460 JOURNAL OF SCIENCE. 
won, and the only way that has yet been discovered to win her 
secrets, is to study her. 
Besides roses, the ranunculus family furnish another excellent 
series showing the evolution of petals. The simplest forms such as 
buttercups,- marsh inarigolds, &e., have usually yellow petals, 
although some buttercups show the transition to white, and our 
finest New Zealand species are pure white. More differentiated 
forms show various stages of white, pink and red, while the most 
highly developed forms such as larkspurs, columbines, and monks- 
hood are mostly blue, though every other shade occurs, as indeed we 
might expect. Numerous other instances might be adduced, but 
there is no need to multiply them indefinitely. The whole balance of 
proof is in favour of the theory that petals are derived from stamens. 
In order to form some idea as to how flowers have arrived at the 
varying stages of development they now exhibit, we must go first to 
the oldest and simplest types of flowers known, as well as to some of 
the higher flowerless plants, and consider what we find there. 
Judging from such materials, it seems probable that the earliest 
flowers consisted only of a pistil containing ovules, and of stamens. 
Similar reproductive organs still occur in the simplest Cycads 
(Gymnosperms), and something analagous occurs in certain orders of 
flowerless plants. Probably also fertilization of the ovules took place 
by pollen from the adjacent anthers being shaken by the action of the 
wind on to the stigmas. In order to protect such flowers, special 
whorls of leaves were developed, and to these we may trace the origin 
of ordinary sepals, and in many cases, of bracts. But the real 
differentiation of the flowers must have commenced when insects took 
to visiting them, and from this stage onwards the evolution of flowers 
and of flower-yisiting insects advanced pari passu, until in the present 
day there are many flowers so developed that they depend for their 
fertilization on one particular species of insect. And corresponding 
with this, though the fact is not so strongly vouched for, there are 
certain insects, chiefly butterflies and moths, which depend for their 
supplies of nectar on one particular species of flower. You may ask 
what first led insects to visit these simple and very rudimentary 
flowers of which I have spoken. The answer is that they went to eat 
the pollen. This taste for pollen is still common to very many insects, 
and many flowers, such as Clematis, appear to produce an immense 
amount of pollen in order to attract their insect visitors. Of course 
it would not do if all their pollen was eaten up, but (if I may be 
allowed the expression) they seem to argue that it is better to 
produce 1,000 grains of pollen and have 990 of them eaten while the 
other ten by adhering to the insect’s body are conveyed to the stigma, 
than it is to produce say 100 and either have them all eaten or run. 
the chance of the flower not being visited at all. 
By thus visiting these simple flowers insects unconsciously took 
to fertilizing them, 7.e., they transpgrted some of the pollen from the 
anther-lobes on to the stigmas, and in many cases they would carry 
pollen from one plant to the stigmas of another, thus producing what 
we term cross-fertilization. Now even before Darwin’s researches it 
was known that cross-fertilization was beneficial to plants, and 
Darwin himself says “Andrew Knight many years ago propounded 
the doctrine that no plant self-fertilizes itself for a perpetuity of 
