i aeinioa 
SEED-EAR CHARACTERS AND PRODUCTIVENESS IN CORN 9 
BASIS OF INTERPRETATION 
Before discussing the coefficients of correlation, it is desirable 
to consider briefly the basis on which they will be interpreted. The 
odds are about 45 to 1 that correlation exists between two variables 
for which the determined coefficient of correlation is three or more 
times its probable error. This limit has been taken as the criterion 
of statistical significance, and no individual coefficient less than 
three times its probable error is considered significant. 
TaBLe 5.—Coefficients of variation for yield for 1,300 check rows that were grown 
in various seasons under conditions similar to those of the ear rows dealt with 
Ty 
I~ ient. || : IN ient 
Variety and season | gab of ees || Variety and season | See Li pena 
l 
C. I. No. 7 | | C. I. No. 120: 
515 oe ie ae ee 7: 10.0 || IOU7~ 2 SSS ee } 60 5.9 
1 Oe eee as eee rae | 162 21.4 || 118 1. Pee 47 18.5 
Wide ee 2 Se 64 10.0 | 1919: eee ea 64 5.5 
git) £25 ee SPs ose ee 19 5.8 | 1920 = 7S BRS. Sree 53 | 11.3 
Pier SS BS 13 4.5-| 
aGU Se 2 See et Se 13 9.4 | Total or mean______| 224 210.9 
PONG 8 er Pa eee 17 6.6 i 
[11 ern o/ seteaehaer 21 7.2 || C.1. No. 133 
|_—______}_ ______ | O11) Ute as 84 12.4 
Total or mean______ ! 383 | 215.5 |, 1962 1 ee ee | 7 7.8 
—-_ —————————— i 1913 1 36st 2 ANS 70 f(r | 
G. I. No. 119: igs S914! Se sees 90 8.4 
{1 i7 pRe eaee 62 14.0 1916... eee gs oe 19 Se 
WRANI Aes 47 13.0 | £9772 PP Ti A 21 | 16.6 
gL LS Rae ee eee 142 9.6 | IGS. Re pe | 15 | 8.2 
1999 25 VE 53 | 2.8 | IHG. We see DET 18 | 12.3 
Total or mean_-___-_- 304 2.10.6 Total or mean_____- 389 | 29.9 
Total or mean for } 
all varieties_______ 1, 300 | 311.91 
| | | 
1 Hill-checked rows. 3 Unweighted average of the four varietal coefficients. 
2 Weighted according to the number of rows. 
The true meaning of a coefficient of correlation depends on far 
more than its statistical significance, however, and arbitrary limits 
can not be set within which it is or is not important for all cases. 
Thus, a coefficient of less than +0.3 between two ear characters 
would be relatively unimportant. A similar coefficient between 
one of the characters of the seed ears and yield, however, would 
indicate a rather important relation. This is evident from a con- 
sideration of the following facts: 
Check rows were grown in a number of the ear-row plats from 
which the data used were obtained. The check rows were planted 
with uniform seed in any given plat and were distributed at regular 
intervals across the field. Their yields should afford a measure of 
the variation due to causes other than seed value. The coefficients 
of variation of 1,300 check rows are shown in Table 5. The mean 
coefficient of variation for the 1,300 rows, computed as indicated in 
the table, is 11.91. The unweighted mean of the standard devia- 
tions on a percentage basis of the individual ear rows of the same 
four varieties is 12.25. Based on the squares of these values, the 
variation among the check rows was 94.5 per cent as large as that 
among the individual ear rows. It is not possible to say that 94.5 
per cent is a statistically accurate measure of the variation in the 
22050°—25j}——2 
