570 THE FERTILISATION OF FLOWERS. [PART Ly. 
of cross- or self-fertilisation, that it will suffice to refer here to a 
few instances only. We have instituted comparisons between 
similar genera of Umbellifer, between species of the genera 
Ranunculus, Geranium, Malva, Polygonum, Cerastium, Epilobium, 
Rubus, Veronica, Carduus, and Hieracium, between different forms 
of Huphrasia officinalis, Rhinanthus crista-galli, Lysimachia vulgaris, 
etc., and in all cases the same result has been obtained ; that, cewteris 
paribus, a species of flower is the more visited by insects the 
more conspicuous it is; and that when closely allied flowers, alike 
in other respects, differ in conspicuousness and also in the degree 
to which cross-fertilisation is insured in case of insect-visits and 
self-fertilisation prevented in their absence, it is always found that 
cross-fertilisation is the more perfectly insured in the more con- 
spicuous and therefore more abundantly visited flowers. And on 
the other hand, under the same conditions, self-fertilisation is best 
insured in those flowers which are least conspicuous, and which 
therefore are least visited by insects and least likely to be cross- 
fertilised. These three statements are fully proved in the case of 
many plants by observations recorded in the foregoing section ; 
the proof rests upon sound facts and direct observations and not 
in any way on speculation or conjecture. 
Let us proceed to examine the last two of these three state- 
ments, which deal more closely with the act of fertilisation. The 
most important deduction to be drawn from them is, that in 
general anthophilous insects are not limited by hereditary instinct 
to certain flowers, but that they wander about getting their food 
on whatever flowers they find it. For if each insect had its own 
species of flower as most caterpillars have their own particular 
food plant, the abundance of insect-visits to the plant would not 
depend at all upon its conspicuousness. 
In isolated cases we may find a particular insect confining its 
visits exclusively or almost exclusively to a particular flower. I 
give the following instances from my own observations. I have 
found Andrena florea only on Bryonia dioica, and A. halictoides 
only in flowers of Campanula: Andrena Hattorfiana confines its 
visits almost entirely to Scabiosa arvensis, Cilissa melanura to 
Lythrum Salicaria, Macropis labiata to Lysimachia vulgaris, Osmia 
adunca and O, cementaria to Echium. But these insects do not form 
1 per cent. of all the species that I have observed, and even 
of these cases the restriction is only complete in two. 
I have already cited numerous instances of insects seeking 
honey in flowers which contain none or which conceal it beyond 
