PART 111. | THE MECHANISMS OF FLOWERS. 155 
flower, so that the anthers, still dusted with pollen, are closely 
pressed together, and lie over the stigmas; they must therefore be 
touched before the stigmas by an insect alighting in the middle of 
the flower. In this stage the flowers are adapted for dusting the 
bellies of insect-visitors with pollen for the fertilisation of other 
flowers ; but their own stigmas are more easily reached by their own 
than by other pollen. 
: The anthers now easily fall off, and one often finds the older 
flowers without anthers, but with five fully outspread stigmas. 
These flowers are therefore again female only, but their stigmas 
have already come in contact with their own pollen, though they 
may still be fruitful with that from younger flowers. 
If we compare the flowers of G. pusillum with those of its 
near ally G. molle, with which it grows in company, and blooms at 
_ the same time, we see that the former are smaller and paler, and 
are accordingly more rarely visited by insects. In each case, 
_ however, the regular occurrence of self-fertilisation compensates 
for the great uncertainty of cross-fertilisation; and further, the 
very early development of the stigmas enables the rare insect-visits 
to be made full use of when they do occur. Finally, the more 
rapid development of the flower permits the number of anthers to 
_ be reduced to one-half. 
In spite of frequent watching, I have only noticed the flowers 
| of G. pusillum to be visited by one insect, a small gnat, Ascia 
| podagrica, F., but by that on several occasions. It settled in the 
_ middle of the flower and sucked honey from it. See also No. 
590, II. 
rr —— 
REVIEW OF THE PRECEDING SPECIES OF GERANIUM. 
If we exclude G. sanguinewm, which grows in more shady 
_ spots, and arrange the other species according to the size and 
conspicuousness of their flowers, we get a series as follows: 
(1) G. palustre and G. pratense ; (2) G. pyrenaicum ; (3) G. molle ; 
(4) G. pusillum. We get practically the same series if we arrange 
| these species according to the decreasing chance of cross-ferti- 
_lisation in case of insect-visits; and so also if we take them 
according to the increasing likelihood of self-fertilisation in 
_ default of insect-visits. For (1) in G@. pratense and G. palustre, 
_ self-fertilisation never takes place in the ordinary flowers,’ while 
_ _ 1 I say “in the ordinary flowers,” because it is possible, and even probable, that 
| homogamic flowers oceur, as Axell has shown in the case of @. silvaticwm, 
