The great difference in the fertility of the long and short-styled flowers, when both are illegitimately fertilised, is a unique case, as far as I have observed with heterostyled plants. The long-styled flowers when thus fertilised are utterly barren, whilst about half of the short-styled ones produce capsules, and these include a little above two-thirds of the number of seeds yielded by them when legitimately fertilised. The sterility of the illegitimately fertilised long-styled flowers is probably increased by the deteriorated condition of their pollen; nevertheless this pollen was highly efficient when applied to the stigmas of the short-styled flowers. With several species of Primula the short- styled flowers are much more sterile than the long-styled, when both are illegitimately fertilised; and it is a tempting view, as formerly remarked, that this greater sterility of the short-styled flowers is a special adaptation to check self-fertilisation, as their stigmas are eminently liable to receive their own pollen. This view is even still more tempting in the case of the long-styled form of Linum grandiflorum. On the other hand, with Pulmonaria angustifolia, it is evident, from the corolla projecting obliquely upwards, that pollen is much more likely to fall on, or to be carried by insects down to the stigma of the short-styled than of the long-styled flowers; yet the short-styled instead of being more sterile, as a protection against self-fertilisation, are far more fertile than the long-styled, when both are illegitimately fertilised.

Pulmonaria azurea, according to Hildebrand, is not heterostyled. (3/12. 'Die Geschlechter-Vertheilung bei den Pflanzen' 1867 page 37.)

[From an examination of dried flowers of Amsinckia spectabilis, sent me by Professor Asa Gray, I formerly thought that this plant, a member of the Boragineae, was heterostyled. The pistil varies to an extraordinary degree in length, being in some specimens twice as long as in others, and the point of insertion of the stamens likewise varies. But on raising many plants from seed, I soon became convinced that the whole case was one of mere variability. The first-formed flowers are apt to have stamens somewhat arrested in development, with very little pollen in their anthers; and in such flowers the stigma projects above the anthers, whilst generally it stands below and sometimes on a level with them. I could detect no difference in the size of the pollen-grain or in the structure of the stigma in the plants which differed most in the above respects; and all of them, when protected from the access of insects, yielded plenty of seeds. Again, from statements made by Vaucher, and from a hasty inspection, I thought at first that the allied Anchusa arvensis and Echium vulgare were heterostyled, but soon saw my error. From information given me, I examined dried flowers of another member of the Boragineae, Arnebia hispidissima, collected from several sites, and though the corolla, together with the included organs, differed much in length, there was no sign of heterostylism.]

Polygonum fagopyrum (Polygonaceae).

(FIGURE 3.7. Polygonum fagopyrum. (From H. Muller.) Upper figure, the long-styled form; lower figure, the short-styled. Some of the anthers have dehisced, others have not.)

Hildebrand has shown that this plant, the common Buck-wheat, is heterostyled. (3/13. 'Die Geschlechter-Vertheilung' etc. 1867 page 34.) In the long-styled form (Figure 3.7), the three stigmas project considerably above the eight short stamens, and stand on a level with the anthers of the eight long stamens in the short-styled form; and so it is conversely with the stigmas and stamens of this latter form. I could perceive no difference in the structure of the stigmas in the two forms. The pollen-grains of the short-styled form are to those of the long-styled as 100 to 82 in diameter. This plant is therefore without doubt heterostyled.

I experimented only in an imperfect manner on the relative fertility of the two forms. Short-styled flowers were dragged several times over two heads of flowers on long-styled plants, protected under a net, which were thus legitimately, though not fully, fertilised. They produced 22 seeds, or 11 per flower-head.

Three flower-heads on long-styled plants received pollen in the same manner from other long-styled plants, and were thus illegitimately fertilised. They produced 14 seeds, or only 4.66 per flower-head.

Two flower-heads on short-styled plants received pollen in like manner from long-styled flowers, and were thus legitimately fertilised. They produced 8 seeds, or 4 per flower-head.

Four heads on short-styled plants similarly received pollen from other short- styled plants, and were thus illegitimately fertilised. They produced 9 seeds, or 2.25 per flower-head.

The results from fertilising the flower-heads in the above imperfect manner cannot be fully trusted; but I may state that the four legitimately fertilised flower-heads yielded on an average 7.50 seeds per head; whereas the seven illegitimately fertilised heads yielded less than half the number, or on an average only 3.28 seeds. The legitimately crossed seeds from the long-styled flowers were finer than those from the illegitimately fertilised flowers on the same plants, in the ratio of 100 to 82, as shown by the weights of an equal number.

About a dozen plants, including both forms, were protected under nets, and early in the season they produced spontaneously hardly any seeds, though at this period the artificially fertilised flowers produced an abundance; but it is a remarkable fact that later in the season, during September, both forms became highly self-fertile. They did not, however, produce so many seeds as some neighbouring uncovered plants which were visited by insects. Therefore the flowers of neither form when left to fertilise themselves late in the season without the aid of insects, are nearly so sterile as most other heterostyled plants. A large number of insects, namely 41 kinds as observed by H. Muller, visit the flowers for the sake of the eight drops of nectar. (3/14. 'Die Befruchtung' etc. page 175 and 'Nature' January 1, 1874 page 166.) He infers from the structure of the flowers that insects would be apt to fertilise them both illegitimately as well as legitimately; but he is mistaken in supposing that the long-styled flowers cannot spontaneously fertilise themselves.

Differently to what occurs in the other genera hitherto noticed, Polygonum, though a very large genus, contains, as far as is at present known, only a single heterostyled species, namely the present one. H. Muller in his interesting description of several other species shows that P. bistorta is so strongly proterandrous (the anthers generally falling off before the stigmas are mature) that the flowers must be cross-fertilised by the many insects which visit them. Other species bear much less conspicuous flowers which secrete little or no nectar, and consequently are rarely visited by insects; these are adapted for self-fertilisation, though still capable of cross-fertilisation. According to Delpino, the Polygonaceae are generally fertilised by the wind, instead of by insects as in the present genus.

[Leucosmia Burnettiana (Thymeliae).

As Professor Asa Gray has expressed his belief that this species and L. acuminata, as well as some species in the allied genus Drymispermum, are dimorphic or heterostyled (3/15. 'American Journal of Science' 1865 page 101 and Seemann's 'Journal of Botany' volume 3 1865 page 305.), I procured from Kew, through the kindness of Dr. Hooker, two dried flowers of the former species, an inhabitant of the Friendly Islands in the Pacific. The pistil of the long-styled form is to that of the short-styled as 100 to 86 in length; the stigma projects just above the throat of the corolla, and is surrounded by five anthers, the tips of which reach up almost to its base; and lower down, within the tubular corolla, five other and rather smaller anthers are seated. In the short-styled form, the stigma stands some way down the tube of the corolla, nearly on a level with the lower anthers of the other form: it differs remarkably from the stigma of the long-styled form, in being more papillose, and in being longer in the ratio of 100 to 60. The anthers of the upper stamens in the short-styled form are supported on free filaments, and project above the throat of the corolla, whilst the anthers of the lower stamens are seated in the throat on a level with the upper stamens of the other form. The diameters of a considerable number of grains from both sets of anthers in both forms were measured, but they did not differ in any trustworthy degree. The mean diameter of twenty-two grains from the short-styled flower was to that of twenty-four grains from the long-styled, as 100 to 99. The anthers of the upper stamens in the short-styled form appeared to be poorly developed, and contained a considerable number of shrivelled grains which were omitted in striking the above average. Notwithstanding the fact of the pollen-grains from the two forms not differing in diameter in any appreciable degree, there can hardly be a doubt from the great difference in the two forms in the length of the pistil, and especially of the stigma, together with its more papillose condition in the short-styled form, that the present species is truly heterostyled. This case resembles that of Linum grandiflorum, in which the sole difference between the two forms consists in the length of the pistils and stigmas. From the great length of the tubular corolla of Leucosmia, it is clear that the flowers are cross-fertilised by large Lepidoptera or by honey-sucking birds, and the position of the stamens in two whorls one beneath the other, which is a character that I have not seen in any other heterostyled dimorphic plant, probably serves to smear the inserted organ thoroughly with pollen.

Menyanthes trifoliata (Gentianeae).

This plant inhabits marshes: my son William gathered 247 flowers from so many distinct plants, and of these 110 were long-styled, and 137 short-styled. The pistil of the long-styled form is in length to that of the short-styled in the ratio of about 3 to 2. The stigma of the former, as my son observed, is decidedly larger than that of the short-styled; but in both forms it varies much in size. The stamens of the short-styled are almost double the length of those of the long-styled; so that their anthers stand rather above the level of the stigma of the long-styled form. The anthers also vary much in size, but seem often to be of larger size in the short-styled flowers. My son made with the camera many drawings of the pollen-grains, and those from the short-styled flowers were in diameter in nearly the ratio of 100 to 84 to those from the long-styled flowers. I know nothing about the capacity for fertilisation in the two forms; but short-styled plants, living by themselves in the gardens at Kew, have produced an abundance of capsules, yet the seeds have never germinated; and this looks as if the short-styled form was sterile with its own pollen.

Limnanthemum Indicum (Gentianeae).

This plant is mentioned by Mr. Thwaites in his Enumeration of the Plants of Ceylon as presenting two forms; and he was so kind as to send me specimens preserved in spirits. The pistil of the long-styled form is nearly thrice as long (i.e. as 14 to 5) as that of the short-styled, and is very much thinner in the ratio of about 3 to 5. The foliaceous stigma is more expanded, and twice as large as that of the short-styled form. In the latter the stamens are about twice as long as those of the long-styled, and their anthers are larger in the ratio of 100 to 70. The pollen-grains, after having been long kept in spirits, were of the same shape and size in both forms. The ovules, according to Mr. Thwaites, are equally numerous (namely from 70 to 80) in the two forms.

Villarsia [sp.?] (Gentianeae).

Fritz Muller sent me from South Brazil dried flowers of this aquatic plant, which is closely allied to Limnanthemum. In the long-styled form the stigma stands some way above the anthers, and the whole pistil, together with the ovary, is in length to that of the short-styled form as about 3 to 2. In the latter form the anthers stand above the stigma, and the style is very short and thick; but the pistil varies a good deal in length, the stigma being either on a level with the tips of the sepals or considerably beneath them. The foliaceous stigma in the long-styled form is larger, with the expansions running farther down the style, than in the other form. One of the most remarkable differences between the two forms is that the anthers of the longer stamens in the short- styled flowers are conspicuously longer than those of the shorter stamens in the long-styled flowers. In the former the sub-triangular pollen-grains are larger; the ratio between their breadth (measured from one angle to the middle of the opposite side) and that of the grains from the long-styled flowers being about 100 to 75. Fritz Muller also informs me that the pollen of the short-styled flowers has a bluish tint, whilst that of the long-styled is yellow. When we treat of Lythrum salicaria we shall find a strongly marked contrast in the colour of the pollen in two of the forms.

The three genera, Menyanthes, Limnanthemum, and Villarsia, now described, constitute a well-marked sub-tribe of the Gentianeae. All the species, as far as at present known, are heterostyled, and all inhabit aquatic or sub-aquatic stations.

Forsythia suspensa (Oleaceae).

Professor Asa Gray states that the plants of this species growing in the Botanic Gardens at Cambridge, U.S., are short-styled, but that Siebold and Zuccarini describe the long-styled form, and give figures of two forms; so that there can be little doubt, as he remarks, about the plant being dimorphic. (3/16. 'The American Naturalist' July 1873 page 422.) I therefore applied to Dr. Hooker, who sent me a dried flower from Japan, another from China, and another from the Botanic Gardens at Kew. The first proved to be long-styled, and the other two short-styled. In the long-styled form, the pistil is in length to that of the short-styled as 100 to 38, the lobes of the stigma being a little longer (as 10 to 9), but narrower and less divergent. This last character, however, may be only a temporary one. There seems to be no difference in the papillose condition of the two stigmas. In the short-styled form, the stamens are in length to those of the long-styled as 100 to 66, but the anthers are shorter in the ratio of 87 to 100; and this is unusual, for when there is any difference in size between the anthers of the two forms, those from the longer stamens of the short-styled are generally the longest. The pollen-grains from the short-styled flowers are certainly larger, but only in a slight degree, than those from the long-styled, namely, as 100 to 94 in diameter. The short-styled form, which grows in the Gardens at Kew, has never there produced fruit.

Forsythia viridissima appears likewise to be heterostyled; for Professor Asa Gray says that although the long-styled form alone grows in the gardens at Cambridge, U.S., the published figures of this species belong to the short- styled form.

Cordia [sp.?] (Cordiaceae).

Fritz Muller sent me dried specimens of this shrub, which he believes to be heterostyled; and I have not much doubt that this is the case, though the usual characteristic differences are not well pronounced in the two forms. Linum grandiflorum shows us that a plant may be heterostyled in function in the highest degree, and yet the two forms may have stamens of equal length, and pollen-grains of equal size. In the present species of Cordia, the stamens of both forms are of nearly equal length, those of the short-styled being rather the longest; and the anthers of both are seated in the mouth of the corolla. Nor could I detect any difference in the size of the pollen-grains, when dry or after being soaked in water. The stigmas of the long-styled form stand clear above the anthers, and the whole pistil is longer than that of the short-styled, in about the ratio of 3 to 2.

The stigmas of the short-styled form are seated beneath the anthers, and they are considerably shorter than those of the long-styled form. This latter difference is the most important one of any between the two forms.

Gilia (Ipomopsis) pulchella vel aggregata (Polemoniaceae).

Professor Asa Gray remarks with respect to this plant: "the tendency to dimorphism, of which there are traces, or perhaps rather incipient manifestations in various portions of the genus, is most marked in G. aggregata." (3/17. 'Proceedings of the American Academy of Arts and Sciences.' June 14, 1870 page 275.) He sent me some dried flowers, and I procured others from Kew. They differ greatly in size, some being nearly twice as long as others (namely as 30 to 17), so that it was not possible to compare, except by calculation, the absolute length of the organs from different plants. Moreover, the relative position of the stigmas and anthers is variable: in some long- styled flowers the stigmas and anthers were exserted only just beyond the throat of the corolla; whilst in others they were exserted as much as 4/10 of an inch. I suspect also that the pistil goes on growing for some time after the anthers have dehisced. Nevertheless it is possible to class the flowers under two forms. In some of the long-styled, the length of pistil to that of the short-styled was as 100 to 82; but this result was gained by reducing the size of the corollas to the same scale. In another pair of flowers the difference in length between the pistils of the two forms was certainly greater, but they were not actually measured. In the short-styled flowers whether large or small, the stigma is seated low down within the tube of the corolla. The papillae on the long-styled stigma are longer than those on the short-styled, in the ratio of 100 to 40. The filaments in some of the short-styled flowers were, to those of the long-styled, as 100 to 25 in length, the free, or unattached portion being alone measured; but this ratio cannot be trusted, owing to the great variability of the stamens. The mean diameter of eleven pollen-grains from long-styled flowers, and of twelve from the short-styled, was exactly the same. It follows from these several statements, that the difference in length and state of surface of the stigmas in the flowers is the sole reliable evidence that this species is heterostyled; for it would be rash to trust to the difference in the length of the pistils, seeing how variable they are. I should have left the case altogether doubtful, had it not been for the observations on the following species; and these leave little doubt on my mind that the present plant is truly heterostyled. Professor Gray informs me that in another species, G. coronopifolia, belonging to the same section of the genus, he can see no sign of dimorphism.

Gilia (Leptosiphon) micrantha.

A few flowers sent me from Kew had been somewhat injured, so that I cannot say anything positively with respect to the position and relative length of the organs in the two forms. But their stigmas differed almost exactly in the same manner as in the last species; the papillae on the long-styled stigma being longer than those on the short-styled, in the ratio of 100 to 42. My son measured nine pollen-grains from the long-styled, and the same number from the short-styled form; and the mean diameter of the former was to that of the latter as 100 to 81. Considering this difference, as well as that between the stigmas of the two forms, there can be no doubt that this species is heterostyled. So probably is Gilia nudicaulis, which likewise belongs to the Leptosiphon section of the genus, for I hear from Professor Asa Gray that in some individuals the style is very long, with the stigma more or less exserted, whilst in others it is deeply included within the tube; the anthers being always seated in the throat of the corolla.

Phlox subulata (Polemoniaceae).

Professor Asa Gray informs me that the greater number of the species in this genus have a long pistil, with the stigma more or less exserted; whilst several other species, especially the annuals, have a short pistil seated low down within the tube of the corolla. In all the species the anthers are arranged one below the other, the uppermost just protruding from the throat of the corolla. In Phlox subulata alone he has "seen both long and short styles; and here the short-styled plant has (irrespective of this character) been described as a distinct species (P. nivalis, P. Hentzii), and is apt to have a pair of ovules in each cell, while the long-styled P. subulata rarely shows more than one." (3/18. 'Proceedings of the American Academy of Arts and Sciences' June 14, 1870 page 248.) Some dried flowers of both forms were sent me by him, and I received others from Kew, but I have failed to make out whether the species is heterostyled. In two flowers of nearly equal size, the pistil of the long-styled form was twice as long as that of the short-styled; but in other cases the difference was not nearly so great. The stigma of the long-styled pistil stands nearly in the throat of the corolla; whilst in the short-styled it is placed low down—sometimes very low down in the tube, for it varies greatly in position. The stigma is more papillose, and of greater length (in one instance in the ratio of 100 to 67), in the short-styled flowers than in the long-styled. My son measured twenty pollen-grains from a short-styled flower, and nine from a long- styled, and the former were in diameter to the latter as 100 to 93; and this difference accords with the belief that the plant is heterostyled. But the grains from the short-styled varied much in diameter. He afterwards measured ten grains from a distinct long-styled flower, and ten from another plant of the same form, and these grains differed in diameter in the ratio of 100 to 90. The mean diameter of these two lots of twenty grains was to that of twelve grains from another short-styled flower as 100 to 75: here, then, the grains from the short-styled form were considerably smaller than those from the long-styled, which is the reverse of what occurred in the former instance, and of what is the general rule with heterostyled plants. The whole case is perplexing in the highest degree, and will not be understood until experiments are tried on living plants. The greater length, and more papillose condition of the stigma in the short-styled than in the long-styled flowers, looks as if the plant was heterostyled; for we know that with some species—for instance, Leucosmia and certain Rubiaceae—the stigma is longer and more papillose in the short-styled form, though the reverse of this holds good in Gilia, a member of the same family with Phlox. The similar position of the anthers in the two forms is somewhat opposed to the present species being heterostyled; as is the great difference in the length of the pistil in several short-styled flowers. But the extraordinary variability in diameter of the pollen-grains, and the fact that in one set of flowers the grains from the long-styled flowers were larger than those from the short-styled, is strongly opposed to the belief that Phlox subulata is heterostyled. Possibly this species was once heterostyled, but is now becoming sub-dioecious; the short-styled plants having been rendered more feminine in nature. This would account for their ovaries usually containing more ovules, and for the variable condition of their pollen-grains. Whether the long- styled plants are now changing their nature, as would appear to be the case from the variability of their pollen-grains, and are becoming more masculine, I will not pretend to conjecture; they might remain as hermaphrodites, for the coexistence of hermaphrodite and female plants of the same species is by no means a rare event.

Erythroxylum [sp.?] (Erythroxylidae).

(FIGURE 3.8. Erythroxylon [sp.?] Left: Long-styled form. Right: Short-styled form. From a sketch by Fritz Muller, magnified five times.)

Fritz Muller sent me from South Brazil dried flowers of this tree, together with the drawings (Figure 3.8.), which show the two forms, magnified about five times, with the petals removed. In the long-styled form the stigmas project above the anthers, and the styles are nearly twice as long as those of the short-styled form, in which the stigmas stand beneath the anthers. The stigmas in many, but not in all the short-styled flowers are larger than those in the long-styled. The anthers of the short-styled flowers stand on a level with the stigmas of the other form; but the stamens are longer by only one-fourth or one- fifth of their own length than those of the long-styled. Consequently the anthers of the latter do not stand on a level with, but rather above the stigmas of the other form. Differently from what occurs in the following closely allied genus, Sethia, the stamens are of nearly equal length in the flowers of the same form. The pollen-grains of the short-styled flowers, measured in their dry state, are a little larger than those from the long-styled flowers in about the ratio of 100 to 93. (3/19. F. Muller remarks in his letter to me that the flowers, of which he carefully examined many specimens, are curiously variable in the number of their parts: 5 sepals and petals, 10 stamens and 3 pistils are the prevailing numbers; but the sepals and petals often vary from 5 to 7; the stamens from 10 to 14, and the pistils from 3 to 4.)

Sethia acuminata (Erythroxylidae).

Mr. Thwaites pointed out several years ago that this plant exists under two forms, which he designated as forma stylosa et staminea; and the flowers sent to me by him are clearly heterostyled. (3/20. 'Enumeratio Plantarum Zeylaniae' 1864 page 54.) In the long-styled form the pistil is nearly twice as long, and the stamens half as long as the corresponding organs in the short-styled form. The stigmas of the long-styled seem rather smaller than those of the short-styled. All the stamens in the short-styled flowers are of nearly equal length, whereas in long-styled they differ in length, being alternately a little longer and shorter; and this difference in the stamens of the two forms is probably related, as we shall hereafter see in the case of the short-styled flowers of Lythrum salicaria, to the manner in which insects can best transport pollen from the long-styled flowers to the stigmas of the short-styled. The pollen-grains from the short-styled flowers, though variable in size, are to those of the long-styled, as far as I could make out, as 100 to 83 in their longer diameter. Sethia obtusifolia is heterostyled like S. acuminata.

Cratoxylon formosum (Hypericineae).

Mr. Thiselton Dyer remarks that this tree, an inhabitant of Malacca and Borneo, appears to be heterostyled. (3/21. 'Journal of Botany' London 1872 page 26.) He sent me dried flowers, and the difference between the two forms is conspicuous. In the short-styled form the pistils are in length to those of the short-styled as 100 to 40, with their globular stigmas about twice as thick. These stand just above the numerous anthers and a little beneath the tips of the petals. In the short-styled form the anthers project high above the pistils, the stigmas of which diverge between the three bundles of stamens, and stand only a little above the tips of the sepals. The stamens in this form are to those of the long- styled as 100 to 86 in length; and therefore they do not differ so much in length as do the pistils. Ten pollen-grains from each form were measured, and those from the short-styled were to those from the long-styled as 100 to 86 in diameter. This plant, therefore, is in all respects a well-characterised heterostyled species.

Aegiphila elata (Verbenaceae).

Mr. Bentham was so kind as to send me dried flowers of this species and of Ae. mollis, both inhabitants of South America. The two forms differ conspicuously, as the deeply bifid stigma of the one, and the anthers of the other project far above the mouth of the corolla. In the long-styled form of the present species, the style is twice and a half as long as that of the short-styled. The divergent stigmas of the two forms do not differ much in length, nor as far as I could perceive in their papillae. In the long-styled flowers the filaments adhere to the corolla close up to the anthers, which are enclosed some way down within the tube. In the short-styled flowers the filaments are free above the point where the anthers are seated in the other form, and they project from the corolla to an equal height with that of the stigmas in the long-styled flowers. It is often difficult to measure with accuracy pollen-grains, which have long been dried and then soaked in water; but they here manifestly differed greatly in size. Those from the short-styled flowers were to those from the long-styled in diameter in about the ratio of 100 to 62. The two forms of Ae. mollis present a like difference in the length of their pistils and stamens.

Aegiphila obdurata.

Flowers of this bush were sent me from St. Catharina in Brazil, by Fritz Muller, and were named for me at Kew. They appeared at first sight grandly heterostyled, as the stigma of the long-styled form projects far out of the corolla, whilst the anthers are seated halfway down within the tube; whereas in the short-styled form the anthers project from the corolla and the stigma is enclosed in the tube at nearly the same level with the anthers of the other form. The pistil of the long-styled is to that of the short-styled as 100 to 60 in length, and the stigmas, taken by themselves, as 100 to 55. Nevertheless, this plant cannot be heterostyled. The anthers in the long-styled form are brown, tough, and fleshy, and less than half the length of those in the short-styled form, strictly as 44 to 100; and what is much more important, they were in a rudimentary condition in the two flowers examined by me, and did not contain a single grain of pollen. In the short-styled form, the divided stigma, which as we have seen is much shortened, is thicker and more fleshy than the stigma of the long-styled, and is covered with small irregular projections, formed of rather large cells. It had the appearance of having suffered from hyperthrophy, and is probably incapable of fertilisation. If this be so the plant is dioecious, and judging from the two species previously described, it probably was once heterostyled, and has since been rendered dioecious by the pistil in the one form, and the stamens in the other having become functionless and reduced in size. It is, however, possible that the flowers may be in the same state as those of the common thyme and of several other Labiatae, in which females and hermaphrodites regularly co-exist. Fritz Muller, who thought that the present plant was heterostyled, as I did at first, informs me that he found bushes in several places growing quite isolated, and that these were completely sterile; whilst two plants growing close together were covered with fruit. This fact agrees better with the belief that the species is dioecious than that it consists of hermaphrodites and females; for if any one of the isolated plants had been an hermaphrodite, it would probably have produced some fruit.]

RUBIACEAE.

This great natural family contains a much larger number of heterostyled genera than any other one, as yet known.

Mitchella repens.

Professor Asa Gray sent me several living plants collected when out of flower, and nearly half of these proved long-styled, and the other half short-styled. The white flowers, which are fragrant and which secrete plenty of nectar, always grow in pairs with their ovaries united, so that the two together produce "a berry-like double drupe." (3/22. A. Gray 'Manual of the Botany of the United States' 1856 page 172.) In my first series of experiments (1864) I did not suppose that this curious arrangement of the flowers would have any influence on their fertility; and in several instances only one of the two flowers in a pair was fertilised; and a large proportion or all of these failed to produce berries. In the ensuing year both flowers of each pair were invariably fertilised in the same manner; and the latter experiments alone serve to show the proportion of flowers which yield berries, when legitimately and illegitimately fertilised; but for calculating the average number of seeds per berry I have used those produced during both seasons.

In the long-styled flowers the stigma projects just above the bearded throat of the corolla, and the anthers are seated some way down the tube. In the short- styled flowers those organs occupy reversed positions. In this latter form the fresh pollen-grains are a little larger and more opaque than those of the long- styled form. The results of my experiments are given in Table 3.21.

TABLE 3.21. Mitchella repens.

Column 1: Nature of the Union. Column 2: Number of Pairs of Flowers fertilised during the second season. Column 3: Number of Drupes produced during the second season. Column 4: Average Number of good Seeds per Drupe in all the Drupes during the two Seasons.

Long-styled by pollen of short-styled. Legitimate union : 9 : 8 : 4.6.

Long-styled by own-form pollen. Illegitimate union : 8 : 3 : 2.2.

Short-styled by pollen of long-styled. Legitimate union: 8 : 7 : 4.1.

Short-styled by own-form pollen. Illegitimate union : 9 : 0 : 2.0.

The two legitimate unions together : 17 : 15 : 4.4.

The two illegitimate unions together : 17 : 3 : 2.1.

It follows from this table that 88 per cent of the paired flowers of both forms, when legitimately fertilised, yielded double berries, nineteen of which contained on an average 4.4 seeds, with a maximum in one of 8 seeds. Of the illegitimately fertilised paired flowers only 18 per cent yielded berries, six of which contained on an average only 2.1 seeds, with a maximum in one of 4 seeds. Thus the two legitimate unions are more fertile than the two illegitimate, according to the proportion of flowers which yielded berries, in the ratio of 100 to 20; and according to the average number of contained seeds as 100 to 47.

Three long-styled and three short-styled plants were protected under separate nets, and they produced altogether only 8 berries, containing on an average only 1.5 seed. Some additional berries were produced which contained no seeds. The plants thus treated were therefore excessively sterile, and their slight degree of fertility may be attributed in part to the action of the many individuals of Thrips which haunted the flowers. Mr. J. Scott informs me that a single plant (probably a long-styled one), growing in the Botanic Gardens at Edinburgh, which no doubt was freely visited by insects, produced plenty of berries, but how many of them contained seeds was not observed.

Borreria, nov. sp. near valerianoides (Rubiaceae).

Fritz Muller sent me seeds of this plant, which is extremely abundant in St. Catharina, in South Brazil; and ten plants were raised, consisting of five long- styled and five short-styled. The pistil of the long-styled flowers projects just beyond the mouth of the corolla, and is thrice as long as that of the short-styled, and the divergent stigmas are likewise rather larger. The anthers in the long-styled form stand low down within the corolla, and are quite hidden. In the short-styled flowers the anthers project just above the mouth of the corolla, and the stigma stands low down within the tube. Considering the great difference in the length of the pistils in the two forms, it is remarkable that the pollen-grains differ very little in size, and Fritz Muller was struck with the same fact. In a dry state the grains from the short-styled flowers could just be perceived to be larger than those from the long-styled, and when both were swollen by immersion in water, the former were to the latter in diameter in the ratio of 100 to 92. In the long-styled flowers beaded hairs almost fill up the mouth of the corolla and project above it; they therefore stand above the anthers and beneath the stigma. In the short-styled flowers a similar brush of hairs is situated low down within the tubular corolla, above the stigma and beneath the anthers. The presence of these beaded hairs in both forms, though occupying such different positions, shows that they are probably of considerable functional importance. They would serve to guard the stigma of each form from its own pollen; but in accordance with Professor Kerner's view their chief use probably is to prevent the copious nectar being stolen by small crawling insects, which could not render any service to the species by carrying pollen from one form to the other. (3/23. 'Die Schutzmittel der Bluthen gegen unberufene Gaste' 1876 page 37.)

The flowers are so small and so crowded together that I was not willing to expend time in fertilising them separately; but I dragged repeatedly heads of short-styled flowers over three long-styled flower-heads, which were thus legitimately fertilised; and they produced many dozen fruits, each containing two good seeds. I fertilised in the same manner three heads on the same long- styled plant with pollen from another long-styled plant, so that these were fertilised illegitimately, and they did not yield a single seed. Nor did this plant, which was of course protected by a net, bear spontaneously any seeds. Nevertheless another long-styled plant, which was carefully protected, produced spontaneously a very few seeds; so that the long-styled form is not always quite sterile with its own pollen.

Faramea [sp.?] (Rubiaceae).

(FIGURE 3.9. Faramea [sp.?] Left: Short-styled form. Right: Long-styled form. Outlines of flowers from dried specimens. Pollen-grains magnified 180 times, by Fritz Muller.)

Fritz Muller has fully described the two forms of this remarkable plant, an inhabitant of South Brazil. (3/24. 'Botanische Zeitung' September 10, 1869 page 606.) In the long-styled form the pistil projects above the corolla, and is almost exactly twice as long as that of the short-styled, which is included within the tube. The former is divided into two rather short and broad stigmas, whilst the short-styled pistil is divided into two long, thin, sometimes much curled stigmas. The stamens of each form correspond in height or length with the pistils of the other form. The anthers of the short-styled form are a little larger than those of the long-styled; and their pollen-grains are to those of the other form as 100 to 67 in diameter. But the pollen-grains of the two forms differ in a much more remarkable manner, of which no other instance is known; those from the short-styled flowers being covered with sharp points; the smaller ones from the long-styled being quite smooth. Fritz Muller remarks that this difference between the pollen-grains of the two forms is evidently of service to the plant; for the grains from the projecting stamens of the short-styled form, if smooth, would have been liable to be blown away by the wind, and would thus have been lost; but the little points on their surfaces cause them to cohere, and at the same time favour their adhesion to the hairy bodies of insects, which merely brush against the anthers of these stamens whilst visiting the flowers. On the other hand, the smooth grains of the long-styled flowers are safely included within the tube of the corolla, so that they cannot be blown away, but are almost sure to adhere to the proboscis of an entering insect, which is necessarily pressed close against the enclosed anthers.

It may be remembered that in the long-styled form of Linum perenne each separate stigma rotates on its own axis, when the flower is mature, so as to turn its papillose surface outwards. There can be no doubt that this movement, which is confined to the long-styled form, is effected in order that the proper surface of the stigma should receive pollen brought by insects from the other form. Now with Faramea, as Fritz Muller shows, it is the stamens which rotate on their axes in one of the two forms, namely, the short-styled, in order that their pollen should be brushed off by insects and transported to the stigmas of the other form. In the long-styled flowers the anthers of the short enclosed stamens do not rotate on their axes, but dehisce on their inner sides, as is the common rule with the Rubiaceae; and this is the best position for the adherence of the pollen-grains to the proboscis of an entering insect. Fritz Muller therefore infers that as the plant became heterostyled, and as the stamens of the short- styled form increased in length, they gradually acquired the highly beneficial power of rotating on their own axes. But he has further shown, by the careful examination of many flowers, that this power has not as yet been perfected; and, consequently, that a certain proportion of the pollen is rendered useless, namely, that from the anthers which do not rotate properly. It thus appears that the development of the plant has not as yet been completed; the stamens have indeed acquired their proper length, but not their full and perfect power of rotation. (3/25. Fritz Muller gives another instance of the want of absolute perfection in the flowers of another member of the Rubiaceae, namely, Posoqueria fragrans, which is adapted in a most wonderful manner for cross-fertilisation by the agency of moths. (See 'Botanische Zeitung' 1866 Number 17.) In accordance with the nocturnal habits of these insects, most of the flowers open only during the night; but some open in the day, and the pollen of such flowers is robbed, as Fritz Muller has often seen, by humble-bees and other insects, without any benefit being thus conferred on the plant.)

The several points of difference in structure between the two forms of Faramea are highly remarkable. Until within a recent period, if any one had been shown two plants which differed in a uniform manner in the length of their stamens and pistils,—in the form of their stigmas,—in the manner of dehiscence and slightly in the size of their anthers,—and to an extraordinary degree in the diameter and structure of their pollen-grains, he would have declared it impossible that the two could have belonged to one and the same species.

[Suteria (species unnamed in the herbarium at Kew.) (Rubiaceae).

I owe to the kindness of Fritz Muller dried flowers of this plant from St. Catharina, in Brazil. In the long-styled form the stigma stands in the mouth of the corolla, above the anthers, which latter are enclosed within the tube, but only a short way down. In the short-styled form the anthers are placed in the mouth of the corolla above the stigma, which occupies the same position as the anthers in the other form, being seated only a short way down the tube. Therefore the pistil of the long-styled form does not exceed in length that of the short-styled in nearly so great a degree as in many other Rubiaceae. Nevertheless there is a considerable difference in the size of the pollen-grains in the two forms; for, as Fritz Muller informs me, those of the short-styled are to those of the long-styled as 100 to 75 in diameter.

Houstonia coerulea (Rubiaceae).

Professor Asa Gray has been so kind as to send me an abstract of some observations made by Dr. Rothrock on this plant. The pistil is exserted in the one form and the stamens in the other, as has long been observed. The stigmas of the long-styled form are shorter, stouter, and far more hispid than in the other form. The stigmatic hairs or papillae on the former are .04 millimetres, and on the latter only .023 millimetres in length. In the short-styled form the anthers are larger, and the pollen-grains, when distended with water, are to those from the long-styled form as 100 to 72 in diameter.

Selected capsules from some long-styled plants growing in the Botanic Gardens at Cambridge, U.S., near where plants of the other form grew, contained on an average 13 seeds; but these plants must have been subjected to unfavourable conditions, for some long-styled plants in a state of nature yielded an average of 21.5 seeds per capsule. Some short-styled plants, which had been planted by themselves in the Botanic Gardens, where it was not likely that they would have been visited by insects that had previously visited long-styled plants, produced capsules, eleven of which were wholly sterile, but one contained 4, and another 8 seeds. So that the short-styled form seems to be very sterile with its own pollen. Professor Asa Gray informs me that the other North American species of this genus are likewise heterostyled.

Oldenlandia [sp.?] (Rubiaceae).

Mr. J. Scott sent me from India dried flowers of a heterostyled species of this genus, which is closely allied to the last. The pistil in the long-styled flowers is longer by about a quarter of its length, and the stamens shorter in about the same proportion, than the corresponding organs in the short-styled flowers. In the latter the anthers are longer, and the divergent stigmas decidedly longer and apparently thinner than in the long-styled form. Owing to the state of the specimens, I could not decide whether the stigmatic papillae were longer in the one form than in the other. The pollen-grains, distended with water, from the short-styled flowers were to those from the long-styled as 100 to 78 in diameter, as deduced from the mean of ten measurements of each kind.

Hedyotis [sp.?] (Rubiaceae).

Fritz Muller sent me from St. Catharina, in Brazil, dried flowers of a small delicate species, which grows on wet sand near the edges of fresh-water pools. In the long-styled form the stigma projects above the corolla, and stands on a level with the projecting anthers of the short-styled form; but in the latter the stigmas stand rather beneath the level of the anthers in the other or long- styled form, these being enclosed within the tube of the corolla. The pistil of the long-styled form is nearly thrice as long as that of the short-styled, or, speaking strictly, as 100 to 39; and the papillae on the stigma of the former are broader, in the ratio of 4 to 3, but whether longer than those of the short- styled, I could not decide. In the short-styled form, the anthers are rather larger, and the pollen-grains are to those from the long-styled flowers, as 100 to 88 in diameter. Fritz Muller sent me a second, small-sized species, which is likewise heterostyled.

Coccocypselum [sp.?] (Rubiaceae).

Fritz Muller also sent me dried flowers of this plant from St. Catharina, in Brazil. The exserted stigma of the long-styled form stands a little above the level of the exserted anthers of the short-styled form; and the enclosed stigma of the latter also stands a little above the level of the enclosed anthers in the long-styled form. The pistil of the long-styled is about twice as long as that of the short-styled, with its two stigmas considerably longer, more divergent, and more curled. Fritz Muller informs me that he could detect no difference in the size of the pollen-grains in the two forms. Nevertheless, there can be no doubt that this plant is heterostyled.

Lipostoma [sp.?] (Rubiaceae).

Dried flowers of this plant, which grows in small wet ditches in St. Catharina, in Brazil, were likewise sent me by Fritz Muller. In the long-styled form the exserted stigma stands rather above the level of the exserted anthers of the other form; whilst in the short-styled form it stands on a level with the anthers of the other form. So that the want of strict correspondence in height between the stigmas and anthers in the two forms is reversed, compared with what occurs in Hedyotis. The long-styled pistil is to that of the short-styled as 100 to 36 in length; and its divergent stigmas are longer by fully one-third of their own length than those of the short-styled form. In the latter the anthers are a little larger, and the pollen-grains are as 100 to 80 in diameter, compared with those from the long-styled form.

Cinchona micrantha (Rubiaceae).

Dried specimens of both forms of this plant were sent me from Kew. (3/26. My attention was called to this plant by a drawing copied from Howard's 'Quinologia' Table 3 given by Mr. Markham in his 'Travels in Peru' page 539.) In the long-styled form the apex of the stigma stands just beneath the bases of the hairy lobes of the corolla; whilst the summits of the anthers are seated about halfway down the tube. The pistil is in length as 100 to 38 to that of the short-styled form. In the latter the anthers occupy the same position as the stigma of the other form, and they are considerably longer than those of the long-styled form. As the summit of the stigma in the short-styled form stands beneath the bases of the anthers, which are seated halfway down the corolla, the style has been extremely shortened in this form, its length to that of the long- styled being, in the specimens examined, only as 5.3 to 100! The stigma, also, in the short-styled form is very much shorter than that in the long-styled, in the ratio of 57 to 100. The pollen grains from the short-styled flowers, after having been soaked in water, were rather larger—in about the ratio of 100 to 91—than those from the long-styled flowers, and they were more triangular, with the angles more prominent. As all the grains from the short-styled flowers were thus characterised, and as they had been left in water for three days, I am convinced that this difference in shape in the two sets of grains cannot be accounted for by unequal distension with water.

Besides the several Rubiaceous genera already mentioned, Fritz Muller informs me that two or three species of Psychotria and Rudgea eriantha, natives of St. Catharina, in Brazil, are heterostyled, as is Manettia bicolor. I may add that I formerly fertilised with their own pollen several flowers on a plant of this latter species in my hothouse, but they did not set a single fruit. From Wight and Arnott's description, there seems to be little doubt that Knoxia in India is heterostyled; and Asa Gray is convinced that this is the case with Diodia and Spermacoce in the United States. Lastly, from Mr. W.W. Bailey's description, it appears that the Mexican Bouvardia leiantha is heterostyled. (3/27. 'Bulletin of the Torrey Bot. Club' 1876 page 106.)]

Altogether we now know of 17 heterostyled genera in the great family of the Rubiaceae; though more information is necessary with respect to some of them, more especially those mentioned in the last paragraph, before we can feel absolutely safe. In the 'Genera Plantarum,' by Bentham and Hooker, the Rubiaceae are divided into 25 tribes, containing 337 genera; and it deserves notice that the genera now known to be heterostyled are not grouped in one or two of these tribes, but are distributed in no less than eight of them. From this fact we may infer that most of the genera have acquired their heterostyled structure independently of one another; that is, they have not inherited this structure from some one or even two or three progenitors in common. It further deserves notice that in the homostyled genera, as I am informed by Professor Asa Gray, the stamens are either exserted or are included within the tube of the corolla, in a nearly constant manner; so that this character, which is not even of specific value in the heterostyled species, is often of generic value in other members of the family.

CHAPTER IV.

HETEROSTYLED TRIMORPHIC PLANTS.

Lythrum salicaria. Description of the three forms. Their power and complex manner of fertilising one another. Eighteen different unions possible. Mid-styled form eminently feminine in nature. Lythrum Graefferi likewise trimorphic. L. thymifolia dimorphic. L. Hyssopifolia homostyled. Nesaea verticillata trimorphic. Lagerstroemia, nature doubtful. Oxalis, trimorphic species of. O. Valdiviana. O. Regnelli, the illegitimate unions quite barren. O. speciosa. O. sensitiva. Homostyled species of Oxalis. Pontederia, the one monocotyledonous genus known to include heterostyled species.

In the previous chapters various heterostyled dimorphic plants have been described, and now we come to heterostyled trimorphic plants, or those which present three forms. These have been observed in three families, and consist of species of Lythrum and of the allied genus Nesaea, of Oxalis and Pontederia. In their manner of fertilisation these plants offer a more remarkable case than can be found in any other plant or animal.

Lythrum salicaria.

(FIGURE 4.10. Diagram of the flowers of the three forms of Lythrum salicaria, in their natural position, with the petals and calyx removed on the near side: enlarged six times. Top: Long-styled. Middle: Mid-styled. Bottom: Short-styled. The dotted lines with the arrows show the directions in which pollen must be carried to each stigma to ensure full fertility.)

The pistil in each form differs from that in either of the other forms, and in each there are two sets of stamens different in appearance and function. But one set of stamens in each form corresponds with a set in one of the other two forms. Altogether this one species includes three females or female organs and three sets of male organs, all as distinct from one another as if they belonged to different species; and if smaller functional differences are considered, there are five distinct sets of males. Two of the three hermaphrodites must coexist, and pollen must be carried by insects reciprocally from one to the other, in order that either of the two should be fully fertile; but unless all three forms coexist, two sets of stamens will be wasted, and the organisation of the species, as a whole, will be incomplete. On the other hand, when all three hermaphrodites coexist, and pollen is carried from one to the other, the scheme is perfect; there is no waste of pollen and no false co-adaptation. In short, nature has ordained a most complex marriage-arrangement, namely a triple union between three hermaphrodites,—each hermaphrodite being in its female organ quite distinct from the other two hermaphrodites and partially distinct in its male organs, and each furnished with two sets of males.

The three forms may be conveniently called, from the unequal lengths of their pistils, the LONG-STYLED, MID-STYLED, and SHORT-STYLED. The stamens also are of unequal lengths, and these may be called the LONGEST, MID-LENGTH, and SHORTEST. Two sets of stamens of different length are found in each form. The existence of the three forms was first observed by Vaucher, and subsequently more carefully by Wirtgen ; but these botanists, not being guided by any theory or even suspicion of their functional differences, did not perceive some of the most curious points of difference in their structure. (4/1. Vaucher 'Hist. Phys. des Plantes d'Europe' tome 2 1841 page 371. Wirtgen "Ueber Lythrum salicaria und dessen Formen" 'Verhand. des naturhist. Vereins fur preuss. Rheinl.' 5 Jahrgang 1848 S. 7.) I will first briefly describe the three forms by the aid of Figure 4.10, which shows the flowers, six times magnified, in their natural position, with their petals and calyx on the near side removed.

LONG-STYLED FORM.

This form can be at once recognised by the length of the pistil, which is (including the ovarium) fully one-third longer than that of the mid-styled, and more than thrice as long as that of the short-styled form. It is so disproportionately long, that it projects in the bud through the folded petals. It stands out considerably beyond the mid-length stamens; its terminal portion depends a little, but the stigma itself is slightly upturned. The globular stigma is considerably larger than that of the other two forms, with the papillae on its surface generally longer. The six mid-length stamens project about two-thirds the length of the pistil, and correspond in length with the pistil of the mid-styled form. Such correspondence in this and the two following forms is generally very close; the difference, where there is any, being usually in a slight excess of length in the stamens. The six shortest stamens lie concealed within the calyx; their ends are turned up, and they are graduated in length, so as to form a double row. The anthers of these stamens are smaller than those of the mid-length ones. The pollen is of the same yellow colour in both sets. H. Muller measured the pollen-grain in all three forms, and his measurements are evidently more trustworthy than those which I formerly made, so I will give them. (4/2. 'Die Befruchtung der Blumen' 1873 page 193.) The numbers refer to divisions of the micrometer equalling 1/300 millimetres. The grains, distended with water, from the mid-length stamens are 7 to 7 1/2, and those from the shortest stamens 6 to 6 1/2 in diameter, or as 100 to 86. The capsules of this form contain on an average 93 seeds: how this average was obtained will presently be explained. As these seeds, when cleaned, seemed larger than those from the mid-styled or short-styled forms, 100 of them were placed in a good balance, and by the double method of weighing were found to equal 121 seeds of the mid-styled or 142 of the short-styled; so that five long-styled seeds very nearly equal six mid-styled or seven short-styled seeds.

MID-STYLED FORM.

The pistil occupies the position represented in Figure 4.10, with its extremity considerably upturned, but to a variable degree; the stigma is seated between the anthers of the longest and the shortest stamens. The six longest stamens correspond in length with the pistil of the long-styled form; their filaments are coloured bright pink; the anthers are dark-coloured, but from containing bright-green pollen and from their early dehiscence they appear emerald-green. Hence in general appearance these stamens are remarkably dissimilar from the mid-length stamens of the long-styled form. The six shortest stamens are enclosed within the calyx, and resemble in all respects the shortest stamens of the long-styled form; both these sets correspond in length with the short pistil of the short-styled form. The green pollen-grains of the longest stamens are 9 to 10 in diameter, whilst the yellow grains from the shortest stamens are only 6; or as 100 to 63. But the pollen-grains from different plants appeared to me, in this case and others, to be in some degree variable in size. The capsules contain on an average 130 seeds; but perhaps, as we shall see, this is rather too high an average. The seeds themselves, as before remarked, are smaller than those of the long-styled form.

SHORT-STYLED FORM.

The pistil is here very short, not one-third of the length of that of the long- styled form. It is enclosed within the calyx, which, differently from that in the other two forms, does not enclose any anthers. The end of the pistil is generally bent upwards at right angles. The six longest stamens, with their pink filaments and green pollen, resemble the corresponding stamens of the mid-styled form. But according to H. Muller, their pollen-grains are a little larger, namely 9 1/2 to 10 1/2, instead of 9 to 10 in diameter. The six mid-length stamens, with their uncoloured filaments and yellow pollen, resemble in the size of their pollen-grains and in all other respects the corresponding stamens of the long-styled form. The difference in diameter between the grains from the two sets of anthers in the short-styled form is as 100 to 73. The capsules contain fewer seeds on an average than those of either of the preceding forms, namely 83.5; and the seeds are considerably smaller. In this latter respect, but not in number, there is a gradation parallel to that in the length of the pistil, the long-styled having the largest seeds, the mid-styled the next in size, and the short-styled the smallest.

We thus see that this plant exists under three female forms, which differ in the length and curvature of the style, in the size and state of the stigma, and in the number and size of the seed. There are altogether thirty-six males or stamens, and these can be divided into three sets of a dozen each, differing from one another in length, curvature, and colour of the filaments—in the size of the anthers, and especially in the colour and diameter of the pollen-grains. Each form bears half-a-dozen of one kind of stamens and half-a-dozen of another kind, but not all three kinds. The three kinds of stamens correspond in length with the three pistils: the correspondence is always between half of the stamens in two of the forms with the pistil of the third form. Table 4.a of the diameters of the pollen-grains, after immersion in water, from both sets of stamens in all three forms is copied from H. Muller; they are arranged in the order of their size:—

TABLE 4.a. Lythrum salicaria. Diameters of pollen-grains after immersion in water.

Column 1: Source of Pollen-grains. Column 2: Minimum diameter. Column 3: Maximum diameter.

Longest stamens of short-styled form : 9 1/2 : 10 1/2. Longest stamens of mid-styled form : 9 : 10. Mid-length stamens of long-styled form : 7 : 7 1/2. Mid-length stamens of short-styled form : 7 : 7 1/2. Shortest stamens of long-styled form : 6 : 6 1/2. Shortest stamens of mid-styled form : 6 : 6.

We here see that the largest pollen-grains come from the longest stamens, and the least (smallest) from the shortest; the extreme difference in diameter between them being as 100 to 60.

The average number of seeds in the three forms was ascertained by counting them in eight fine selected capsules taken from plants growing wild, and the result was, as we have seen, for the long-styled (neglecting decimals) 93, mid-styled 130, and short-styled 83. I should not have trusted in these ratios had I not possessed a number of plants in my garden which, owing to their youth, did not yield the full complement of seed, but were of the same age and grew under the same conditions, and were freely visited by bees. I took six fine capsules from each, and found the average to be for the long-styled 80, for the mid-styled 97, and for the short-styled 61. Lastly, legitimate unions effected by me between the three forms gave, as may be seen in the following tables, for the long- styled an average of 90 seeds, for the mid-styled 117, and for the short-styled 71. So that we have good concurrent evidence of a difference in the average production of seed by the three forms. To show that the unions effected by me often produced their full effect and may be trusted, I may state that one mid- styled capsule yielded 151 good seeds, which is the same number as in the finest wild capsule which I examined. Some artificially fertilised short- and long- styled capsules produced a greater number of seeds than was ever observed by me in wild plants of the same forms, but then I did not examine many of the latter. This plant, I may add, offers a remarkable instance, how profoundly ignorant we are of the life-conditions of a species. Naturally it grows "in wet ditches, watery places, and especially on the banks of streams," and though it produces so many minute seeds, it never spreads on the adjoining land; yet, when planted in my garden, on clayey soil lying over chalk, and which is so dry that a rush cannot be found, it thrives luxuriantly, grows to above 6 feet in height, produces self-sown seedlings, and (which is a severer test) is as fertile as in a state of nature. Nevertheless it would be almost a miracle to find this plant growing spontaneously on such land as that in my garden.

According to Vaucher and Wirtgen, the three forms coexist in all parts of Europe. Some friends gathered for me in North Wales a number of twigs from separate plants growing near one another, and classified them. My son did the same in Hampshire, and here is the result:—

TABLE 4.22. Lythrum salicaria. Classification according to form of flower.

Column 1: Place of origin. Column 2: Long-styled. Column 3: Mid-styled. Column 4: Short-styled. Column 5: Total.

North Wales : 95 : 97 : 72 : 264. Hampshire : 53 : 38 : 38 : 129. Total : 148 : 135 : 110 : 393.

If twice or thrice the number had been collected, the three forms would probably have been found nearly equal; I infer this from considering the above figures, and from my son telling me that if he had collected in another spot, he felt sure that the mid-styled plants would have been in excess. I several times sowed small parcels of seed, and raised all three forms; but I neglected to record the parent-form, excepting in one instance, in which I raised from short-styled seed twelve plants, of which only one turned out long-styled, four mid-styled, and seven short-styled.

Two plants of each form were protected from the access of insects during two successive years, and in the autumn they yielded very few capsules and presented a remarkable contrast with the adjoining uncovered plants, which were densely covered with capsules. In 1863 a protected long-styled plant produced only five poor capsules; two mid-styled plants produced together the same number; and two short-styled plants only a single one. These capsules contained very few seeds; yet the plants were fully productive when artificially fertilised under the net. In a state of nature the flowers are incessantly visited for their nectar by hive- and other bees, various Diptera and Lepidoptera. (4/3. H. Muller gives a list of the species 'Die Befruchtung der Blumen' page 196. It appears that one bee, the Cilissa melanura, almost confines its visits to this plant.) The nectar is secreted all round the base of the ovarium; but a passage is formed along the upper and inner side of the flower by the lateral deflection (not represented in the diagram) of the basal portions of the filaments; so that insects invariably alight on the projecting stamens and pistil, and insert their proboscides along the upper and inner margin of the corolla. We can now see why the ends of the stamens with their anthers, and the ends of the pistils with their stigmas, are a little upturned, so that they may be brushed by the lower hairy surfaces of the insects' bodies. The shortest stamens which lie enclosed within the calyx of the long- and mid-styled forms can be touched only by the proboscis and narrow chin of a bee; hence they have their ends more upturned, and they are graduated in length, so as to fall into a narrow file, sure to be raked by the thin intruding proboscis. The anthers of the longer stamens stand laterally farther apart and are more nearly on the same level, for they have to brush against the whole breadth of the insect's body. In very many other flowers the pistil, or the stamens, or both, are rectangularly bent to one side of the flower. This bending may be permanent, as with Lythrum and many others, or may be effected, as in Dictamnus fraxinella and others, by a temporary movement, which occurs in the case of the stamens when the anthers dehisce, and in the case of the pistil when the stigma is mature; but these two movements do not always take place simultaneously in the same flower. Now I have found no exception to the rule, that when the stamens and pistil are bent, they bend to that side of the flower which secretes nectar, even though there be a rudimentary nectary of large size on the opposite side, as in some species of Corydalis. When nectar is secreted on all sides, they bend to that side where the structure of the flower allows the easiest access to it, as in Lythrum, various Papilionaceae, and others. The rule consequently is, that when the pistils and stamens are curved or bent, the stigma and anthers are thus brought into the pathway leading to the nectary. There are a few cases which seem to be exceptions to this rule, but they are not so in truth; for instance, in the Gloriosa lily, the stigma of the grotesque and rectangularly bent pistil is brought, not into any pathway from the outside towards the nectar-secreting recesses of the flower, but into the circular route which insects follow in proceeding from one nectary to the other. In Scrophularia aquatica the pistil is bent downwards from the mouth of the corolla, but it thus strikes the pollen-dusted breast of the wasps which habitually visit these ill-scented flowers. In all these cases we see the supreme dominating power of insects on the structure of flowers, especially of those which have irregular corollas. Flowers which are fertilised by the wind must of course be excepted; but I do not know of a single instance of an irregular flower which is thus fertilised.

Another point deserves notice. In each of the three forms two sets of stamens correspond in length with the pistils in the other two forms. When bees suck the flowers, the anthers of the longest stamens, bearing the green pollen, are rubbed against the abdomen and the inner sides of the hind legs, as is likewise the stigma of the long-styled form. The anthers of the mid-length stamens and the stigma of the mid-styled form are rubbed against the under side of the thorax and between the front pair of legs. And, lastly, the anthers of the shortest stamens and the stigma of the short-styled form are rubbed against the proboscis and chin: for the bees in sucking the flowers insert only the front part of their heads into the flower. On catching bees, I observed much green pollen on the inner sides of the hind legs and on the abdomen, and much yellow pollen on the under side of the thorax. There was also pollen on the chin, and, it may be presumed, on the proboscis, but this was difficult to observe. I had, however, independent proof that pollen is carried on the proboscis; for a small branch of a protected short-styled plant (which produced spontaneously only two capsules) was accidentally left during several days pressing against the net, and bees were seen inserting their proboscides through the meshes, and in consequence numerous capsules were formed on this one small branch. From these several facts it follows that insects will generally carry the pollen of each form from the stamens to the pistil of corresponding length; and we shall presently see the importance of this adaptation. It must not, however, be supposed that the bees do not get more or less dusted all over with the several kinds of pollen; for this could be seen to occur with the green pollen from the longest stamens. Moreover a case will presently be given of a long-styled plant producing an abundance of capsules, though growing quite by itself, and the flowers must have been fertilised by their own kinds of pollen; but these capsules contained a very poor average of seed. Hence insects, and chiefly bees, act both as general carriers of pollen, and as special carriers of the right sort.

Wirtgen remarks on the variability of this plant in the branching of the stem, in the length of the bracteae, size of the petals, and in several other characters. (4/4. 'Verhand. des naturhist. Vereins fur Pr. Rheinl.' 5 Jahrgang 1848 pages 11, 13.) The plants which grew in my garden had their leaves, which differed much in shape, arranged oppositely, alternately, or in whorls of three. In this latter case the stems were hexagonal; those of the other plants being quadrangular. But we are concerned chiefly, with the reproductive organs: the upward bending of the pistil is variable, and especially in the short-styled form, in which it is sometimes straight, sometimes slightly curved, but generally bent at right angles. The stigma of the long-styled pistil frequently has longer papillae or is rougher than that of the mid-styled, and the latter than that of the short-styled; but this character, though fixed and uniform in the two forms of Primula veris, etc., is here variable, for I have seen mid- styled stigmas rougher than those of the long-styled. (4/5. The plants which I observed grew in my garden, and probably varied rather more than those growing in a state of nature. H. Muller has described the stigmas of all three forms with great care, and he appears to have found the stigmatic papillae differing constantly in length and structure in the three forms, being longest in the long-styled form.) The degree to which the longest and mid-length stamens are graduated in length and have their ends upturned is variable; sometimes all are equally long. The colour of the green pollen in the longest stamens is variable, being sometimes pale greenish-yellow; in one short-styled plant it was almost white. The grains vary a little in size: I examined one short-styled plant with the grains from the mid-length and shortest anthers of the same size. We here see great variability in many important characters; and if any of these variations were of service to the plant, or were correlated with useful functional differences, the species is in that state in which natural selection might readily do much for its modification.

ON THE POWER OF MUTUAL FERTILISATION BETWEEN THE THREE FORMS.

Nothing shows more clearly the extraordinary complexity of the reproductive system of this plant, than the necessity of making eighteen distinct unions in order to ascertain the relative fertilising power of the three forms. Thus the long-styled form has to be fertilised with pollen from its own two kinds of anthers, from the two in the mid-styled, and from the two in the short-styled form. The same process has to be repeated with the mid-styled and short-styled forms. It might have been thought sufficient to have tried on each stigma the green pollen, for instance, from either the mid- or short-styled longest stamens, and not from both; but the result proves that this would have been insufficient, and that it was necessary to try all six kinds of pollen on each stigma. As in fertilising flowers there will always be some failures, it would have been advisable to have repeated each of the eighteen unions a score of times; but the labour would have been too great; as it was, I made 223 unions, i.e. on an average I fertilised above a dozen flowers in the eighteen different methods. Each flower was castrated; the adjoining buds had to be removed, so that the flowers might be safely marked with thread, wool, etc.; and after each fertilisation the stigma was examined with a lens to see that there was sufficient pollen on it. Plants of all three forms were protected during two years by large nets on a framework; two plants were used during one or both years, in order to avoid any individual peculiarity in a particular plant. As soon as the flowers had withered, the nets were removed; and in the autumn the capsules were daily inspected and gathered, the ripe seeds being counted under the microscope. I have given these details that confidence may be placed in the following tables, and as some excuse for two blunders which, I believe, were made. These blunders are referred to, with their probable cause, in two footnotes to the tables. The erroneous numbers, however, are entered in the tables, that it may not be supposed that I have in any one instance tampered with the results.

A few words explanatory of the three tables must be given. Each is devoted to one of the three forms, and is divided into six compartments. The two upper ones in each table show the number of good seeds resulting from the application to the stigma of pollen from the two sets of stamens which correspond in length with the pistil of that form, and which are borne by the other two forms. Such unions are of a legitimate nature. The two next lower compartments show the result of the application of pollen from the two sets of stamens, not corresponding in length with the pistil, and which are borne by the other two forms. These unions are illegitimate. The two lowest compartments show the result of the application of each form's own two kinds of pollen from the two sets of stamens belonging to the same form, and which do not equal the pistil in length. These unions are likewise illegitimate. The term own-form pollen here used does not mean pollen from the flower to be fertilised—for this was never used—but from another flower on the same plant, or more commonly from a distinct plant of the same form. The figure "0" means that no capsule was produced, or if a capsule was produced that it contained no good seed. In some part of each row of figures in each compartment, a short horizontal line may be seen; the unions above this line were made in 1862, and below it in 1863. It is of importance to observe this, as it shows that the same general result was obtained during two successive years; but more especially because 1863 was a very hot and dry season, and the plants had occasionally to be watered. This did not prevent the full complement of seed being produced from the more fertile unions; but it rendered the less fertile ones even more sterile than they otherwise would have been. I have seen striking instances of this fact in making illegitimate and legitimate unions with Primula; and it is well known that the conditions of life must be highly favourable to give any chance of success in producing hybrids between species which are crossed with difficulty.

TABLE 4.23. Lythrum salicaria, long-styled form.

TABLE 4.23.1. Legitimate union.

13 flowers fertilised by the longest stamens of the mid-styled. These stamens equal in length the pistil of the long-styled.

Product of good seed in each capsule.

36 53 81 0 0 0 0 0 0 0 - 0 45 41

38 percent of these flowers yielded capsules. Each capsule contained, on an average, 51.2 seeds.

TABLE 4.23.2. Legitimate union.

13 flowers fertilised by the longest stamens of the short-styled. These stamens equal in length the pistil of the long-styled.

Product of good seed in each capsule.

159 104 43 119 96 poor seed. 96 103 99 0 131 0 116 - 114

84 percent of these flowers yielded capsules. Each capsule contained, on an average, 107.3 seeds.

TABLE 4.23.3. Illegitimate union.

14 flowers fertilised by the shortest stamens of the mid-styled.

3 0 0 0 0 0 0 0 0 0 - 0 0 0 0

Too sterile for any average.

TABLE 4.23.4. Illegitimate union.

12 flowers fertilised by the mid-length stamens of the short-styled.

20 0 0 0 0 0 0 0 - 0 0 0 0

Too sterile for any average.

TABLE 4.23.5. Illegitimate union.

15 flowers fertilised by own-form mid-length stamens.

2 - 10 0 23 0 0 0 0 0 0 0 0 0 0 0

Too sterile for any average.

TABLE 4.23.6. Illegitimate union.

15 flowers fertilised by own-form shortest stamens.

4 - 8 0 4 0 0 0 0 0 0 0 0 0 0 0

Too sterile for any average.

Besides the above experiments, I fertilised a considerable number of long-styled flowers with pollen, taken by a camel's-hair brush, from both the mid-length and shortest stamens of their own form: only 5 capsules were produced, and these yielded on an average 14.5 seeds. In 1863 I tried a much better experiment: a long-styled plant was grown by itself, miles away from any other plant, so that the flowers could have received only their own two kinds of pollen. The flowers were incessantly visited by bees, and their stigmas must have received successive applications of pollen on the most favourable days and at the most favourable hours: all who have crossed plants know that this highly favours fertilisation. This plant produced an abundant crop of capsules; I took by chance 20 capsules, and these contained seeds in number as follows:—

20 20 35 21 19 26 24 12 23 10 7 30 27 29 13 20 12 29 19 35

This gives an average of 21.5 seeds per capsule. As we know that the long-styled form, when standing near plants of the other two forms and fertilised by insects, produces on an average 93 seeds per capsule, we see that this form, fertilised by its own two pollens, yields only between one-fourth and one-fifth of the full number of seed. I have spoken as if the plant had received both its own kinds of pollen, and this is, of course, possible; but, from the enclosed position of the shortest stamens, it is much more probable that the stigma received exclusively pollen from the mid-length stamens; and this, as may be seen in Table 4.23.5, is the more fertile of the two self-unions.

TABLE 4.24. Lythrum salicaria, mid-styled form.

TABLE 4.24.1. Legitimate union.

12 flowers fertilised by the mid-length stamens of the long-styled. These stamens equal in length the pistil of the mid-styled.

Product of good seed in each capsule.

138 122 149 50 147 151 109 119 133 138 144 0 -

92 percent of these flowers (probably 100 per cent) yielded capsules. Each capsule contained, on an average, 127.3 seeds.

TABLE 4.24.2. Legitimate union.

12 flowers fertilised by the mid-length stamens of the short-styled. These stamens equal in length the pistil of the mid-styled.

Product of good seed in each capsule.

112 109 130 143 143 124 100 145 33 12 - 141 104

100 percent of these flowers yielded capsules. Each capsule contained, on an average, 108.0 seeds; or, excluding capsules with less than 20 seeds, the average is 116.7 seeds.

TABLE 4.24.3. Illegitimate union.

13 flowers fertilised by the shortest stamens of the long-styled.

83 12 0 19 0 85 seeds small and poor. - 0 44 0 44 0 45 0

54 percent of these flowers yielded capsules. Each capsule contained, on an average, 47.4 seeds; or, excluding capsules with less than 20 seeds, the average is 60.2 seeds.

TABLE 4.24.4. Illegitimate union.

15 flowers fertilised by the longest stamens of the short-styled.

130 86 115 113 14 29 6 17 2 113 9 79 - 128 132 0

93 percent of these flowers yielded capsules. Each capsule contained, on an average, 69.5 seeds; or, excluding capsules with less than 20 seeds, the average is 102.8 seeds.

TABLE 4.24.5. Illegitimate union.

12 flowers fertilised by own-form longest stamens.

92 0 9 0 63 0 - 0 136?* 0 0 0 0

(4/6. * I have hardly a doubt that this result of 136 seeds in Table 4.24.5 was due to a gross error. The flowers to be fertilised by their own longest stamens were first marked by "white thread," and those by the mid-length stamens of the long-styled form by "white silk;" a flower fertilised in the later manner would have yielded about 136 seeds, and it may be observed that one such pod is missing, namely at the bottom of Table 4.24.1. Therefore I have hardly any doubt that I fertilised a flower marked with "white thread" as if it had been marked with "white silk." With respect to the capsule which yielded 92 seeds, in the same column with that which yielded 136, I do not know what to think. I endeavoured to prevent pollen dropping from an upper to a lower flower, and I tried to remember to wipe the pincers carefully after each fertilisation; but in making eighteen different unions, sometimes on windy days, and pestered by bees and flies buzzing about, some few errors could hardly be avoided. One day I had to keep a third man by me all the time to prevent the bees visiting the uncovered plants, for in a few seconds' time they might have done irreparable mischief. It was also extremely difficult to exclude minute Diptera from the net. In 1862 I made the great mistake of placing a mid-styled and long-styled under the same huge net: in 1863 I avoided this error.)

Excluding the capsule with 136 seeds, 25 percent of the flowers yielded capsules, and each capsule contained, on an average, 54.6 seeds; or, excluding capsules with less than 20 seeds, the average is 77.5.

TABLE 4.24.6. Illegitimate union.

12 flowers fertilised by own-form shortest stamens.

0 0 0 0 0 0 - 0 0 0 0 0 0

Not one flower yielded a capsule.

Besides the experiments in Table 4.24, I fertilised a considerable number of mid-styled flowers with pollen, taken by a camel's-hair brush, from both the longest and shortest stamens of their own form: only 5 capsules were produced, and these yielded on an average 11.0 seeds.

TABLE 4.25. Lythrum salicaria, short-styled form.

TABLE 4.25.1. Legitimate union.

12 flowers fertilised by the shortest stamens of the long-styled. These stamens equal in length the pistil of the short-styled.

69 56 61 88 88 112 66 111 0 62 0 100 -

83 percent of the flowers yielded capsules. Each capsule contained, on an average, 81.3 seeds.

TABLE 4.25.2. Legitimate union.

13 flowers fertilised by the shortest stamens of the mid-styled. These stamens equal in length the pistil of the short-styled.

93 69 77 69 48 53 43 9 0 0 0 0 - 0

61 percent of the flowers yielded capsules. Each capsule contained, on an average, 64.6 seeds.

TABLE 4.25.3. Illegitimate union.

10 flowers fertilised by the mid-length stamens of the long-styled.

0 14 0 0 0 0 0 0 - 0 23

Too sterile for any average.

TABLE 4.25.4. Illegitimate union. 10 flowers fertilised by the longest stamens of the mid-styled.

0 0 0 0 0 0 0 0 - 0 0

Too sterile for any average.

TABLE 4.25.5. Illegitimate union.

10 flowers fertilised by own-form longest stamens.

0 0 0 0 0 0 - 0 0 0 0

Too sterile for any average.

TABLE 4.25.6. Illegitimate union.

10 flowers fertilised by own-form mid-length stamens.

64?* 0 0 0 0 0 - 0 21 0 9

(4/7. *I suspect that by mistake I fertilised this flower in Table 4.25.6 with pollen from the shortest stamens of the long-styled form, and it would then have yielded about 64 seeds. Flowers to be thus fertilised were marked with black silk; those with pollen from the mid-length stamens of the short-styled with black thread; and thus probably the mistake arose.)

Too sterile for any average.

Besides the experiments in the table, I fertilised a number of flowers without particular care with their own two kinds of pollen, but they did not produce a single capsule.

SUMMARY OF THE RESULTS.

LONG-STYLED FORM.

Twenty-six flowers fertilised legitimately by the stamens of corresponding length, borne by the mid-and short-styled forms, yielded 61.5 per cent of capsules, which contained on an average 89.7 seeds.

Twenty-six long-styled flowers fertilised illegitimately by the other stamens of the mid-and short-styled forms yielded only two very poor capsules.

Thirty long-styled flowers fertilised illegitimately by their own-form two sets of stamens yielded only eight very poor capsules; but long-styled flowers fertilised by bees with pollen from their own stamens produced numerous capsules containing on an average 21.5 seeds.

MID-STYLED FORM.

Twenty-four flowers legitimately fertilised by the stamens of corresponding length, borne by the long and short-styled forms, yielded 96 (probably 100) per cent of capsules, which contained (excluding one capsule with 12 seeds) on an average 117.2 seeds.

Fifteen mid-styled flowers fertilised illegitimately by the longest stamens of the short-styled form yielded 93 per cent of capsules, which (excluding four capsules with less than 20 seeds) contained on an average 102.8 seeds.

Thirteen mid-styled flowers fertilised illegitimately by the mid-length stamens of the long-styled form yielded 54 per cent of capsules, which (excluding one with 19 seeds) contained on an average 60.2 seeds.

Twelve mid-styled flowers fertilised illegitimately by their own-form longest stamens yielded 25 per cent of capsules, which (excluding one with 9 seeds) contained on an average 77.5 seeds.

Twelve mid-styled flowers fertilised illegitimately by their own-form shortest stamens yielded not a single capsule.

SHORT-STYLED FORM.

Twenty-five flowers fertilised legitimately by the stamens of corresponding length, borne by the long and mid-styled forms, yielded 72 per cent of capsules, which (excluding one capsule with only 9 seeds) contained on an average 70.8 seeds.

Twenty short-styled flowers fertilised illegitimately by the other stamens of the long and mid-styled forms yielded only two very poor capsules.

Twenty short-styled flowers fertilised illegitimately by their own stamens yielded only two poor (or perhaps three) capsules.

If we take all six legitimate unions together, and all twelve illegitimate unions together, we get the following results:

TABLE 4.26.

Column 1: Nature of union. Column 2: Number of Flowers fertilised. Column 3: Number of Capsules produced. Column 4: Average Number of Seeds per Capsule. Column 5: Average Number of Seeds per Flower fertilised.

The six legitimate unions : 75 : 56 : 96.29 : 71.89. The twelve illegitimate unions : 146 : 36 : 44.72 : 11.03.

Therefore the fertility of the legitimate unions to that of the illegitimate, as judged by the proportion of the fertilised flowers which yielded capsules, is as 100 to 33; and judged by the average number of seeds per capsule, as 100 to 46.

From this summary and the several foregoing tables we see that it is only pollen from the longest stamens which can fully fertilise the longest pistil; only that from the mid-length stamens, the mid-length pistil; and only that from the shortest stamens, the shortest pistil. And now we can comprehend the meaning of the almost exact correspondence in length between the pistil in each form and a set of six stamens in two of the other forms; for the stigma of each form is thus rubbed against that part of the insect's body which becomes charged with the proper pollen. It is also evident that the stigma of each form, fertilised in three different ways with pollen from the longest, mid-length, and shortest stamens, is acted on very differently, and conversely that the pollen from the twelve longest, twelve mid-length, and twelve shortest stamens acts very differently on each of the three stigmas; so that there are three sets of female and of male organs. Moreover, in most cases the six stamens of each set differ somewhat in their fertilising power from the six corresponding ones in one of the other forms. We may further draw the remarkable conclusion that the greater the inequality in length between the pistil and the set of stamens, the pollen of which is employed for its fertilisation, by so much is the sterility of the union increased. There are no exceptions to this rule. To understand what follows the reader should look to Tables 4.23, 4.24 and 4.25, and to the diagram Figure 4.10. In the long-styled form the short stamens obviously differ in length from the pistil to a greater degree than do the mid-length stamens; and the capsules produced by the use of pollen from the shortest stamens contain fewer seeds than those produced by the pollen from the mid-length stamens. The same result follows with the long-styled form, from the use of the pollen of shortest stamens of the mid-styled form and of the mid-length stamens of the short-styled form. The same rule also holds good with the mid-styled and short- styled forms, when illegitimately fertilised with pollen from the stamens more or less unequal in length to their pistils. Certainly the difference in sterility in these several cases is slight; but, as far as we are enabled to judge, it always increases with the increasing inequality of length between the pistil and the stamens which are used in each case.

The correspondence in length between the pistil in each form and a set of stamens in the other two forms, is probably the direct result of adaptation, as it is of high service to the species by leading to full and legitimate fertilisation. But the rule of the increased sterility of the illegitimate unions according to the greater inequality in length between the pistils and stamens employed for the union can be of no service. With some heterostyled dimorphic plants the difference of fertility between the two illegitimate unions appears at first sight to be related to the facility of self-fertilisation; so that when from the position of the parts the liability in one form to self- fertilisation is greater than in the other, a union of this kind has been checked by having been rendered the more sterile of the two. But this explanation does not apply to Lythrum; thus the stigma of the long-styled form is more liable to be illegitimately fertilised with pollen from its own mid- length stamens, or with pollen from the mid-length stamens of the short-styled form, than by its own shortest stamens or those of the mid-styled form; yet the two former unions, which it might have been expected would have been guarded against by increased sterility, are much less likely to be effected. The same relation holds good even in a more striking manner with the mid-styled form, and with the short-styled form as far as the extreme sterility of all its illegitimate unions allows of any comparison. We are led, therefore, to conclude that the rule of increased sterility in accordance with increased inequality in length between the pistils and stamens, is a purposeless result, incidental on those changes through which the species has passed in acquiring certain characters fitted to ensure the legitimate fertilisation of the three forms.

Another conclusion which may be drawn from Tables 4.23, 4.24, and 4.25, even from a glance at them, is that the mid-styled form differs from both the others in its much higher capacity for fertilisation in various ways. Not only did the twenty-four flowers legitimately fertilised by the stamens of corresponding lengths, all, or all but one, yield capsules rich in seed; but of the other four illegitimate unions, that by the longest stamens of the short-styled form was highly fertile, though less so than the two legitimate unions, and that by the mid-length stamens of the long-styled form was fertile to a considerable degree; the remaining two illegitimate unions, namely, with this form's own pollen, were sterile, but in different degrees. So that the mid-styled form, when fertilised in the six different possible methods, evinces five grades of fertility. By comparing Tables 4.24.3 and 4.24.6 we may see that the action of the pollen from the shortest stamens of the long-styled and mid-styled forms is widely different; in the one case above half the fertilised flowers yielded capsules containing a fair number of seeds; in the other case not one capsule was produced. So, again, the green, large-grained pollen from the longest stamens of the short-styled and mid-styled forms (in Tables 4.24.4 and 4.24.5) is widely different. In both these cases the difference in action is so plain that it cannot be mistaken, but it can be corroborated. If we look to Table 4.25 to the legitimate action of the shortest stamens of the long- and mid-styled forms on the short-styled form, we again see a similar but slighter difference, the pollen of the shortest stamens of the mid-styled form yielding a smaller average of seed during the two years of 1862 and 1863 than that from the shortest stamens of the long-styled form. Again, if we look to Table 4.23, to the legitimate action on the long-styled form of the green pollen of the two sets of longest stamens, we shall find exactly the same result, namely, that the pollen from the longest stamens of the mid-styled form yielded during both years fewer seeds than that from the longest stamens of the short-styled form. Hence it is certain that the two kinds of pollen produced by the mid-styled form are less potent than the two similar kinds of pollen produced by the corresponding stamens of the other two forms.