EFFECT OF SELECTION 139 
gradually slower and slower. At the end of the series 
two sudden steps upward break the general regularity 
of the series of records. But on examination of the 
evidence it is found that these are associated with 
special conditions, and are not really exceptional. The 
first of these breaks—that which occurs in 1892—is 
coincident with the introduction of a new type of sulky, 
having ball bearings and other improvements ; whilst 
the record of 1903 was accomplished behind a pace- 
maker carrying a wind-shield. Neither of these records, 
therefore, is strictly comparable with the rest of the 
series. 
The observations in this case do not, indeed, seem 
to be sufficient to afford the basis for a final decision 
against the theory of the indeterminate power of selec- 
tion. Yet Malthus’ criticism clearly applies very 
definitely to such a case—i.e., there must be a limit 
beyond which the speed of the trotting-horse will 
never improve without a fundamental change taking 
place in his organization. It seems, therefore, safe to 
conclude that the curve to which the series of records 
approaches is of the character of a parabola—+.e., one 
which continually becomes more and more nearly 
horizontal—as the speed of the horse gradually ap- 
proaches its highest possible limit. 
De Vries, then, contends that all new domestic breeds 
have arisen by the discontinuous method as definite 
novelties. Darwin himself was perfectly aware that 
this is usually the case, but the conclusion which he 
drew from the fact was a different one, as the following 
passage shows : 
