310 LATE ORIGIN OF BLUE FLOWERS 
Gentians. Here, also, while the well-known deep blue 
species have long tubular flowers, specially adapted to 
bees and butterflies, the yellow Gentiana lutea has 
a simple open flower with exposed honey. 
Miillerand Hildebrand! havealso pointed out that the 
blue flowers, which, according to this view, are descended 
from white or yellow ancestors, passing in many cases 
through a red stage, frequently vary, as if the colours had 
not had time to fix themselves, and by atavism assume 
their original colour. Thus Aquilegia vulgaris, Ajuga 
Genevensis, Polygala vulgaris, P. comosa, Salvia pra- 
tensis, Myosotis alpestris, and many other blue flowers, 
are often reddish or white; Viola calcarata is normally 
blue, but. occasionally yellow. On the other hand, flowers 
which are normally white or yellow, rarely, I might 
almost say never, vary to blue. Moreover, though it is 
true that there are comparatively few blue flowers, still, 
if we consider only those in which the honey is con- 
cealed, and which are, as we know, specially suited to 
and frequented by bees and butterflies, we find a larger 
proportion. Thus, of 150 flowers with concealed honey 
observed by Miller in the Swiss Alps,? 68 were white 
or yellow, 52 more or less red, and 30 blue or violet. 
However this may be, it seems to me that the 
preceding experiments show conclusively that bees do 
prefer one colour to another, and that: blue is distinetly 
their favourite. 
' Die Farben der Blithen, p. 26. 
2 Alpenblumen, p. 492. 
