16 BULLETIN 1321, U. S. DEPARTMENT OF AGRICULTURE 
the varieties. Similarly, the relation of yield to butt circumference 
and weight of cob is positive and to tip circumference is negative in 
three of the four varieties. 
TasBLe 10.—Direction of the relations between yield and seven characters of seed ears, 
as determined from the multiple regression equations for yield on these characters 
Positive | Negative 
Variety relations | relations ~ 
Oty Nissin 2s a ee ae ES OS SR ge pee he ee BE ea ee ee pe ee BCE F\| AID 
GENE REG eae er ee a RAIMI | ae eR RCN rend ae LRA GND) AHIEF 
st? NOUTQ0Ss aie see bad Le ee ed a ee EEE eS Gee oy ye RES Pe eee BCDF\| HIE 
Gel IN O2133 eos Sa cee RE 0g ee ae ate 2 ee BC DEE CHF OR 
. he differences between the multiple correlation coefficients in- 
volving different combinations of characters provide a measure of 
the relative importance of the different characters. It is evident 
that only a few ear characters contribute significantly to the relations 
shown by the higher order coefficients. On the basis of this informa- 
tion the following additional coefficients have been computed. In 
C. 1. No. 77 Ragan is 0.2062, in C. I. No. 119 Risen is 0.2201, and 
in C. I. No. 133 Raceox) is 0.1427. In C.I. No. 120 weight of 
ear is the one important factor (74:=0.1526), and there is little 
gain from including any other characters in the regression equation. 
DISCUSSION 
Considering the data as a whole and bearing in mind the varietal 
differences, several rather general tendencies are apparent. A posi- 
tive relation between yield and weight of ear and negative relations 
between yield and number of rows per unit of circumference and 
number of kernels per unit of length are the most important of these. 
Positive relations between yield and length of ear, weight of cob, and 
butt circumference are less important or consistent. Of the varia- 
tion in yield other than that due to experimental fluctuation, from 
20.4 per cent in C. I. No. 133 to 48.4 per cent in C. I. No. 119 could 
be predicted from regression equations involving one to three ear 
characters on the basis of the previous interpretation. Disregard- 
ing this interpretation, from 2.04 to 4.84 per cent of the total varia- 
tion in yield was a function of variation in one to three characters of 
the seed ears. 
Perhaps the fact of greatest interest from the breeder’s point of 
view is the negative relation between yield and number of rows. 
Relatively few rows certainly mean relatively wide kernels, but it is 
far from certain that this is the fundamental connection between 
number of rows and yield. The volume of an ear increases with the 
square of the diameter and only directly with the length. It there- 
fore should be possible to increase yield more readily by adding to 
the circumference than to the length of the ear. Kyle and Stone- 
berg (4) have shown that this is true, in fact, when only the bearing 
plants were considered. When all plants were considered, the larger 
numbers of barren plants in the many-rowed strains were enough to 
result in larger yields from ears of smaller circumference. Their 
