62 MR. J. C. WILLIS ON THE 
been separated, by insects or otherwise, the filaments simply bend 
backwards, one on each side of the style, like those of Teucrium. 
If, however, as was usually the case in Cambridge, the anthers 
still cohere, the filaments have to go through more contortions 
to effect their object (fig. 15). 
The flower appears to be adapted specially to long-tongued 
Lepidoptera, but none were observed visiting it in Cambridge 
(no night observations were made). Only a large insect would 
touch the stamens when standing on the lower lip. Various 
Syrphide were observed creeping over the flowers and eating 
pollen; but the only insect noticed attempting to reach the 
honey was a large yellow humble bee, which visited several 
flowers and made desperate efforts to obtain the nectar, though 
apparently without any success, its proboscis not being nearly 
long enough. Its back did not touch the stamens at all, though 
sometimes coming within a few millimetres. 
The flower occasionally becomes self-fertilized during the 
movements of the stamens, &c., but very often this does not 
occur. Only a few of those examined set seed, and of these 
many had only one or two. 
M. kalmiana and M. fistulosa.—These species have smaller 
flowers of a purplish colour. In the principal points of structure 
they resemble the preceding species, but the anthers are turned 
so as to stand with the long axes horizontal, as in M. ciliata (E. 
and G. loc. cit.). Kerner* has observed honey-bees visiting 
M. fistulosa, while avoiding the scarlet M. didyma, and suggests 
that the latter may be visited by humming-birds. 
The Cohesion of the Anthers.—The manner in which the 
anthers of Monarda cohere was pointed out by Mr. F. Darwin, 
who suggested also that it should be worked out. In the young 
bud the epidermis of the anther is quite smooth (fig. 18), but 
after the bud has reached a length of about 5 mm. the cells of 
the epidermis swell out into papille of a conical shape with a 
rounded apex. This process occurs on those parts of the anthers 
which are contiguous to one another, but not only on the parts 
which are in contact. The wall of the papilla becomes slightly 
cuticularized, and is also transversely ridged on the surface of the 
cuticle. As the anthers are pressed together in the bud, the 
papille of the one become interlocked with those of the other, 
and thus the cohesion is effected. 
* * Pflanzenleben,’ ii. p. 191. 
