opposite direction? How do you know | taken their origin from stamens, while it

puts it, "obviously adapted to a less æsthetically cultivated circle of visitors." And some livid red flowers actually resemble in their color and odor decaying raw meat, thus inducing bluebottle flies to visit them and so carry their pollen from head to head.

cise such a selective action, the petals must themselves have shown some tendency to vary in certain fixed directions. How could such an original tendency arise? For, of course, if the insects never saw any pink, purple, or blue petals, they could not specially favor and select them; so that we are as yet hardly nearer the solution of the problem than ever.

he says, "are in many cases exactly the same as those in the foliage from which chlorophyll has disappeared; so that the petals are often exactly like leaves which have turned yellow and red in autumn, or the very yellow or red leaves of early spring." "The color of many crimson, pink, and red flowers is due to the devel opment of substances belonging to the erythrophyll group, and not unfrequently to exactly the same kind as that so often found in leaves. The facts seem to indicate that these various substances may be due to an alteration of the normal constituents of leaves. So far as I have been able to ascertain, their development seems as if related to extra oxidization, modified by light and other varying conditions not yet understood."

Down to the minutest distinctions between species, this correlation of flowers to the tastes of their particular guests Here Mr. Sorby, who has chemically seems to hold good. Hermann Müller studied the coloring matter of leaves and notes that the common galium of our flowers far more deeply than any other heaths and hedges is white, and therefore investigator, supplies us with a useful visited by small flies; while the lady's hint. He tells us that the various pigbedstraw, its near relative, is yellow, and ments of bright petals are already conowes its fertilization to little beetles. Mr. tained in the ordinary tissues of the plant, H. O. Forbes counted on one occasion whose juices only need to be slightly the visits he saw paid to the flowers on a modified in chemical constitution in order single bank; and he found that a particu- to make them into the blues, pinks, and lar bumble-bee sucked the honey of thirty purples with which we are so familiar. purple dead-nettles in succession, passing"The colored substances in the petals," over without notice all the other plants in the neighborhood; two other species of bumble-bee and a cabbage butterfly also patronized the same dead-nettles exclusively. Fritz Müller noticed a lantana in South America which changes color as its flowering advances; and he observed that each kind of butterfly which visited it stuck rigidly to its own favorite color, waiting to pay its addresses until that color appeared. Mr. Darwin cut off the petals of a lobelia and found that the hive-bees never went near it, though they were very busy with the surrounding flowers. But perhaps Sir John Lubbock's latest experiments on bees are the most conclusive of all. He had long ago convinced himself, by trials with honey placed on slips of glass above yellow, pink, or blue paper, that bees could discriminate the different colors; and he has now shown in the same way that they display a marked preference for blue over all others. The fact is, blue flowers are, as a rule, specialized for fertilization by bees, and bees therefore prefer this color; while conversely the flowers have at the same time become blue because that was the color which the bees prefer. As in most other cases, the adaptation must have gone on pari passu on both sides. As the bee-flowers grew bluer, the bees must have grown fonder and fonder of blue; and as they grew fonder of blue, they must have more and more constantly preferred the bluest flowers.

We thus see how the special tastes of insects may have become the selective agency for developing white, pink, red, purple, and blue petals from the original yellow ones. But before they could exer

The different hues assumed by petals are all thus, as it were, laid up beforehand in the tissues of the plant, ready to be brought out at a moment's notice. And all flowers, as we know, easily sport a little in color. But the question is, do their changes tend to follow any regular and definite order? Is there any reason to believe that the modification runs from yellow through red to blue, rather than vice verså? I believe there is; and we get hints of it in the following fashion.

One of our common little English forget-me-nots, by name Myosotis versicolor (may I be pardoned for using a few scientific names just this once?) is pale yellow when it first opens; but as it grows older, it becomes faintly pinkish, and ends by being blue like the others of its race. Now, this sort of color-change is by no means uncommon; and in all the cases that I know of it is always in the same.

direction, from yellow or white, through | On the other hand, Mr. Wallace's prin pink, orange, or red, to purple or blue. ciple also explains why the bees and but For example, one of the wall-flower tribe, terflies should prefer these specialized Cheiranthus chamæleo, has at first a whit- colors to all others, and should therefore ish flower, then a citron-yellow, and finally select the flowers which display them by emerges into red or violet. The petals preference over any less developed types. of Stylidium fruticosum are pale yellow For bees and butterflies are the most to begin with, and afterwards become highly adapted of all insects to honeylight rose-colored. An evening primrose, Enothera tetraptera, has white flowers in its first stage and red ones at a later period of development. Cobaa scandens goes from white to violet; Hibiscus mutabilis from white through flesh-colored to red. Fritz Müller's lantana is yellow on its first day, orange on the second, and purple on the third. The whole tribe of borages begin by being pink and end with being blue. The garden convolvulus opens a blushing white and passes into full purple. In all these and many other cases the general direction of the changes is the same. They are usually set down as due to oxidation of the pigmentary

matter.

If this be so, there is a good reason why bees should be specially fond of blue, and why blue flowers should be specially adapted for fertilization by their aid. For Mr. A. R. Wallace has shown that color is most apt to appear or to vary in those parts of plants or animals which have undergone the highest amount of modification. The markings of the peacock and the argus pheasant come out upon their immensely developed secondary tailfeathers or wing-plumes; the metallic hues of sun-birds and humming-birds show themselves upon their highly specialized crests, gorgets, or lappets. It is the same with the hackles of fowls, the head-ornaments of fruit-pigeons, and the bills of toucans. The most exquisite colors in the insect world are those which are developed on the greatly expanded and delicately feathered wings of butterflies; and the eye-spots which adorn a few species are usually found on their very highly modified_swallow-tail appen dages. So, too, with flowers; those which have undergone most modification have their colors most profoundly altered. In this way, we may put it down as a general rule (to be tested hereafter) that the least developed flowers are usually yellow or white; those which have undergone a little more modification are usually pink or red; and those which have been most highly specialized of any are usually purple, lilac, or blue. Absolute deep ultramarine, like that of this harebell, probably marks the highest level of all.

seeking and flower-feeding. They have themselves on their side undergone the largest amount of specialization for that particular function. And if the more specialized and modified flowers, which gradually fitted their forms and the posi tion of their honey-glands to the forms of the bees or butterflies, showed a natural tendency to pass from yellow through pink and red to purple and blue, it would follow that the insects which were being evolved side by side with them, and which were aiding at the same time in their evolution, would grow to recognize these developed colors as the visible symbols of those flowers from which they could obtain the largest amount of honey with the least possible trouble. Thus it would finally result that the ordinary unspecialized flowers, which depended upon small insect riff-raff, would be mostly left yellow or white; those which appealed to rather higher insects would become pink or red; and those which laid themselves out for bees and butterflies, the aristocrats of the arthropodous world, would grow for the most part to be purple or blue.

The

Now, this is very much what we ac tually find to be the case in nature. simplest and earliest flowers are those with regular, symmetrical, open cups, which can be visited by any insects whatsoever; and these are in large part yellow or white. A little higher are the flowers with more or less closed cups, whose honey can only be reached by more specialized insects; and these are oftener pink or reddish. More profoundly modified are those irregular one-sided flowers, which have assumed special shapes to accommodate bees or other specific honey-seekers; and these are often purple and not infrequently blue. Highly specialized in another way are the flowers whose petals have all coalesced into a tubular corolla; and these might almost be said to be usually purple or blue. And, finally, highest of all are the flowers whose tubular corolla has been turned to one side, thus combining the united petals with the irregular shape; and these are almost invariably purple or blue. proceed in the sequel to give examples. One may say that the most profoundly

I shall

modified of all existing flowers are the families of the composites, the labiates, the snapdragons, and the orchids. Now these are exactly the families in which blue and purple flowers are commonest; while in all of them, except the composites, white flowers are rare, and unmixed yellow flowers almost unknown. But perhaps the best way to test the principle will be to look at one or two families in detail, remembering of course that we can only expect approximate results, owing to the natural complexity of the conditions. Not to overburden the subject with unfamiliar names I shall seldom go beyond the limits of our own native English flora.

rosy pink or bright red. Finally, the true roses, whose flowers are the most developed of all, have usually extremely broad pink petals (like those of our own dogrose), which in some still bigger exotic species become crimson or damask of the deepest dye. They are more sought after by insects than any others of their family.

At the same time, the roses as a whole, being a relatively simple family, with regular symmetrical flowers of the separate type, have never risen to the stage of producing blue petals. That is why our florists cannot turn out a blue rose. It is easy enough to make roses or any other blossoms vary within their own natural limits, revert to any earlier form or color through which they have previously passed; but it is difficult or impossible to make them take a step which they have never yet naturally taken. Hence florists generally find the most developed flowers are also the most variable and plastic in color; and hence, too, we can get red, pink, white, straw-colored, or yellow roses, but not blue ones. This, I believe, is the historical truth underlying De Candolle's division of flowers into a xanthic and a cyanic series.

The roses form a most instructive family to begin with. As a whole they are not very highly developed, since all of them have simple, open, symmetrical flowers, generally with five distinct petals. But of all the rose tribe, as I have endeavored to show elsewhere, the potentilla group, including our common English cinquefoils and silver-weed, seem to make up the most central, simple, and primitive members. They are chiefly low, creeping weeds, and their flowers are of the earliest pattern, without any specialization of Still more interesting, because covering form, or any peculiar adaptation to in- a wider range of color, are the buttercup sect visitors. Now among the potentilla family, whose petals vary from yellow to group, nearly all the blossoms are yellow, every shade of crimson, purple, and blue. as are also those of the other early allied Here, the simplest and least differentiated forms, such as agrimony and herb-bennet. members of the group are the common Almost the only white potentillas in En- meadow buttercups, which, as everybody gland are the barren strawberry and the knows, have five open petals of a brilliant true strawberry, which have diverged golden hue. Nowhere else is the exact more than any other species from the accordance in color between stamens and norma of the race. Water-avens, how-petals more noticeable than in these flowever, a close relative of herb-bennet, has ers. There are two kinds of buttercup in a dusky purplish tinge; and Sir John Lubbock notes that it secretes honey, and is far oftener visited by insects than its kinsman. The bramble tribe, including the blackberry, raspberry, and dewberry, have much larger flowers than the potentillas, and are very greatly frequented by winged visitors. Their petals are pure white, often with a pinky tinge, especially on big, well-grown blossoms. But there is one low, little-developed member of the blackberry group, the stone-bramble, with narrow, inconspicuous petals of a greenish yellow, merging into dirty white; and this humble form seems to preserve for us the transitional stage from the yellow potentilla to the true white brambles. One step higher, the cherries, apples, and pears have very large and expanded petals, white toward the centre, but blushing at the edges into

England, however, which show us the transition from yellow to white actually taking place under our very eyes. These are the water crowfoot and its close ally the ivy-leaved crowfoot, whose petals are still faintly yellow toward the centre, but fade away into primrose and white as they approach the edge. The clematis and anemone, which are more highly developed, have white sepals (for the petals here are suppressed), even in our English species; and exotic kinds varying from pink to purple are cultivated in our flower-gardens. Columbines are very specialized forms of the buttercup type, both sepals and petals being brightly colored, while the former organs are produced above into long, bow-shaped spurs, each of which secretes a drop of honey; and various columbines accordingly range from red to purple and dark blue. Even

the columbine, however, though so highly | here is that of the pea-blossoms. These specialized, is not bilaterally but circularly are all bilateral in shape, as everybody symmetrical. This last and highest mode of adaptation to insect visits is found in larkspur, and still more developed in the curious monkshood. Now larkspur is usually blue, though white or red blossoms sometimes occur by reversion; while monkshood is one of the deepest blue flowers we possess. Sir John Lubbock has shown that a particular bumble-bee (Bombus hortorum), is the only north European insect capable of fertilizing the larkspur.

The violets are a whole family of bilateral flowers, highly adapted to fertilization by insects, and as a rule they are blue. Here, too, however, white varieties easily arise by reversion; while one member of the group, the common pansy, is perhaps the most variable flower in all nature.

knows; but the lower and smaller species, such as the medick, lotus, and lady's fingers, are usually yellow. So also are broom and gorse. Among the more specialized clovers, some of which are fertilized by bees alone, white, red, and purple predominate. Even with the smaller and earlier types, the most developed species, like lucerne, are likewise purple. But in the largest and most advanced types, the peas, beans, vetches, and scarlet runners, we get much brighter and deeper colors, often with more or less tinge of blue. In the sweet peas and many others, the standard frequently dif fers in hue from the keel or the wings a still further advance in heterogeneity of coloration. Lupines, sainfoin, everlasting pea, and wistaria are highly evolved members of the same family, in which purple, lilac, mauve, or blue tints become distinctly pronounced.

blue, so rare elsewhere, become almost the rule. For example, in the great family of the heaths, which is highly adapted to insect fertilization, more particularly by bees, purple and blue are the prevailing tints, so much so that, as we all have noticed a hundred times over, they often color whole tracts of hillside together. So far as I know, there are no really yellow heaths at all. The bell-shaped blossoms mark at once the position of the heaths with reference to insects; and the order, according to Mr. Bentham, supplies us with more ornamental plants than any other in the whole world.

Pinks do not display so wide a range in either direction. They begin as high up as white, and never get any higher than red or carnation. The small, undevel- When we pass on, however, to the flowoped field species, such as the chick- ers in which (as in this harebell) the petals weeds, stitchworts, and corn-spurries, have have all coalesced into a tubular or camopen flowers of very primitive character, panulate corolla, we get even more strikand almost all of them are white. They ing results. Here, where the very shape are fertilized by miscellaneous small flies. at once betokens high modification, yellow But the campions and true pinks have a is a comparatively rare color (especially tubular calyx, and the petals are raised on as a ground-tone, though it often comes. long claws, while most of them also dis-out in spots or patches), while purple and play special adaptations for a better class of insect fertilization in the way of fringes or crowns on the petals. These higher kinds are generally pink or red. Our own beautiful purple English corn-cockle is a highly developed campion, so specialized that only butterflies can reach its honey with their long tongues, as the nectaries are situated at the bottom of the tube. Two other species of campion, however, show us interestingly the way in which variations of color may occur in a retrograde direction even among highly evolved forms. One of them, the day lychnis, has red, scentless flowers, opening in the morning, and it is chiefly fertilized by diurnal butterflies. But its descendant, the night lychnis, has taken to fertilization by means of moths; and as moths can only see white flowers, it has become white, and has acquired a faint perfume as an extra attraction. Still, the change has not yet become fully organized in the species, for one may often find a night lychnis at the present time which is only pale pink, instead of being pure white.

It is the same with the families allied to my harebell here. They are, in fact, for the most part larger and handsomer blossoms of the same type as the heaths; and the greater number of them, like the harebell itself and the Canterbury bell, are deep blue. Rampion and sheep's bit, also blue, are clustered heads of similar blossoms. The little blue lobelia of our borders, which is bilateral as well as tubular, belongs to a closely related tribe. Not far from them are the lilac scabious, the The only other family of flowers with blue devil's bit, and the mauve teasel. separate petals which I shall consider | Amongst all these very highly evolved