New York Agricultural Experiment Station. 
61 
Grouping the ten varieties that passed the greatest percentage of 
plants safely through the winter, no apparent relation is to be seen 
between the per cent of plants which survived the winter and the yield 
of grain. Carrying this comparison further, no relation seems to exist 
between the yield, number and per cent of plants which survived the 
winter, as is seen below. 
NAME OF VARIETY. 
Rank in 
produc- 
tiveness. 
No. plants 
survived 
winter. 
Per cent 
plants 
survived 
winter. 
Yield. 
White Blue Stem 
530 
59.21 
5 lbs. 7 ozs. 
Velvet Chaff 
\ 
541 
64.79 
5 " 6 " 
Swamp 
3 
547 
55.92 
5 " 5 M 
Diehl-Mediterranean 
4 
G44 
63.53 
5 " 2 " 
Scott (bearded) 
5 
495 
51.40 
4 " 12 " 
California Blue Stem 
6 
639 
66.08 
4 " 11 " 
York White Chaff 
7 
396 
51.29 
4 " 5 " 
8 
416 
57.22 
4 " 3 " 
AVhite Eldorado 
9 
320 
51.44 
3 " 7 " 
10 
451 
55.05 
3 " " 
In order to determine, however, the value of duplicate results in 
testing one variety after this manner, twenty rows of Clawson wheat 
were planted in a manner indentical with the varieties. The soil in 
which these twenty rows were placed, was very even in its character, 
so that as far as external features went, the ground seemed very 
suitable. In the twenty rows were planted 21,681 seeds, an average 
of 1,084 to each row, and of these 15,865 vegetated, or 793 on an 
average to each row, or 72.59 per cent per row. Of the total 
number of seeds that vegetated, 5,090 survived the winter, aver- 
aging 254 to the row, or 32.03 per cent per row. The yield of grain 
from the 5,090 plants was 53 lbs. 7 ozs., an average of 3 lbs. 9 ozs. 
per row; of straw, 127 lbs. 9 ozs., an average of 6 lbs. 7 ozs. per row. 
In this experiment of duplicates, the results are fully as contradic- 
tory as in a duplicate set of plat experiments, the extremes in yield 
ranging from 1 lb. 11 ozs. in one row to 4 lbs. 5 ozs. of grain in 
another. A like variation exists in both seed vegetation and winter 
survival of plants. In the former case the extremes are 719 — 853; in 
the latter 165 — 372. 
Confirming results by means of duplication is the only method by 
which we can hope to secure reasonably correct returns, as thereby 
the limit of errors may be carefully jirawn. The case of this experi- 
ment simply offers additional evidence to prove the weakness of plat 
experimentation, and supplies data for explaining wherein this weak- 
ness lies. Until we shall be able to prove what factors are essential 
to correctly test the productiveness of varieties, and be able to con- 
trol or interpret the workings of these factors, we cannot hope to 
