Nov. 26,1917 
Field Technic in Rod Row Tests 
417 
and 3-row plots seem significantly better than three, although there is 
some doubt about the accuracy of this conclusion when one notes that 
four replications give a lower variability than six. Twelve replications 
as compared with three show a difference of 3.71 ±1.041. 
SUMMARY OF RESULTS 
(1) In a study of competition between rod rows of small grains grown 
1 foot apart there was some effect on the yield of border rows of the same 
variety due to height of adjacent rows of barley, winter wheat, and in 
one of two tests an indication of such effect in oats. The results were 
variable in different plots, such variation being due possibly to the envi¬ 
ronmental conditions. There was no apparent effect of height of adjacent 
rows on the yield of border rows of the same variety in spring wheat. 
(2) The yield of adjacent rows appeared to be of some importance in 
barley tests and in the Farm Crops spring-wheat tests. Correlations 
obtained for other tests indicated considerable soil heterogeneity. 
(3) The effects of the height of adjacent rows for the barley Plant- 
Breeding tests was unquestionable. The effects of the height of adjacent 
rows were sufficient to often cause differences of 4 or 5 bushels per acre 
in the yield of border rows of the same variety of barley. 
(4) The comparison of yield variability of border and central rows of 
check plots of barley, oats, spring and winter wheat was further evi¬ 
dence of the competitive effect of rod rows of small grains when grown 
1 foot apart. In nearly all tests the border rows proved to be more vari¬ 
able in yield than the central rows. 
(5) The check plots of oats, spring and winter wheat, and barley were 
used for a study of the value of replications in reducing error. In nearly 
all tests three replications as compared with a single plot reduced error 
by 25 to 50 per cent. 
(a) In the spring-wheat Farm Crops tests for central rows four 
replications gave about the same coefficient of variability as nine 
replications for the Plant Breeding test. In the Farm Crops series 
six replications reduced the error by over 50 per cent as compared 
with three replications. The lack of agreement between the Plant 
Breeding and Farm Crops test may be explained by the serious 
black-stemrust epidemic which attacked the Plant Breeding Nursery. 
(b) For the Farm Crops oat test six replications reduced the error 
by about 25 per cent as compared with three replications. For the 
Plant Breeding tests three replications were about as valuable as 
any higher number below nine. 
(c) For the barley tests three replications seemed to be about as 
accurate as any number below six. 
(d) For the winter-wheat results three or four replications reduced 
the error about 50 per cent and appeared to be about as accurate as 
any number below 12. 
