Experimental Error in Crop Tests 
75 
The proper conduct of exparimental work in crop produc- 
tion in light of our present knowledge requires either a large 
extension in land area and labor facilities or else a marked 
restriction in the amount of investigation carried on. 
MEASURING IMPROVEMENT IN YIELD THRU BREEDING 
Comparing the yield of corn for one period of years with 
the yield of another period is an unreliable method for not- 
ing improvement thru corn breeding. An illustration of this 
method is found in a circular of the United States Depart- 
ment of Agriculture Office of Corn Investigations, August 20, 
1914. The data in Table 34 were given in this circular as 
Table 34 — Daia given in Circular of Office of Corn Investiga- 
tions, U. S. Department of Agriculture, August 20, 191A, 
to show improvement from ear-to-row breeding conducted 
at Piketon, Pike County, Ohio 
Average for 
first seven 
years, 1901- 
1907 inclusive 
Average for 
second seven 
years, 1907- 
1913 inclusive 
Ratio first 
period to 
second period 
Yield per acre as weighed in the fall 
(70 lbs. of ears to the bushel) .... 
Yield per acre of dry shelled grain 
(56 lb?, to the bushel) 
Bushels 
77 
63 
Bushels 
85 
75 
100:110.4 
100:119 
indicating 19 per cent increase in yield of dry shelled corn 
per acre by ear-to-row breeding. The increase in yield of 
ear corn as weighed at husking time was 10.4 per cent. The 
measure of improvement by breeding was the average in- 
creased yield during a seven year period, 1907-1913, over the 
previous seven-year period. 
A comparison of the yields in Table 35 during these same 
two periods for the state of Ohio as compiled from the United 
States Yearbook indicates a similar increase in yield for the 
state in general. During the last period of seven years, the 
Ohio state yield was 11.4 per cent higher than during the 
previous seven years. Likewise data compiled from the re- 
ports of the Ohio State Secretary of Agriculture, indicate 9.4 
per cent greater yield for Pike County, in which the experi- 
ments were conducted, during the last seven years than dur- 
ing the previous seven years. This suggests that more favor- 
