1898] A THEORY OF RETROGRESSION 399 
evolutionary variation in other particulars. It is slight because, 
since such species have evolved but slowly, reversion to a not very 
remote ancestor does not result in any appreciable change of type. 
Thus, under ordinary circumstances, if a man reverted in any parti- 
cular to an ancestor of a thousand years ago, no one would recognise 
to what the change of type was due. Not only would the change 
be too slight, but the observer would need to have a knowledge 
of the ancestral form, and such knowledge is usually impossible. 
Sometimes, however, recognisable reversion does occur even among 
such beings. Thus a man may resemble the portrait of some far- 
away ancestor, or again the progeny of an ordinary pair of horses 
may exhibit the zebra-like stripes of a remote ancestor. It is not, 
however, among complex beings, slowly evolved in every particular, 
that we must seek our proofs. We must turn to plants and animals 
that have undergone swift evolution in some one particular, and 
this, so far as I know, occurs only under stringent Artificial Selec- 
tion. For Natural Selection, having care for many characters, re- 
sults in but slow evolution; but Artificial Selection, having care for 
only one or only a few characters, results in much swifter evolution. 
Supposing, then, we take any breed of domesticated animals or culti- 
vated plants, and, after choosing the finest specimens, henceforward 
breed indiscriminately from these and their descendants ; what then 
happens? It is notorious that under such circumstances cessation 
of selection is marked by a reversion towards the ancestral type, a re- 
version swift in proportion to the swiftness of the antecedent evolution. 
Thus, without continued stringent selection, the speed of race-horses 
cannot be maintained; they tend to lose their special characters, 
and revert to the ordinary horse. The same is true of all other 
prize breeds. Again, careful breeding from ordinary horses readily 
evolves a speedier race, for the offspring of ordinary horses in many 
instances surpass the parents. But, in proportion to the success of 
the breeder, further improvement grows continually more and more 
difficult, till at length evolution practically reaches a standstill. 
Improvement thereafter is very slow indeed. For this reason it is 
now very difficult to improve our breed of race-horses. The off- 
spring of a pair of the finest animals are in the great majority of 
cases inferior to their parents, and, therefore, practically all that the 
most stringent selection is now able to achieve is to preserve, not to 
improve, the race. It is, therefore, plain that, owing to the increas- 
ing tendency towards reversion, rapid evolution quickly slows down, 
till, even in the presence of stringent selection, it practically ceases. 
But perhaps the most striking proofs of the present theory are 
furnished by certain cultivated plants (for instance the apple), which 
are usually propagated by means of slips or suckers—that is, by 
detached portions of the individual. Practically speaking, the most 
