As most plants are at least occasionally cross-fertilised by the aid of insects, we may assume that this was the case with our supposed varying plant; but that it would have been beneficial to it to have been more regularly cross- fertilised. We should bear in mind how important an advantage it has been proved to be to many plants, though in different degrees and ways, to be cross- fertilised. It might well happen that our supposed species did not vary in function in the right manner, so as to become either dichogamous or completely self-sterile, or in structure so as to ensure cross-fertilisation. If it had thus varied, it would never have been rendered heterostyled, as this state would then have been superfluous. But the parent-species of our several existing heterostyled plants may have been, and probably were (judging from their present constitution) in some degree self-sterile; and this would have made regular cross-fertilisation still more desirable.

Now let us take a highly varying species with most or all of the anthers exserted in some individuals, and in others seated low down in the corolla; with the stigma also varying in position in like manner. Insects which visited such flowers would have different parts of their bodies dusted with pollen, and it would be a mere chance whether this were left on the stigma of the next flower which was visited. If all the anthers could have been placed on the same level in all the plants, then abundant pollen would have adhered to the same part of the body of the insects which frequented the flowers, and would afterwards have been deposited without loss on the stigma, if it likewise stood on the same unvarying level in all the flowers. But as the stamens and pistils are supposed to have already varied much in length and to be still varying, it might well happen that they could be reduced much more easily through natural selection into two sets of different lengths in different individuals, than all to the same length and level in all the individuals. We know from innumerable instances, in which the two sexes and the young of the same species differ, that there is no difficulty in two or more sets of individuals being formed which inherit different characters. In our particular case the law of compensation or balancement (which is admitted by many botanists) would tend to cause the pistil to be reduced in those individuals in which the stamens were greatly developed, and to be increased in length in those which had their stamens but little developed.

Now if in our varying species the longer stamens were to be nearly equalised in length in a considerable body of individuals, with the pistil more or less reduced; and in another body, the shorter stamens to be similarly equalised, with the pistil more or less increased in length, cross-fertilisation would be secured with little loss of pollen; and this change would be so highly beneficial to the species, that there is no difficulty in believing that it could be effected through natural selection. Our plant would then make a close approach in structure to a heterostyled dimorphic species; or to a trimorphic species, if the stamens were reduced to two lengths in the same flower in correspondence with that of the pistils in the other two forms. But we have not as yet even touched on the chief difficulty in understanding how heterostyled species could have originated. A completely self-sterile plant or a dichogamous one can fertilise and be fertilised by any other individual of the same species; whereas the essential character of a heterostyled plant is that an individual of one form cannot fully fertilise or be fertilised by an individual of the same form, but only by one belonging to another form.

H. Muller has suggested that ordinary or homostyled plants may have been rendered heterostyled merely through the effects of habit. (6/5. 'Die Befruchtung der Blumen' page 352.) Whenever pollen from one set of anthers is habitually applied to a pistil of particular length in a varying species, he believes that at last the possibility of fertilisation in any other manner will be nearly or completely lost. He was led to this view by observing that Diptera frequently carried pollen from the long-styled flowers of Hottonia to the stigma of the same form, and that this illegitimate union was not nearly so sterile as the corresponding union in other heterostyled species. But this conclusion is directly opposed by some other cases, for instance by that of Linum grandiflorum; for here the long-styled form is utterly barren with its own-form pollen, although from the position of the anthers this pollen is invariably applied to the stigma. It is obvious that with heterostyled dimorphic plants the two female and the two male organs differ in power; for if the same kind of pollen be placed on the stigmas of the two forms, and again if the two kinds of pollen be placed on the stigmas of the same form, the results are in each case widely different. Nor can we see how this differentiation of the two female and two male organs could have been effected merely through each kind of pollen being habitually placed on one of the two stigmas.

Another view seems at first sight probable, namely, that an incapacity to be fertilised in certain ways has been specially acquired by heterostyled plants. We may suppose that our varying species was somewhat sterile (as is often the case) with pollen from its own stamens, whether these were long or short; and that such sterility was transferred to all the individuals with pistils and stamens of the same length, so that these became incapable of intercrossing freely; but that such sterility was eliminated in the case of the individuals which differed in the length of their pistils and stamens. It is, however, incredible that so peculiar a form of mutual infertility should have been specially acquired unless it were highly beneficial to the species; and although it may be beneficial to an individual plant to be sterile with its own pollen, cross-fertilisation being thus ensured, how can it be any advantage to a plant to be sterile with half its brethren, that is, with all the individuals belonging to the same form? Moreover, if the sterility of the unions between plants of the same form had been a special acquirement, we might have expected that the long-styled form fertilised by the long-styled would have been sterile in the same degree as the short-styled fertilised by the short-styled; but this is hardly ever the case. On the contrary, there is sometimes the widest difference in this respect, as between the two illegitimate unions of Pulmonaria angustifolia and of Hottonia palustris.

It is a more probable view that the male and female organs in two sets of individuals have been by some means specially adapted for reciprocal action; and that the sterility between the individuals of the same set or form is an incidental and purposeless result. The meaning of the term "incidental" may be illustrated by the greater or less difficulty in grafting or budding together two plants belonging to distinct species; for as this capacity is quite immaterial to the welfare of either, it cannot have been specially acquired, and must be the incidental result of differences in their vegetative systems. But how the sexual elements of heterostyled plants came to differ from what they were whilst the species was homostyled, and how they became co-adapted in two sets of individuals, are very obscure points. We know that in the two forms of our existing heterostyled plants the pistil always differs, and the stamens generally differ in length; so does the stigma in structure, the anthers in size, and the pollen-grains in diameter. It appears, therefore, at first sight probable that organs which differ in such important respects could act on one another only in some manner for which they had been specially adapted. The probability of this view is supported by the curious rule that the greater the difference in length between the pistils and stamens of the trimorphic species of Lythrum and Oxalis, the products of which are united for reproduction, by so much the greater is the infertility of the union. The same rule applies to the two illegitimate unions of some dimorphic species, namely, Primula vulgaris and Pulmonaria angustifolia; but it entirely fails in other cases, as with Hottonia palustris and Linum grandiflorum. We shall, however, best perceive the difficulty of understanding the nature and origin of the co-adaptation between the reproductive organs of the two forms of heterostyled plants, by considering the case of Linum grandiflorum: the two forms of this plant differ exclusively, as far as we can see, in the length of their pistils; in the long-styled form, the stamens equal the pistil in length, but their pollen has no more effect on it than so much inorganic dust; whilst this pollen fully fertilises the short pistil of the other form. Now, it is scarcely credible that a mere difference in the length of the pistil can make a wide difference in its capacity for being fertilised. We can believe this the less because with some plants, for instance, Amsinckia spectabilis, the pistil varies greatly in length without affecting the fertility of the individuals which are intercrossed. So again I observed that the same plants of Primula veris and vulgaris differed to an extraordinary degree in the length of their pistils during successive seasons; nevertheless they yielded during these seasons exactly the same average number of seeds when left to fertilise themselves spontaneously under a net.

We must therefore look to the appearance of inner or hidden constitutional differences between the individuals of a varying species, of such a nature that the male element of one set is enabled to act efficiently only on the female element of another set. We need not doubt about the possibility of variations in the constitution of the reproductive system of a plant, for we know that some species vary so as to be completely self-sterile or completely self-fertile, either in an apparently spontaneous manner or from slightly changed conditions of life. Gartner also has shown that the individual plants of the same species vary in their sexual powers in such a manner that one will unite with a distinct species much more readily than another. (6/6. Gartner 'Bastarderzeugung im Pflanzenreich' 1849 page 165.) But what the nature of the inner constitutional differences may be between the sets or forms of the same varying species, or between distinct species, is quite unknown. It seems therefore probable that the species which have become heterostyled at first varied so that two or three sets of individuals were formed differing in the length of their pistils and stamens and in other co-adapted characters, and that almost simultaneously their reproductive powers became modified in such a manner that the sexual elements in one set were adapted to act on the sexual elements of another set; and consequently that these elements in the same set or form incidentally became ill-adapted for mutual interaction, as in the case of distinct species. I have elsewhere shown that the sterility of species when first crossed and of their hybrid offspring must also be looked at as merely an incidental result, following from the special co-adaptation of the sexual elements of the same species. (6/7. 'Origin of Species' 6th edition page 247; 'Variation of Animals and Plants under Domestication' 2nd edition volume 2 page 169; 'The Effects of Cross and Self-fertilisation' page 463. It may be well here to remark that, judging from the remarkable power with which abruptly changed conditions of life act on the reproductive system of most organisms, it is probable that the close adaptation of the male to the female elements in the two forms of the same heterostyled species, or in all the individuals of the same ordinary species, could be acquired only under long-continued nearly uniform conditions of life.) We can thus understand the striking parallelism, which has been shown to exist between the effects of illegitimately uniting heterostyled plants and of crossing distinct species. The great difference in the degree of sterility between the various heterostyled species when illegitimately fertilised, and between the two forms of the same species when similarly fertilised, harmonises well with the view that the result is an incidental one which follows from changes gradually effected in their reproductive systems, in order that the sexual elements of the distinct forms should act perfectly on one another.

TRANSMISSION OF THE TWO FORMS BY HETEROSTYLED PLANTS.

The transmission of the two forms by heterostyled plants, with respect to which many facts were given in the last chapter, may perhaps be found hereafter to throw some light on their manner of development. Hildebrand observed that seedlings from the long-styled form of Primula Sinensis when fertilised with pollen from the same form were mostly long-styled, and many analogous cases have since been observed by me. All the known cases are given in Tables 6.36 and 6.37.

TABLE 6.36. Nature of the offspring from illegitimately fertilised dimorphic plants.

Column 1: Species and form. Column 2: Number of long-styled offspring. Column 3: Number of short-styled offspring.

Primula veris. Long-styled form, fertilised by own-form pollen during five successive generations : 156 : 6.

Primula veris. Short-styled form, fertilised by own-form pollen : 5 : 9.

Primula vulgaris. Long-styled form, fertilised by own-form pollen during two successive generations : 69 : 0.

Primula auricula. Short-styled form, fertilised by own-form pollen, is said to produce during successive generations offspring in about the following proportions : 25 : 75.

Primula Sinensis. Long-styled form, fertilised by own-form pollen during two successive generations : 52 : 0.

Primula Sinensis. Long-styled form, fertilised by own-form pollen (Hildebrand) : 14 : 3.

Primula Sinensis. Short-styled form, fertilised by own-form pollen: 1 : 24.

Pulmonaria officinalis. Long-styled form, fertilised by own-form pollen : 11 : 0.

Polygonum fagopyrum. Long-styled form, fertilised by own-form pollen : 45 : 4.

Polygonum fagopyrum. Short-styled form, fertilised by own-form pollen : 13 : 20.

TABLE 6.37. Nature of the offspring from illegitimately fertilised trimorphic plants.

Column 1: Species and form. Column 2: Number of long-styled offspring. Column 3: Number of mid-styled offspring. Column 4: Number of short-styled offspring.

Lythrum salicaria. Long-styled form, fertilised by own-form pollen : 56 : 0 : 0.

Lythrum salicaria. Short-styled form, fertilised by own-form pollen : 1 : 0 : 8.

Lythrum salicaria. Short-styled form, fertilised by pollen from mid-length stamens of long-styled form : 4 : 0 : 8.

Lythrum salicaria. Mid-styled form, fertilised by own-form pollen : 1 : 3 : 0.

Lythrum salicaria. Mid-styled form, fertilised by pollen from shortest stamens of long-styled form : 17 : 8 : 0.

Lythrum salicaria. Mid-styled form, fertilised by pollen from longest stamens of short-styled form : 14 : 8 : 18.

Oxalis rosea. Long-styled form, fertilised during several generations by own- form pollen, produced offspring in the ratio of : 100 : 0 : 0.

Oxalis hedysaroides. Mid-styled form, fertilised by own-form pollen : 0 : 17 : 0.

We see in these two tables that the offspring from a form illegitimately fertilised with pollen from another plant of the same form belong, with a few exceptions, to the same form as their parents. For instance, out of 162 seedlings from long-styled plants of Primula veris fertilised during five generations in this manner, 156 were long-styled and only 6 short-styled. Of 69 seedlings from P. vulgaris similarly raised all were long-styled. So it was with 56 seedlings from the long-styled form of the trimorphic Lythrum salicaria, and with numerous seedlings from the long-styled form of Oxalis rosea. The offspring from the short-styled forms of dimorphic plants, and from both the mid-styled and short-styled forms of trimorphic plants, fertilised with their own-form pollen, likewise tend to belong to the same form as their parents, but not in so marked a manner as in the case of the long-styled form. There are three cases in Table 6.37, in which a form of Lythrum was fertilised illegitimately with pollen from another form; and in two of these cases all the offspring belonged to the same two forms as their parents, whilst in the third case they belonged to all three forms.

The cases hitherto given relate to illegitimate unions, but Hildebrand, Fritz Muller, and myself found that a very large proportion, or all of the offspring, from a legitimate union between any two forms of the trimorphic species of Oxalis belonged to the same two forms. A similar rule therefore holds good with unions which are fully fertile, as with those of an illegitimate nature which are more or less sterile. When some of the seedlings from a heterostyled plant belong to a different form from that of its parents, Hildebrand accounts for the fact by reversion. For instance, the long-styled parent-plant of Primula veris, from which the 162 illegitimate seedlings in Table 6.36 were derived in the course of five generations, was itself no doubt derived from the union of a long-styled and a short-styled parent; and the 6 short-styled seedlings may be attributed to reversion to their short-styled progenitor. But it is a surprising fact in this case, and in other similar ones, that the number of the offspring which thus reverted was not larger. The fact is rendered still more strange in the particular instance of P. veris, for there was no reversion until four or five generations of long-styled plants had been raised. It may be seen in both tables that the long-styled form transmits its form much more faithfully than does the short-styled, when both are fertilised with their own-form pollen; and why this should be so it is difficult to conjecture, unless it be that the aboriginal parent-form of most heterostyled species possessed a pistil which exceeded its own stamens considerably in length. (6/8. It may be suspected that this was the case with Primula, judging from the length of the pistil in several allied genera (see Mr. J. Scott 'Journal of the Linnean Society Botany' volume 8 1864 page 85). Herr Breitenbach found many specimens of Primula elatior growing in a state of nature with some flowers on the same plant long-styled, others short-styled and others equal-styled; and the long-styled form greatly preponderated in number; there being 61 of this form to 9 of the short-styled and 15 of the equal-styled.) I will only add that in a state of nature any single plant of a trimorphic species no doubt produces all three forms; and this may be accounted for either by its several flowers being separately fertilised by both the other forms, as Hildebrand supposes; or by pollen from both the other forms being deposited by insects on the stigma of the same flower.

EQUAL-STYLED VARIETIES.

The tendency of the dimorphic species of Primula to produce equal-styled varieties deserves notice. Cases of this kind have been observed, as shown in the last chapter, in no less than six species, namely, P. veris, vulgaris, Sinensis, auricula, farinosa, and elatior. In the case of P. veris, the stamens resemble in length, position and size of their pollen-grains the stamens of the short-styled form; whilst the pistil closely resembles that of the long-styled, but as it varies much in length, one proper to the short-styled form appears to have been elongated and to have assumed at the same time the functions of a long-styled pistil. Consequently the flowers are capable of spontaneous self- fertilisation of a legitimate nature and yield a full complement of seed, or even more than the number produced by ordinary flowers legitimately fertilised. With P. Sinensis, on the other hand, the stamens resemble in all respects the shorter ones proper to the long-styled form, whilst the pistil makes a near approach to that of the short-styled, but as it varies in length, it would appear as if a long-styled pistil had been reduced in length and modified in function. The flowers in this case as in the last are capable of spontaneous legitimate fertilisation, and are rather more productive than ordinary flowers legitimately fertilised. With P. auricula and farinosa the stamens resemble those of the short-styled form in length, but those of the long-styled in the size of their pollen-grains; the pistil also resembles that of the long-styled, so that although the stamens and pistil are of nearly equal length, and consequently pollen is spontaneously deposited on the stigma, yet the flowers are not legitimately fertilised and yield only a very moderate supply of seed. We thus see, firstly, that equal-styled varieties have originated in various ways, and, secondly, that the combination of the two forms in the same flower differs in completeness. With P. elatior some of the flowers on the same plant have become equal-styled, instead of all of them as in the other species.

Mr. Scott has suggested that the equal-styled varieties arise through reversion to the former homostyled condition of the genus. This view is supported by the remarkable fidelity with which the equal-styled variation is transmitted after it has once appeared. I have shown in Chapter 13 of my 'Variation of Animals and Plants under Domestication,' that any cause which disturbs the constitution tends to induce reversion, and it is chiefly the cultivated species of Primula which become equal-styled. Illegitimate fertilisation, which is an abnormal process, is likewise an exciting cause; and with illegitimately descended long- styled plants of P. Sinensis, I have observed the first appearance and subsequent stages of this variation. With some other plants of P. Sinensis of similar parentage the flowers appeared to have reverted to their original wild condition. Again, some hybrids between P. veris and vulgaris were strictly equal-styled, and others made a near approach to this structure. All these facts support the view that this variation results, at least in part, from reversion to the original state of the genus, before the species had become heterostyled. On the other hand, some considerations indicate, as previously remarked, that the aboriginal parent-form of Primula had a pistil which exceeded the stamens in length. The fertility of the equal-styled varieties has been somewhat modified, being sometimes greater and sometimes less than that of a legitimate union. Another view, however, may be taken with respect to the origin of the equal- styled varieties, and their appearance may be compared with that of hermaphrodites amongst animals which properly have their sexes separated; for the two sexes are combined in a monstrous hermaphrodite in a somewhat similar manner as the two sexual forms are combined in the same flower of an equal- styled variety of a heterostyled species.

FINAL REMARKS.

The existence of plants which have been rendered heterostyled is a highly remarkable phenomenon, as the two or three forms of the same undoubted species differ not only in important points of structure, but in the nature of their reproductive powers. As far as structure is concerned, the two sexes of many animals and of some plants differ to an extreme degree; and in both kingdoms the same species may consist of males, females, and hermaphrodites. Certain hermaphrodite cirripedes are aided in their reproduction by a whole cluster of what I have called complemental males, which differ wonderfully from the ordinary hermaphrodite form. With ants we have males and females, and two or three castes of sterile females or workers. With Termites there are, as Fritz Muller has shown, both winged and wingless males and females, besides the workers. But in none of these cases is there any reason to believe that the several males or several females of the same species differ in their sexual powers, except in the atrophied condition of the reproductive organs in the workers of social insects. Many hermaphrodite animals must unite for reproduction, but the necessity of such union apparently depends solely on their structure. On the other hand, with heterostyled dimorphic species there are two females and two sets of males, and with trimorphic species three females and three sets of males, which differ essentially in their sexual powers. We shall, perhaps, best perceive the complex and extraordinary nature of the marriage arrangements of a trimorphic plant by the following illustration. Let us suppose that the individuals of the same species of ant always lived in triple communities; and that in one of these, a large-sized female (differing also in other characters) lived with six middle-sized and six small-sized males; in the second community a middle-sized female lived with six large- and six small-sized males; and in the third, a small-sized female lived with six large- and six middle-sized males. Each of these three females, though enabled to unite with any male, would be nearly sterile with her own two sets of males, and likewise with two other sets of males of the same size with her own which lived in the other two communities; but she would be fully fertile when paired with a male of her own size. Hence the thirty-six males, distributed by half-dozens in the three communities, would be divided into three sets of a dozen each; and these sets, as well as the three females, would differ from one another in their reproductive powers in exactly the same manner as do the distinct species of the same genus. But it is a still more remarkable fact that young ants raised from any one of the three female ants, illegitimately fertilised by a male of a different size would resemble in a whole series of relations the hybrid offspring from a cross between two distinct species of ants. They would be dwarfed in stature, and more or less, or even utterly barren. Naturalists are so much accustomed to behold great diversities of structure associated with the two sexes, that they feel no surprise at almost any amount of difference; but differences in sexual nature have been thought to be the very touchstone of specific distinction. We now see that such sexual differences—the greater or less power of fertilising and being fertilised—may characterise the co-existing individuals of the same species, in the same manner as they characterise and have kept separate those groups of individuals, produced during the lapse of ages, which we rank and denominate as distinct species.

CHAPTER VII. POLYGAMOUS, DIOECIOUS, AND GYNO-DIOECIOUS PLANTS.

The conversion in various ways of hermaphrodite into dioecious plants. Heterostyled plants rendered dioecious. Rubiaceae. Verbenaceae. Polygamous and sub-dioecious plants. Euonymus. Fragaria. The two sub-forms of both sexes of Rhamnus and Epigaea. Ilex. Gyno-dioecious plants. Thymus, difference in fertility of the hermaphrodite and female individuals. Satureia. Manner in which the two forms probably originated. Scabiosa and other gyno-dioecious plants. Difference in the size of the corolla in the forms of polygamous, dioecious, and gyno-dioecious plants.

There are several groups of plants in which all the species are dioecious, and these exhibit no rudiments in the one sex of the organs proper to the other. About the origin of such plants nothing is known. It is possible that they may be descended from ancient lowly organised forms, which had from the first their sexes separated; so that they have never existed as hermaphrodites. There are, however, many other groups of species and single ones, which from being allied on all sides to hermaphrodites, and from exhibiting in the female flowers plain rudiments of male organs, and conversely in the male flowers rudiments of female organs, we may feel sure are descended from plants which formerly had the two sexes combined in the same flower. It is a curious and obscure problem how and why such hermaphrodites have been rendered bisexual.

If in some individuals of a species the stamens alone were to abort, females and hermaphrodites would be left existing, of which many instances occur; and if the female organs of the hermaphrodite were afterwards to abort, the result would be a dioecious plant. Conversely, if we imagine the female organs alone to abort in some individuals, males and hermaphrodites would be left; and the hermaphrodites might afterwards be converted into females.

In other cases, as in that of the common Ash-tree mentioned in the Introduction, the stamens are rudimentary in some individuals, the pistils in others, others again remaining as hermaphrodites. Here the modification of the two sets of organs appears to have occurred simultaneously, as far as we can judge from their equal state of abortion. If the hermaphrodites were supplanted by the individuals having separated sexes, and if these latter were equalised in number, a strictly dioecious species would be formed.

There is much difficulty in understanding why hermaphrodite plants should ever have been rendered dioecious. There would be no such conversion, unless pollen was already carried regularly by insects or by the wind from one individual to the other; for otherwise every step towards dioeciousness would lead towards sterility. As we must assume that cross-fertilisation was assured before an hermaphrodite could be changed into a dioecious plant, we may conclude that the conversion has not been effected for the sake of gaining the great benefits which follow from cross-fertilisation. We can, however, see that if a species were subjected to unfavourable conditions from severe competition with other plants, or from any other cause, the production of the male and female elements and the maturation of the ovules by the same individual, might prove too great a strain on its powers, and the separation of the sexes would then be highly beneficial. This, however, would be effected only under the contingency of a reduced number of seeds, produced by the females alone, being sufficient to keep up the stock.

There is another way of looking at the subject which partially removes a difficulty that appears at first sight insuperable, namely, that during the conversion of an hermaphrodite into a dioecious plant, the male organs must abort in some individuals and the female organs in others. Yet as all are exposed to the same conditions, it might have been expected that those which varied would tend to vary in the same manner. As a general rule only a few individuals of a species vary simultaneously in the same manner; and there is no improbability in the assumption that some few individuals might produce larger seeds than the average, better stocked with nourishment. If the production of such seeds were highly beneficial to a species, and on this head there can be little doubt, the variety with the large seeds would tend to increase. (7/1. See the facts given in 'The Effects of Cross and Self-fertilisation' page 353.) But in accordance with the law of compensation we might expect that the individuals which produced such seeds would, if living under severe conditions, tend to produce less and less pollen, so that their anthers would be reduced in size and might ultimately become rudimentary. This view occurred to me owing to a statement by Sir J.E. Smith that there are female and hermaphrodite plants of Serratula tinctoria, and that the seeds of the former are larger than those of the hermaphrodite form. (7/2. 'Transactions of the Linnean Society' volume 8 page 600.) It may also be worth while to recall the case of the mid-styled form of Lythrum salicaria, which produces a larger number of seeds than the other forms, and has somewhat smaller pollen-grains which have less fertilising power than those of the corresponding stamens in the other two forms; but whether the larger number of seeds is the indirect cause of the diminished power of the pollen, or vice versa, I know not. As soon as the anthers in a certain number of individuals became reduced in size in the manner just suggested or from any other cause, the other individuals would have to produce a larger supply of pollen; and such increased development would tend to reduce the female organs through the law of compensation, so as ultimately to leave them in a rudimentary condition; and the species would then become dioecious.

Instead of the first change occurring in the female organs we may suppose that the male ones first varied, so that some individuals produced a larger supply of pollen. This would be beneficial under certain circumstances, such as a change in the nature of the insects which visited the flowers, or in their becoming more anemophilous, for such plants require an enormous quantity of pollen. The increased action of the male organs would tend to affect through compensation the female organs of the same flower; and the final result would be that the species would consist of males and hermaphrodites. But it is of no use considering this case and other analogous ones, for, as stated in the Introduction, the coexistence of male and hermaphrodite plants is excessively rare.

It is no valid objection to the foregoing views that changes of such a nature would be effected with extreme slowness, for we shall presently see good reason to believe that various hermaphrodite plants have become or are becoming dioecious by many and excessively small steps. In the case of polygamous species, which exist as males, females and hermaphrodites, the latter would have to be supplanted before the species could become strictly dioecious; but the extinction of the hermaphrodite form would probably not be difficult, as a complete separation of the sexes appears often to be in some way beneficial. The males and females would also have to be equalised in number, or produced in some fitting proportion for the effectual fertilisation of the females.

There are, no doubt, many unknown laws which govern the suppression of the male or female organs in hermaphrodite plants, quite independently of any tendency in them to become monoecious, dioecious, or polygamous. We see this in those hermaphrodites which from the rudiments still present manifestly once possessed more stamens or pistils than they now do,—even twice as many, as a whole verticil has often been suppressed. Robert Brown remarks that "the order of reduction or abortion of the stamina in any natural family may with some confidence be predicted," by observing in other members of the family, in which their number is complete, the order of the dehiscence of the anthers (7/3. 'Transactions of the Linnean Society' volume 12 page 98 or 'Miscellaneous Works' volume 2 pages 278-81.); for the lesser permanence of an organ is generally connected with its lesser perfection, and he judges of perfection by priority of development. He also states that whenever there is a separation of the sexes in an hermaphrodite plant, which bears flowers on a simple spike, it is the females which expand first; and this he likewise attributes to the female sex being the more perfect of the two, but why the female should be thus valued he does not explain.

Plants under cultivation or changed conditions of life frequently become sterile; and the male organs are much oftener affected than the female, though the latter alone are sometimes affected. The sterility of the stamens is generally accompanied by a reduction in their size; and we may feel sure, from a wide-spread analogy, that both the male and female organs would become rudimentary in the course of many generations if they failed altogether to perform their proper functions. According to Gartner, if the anthers on a plant are contabescent (and when this occurs it is always at a very early period of growth) the female organs are sometimes precociously developed. (7/4. 'Beitrage zur Kenntniss' etc. page 117 et seq. The whole subject of the sterility of plants from various causes has been discussed in my 'Variation of Animals and Plants under Domestication' chapter 18 2nd edition volume 2 pages 146-56.) I mention this case as it appears to be one of compensation. So again is the well- known fact, that plants which increase largely by stolons or other such means are often utterly barren, with a large proportion of their pollen-grains in a worthless condition.

Hildebrand has shown that with hermaphrodite plants which are strongly proterandrous, the stamens in the flowers which open first sometimes abort; and this seems to follow from their being useless, as no pistils are then ready to be fertilised. Conversely the pistils in the flowers which open last sometimes abort; as when they are ready for fertilisation all the pollen has been shed. He further shows by means of a series of gradations amongst the Compositae, that a tendency from the causes just specified to produce either male or female florets, sometimes spreads to all the florets on the same head, and sometimes even to the whole plant (7/5. 'Ueber die Geschlechtsverhaltnisse bei den Compositen' 1869 page 89.); and in this latter case the species becomes dioecious. In those rare instances mentioned in the Introduction, in which some of the individuals of both monoecious and hermaphrodite plants are proterandrous, others being proterogynous, their conversion into a dioecious condition would probably be much facilitated, as they already consist of two bodies of individuals, differing to a certain extent in their reproductive functions.

Dimorphic heterostyled plants offer still more strongly marked facilities for becoming dioecious; for they likewise consist of two bodies of individuals in approximately equal numbers, and what probably is more important, both the male and female organs differ in the two forms, not only in structure but in function, in nearly the same manner as do the reproductive organs of two distinct species belonging to the same genus. Now if two species are subjected to changed conditions, though of the same nature, it is notorious that they are often affected very differently; therefore the male organs, for instance, in one form of a heterostyled plant might be affected by those unknown causes which induce abortion, differently from the homologous but functionally different organs in the other form; and so conversely with the female organs. Thus the great difficulty before alluded to is much lessened in understanding how any cause whatever could lead to the simultaneous reduction and ultimate suppression of the male organs in half the individuals of a species, and of the female organs in the other half, whilst all were subjected to exactly the same conditions of life.

That such reduction or suppression has occurred in some heterostyled plants is almost certain. The Rubiaceae contain more heterostyled genera than any other family, and from their wide distribution we may infer that many of them became heterostyled at a remote period, so that there will have been ample time for some of the species to have been since rendered dioecious. Asa Gray informs me that Coprosma is dioecious, and that it is closely allied through Nertera to Mitchella, which as we know is a heterostyled dimorphic species. In the male flowers of Coprosma the stamens are exserted, and in the female flowers the stigmas; so that, judging from the affinities of the above three genera, it seems probable that an ancient short-styled form bearing long stamens with large anthers and large pollen-grains (as in the case of several Rubiaceous genera) has been converted into the male Coprosma; and that an ancient long-styled form with short stamens, small anthers and small pollen-grains has been converted into the female form. But according to Mr. Meehan, Mitchella itself is dioecious in some districts; for he says that one form has small sessile anthers without a trace of pollen, the pistil being perfect; while in another form the stamens are perfect and the pistil rudimentary. (7/6. 'Proceedings of the Academy of Sciences of Philadelphia' July 28, 1868 page 183.) He adds that plants may be observed in the autumn bearing an abundant crop of berries, and others without a single one. Should these statements be confirmed, Mitchella will be proved to be heterostyled in one district and dioecious in another.

Asperula is likewise a Rubiaceous genus, and from the published description of the two forms of A. scoparia, an inhabitant of Tasmania, I did not doubt that it was heterostyled; but on examining some flowers sent me by Dr. Hooker they proved to be dioecious. The male flowers have large anthers and a very small ovarium, surmounted by a mere vestige of a stigma without any style; whilst the female flowers possess a large ovarium, the anthers being rudimentary and apparently quite destitute of pollen. Considering how many Rubiaceous genera are heterostyled, it is a reasonable suspicion that this Asperula is descended from a heterostyled progenitor; but we should be cautious on this head, for there is no improbability in a homostyled Rubiaceous plant becoming dioecious. Moreover, in an allied plant, Galium cruciatum, the female organs have been suppressed in most of the lower flowers, whilst the upper ones remain hermaphrodite; and here we have a modification of the sexual organs without any connection with heterostylism.

Mr. Thwaites informs me that in Ceylon various Rubiaceous plants are heterostyled; but in the case of Discospermum one of the two forms is always barren, the ovary containing about two aborted ovules in each loculus; whilst in the other form each loculus contains several perfect ovules; so that the species appears to be strictly dioecious.

Most of the species of the South American genus Aegiphila, a member of the Verbenaceae, apparently are heterostyled; and both Fritz Muller and myself thought that this was the case with Ae. obdurata, so closely did its flowers resemble those of the heterostyled species. But on examining the flowers, the anthers of the long-styled form were found to be entirely destitute of pollen and less than half the size of those in the other form, the pistil being perfectly developed. On the other hand, in the short-styled form the stigmas are reduced to half their proper length, having also an abnormal appearance; whilst the stamens are perfect. This plant therefore is dioecious; and we may, I think, conclude that a short-styled progenitor, bearing long stamens exserted beyond the corolla, has been converted into the male; and a long-styled progenitor with fully developed stigmas into the female.

From the number of bad pollen-grains in the small anthers of the short stamens of the long-styled form of Pulmonaria angustifolia, we may suspect that this form is tending to become female; but it does not appear that the other or short-styled form is becoming more masculine. Certain appearances countenance the belief that the reproductive system of Phlox subulata is likewise undergoing a change of some kind.

I have now given the few cases known to me in which heterostyled plants appear with some considerable degree of probability to have been rendered dioecious. Nor ought we to expect to find many such cases, for the number of heterostyled species is by no means large, at least in Europe, where they could hardly have escaped notice. Therefore the number of dioecious species which owe their origin to the transformation of heterostyled plants is probably not so large as might have been anticipated from the facilities which they offer for such conversion.

In searching for cases like the foregoing ones, I have been led to examine some dioecious or sub-dioecious plants, which are worth describing, chiefly as they show by what fine gradations hermaphrodites may pass into polygamous or dioecious species.

POLYGAMOUS, DIOECIOUS, AND SUB-DIOECIOUS PLANTS.

Euonymus Europaeus (CELASTRINEAE).

(Figure 7.12. Euonymus Europaeus Left: Hermaphrodite or male. Right: Female.)

The spindle-tree is described in all the botanical works which I have consulted as an hermaphrodite. Asa Gray speaks of the flowers of the American species as perfect, whilst those in the allied genus Celastrus are said to be "polygamo- dioecious." If a number of bushes of our spindle-tree be examined, about half will be found to have stamens equal in length to the pistil, with well-developed anthers; the pistil being likewise to all appearance well developed. The other half have a perfect pistil, with the stamens short, bearing rudimentary anthers destitute of pollen; so that these bushes are females. All the flowers on the same plant present the same structure. The female corolla is smaller than that on the polleniferous bushes. The two forms are shown in Figure 7.12.

I did not at first doubt that this species existed under an hermaphrodite and female form; but we shall presently see that some of the bushes which appear to be hermaphrodites never produce fruit, and these are in fact males. The species, therefore, is polygamous in the sense in which I use the term, and trioecious. The flowers are frequented by many Diptera and some small Hymenoptera for the sake of the nectar secreted by the disc, but I did not see a single bee at work; nevertheless the other insects sufficed to fertilise effectually female bushes growing at a distance of even 30 yards from any polleniferous bush.

The small anthers borne by the short stamens of the female flowers are well formed and dehisce properly, but I could never find in them a single grain of pollen. It is somewhat difficult to compare the length of the pistils in the two forms, as they vary somewhat in this respect and continue to grow after the anthers are mature. The pistils, therefore, in old flowers on a polleniferous plant are often of considerably greater length than in young flowers on a female plant. On this account the pistils from five flowers from so many hermaphrodite or male bushes were compared with those from five female bushes, before the anthers had dehisced and whilst the rudimentary ones were of a pink colour and not at all shrivelled. These two sets of pistils did not differ in length, or if there was any difference those of the polleniferous flowers were rather the longest. In one hermaphrodite plant, which produced during three years very few and poor fruit, the pistil much exceeded in length the stamens bearing perfect and as yet closed anthers; and I never saw such a case on any female plant. It is a surprising fact that the pistil in the male and in the semi-sterile hermaphrodite flowers has not been reduced in length, seeing that it performs very poorly or not at all its proper function. The stigmas in the two forms are exactly alike; and in some of the polleniferous plants which never produced any fruit I found that the surface of the stigma was viscid, so that pollen-grains adhered to it and had exserted their tubes. The ovules are of equal size in the two forms. Therefore the most acute botanist, judging only by structure, would never have suspected that some of the bushes were in function exclusively males.

Thirteen bushes growing near one another in a hedge consisted of eight females quite destitute of pollen and of five hermaphrodites with well-developed anthers. In the autumn the eight females were well covered with fruit, excepting one, which bore only a moderate number. Of the five hermaphrodites, one bore a dozen or two fruits, and the remaining four bushes several dozen; but their number was as nothing compared with those on the female bushes, for a single branch, between two and three feet in length, from one of the latter, yielded more than any one of the hermaphrodite bushes. The difference in the amount of fruit produced by the two sets of bushes is all the more striking, as from the sketches above given it is obvious that the stigmas of the polleniferous flowers can hardly fail to receive their own pollen; whilst the fertilisation of the female flowers depends on pollen being brought to them by flies and the smaller Hymenoptera, which are far from being such efficient carriers as bees.

I now determined to observe more carefully during successive seasons some bushes growing in another place about a mile distant. As the female bushes were so highly productive, I marked only two of them with the letters A and B, and five polleniferous bushes with the letters C to G. I may premise that the year 1865 was highly favourable for the fruiting of all the bushes, especially for the polleniferous ones, some of which were quite barren except under such favourable conditions. The season of 1864 was unfavourable. In 1863 the female A produced "some fruit;" in 1864 only 9; and in 1865, 97 fruit. The female B in 1863 was "covered with fruit;" in 1864 it bore 28; and in 1865 "innumerable very fine fruits." I may add, that three other female trees growing close by were observed, but only during 1863, and they then bore abundantly. With respect to the polleniferous bushes, the one marked C did not bear a single fruit during the years 1863 and 1864, but during 1865 it produced no less than 92 fruit, which, however, were very poor. I selected one of the finest branches with 15 fruit, and these contained 20 seeds, or on an average 1.33 per fruit. I then took by hazard 15 fruit from an adjoining female bush, and these contained 43 seeds; that is, more than twice as many, or on an average 2.86 per fruit. Many of the fruits from the female bushes included four seeds, and only one had a single seed; whereas not one fruit from the polleniferous bushes contained four seeds. Moreover when the two lots of seeds were compared, it was manifest that those from the female bushes were the larger. The second polleniferous bush, D, bore in 1863 about two dozen fruit,—in 1864 only 3 very poor fruit, each containing a single seed,—and in 1865, 20 equally poor fruit. Lastly, the three polleniferous bushes, E, F, and G, did not produce a single fruit during the three years 1863, 1864, and 1865.

We thus see that the female bushes differ somewhat in their degree of fertility, and the polleniferous ones in the most marked manner. We have a perfect gradation from the female bush, B, which in 1865 was covered with "innumerable fruits,"—through the female A, which produced during the same year 97,—through the polleniferous bush C, which produced this year 92 fruits, these, however, containing a very low average number of seeds of small size,—through the bush D, which produced only 20 poor fruit,—to the three bushes, E, F, and G, which did not this year, or during the two previous years, produce a single fruit. If these latter bushes and the more fertile female ones were to supplant the others, the spindle-tree would be as strictly dioecious in function as any plant in the world. This case appears to me very interesting, as showing how gradually an hermaphrodite plant may be converted into a dioecious one. (7/7. According to Fritz Muller 'Botanische Zeitung' 1870 page 151, a Chamissoa (Amaranthaceae) in Southern Brazil is in nearly the same state as our Euonymus. The ovules are equally developed in the two forms. In the female the pistil is perfect, whilst the anthers are entirely destitute of pollen. In the polleniferous form, the pistil is short and the stigmas never separate from one another, so that, although their surfaces are covered with fairly well-developed papillae, they cannot be fertilised, these latter plants do not commonly yield any fruit, and are therefore in function males. Nevertheless, on one occasion Fritz Muller found flowers of this kind in which the stigmas had separated, and they produced some fruit.)

Seeing how general it is for organs which are almost or quite functionless to be reduced in size, it is remarkable that the pistils of the polleniferous plants should equal or even exceed in length those of the highly fertile female plants. This fact formerly led me to suppose that the spindle-tree had once been heterostyled; the hermaphrodite and male plants having been originally long- styled, with the pistils since reduced in length, but with the stamens retaining their former dimensions; whilst the female plant had been originally short- styled, with the pistil in its present state, but with the stamens since greatly reduced and rendered rudimentary. A conversion of this kind is at least possible, although it is the reverse of that which appears actually to have occurred with some Rubiaceous genera and Aegiphila; for with these plants the short-styled form has become the male, and the long-styled the female. It is, however, a more simple view that sufficient time has not elapsed for the reduction of the pistil in the male and hermaphrodite flowers of our Euonymus; though this view does not account for the pistils in the polleniferous flowers being sometimes longer than those in the female flowers.

Fragaria vesca, Virginiana, chiloensis, etc. (ROSACEAE).

A tendency to the separation of the sexes in the cultivated strawberry seems to be much more strongly marked in the United States than in Europe; and this appears to be the result of the direct action of climate on the reproductive organs. In the best account which I have seen, it is stated that many of the varieties in the United States consist of three forms, namely, females, which produce a heavy crop of fruit,—of hermaphrodites, which "seldom produce other than a very scanty crop of inferior and imperfect berries,"—and of males, which produce none. (7/8. Mr. Leonard Wray 'Gardener's Chronicle' 1861 page 716.) The most skilful cultivators plant "seven rows of female plants, then one row of hermaphrodites, and so on throughout the field." The males bear large, the hermaphrodites mid-sized, and the females small flowers. The latter plants produce few runners, whilst the two other forms produce many; consequently, as has been observed both in England and in the United States, the polleniferous forms increase rapidly and tend to supplant the females. We may therefore infer that much more vital force is expended in the production of ovules and fruit than in the production of pollen. Another species, the Hautbois strawberry (F. elatior), is more strictly dioecious; but Lindley made by selection an hermaphrodite stock. (7/9. For references and further information on this subject, see 'Variation under Domestication' chapter 10 2nd edition volume 1 page 375.)

Rhamnus catharticus (RHAMNEAE).

(FIGURE 7.13. Rhamnus catharticus (from Caspary.) Left: Long-styled male. Right: Short-styled male.)

(FIGURE 7.14. Rhamnus catharticus. Left: Long-styled female. Right: Short-styled female.)

This plant is well known to be dioecious. My son William found the two sexes growing in about equal numbers in the Isle of Wight, and sent me specimens, together with observations on them. Each sex consists of two sub-forms. The two forms of the male differ in their pistils: in some plants it is quite small, without any distinct stigma; in others the pistil is much more developed, with the papillae on the stigmatic surfaces moderately large. The ovules in both kinds of males are in an aborted condition. On my mentioning this case to Professor Caspary, he examined several male plants in the botanic gardens at Konigsberg, where there were no females, and sent me the drawings in Figure 7.13.

In the English plants the petals are not so greatly reduced as represented in this drawing. My son observed that those males which had their pistils moderately well-developed bore slightly larger flowers, and, what is very remarkable, their pollen-grains exceeded by a little in diameter those of the males with greatly reduced pistils. This fact is opposed to the belief that the present species was once heterostyled; for in this case it might have been expected that the shorter-styled plants would have had larger pollen-grains.

In the female plants the stamens are in an extremely rudimentary condition, much more so than the pistils in the males. The pistil varies considerably in length in the female plants, so that they may be divided into two sub-forms according to the length of this organ. Both the petals and sepals are decidedly smaller in the females than in the males; and the sepals do not turn downwards, as do those of the male flowers when mature. All the flowers on the same male or same female bush, though subject to some variability, belong to the same sub-form; and as my son never experienced any difficulty in deciding under which class a plant ought to be included, he believes that the two sub-forms of the same sex do not graduate into one another. I can form no satisfactory theory how the four forms of this plant originated.

Rhamnus lanceolatus.

This plant exists in the United States, as I am informed by Professor Asa Gray, under two hermaphrodite forms. In the one, which may be called the short-styled, the flowers are sub-solitary, and include a pistil about two-thirds or only half as long as that in the other form; it has also shorter stigmas. The stamens are of equal length in the two forms; but the anthers of the short-styled contain rather less pollen, as far as I could judge from a few dried flowers. My son compared the pollen-grains from the two forms, and those from the long-styled flowers were to those from the short-styled, on an average from ten measurements, as 10 to 9 in diameter; so that the two hermaphrodite forms of this species resemble in this respect the two male forms of R. catharticus. The long-styled form is not so common as the short-styled. The latter is said by Asa Gray to be the more fruitful of the two, as might have been expected from its appearing to produce less pollen, and from the grains being of smaller size; it is therefore the more highly feminine of the two. The long-styled form produces a greater number of flowers, which are clustered together instead of being sub- solitary; they yield some fruit, but as just stated are less fruitful than the other form, so that this form appears to be the more masculine of the two. On the supposition that we have here an hermaphrodite plant becoming dioecious, there are two points deserving notice; firstly, the greater length of the pistil in the incipient male form; and we have met with a nearly similar case in the male and hermaphrodite forms of Euonymus compared with the females. Secondly, the larger size of the pollen-grains in the more masculine flowers, which perhaps may be attributed to their having retained their normal size; whilst those in the incipient female flowers have been reduced. The long-styled form of R. lanceolatus seems to correspond with the males of R. catharticus which have a longer pistil and larger pollen-grains. Light will perhaps be thrown on the nature of the forms in this genus, as soon as the power of both kinds of pollen on both stigmas is ascertained. Several other species of Rhamnus are said to be dioecious or sub-dioecious. (7/10. Lecoq 'Geogr. Bot.' tome 5 1856 pages 420- 26.) On the other hand, R. frangula is an ordinary hermaphrodite, for my son found a large number of bushes all bearing an equal profusion of fruit.

Epigaea repens (ERICACEAE).

This plant appears to be in nearly the same state as Rhamnus catharticus. It is described by Asa Gray as existing under four forms. (7/11. 'American Journal of Science' July 1876. Also 'The American Naturalist' 1876 page 490.) (1.) With long style, perfect stigma, and short abortive stamens. (2.) Shorter style, but with stigma equally perfect, short abortive stamens. These two female forms amounted to 20 per cent of the specimens received from one locality in Maine; but all the fruiting specimens belonged to the first form. (3.) Style long, as in Number 1, but with stigma imperfect, stamens perfect. (4.) Style shorter than in the last, stigma imperfect, stamens perfect. These two latter forms are evidently males. Therefore, as Asa Gray remarks, "the flowers may be classified into two kinds, each with two modifications; the two main kinds characterised by the nature and perfection of the stigma, along with more or less abortion of the stamens; their modifications, by the length of the style." Mr. Meehan has described the extreme variability of the corolla and calyx in this plant, and shows that it is dioecious. (7/12. "Variations in Epigaea repens" 'Proc. Acad. Nat. Soc. of Philadelphia' May 1868 page 153.) It is much to be wished that the pollen-grains in the two male forms should be compared, and their fertilising power tried on the two female forms.

Ilex aquifolium (AQUIFOLIACEAE).

In the several works which I have consulted, one author alone says that the holly is dioecious. (7/13. Vaucher 'Hist. Phys. des Plantes d'Europe' 1841 tome 2 page 11.) During several years I have examined many plants, but have never found one that was really hermaphrodite. I mention this genus because the stamens in the female flowers, although quite destitute of pollen, are but slightly and sometimes not at all shorter than the perfect stamens in the male flowers. In the latter the ovary is small and the pistil is almost aborted. The filaments of the perfect stamens adhere for a greater length to the petals than in the female flowers. The corolla of the latter is rather smaller than that of the male. The male trees produce a greater number of flowers than the females. Asa Gray informs me that I. opaca, which represents in the United States our common holly, appears (judging from dried flowers) to be in a similar state; and so it is, according to Vaucher, with several other but not with all the species of the genus.

GYNO-DIOECIOUS PLANTS.

The plants hitherto described either show a tendency to become dioecious, or apparently have become so within a recent period. But the species now to be considered consist of hermaphrodites and females without males, and rarely show any tendency to be dioecious, as far as can be judged from their present condition and from the absence of species having separated sexes within the same groups. Species belonging to the present class, which I have called gyno- dioecious, are found in various widely distinct families; but are much more common in the Labiatae (as has long been noticed by botanists) than in any other group. Such cases have been noticed by myself in Thymus serpyllum and vulgaris, Satureia hortensis, Origanum vulgare, and Mentha hirsuta; and by others in Nepeta glechoma, Mentha vulgaris and aquatica, and Prunella vulgaris. In these two latter species the female form, according to H. Muller, is infrequent. To these must be added Dracocephalum Moldavicum, Melissa officinalis and clinopodium, and Hyssopus officinalis. (7/14. H. Muller 'Die Befruchtung der Blumen' 1873 and 'Nature' 1873 page 161. Vaucher 'Plantes d'Europe' tome 3 page 611. For Dracocephalum Schimper as quoted by Braun 'Annals and Magazine of Natural History' 2nd series volume 18 1856 page 380. Lecoq 'Geographie Bot. de l'Europe' tome 8 pages 33, 38, 44, etc. Both Vaucher and Lecoq were mistaken in thinking that several of the plants named in the text are dioecious. They appear to have assumed that the hermaphrodite form was a male; perhaps they were deceived by the pistil not becoming fully developed and of proper length until some time after the anthers have dehisced.) In the two last-named plants the female form likewise appears to be rare, for I raised many seedlings of both, and all were hermaphrodites. It has already been remarked in the Introduction that andro-dioecious species, as they may be called, or those which consist of hermaphrodites and males, are extremely rare, or hardly exist.

Thymus serpyllum.

The hermaphrodite plants present nothing particular in the state of their reproductive organs; and so it is in all the following cases. The females of the present species produce rather fewer flowers and have somewhat smaller corollas than the hermaphrodites; so that near Torquay, where this plant abounds, I could, after a little practice, distinguish the two forms whilst walking quickly past them. According to Vaucher, the smaller size of the corolla is common to the females of most or all of the above-mentioned Labiatae. The pistil of the female, though somewhat variable in length, is generally shorter, with the margins of the stigma broader and formed of more lax tissue, than that of the hermaphrodite. The stamens in the female vary excessively in length; they are generally enclosed within the tube of the corolla, and their anthers do not contain any sound pollen; but after long search I found a single plant with the stamens moderately exserted, and their anthers contained a very few full-sized grains, together with a multitude of minute empty ones. In some females the stamens are extremely short, and their minute anthers, though divided into the two normal cells or loculi, contained not a trace of pollen: in others again the anthers did not exceed in diameter the filaments which supported them, and were not divided into two loculi. Judging from what I have myself seen and from the descriptions of others, all the plants in Britain, Germany, and near Mentone, are in the state just described; and I have never found a single flower with an aborted pistil. It is, therefore, remarkable that, according to Delpino, this plant near Florence is generally trimorphic, consisting of males with aborted pistils, females with aborted stamens, and hermaphrodites. (7/15. 'Sull' Opera, la Distribuzione dei Sessi nelle Piante, etc' 1867 page 7. With respect to Germany H. Muller 'Die Befruchtung etc.' page 327.)

I found it very difficult to judge of the proportional number of the two forms at Torquay. They often grow mingled together, but with large patches consisting of one form alone. At first I thought that the two were nearly equal in number; but on examining every plant which grew close to the edge of a little overhanging dry cliff, about 200 yards in length, I found only 12 females; all the rest, some hundreds in number, being hermaphrodites. Again, on an extensive gently sloping bank, which was so thickly covered with this plant that, viewed from the distance of half a mile it appeared of a pink colour, I could not discover a single female. Therefore the hermaphrodites must greatly exceed in number the females, at least in the localities examined by me. A very dry station apparently favours the presence of the female form. With some of the other above-named Labiatae the nature of the soil or climate likewise seems to determine the presence of one or both forms; thus with Nepeta glechoma, Mr. Hart found in 1873 that all the plants which he examined near Kilkenny in Ireland were females; whilst all near Bath were hermaphrodites, and near Hertford both forms were present, but with a preponderance of hermaphrodites. (7/16. 'Nature' June 1873 page 162.) It would, however, be a mistake to suppose that the nature of the conditions determines the form independently of inheritance; for I sowed in the same small bed seeds of T. serpyllum, gathered at Torquay from the female alone, and these produced an abundance of both forms. There is every reason to believe, from large patches consisting of the same form, that the same individual plant, however much it may spread, always retains the same form. In two distant gardens I found masses of the lemon-thyme (T. citriodorus, a var. of T. serpyllum, which I was informed had grown there during many years, and every flower was female.

With respect to the fertility of the two forms, I marked at Torquay a large hermaphrodite and a large female plant of nearly equal sizes, and when the seeds were ripe I gathered all the heads. The two heaps were of very nearly equal bulk; but the heads from the female plant numbered 160, and their seeds weighed 8.7 grains; whilst those from the hermaphrodite plant numbered 200, and their seeds weighed only 4.9 grains; so that the seeds from the female plant were to those from the hermaphrodite as 100 to 56 in weight. If the relative weight of the seeds from an equal number of flower-heads from the two forms be compared, the ratio is as 100 for the female to 45 for the hermaphrodite form.

Thymus vulgaris.

(FIGURE 7.15. Thymus vulgaris (magnified). Left: Hermaphrodite. Right: Two females.)

The common garden thyme resembles in almost every respect T. serpyllum. The same slight differences between the stigmas of the two forms could be perceived. In the females the stamens are not generally quite so much reduced as in the same form of T. serpyllum. In some specimens sent me from Mentone by Mr. Moggridge, together with the sketches in Figure 7.15, the anthers of the female, though small, were well formed, but they contained very little pollen, and not a single sound grain could be detected. Eighteen seedlings were raised from purchased seed, sown in the same small bed; and these consisted of seven hermaphrodites and eleven females. They were left freely exposed to the visits of bees, and no doubt every female flower was fertilised; for on placing under the microscope a large number of stigmas from female plants, not one could be found to which pollen-grains of thyme did not adhere. The seeds were carefully collected from the eleven female plants, and they weighed 98.7 grains; and those from the seven hermaphrodites 36.5 grains. This gives for an equal number of plants the ratio of 100 to 58; and we here see, as in the last case, how much more fertile the females are than the hermaphrodites. These two lots of seeds were sown separately in two adjoining beds, and the seedlings from both the hermaphrodite and female parent-plants consisted of both forms.

Satureia hortensis.

Eleven seedlings were raised in separate pots in a hotbed and afterwards kept in the greenhouse. They consisted of ten females and of a single hermaphrodite. Whether or not the conditions to which they had been subjected caused the great excess of females I do not know. In the females the pistil is rather longer than that of the hermaphrodite, and the stamens are mere rudiments, with minute colourless anthers destitute of pollen. The windows of the greenhouse were left open, and the flowers were incessantly visited by humble and hive bees. Although the ten females did not produce a single grain of pollen, yet they were all thoroughly well fertilised by the one hermaphrodite plant, and this is an interesting fact. It should be added that no other plant of this species grew in my garden. The seeds were collected from the finest female plant, and they weighed 78 grains; whilst those from the hermaphrodite, which was a rather larger plant than the female, weighed only 33.2 grains; that is, in the ratio of 100 to 43. The female form, therefore, is very much more fertile than the hermaphrodite, as in the two last cases; but the hermaphrodite was necessarily self-fertilised, and this probably diminished its fertility.

We may now consider the probable means by which so many of the Labiatae have been separated into two forms, and the advantages thus gained. H. Muller supposes that originally some individuals varied so as to produce more conspicuous flowers; and that insects habitually visited these first, and then dusted with their pollen visited and fertilised the less conspicuous flowers. (7/17. 'Die Befruchtung der Blumen' pages 319, 326.) The production of pollen by the latter plants would thus be rendered superfluous, and it would be advantageous to the species that their stamens should abort, so as to save useless expenditure. They would thus be converted into females. But another view may be suggested: as the production of a large supply of seeds evidently is of high importance to many plants, and as we have seen in the three foregoing cases that the females produce many more seeds than the hermaphrodites, increased fertility seems to me the more probable cause of the formation and separation of the two forms. From the data above given it follows that ten plants of Thymus serpyllum, if half consisted of hermaphrodites and half of females, would yield seeds compared with ten hermaphrodite plants in the ratio of 100 to 72. Under similar circumstances the ratio with Satureia hortensis (subject to the doubt from the self-fertilisation of the hermaphrodite) would be as 100 to 60. Whether the two forms originated in certain individuals varying and producing more seed than usual, and consequently producing less pollen; or in the stamens of certain individuals tending from some unknown cause to abort, and consequently producing more seed, it is impossible to decide; but in either case, if the tendency to the increased production of seed were steadily favoured, the result would be the complete abortion of the male organs. I shall presently discuss the cause of the smaller size of the female corolla.

[Scabiosa arvensis (DIPSACEAE).

It has been shown by H. Muller that this species exists in Germany under an hermaphrodite and female form. (7/18. 'Die Befruchtung der Blumen' page 368. The two forms occur not only in Germany, but in England and France. Lecoq 'Geographie Bot.' 1857 tome 6 pages 473, 477, says that male plants as well as hermaphrodites and females coexist; it is, however, possible that he may have been deceived by the flowers being so strongly proterandrous. From what Lecoq says, S. succisa likewise appears to occur under two forms in France.) In my neighbourhood (Kent) the female plants do not nearly equal in number the hermaphrodites. The stamens of the females vary much in their degree of abortion; in some plants they are quite short and produce no pollen; in others they reach to the mouth of the corolla, but their anthers are not half the proper size, never dehisce, and contain but few pollen-grains, these being colourless and of small diameter. The hermaphrodite flowers are strongly proterandrous, and H. Muller shows that, whilst all the stigmas on the same flower-head are mature at nearly the same time, the stamens dehisce one after the other; so that there is a great excess of pollen, which serves to fertilise the female plants. As the production of pollen by one set of plants is thus rendered superfluous, their male organs have become more or less completely aborted. Should it be hereafter proved that the female plants yield, as is probable, more seeds than the hermaphrodites, I should be inclined to extend the same view to this plant as to the Labiatae. I have also observed the existence of two forms in our endemic S. succisa, and in the exotic S. atro-purpurea. In the latter plant, differently to what occurs in S. arvensis, the female flowers, especially the larger circumferential ones, are smaller than those of the hermaphrodite form. According to Lecoq, the female flower-heads of S. succisa are likewise smaller than those of what he calls the male plants, but which are probably hermaphrodites.

Echium vulgare (BORAGINEAE).

The ordinary hermaphrodite form appears to be proterandrous, and nothing more need be said about it. The female differs in having a much smaller corolla and shorter pistil, but a well-developed stigma. The stamens are short; the anthers do not contain any sound pollen-grains, but in their place yellow incoherent cells which do not swell in water. Some plants were in an intermediate condition; that is, had one or two or three stamens of proper length with perfect anthers, the other stamens being rudimentary. In one such plant half of one anther contained green perfect pollen-grains, and the other half yellowish- green imperfect grains. Both forms produced seed, but I neglected to observe whether in equal numbers. As I thought that the state of the anthers might be due to some fungoid growth, I examined them both in the bud and mature state, but could find no trace of mycelium. In 1862 many female plants were found; and in 1864, 32 plants were collected in two localities, exactly half of which were hermaphrodites, fourteen were females, and two in an intermediate condition. In 1866, 15 plants were collected in another locality, and these consisted of four hermaphrodites and eleven females. I may add that this season was a wet one, which shows that the abortion of the stamens can hardly be due to the dryness of the sites where the plants grew, as I at one time thought probable. Seeds from an hermaphrodite were sown in my garden, and of the 23 seedlings raised, one belonged to the intermediate form, all the others being hermaphrodites, though two or three of them had unusually short stamens. I have consulted several botanical works, but have found no record of this plant varying in the manner here described.

Plantago lanceolata (PLANTAGINEAE).

Delpino states that this plant presents in Italy three forms, which graduate from an anemophilous into an entomophilous condition. According to H. Muller, there are only two forms in Germany, neither of which show any special adaptation for insect fertilisation, and both appear to be hermaphrodites. (7/19. 'Die Befruchtung' etc. page 342.) But I have found in two localities in England female and hermaphrodite forms existing together; and the same fact has been noticed by others. (7/20. Mr. C.W. Crocker 'The Gardener's Chronicle' 1864 page 294. Mr. W. Marshall writes to me to the same effect from Ely.) The females are less frequent than the hermaphrodites; their stamens are short, and their anthers, which are of a brighter green whilst young than those of the other form, dehisce properly, yet contain either no pollen, or a small amount of imperfect grains of variable size. All the flower-heads on a plant belong to the same form. It is well known that this species is strongly proterogynous, and I found that the protruding stigmas of both the hermaphrodite and female flowers were penetrated by pollen-tubes, whilst their own anthers were immature and had not escaped out of the bud. Plantago media does not present two forms; but it appears from Asa Gray's description, that such is the case with four of the North American species. (7/21. 'Manual of the Botany of the Northern United States' 2nd edition 1856 page 269. See also 'American Journal of Science' November 1862 page 419 and 'Proceedings of the American Academy of Science' October 14, 1862 page 53.) The corolla does not properly expand in the short- stamened form of these plants.

Cnicus, Serratula, Eriophorum.

In the Compositae, Cnicus palustris and acaulis are said by Sir J.E. Smith to exist as hermaphrodites and females, the former being the more frequent. With Serratula tinctoria a regular gradation may be followed from the hermaphrodite to the female form; in one of the latter plants the stamens were so tall that the anthers embraced the style as in the hermaphrodites, but they contained only a few grains of pollen, and these in an aborted condition; in another female, on the other hand, the anthers were much more reduced in size than is usual. Lastly, Dr. Dickie has shown that with Eriophorum angustifolium (Cyperaceae) hermaphrodite and female forms exist in Scotland and the Arctic regions, both of which yield seed. (7/22. Sir J.E. Smith 'Transactions of the Linnean Society' volume 13 page 599. Dr. Dickie 'Journal of the Linnean Society Botany' volume 9 1865 page 161.)]

It is a curious fact that in all the foregoing polygamous, dioecious, and gyno- dioecious plants in which any difference has been observed in the size of the corolla in the two or three forms, it is rather larger in the females, which have their stamens more or less or quite rudimentary, than in the hermaphrodites or males. This holds good with Euonymus, Rhamnus catharticus, Ilex, Fragaria, all or at least most of the before-named Labiatae, Scabiosa atro-purpurea, and Echium vulgare. So it is, according to Von Mohl, with Cardamine amara, Geranium sylvaticum, Myosotis, and Salvia. On the other hand, as Von Mohl remarks, when a plant produces hermaphrodite flowers and others which are males owing to the more or less complete abortion of the female organs, the corollas of the males are not at all increased in size, or only exceptionally and in a slight degree, as in Acer. (7/23. 'Botanische Zeitung' 1863 page 326.) It seems therefore probable that the decreased size of the female corollas in the foregoing cases is due to a tendency to abortion spreading from the stamens to the petals. We see how intimately these organs are related in double flowers, in which the stamens are readily converted into petals. Indeed some botanists believe that petals do not consist of leaves directly metamorphosed, but of metamorphosed stamens. That the lessened size of the corolla in the above case is in some manner an indirect result of the modification of the reproductive organs is supported by the fact that in Rhamnus catharticus not only the petals but the green and inconspicuous sepals of the female have been reduced in size; and in the strawberry the flowers are largest in the males, mid-sized in the hermaphrodites, and smallest in the females. These latter cases,—the variability in the size of the corolla in some of the above species, for instance in the common thyme,—together with the fact that it never differs greatly in size in the two forms—make me doubt much whether natural selection has come into play;—that is whether, in accordance with H. Muller's belief, the advantage derived from the polleniferous flowers being visited first by insects has been sufficient to lead to a gradual reduction of the corolla of the female. We should bear in mind that as the hermaphrodite is the normal form, its corolla has probably retained its original size. (7/24. It does not appear to me that Kerner's view 'Die Schutzmittel des Pollens' 1873 page 56, can be accepted in the present cases, namely that the larger corolla in the hermaphrodites and males serves to protect their pollen from rain. In the genus Thymus, for instance, the aborted anthers of the female are much better protected than the perfect ones of the hermaphrodite.) An objection to the above view should not be passed over; namely, that the abortion of the stamens in the females ought to have added through the law of compensation to the size of the corolla; and this perhaps would have occurred, had not the expenditure saved by the abortion of the stamens been directed to the female reproductive organs, so as to give to this form increased fertility.

CHAPTER VIII. CLEISTOGAMIC FLOWERS.

General character of cleistogamic flowers. List of the genera producing such flowers, and their distribution in the vegetable series. Viola, description of the cleistogamic flowers in the several species; their fertility compared with that of the perfect flowers. Oxalis acetosella. O. sensitiva, three forms of cleistogamic flowers. Vandellia. Ononis. Impatiens. Drosera. Miscellaneous observations on various other cleistogamic plants. Anemophilous species producing cleistogamic flowers. Leersia, perfect flowers rarely developed. Summary and concluding remarks on the origin of cleistogamic flowers. The chief conclusions which may be drawn from the observations in this volume.

It was known even before the time of Linnaeus that certain plants produced two kinds of flowers, ordinary open, and minute closed ones; and this fact formerly gave rise to warm controversies about the sexuality of plants. These closed flowers have been appropriately named cleistogamic by Dr. Kuhn. (8/1. 'Botanische Zeitung' 1867 page 65.) They are remarkable from their small size and from never opening, so that they resemble buds; their petals are rudimentary or quite aborted; their stamens are often reduced in number, with the anthers of very small size, containing few pollen-grains, which have remarkably thin transparent coats, and generally emit their tubes whilst still enclosed within the anther-cells; and, lastly, the pistil is much reduced in size, with the stigma in some cases hardly at all developed. These flowers do not secrete nectar or emit any odour; from their small size, as well as from the corolla being rudimentary, they are singularly inconspicuous. Consequently insects do not visit them; nor if they did, could they find an entrance. Such flowers are therefore invariably self-fertilised; yet they produce an abundance of seed. In several cases the young capsules bury themselves beneath the ground, and the seeds are there matured. These flowers are developed before, or after, or simultaneously with the perfect ones. Their development seems to be largely governed by the conditions to which the plants are exposed, for during certain seasons or in certain localities only cleistogamic or only perfect flowers are produced.

Dr. Kuhn, in the article above referred to, gives a list of 44 genera including species which bear flowers of this kind. To this list I have added some genera, and the authorities are appended in a footnote. I have omitted three names, from reasons likewise given in the footnote. But it is by no means easy to decide in all cases whether certain flowers ought to be ranked as cleistogamic. For instance, Mr. Bentham informs me that in the South of France some of the flowers on the vine do not fully open and yet set fruit; and I hear from two experienced gardeners that this is the case with the vine in our hothouses; but as the flowers do not appear to be completely closed it would be imprudent to consider them as cleistogamic. The flowers of some aquatic and marsh plants, for instance of Ranunculus aquatalis, Alisma natans, Subularia, Illecebrum, Menyanthes, and Euryale, remain closely shut as long as they are submerged, and in this condition fertilise themselves. (8/2. Delpino 'Sull' Opera, la Distribuzione dei Sessi nelle Piante' etc. 1867 page 30. Subularia, however, sometimes has its flowers fully expanded beneath the water, see Sir J.E. Smith 'English Flora' volume 3 1825 page 157. For the behaviour of Menyanthes in Russia see Gillibert in 'Act. Acad. St. Petersb.' 1777 part 2 page 45.—On Euryale 'Gardener's Chronicle' 1877 page 280.) They behave in this manner, apparently as a protection to their pollen, and produce open flowers when exposed to the air; so that these cases seem rather different from those of true cleistogamic flowers, and have not been included in the list. Again, the flowers of some plants which are produced very early or very late in the season do not properly expand; and these might perhaps be considered as incipiently cleistogamic; but as they do not present any of the remarkable peculiarities proper to the class, and as I have not found any full record of such cases, they are not entered in the list. When, however, it is believed on fairly good evidence that the flowers on a plant in its native country do not open at any hour of the day or night, and yet set seeds capable of germination, these may fairly be considered as cleistogamic, notwithstanding that they present no peculiarities of structure. I will now give as complete a list of the genera containing cleistogamic species as I have been able to collect.

TABLE 8.38. List of genera including cleistogamic species (chiefly after Kuhn). (8/3. I have omitted Trifolium and Arachis from the list, because Von Mohl says 'Botanische Zeitung' 1863 page 312, that the flower-stems merely draw the flowers beneath the ground, and that these do not appear to be properly cleistogamic. Correa de Mello 'Journal of the Linnean Society Botany' volume 11 1870 page 254, observed plants of Arachis in Brazil, and could never find such flowers. Plantago has been omitted because as far as I can discover it produces hermaphrodite and female flower-heads, but not cleistogamic flowers. Krascheninikowia (vel Stellaria) has been omitted because it seems very doubtful from Maximowicz' description whether the lower flowers which have no petals or very small ones, and barren stamens or none, are cleistogamic; the upper hermaphrodite flowers are said never to produce fruit, and therefore probably act as males. Moreover in Stellaria graminea, as Babington remarks 'British Botany' 1851 page 51, "shorter and longer petals accompany an imperfection of the stamens or germen."

I have added to the list the following cases: Several Acanthaceae, for which see J. Scott in 'Journal of Botany' London new series volume 1 1872 page 161. With respect to salvia see Dr. Ascherson in 'Botanische Zeitung' 1871 page 555. For Oxybaphus and Nyctaginia see Asa Gray in 'American Naturalist' November 1873 page 692. From Dr. Torrey's account of Hottonia inflata 'Bulletin of the Torrey Botanical Club' volume 2 June 1871, it is manifest that this plant produces true cleistogamic flowers. For Pavonia see Bouche in 'Sitzungsberichte d. Gesellsch. Natur. Freunde' October 20, 1874 page 90. I have added Thelymitra, as from the account given by Mr. Fitzgerald in his magnificent work on 'Australian Orchids' it appears that the flowers of this plant in its native home never open, but they do not appear to be reduced in size. Nor is this the case with the flowers of certain species of Epidendron, Cattleya, etc. see second edition of my 'Fertilisation of Orchids' page 147, which without expanding produce capsules. It is therefore doubtful whether these Orchideae ought to have been included in the list. From what Duval-Jouve says about Cryptostachys in 'Bulletin of the Soc. Bot. de France' tome 10 1863 page 195, this plant appears to produce cleistogamic flowers. the other additions to the list are noticed in my text.)

DICOTYLEDONS.

BORAGINEAE: Eritrichium.

CONVOLVULACEAE: Cuscuta.

SCROPHULARINEAE: Scrophularia. Linaria. Vandellia.

ACANTHACEAE: Cryphiacanthus. Eranthemum. Daedalacanthus. Dipteracanthus. Aechmanthera. Ruellia.

LABIATAE: Lamium. Salvia.

NYCTAGINEAE: Oxybaphus. Nyctaginia.

ASCLEPIADAE: Stapelia.

CAMPANULACEAE: Specularia. Campanula.

PRIMULACEAE: Hottonia.

COMPOSITAE: Anandria.

CRUCIFERAE: Heterocarpaea.

VIOLACEAE: Viola.

CISTINEAE: Helianthemum. Lechea.

MALVEACEAE: Pavonia.

MALPIGHIACEAE: Gaudichaudia. Aspicarpa. Camarea. Janusia.

POLYGALEAE: Polygala.

BALSAMINEAE: Impatiens.

GERANIACAEA: Oxalis.

LEGUMINOSAE: Ononis. Parochaetus. Chapmannia. Stylosanthus. Lespedeza. Vicia. Lathyrus. Martinsia vel Neurocarpum. Amphicarpaea. Glycine. Galactia. Voandzeia.

DROSERACEAE: Drosera.

MONOCOTYLEDONS.

JUNCEAE: Juncus.

GRAMINEAE: Leersia. Hordeum. Cryptostachys.

COMMELINEAE: Commelina.

PONTEDERACEAE: Monochoria.

ORCHIDEAE: Schomburgkia. Cattleya. Epidendron. Thelymitra.

The first point that strikes us in considering this list of 55 genera, is that they are very widely distributed in the vegetable series. They are more common in the family of the Leguminosae than in any other, and next in order in that of the Acanthaceae and Malpighiaceae. A large number, but not all the species, of certain genera, as of Oxalis and Viola, bear cleistogamic as well as ordinary flowers. A second point which deserves notice is that a considerable proportion of the genera produce more or less irregular flowers; this is the case with about 32 out of the 55 genera, but to this subject I shall recur.

I formerly made many observations on cleistogamic flowers, but only a few of them are worth giving, since the appearance of an admirable paper by Hugo Von Mohl, whose examination was in some respects much more complete than mine. (8/4. 'Botanische Zeitung' 1863 page 309-28.) His paper includes also an interesting history of our knowledge on the subject.

Viola canina.

The calyx of the cleistogamic flowers differs in no respect from that of the perfect ones. The petals are reduced to five minute scales; the lower one, which represents the lower lip, is considerably larger than the others, but with no trace of the spur-like nectary; its margins are smooth, whilst those of the other four scale-like petals are papillose. D. Muller of Upsala says that in the specimens which he observed the petals were completely aborted. (8/5. Ibid. 1857 page 730. This paper contains the first full and satisfactory account of any cleistogamic flower.) The stamens are very small, and only the two lower ones are provided with anthers, which do not cohere together as in the perfect flowers. The anthers are minute, with the two cells or loculi remarkably distinct; they contain very little pollen in comparison with those of the perfect flowers. The connective expands into a membranous hood-like shield which projects above the anther-cells. These two lower stamens have no vestige of the curious appendages which secrete nectar in the perfect flowers. The three other stamens are destitute of anthers and have broader filaments, with their terminal membranous expansions flatter or not so hood-like as those of the two antheriferous stamens. The pollen-grains have remarkably thin transparent coats; when exposed to the air they shrivel up quickly; when placed in water they swell, and are then 8-10/7000 of an inch in diameter, and therefore of smaller size than the ordinary pollen-grains similarly treated, which have a diameter of 13-14/7000 of an inch. In the cleistogamic flowers, the pollen-grains, as far as I could see, never naturally fall out of the anther-cells, but emit their tubes through a pore at the upper end. I was able to trace the tubes from the grains some way down the stigma. The pistil is very short, with the style hooked, so that its extremity, which is a little enlarged or funnel-shaped and represents the stigma, is directed downwards, being covered by the two membranous expansions of the antheriferous stamens. It is remarkable that there is an open passage from the enlarged funnel-shaped extremity to within the ovarium; this was evident, as slight pressure caused a bubble of air, which had been drawn in by some accident, to travel freely from one end to the other: a similar passage was observed by Michalet in V. alba. The pistil therefore differs considerably from that of the perfect flower; for in the latter it is much longer, and straight with the exception of the rectangularly bent stigma; nor is it perforated by an open passage.

The ordinary or perfect flowers have been said by some authors never to produce capsules; but this is an error, though only a small proportion of them do so. This appears to depend in some cases on their anthers not containing even a trace of pollen, but more generally on bees not visiting the flowers. I twice covered with a net a group of flowers, and marked with threads twelve of them which had not as yet expanded. This precaution is necessary, for though as a general rule the perfect flowers appear considerably before the cleistogamic ones, yet occasionally some of the latter are produced early in the season, and their capsules might readily be mistaken for those produced by the perfect flowers. Not one of the twelve marked perfect flowers yielded a capsule, whilst others under the net which had been artificially fertilised produced five capsules; and these contained exactly the same average number of seeds as some capsules from flowers outside the net which had been fertilised by bees. I have repeatedly seen Bombus hortorum, lapidarius, and a third species, as well as hive-bees, sucking the flowers of this violet: I marked six which were thus visited, and four of them produced fine capsules; the two others were gnawed off by some animal. I watched Bombus hortorum for some time, and whenever it came to a flower which did not stand in a convenient position to be sucked, it bit a hole through the spur-like nectary. Such ill-placed flowers would not yield any seed or leave descendants; and the plants bearing them would thus tend to be eliminated through natural selection.