BOTANICAL EVOLUTION. A59 
are most probably modified stamens. Of course there is uo absolute 
proof that bracts or sepals have not developed into petals in some 
cases, for it is a common enough occurrence for them to become 
brightly coloured or petaloid as we call them, as ¢.g., the fine white 
bract of Calla Hthiopica, and the brightly coloured sepals ef Clematis, 
Anemone and Fuchsia. But in such cases comes in the economical 
tendency of which I spoke in my first lecture. Thus in Calla the 
individual flowers are very small and imperfect, and have neither 
sepals nor petals; in Clematis and Anemone there are no petals, and 
in Fuchsia they are relatively very small, and in one of our New 
Zealand species have completely disappeared. But against this 
negative theory that petals have been developed from sepals or 
bracts, we have the very strong positive evidence that whenever 
flowers either in a state of nature and still more commonly under 
cultivation tend to become double, this process always commences in 
the outer stamens. In these the filament commences first to flatten 
out and the anther-lobes become smaller and more or less imperfect,— 
and the process goes on often until all the stamens become converted 
into petals. There is hardly one of our cultivated garden flowers, of 
those at least of which double varietics are grown, in which you 
cannot see this process in greater or less perfection. A further very 
strong argument is this, that among coloured flowers, those of simplest 
structure have yellow petals, and this is what we should expect if 
petals are derived from stamens, for by far the largest number of 
known flowers have yellow stamens and yellow pollen. But if the 
petals are themselves derived from sepals they would have probably 
assumed any other colour as soon as yellow which is certainly not 
what we see. This is very readily seen if we examine the colours of 
the flowers of certain orders of plants, Thus, as Grant Allen has 
shewn, if we take the Rose family, we find the simplest forms of it, 
represented by the Potentillas or Cinquefoils, to have yellow petals, 
while the more developed forms such as the Strawberry, Brambles, 
Cherries, Apples, and Roses, tend to pass first to white and then to 
pink petals. It is no exception to this rule that some of our most 
highly cultivated garden roses are yellow. There is no stronger 
tendency under conditions favouring variation than for a reversion to 
_@ parental trait or character, and if a horticulturist chooses to seize 
on such a variation and seeks to perpetuate and intensify it, he will 
probably have no difficulty in doing so. But he will find it a more 
difficult matter to produce a flower of a colour not included in the 
series through which the plant has been evolved. Roses have 
probably passed from the yellow to the white and thence to the red 
colours, the two latter of which characterize wild roses, but we have 
yet to see a blue rose. There is no impossibility in the idea of a 
blue variation arising among roses, but there is at the same time a 
strong improbability. And here I may just diverge for a moment to 
say that if gardeners, whether professional or amateur, knew a little 
more about the principles of botanical science, they would not try the 
absurd experiments they sometimes do, such as attempting to obtain 
blue roses by placing pounded charcoal round their roots, and other 
such ignorant devices. Such means are on 4 par with those adopted 
by a small boy, who in his ambition to get a crop of gunpowder, 
sowed his little garden bed with it. Nature does not yield her 
secrets to such hap-hazard work; she must be wooed if she is to be 
