A COMPARISON OF MAIZE-BREEDING METHODS ja 
both yield per section and yield per plant, but these differences are 
but three or four times the probable error. 
‘The yields of crossbred and F, were practically the same. The 
crossbred exceeded the F, slightly in yield per plant but occupied 
the reverse position in yield per acre. The F, and crib selected were 
not directly compared, but both were compared with the crossed 
block, and judged by this indirect comparison the crib selected ex- 
ceeded the F, in both yield per plant and yield per acre by an 
amount that is more than six times the probable error of the differ- 
ence. Arranged in a descending order the standing of the five seed 
stocks was as follows: F,, F; and crossed block, crib selected, Fy. 
A check on the direct comparison is possible in the cases where 
both seed stocks of an experiment are elsewhere compared with a 
third kind. The results are consistent within the limits of errors of 
sampling. 
CORRELATION OF NUMBER OF PLANTS WITH YIELD PER PLANT 
In comparing the yields of the various seed stocks, both the yield 
per row and the yield per plant have been considered. If the space 
afforded by missing 
plants is utilized by 1924 923 1922 i921 . 1920 1919 1918 
those remaining to in- Gach Ores ar nanan 
crease their yield, the ie < 
yield per plant would NN 
favor progenies with a rx Vee 
low germination and 
would vitiate conclu- . 
sions based on this ee 7 
measuremen t. On Fic. 4.—Representative pedigree of progenies in the F, seed stock of 
the other hand, if poor maize tested in 1924. The numbers indicate individual plants. 
Selfed generations are shown by double lines. F=87.5 
stands result from un- 
favorable soil conditions, the same factors that result in failure to 
germinate or the death of plants might be expected to reduce the 
yield of the surviving plants. If the former is the correct explana- 
tion, there should be a negative correlation between the number of 
plants and the yield per plant. In the latter case the correlation 
should be reversed, and the sections with the fewest plants should 
have the lowest yield per plant. 
_ In calculating the correlation coefficient of number of plants with 
yield per plant, differences in the yield of the various stocks were 
allowed for by using the departures of the sections from the mean of 
the row for both yield and number of plants. The result was a co- 
efficient of 0.063+0.037, not a significant correlation. From the 
probable error it would appear that no correlation greater than 0.1 
4s to be expected, and with a relation of this order the effect of the 
number of plants on the yield per plant would be negligible. 
MEASURES OF INBREEDING 
_ It is obvious that in both the crossed and the selfed breeding 
methods followed in the present experiments the portion of the 
foundation stock from which the seed has descended will become 
reduced as the selection proceeds. In the crossed experiment, re- 
