148 
But if the shape of the handle and the nature of the catch 
occurred in an accidental sort of way among umbrellas of 
all prices, and if he took care to mention his different 
requirements one at a time—then first he might say that 
he would not have an umbrella with a crooked handle, 
so all the umbrellas with crooked handles would be put 
out of his way. Then suppose that among the straightest 
handled umbrellas he asked for a catch of a particular 
kind—which might be old-fashioned or a new fashion 
hardly yet come in, and so not to be found in any great 
number—then all the other umbrellas with all other kinds 
of catches would be put aside ; and then it is quite con 
ceivable that he might not like the look or the price of 
the remaining umbrellas. Of course when this was dis 
covered, the obliging shopman would begin de novo. 
But selection by death as it is supposed to occur in 
nature knows nothing of da capo operations ; and hence, 
as it seems to me, the difficulty presented by successive 
tests of fitness is not overcome by the parable of the 
umbrella. It does not prove that “animals will, in each 
generation, have the best chance of survival which are 
most in harmony with their environment.” Hence we 
have a fourth most important difference between Natural 
and Artificial Selection which has been overlooked by 
Mr. Romanes. 
We have seen that Natural Selection is handicapped by 
co-operation among animals which often prevents that 
rigid and pitiless competition which the theory presupposes. 
Now, it is quite clear that domestic animals cannot com 
bine in a similar way to resist the will of the breeder. 
Mr. Romanes incidentally and parenthetically remarks 
that the facts of domestication illustrate not only the 
action of Natural Selection, but also of Sexual Selection. 
But I venture to think nothing can be more different than