Colour in Flowers. 
5 1 
of the Gnetum and the primitive Angiospermic nucelli, especially the 
extent of the sporogenous tissue in both, we seem to have grounds 
for regarding Gnetum as nearer the Angiosperm than any other 
phylum. 
In Anatomy we know that Gnetum has characters approximating 
to those of Angiosperms and that its habit (foliage, etc.) is more 
Angiospermic than Coniferous. 
In Drimys and other genera allied to Liriodendron homoxylous 
secondary wood is recorded, so that each phylum seems to approxi¬ 
mate to the other on anatomical grounds. 
The theory of the sudden appearance and rapid differentiation 
of Angiosperms which seems to be the chief outcome so far of their 
geological record would compel us to expect the immediate ancestor 
to have attained considerable vegetative differentiation. 
It is a curious fact that the distichously branching spikes of 
Gnetum latifolium, if telescoped down as suggested above, would 
yield the partial inflorescence chararacteristic of Castanea , the 
basis according to Eichler of that of other Amentiferae. 
Is not then the hypothesis a tenable one that some race which 
in respect of its numerous free sporophylls, micellar structure, 
embryo-sac. development, and vegetative characters bore considerable 
resemblance to Gnetum, finally attained the full characters of the 
Angiospermic embryo-sac, and that with these a new impetus was 
given which has resulted in the dominance of the Angiosperm ? 
M. Benson. 
WHAT PART DOES COLOUR PLAY IN THE ATTRACTION 
OF INSECTS TO FLOWERS? 
O conclusion relating to that most fascinating subject the 
natural history of the flower appears more certain than the 
doctrine of “flag’’-function of the coloured portion (usually the 
corolla) of an entomophilous flower in attracting insects for the 
purpose of carrying out cross-pollination. It seems, indeed, impos¬ 
sible to understand why entomophilous flowers (i.e. flowers which 
can be shewn to be adapted for cross-pollination by insects) should 
in general possess conspicuous coloured corollas if these do not 
perform the function in question. Sprengel, the great pioneer of 
