604 
ciency possible. The failure to understand 
this difference in purpose has frequently 
led to confusion in our discussions. 
It is beyond the scope of this paper to 
discuss the kinds of variation used in these 
different types of selection, even if we pos- 
sessed the requisite knowledge, which is 
doubtful. The speaker may be pardoned, 
however, for digressing far enough to state 
that it is his conviction that there is no 
very hard and fast line between that varia- 
tion which is in considerable degree in- 
herited, such as is found frequently in high 
milk-producing cows in selection within the 
breed, and the mutation which gives abso- 
lute inheritance and establishes a perma- 
nent new mode. The great difficulty in 
determining whether there is any true 
cumulative action of selection which will 
extend a character beyond the limits of the 
race or species is met in determining what 
are and what are not mutations. My ex- 
perience has led me to conclude that the 
continuous selection of maximum fluctua- 
tions in a certain direction may in some 
eases lead to the gradual strengthening of 
the character until finally it may become, 
more or less suddenly, fully heritable and 
it would then be recognized as a mutation. 
In many cases we find exceedingly small 
differences maintaining themselves genera- 
tion after generation under different en- 
vironments when the lines of descent are 
kept pure. A marked illustration of this 
is afforded by Mr. Evans’s studies on pure 
lines of Stellaria reported at-this meeting. 
The segregation of such characters in hy- 
bridization would be exceedingly difficult 
to recognize if it did occur. Again the 
occurrence of such small mutants, if we 
may so designate them, within a breed 
under selection, if not recognized and iso- 
lated, would be crossed with fluctuations 
and cause variations which would be recog- 
SCIENCE 
[N.S. Vou. XXXV. No. 903 
nized as regressions in the highly selected 
strain. 
I think it will have become clear from 
the above discussion that in the present 
state of our knowledge of selection we can 
only advocate that practical breeders con- 
tinue their selections as in the past. This 
is particularly true in the cases where it 
is the idea to maintain the race or breed at 
its highest efficiency. In the case of plant 
breeders working to produce new races, the 
mutation theory introduces a new element 
and leads the breeder to search for a 
mutant possessing desirable characters 
which he can isolate and which he may 
expect will reproduce its characters as soon 
as he has purified the type from mixtures 
derived through hybridization with other 
types. He will select the type to purify it 
rather than to augment its good qualities. 
Returning again to the question of new 
characters, we may profitably question 
more definitely where such new characters 
come from, if they are not produced by 
selection. Clearly, no problem is of more 
importance to the breeder than to be able 
to definitely produce or cause such new 
characters to appear. If the breeder must 
await the pleasure of nature to secure the 
changes he desires, the waiting may be long 
and tedious. If he must watch thousands 
of plants of a certain race or species every 
year in order to find the apparently acci- 
dental variation or mutation in the direc- 
tion of the improvement he has in mind 
which may rarely or never be found, the 
process will be so hazardous that we should 
have to await the accidental discovery of 
any new characters. Indeed, up to the 
present time we have had practically no 
other recourse than to await the accidental 
discovery of such new characters. ‘We, 
however, have had many theorists and in- 
vestigators who believed that changed en- 
vironment would stimulate the production 
