plats should be at least twenty 

 times the distance between the centers of the check plates, because of the soil variability, and 

 to permit the calculation of a check yield for each and every plat of the series by interpolation 

 between the yields of adjacent checks. This method (it will be noticed) combines the benefits 

 derived from two principles, viz., reduplication and checks. As each plat crosses a large 

 number of soil conditions, it is in reality a reduplication. The series are also duplicated an- 

 nually. 



Figure 1 illustrated the method of variety testing now in use. Each 

 plat is a long narrow strip and therefore crosses a great variety of 

 soil areas, it is as efficient as a reduplication series. It has the advantage, 

 because the elements are end to end and can be cut collectively at one trip 

 of the binder.^ The shocks belonging to a plat are set up in a row and 

 can easily be brought together at threshing time. The series are dupli- 

 cated each year. The duplicate series usually follow each other. 



After threshing the grain is recleaned in a fanning mill and weighed. 

 The results are recorded in pounds per plat. In order to reduce these 

 results to a uniform basis the corresponding yield of the standard 

 or check variety is also calculated for every plat of the series. This calcula- 



4The sum of the yields of the reduplication plats is found by means of the binder instead 

 of by means of the adding machine. 



