412 
Journal of Agricultural Research 
Vol. XI, No. 9 
winter wheat where there is apparently little competition between adja¬ 
cent rows than for the barley and oat tests. 
Table X gives the comparison for the coefficients of variability for 
yield of the central rows and of the 3-row plots. 
Table X .—Comparison of the difference between the coefficients of variability of central 
rows and j-row plats 
Cereal* 
Designation of test. 
Difference be¬ 
tween coefficient 
of variability of 
central rows and 
3-row plots. 
Difference 
prob¬ 
able error. 
Wheat (Minnesota 169). 
Do. 
Oats (Minnesota 281). 
Do. 
Plant Breeding Nursery.... 
Farm Crops variety test .... 
Plant Breeding Nursery.... 
Farm Crops variety test .... 
Plant Breeding Nursery.... 
Farm Crops variety test .... 
Plant Breeding Nursery.... 
Farm Crops variety test. 
2. 9 S±i- 535 
. 6 i±i. 215 
— . oidzi. 036 
- -33±i*4i2 
. $S± 2 . 596 
• 4 ®± .993 
1. 6$dii. 313 
2. 94 ±i. 699 
T -9 
• 5 
Barley (Minnesota 105). 
Do. 
Winter wheat (Minnesota 529). 
Do. 
. 2 
* 5 
r * 3 
7 
The results as reported for the comparison of central rows and 3-row 
plots (Table X) are in substantial agreement with the results for a 
comparison of central and border rows (Table IX). For the oats and 
barley the central rows are no more variable for yield than the 3-row 
plots. For these crops the border rows had a significantly greater co¬ 
efficient of variability for yield than was obtained for the central rows. 
In the winter- and spring-wheat tests there was little indication 
of competitive effect of adjacent rows, with the exception of the winter- 
wheat Farm Crops test, as determined by the coefficient of variability 
of the border and central rows. Table X shows that with winter and 
spring wheat the use of all three rows for each check plot in computing 
yield was somewhat more accurate than the use of central rows. The 
difference between the coefficient of variability for yield for the central 
rows and 3-row plots is 1.9, 0.5, 1.3, and 1.7 times the probable error. 
The probability of the occurrence of deviations as large as 1.7 and 1.9 
times the probable error are, respectively, 2.98 to 1 and 4.00 to 1. The 
slight reduction in error im winter and spring wheat due to the use of 
all three rows of the 3-row plots does not appear to be of sufficient size 
to warrant the work involved in harvesting, threshing, and computing 
the border-row yields. 
VALUE OF REPLICATIONS 
The two methods of overcoming soil variations of any particular 
season's test are the determination of the sort and size of plot to use 
and systematic plot replications. Plots consisting of 3-rod rows each 
were used for our test. Each row was then harvested separately and the 
yield per acre computed. The value of replications was determined 
