A COMPARISON OF MAIZE-BREEDING METHODS : i) 
At harvest the borders were divided transversely into eight sec- 
tions of equal length, and the number of plants and the weight of 
good ears in each row of each section were recorded. Omitting the 
outside rows, there are in each border three rows of one kind of seed, 
occupying the second, fourth, and sixth rows of the border, compared 
with two rows of another kind, occupying the third and fifth rows. 
A series of differences for each of the sections was determined as fol- 
lows: (1) The difference between row 3 and the mean of the two ad- 
jacent rows 2 and 4 and (2) the difference between row 5 and the 
mean of the two adjacent rows 4 and 6. There were thus 16 sepa- 
rate comparisons in each border. The 16 differences were treated as 
an array and the mean with its probable error determined. LBor- 
ders 3 and 4 and borders 5 and 6 were duplicates, so that in these two 
instances there were 32 differences to be averaged. By treating dif- 
1924 1923 1922 1921 1920 1919 1918 1917 
7-F 
FF. 
227, JE-#. 
SA7CE 
SOLE, 
PIIZ 9-/ 
fe SOLF, 
#YO 1-7 & 
PLDI 
LA 
P/E 
E27 
FXF 
IY AY 7 
I-E 
PILP CY 
a7, 
/ Vie SSF 
W418 
Pll S18 
SL, 
Fic. 3.— Representative pedigree of progenies in the crossbred seed stock of maize tested in 1924. Numbers 
in the pedigree indicate individual plants. The numbers in 1917 are those assigned to the original ears, 
first planted in 1918. F=16.8 
ferences between adjacent sections instead of absolute yields the dis- 
turbing effect of soil diversity is very largely eliminated. By this 
system the middle row of the border figures in the comparison with 
the rows on either side and is thus given a double weight. There 
was no evidence that the middle row yielded either more or less than 
rows 2 and 4, and the only error this practice would involve would 
bea very slight spurious reduction in the probable error. Themethod 
has the advantage of eliminating any error due to a trend in fertil- 
ity across the field. 
The distance between the outside rows of adjacent borders was 
slightly greater than that between the rows inside the borders, 
and there is a general belief that the outside rows receive a little 
less water than the other rows. For these reasons the two outside 
71 (31—26|——2 
