PART III. | THE MECHANISMS OF FLOWERS. 479 
described above, and effect cross-fertilisation regularly, I have 
only noted the following :— 
(1) Bombus silvarum, L. 9 and $; (2) Osmia rufa, L. 9 ; (3) Anthidium 
manicatum, L. 9 (very freq.) ; (4) Megachile pyrina, Lep. (fasciata, Sm.), ¢, 
all sucking normally. 
Among useless visitors Sprengel mentions (p. 61) cabbage-white butterflies, 
which insert their proboscis in the small space between the lower anther-lobes 
and the corolla, and obtain the honey without effecting fertilisation. I have 
seen Plusia gamma, L., acting in the same way, I have also seen small bees in 
abundance (Prosopis communis, Nyl. ¢ 9, Halictus sexstriatus, Schck. 9, H 
nitidus, Schck. 9, H. nitidiusculus, K. 9, H. morio, F. 2) creeping into the 
flowers and reaching the honey without causing more than a slight and ineffectual 
movement of the levers. The position of the anthers quite prevents flies from 
stealing the pollen of this flower, and I have never observed a bee sweeping 
the anthers with its tarsal brushes. Delpino (No. 567, pp. 9, 10) unjustly 
questions my statement “that in Salvia the anthers are more or less protected 
from insects, which are restricted mainly or exclusively to the honey.” A 
further list of visitors in Low Germany (twelve Apidae, two Bombylide) is 
given in No. 590, mt. On the Alps I have seen the flowers visited by the 
hive-bee, by six humble-bees, eleven Lepidoptera, and Rhingia (609). 
This species is gynodiecious. In addition to the large-flowered 
hermaphrodite plauts, small-flowered female plants occur, in whose 
flowers the now functionless lever-apparatus shows all degrees of 
abortion. 
This species is remarkable above all other cases of gynodicecism, 
because it shows us the gradual abortion of all the stamens of a 
flower in four distinct stages: (1) The first of the five stamens, 
which must have been placed in the median line of the corolla 
superiorly, and is still present, more or less altered, in many Scro- 
pbularinez (Scrophularia, Pentstemon), had already disappeared in 
the common ancestors of the Labiates. (2) The two upper stamens 
of the remaining four were reduced to tiny stalked knobs in the 
ancestors of the genus Salvia. (3) The inferior anther-lobes of 
the two remaining stamens produce pollen to a greater or less 
extent in S. officinalis, S. porphyrantha, and S. triangularis, but in 
S. pratensis they are transformed into two hollow laminze which 
coalesce anteriorly. (4) In the small-flowered female form of 8. 
pratensis, the superior anther-lobes also have become useless, and 
subject to degeneration in common with the whole of the lever- 
apparatus (570, vol. xvi.; 609). 
Salvia Sclarea, L., ‘s athiopica, S. argentea, L., S. ieee 
Ait., 8. pendula, Vahl., and. S. rubra, Spr., have the same.mechan- 
ism according to ‘Hildebaad as, S._ pratensis... In. 8. -natans, Le 
(Hild. figs. 4-7), on the other hand, the rotation of the connectives 
