[Vor. 4 
64 ANNALS OF THE MISSOURI BOTANICAL GARDEN 
red and green, and the petioles of the leaves vary much in 
length. 
In Т. nivale Riddell the peduncle may be erect, declined or 
nodding, as in the three species, Т. erectum, Т. declinatum, 
and T. cernuum respectively. These conditions іп T. nivale 
probably represent unit varieties which would breed true in 
cultivation, and it is reasonable to suppose that the differ- 
ences between the above three species have also originated 
through unit variations. 7. nivale and Т. rivale are the most 
aberrant of the North American Trillia. The dwarf character 
of both may be supposed to have originated through muta- 
tions. The leaves of T. nivale most nearly resemble in shape 
those of T. viride Beck, though much smaller. In the latter, 
however, both the leaves and flowers are sessile, while in T. 
nivale the leaves are short-petioled and the flowers rather 
short-peduncled, so that a close relationship cannot be 
assumed. 
T. nivale differs chiefly from its probable ancestor, T. 
grandiflorum, (1) in being a dwarf, (2) in the shape of the 
leaves, which are oval, obtuse, with short petioles, instead of 
rhombic-oval, acuminate, sessile, and (3) in the shape of the 
petals, which are oblong or oval instead of oblanceolate or 
obovate-oblanceolate and mucronate. If we compare these 
differences with those between Oenothera Lamarckiana and 
its dwarf mutant (E. nanella, we see that the differences (1) 
and (2) above might have originated at one stroke, though 
as regards (2) the condition is reversed, for in Œ. nanella the 
leaves are mostly sessile, while in Œ. Lamarckiana they are 
petioled. The difference (3) in the petals of T. nivale would 
probably have required another and independent step. At 
any rate, although the species is so aberrant in the genus, two 
mutations are sufficient to account for its origin. According 
to the older views, one must have assumed a long period of 
isolation and gradual selection to produce such a form. Now 
we know that there is no necessary relation between the 
length of a step and the time taken to produce it. A relatively 
wide mutation wil happen just as quickly as a narrow one, 
and, indeed, if the wider difference has any survival value it 
