48 (249) DOMINANCE. REVERSED DOMINANCE RECESSIVENESS OF 
yet enough been specialized, to show the supposed difference in vari- 
ability. Perhaps a difference of headlength of the parents of 0.5 cM. is 
already to large; perhaps also it is unfavourable, that sons and daught- - 
ers have been taken together. 
Now let us still look for the results of the inquiry of the variability 
of the indices of the children of parents with different high indices 
(tab. XV, p. 182). 
The average index of females being larger than of males, we comput- 
ated the mean index of fathers and of mothers. We found for 353 
fathers M = 79.5 and for 357 mothers M = 80.7. (cf. also 1920c, Tab. 1 
p.24, M; = 79.56, also = 79.45; M, = 80.78, also = 80.73). For our 
whole material we had found (1920b) M, = 80.4 and M, = 
81.04. For the correction of mothers to fathers we have diminished the 
indices of mothers with one unity. 
Successively there have been made tables, where the corrected in- 
dices of mothers are less than one unity, less than two, than three and 
than four unities larger, resp. smaller than those of fathers (Tab. XV, 
18). | 
In these eight tables we see in general for sons and for daughters, 
that these indices agree in some degree with those of parents. 
For the computation of the standarddeviations the numbers are too 
small. Classes have therefore been joined (tab. 15a and 15b). One class 
comprises the 4 or 5 (or 3) classes of the high indices of tab. XV, one 
class is the class of tab. XV of the average indices and the third class 
comprises the 3 or 4 (or 5) classes of tab. XV of the low indices. So we 
have one class for high, one class for average and one class for low in- 
dices. In this way it is attained thatthetwo classesofextremespecimens 
contain large numbers. For the class of parents with average indices 
that class of tab. XV has nearly always been taken, of which the in- 
dices approach nearest to the calculated mean of the indices. 
Joined in this manner the numbers for each table are still small. 
Therefore, as appears from tab. 15a and 15b, the tables are 2 oined once 
more in different ways. 
From tab. 15a, of the sons, we see, that in tab. XV, the standard- 
deviation of the sons of parents with average indices is larger than 
that of the sons of parents with high or with low indices. This is also — 
true for tab. XV, and for tab. XV, and XV, together. So for those fam- 
ilies, where the indices of parents differ less than one unity the vari- 
FOR 





