600 
lish the study of evolution on an experi- 
mental basis. With the demonstration that 
evolution could be studied experimentally, 
the question of the effectiveness of selection 
was taken up, and we are now doubtless on 
the road to a solution of the problem. It 
is only possible for us here to call attention 
to a few of the researches in this direction. 
The classical researches of De Vries, now 
familiar to us all, challenged the correct- 
ness of the selection theory and sought to 
show that species originated by sudden 
jumps or mutations. We may admit that 
De Vries proved that species or new char- 
acters were formed suddenly as mutations, 
but this would not prove that they might 
not also be formed or actually induced to mu- 
tate by a continuous -process of selection. 
Indeed, in his experiments on the produc- 
tion of a double-flowered variety of Chrys- 
anthemum segetum (‘‘Mutationstheorie,’’ 
Vol. I., p. 523), a few generations of selec- 
tion led to markedly increasing the number 
of ray-florets before the ligulate corollas 
appeared among the disk-florets, the change 
which he interpreted as the mutation that 
gave him the double variety. 
Johannsen has contributed much to our 
knowledge of selection and has given us a 
more exact method of experimentation by 
his conception of pure lines, biotypes, geno- 
types and phenotypes. His experiments in 
the selection of pure lines of beans in an 
attempt to produce large and small seeded 
types, have led him to conclude that selec- 
tion within a pure line is ineffective in pro- 
ducing changes. He did, however, secure 
new types from pure lines through muta- 
tions. 
Tower’s experiments with the potato 
beetle in attempting to create by selection, 
large and small races, albinie and melanic 
races, and races with changed color-pat- 
tern, although conducted carefully for 
from ten to twelve generations, failed to 
SCIENCE 
[N.S. Vou. XXXV. No. 903 
give any evidence of producing perma- 
nently changed types. While strains of 
plus and minus variates gave populations 
with a range of variation apparently mark- 
edly restricted to their respective sides of 
the normal variation range, still these se- 
lected strains did not greatly exceed the 
normal range of variation in either direc- 
tion, and when the selection was discon- 
tinued, in two or three generations, again 
produced populations exhibiting the nor- 
mal range of variation. Clearly no new 
unit characters had been added by the 
selection. Tower, however, found that by 
subjecting the beetles, during the process 
of the formation of gametes, to certain ab- 
normal conditions, he was likely to obtain 
mutations in the progeny that would imme- 
diately form the beginnings of new races. 
Jennings in a series of selection experi- 
ments conducted with paramecium, which 
were continued for over twenty genera- 
tions, obtained no evidence of a permanent 
modification of the type. 
Pearl has conducted an extended experi- 
ment in the selection of chickens in the 
attempt to produce a breed of high egg- 
laying capacity. His results have led him 
to the conclusion that selection alone has no 
effect in producing a permanent improve- 
ment or a change of type. 
Up to the present time these are the 
principal contributions to the subject, 
that discredit the effectiveness of selec- 
tion as an active agency. 
On the opposite side of the controversy 
we have the very careful and extensive re- 
searches of Castle and MacCurdy in the 
selection of Irish rats to increase the black- 
colored dorsal band on the one hand and to 
decrease or obliterate it on the other. 
Castle appears to have obtained very posi- 
tive results favoring the gradual cumulative 
action of the selection, as he succeeded in 
markedly increasing the amount of black 
