As both these tables relate to the fertility of flowers fertilised by pollen from another plant and by their own pollen, they may be considered together. The difference between them consists in the self-fertilised flowers in Table 9/G, being produced by self-fertilised parents, and the crossed flowers by crossed parents, which in the later generations had become somewhat closely inter-related, and had been subjected all the time to nearly the same conditions. These two tables include fifty cases relating to thirty-two species. The flowers on many other species were crossed and self-fertilised, but as only a few were thus treated, the results cannot be trusted, as far as fertility is concerned, and are not here given. Some other cases have been rejected, as the plants were in an unhealthy condition. If we look to the figures in the two tables expressing the ratios between the mean relative fertility of the crossed and self-fertilised flowers, we see that in a majority of cases (i.e., in thirty-five out of fifty) flowers fertilised by pollen from a distinct plant yield more, sometimes many more, seeds than flowers fertilised with their own pollen; and they commonly set a larger proportion of capsules. The degree of infertility of the self-fertilised flowers differs extremely in the different species, and even, as we shall see in the section on self-sterile plants, in the individuals of the same species, as well as under slightly changed conditions of life. Their fertility ranges from zero to fertility equalling that of the crossed flowers; and of this fact no explanation can be offered. There are fifteen cases in the two tables in which the number of seeds per capsule produced by the self-fertilised flowers equals or even exceeds that yielded by the crossed flowers. Some few of these cases are, I believe, accidental; that is, would not recur on a second trial. This was apparently the case with the plants of the fifth generation of Ipomoea, and in one of the experiments with Dianthus. Nicotiana offers the most anomalous case of any, as the self-fertilised flowers on the parent-plants, and on their descendants of the second and third generations, produced more seeds than did the crossed flowers; but we shall recur to this case when we treat of highly self-fertile varieties.

It might have been expected that the difference in fertility between the crossed and self-fertilised flowers would have been more strongly marked in Table 9/G, in which the plants of one set were derived from self-fertilised parents, than in Table 9/F, in which flowers on the parent-plants were self-fertilised for the first time. But this is not the case, as far as my scanty materials allow of any judgment. There is therefore no evidence at present, that the fertility of plants goes on diminishing in successive self-fertilised generations, although there is some rather weak evidence that this does occur with respect to their height or growth. But we should bear in mind that in the later generations the crossed plants had become more or less closely inter-related, and had been subjected all the time to nearly uniform conditions.

It is remarkable that there is no close correspondence, either in the parent-plants or in the successive generations, between the relative number of seeds produced by the crossed and self-fertilised flowers, and the relative powers of growth of the seedlings raised from such seeds. Thus, the crossed and self-fertilised flowers on the parent-plants of Ipomoea, Gesneria, Salvia, Limnanthes, Lobelia fulgens, and Nolana produced a nearly equal number of seeds, yet the plants raised from the crossed seeds exceeded considerably in height those raised from the self-fertilised seeds. The crossed flowers of Linaria and Viscaria yielded far more seeds than the self-fertilised flowers; and although the plants raised from the former were taller than those from the latter, they were not so in any corresponding degree. With Nicotiana the flowers fertilised with their own pollen were more productive than those crossed with pollen from a slightly different variety; yet the plants raised from the latter seeds were much taller, heavier, and more hardy than those raised from the self-fertilised seeds. On the other hand, the crossed seedlings of Eschscholtzia were neither taller nor heavier than the self-fertilised, although the crossed flowers were far more productive than the self-fertilised. But the best evidence of a want of correspondence between the number of seeds produced by crossed and self-fertilised flowers, and the vigour of the offspring raised from them, is afforded by the plants of the Brazilian and European stocks of Eschscholtzia, and likewise by certain individual plants of Reseda odorata; for it might have been expected that the seedlings from plants, the flowers of which were excessively self-sterile, would have profited in a greater degree by a cross, than the seedlings from plants which were moderately or fully self-fertile, and therefore apparently had no need to be crossed. But no such result followed in either case: for instance, the crossed and self-fertilised offspring from a highly self-fertile plant of Reseda odorata were in average height to each other as 100 to 82; whereas the similar offspring from an excessively self-sterile plant were as 100 to 92 in average height.

With respect to the innate fertility of the plants of crossed and self-fertilised parentage, given in the previous Table 9/D—that is, the number of seeds produced by both lots when their flowers were fertilised in the same manner,—nearly the same remarks are applicable, in reference to the absence of any close correspondence between their fertility and powers of growth, as in the case of the plants in the Tables 9/F and 9/G, just considered. Thus the crossed and self-fertilised plants of Ipomoea, Papaver, Reseda odorata, and Limnanthes were almost equally fertile, yet the former exceeded considerably in height the self-fertilised plants. On the other hand, the crossed and self-fertilised plants of Mimulus and Primula differed to an extreme degree in innate fertility, but by no means to a corresponding degree in height or vigour.

In all the cases of self-fertilised flowers included in Tables 9/E, 9/F, and 9/G, these were fertilised with their own pollen; but there is another form of self-fertilisation, namely, by pollen from other flowers on the same plant; but this latter method made no difference in comparison with the former in the number of seeds produced, or only a slight difference. Neither with Digitalis nor Dianthus were more seeds produced by the one method than by the other, to any trustworthy degree. With Ipomoea rather more seeds, in the proportion of 100 to 91, were produced from a crossed between flowers on the same plant than from strictly self-fertilised flowers; but I have reason to suspect that the result was accidental. With Origanum vulgare, however, a cross between flowers on plants propagated by stolons from the same stock certainly increased slightly their fertility. This likewise occurred, as we shall see in the next section, with Eschscholtzia, perhaps with Corydalis cava and Oncidium; but not so with Bignonia, Abutilon, Tabernaemontana, Senecio, and apparently Reseda odorata.

SELF-STERILE PLANTS.

The cases here to be described might have been introduced in Table 9/F, which gives the relative fertility of flowers fertilised with their own pollen, and with that from a distinct plant, but it has been found more convenient to keep them for separate discussion. The present cases must not be confounded with those to be given in the next chapter relatively to flowers which are sterile when insects are excluded; for such sterility depends not merely on the flowers being incapable of fertilisation with their own pollen, but on mechanical causes, by which their pollen is prevented from reaching the stigma, or on the pollen and stigma of the same flower being matured at different periods.

In the seventeenth chapter of my 'Variation of Animals and Plants under Domestication' I had occasion to enter fully on the present subject; and I will therefore here give only a brief abstract of the cases there described, but others must be added, as they have an important bearing on the present work. Kolreuter long ago described plants of Verbascum phoeniceum which during two years were sterile with their own pollen, but were easily fertilised by that of four other species; these plants however afterwards became more or less self-fertile in a strangely fluctuating manner. Mr. Scott also found that this species, as well as two of its varieties, were self-sterile, as did Gartner in the case of Verbascum nigrum. So it was, according to this latter author, with two plants of Lobelia fulgens, though the pollen and ovules of both were in an efficient state in relation to other species. Five species of Passiflora and certain individuals of a sixth species have been found sterile with their own pollen; but slight changes in their conditions, such as being grafted on another stock or a change of temperature, rendered them self-fertile. Flowers on a completely self-impotent plant of Passiflora alata fertilised with pollen from its own self-impotent seedlings were quite fertile. Mr. Scott, and afterwards Mr. Munro, found that some species of Oncidium and of Maxillaria cultivated in a hothouse in Edinburgh were quite sterile with their own pollen; and Fritz Muller found this to be the case with a large number of Orchidaceous genera growing in their native home of South Brazil. (9/2. 'Botanische Zeitung' 1868 page 114.) He also discovered that the pollen-masses of some orchids acted on their own stigmas like a poison; and it appears that Gartner formerly observed indications of this extraordinary fact in the case of some other plants.

Fritz Muller also states that a species of Bignonia and Tabernaemontana echinata are both sterile with their own pollen in their native country of Brazil. (9/3. Ibid 1868 page 626 and 1870 page 274.) Several Amaryllidaceous and Liliaceous plants are in the same predicament. Hildebrand observed with care Corydalis cava, and found it completely self-sterile (9/4. 'Report of the International Horticultural Congress' 1866.); but according to Caspary a few self-fertilised seeds are occasionally produced: Corydalis halleri is only slightly self-sterile, and C. intermedia not at all so. (9/5. 'Botanische Zeitung' June 27, 1873.) In another Fumariaceous genus, Hypecoum, Hildebrand observed that H. grandiflorum was highly self-sterile, whilst H. procumbens was fairly self-fertile. (9/6. 'Jahrb. fur wiss. Botanik' B. 7 page 464.) Thunbergia alata kept by me in a warm greenhouse was self-sterile early in the season, but at a later period produced many spontaneously self-fertilised fruits. So it was with Papaver vagum: another species, P. alpinum, was found by Professor H. Hoffmann to be quite self-sterile excepting on one occasion (9/7. 'Zur Speciesfrage' 1875 page 47.); whilst P. somniferum has been with me always completely self-sterile.

Eschscholtzia californica.

This species deserves a fuller consideration. A plant cultivated by Fritz Muller in South Brazil happened to flower a month before any of the others, and it did not produce a single capsule. This led him to make further observations during the next six generations, and he found that all his plants were completely sterile, unless they were crossed by insects or were artificially fertilised with pollen from a distinct plant, in which case they were completely fertile. (9/8. 'Botanische Zeitung' 1868 page 115 and 1869 page 223.) I was much surprised at this fact, as I had found that English plants, when covered by a net, set a considerable number of capsules; and that these contained seeds by weight, compared with those on plants intercrossed by the bees, as 71 to 100. Professor Hildebrand, however, found this species much more self-sterile in Germany than it was with me in England, for the capsules produced by self-fertilised flowers, compared with those from intercrossed flowers, contained seeds in the ratio of only 11 to 100. At my request Fritz Muller sent me from Brazil seeds of his self-sterile plants, from which I raised seedlings. Two of these were covered with a net, and one produced spontaneously only a single capsule containing no good seeds, but yet, when artificially fertilised with its own pollen, produced a few capsules. The other plant produced spontaneously under the net eight capsules, one of which contained no less than thirty seeds, and on an average about ten seeds per capsule. Eight flowers on these two plants were artificially self-fertilised, and produced seven capsules, containing on an average twelve seeds; eight other flowers were fertilised with pollen from a distinct plant of the Brazilian stock, and produced eight capsules, containing on an average about eighty seeds: this gives a ratio of 15 seeds for the self-fertilised capsules to 100 for the crossed capsules. Later in the season twelve other flowers on these two plants were artificially self-fertilised; but they yielded only two capsules, containing three and six seeds. It appears therefore that a lower temperature than that of Brazil favours the self-fertility of this plant, whilst a still lower temperature lessens it. As soon as the two plants which had been covered by the net were uncovered, they were visited by many bees,and it was interesting to observe how quickly they became, even the more sterile plant of the two, covered with young capsules. On the following year eight flowers on plants of the Brazilian stock of self-fertilised parentage (i.e., grandchildren of the plants which grew in Brazil) were again self-fertilised, and produced five capsules, containing on an average 27.4 seeds, with a maximum in one of forty-two seeds; so that their self-fertility had evidently increased greatly by being reared for two generations in England. On the whole we may conclude that plants of the Brazilian stock are much more self-fertile in this country than in Brazil, and less so than plants of the English stock in England; so that the plants of Brazilian parentage retained by inheritance some of their former sexual constitution. Conversely, seeds from English plants sent by me to Fritz Muller and grown in Brazil, were much more self-fertile than his plants which had been cultivated there for several generations; but he informs me that one of the plants of English parentage which did not flower the first year, and was thus exposed for two seasons to the climate of Brazil, proved quite self-sterile, like a Brazilian plant, showing how quickly the climate had acted on its sexual constitution.

Abutilon darwinii.

Seeds of this plant were sent me by Fritz Muller, who found it, as well as some other species of the same genus, quite sterile in its native home of South Brazil, unless fertilised with pollen from a distinct plant, either artificially or naturally by humming-birds. (9/9. 'Jenaische Zeitschr. fur Naturwiss' B. 7 1872 page 22 and 1873 page 441.) Several plants were raised from these seeds and kept in the hothouse. They produced flowers very early in the spring, and twenty of them were fertilised, some with pollen from the same flower, and some with pollen from other flowers on the same plants; but not a single capsule was thus produced, yet the stigmas twenty-seven hours after the application of the pollen were penetrated by the pollen-tubes. At the same time nineteen flowers were crossed with pollen from a distinct plant, and these produced thirteen capsules, all abounding with fine seeds. A greater number of capsules would have been produced by the cross, had not some of the nineteen flowers been on a plant which was afterwards proved to be from some unknown cause completely sterile with pollen of any kind. Thus far these plants behaved exactly like those in Brazil; but later in the season, in the latter part of May and in June, they began to produce under a net a few spontaneously self-fertilised capsules. As soon as this occurred, sixteen flowers were fertilised with their own pollen, and these produced five capsules, containing on an average 3.4 seeds. At the same time I selected by chance four capsules from the uncovered plants growing close by, the flowers of which I had seen visited by humble-bees, and these contained on an average 21.5 seeds; so that the seeds in the naturally intercrossed capsules to those in the self-fertilised capsules were as 100 to 16. The interesting point in this case is that these plants, which were unnaturally treated by being grown in pots in a hothouse, under another hemisphere, with a complete reversal of the seasons, were thus rendered slightly self-fertile, whereas they seem always to be completely self-sterile in their native home.

Senecio cruentus (greenhouse varieties, commonly called Cinerarias, probably derived from several fruticose or herbaceous species much intercrossed (9/10. I am much obliged to Mr. Moore and to Mr. Thiselton Dyer for giving me information with respect to the varieties on which I experimented. Mr. Moore believes that Senecio cruentas, tussilaginis, and perhaps heritieri, maderensis and populifolius have all been more or less blended together in our Cinerarias.))

Two purple-flowered varieties were placed under a net in the greenhouse, and four corymbs on each were repeatedly brushed with flowers from the other plant, so that their stigmas were well covered with each other's pollen. Two of the eight corymbs thus treated produced very few seeds, but the other six produced on an average 41.3 seeds per corymb, and these germinated well. The stigmas on four other corymbs on both plants were well smeared with pollen from the flowers on their own corymbs; these eight corymbs produced altogether ten extremely poor seeds, which proved incapable of germinating. I examined many flowers on both plants, and found the stigmas spontaneously covered with pollen; but they produced not a single seed. These plants were afterwards left uncovered in the same house where many other Cinerarias were in flower; and the flowers were frequently visited by bees. They then produced plenty of seed, but one of the two plants less than the other, as this species shows some tendency to be dioecious.

The trial was repeated on another variety with white petals tipped with red. Many stigmas on two corymbs were covered with pollen from the foregoing purple variety, and these produced eleven and twenty-two seeds, which germinated well. A large number of the stigmas on several of the other corymbs were repeatedly smeared with pollen from their own corymb; but they yielded only five very poor seeds, which were incapable of germination. Therefore the above three plants belonging to two varieties, though growing vigorously and fertile with pollen from either of the other two plants, were utterly sterile with pollen from other flowers on the same plant.

Reseda odorata.

Having observed that certain individuals were self-sterile, I covered during the summer of 1868 seven plants under separate nets, and will call these plants A, B, C, D, E, F, G. They all appeared to be quite sterile with their own pollen, but fertile with that of any other plant.

Fourteen flowers on A were crossed with pollen from B or C, and produced thirteen fine capsules. Sixteen flowers were fertilised with pollen from other flowers on the same plant, but yielded not a single capsule.

Fourteen flowers on B were crossed with pollen from A, C or D, and all produced capsules; some of these were not very fine, yet they contained plenty of seeds. Eighteen flowers were fertilised with pollen from other flowers on the same plant, and produced not one capsule.

Ten flowers on C were crossed with pollen from A, B, D or E, and produced nine fine capsules. Nineteen flowers were fertilised with pollen from other flowers on the same plant, and produced no capsules.

Ten flowers on D were crossed with pollen from A, B, C or E, and produced nine fine capsules. Eighteen flowers were fertilised with pollen from other flowers on the same plant, and produced no capsules.

Seven flowers on E were crossed with pollen from A, C, or D, and all produced fine capsules. Eight flowers were fertilised with pollen from other flowers on the same plant, and produced no capsules.

On the plants F and G no flowers were crossed, but very many (number not recorded) were fertilised with pollen from other flowers on the same plants, and these did not produce a single capsule.

We thus see that fifty-five flowers on five of the above plants were reciprocally crossed in various ways; several flowers on each of these plants being fertilised with pollen from several of the other plants. These fifty-five flowers produced fifty-two capsules, almost all of which were of full size and contained an abundance of seeds. On the other hand, seventy-nine flowers (besides many others not recorded) were fertilised with pollen from other flowers on the same plants, and these did not produce a single capsule. In one case in which I examined the stigmas of the flowers fertilised with their own pollen, these were penetrated by the pollen-tubes, although such penetration produced no effect. Pollen falls generally, and I believe always, from the anthers on the stigmas of the same flower; yet only three out of the above seven protected plants produced spontaneously any capsules, and these it might have been thought must have been self-fertilised. There were altogether seven such capsules; but as they were all seated close to the artificially crossed flowers, I can hardly doubt that a few grains of foreign pollen had accidentally fallen on their stigmas. Besides the above seven plants, four others were kept covered under the SAME large net; and some of these produced here and there in the most capricious manner little groups of capsules; and this makes me believe that a bee, many of which settled on the outside of the net, being attracted by the odour, had on some one occasion found an entrance, and had intercrossed a few of the flowers.

In the spring of 1869 four plants raised from fresh seeds were carefully protected under separate nets; and now the result was widely different to what it was before. Three of these protected plants became actually loaded with capsules, especially during the early part of the summer; and this fact indicates that temperature produces some effect, but the experiment given in the following paragraph shows that the innate constitution of the plant is a far more important element. The fourth plant produced only a few capsules, many of them of small size; yet it was far more self-fertile than any of the seven plants tried during the previous year. The flowers on four small branches of this semi-self-sterile plant were smeared with pollen from one of the other plants, and they all produced fine capsules.

As I was much surprised at the difference in the results of the trials made during the two previous years, six fresh plants were protected by separate nets in the year 1870. Two of these proved almost completely self-sterile, for on carefully searching them I found only three small capsules, each containing either one or two seeds of small size, which, however, germinated. A few flowers on both these plants were reciprocally fertilised with each other's pollen, and a few with pollen from one of the following self-fertile plants, and all these flowers produced fine capsules. The four other plants whilst still remaining protected beneath the nets presented a wonderful contrast (though one of them in a somewhat less degree than the others), for they became actually covered with spontaneously self-fertilised capsules, as numerous as, or very nearly so, and as fine as those on the unprotected plants growing near.

The above three spontaneously self-fertilised capsules produced by the two almost completely self-sterile plants, contained altogether five seeds; and from these I raised in the following year (1871) five plants, which were kept under separate nets. They grew to an extraordinarily large size, and on August 29th were examined. At first sight they appeared entirely destitute of capsules; but on carefully searching their many branches, two or three capsules were found on three of the plants, half-a-dozen on the fourth, and about eighteen on the fifth plant. But all these capsules were small, some being empty; the greater number contained only a single seed, and very rarely more than one. After this examination the nets were taken off, and the bees immediately carried pollen from one of these almost self-sterile plants to the other, for no other plants grew near. After a few weeks the ends of the branches on all five plants became covered with capsules, presenting a curious contrast with the lower and naked parts of the same long branches. These five plants therefore inherited almost exactly the same sexual constitution as their parents; and without doubt a self-sterile race of Mignonette could have been easily established.

Reseda lutea.

Plants of this species were raised from seeds gathered from a group of wild plants growing at no great distance from my garden. After casually observing that some of these plants were self-sterile, two plants taken by hazard were protected under separate nets. One of these soon became covered with spontaneously self-fertilised capsules, as numerous as those on the surrounding unprotected plants; so that it was evidently quite self-fertile. The other plant was partially self-sterile, producing very few capsules, many of which were of small size. When, however, this plant had grown tall, the uppermost branches became pressed against the net and grew crooked, and in this position the bees were able to suck the flowers through the meshes, and brought pollen to them from the neighbouring plants. These branches then became loaded with capsules; the other and lower branches remaining almost bare. The sexual constitution of this species is therefore similar to that of Reseda odorata.

CONCLUDING REMARKS ON SELF-STERILE PLANTS.

In order to favour as far as possible the self-fertilisation of some of the foregoing plants, all the flowers on Reseda odorata and some of those on the Abutilon were fertilised with pollen from other flowers on the same plant, instead of with their own pollen, and in the case of the Senecio with pollen from other flowers on the same corymb; but this made no difference in the result. Fritz Muller tried both kinds of self-fertilisation in the case of Bignonia, Tabernaemontana and Abutilon, likewise with no difference in the result. With Eschscholtzia, however, he found that pollen from other flowers on the same plant was a little more effective than pollen from the same flower. So did Hildebrand in Germany; as thirteen out of fourteen flowers of Eschscholtzia thus fertilised set capsules, these containing on an average 9.5 seeds; whereas only fourteen flowers out of twenty-one fertilised with their own pollen set capsules, these containing on an average 9.0 seeds. (9/11. 'Pringsheim's Jahrbuch fur wiss. Botanik' 7 page 467.) Hildebrand found a trace of a similar difference with Corydalis cava, as did Fritz Muller with an Oncidium. (9/12. 'Variation under Domestication' chapter 17 2nd edition volume 2 pages 113-115.)

In considering the several cases above given of complete or almost complete self-sterility, we are first struck with their wide distribution throughout the vegetable kingdom. Their number is not at present large, for they can be discovered only by protecting plants from insects and then fertilising them with pollen from another plant of the same species and with their own pollen; and the latter must be proved to be in an efficient state by other trials. Unless all this be done, it is impossible to know whether their self-sterility may not be due to the male or female reproductive organs, or to both, having been affected by changed conditions of life. As in the course of my experiments I have found three new cases, and as Fritz Muller has observed indications of several others, it is probable that they will hereafter be proved to be far from rare. (9/13. Mr. Wilder, the editor of a horticultural journal in the United States quoted in 'Gardeners' Chronicle' 1868 page 1286, states that Lilium auratum, Impatiens pallida and fulva, and Forsythia viridissima, cannot be fertilised with their own pollen.)

As with plants of the same species and parentage, some individuals are self-sterile and others self-fertile, of which fact Reseda odorata offers the most striking instances, it is not at all surprising that species of the same genus differ in this same manner. Thus Verbascum phoeniceum and nigrum are self-sterile, whilst V. thapsus and lychnitis are quite self-fertile, as I know by trial. There is the same difference between some of the species of Papaver, Corydalis, and of other genera. Nevertheless, the tendency to self-sterility certainly runs to a certain extent in groups, as we see in the genus Passiflora, and with the Vandeae amongst Orchids.

Self-sterility differs much in degree in different plants. In those extraordinary cases in which pollen from the same flower acts on the stigma like a poison, it is almost certain that the plants would never yield a single self-fertilised seed. Other plants, like Corydalis cava, occasionally, though very rarely, produce a few self-fertilised seeds. A large number of species, as may be seen in Table 9/F, are less fertile with their own pollen than with that from another plant; and lastly, some species are perfectly self-fertile. Even with the individuals of the same species, as just remarked, some are utterly self-sterile, others moderately so, and some perfectly self-fertile. The cause, whatever it may be, which renders many plants more or less sterile with their own pollen, that is, when they are self-fertilised, must be different, at least to a certain extent, from that which determines the difference in height, vigour, and fertility of the seedlings raised from self-fertilised and crossed seeds; for we have already seen that the two classes of cases do not by any means run parallel. This want of parallelism would be intelligible, if it could be shown that self-sterility depended solely on the incapacity of the pollen-tubes to penetrate the stigma of the same flower deeply enough to reach the ovules; whilst the greater or less vigorous growth of the seedlings no doubt depends on the nature of the contents of the pollen-grains and ovules. Now it is certain that with some plants the stigmatic secretion does not properly excite the pollen-grains, so that the tubes are not properly developed, if the pollen is taken from the same flower. This is the case according to Fritz Muller with Eschscholtzia, for he found that the pollen-tubes did not penetrate the stigma deeply; and with the Orchidaceous genus Notylia they failed altogether to penetrate it. (9/14. 'Botanische Zeitung' 1868 pages 114, 115.)

With dimorphic and trimorphic species, an illegitimate union between plants of the same form presents the closest analogy with self-fertilisation, whilst a legitimate union closely resembles cross-fertilisation; and here again the lessened fertility or complete sterility of an illegitimate union depends, at least in part, on the incapacity for interaction between the pollen-grains and stigma. Thus with Linum grandiflorum, as I have elsewhere shown, not more than two or three out of hundreds of pollen-grains, either of the long-styled or short-styled form, when placed on the stigma of their own form, emit their tubes, and these do not penetrate deeply; nor does the stigma itself change colour, as occurs when it is legitimately fertilised. (9/15. 'Journal of the Linnean Society Botany' volume 7 1863 pages 73-75.)

On the other hand the difference in innate fertility, as well as in growth between plants raised from crossed and self-fertilised seeds, and the difference in fertility and growth between the legitimate and illegitimate offspring of dimorphic and trimorphic plants, must depend on some incompatibility between the sexual elements contained within the pollen-grains and ovules, as it is through their union that new organisms are developed.

If we now turn to the more immediate cause of self-sterility, we clearly see that in most cases it is determined by the conditions to which the plants have been subjected. Thus Eschscholtzia is completely self-sterile in the hot climate of Brazil, but is perfectly fertile there with the pollen of any other individual. The offspring of Brazilian plants became in England in a single generation partially self-fertile, and still more so in the second generation. Conversely, the offspring of English plants, after growing for two seasons in Brazil, became in the first generation quite self-sterile. Again, Abutilon darwinii, which is self-sterile in its native home of Brazil, became moderately self-fertile in a single generation in an English hothouse. Some other plants are self-sterile during the early part of the year, and later in the season become self-fertile. Passiflora alata lost its self-sterility when grafted on another species. With Reseda, however, in which some individuals of the same parentage are self-sterile and others are self-fertile, we are forced in our ignorance to speak of the cause as due to spontaneous variability; but we should remember that the progenitors of these plants, either on the male or female side, may have been exposed to somewhat different conditions. The power of the environment thus to affect so readily and in so peculiar a manner the reproductive organs, is a fact which has many important bearings; and I have therefore thought the foregoing details worth giving. For instance, the sterility of many animals and plants under changed conditions of life, such as confinement, evidently comes within the same general principle of the sexual system being easily affected by the environment. It has already been proved, that a cross between plants which have been self-fertilised or intercrossed during several generations, having been kept all the time under closely similar conditions, does not benefit the offspring; and on the other hand, that a cross between plants that have been subjected to different conditions benefits the offspring to an extraordinary degree. We may therefore conclude that some degree of differentiation in the sexual system is necessary for the full fertility of the parent-plants and for the full vigour of their offspring. It seems also probable that with those plants which are capable of complete self-fertilisation, the male and female elements and organs already differ to an extent sufficient to excite their mutual interaction; but that when such plants are taken to another country, and become in consequence self-sterile, their sexual elements and organs are so acted on as to be rendered too uniform for such interaction, like those of a self-fertilised plant long cultivated under the same conditions. Conversely, we may further infer that plants which are self-sterile in their native country, but become self-fertile under changed conditions, have their sexual elements so acted on, that they become sufficiently differentiated for mutual interaction.

We know that self-fertilised seedlings are inferior in many respects to those from a cross; and as with plants in a state of nature pollen from the same flower can hardly fail to be often left by insects or by the wind on the stigma, it seems at first sight highly probable that self-sterility has been gradually acquired through natural selection in order to prevent self-fertilisation. It is no valid objection to this belief that the structure of some flowers, and the dichogamous condition of many others, suffice to prevent the pollen reaching the stigma of the same flower; for we should remember that with most species many flowers expand at the same time, and that pollen from the same plant is equally injurious or nearly so as that from the same flower. Nevertheless, the belief that self-sterility is a quality which has been gradually acquired for the special purpose of preventing self-fertilisation must, I believe, be rejected. In the first place, there is no close correspondence in degree between the sterility of the parent-plants when self-fertilised, and the extent to which their offspring suffer in vigour by this process; and some such correspondence might have been expected if self-sterility had been acquired on account of the injury caused by self-fertilisation. The fact of individuals of the same parentage differing greatly in their degree of self-sterility is likewise opposed to such a belief; unless, indeed, we suppose that certain individuals have been rendered self-sterile to favour intercrossing, whilst other individuals have been rendered self-fertile to ensure the propagation of the species. The fact of self-sterile individuals appearing only occasionally, as in the case of Lobelia, does not countenance this latter view. But the strongest argument against the belief that self-sterility has been acquired to prevent self-fertilisation, is the immediate and powerful effect of changed conditions in either causing or in removing self-sterility. We are not therefore justified in admitting that this peculiar state of the reproductive system has been gradually acquired through natural selection; but we must look at it as an incidental result, dependent on the conditions to which the plants have been subjected, like the ordinary sterility caused in the case of animals by confinement, and in the case of plants by too much manure, heat, etc. I do not, however, wish to maintain that self-sterility may not sometimes be of service to a plant in preventing self-fertilisation; but there are so many other means by which this result might be prevented or rendered difficult, including as we shall see in the next chapter the prepotency of pollen from a distinct individual over a plant's own pollen, that self-sterility seems an almost superfluous acquirement for this purpose.

Finally, the most interesting point in regard to self-sterile plants is the evidence which they afford of the advantage, or rather of the necessity, of some degree or kind of differentiation in the sexual elements, in order that they should unite and give birth to a new being. It was ascertained that the five plants of Reseda odorata which were selected by chance, could be perfectly fertilised by pollen taken from any one of them, but not by their own pollen; and a few additional trials were made with some other individuals, which I have not thought worth recording. So again, Hildebrand and Fritz Muller frequently speak of self-sterile plants being fertile with the pollen of any other individual; and if there had been any exceptions to the rule, these could hardly have escaped their observation and my own. We may therefore confidently assert that a self-sterile plant can be fertilised by the pollen of any one out of a thousand or ten thousand individuals of the same species, but not by its own. Now it is obviously impossible that the sexual organs and elements of every individual can have been specialised with respect to every other individual. But there is no difficulty in believing that the sexual elements of each differ slightly in the same diversified manner as do their external characters; and it has often been remarked that no two individuals are absolutely alike. Therefore we can hardly avoid the conclusion, that differences of an analogous and indefinite nature in the reproductive system are sufficient to excite the mutual action of the sexual elements, and that unless there be such differentiation fertility fails.

THE APPEARANCE OF HIGHLY SELF-FERTILE VARIETIES.

We have just seen that the degree to which flowers are capable of being fertilised with their own pollen differs much, both with the species of the same genus, and sometimes with the individuals of the same species. Some allied cases of the appearance of varieties which, when self-fertilised, yield more seed and produce offspring growing taller than their self-fertilised parents, or than the intercrossed plants of the corresponding generation, will now be considered.

Firstly, in the third and fourth generations of Mimulus luteus, a tall variety, often alluded to, having large white flowers blotched with crimson, appeared amongst both the intercrossed and self-fertilised plants. It prevailed in all the later self-fertilised generations to the exclusion of every other variety, and transmitted its characters faithfully, but disappeared from the intercrossed plants, owing no doubt to their characters being repeatedly blended by crossing. The self-fertilised plants belonging to this variety were not only taller, but more fertile than the intercrossed plants; though these latter in the earlier generations were much taller and more fertile than the self-fertilised plants. Thus in the fifth generation the self-fertilised plants were to the intercrossed in height as 126 to 100. In the sixth generation they were likewise much taller and finer plants, but were not actually measured; they produced capsules compared with those on the intercrossed plants, in number, as 147 to 100; and the self-fertilised capsules contained a greater number of seeds. In the seventh generation the self-fertilised plants were to the crossed in height as 137 to 100; and twenty flowers on these self-fertilised plants fertilised with their own pollen yielded nineteen very fine capsules,—a degree of self-sterility which I have not seen equalled in any other case. This variety seems to have become specially adapted to profit in every way by self-fertilisation, although this process was so injurious to the parent-plants during the first four generations. It should however be remembered that seedlings raised from this variety, when crossed by a fresh stock, were wonderfully superior in height and fertility to the self-fertilised plants of the corresponding generation.

Secondly, in the sixth self-fertilised generation of Ipomoea a single plant named the Hero appeared, which exceeded by a little in height its intercrossed opponent,—a case which had not occurred in any previous generation. Hero transmitted the peculiar colour of its flowers, as well as its increased tallness and a high degree of self-fertility, to its children, grandchildren, and great-grandchildren. The self-fertilised children of Hero were in height to other self-fertilised plants of the same stock as 100 to 85. Ten self-fertilised capsules produced by the grandchildren contained on an average 5.2 seeds; and this is a higher average than was yielded in any other generation by the capsules of self-fertilised flowers. The great-grandchildren of Hero derived from a cross with a fresh stock were so unhealthy, from having been grown at an unfavourable season, that their average height in comparison with that of the self-fertilised plants cannot be judged of with any safety; but it did not appear that they had profited even by a cross of this kind.

Thirdly, the plants of Nicotiana on which I experimented appear to come under the present class of cases; for they varied in their sexual constitution and were more or less highly self-fertile. They were probably the offspring of plants which had been spontaneously self-fertilised under glass for several generations in this country. The flowers on the parent-plants which were first fertilised by me with their own pollen yielded half again as many seeds as did those which were crossed; and the seedlings raised from these self-fertilised seeds exceeded in height those raised from the crossed seeds to an extraordinary degree. In the second and third generations, although the self-fertilised plants did not exceed the crossed in height, yet their self-fertilised flowers yielded on two occasions considerably more seeds than the crossed flowers, even than those which were crossed with pollen from a distinct stock or variety.

Lastly, as certain individual plants of Reseda odorata and lutea are incomparably more self-fertile than other individuals, the former might be included under the present heading of the appearance of new and highly self-fertile varieties. But in this case we should have to look at these two species as normally self-sterile; and this, judging by my experience, appears to be the correct view.

We may therefore conclude from the facts now given, that varieties sometimes arise which when self-fertilised possess an increased power of producing seeds and of growing to a greater height, than the intercrossed or self-fertilised plants of the corresponding generation—all the plants being of course subjected to the same conditions. The appearance of such varieties is interesting, as it bears on the existence under nature of plants which regularly fertilise themselves, such as Ophrys apifera and a few other orchids, or as Leersia oryzoides, which produces an abundance of cleistogene flowers, but most rarely flowers capable of cross-fertilisation.

Some observations made on other plants lead me to suspect that self-fertilisation is in some respects beneficial; although the benefit thus derived is as a rule very small compared with that from a cross with a distinct plant. Thus we have seen in the last chapter that seedlings of Ipomoea and Mimulus raised from flowers fertilised with their own pollen, which is the strictest possible form of self-fertilisation, were superior in height, weight, and in early flowering to the seedlings raised from flowers crossed with pollen from other flowers on the same plant; and this superiority apparently was too strongly marked to be accidental. Again, the cultivated varieties of the common pea are highly self-fertile, although they have been self-fertilised for many generations; and they exceeded in height seedlings from a cross between two plants belonging to the same variety in the ratio of 115 to 100; but then only four pairs of plants were measured and compared. The self-fertility of Primula veris increased after several generations of illegitimate fertilisation, which is a process closely analogous to self-fertilisation, but only as long as the plants were cultivated under the same favourable conditions. I have also elsewhere shown that with Primula veris and sinensis, equal-styled varieties occasionally appear which possess the sexual organs of the two forms combined in the same flower. (9/16. 'Journal of the Linnean Society Botany' volume 10 1867 pages 417, 419.) Consequently they fertilise themselves in a legitimate manner and are highly self-fertile; but the remarkable fact is that they are rather more fertile than ordinary plants of the same species legitimately fertilised by pollen from a distinct individual. Formerly it appeared to me probable, that the increased fertility of these dimorphic plants might be accounted for by the stigma lying so close to the anthers that it was impregnated at the most favourable age and time of the day; but this explanation is not applicable to the above given cases, in which the flowers were artificially fertilised with their own pollen.

Considering the facts now adduced, including the appearance of those varieties which are more fertile and taller than their parents and than the intercrossed plants of the corresponding generation, it is difficult to avoid the suspicion that self-fertilisation is in some respects advantageous; though if this be really the case, any such advantage is as a rule quite insignificant compared with that from a cross with a distinct plant, and especially with one of a fresh stock. Should this suspicion be hereafter verified, it would throw light, as we shall see in the next chapter, on the existence of plants bearing small and inconspicuous flowers which are rarely visited by insects, and therefore are rarely intercrossed.

RELATIVE WEIGHT AND PERIOD OF GERMINATION OF SEEDS FROM CROSSED AND SELF-FERTILISED FLOWERS.

An equal number of seeds from flowers fertilised with pollen from another plant, and from flowers fertilised with their own pollen, were weighed, but only in sixteen cases. Their relative weights are given in the following list; that of the seeds from the crossed flowers being taken as 100.

Column 1: Name of Plant.

Column 2: x, in the expression, 100 to x.

Ipomoea purpurea (parent plants): 127. Ipomoea purpurea (third generation): 87. Salvia coccinea: 100. Brassica oleracea: 103. Iberis umbellata (second generation): 136. Delphinium consolida: 45. Hibiscus africanus: 105. Tropaeolum minus: 115. Lathyrus odoratus (about): 100. Sarothamnus scoparius: 88. Specularia speculum: 86. Nemophila insignis: 105. Borago officinalis: 111. Cyclamen persicum (about): 50. Fagopyrum esculentum: 82. Canna warscewiczi (3 generations): 102.

It is remarkable that in ten out of these sixteen cases the self-fertilised seeds were either superior or equal to the crossed in weight; nevertheless, in six out of the ten cases (namely, with Ipomoea, Salvia, Brassica, Tropaeolum, Lathyrus, and Nemophila) the plants raised from these self-fertilised seeds were very inferior in height and in other respects to those raised from the crossed seeds. The superiority in weight of the self-fertilised seeds in at least six out of the ten cases, namely, with Brassica, Hibiscus, Tropaeolum, Nemophila, Borago, and Canna, may be accounted for in part by the self-fertilised capsules containing fewer seeds; for when a capsule contains only a few seeds, these will be apt to be better nourished, so as to be heavier, than when many are contained in the same capsule. It should, however, be observed that in some of the above cases, in which the crossed seeds were the heaviest, as with Sarothamnus and Cyclamen, the crossed capsules contained a larger number of seeds. Whatever may be the explanation of the self-fertilised seeds being often the heaviest, it is remarkable in the case of Brassica, Tropaeolum, Nemophila, and of the first generation of Ipomoea, that the seedlings raised from them were inferior in height and in other respects to the seedlings raised from the crossed seeds. This fact shows how superior in constitutional vigour the crossed seedlings must have been, for it cannot be doubted that heavy and fine seeds tend to yield the finest plants. Mr. Galton has shown that this holds good with Lathyrus odoratus; as has Mr. A.J. Wilson with the Swedish turnip, Brassica campestris ruta baga. Mr. Wilson separated the largest and smallest seeds of this latter plant, the ratio between the weights of the two lots being as 100 to 59, and he found that the seedlings "from the larger seeds took the lead and maintained their superiority to the last, both in height and thickness of stem." (9/17. 'Gardeners' Chronicle' 1867 page 107. Loiseleur-Deslongchamp 'Les Cereales' 1842 pages 208-219, was led by his observations to the extraordinary conclusion that the smaller grains of cereals produce as fine plants as the large. This conclusion is, however, contradicted by Major Hallet's great success in improving wheat by the selection of the finest grains. It is possible, however, that man, by long-continued selection, may have given to the grains of the cereals a greater amount of starch or other matter, than the seedlings can utilise for their growth. There can be little doubt, as Humboldt long ago remarked, that the grains of cereals have been rendered attractive to birds in a degree which is highly injurious to the species.) Nor can this difference in the growth of the seedling turnips be attributed to the heavier seeds having been of crossed, and the lighter of self-fertilised origin, for it is known that plants belonging to this genus are habitually intercrossed by insects.

With respect to the relative period of germination of crossed and self-fertilised seeds, a record was kept in only twenty-one cases; and the results are very perplexing. Neglecting one case in which the two lots germinated simultaneously, in ten cases or exactly one-half many of the self-fertilised seeds germinated before the crossed, and in the other half many of the crossed before the self-fertilised. In four out of these twenty cases, seeds derived from a cross with a fresh stock were compared with self-fertilised seeds from one of the later self-fertilised generations; and here again in half the cases the crossed seeds, and in the other half the self-fertilised seeds, germinated first. Yet the seedlings of Mimulus raised from such self-fertilised seeds were inferior in all respects to the crossed seedlings, and in the case of Eschscholtzia they were inferior in fertility. Unfortunately the relative weight of the two lots of seeds was ascertained in only a few instances in which their germination was observed; but with Ipomoea and I believe with some of the other species, the relative lightness of the self-fertilised seeds apparently determined their early germination, probably owing to the smaller mass being favourable to the more rapid completion of the chemical and morphological changes necessary for germination. On the other hand, Mr. Galton gave me seeds (no doubt all self-fertilised) of Lathyrus odoratus, which were divided into two lots of heavier and lighter seeds; and several of the former germinated first. It is evident that many more observations are necessary before anything can be decided with respect to the relative period of germination of crossed and self-fertilised seeds.



CHAPTER X.

MEANS OF FERTILISATION.

Sterility and fertility of plants when insects are excluded. The means by which flowers are cross-fertilised. Structures favourable to self-fertilisation. Relation between the structure and conspicuousness of flowers, the visits of insects, and the advantages of cross-fertilisation. The means by which flowers are fertilised with pollen from a distinct plant. Greater fertilising power of such pollen. Anemophilous species. Conversion of anemophilous species into entomophilous. Origin of nectar. Anemophilous plants generally have their sexes separated. Conversion of diclinous into hermaphrodite flowers. Trees often have their sexes separated.

In the introductory chapter I briefly specified the various means by which cross-fertilisation is favoured or ensured, namely, the separation of the sexes,—the maturity of the male and female sexual elements at different periods,—the heterostyled or dimorphic and trimorphic condition of certain plants,—many mechanical contrivances,—the more or less complete inefficiency of a flower's own pollen on the stigma,—and the prepotency of pollen from any other individual over that from the same plant. Some of these points require further consideration; but for full details I must refer the reader to the several excellent works mentioned in the introduction. I will in the first place give two lists: the first, of plants which are either quite sterile or produce less than about half the full complement of seeds, when insects are excluded; and a second list of plants which, when thus treated, are fully fertile or produce at least half the full complement of seeds. These lists have been compiled from the several previous tables, with some additional cases from my own observations and those of others. The species are arranged nearly in the order followed by Lindley in his 'Vegetable Kingdom.' The reader should observe that the sterility or fertility of the plants in these two lists depends on two wholly distinct causes; namely, the absence or presence of the proper means by which pollen is applied to the stigma, and its less or greater efficiency when thus applied. As it is obvious that with plants in which the sexes are separate, pollen must be carried by some means from flower to flower, such species are excluded from the lists; as are likewise dimorphic and trimorphic plants, in which the same necessity occurs to a limited extent. Experience has proved to me that, independently of the exclusion of insects, the seed-bearing power of a plant is not lessened by covering it while in flower under a thin net supported on a frame; and this might indeed have been inferred from the consideration of the two following lists, as they include a considerable number of species belonging to the same genera, some of which are quite sterile and others quite fertile when protected by a net from the access of insects.

[LIST OF PLANTS WHICH, WHEN INSECTS ARE EXCLUDED, ARE EITHER QUITE STERILE, OR PRODUCE, AS FAR AS I COULD JUDGE, LESS THAN HALF THE NUMBER OF SEEDS PRODUCED BY UNPROTECTED PLANTS.

Passiflora alata, racemosa, coerulea, edulis, laurifolia, and some individuals of P. quadrangularis (Passifloraceae), are quite sterile under these conditions: see 'Variation of Animals and Plants under Domestication' chapter 17 2nd edition volume 2 page 118.

Viola canina (Violaceae).—Perfect flowers quite sterile unless fertilised by bees, or artificially fertilised.

Viola tricolor.—Sets very few and poor capsules.

Reseda odorata (Resedaceae).—Some individuals quite sterile.

Reseda lutea.—Some individuals produce very few and poor capsules.

Abutilon darwinii (Malvaceae).—Quite sterile in Brazil: see previous discussion on self-sterile plants.

Nymphaea (Nymphaeaceae).—Professor Caspary informs me that some of the species are quite sterile if insects are excluded.

Euryale amazonica (Nymphaeaceae).—Mr. J. Smith, of Kew, informs me that capsules from flowers left to themselves, and probably not visited by insects, contained from eight to fifteen seeds; those from flowers artificially fertilised with pollen from other flowers on the same plant contained from fifteen to thirty seeds; and that two flowers fertilised with pollen brought from another plant at Chatsworth contained respectively sixty and seventy-five seeds. I have given these statements because Professor Caspary advances this plant as a case opposed to the doctrine of the necessity or advantage of cross-fertilisation: see Sitzungsberichte der Phys.-okon. Gesell.zu Konigsberg, B.6 page 20.)

Delphinium consolida (Ranunculaceae).—Produces many capsules, but these contain only about half the number of seeds compared with capsules from flowers naturally fertilised by bees.

Eschscholtzia californica (Papaveraceae).—Brazilian plants quite sterile: English plants produce a few capsules.

Papaver vagum (Papaveraceae).—In the early part of the summer produced very few capsules, and these contained very few seeds.

Papaver alpinum.—H. Hoffmann ('Speciesfrage' 1875 page 47) states that this species produced seeds capable of germination only on one occasion.

Corydalis cava (Fumariaceae).—Sterile: see the previous discussion on self-sterile plants.

Corydalis solida.—I had a single plant in my garden (1863), and saw many hive-bees sucking the flowers, but not a single seed was produced. I was much surprised at this fact, as Professor Hildebrand's discovery that C. cava is sterile with its own pollen had not then been made. He likewise concludes from the few experiments which he made on the present species that it is self-sterile. The two foregoing cases are interesting, because botanists formerly thought (see, for instance, Lecoq, 'De la Fecondation et de l'Hybridation' 1845 page 61 and Lindley 'Vegetable Kingdom' 1853 page 436) that all the species of the Fumariaceae were specially adapted for self-fertilisation.

Corydalis lutea.—A covered-up plant produced (1861) exactly half as many capsules as an exposed plant of the same size growing close alongside. When humble-bees visit the flowers (and I repeatedly saw them thus acting) the lower petals suddenly spring downwards and the pistil upwards; this is due to the elasticity of the parts, which takes effect, as soon as the coherent edges of the hood are separated by the entrance of an insect. Unless insects visit the flowers the parts do not move. Nevertheless, many of the flowers on the plants which I had protected produced capsules, notwithstanding that their petals and pistils still retained their original position; and I found to my surprise that these capsules contained more seeds than those from flowers, the petals of which had been artificially separated and allowed to spring apart. Thus, nine capsules produced by undisturbed flowers contained fifty-three seeds; whilst nine capsules from flowers, the petals of which had been artificially separated, contained only thirty-two seeds. But we should remember that if bees had been permitted to visit these flowers, they would have visited them at the best time for fertilisation. The flowers, the petals of which had been artificially separated, set their capsules before those which were left undisturbed under the net. To show with what certainty the flowers are visited by bees, I may add that on one occasion all the flowers on some unprotected plants were examined, and every single one had its petals separated; and, on a second occasion, forty-one out of forty-three flowers were in this state. Hildebrand states (Pring. Jahr. f. wiss. Botanik, B. 7 page 450) that the mechanism of the parts in this species is nearly the same as in C. ochroleuca, which he has fully described.

Hypecoum grandiflorum (Fumariaceae).—Highly self-sterile (Hildebrand, ibid.).

Kalmia latifolia (Ericaceae).—Mr. W.J. Beal says ('American Naturalist' 1867) that flowers protected from insects wither and drop off, with "most of the anthers still remaining in the pockets."

Pelargonium zonale (Geraniaceae).—Almost sterile; one plant produced two fruits. It is probable that different varieties would differ in this respect, as some are only feebly dichogamous.

Dianthus caryophyllus (Caryophyllaceae).—Produces very few capsules which contain any good seeds.

Phaseolus multiflorus (Leguminosae).—Plants protected from insects produced on two occasions about one-third and one-eighth of the full number of seeds: see my article in 'Gardeners' Chronicle' 1857 page 225 and 1858 page 828; also 'Annals and Magazine of Natural History' 3rd series volume 2 1858 page 462. Dr. Ogle ('Popular Science Review' 1870 page 168) found that a plant was quite sterile when covered up. The flowers are not visited by insects in Nicaragua, and, according to Mr. Belt, the species is there quite sterile: 'The Naturalist in Nicaragua' page 70.

Vicia faba (Leguminosae).—Seventeen covered-up plants yielded 40 beans, whilst seventeen plants left unprotected and growing close alongside produced 135 beans; these latter plants were, therefore, between three and four times more fertile than the protected plants: see 'Gardeners' Chronicle' for fuller details, 1858 page 828.

Erythrina (sp.?) (Leguminosae).—Sir W. MacArthur informed me that in New South Wales the flowers do not set, unless the petals are moved in the same manner as is done by insects.

Lathyrus grandiflorus (Leguminosae).—Is in this country more or less sterile. It never sets pods unless the flowers are visited by humble-bees (and this happens only rarely), or unless they are artificially fertilised: see my article in 'Gardeners' Chronicle' 1858 page 828.

Sarothamnus scoparius (Leguminosae).—Extremely sterile when the flowers are neither visited by bees, nor disturbed by being beaten by the wind against the surrounding net.

Melilotus officinalis (Leguminosae).—An unprotected plant visited by bees produced at least thirty times more seeds than a protected one. On this latter plant many scores of racemes did not produce a single pod; several racemes produced each one or two pods; five produced three; six produced four; and one produced six pods. On the unprotected plant each of several racemes produced fifteen pods; nine produced between sixteen and twenty-two pods, and one produced thirty pods.

Lotus corniculatus (Leguminosae).—Several covered-up plants produced only two empty pods, and not a single good seed.

Trifolium repens (Leguminosae).—Several plants were protected from insects, and the seeds from ten flowers-heads on these plants, and from ten heads on other plants growing outside the net (which I saw visited by bees), were counted; and the seeds from the latter plants were very nearly ten times as numerous as those from the protected plants. The experiment was repeated on the following year; and twenty protected heads now yielded only a single aborted seed, whilst twenty heads on the plants outside the net (which I saw visited by bees) yielded 2290 seeds, as calculated by weighing all the seed, and counting the number in a weight of two grains.

Trifolium pratense.—One hundred flower-heads on plants protected by a net did not produce a single seed, whilst 100 heads on plants growing outside, which were visited by bees, yielded 68 grains weight of seeds; and as eighty seeds weighed two grains, the 100 heads must have yielded 2720 seeds. I have often watched this plant, and have never seen hive-bees sucking the flowers, except from the outside through holes bitten by humble-bees, or deep down between the flowers, as if in search of some secretion from the calyx, almost in the same manner as described by Mr. Farrer, in the case of Coronilla ('Nature' 1874 July 2 page 169). I must, however, except one occasion, when an adjoining field of sainfoin (Hedysarum onobrychis) had just been cut down, and when the bees seemed driven to desperation. On this occasion most of the flowers of the clover were somewhat withered, and contained an extraordinary quantity of nectar, which the bees were able to suck. An experienced apiarian, Mr. Miner, says that in the United States hive-bees never suck the red clover; and Mr. R. Colgate informs me that he has observed the same fact in New Zealand after the introduction of the hive-bee into that island. On the other hand, H. Muller ('Befruchtung' page 224) has often seen hive-bees visiting this plant in Germany, for the sake both of pollen and nectar, which latter they obtained by breaking apart the petals. It is at least certain that humble-bees are the chief fertilisers of the common red clover.

Trifolium incarnatum.—The flower-heads containing ripe seeds, on some covered and uncovered plants, appeared equally fine, but this was a false appearance; 60 heads on the latter yielded 349 grains weight of seeds, whereas 60 on the covered-up plants yielded only 63 grains, and many of the seeds in the latter lot were poor and aborted. Therefore the flowers which were visited by bees produced between five and six times as many seeds as those which were protected. The covered-up plants not having been much exhausted by seed-bearing, bore a second considerable crop of flower-stems, whilst the exposed plants did not do so.

Cytisus laburnum (Leguminosae).—Seven flower-racemes ready to expand were enclosed in a large bag made of net, and they did not seem in the least injured by this treatment. Only three of them produced any pods, each a single one; and these three pods contained one, four, and five seeds. So that only a single pod from the seven racemes included a fair complement of seeds.

Cuphea purpurea (Lythraceae).—Produced no seeds. Other flowers on the same plant artificially fertilised under the net yielded seeds.

Vinca major (Apocynaceae).—Is generally quite sterile, but sometimes sets seeds when artificially cross-fertilised: see my notice 'Gardeners' Chronicle' 1861 page 552.

Vinca rosea.—Behaves in the same manner as the last species: 'Gardeners' Chronicle' 1861 page 699, 736, 831.

Tabernaemontana echinata (Apocynaceae).—Quite sterile.

Petunia violacea (Solanaceae).—Quite sterile, as far as I have observed.

Solanum tuberosum (Solanaceae).—Tinzmann says ('Gardeners' Chronicle' 1846 page 183) that some varieties are quite sterile unless fertilised by pollen from another variety.

Primula scotica (Primulaceae).—A non-dimorphic species, which is fertile with its own pollen, but is extremely sterile if insects are excluded. J. Scott in 'Journal of the Linnean Society Botany' volume 8 1864 page 119.

Cortusa matthioli (Primulaceae).—Protected plants completely sterile; artificially self-fertilised flowers perfectly fertile. J. Scott ibid. page 84.

Cyclamen persicum (Primulaceae).—During one season several covered-up plants did not produce a single seed.

Borago officinalis (Boraginaceae).—Protected plants produced about half as many seeds as the unprotected.

Salvia tenori (Labiatae).—Quite sterile; but two or three flowers on the summits of three of the spikes, which touched the net when the wind blew, produced a few seeds. This sterility was not due to the injurious effects of the net, for I fertilised five flowers with pollen from an adjoining plant, and these all yielded fine seeds. I removed the net, whilst one little branch still bore a few not completely faded flowers, and these were visited by bees and yielded seeds.

Salvia coccinea.—Some covered-up plants produced a good many fruits, but not, I think, half as many as did the uncovered plants; twenty-eight of the fruits spontaneously produced by the protected plant contained on an average only 1.45 seeds, whilst some artificially self-fertilised fruits on the same plant contained more than twice as many, namely 3.3 seeds.

Bignonia (unnamed species) (Bignoniaceae).—Quite sterile: see my account of self-sterile plants.

Digitalis purpurea (Scrophulariaceae).—Extremely sterile, only a few poor capsules being produced.

Linaria vulgaris (Scrophulariaceae).—Extremely sterile.

Antirrhinum majus, red var. (Scrophulariaceae).—Fifty pods gathered from a large plant under a net contained 9.8 grains weight of seeds; but many (unfortunately not counted) of the fifty pods contained no seeds. Fifty pods on a plant fully exposed to the visits of humble-bees contained 23.1 grains weight of seed, that is, more than twice the weight; but in this case again, several of the fifty pods contained no seeds.

Antirrhinum majus (white var., with a pink mouth to the corolla).—Fifty pods, of which only a very few were empty, on a covered-up plant contained 20 grains weight of seed; so that this variety seems to be much more self-fertile than the previous one. With Dr. W. Ogle ('Popular Science Review' January 1870 page 52) a plant of this species was much more sterile when protected from insects than with me, for it produced only two small capsules. As showing the efficiency of bees, I may add that Mr. Crocker castrated some young flowers and left them uncovered; and these produced as many seeds as the unmutilated flowers.

Antirrhinum majus (peloric var.).—This variety is quite fertile when artificially fertilised with its own pollen, but is utterly sterile when left to itself and uncovered, as humble-bees cannot crawl into the narrow tubular flowers.

Verbascum phoeniceum (Scrophulariaceae).—Quite sterile. See my account of self-sterile plants.

Verbascum nigrum.—Quite sterile. See my account of self-sterile plants.

Campanula carpathica (Lobeliaceae).—Quite sterile.

Lobelia ramosa (Lobeliaceae).—Quite sterile.

Lobelia fulgens.—This plant is never visited in my garden by bees, and is quite sterile; but in a nursery-garden at a few miles' distance I saw humble-bees visiting the flowers, and they produced some capsules.

Isotoma (a white-flowered var.) (Lobeliaceae).—Five plants left unprotected in my greenhouse produced twenty-four fine capsules, containing altogether 12.2 grains weight of seed, and thirteen other very poor capsules, which were rejected. Five plants protected from insects, but otherwise exposed to the same conditions as the above plants, produced sixteen fine capsules, and twenty other very poor and rejected ones. The sixteen fine capsules contained seeds by weight in such proportion that twenty-four would have yielded 4.66 grains. So that the unprotected plants produced nearly thrice as many seeds by weight as the protected plants.

Leschenaultia formosa (Goodeniaceae).—Quite sterile. My experiments on this plant, showing the necessity of insect aid, are given in the 'Gardeners' Chronicle' 1871 page 1166.

Senecio cruentus (Compositae).—Quite sterile: see my account of self-sterile plants.

Heterocentron mexicanum (Malastomaceae).—Quite sterile; but this species and the following members of the group produce plenty of seed when artificially self-fertilised.

Rhexia glandulosa (Melastomaceae).—Set spontaneously only two or three capsules.

Centradenia floribunda (Melastomaceae).—During some years produced spontaneously two or three capsules, sometimes none.

Pleroma (unnamed species from Kew) (Melastomaceae).—During some years produced spontaneously two or three capsules, sometimes none.

Monochaetum ensiferum (Melastomaceae).—During some years produced spontaneously two or three capsules, sometimes none.

Hedychium (unnamed species) (Marantaceae).—Almost self-sterile without aid.

Orchideae.—An immense proportion of the species sterile, if insects are excluded.

LIST OF PLANTS, WHICH WHEN PROTECTED FROM INSECTS ARE EITHER QUITE FERTILE, OR YIELD MORE THAN HALF THE NUMBER OF SEEDS PRODUCED BY UNPROTECTED PLANTS.

Passiflora gracilis (Passifloraceae).—Produces many fruits, but these contain fewer seeds than fruits from intercrossed flowers.

Brassica oleracea (Cruciferae).—Produces many capsules, but these generally not so rich in seed as those on uncovered plants.

Raphanus sativus (Cruciferae).—Half of a large branching plant was covered by a net, and was as thickly covered with capsules as the other and unprotected half; but twenty of the capsules on the latter contained on an average 3.5 seeds, whilst twenty of the protected capsules contained only 1.85 seeds, that is, only a little more than half the number. This plant might perhaps have been more properly included in the former list.

Iberis umbellata (Cruciferae).—Highly fertile.

Iberis amara.—Highly fertile.

Reseda odorata and lutea (Resedaceae).—Certain individuals completely self-fertile.

Euryale ferox (Nymphaeaceae).—Professor Caspary informs me that this plant is highly self-fertile when insects are excluded. He remarks in the paper before referred to, that his plants (as well as those of the Victoria regia) produce only one flower at a time; and that as this species is an annual, and was introduced in 1809, it must have been self-fertilised for the last fifty-six generations; but Dr. Hooker assures me that to his knowledge it has been repeatedly introduced, and that at Kew the same plant both of the Euryale and of the Victoria produce several flowers at the same time.

Nymphaea (Nymphaeaceae).—Some species, as I am informed by Professor Caspary, are quite self-fertile when insects are excluded.

Adonis aestivalis (Ranunculaceae).—Produces, according to Professor H. Hoffmann ('Speciesfrage' page 11), plenty of seeds when protected from insects.

Ranunculus acris (Ranunculaceae).—Produces plenty of seeds under a net.

Papaver somniferum (Papaveraceae).—Thirty capsules from uncovered plants yielded 15.6 grains weight of seed, and thirty capsules from covered-up plants, growing in the same bed, yielded 16.5 grains weight; so that the latter plants were more productive than the uncovered. Professor H. Hoffmann ('Speciesfrage' 1875 page 53) also found this species self-fertile when protected from insects.

Papaver vagum.—Produced late in the summer plenty of seeds, which germinated well.

Papaver argemonoides.—According to Hildebrand ('Jahrbuch fur w. Bot.' B.7 page 466), spontaneously self-fertilised flowers are by no means sterile.

Glaucium luteum (Papaveraceae).—According to Hildebrand ('Jahrbuch fur w. Bot.' B.7 page 466), spontaneously self-fertilised flowers are by no means sterile.

Argemone ochroleuca (Papaveraceae).—According to Hildebrand ('Jahrbuch fur w. Bot.' B.7 page 466), spontaneously self-fertilised flowers are by no means sterile.

Adlumia cirrhosa (Fumariaceae).—Sets an abundance of capsules.

Hypecoum procumbens (Fumariaceae).—Hildebrand says (idem), with respect to protected flowers, that "eine gute Fruchtbildung eintrete."

Fumaria officinalis (Fumariaceae).—Covered-up and unprotected plants apparently produced an equal number of capsules, and the seeds of the former seemed to the eye equally good. I have often watched this plant, and so has Hildebrand, and we have never seen an insect visit the flowers. Hermann Muller has likewise been struck with the rarity of the visits of insects to it, though he has sometimes seen hive-bees at work. The flowers may perhaps be visited by small moths, as is probably the case with the following species.

Fumaria capreolata.—Several large beds of this plant growing wild were watched by me during many days, but the flowers were never visited by any insects, though a humble-bee was once seen closely to inspect them. Nevertheless, as the nectary contains much nectar, especially in the evening, I felt convinced that they were visited, probably by moths. The petals do not naturally separate or open in the least; but they had been opened by some means in a certain proportion of the flowers, in the same manner as follows when a thick bristle is pushed into the nectary; so that in this respect they resemble the flowers of Corydalis lutea. Thirty-four heads, each including many flowers, were examined, and twenty of them had from one to four flowers, whilst fourteen had not a single flower thus opened. It is therefore clear that some of the flowers had been visited by insects, while the majority had not; yet almost all produced capsules.

Linum usitatissimum (Linaceae).—Appears to be quite fertile. H. Hoffmann 'Botanische Zeitung' 1876 page 566.

Impatiens barbigerum (Balsaminaceae).—The flowers, though excellently adapted for cross-fertilisation by the bees which freely visit them, set abundantly under a net.

Impatiens noli-me-tangere (Balsaminaceae).—This species produces cleistogene and perfect flowers. A plant was covered with a net, and some perfect flowers, marked with threads, produced eleven spontaneously self-fertilised capsules, which contained on an average 3.45 seeds. I neglected to ascertain the number of seeds produced by perfect flowers exposed to the visits of insects, but I believe it is not greatly in excess of the above average. Mr. A.W. Bennett has carefully described the structure of the flowers of I. fulva in 'Journal of the Linnean Society' volume 13 Bot. 1872 page 147. This latter species is said to be sterile with its own pollen ('Gardeners' Chronicle' 1868 page 1286), and if so, it presents a remarkable contrast with I. barbigerum and noli-me-tangere.

Limnanthes douglasii (Geraniaceae).—Highly fertile.

Viscaria oculata (Caryophyllaceae).—Produces plenty of capsules with good seeds.

Stellaria media (Caryophyllaceae).—Covered-up and uncovered plants produced an equal number of capsules, and the seeds in both appeared equally numerous and good.

Beta vulgaris (Chenopodiaceae).—Highly self-fertile.

Vicia sativa (Leguminosae).—Protected and unprotected plants produced an equal number of pods and equally fine seeds. If there was any difference between the two lots, the covered-up plants were the most productive.

Vicia hirsuta.—This species bears the smallest flowers of any British leguminous plant. The result of covering up plants was exactly the same as in the last species.

Pisum sativum (Leguminosae).—Fully fertile.

Lathyrus odoratus (Leguminosae).—Fully fertile.

Lathyrus nissolia.—Fully fertile.

Lupinus luteus (Leguminosae).—Fairly productive.

Lupinus pilosus.—Produced plenty of pods.

Ononis minutissima (Leguminosae).—Twelve perfect flowers on a plant under a net were marked by threads, and produced eight pods, containing on an average 2.38 seeds. Pods produced by flowers visited by insects would probably have contained on an average 3.66 seeds, judging from the effects of artificial cross-fertilisation.

Phaseolus vulgaris (Leguminosae).—Quite fertile.

Trifolium arvense (Leguminosae).—The excessively small flowers are incessantly visited by hive and humble-bees. When insects were excluded the flower-heads seemed to produce as many and as fine seeds as the exposed heads.

Trifolium procumbens.—On one occasion covered-up plants seemed to yield as many seeds as the uncovered. On a second occasion sixty uncovered flower-heads yielded 9.1 grains weight of seeds, whilst sixty heads on protected plants yielded no less than 17.7 grains; so that these latter plants were much more productive; but this result I suppose was accidental. I have often watched this plant, and have never seen the flowers visited by insects; but I suspect that the flowers of this species, and more especially of Trifolium minus, are frequented by small nocturnal moths which, as I hear from Mr. Bond, haunt the smaller clovers.

Medicago lupulina (Leguminosae).—On account of the danger of losing the seeds, I was forced to gather the pods before they were quite ripe; 150 flower-heads on plants visited by bees yielded pods weighing 101 grains; whilst 150 heads on protected plants yielded pods weighing 77 grains. The inequality would probably have been greater if the mature seeds could have been all safely collected and compared. Ig. Urban (Keimung, Bluthen, etc., bei Medicago 1873) has described the means of fertilisation in this genus, as has the Reverend G. Henslow in the 'Journal of the Linnean Society Botany' volume 9 1866 pages 327 and 355.

Nicotiana tabacum (Solanaceae).—Fully self-fertile.

Ipomoea purpurea (Convolvulaceae).—Highly self-fertile.

Leptosiphon androsaceus (Polemoniacae).—Plants under a net produced a good many capsules.

Primula mollis (Primulaceae).—A non-dimorphic species, self-fertile: J. Scott, in 'Journal of the Linnean Society Botany' volume 8 1864 page 120.

Nolana prostrata (Nolanaceae).—Plants covered up in the greenhouse, yielded seeds by weight compared with uncovered plants, the flowers of which were visited by many bees, in the ratio of 100 to 61.

Ajuga reptans (Labiatae).—Set a good many seeds; but none of the stems under a net produced so many as several uncovered stems growing closely by.

Euphrasia officinalis (Scrophulariaceae).—Covered-up plants produced plenty of seed; whether less than the exposed plants I cannot say. I saw two small Dipterous insects (Dolichopos nigripennis and Empis chioptera) repeatedly sucking the flowers; as they crawled into them, they rubbed against the bristles which project from the anthers, and became dusted with pollen.

Veronica agrestis (Scrophulariaceae).—Covered-up plants produced an abundance of seeds. I do not know whether any insects visit the flowers; but I have observed Syrphidae repeatedly covered with pollen visiting the flowers of V. hederaefolia and chamoedrys.

Mimulus luteus (Scrophulariaceae).—Highly self-fertile.

Calceolaria (greenhouse variety) (Scrophulariaceae).—Highly self-fertile.

Verbascum thapsus (Scrophulariaceae).—Highly self-fertile.

Verbascum lychnitis.—Highly self-fertile.

Vandellia nummularifolia (Scrophulariaceae).—Perfect flowers produce a good many capsules.

Bartsia odontites (Scrophulariaceae).—Covered-up plants produced a good many seeds; but several of these were shrivelled, nor were they so numerous as those produced by unprotected plants, which were incessantly visited by hive and humble-bees.

Specularia speculum (Lobeliaceae).—Covered plants produced almost as many capsules as the uncovered.

Lactuca sativa (Compositae).—Covered plants produced some seeds, but the summer was wet and unfavourable.

Galium aparine (Rubiaceae).—Covered plants produced quite as many seeds as the uncovered.

Apium petroselinum (Umbelliferae).—Covered plants apparently were as productive as the uncovered.

Zea mays (Gramineae).—A single plant in the greenhouse produced a good many grains.

Canna warscewiczi (Marantaceae).—Highly self-fertile.

Orchidaceae.—In Europe Ophrys apifera is as regularly self-fertilised as is any cleistogene flower. In the United States, South Africa, and Australia there are a few species which are perfectly self-fertile. These several cases are given in the second edition of my work on the Fertilisation of Orchids.

Allium cepa (blood red var.) (Liliaceae).—Four flower-heads were covered with a net, and they produced somewhat fewer and smaller capsules than those on the uncovered heads. The capsules were counted on one uncovered head, and were 289 in number; whilst those on a fine head from under the net were only 199.]

Each of these lists contains by a mere accident the same number of genera, namely, forty-nine. The genera in the first list include sixty-five species, and those in the second sixty species; the Orchideae in both being excluded. If the genera in this latter order, as well as in the Asclepiadae and Apocynaceae, had been included, the number of species which are sterile if insects are excluded would have been greatly increased; but the lists are confined to species which were actually experimented on. The results can be considered as only approximately accurate, for fertility is so variable a character, that each species ought to have been tried many times. The above number of species, namely, 125, is as nothing to the host of living plants; but the mere fact of more than half of them being sterile within the specified degree, when insects are excluded, is a striking one; for whenever pollen has to be carried from the anthers to the stigma in order to ensure full fertility, there is at least a good chance of cross-fertilisation. I do not, however, believe that if all known plants were tried in the same manner, half would be found to be sterile within the specified limits; for many flowers were selected for experiment which presented some remarkable structure; and such flowers often require insect-aid. Thus out of the forty-nine genera in the first list, about thirty-two have flowers which are asymmetrical or present some remarkable peculiarity; whilst in the second list, including species which are fully or moderately fertile when insects were excluded, only about twenty-one out of the forty-nine are asymmetrical or present any remarkable peculiarity.

MEANS OF CROSS-FERTILISATION.

The most important of all the means by which pollen is carried from the anthers to the stigma of the same flower, or from flower to flower, are insects, belonging to the orders of Hymenoptera, Lepidoptera, and Diptera; and in some parts of the world, birds. (10/1. I will here give all the cases known to me of birds fertilising flowers. In South Brazil, humming-birds certainly fertilise the various species of Abutilon, which are sterile without their aid (Fritz Muller 'Jenaische Zeitschrift f. Naturwiss.' B. 7 1872 page 24.) Long-beaked humming-birds visit the flowers of Brugmansia, whilst some of the short-beaked species often penetrate its large corolla in order to obtain the nectar in an illegitimate manner, in the same manner as do bees in all parts of the world. It appears, indeed, that the beaks of humming-birds are specially adapted to the various kinds of flowers which they visit: on the Cordillera they suck the Salviae, and lacerate the flowers of the Tacsoniae; in Nicaragua, Mr. Belt saw them sucking the flowers of Marcgravia and Erythina, and thus they carried pollen from flower to flower. In North America they are said to frequent the flowers of Impatiens: (Gould 'Introduction to the Trochilidae' 1861 pages 15, 120; 'Gardeners' Chronicle' 1869 page 389; 'The Naturalist in Nicaragua' page 129; 'Journal of the Linnean Society Botany' volume 13 1872 page 151.) I may add that I often saw in Chile a Mimus with its head yellow with pollen from, as I believe, a Cassia. I have been assured that at the Cape of Good Hope, Strelitzia is fertilised by the Nectarinidae. There can hardly be a doubt that many Australian flowers are fertilised by the many honey-sucking birds of that country. Mr. Wallace remarks (address to the Biological Section, British Association 1876) that he has "often observed the beaks and faces of the brush-tongued lories of the Moluccas covered with pollen." In New Zealand, many specimens of the Anthornis melanura had their heads coloured with pollen from the flowers of an endemic species of Fuchsia (Potts 'Transactions of the New Zealand Institute' volume 3 1870 page 72.) Next in importance, but in a quite subordinate degree, is the wind; and with some aquatic plants, according to Delpino, currents of water. The simple fact of the necessity in many cases of extraneous aid for the transport of the pollen, and the many contrivances for this purpose, render it highly probable that some great benefit is thus gained; and this conclusion has now been firmly established by the proved superiority in growth, vigour, and fertility of plants of crossed parentage over those of self-fertilised parentage. But we should always keep in mind that two somewhat opposed ends have to be gained; the first and more important one being the production of seeds by any means, and the second, cross-fertilisation.

The advantages derived from cross-fertilisation throw a flood of light on most of the chief characters of flowers. We can thus understand their large size and bright colours, and in some cases the bright tints of the adjoining parts, such as the peduncles, bracteae, etc. By this means they are rendered conspicuous to insects, on the same principle that almost every fruit which is devoured by birds presents a strong contrast in colour with the green foliage, in order that it may be seen, and its seeds freely disseminated. With some flowers conspicuousness is gained at the expense even of the reproductive organs, as with the ray-florets of many Compositae, the exterior flowers of Hydrangea, and the terminal flowers of the Feather-hyacinth or Muscari. There is also reason to believe, and this was the opinion of Sprengel, that flowers differ in colour in accordance with the kinds of insects which frequent them.

Not only do the bright colours of flowers serve to attract insects, but dark-coloured streaks and marks are often present, which Sprengel long ago maintained served as guides to the nectary. These marks follow the veins in the petals, or lie between them. They may occur on only one, or on all excepting one or more of the upper or lower petals; or they may form a dark ring round the tubular part of the corolla, or be confined to the lips of an irregular flower. In the white varieties of many flowers, such as of Digitalis purpurea, Antirrhinum majus, several species of Dianthus, Phlox, Myosotis, Rhododendron, Pelargonium, Primula and Petunia, the marks generally persist, whilst the rest of the corolla has become of a pure white; but this may be due merely to their colour being more intense and thus less readily obliterated. Sprengel's notion of the use of these marks as guides appeared to me for a long time fanciful; for insects, without such aid, readily discover and bite holes through the nectary from the outside. They also discover the minute nectar-secreting glands on the stipules and leaves of certain plants. Moreover, some few plants, such as certain poppies, which are not nectariferous, have guiding marks; but we might perhaps expect that some few plants would retain traces of a former nectariferous condition. On the other hand, these marks are much more common on asymmetrical flowers, the entrance into which would be apt to puzzle insects, than on regular flowers. Sir J. Lubbock has also proved that bees readily distinguish colours, and that they lose much time if the position of honey which they have once visited be in the least changed. (10/2. 'British Wild Flowers in relation to Insects' 1875 page 44.) The following case affords, I think, the best evidence that these marks have really been developed in correlation with the nectary. The two upper petals of the common Pelargonium are thus marked near their bases; and I have repeatedly observed that when the flowers vary so as to become peloric or regular, they lose their nectaries and at the same time the dark marks. When the nectary is only partially aborted, only one of the upper petals loses its mark. Therefore the nectary and these marks clearly stand in some sort of close relation to one another; and the simplest view is that they were developed together for a special purpose; the only conceivable one being that the marks serve as a guide to the nectary. It is, however, evident from what has been already said, that insects could discover the nectar without the aid of guiding marks. They are of service to the plant, only by aiding insects to visit and suck a greater number of flowers within a given time than would otherwise be possible; and thus there will be a better chance of fertilisation by pollen brought from a distinct plant, and this we know is of paramount importance.

The odours emitted by flowers attract insects, as I have observed in the case of plants covered by a muslin net. Nageli affixed artificial flowers to branches, scenting some with essential oils and leaving others unscented; and insects were attracted to the former in an unmistakable manner. (10/3. 'Enstehung etc. der Naturhist. Art.' 1865 page 23.) Not a few flowers are both conspicuous and odoriferous. Of all colours, white is the prevailing one; and of white flowers a considerably larger proportion smell sweetly than of any other colour, namely, 14.6 per cent; of red, only 8.2 per cent are odoriferous. (10/4. The colours and odours of the flowers of 4200 species have been tabulated by Landgrabe and by Schubler and Kohler. I have not seen their original works, but a very full abstract is given in Loudon's 'Gardeners' Magazine' volume 13 1837 page 367.) The fact of a larger proportion of white flowers smelling sweetly may depend in part on those which are fertilised by moths requiring the double aid of conspicuousness in the dusk and of odour. So great is the economy of nature, that most flowers which are fertilised by crepuscular or nocturnal insects emit their odour chiefly or exclusively in the evening. Some flowers, however, which are highly odoriferous depend solely on this quality for their fertilisation, such as the night-flowering stock (Hesperis) and some species of Daphne; and these present the rare case of flowers which are fertilised by insects being obscurely coloured.