370 
this sentence which leads me to suppose that I 
haye not adequately explained myself, for 
clearly Professor Cockerell does not understand 
my theory as I understand it. Supposing, as 
regards any character, A BCD represent a 
line of individuals, then, if D reverts to B, that 
is, if D varies from C so that in his ontogeny he 
repeats the life history of the race only up to 
the point reached by B, omitting the additional 
characteristics of C, my contention is that C 
disappears altogether from the series, which, 
from the point of view of heredity, becomes 
ABD and, therefore, if the characters of C 
ever reappear in H, or any subsequent member 
of the series, they must do so (as they did in 
C’s case) as a result of evolution (variation 
away from the ancestry), not as a result of re- 
trogression, of atavism (i. e., variation towards 
the ancestry).* To test the truth of the theory 
we must consider a species in which evolution 
has been very rapid, as in that case atavistic 
changes can easily be observed with compara- 
tive ease; for example, any one of the various 
garden plants which are propagated by means 
of slips, it being probable that but a few semi- 
nal generations intervene between these widely 
divergent types and the ancestral wild varieties. 
The seeds of such a cultivated plant usually 
give origin to plants, which, in the great ma- 
jority of instances, resemble much more nearly 
the wild than the cultivated variety. Now, if 
Professor Cockerell is right, the continual semi- 
nal propagation of such reverted plants should 
often result in a ‘reversion’ to the cultivated 
type; whereas, if Iam right, this ‘reversion’ 
should never occur, since (for purposes of hered- 
ity) the cultivated variety has been eliminated 
from the ancestral line. Such reappearance 
would purport, therefore, not reversion, but a 
fresh evolution, which evolution would be too 
great to be accomplished in a single generation. 
So also the race horse, in the absence of selec- 
tion, would revert to the ordinary horse, after 
which a race horse should be as rare among the 
* For the sake of simplicity I do not here consider 
such a case as when D repeats the life history up to C 
and then reverts back to B, but it is dealt with to some 
extent in my book, and the process is of importance 
when considering the phenomena connected with ‘re- 
versed selection.’ 
SCIENCE. 
[N. S. Vou. VI. No. 140. 
progeny as it is at present among the progeny 
of ordinary horses. 
It follows as a logical conclusion from the 
foregoing that cessation of selection, as regards 
any structure or character, during unlimited 
time must result in unlimited atavism. First 
recent ancestors would be approximated to, 
then ancestors more remote, and lastly that re- 
mote ancestral condition when the structure or 
character did not exist. It would then disap- 
pear as utterly as have done, from this cause, 
the limbs of serpents. In fact, atavism effects the 
same result as disuse is said to do by Neo-La- 
marckians, and, unless we accept the Lamarck- 
ian doctrine, I confess Iam unable to perceive 
any other probable reason for retrogression. 
Professor Cockerell further writes: ‘‘The 
germ, it must be supposed, contains units rep- 
resenting many phases of existence, while 
others are new. When one of the latter de- 
velop we say the variation is a progressive one ; 
when the former develop we call the result 
atavism. It is reasonable to suppose that enyi- 
ronmental and germinal selection are the factors 
which determine which of its possible develop- 
ments the germ shal]l undergo.’’ Here, again, is 
proof that I have not made myself clear on this 
point to Professor Cockerell. He has been kind 
enough to speak with some approval of my 
theory, but he has not perceived that it is ab- 
solutely incompatible with the theory of ger- 
minal selection. Surely in the present state of 
our knowledge it is too much to affirm that 
the germ must contains units ‘ representing 
phases of existence.’ During the ontogeny 
many phases of existence are represented in 
fleeting but orderly succession, but again, surely 
it is not essential to suppose that the develop- 
ment of each is due to one or more units; and, 
moreover, does not the regular recapitulation 
of the phylogeny during the ontogeny negative 
the idea that any ‘units’ are ‘held over?’ 
I must be brief in my comments on the rest 
of Professor Cockerell’s review. He says that 
in considering man’s present evolution I haye 
neglected to take into account what Mr. B. 
Kidd has called ‘social efficiency.’ If he turns 
to my work (pp. 178-200) he will find that I 
have not altogether neglected the matter. 
Briefly, I have attributed differences in social 
