PART IY. | GENERAL RETROSPECT. 591 
had once been insured becomes uncertain owing to the competition 
of other flowers (¢.g. Malva rotundifolia, species of Geranium), or 
owing to the unfavourable locality (cf. Lysimaehia nemorum) or 
unfavourable weather (cf. Veronica Beccabunga), the mechanisms of 
the flowers have in many cases undergone a change such as to 
render self-fertilisation again possible ; in a few cases reversion to 
anemophily has taken place (Artemisiacer, Zhalictrum). As 
examples of the countless ways in which plants revert to self- 
fertilisation in default of sufficient insect-visits, I may mention the 
following :—In some dichogamic flowers the stigmas curl back upon 
the anthers or other parts which still retain some pollen (Stellaria 
graminea, Malva rotundifolia, species of Geraniwm, Composite) ; 
the stigmas sometimes curl back till they come in the line of fall of 
the pollen (Melampyrum pratense) or even place themselves between 
the anthers (thinanthus minor); anthers which stand in a ring 
round the stigma may, in default of sufficient insect-visits, con- 
verge above the stigma, applying their pollen to it (Myosotis, 
Lithospermum, Cruciferze); even mechanisms which effect cross- 
fertilisation with astonishing precision in case of insect-visits are 
not unfrequently transformed so as to render self-fertilisation 
inevitable when insect-visits are few (certain Orchidez, Fuma- 
riacee, Salvia); or the production of cleistogamic, self-fertilised 
flowers may compensate for the loss of the power of self-fertilisation 
in the ordinary flowers ( Viola). 
Unlike Axell, who brings his book to a conclusion by saying 
“ We see thus that the development of mechanisms for fertilisation 
in Phanerogams has gone on and still goes on in the same 
direction,” + I should say : The dependence of entomophilous flowers 
on guests so infinitely various in habits, tastes, and numbers, in 
their food and in the means of obtaining it, must have rendered 
possible not one but countless paths towards perfection, paths 
leading not always forwards but sometimes backwards; and only in 
such a way could the infinite variety of existing flowers have 
come into existence. 
My brother Fritz Miiller has sent me the following instances 
of the general principle that whenever in a variable species 
selection occurs in a definite way, then that selection, apart from 
other relations, will lead to a continued increase of the variation 
in the same direction from generation to generation :— 
1 *¢ Vi ansa saledes, att utvecklingen i anordningarna for kénens forening hos de 
fanerogama vixterna fortgatt och fortgar i nimmda riktning.”—No. 17, p. 95. 
