SEPTEMBER 3, 1897. ] 
briefly, that theory is as follows: It is known 
that the ontogeny recapitulates the phylogeny, 
though in a rapid, blurred and indistinct way.* 
In other words, almost every individual reca- 
pitulates the traits of all his ancestors, beginning 
with those of the first (as, for convenience, we 
may call the unicellular organism) and ending 
with those of the last. I use the word ‘almost’ 
because, though the earlier stages must, of 
course, be recapitulated, for otherwise the indi- 
vidual would not.develop, yet sometimes (indeed 
often) the very last steps made in the evolution 
of the race fail to be reproduced by the indi- 
vidual in his development. He then resembles 
some remote ancestor more than he does his 
parents, and presents, in fact, an example of 
what is known as atavism. When the evolu- 
tion of the race has been rapid, reversion to 
any given ancestor results, of course, in a much 
greater and more observable degree of retro- 
gression than when the evolution has been 
slower, and therefore, while offspring of race 
horses often exhibit far-reaching reversion to- 
wards the ordinary horse, and while the seed- 
lings of various garden plants (apple, peach, 
rose, etc.), which have been evolved under ex- 
cessively stringent selection, generally revert 
to the ancestral type, the offspring of wild 
plants and animals generally ‘breed true.’ 
An individual may vary from his parent in 
two ways—towards the ancestry when some of 
the last steps made in the phylogeny are omitted 
in his ontogeny, or away from it, the former 
variation being atavistic, the latter evolution- 
ary; and so far as we are aware the chances of 
his doing the one or the other are equal. But 
while every variation towards the ancestry, 
that is, every failure to repeat in the ontogeny 
the last steps made in the phylogeny, produce 
atavistic retrogression, every variation from it 
need not be an extension of the previous evolu- 
tion ; it may constitute a reversal of it (as will 
be seen presently) or be in some other different 
direction, whence it follows that in the absence 
of selection a species must always undergo 
retrogression. 
Up to this point, or nearly up to it (for as to 
*J have given reasons why the recapitulation is 
blurred and indistinct, but these need not detain us 
here. 
SCIENCE. 
369 
this I am not very clear), my critic and I seem 
to be in agreement; but beyond this we differ, 
for while he thinks that such atavism can result 
in very limited retrogression only, I am of the 
opinion that in the entire and continued ab- 
sence of selection it must result in absolutely 
unlimited retrogression. Before dealing with 
his objections I ought to say that in my book I 
quite reject the Lamarckian doctrine of the 
transmissibility of acquired traits, and since 
Professor Cockerell is good enough to say that 
he regards the arguments there used as conclu- 
sive, the matter is not in dispute between us. 
Moreover, he seems to agree with me in attrib- 
uting evolution to the accumulation of small 
‘normal’ variations, not to the accumulation 
of great and abnormal variations, for the reason 
that the latter tend to be swamped, owing to 
their infrequency. Ishould also make it clear 
that by atavism I do not necessarily mean ata- 
vism of the whole organism. Such wholesale 
reversion to the ancestral type must be ex- 
tremely rare in the case of all species slowly 
evolved under the ordinary conditions of na- 
ture; it occurs only, so far as I am aware, in 
such species as have evolved under very strin- 
gent artificial selection. Under natural condi- 
tions an individual may exhibit evolution in 
some particulars and atavistic retrogression in 
others, but these latter, owing to the slowness 
of the antecedent evolution, must generally be 
minute in amount, and, therefore, when asso- 
ciated with evolutionary changes, unrecogniz- 
able in the individual, though recognizable after 
the lapse of generations in a line of individuals. 
Thus, while selection may result in evolution in 
the legs of a bird to which flight has become use- 
less, atavistic reversion in the absence of selec- 
tion as regards them would result in the retro- 
gression of its wings. But here again, the wings, 
being complex organs, would not equally retro- 
gress in all their parts, and, therefore, would 
never very closely approximate to the ancestral 
type. 
Professor Cockerell says: ‘I cannot see, with 
Mr. Reid, that there would be unlimited ata- 
vism, because when the atavistic changes had 
proceeded from B to A the B features would be- 
come ancestral and a new atayism from A to B 
would appear.’’ (The italics are mine.) It is 
