1921-22.] 
On the Dominance Katio. 
337 
which are in the ratio 3 : 5. 
The dominance ratio is therefore 
? = -2308; 
3 + 2(5) 
the greater variation in the ratio — showing itself in a lower dominance 
ratio. 
3. In the third symmetrical case, when genetic selection is at work, the 
variation of is even greater (fig. 3) ; since both and contain the 
factor the factors in which p or q is very small, make no appreciable 
contribution to these quantities, consequently we only consider the central 
portion of the distribution, where 
dfa , 
sin (ji cos 
the intensity of selection appearing only as a constant factor, and therefore 
influencing the range of variation of the species, but not its dominance 
ratio. Here we have the integrals 
Utt 
i sin^ </) 
Jo 
cos^ cf)d(f) and 
leading to a dominance ratio 
1 
1+4 
• 2000 . 
sin (fi cos^ cfidcjy 
4. In the case of genotypic selection, which case most nearly reproduces 
natural conditions, the distribution in the centre of the range is 
■ 5 "^ 3 , > 
Sin <J> cos*^ <p 
consequently the two integrals with which we are concerned 
I sin^ cfi cos , I sin ^ cos^ cfidcji 
Jo *'o 
are now equal, and the dominance ratio is raised to 
In considering the interpretation of the dominance ratio, in our 
previous inquiry, we found that for symmetrical distribution the value 
J occurred as a limiting value when the standard deviation of 0 ^=log^^ 
was made zero. Since the dominance ratio calculated from observed 
human correlations averaged *32, with a standard error about *03, we 
were led to consider that either the allelomorphs concerned occurred 
usually in nearly equal numbers, a supposition for which we saw no 
VOL. XLII. 22 
