COLLINSIA. 53 
tered pale blue flowers I was observing that not a few, indeed 
nearly all of the truly terminal ones were hexamerous, having 
a corolla-lobe and a stamen too many. Stopping to observe 
these not altogether insignificant anomalies, a smaller plant 
with larger flowers caught my eye, a plant which I might 
atherwise have missed, and one which proved itself a greater 
wonder. At the distance of a couple of rods I thought it a 
small full blown specimen of Gilia densiflora, and should 
have passed it by for that but that at a second glance I seemed 
to note a slight irregularity in the corollas. Clambering down 
close to it I saw, first of all that, of whatever species, it was a 
stunted individual the top of which had been eaten off by 
some grazing animal, and that the mere stub which was left 
was crowded with flowers. I then discovered to my amaze- 
ment that the plant which bore these exceedingly gilia-like 
corollas was Collinsia bicolor. 
Before proceeding to describe these anomalies I must speak 
of that normal form of which they are but a casual variation. 
The corollas of the collinsias are, as a rule, strongly bilabiate, 
the limb consisting of two lips, the upper an erect and plane 
two-lobed lamina, the lower a considerably different three- 
lobed one. In the lower, the lateral lobes are plane and 
spreading, and, as regards the upper lip, exactly divaricate, 
while the middle one of the three is closely folded upon itself 
into a narrow keel which encloses the stamens and lies con- 
cealed beneath and behind the two broad and showy lateral 
ones, in this respect imitating the papilionaceous corollas of 
some Leguminosw. Indeed, beginners in botany are apt to 
mistake the first collinsia corolla they see for that of a legumi- 
nous plant, being deceived by this rather striking imitation. 
It is in reality a corolla of quite extreme irregularity between 
the bilabiate and the personate types. 
In the specimen under consideration there were some thirty 
well expanded corollas. In about twenty of them there was a 
manifest deviation from strict regularity, but only in this, that 
