1891.] Are Acquired Variations Inherited ? 213; 
any other modes of origin. They must start, therefore, indefi- 
nitely, but secure a definite difection by the selection of those in 
favorable, and elimination of those in unfavorable, directions. 
This direction must be continuously plus where the characters 
are developed by direct Selection,” or neutralized where the char- 
acters are under the. sustaining power of Selection, or minus 
where the characters are degenerating under the influence of 
Panmixia (free intercrossing), or even of reversed Selection.” 
Every union of new individuals, according to Galton’s law of 
regression, however, will tend to draw back all the plus- and minus- 
variations to mediocrity, even where both parents show a tendency 
in the same direction. This regressive tendency to mediocrity, 
seen in the union of a single pair, will be further hastened by 
Panmixia.* We have assumed the continuous operation of 
Selection and abundant favorable variations to draw from, but we 
have seen, under Query 3, that variability is generally greatest 
when external conditions are most favorable; at the same time 
Selection must be least active, for the struggle for existence is ` 
least severe,—that is, Selection is least rapid when its materials 
are most abundant. So much for the probabilities of the produc- 
tion of definite lines of variation in single characters of this class. 
Evolution is not, however, a “log-rolling” process, in which some 
parts lag behind while others are improved by selection; in the 
fossil series, as all parts of the skeletal organism are observed in 
course of evolution at the same time, we must assume indefinite 
variability in every part, and admit the probability that, especially 
in uncorrelated parts, the sum of favorable variations, will 
be equal to the sum of unfavorable variations, and thus neu- 
tralize each other, so far as Selection is concerned. We must, 
therefore, add the assumption that these definite lines will be 
selected in correlation with those observed to occur in all the sur- 
rounding parts, and granting that groups’ of correlated parts may 
61 See Biol. Memoirs, pp. 264, 85, Tor, 275. 
6& Romanes has endeavored to show that where a character becomes detrimental 
Selection will tend actively to eliminate it. 
63 Galton has shown that in the union of two individuals showing exceptional charac- 
teristics only a few of the offspring would be likely to differ from mediocrity so widely as 
- the mid-parent. “ Natural Inheritance," p. 106. 
