From some observations by Dr. Torrey, Hottonia inflata, an inhabitant of the United States, does not appear to be heterostyled, but is remarkable from producing cleistogamic flowers, as will be seen in the last chapter of this volume.

Besides the genera Primula and Hottonia, Androsace (vel Gregoria, vel Aretia) vitalliana is heterostyled. Mr. Scott fertilised with their own pollen 21 flowers on three short-styled plants in the Edinburgh Botanic Gardens, and not one yielded a single seed; but eight of them which were fertilised with pollen from one of the other plants of the same form, set two empty capsules. (1/19. See also Treviranus in 'Botanische Zeitung' 1863 page 6 on this plant being dimorphic.) He was able to examine only dried specimens of the long-styled forms. But the evidence seems sufficient to leave hardly a doubt that Androsace is heterostyled. Fritz Muller sent me from South Brazil dried flowers of a Statice which he believed to be heterostyled. In the one form the pistil was considerably longer and the stamens slightly shorter than the corresponding organs in the other form. But as in the shorter-styled form the stigmas reached up to the anthers of the same flower, and as I could not detect in the dried specimens of the two forms any difference in their stigmas, or in the size of their pollen-grains, I dare not rank this plant as heterostyled. From statements made by Vaucher I was led to think that Soldanella alpina was heterostyled, but it is impossible that Kerner, who has closely studied this plant, could have overlooked the fact. So again from other statements it appeared probable that Pyrola might be heterostyled, but H. Muller examined for me two species in North Germany, and found this not to be the case.

CHAPTER II.

HYBRID PRIMULAS.

The oxlip a hybrid naturally produced between Primula veris and vulgaris. The differences in structure and function between the two parent-species. Effects of crossing long-styled and short-styled oxlips with one another and with the two forms of both parent-species. Character of the offspring from oxlips artificially self-fertilised and cross- fertilised in a state of nature. Primula elatior shown to be a distinct species. Hybrids between other heterostyled species of Primula. Supplementary note on spontaneously produced hybrids in the genus Verbascum.

The various species of Primula have produced in a state of nature throughout Europe an extraordinary number of hybrid forms. For instance, Professor Kerner has found no less than twenty-five such forms in the Alps. (2/1. "Die Primulaceen-Bastarten" 'Oesterr. Botanische Zeitschrift' Jahr 1875 Numbers 3, 4 and 5. See also Godron on hybrid Primulas in 'Bull. Soc. Bot. de France' tome 10 1853 page 178. Also in 'Revue des Sciences Nat.' 1875 page 331.) The frequent occurrence of hybrids in this genus no doubt has been favoured by most of the species being heterostyled, and consequently requiring cross-fertilisation by insects; yet in some other genera, species which are not heterostyled and which in some respects appear not well adapted for hybrid-fertilisation, have likewise been largely hybridised. In certain districts of England, the common oxlip—a hybrid between the cowslip (P. veris, vel officinalis) and the primrose (P. vulgaris, vel acaulis)—is frequently found, and it occurs occasionally almost everywhere. Owing to the frequency of this intermediate hybrid form, and to the existence of the Bardfield oxlip (P. elatior), which resembles to a certain extent the common oxlip, the claim of the three forms to rank as distinct species has been discussed oftener and at greater length than that of almost any other plant. Linnaeus considered P. veris, vulgaris and elatior to be varieties of the same species, as do some distinguished botanists at the present day; whilst others who have carefully studied these plants do not doubt that they are distinct species. The following observations prove, I think, that the latter view is correct; and they further show that the common oxlip is a hybrid between P. veris and vulgaris.

The cowslip differs so conspicuously in general appearance from the primrose, that nothing need here be said with respect to their external characters. (2/2. The Reverend W.A. Leighton has pointed out certain differences in the form of the capsules and seed in 'Annals and Magazine of Natural History' 2nd series volume 2 1848 page 164.) But some less obvious differences deserve notice. As both species are heterostyled, their complete fertilisation depends on insects. The cowslip is habitually visited during the day by the larger humble-bees (namely Bombus muscorum and hortorum), and at night by moths, as I have seen in the case of Cucullia. The primrose is never visited (and I speak after many years' observation) by the larger humble-bees, and only rarely by the smaller kinds; hence its fertilisation must depend almost exclusively on moths. There is nothing in the structure of the flowers of the two plants which can determine the visits of such widely different insects. But they emit a different odour, and perhaps their nectar may have a different taste. Both the long-styled and short-styled forms of the primrose, when legitimately and naturally fertilised, yield on an average many more seeds per capsule than the cowslip, namely, in the proportion of 100 to 55. When illegitimately fertilised they are likewise more fertile than the two forms of the cowslip, as shown by the larger proportion of their flowers which set capsules, and by the larger average number of seeds which the capsules contain. The difference also between the number of seeds produced by the long-styled and short-styled flowers of the primrose, when both are illegitimately fertilised, is greater than that between the number produced under similar circumstances by the two forms of the cowslip. The long-styled flowers of the primrose when protected from the access of all insects, except such minute ones as Thrips, yield a considerable number of capsules containing on an average 19.2 seeds per capsule; whereas 18 plants of the long-styled cowslip similarly treated did not yield a single seed.

The primrose, as every one knows, flowers a little earlier in the spring than the cowslip, and inhabits slightly different stations and districts. The primrose generally grows on banks or in woods, whilst the cowslip is found in more open places. The geographical range of the two forms is different. Dr. Bromfield remarks that "the primrose is absent from all the interior region of northern Europe, where the cowslip is indigenous." (2/3. 'Phytologist' volume 3 page 694.) In Norway, however, both plants range to the same degree of north latitude. (2/4. H. Lecoq 'Geograph. Bot. de l'Europe' tome 8 1858 pages 141, 144. See also 'Annals and Magazine of Natural History' 9 1842 pages 156, 515. Also Boreau 'Flore du centre de la France' 1840 tome 2 page 376. With respect to the rarity of P. veris in western Scotland, see H.C. Watson 'Cybele Britannica' 2 page 293.)

The cowslip and primrose, when intercrossed, behave like distinct species, for they are far from being mutually fertile. Gartner crossed 27 flowers of P. vulgaris with pollen of P. veris, and obtained 16 capsules; but these did not contain any good seed. (2/5. 'Bastarderzeugung' 1849 page 721.) He also crossed 21 flowers of P. veris with pollen of P. vulgaris; and now he got only five capsules, containing seed in a still less perfect condition. Gartner knew nothing about heterostylism; and his complete failure may perhaps be accounted for by his having crossed together the same forms of the cowslip and primrose; for such crosses would have been of an illegitimate as well as of a hybrid nature, and this would have increased their sterility. My trials were rather more fortunate. Twenty-one flowers, consisting of both forms of the cowslip and primrose, were intercrossed legitimately, and yielded seven capsules (i.e. 33 per cent), containing on an average 42 seeds; some of these seeds, however, were so poor that they probably would not have germinated. Twenty-one flowers on the same cowslip and primrose plants were also intercrossed illegitimately, and they likewise yielded seven capsules (or 33 per cent), but these contained on an average only 13 good and bad seeds. I should, however, state that some of the above flowers of the primrose were fertilised with pollen from the polyanthus, which is certainly a variety of the cowslip, as may be inferred from the perfect fertility inter se of the crossed offspring from these two plants. (2/6. Mr. Scott has discussed the nature of the polyanthus ('Proceedings of the Linnean Society' 8 Botany 1864 page 103), and arrives at a different conclusion; but I do not think that his experiments were sufficiently numerous. The degree of infertility of a cross is liable to much fluctuation. Pollen from the cowslip at first appears rather more efficient on the primrose than that of the polyanthus; for 12 flowers of both forms of the primrose, fertilised legitimately and illegitimately with pollen of the cowslip gave five capsules, containing on an average 32.4 seeds; whilst 18 flowers similarly fertilised by polyanthus-pollen yielded only five capsules, containing only 22.6 seeds. On the other hand, the seeds produced by the polyanthus-pollen were much the finest of the whole lot, and were the only ones which germinated.) To show how sterile these hybrid unions were I may remind the reader that 90 per cent of the flowers of the primrose fertilised legitimately with primrose-pollen yielded capsules, containing on an average 66 seeds; and that 54 per cent of the flowers fertilised illegitimately yielded capsules containing on an average 35.5 seeds per capsule. The primrose, especially the short-styled form, when fertilised by the cowslip, is less sterile, as Gartner likewise observed, than is the cowslip when fertilised by the primrose. The above experiments also show that a cross between the same forms of the primrose and cowslip is much more sterile than that between different forms of these two species.

The seeds from the several foregoing crosses were sown, but none germinated except those from the short-styled primrose fertilised with pollen of the polyanthus; and these seeds were the finest of the whole lot. I thus raised six plants, and compared them with a group of wild oxlips which I had transplanted into my garden. One of these wild oxlips produced slightly larger flowers than the others, and this one was identical in every character (in foliage, flower- peduncle, and flowers) with my six plants, excepting that the flowers of the latter were tinged of a dingy red colour, from being descended from the polyanthus.

We thus see that the cowslip and primrose cannot be crossed either way except with considerable difficulty, that they differ conspicuously in external appearance, that they differ in various physiological characters, that they inhabit slightly different stations and range differently. Hence those botanists who rank these plants as varieties ought to be able to prove that they are not as well fixed in character as are most species; and the evidence in favour of such instability of character appears at first sight very strong. It rests, first, on statements made by several competent observers that they have raised cowslips, primroses, and oxlips from seeds of the same plant; and, secondly, on the frequent occurrence in a state of nature of plants presenting every intermediate gradation between the cowslip and primrose.

The first statement, however, is of little value; for, heterostylism not being formerly understood, the seed-bearing plants were in no instance protected from the visits of insects (2/7. One author states in the 'Phytologist' volume 3 page 703 that he covered with bell-glasses some cowslips, primroses, etc., on which he experimented. He specifies all the details of his experiment, but does not say that he artificially fertilised his plants; yet he obtained an abundance of seed, which is simply impossible. Hence there must have been some strange error in these experiments, which may be passed over as valueless.); and there would be almost as much risk of an isolated cowslip, or of several cowslips if consisting of the same form, being crossed by a neighbouring primrose and producing oxlips, as of one sex of a dioecious plant, under similar circumstances, being crossed by the opposite sex of an allied and neighbouring species. Mr. H.C. Watson, a critical and most careful observer, made many experiments by sowing the seeds of cowslips and of various kinds of oxlips, and arrived at the following conclusion, namely, "that seeds of a cowslip can produce cowslips and oxlips, and that seeds of an oxlip can produce cowslips, oxlips, and primroses." (2/8. 'Phytologist' 2 pages 217, 852; 3 page 43.) This conclusion harmonises perfectly with the view that in all cases, when such results have been obtained, the unprotected cowslips have been crossed by primroses, and the unprotected oxlips by either cowslips or primroses; for in this latter case we might expect, by the aid of reversion, which notoriously comes into powerful action with hybrids, that the two parent-forms in appearance pure, as well as many intermediate gradations, would be occasionally produced. Nevertheless the two following statements offer considerable difficulty. The Reverend Professor Henslow raised from seeds of a cowslip growing in his garden, various kinds of oxlips and one perfect primrose; but a statement in the same paper perhaps throws light on this anomalous result. (2/9. Loudon's 'Magazine of Natural History' 3 1830 page 409.) Professor Henslow had previously transplanted into his garden a cowslip, which completely changed its appearance during the following year, and now resembled an oxlip. Next year again it changed its character, and produced, in addition to the ordinary umbels, a few single- flowered scapes, bearing flowers somewhat smaller and more deeply coloured than those of the common primrose. From what I have myself observed with oxlips, I cannot doubt that this plant was an oxlip in a highly variable condition, almost like that of the famous Cytisus adami. This presumed oxlip was propagated by offsets, which were planted in different parts of the garden; and if Professor Henslow took by mistake seeds from one of these plants, especially if it had been crossed by a primrose, the result would be quite intelligible. Another case is still more difficult to understand: Dr. Herbert raised, from the seeds of a highly cultivated red cowslip, cowslips, oxlips of various kinds, and a primrose. (2/10. 'Transactions of the Horticultural Society' 4 page 19.) This case, if accurately recorded, which I much doubt, is explicable only on the improbable assumption that the red cowslip was not of pure parentage. With species and varieties of many kinds, when intercrossed, one is sometimes strongly prepotent over the other; and instances are known of a variety crossed by another, producing offspring which in certain characters, as in colour, hairiness, etc., have proved identical with the pollen-bearing parent, and quite dissimilar to the mother-plant (2/11. I have given instances in my work 'On the Variation of Animals and Plants under Domestication' chapter 15 2nd edition volume 2 page 69.); but I do not know of any instance of the offspring of a cross perfectly resembling, in a considerable number of important characters, the father alone. It is, therefore, very improbable that a pure cowslip crossed by a primrose should ever produce a primrose in appearance pure. Although the facts given by Dr. Herbert and Professor Henslow are difficult to explain, yet until it can be shown that a cowslip or a primrose, carefully protected from insects, will give birth to at least oxlips, the cases hitherto recorded have little weight in leading us to admit that the cowslip and primrose are varieties of one and the same species.

Negative evidence is of little value; but the following facts may be worth giving:—Some cowslips which had been transplanted from the fields into a shrubbery were again transplanted into highly manured land. In the following year they were protected from insects, artificially fertilised, and the seed thus procured was sown in a hotbed. The young plants were afterwards planted out, some in very rich soil, some in stiff poor clay, some in old peat, and some in pots in the greenhouse; so that these plants, 765 in number, as well as their parents, were subjected to diversified and unnatural treatment; but not one of them presented the least variation except in size—those in the peat attaining almost gigantic dimensions, and those in the clay being much dwarfed.

I do not, of course, doubt that cowslips exposed during SEVERAL successive generations to changed conditions would vary, and that this might occasionally occur in a state of nature. Moreover, from the law of analogical variation, the varieties of any one species of Primula would probably in some cases resemble other species of the genus. For instance I raised a red primrose from seed from a protected plant, and the flowers, though still resembling those of the primrose, were borne during one season in umbels on a long foot-stalk like that of a cowslip.

With regard to the second class of facts in support of the cowslip and primrose being ranked as mere varieties, namely, the well-ascertained existence in a state of nature of numerous linking forms (2/12. See an excellent article on this subject by Mr. H.C. Watson in the 'Phytologist' volume 3 page 43.):—If it can be shown that the common wild oxlip, which is intermediate in character between the cowslip and primrose, resembles in sterility and other essential respects a hybrid plant, and if it can further be shown that the oxlip, though in a high degree sterile, can be fertilised by either parent-species, thus giving rise to still finer gradational links, then the presence of such linking forms in a state of nature ceases to be an argument of any weight in favour of the cowslip and primrose being varieties, and becomes, in fact, an argument on the other side. The hybrid origin of a plant in a state of nature can be recognised by four tests: first, by its occurrence only where both presumed parent-species exist or have recently existed; and this holds good, as far as I can discover, with the oxlip; but the P. elatior of Jacq., which, as we shall presently see, constitutes a distinct species, must not be confounded with the common oxlip. Secondly, by the supposed hybrid plant being nearly intermediate in character between the two parent-species, and especially by its resembling hybrids artificially made between the same two species. Now the oxlip is intermediate in character, and resembles in every respect, except in the colour of the corolla, hybrids artificially produced between the primrose and the polyanthus, which latter is a variety of the cowslip. Thirdly, by the supposed hybrids being more or less sterile when crossed inter se: but to try this fairly two distinct plants of the same parentage, and not two flowers on the same plant, should be crossed; for many pure species are more or less sterile with pollen from the same individual plant; and in the case of hybrids from heterostyled species the opposite forms should be crossed. Fourthly and lastly, by the supposed hybrids being much more fertile when crossed with either pure parent-species than when crossed inter se, but still not as fully fertile as the parent-species.

For the sake of ascertaining the two latter points, I transplanted a group of wild oxlips into my garden. They consisted of one long-styled and three short- styled plants, which, except in the corolla of one being slightly larger, resembled each other closely. The trials which were made, and the results obtained, are shown in tables 2.14, 2.15, 2.16, 2.17 and 2.18. No less than twenty different crosses are necessary in order to ascertain fully the fertility of hybrid heterostyled plants, both inter se and with their two parent-species. In this instance 256 flowers were crossed in the course of four seasons. I may mention, as a mere curiosity, that if any one were to raise hybrids between two trimorphic heterostyled species, he would have to make 90 distinct unions in order to ascertain their fertility in all ways; and as he would have to try at least 10 flowers in each case, he would be compelled to fertilise 900 flowers and count their seeds. This would probably exhaust the patience of the most patient man.

TABLE 2.14. Crosses inter se between the two forms of the common Oxlip.

Column 1: Illegitimate union. Short-styled oxlip, by pollen of short-styled oxlip: 20 flowers fertilised, did not produce one capsule.

Column 2: Legitimate union. Short-styled oxlip, by pollen of long-styled oxlip: 10 flowers fertilised, did not produce one capsule.

Column 3: Illegitimate union. Long-styled oxlip, by its own pollen: 24 flowers fertilised, produced five capsules, containing 6, 10, 20, 8, and 14 seeds. Average 11.6.

Column 4: Legitimate union. Long-styled oxlip, by pollen of short-styled oxlip: 10 flowers fertilised, did not produce one capsule.

TABLE 2.15. Both forms of the Oxlip crossed with Pollen of both forms of the Cowslip, P. veris.

Column 1: Illegitimate union. Short-styled oxlip, by pollen of short-styled cowslip: 18 flowers fertilised, did not produce one capsule.

Column 2: Legitimate union. Short-styled oxlip, by pollen of long-styled cowslip: 18 flowers fertilised, produced three capsules, containing 7, 3, and 3 wretched seeds, apparently incapable of germination.

Column 3: Illegitimate union. Long-styled oxlip, by pollen of long-styled cowslip: 11 flowers fertilised, produced one capsule, containing 13 wretched seeds.

Column 4: Legitimate union. Long-styled oxlip, by pollen of short-styled cowslip: 5 flowers fertilised, produced two capsules, containing 21 and 28 very fine seeds.

TABLE 2.16. Both forms of the Oxlip crossed with Pollen of both forms of the Primrose, P. vulgaris.

Column 1: Illegitimate union. Short-styled oxlip, by pollen of short-styled primrose: 34 flowers fertilised, produced two capsules, containing 5 and 12 seeds.

Column 2: Legitimate union. Short-styled oxlip, by pollen of long-styled primrose: 26 flowers fertilised, produced six capsules, containing 16, 20, 5, 10, 19, and 24 seeds. Average 15.7. Many of the seeds very poor, some good.

Column 3: Illegitimate union. Long-styled oxlip, by pollen of long-styled primrose: 11 flowers fertilised, produced four capsules, containing 10, 7, 5, and 6 wretched seeds. Average 7.0.

Column 4: Legitimate union. Long-styled oxlip, by pollen of short-styled primrose: 5 flowers fertilised, produced five capsules, containing 26, 32, 23, 28, and 34 seeds. Average 28.6.

TABLE 2.17. Both forms of the Cowslip crossed with Pollen of both forms of the Oxlip.

Column 1: Illegitimate union. Short-styled cowslip, by pollen of short-styled oxlip: 8 flowers fertilised, did not produce one capsule.

Column 2: Legitimate union. Long-styled cowslip, by pollen of short-styled oxlip: 8 flowers fertilised, produced one capsule, containing 26 seeds.

Column 3: Illegitimate union. Long-styled cowslip, by pollen of long-styled oxlip: 8 flowers fertilised, produced three capsules, containing 5, 6 and 14 seeds. Average 8.3.

Column 4: Legitimate union. Short-styled cowslip, by pollen of long-styled oxlip: 8 flowers fertilised, produced 8 capsules, containing 58, 38, 31, 44, 23, 26, 37, and 66 seeds. Average 40.4.

TABLE 2.18. Both forms of the Primrose crossed with Pollen of both forms of the Oxlip.

Column 1: Illegitimate union. Short-styled primrose, by pollen of short-styled oxlip: 8 flowers fertilised, did not produce one capsule.

Column 2: Legitimate union. Long-styled primrose, by pollen of short-styled oxlip: 8 flowers fertilised, produced two capsules, containing 5 and 2 seeds.

Column 3: Illegitimate union. Long-styled primrose, by pollen of long-styled oxlip: 8 flowers fertilised, produced 8 capsules, containing 15, 7, 12, 20, 22, 7, 16, and 13 seeds. Average 14.0.

Column 4: Legitimate union. Short-styled primrose, by pollen of long-styled oxlip: 8 flowers fertilised, produced 4 capsules, containing 52, 52, 42, and 49 seeds, some good and some bad. Average 48.7.

We see in Tables 2/14 to 2/18 the number of capsules and of seeds produced, by crossing both forms of the oxlip in a legitimate and illegitimate manner with one another, and with the two forms of the primrose and cowslip. I may premise that the pollen of two of the short-styled oxlips consisted of nothing but minute aborted whitish cells; but in the third short-styled plant about one- fifth of the grains appeared in a sound condition. Hence it is not surprising that neither the short-styled nor the long-styled oxlip produced a single seed when fertilised with this pollen. Nor did the pure cowslips or primroses when illegitimately fertilised with it; but when thus legitimately fertilised they yielded a few good seeds. The female organs of the short-styled oxlips, though greatly deteriorated in power, were in a rather better condition than the male organs; for though the short-styled oxlips yielded no seed when fertilised by the long-styled oxlips, and hardly any when illegitimately fertilised by pure cowslips or primroses, yet when legitimately fertilised by these latter species, especially by the long-styled primrose, they yielded a moderate supply of good seed.

The long-styled oxlip was more fertile than the three short-styled oxlips, and about half its pollen-grains appeared sound. It bore no seed when legitimately fertilised by the short-styled oxlips; but this no doubt was due to the badness of the pollen of the latter; for when illegitimately fertilised (Table 2.14) by its own pollen it produced some good seeds, though much fewer than self- fertilised cowslips or primroses would have produced. The long-styled oxlip likewise yielded a very low average of seed, as may be seen in the third compartment of Tables 2.15 to 2.18, when illegitimately fertilised by, and when illegitimately fertilising, pure cowslips and primroses. The four corresponding legitimate unions, however, were moderately fertile, and one (namely that between a short-styled cowslip and the long-styled oxlip in Table 2.17) was nearly as fertile as if both parents had been pure. A short-styled primrose legitimately fertilised by the long-styled oxlip (Table 2.18) also yielded a moderately good average, namely 48.7 seeds; but if this short-styled primrose had been fertilised by a long-styled primrose it would have yielded an average of 65 seeds. If we take the ten legitimate unions together, and the ten illegitimate unions together, we shall find that 29 per cent of the flowers fertilised in a legitimate manner yielded capsules, these containing on an average 27.4 good and bad seeds; whilst only 15 per cent of the flowers fertilised in an illegitimate manner yielded capsules, these containing on an average only 11.0 good and bad seeds.

In a previous part of this chapter it was shown that illegitimate crosses between the long-styled form of the primrose and the long-styled cowslip, and between the short-styled primrose and short-styled cowslip, are more sterile than legitimate crosses between these two species; and we now see that the same rule holds good almost invariably with their hybrid offspring, whether these are crossed inter se, or with either parent-species; so that in this particular case, but not as we shall presently see in other cases, the same rule prevails with the pure unions between the two forms of the same heterostyled species, with crosses between two distinct heterostyled species, and with their hybrid offspring.

Seeds from the long-styled oxlip fertilised by its own pollen were sown, and three long-styled plants raised. The first of these was identical in every character with its parent. The second bore rather smaller flowers, of a paler colour, almost like those of the primrose; the scapes were at first single- flowered, but later in the season a tall thick scape, bearing many flowers, like that of the parent oxlip, was thrown up. The third plant likewise produced at first only single-flowered scapes, with the flowers rather small and of a darker yellow; but it perished early. The second plant also died in September; and the first plant, though all three grew under very favourable conditions, looked very sickly. Hence we may infer that seedlings from self-fertilised oxlips would hardly be able to exist in a state of nature. I was surprised to find that all the pollen-grains in the first of these seedling oxlips appeared sound; and in the second only a moderate number were bad. These two plants, however, had not the power of producing a proper number of seeds; for though left uncovered and surrounded by pure primroses and cowslips, the capsules were estimated to include an average of only from fifteen to twenty seeds.

From having many experiments in hand, I did not sow the seed obtained by crossing both forms of the primrose and cowslip with both forms of the oxlip, which I now regret; but I ascertained an interesting point, namely, the character of the offspring from oxlips growing in a state of nature near both primroses and cowslips. The oxlips were the same plants which, after their seeds had been collected, were transplanted and experimented on. From the seeds thus obtained eight plants were raised, which, when they flowered, might have been mistaken for pure primroses; but on close comparison the eye in the centre of the corolla was seen to be of a darker yellow, and the peduncles more elongated. As the season advanced, one of these plants threw up two naked scapes, 7 inches in height, which bore umbels of flowers of the same character as before. This fact led me to examine the other plants after they had flowered and were dug up; and I found that the flower-peduncles of all sprung from an extremely short common scape, of which no trace can be found in the pure primrose. Hence these plants are beautifully intermediate between the oxlip and the primrose, inclining rather towards the latter; and we may safely conclude that the parent oxlips had been fertilised by the surrounding primroses.

From the various facts now given, there can be no doubt that the common oxlip is a hybrid between the cowslip (P. veris, Brit. Fl.) and the primrose (P. vulgaris, Brit. Fl.), as has been surmised by several botanists. It is probable that oxlips may be produced either from the cowslip or the primrose as the seed- bearer, but oftenest from the latter, as I judge from the nature of the stations in which oxlips are generally found (2/13. See also on this head Hardwicke's 'Science Gossip' 1867 pages 114, 137.), and from the primrose when crossed by the cowslip being more fertile than, conversely, the cowslip by the primrose. The hybrids themselves are also rather more fertile when crossed with the primrose than with the cowslip. Whichever may be the seed-bearing plant, the cross is probably between different forms of the two species; for we have seen that legitimate hybrid unions are more fertile than illegitimate hybrid unions. Moreover a friend in Surrey found that 29 oxlips which grew in the neighbourhood of his house consisted of 13 long-styled and 16 short-styled plants; now, if the parent-plants had been illegitimately united, either the long- or short-styled form would have greatly preponderated, as we shall hereafter see good reason to believe. The case of the oxlip is interesting; for hardly any other instance is known of a hybrid spontaneously arising in such large numbers over so wide an extent of country. The common oxlip (not the P. elatior of Jacq.) is found almost everywhere throughout England, where both cowslips and primroses grow. In some districts, as I have seen near Hartfield in Sussex and in parts of Surrey, specimens may be found on the borders of almost every field and small wood. In other districts the oxlip is comparatively rare: near my own residence I have found, during the last twenty-five years, not more than five or six plants or groups of plants. It is difficult to conjecture what is the cause of this difference in their number. It is almost necessary that a plant, or several plants belonging to the same form, of one parent-species, should grow near the opposite form of the other parent-species; and it is further necessary that both species should be frequented by the same kind of insect, no doubt a moth. The cause of the rare appearance of the oxlip in certain districts may be the rarity of some moth, which in other districts habitually visits both the primrose and cowslip.

Finally, as the cowslip and primrose differ in the various characters above specified,—as they are in a high degree sterile when intercrossed,—as there is no trustworthy evidence that either species, when uncrossed, has ever given birth to the other species or to any intermediate form,—and as the intermediate forms which are often found in a state of nature have been shown to be more or less sterile hybrids of the first or second generation,—we must for the future look at the cowslip and primrose as good and true species.

Primula elatior, Jacq., or the Bardfield Oxlip, is found in England only in two or three of the eastern counties. On the Continent it has a somewhat different range from that of the cowslip and primrose; and it inhabits some districts where neither of these species live. (2/14. For England, see Hewett C. Watson 'Cybele Britannica' volume 2 1849 page 292. For the Continent, see Lecoq 'Geograph. Botanique de l'Europe' tome 8 1858 page 142. For the Alps see 'Annals and Magazine of Natural History' volume 9 1842 pages 156 and 515.) In general appearance it differs so much from the common oxlip, that no one accustomed to see both forms in the living state could afterwards confound them; but there is scarcely more than a single character by which they can be distinctly defined, namely, their linear-oblong capsules equalling the calyx in length. (2/15. Babington 'Manual of British Botany' 1851 page 258.) The capsules when mature differ conspicuously, owing to their length, from those of the cowslip and primrose. With respect to the fertility of the two forms when these are united in the four possible methods, they behave like the other heterostyled species of the genus, but differ somewhat (see Tables 1.8 and 1.12.) in the smaller proportion of the illegitimately fertilised flowers which set capsules. That P. elatior is not a hybrid is certain, for when the two forms were legitimately united they yielded the large average of 47.1 seeds, and when illegitimately united 35.5 per capsule; whereas, of the four possible unions (Table 2.14) between the two forms of the common oxlip which we know to be a hybrid, one alone yielded any seed; and in this case the average number was only 11.6 per capsule. Moreover I could not detect a single bad pollen-grain in the anthers of the short-styled P. elatior; whilst in two short-styled plants of the common oxlip all the grains were bad, as were a large majority in a third plant. As the common oxlip is a hybrid between the primrose and cowslip, it is not surprising that eight long-styled flowers of the primrose, fertilised by pollen from the long-styled common oxlip, produced eight capsules (Table 1.18), containing, however, only a low average of seeds; whilst the same number of flowers of the primrose, similarly fertilised by the long-styled Bardfield oxlip, produced only a single capsule; this latter plant being an altogether distinct species from the primrose. Plants of P. elatior have been propagated by seed in a garden for twenty-five years, and have kept all this time quite constant, excepting that in some cases the flowers varied a little in size and tint. (2/16. See Mr. H. Doubleday in the 'Gardener's Chronicle' 1867 page 435, also Mr. W. Marshall ibid. page 462.) Nevertheless, according to Mr. H.C. Watson and Dr. Bromfield (2/17. 'Phytologist' volume 1 page 1001 and volume 3 page 695.), plants may be occasionally found in a state of nature, in which most of the characters by which this species can be distinguished from P. veris and vulgaris fail; but such intermediate forms are probably due to hybridisation; for Kerner states, in the paper before referred to, that hybrids sometimes, though rarely, arise in the Alps between P. elatior and veris.

Finally, although we may freely admit that Primula veris, vulgaris, and elatior, as well as all the other species of the genus, are descended from a common primordial form, yet from the facts above given, we must conclude that these three forms are now as fixed in character as are many others which are universally ranked as true species. Consequently they have as good a right to receive distinct specific names as have, for instance, the ass, quagga, and zebra.

Mr. Scott has arrived at some interesting results by crossing other heterostyled species of Primula. (2/18. 'Journal of the Linnean Society Botany' volume 8 1864 page 93 to end.) I have already alluded to his statement, that in four instances (not to mention others) a species when crossed with a distinct one yielded a larger number of seeds than the same species fertilised illegitimately with its own-form pollen, though taken from a distinct plant. It has long been known from the researches of Kolreuter and Gartner, that two species when crossed reciprocally sometimes differ as widely as is possible in their fertility: thus A when crossed with the pollen of B will yield a large number of seeds, whilst B may be crossed repeatedly with pollen of A, and will never yield a single seed. Now Mr. Scott shows in several cases that the same law holds good when two heterostyled species of Primula are intercrossed, or when one is crossed with a homostyled species. But the results are much more complicated than with ordinary plants, as two heterostyled dimorphic species can be intercrossed in eight different ways. I will give one instance from Mr. Scott. The long-styled P. hirsuta fertilised legitimately and illegitimately with pollen from the two forms of P. auricula, and reciprocally the long-styled P. auricula fertilised legitimately and illegitimately with pollen from the two forms of P. hirsuta, did not produce a single seed. Nor did the short-styled P. hirsuta when fertilised legitimately and illegitimately with the pollen of the two forms of P. auricula. On the other hand, the short-styled P. auricula fertilised with pollen from the long-styled P. hirsuta yielded capsules containing on an average no less than 56 seeds; and the short-styled P. auricula by pollen of the short- styled P. hirsuta yielded capsules containing on an average 42 seeds per capsule. So that out of the eight possible unions between the two forms of these two species, six were utterly barren, and two fairly fertile. We have seen also the same sort of extraordinary irregularity in the results of my twenty different crosses (Tables 2.14 to 2.18), between the two forms of the oxlip, primrose, and cowslip. Mr. Scott remarks, with respect to the results of his trials, that they are very surprising, as they show us that "the sexual forms of a species manifest in their respective powers for conjunction with those of another species, physiological peculiarities which might well entitle them, by the criterion of fertility, to specific distinction."

Finally, although P. veris and vulgaris, when crossed legitimately, and especially when their hybrid offspring are crossed in this manner with both parent-species, were decidedly more fertile, than when crossed in an illegitimate manner, and although the legitimate cross effected by Mr. Scott between P. auricula and hirsuta was more fertile, in the ratio of 56 to 42, than the illegitimate cross, nevertheless it is very doubtful, from the extreme irregularity of the results in the various other hybrid crosses made by Mr. Scott, whether it can be predicted that two heterostyled species are generally more fertile if crossed legitimately (i.e. when opposite forms are united) than when crossed illegitimately.

SUPPLEMENTARY NOTE ON SOME WILD HYBRID VERBASCUMS.

In an early part of this chapter I remarked that few other instances could be given of a hybrid spontaneously arising in such large numbers, and over so wide an extent of country, as that of the common oxlip; but perhaps the number of well-ascertained cases of naturally produced hybrid willows is equally great. (2/19. Max Wichura 'Die Bastardbefruchtung etc. der Weiden' 1865.) Numerous spontaneous hybrids between several species of Cistus, found near Narbonne, have been carefully described by M. Timbal-Lagrave (2/20. 'Mem. de l'Acad. des Sciences de Toulouse' 5e serie tome 5 page 28.), and many hybrids between an Aceras and Orchis have been observed by Dr. Weddell. (2/21. 'Annales des Sc. Nat.' 3e serie Bot. tome 18 page 6.) In the genus Verbascum, hybrids are supposed to have often originated in a state of nature (2/22. See for instance the 'English Flora' by Sir J.E. Smith 1824 volume 1 page 307.); some of these undoubtedly are hybrids, and several hybrids have originated in gardens; but most of these cases require, as Gartner remarks, verification. (2/23. See Gartner 'Bastarderzeugung' 1849 page 590.) Hence the following case is worth recording, more especially as the two species in question, V. thapsus and lychnitis, are perfectly fertile when insects are excluded, showing that the stigma of each flower receives its own pollen. Moreover the flowers offer only pollen to insects, and have not been rendered attractive to them by secreting nectar.

I transplanted a young wild plant into my garden for experimental purposes, and when it flowered it plainly differed from the two species just mentioned and from a third which grows in this neighbourhood. I thought that it was a strange variety of V. thapsus. It attained the height (by measurement) of 8 feet! It was covered with a net, and ten flowers were fertilised with pollen from the same plant; later in the season, when uncovered, the flowers were freely visited by pollen-collecting bees; nevertheless, although many capsules were produced, not one contained a single seed. During the following year this same plant was left uncovered near plants of V. thapsus and lychnitis; but again it did not produce a single seed. Four flowers, however, which were repeatedly fertilised with pollen of V. lychnitis, whilst the plant was temporarily kept under a net, produced four capsules, which contained five, one, two, and two seeds; at the same time three flowers were fertilised with pollen of V. thapsus, and these produced two, two, and three seeds. To show how unproductive these seven capsules were, I may state that a fine capsule from a plant of V. thapsus growing close by contained above 700 seeds. These facts led me to search the moderately-sized field whence my plant had been removed, and I found in it many plants of V. thapsus and lychnitis as well as thirty-three plants intermediate in character between these two species. These thirty-three plants differed much from one another. In the branching of the stem they more closely resembled V. lychnitis than V. thapsus, but in height the latter species. In the shape of their leaves they often closely approached V. lychnitis, but some had leaves extremely woolly on the upper surface and decurrent like those of V. thapsus; yet the degree of woolliness and of decurrency did not always go together. In the petals being flat and remaining open, and in the manner in which the anthers of the longer stamens were attached to the filaments, these plants all took more after V. lychnitis than V. thapsus. In the yellow colour of the corolla they all resembled the latter species. On the whole, these plants appeared to take rather more after V. lychnitis than V. thapsus. On the supposition that they were hybrids, it is not an anomalous circumstance that they should all have produced yellow flowers; for Gartner crossed white and yellow-flowered varieties of Verbascum, and the offspring thus produced never bore flowers of an intermediate tint, but either pure white or pure yellow flowers, generally of the latter colour. (2/24. 'Bastardzeugung' page 307.)

My observations were made in the autumn; so that I was able to collect some half-matured capsules from twenty of the thirty-three intermediate plants, and likewise capsules of the pure V. lychnitis and thapsus growing in the same field. All the latter were filled with perfect but immature seeds, whilst the capsules of the twenty intermediate plants did not contain one single perfect seed. These plants, consequently, were absolutely barren. From this fact,—from the one plant which was transplanted into my garden yielding when artificially fertilised with pollen from V. lychnitis and thapsus some seeds, though extremely few in number,—from the circumstance of the two pure species growing in the same field,—and from the intermediate character of the sterile plants, there can be no doubt that they were hybrids. Judging from the position in which they were chiefly found, I am inclined to believe they were descended from V. thapsus as the seed-bearer, and V. lychnitis as the pollen-bearer.

It is known that many species of Verbascum, when the stem is jarred or struck by a stick, cast off their flowers. (2/25. This was first observed by Correa de Serra: see Sir J.E. Smith's 'English Flora' 1824 volume 1 page 311; also 'Life of Sir J.E. Smith' volume 2 page 210. I was guided to these references by the Reverend W.A. Leighton, who observed this same phenomenon with V. virgatum.) This occurs with V. thapsus, as I have repeatedly observed. The corolla first separates from its attachment, and then the sepals spontaneously bend inwards so as to clasp the ovarium, pushing off the corolla by their movement, in the course of two or three minutes. Nothing of this kind takes place with young barely expanded flowers. With Verbascum lychnitis and, as I believe, V. phoeniceum the corolla is not cast off, however often and severely the stem may be struck. In this curious property the above-described hybrids took after V. thapsus; for I observed, to my surprise, that when I pulled off the flower-buds round the flowers which I wished to mark with a thread, the slight jar invariably caused the corollas to fall off.

These hybrids are interesting under several points of view. First, from the number found in various parts of the same moderately-sized field. That they owed their origin to insects flying from flower to flower, whilst collecting pollen, there can be no doubt. Although insects thus rob the flowers of a most precious substance, yet they do great good; for, as I have elsewhere shown, the seedlings of V. thapsus raised from flowers fertilised with pollen from another plant, are more vigorous than those raised from self-fertilised flowers. (2/26. 'The Effects of Cross and Self-fertilisation' 1876 page 89.) But in this particular instance the insects did great harm, as they led to the production of utterly barren plants. Secondly, these hybrids are remarkable from differing much from one another in many of their characters; for hybrids of the first generation, if raised from uncultivated plants, are generally uniform in character. That these hybrids belonged to the first generation we may safely conclude, from the absolute sterility of all those observed by me in a state of nature and of the one plant in my garden, excepting when artificially and repeatedly fertilised with pure pollen, and then the number of seeds produced was extremely small. As these hybrids varied so much, an almost perfectly graduated series of forms, connecting together the two widely distinct parent-species, could easily have been selected. This case, like that of the common oxlip, shows that botanists ought to be cautious in inferring the specific identity of two forms from the presence of intermediate gradations; nor would it be easy in the many cases in which hybrids are moderately fertile to detect a slight degree of sterility in such plants growing in a state of nature and liable to be fertilised by either parent-species. Thirdly and lastly, these hybrids offer an excellent illustration of a statement made by that admirable observer Gartner, namely, that although plants which can be crossed with ease generally produce fairly fertile offspring, yet well-pronounced exceptions to this rule occur; and here we have two species of Verbascum which evidently cross with the greatest ease, but produce hybrids which are excessively sterile.

CHAPTER III. HETEROSTYLED DIMORPHIC PLANTS—continued.

Linum grandiflorum, long-styled form utterly sterile with own-form pollen. Linum perenne, torsion of the pistils in the long-styled form alone. Homostyled species of Linum. Pulmonaria officinalis, singular difference in self-fertility between the English and German long-styled plants. Pulmonaria angustifolia shown to be a distinct species, long-styled form completely self-sterile. Polygonum fagopyrum. Various other heterostyled genera. Rubiaceae. Mitchella repens, fertility of the flowers in pairs. Houstonia. Faramea, remarkable difference in the pollen-grains of the two forms; torsion of the stamens in the short-styled form alone; development not as yet perfect. The heterostyled structure in the several Rubiaceous genera not due to descent in common.

(FIGURE 3.4. Linum grandiflorum. Left: Long-styled form. Right: Short-styled form. s, s: stigmas.)

It has long been known that several species of Linum present two forms (3/1. Treviranus has shown that this is the case in his review of my original paper 'Botanische Zeitung' 1863 page 189.), and having observed this fact in L. flavum more than thirty years ago, I was led, after ascertaining the nature of heterostylism in Primula, to examine the first species of Linum which I met with, namely, the beautiful L. grandiflorum. This plant exists under two forms, occurring in about equal numbers, which differ little in structure, but greatly in function. The foliage, corolla, stamens, and pollen-grains (the latter examined both distended with water and dry) are alike in the two forms (Figure 3.4). The difference is confined to the pistil; in the short-styled form the styles and the stigmas are only about half the length of those in the long- styled. A more important distinction is, that the five stigmas in the short- styled form diverge greatly from one another, and pass out between the filaments of the stamens, and thus lie within the tube of the corolla. In the long-styled form the elongated stigmas stand nearly upright, and alternate with the anthers. In this latter form the length of the stigmas varies considerably, their upper extremities projecting even a little above the anthers, or reaching up only to about their middle. Nevertheless, there is never the slightest difficulty in distinguishing between the two forms; for, besides the difference in the divergence of the stigmas, those of the short-styled form never reach even to the bases of the anthers. In this form the papillae on the stigmatic surfaces are shorter, darker-coloured, and more crowded together than in the long-styled form; but these differences seem due merely to the shortening of the stigma, for in the varieties of the long-styled form with shorter stigmas, the papillae are more crowded and darker-coloured than in those with the longer stigmas. Considering the slight and variable differences between the two forms of this Linum, it is not surprising that hitherto they have been overlooked.

In 1861 I had eleven plants in my garden, eight of which were long-styled, and three short-styled. Two very fine long-styled plants grew in a bed a hundred yards off all the others, and separated from them by a screen of evergreens. I marked twelve flowers, and placed on their stigmas a little pollen from the short-styled plants. The pollen of the two forms is, as stated, identical in appearance; the stigmas of the long-styled flowers were already thickly covered with their own pollen—so thickly that I could not find one bare stigma, and it was late in the season, namely, September 15th. Altogether, it seemed almost childish to expect any result. Nevertheless from my experiments on Primula, I had faith, and did not hesitate to make the trial, but certainly did not anticipate the full result which was obtained. The germens of these twelve flowers all swelled, and ultimately six fine capsules (the seed of which germinated on the following year) and two poor capsules were produced; only four capsules shanking off. These same two long-styled plants produced, in the course of the summer, a vast number of flowers, the stigmas of which were covered with their own pollen; but they all proved absolutely barren, and their germens did not even swell.

The nine other plants, six long-styled and three short-styled, grew not very far apart in my flower-garden. Four of these long-styled plants produced no seed- capsules; the fifth produced two; and the remaining one grew so close to a short-styled plant that their branches touched, and this produced twelve capsules, but they were poor ones. The case was different with the short-styled plants. The one which grew close to the long-styled plant produced ninety-four imperfectly fertilised capsules containing a multitude of bad seeds, with a moderate number of good ones. The two other short-styled plants growing together were small, being partly smothered by other plants; they did not stand very close to any long-styled plants, yet they yielded together nineteen capsules. These facts seem to show that the short-styled plants are more fertile with their own pollen than are the long-styled, and we shall immediately see that this probably is the case. But I suspect that the difference in fertility between the two forms was in this instance in part due to a distinct cause. I repeatedly watched the flowers, and only once saw a humble-bee momentarily alight on one, and then fly away. If bees had visited the several plants, there cannot be a doubt that the four long-styled plants, which did not produce a single capsule, would have borne an abundance. But several times I saw small diptera sucking the flowers; and these insects, though not visiting the flowers with anything like the regularity of bees, would carry a little pollen from one form to the other, especially when growing near together; and the stigmas of the short-styled plants, diverging within the tube of the corolla, would be more likely than the upright stigmas of the long-styled plants, to receive a small quantity of pollen if brought to them by small insects. Moreover from the greater number of the long-styled than of the short-styled plants in the garden, the latter would be more likely to receive pollen from the long-styled, than the long-styled from the short-styled.

In 1862 I raised thirty-four plants of this Linum in a hot-bed; and these consisted of seventeen long-styled and seventeen short-styled forms. Seed sown later in the flower-garden yielded seventeen long-styled and twelve short-styled forms. These facts justify the statement that the two forms are produced in about equal numbers. The thirty-four plants of the first lot were kept under a net which excluded all insects, except such minute ones as Thrips. I fertilised fourteen long-styled flowers legitimately with pollen from the short-styled, and got eleven fine seed-capsules, which contained on an average 8.6 seeds per capsule, but only 5.6 appeared to be good. It may be well to state that ten seeds is the maximum production for a capsule, and that our climate cannot be very favourable to this North-African plant. On three occasions the stigmas of nearly a hundred flowers were fertilised illegitimately with their own-form pollen, taken from separate plants, so as to prevent any possible ill effects from close inter-breeding. Many other flowers were also produced, which, as before stated, must have received plenty of their own pollen; yet from all these flowers, borne by the seventeen long-styled plants, only three capsules were produced. One of these included no seed, and the other two together gave only five good seeds. It is probable that this miserable product of two half-fertile capsules from the seventeen plants, each of which must have produced at least fifty or sixty flowers, resulted from their fertilisation with pollen from the short-styled plants by the aid of Thrips; for I made a great mistake in keeping the two forms under the same net, with their branches often interlocking; and it is surprising that a greater number of flowers were not accidentally fertilised.

Twelve short-styled flowers were in this instance castrated, and afterwards fertilised legitimately with pollen from the long-styled form; and they produced seven fine capsules. These included on an average 7.6 seeds, but of apparently good seed only 4.3 per capsule. At three separate times nearly a hundred flowers were fertilised illegitimately with their own-form pollen, taken from separate plants; and numerous other flowers were produced, many of which must have received their own pollen. From all these flowers on the seventeen short-styled plants only fifteen capsules were produced, of which only eleven contained any good seed, on an average 4.2 per capsule. As remarked in the case of the long- styled plants, some even of these capsules were perhaps the product of a little pollen accidentally fallen from the adjoining flowers of the other form on to the stigmas, or transported by Thrips. Nevertheless the short-styled plants seem to be slightly more fertile with their own pollen than the long-styled, in the proportion of fifteen capsules to three; nor can this difference be accounted for by the short-styled stigmas being more liable to receive their own pollen than the long-styled, for the reverse is the case. The greater self-fertility of the short-styled flowers was likewise shown in 1861 by the plants in my flower- garden, which were left to themselves, and were but sparingly visited by insects.

On account of the probability of some of the flowers on the plants of both forms, which were covered under the same net, having been legitimately fertilised in an accidental manner, the relative fertility of the two legitimate and two illegitimate unions cannot be compared with certainty; but judging from the number of good seeds per capsule, the difference was at least in the ratio of 100 to 7, and probably much greater.

Hildebrand tested my results, but only on a single short-styled plant, by fertilising many flowers with their own-form pollen; and these did not produce any seed. This confirms my suspicion that some of the few capsules produced by the foregoing seventeen short-styled plants were the product of accidental legitimate fertilisation. Other flowers on the same plant were fertilised by Hildebrand with pollen from the long-styled form, and all produced fruit. (3/2. 'Botanische Zeitung' January 1, 1864 page 2.)

The absolute sterility (judging from the experiments of 1861) of the long-styled plants with their own-form pollen led me to examine into its apparent cause; and the results are so curious that they are worth giving in detail. The experiments were tried on plants grown in pots and brought successively into the house.

FIRST.

Pollen from a short-styled plant was placed on the five stigmas of a long-styled flower, and these, after thirty hours, were found deeply penetrated by a multitude of pollen-tubes, far too numerous to be counted; the stigmas had also become discoloured and twisted. I repeated this experiment on another flower, and in eighteen hours the stigmas were penetrated by a multitude of long pollen- tubes. This is what might have been expected, as the union is a legitimate one. The converse experiment was likewise tried, and pollen from a long-styled flower was placed on the stigmas of a short-styled flower, and in twenty-four hours the stigmas were discoloured, twisted, and penetrated by numerous pollen-tubes; and this, again, is what might have been expected, as the union was a legitimate one.

SECONDLY.

Pollen from a long-styled flower was placed on all five stigmas of a long-styled flower on a separate plant: after nineteen hours the stigmas were dissected, and only a single pollen-grain had emitted a tube, and this was a very short one. To make sure that the pollen was good, I took in this case, and in most of the other cases, pollen either from the same anther or from the same flower, and proved it to be good by placing it on the stigma of a short-styled plant, and found numerous pollen-tubes emitted.

THIRDLY.

Repeated last experiment, and placed own-form pollen on all five stigmas of a long-styled flower; after nineteen hours and a half, not one single grain had emitted its tube.

FOURTHLY.

Repeated the experiment, with the same result after twenty-four hours.

FIFTHLY.

Repeated last experiment, and, after leaving pollen on for nineteen hours, put on an additional quantity of own-form pollen on all five stigmas. After an interval of three days, the stigmas were examined, and, instead of being discoloured and twisted, they were straight and fresh-coloured. Only one grain had emitted a quite short tube, which was drawn out of the stigmatic tissue without being ruptured.

The following experiments are more striking:—

SIXTHLY.

I placed own-form pollen on three of the stigmas of a long-styled flower, and pollen from a short-styled flower on the other two stigmas. After twenty-two hours these two stigmas were discoloured, slightly twisted, and penetrated by the tubes of numerous pollen-grains: the other three stigmas, covered with their own-form pollen, were fresh, and all the pollen-grains were loose; but I did not dissect the whole stigma.

SEVENTHLY.

Experiment repeated in the same manner, with the same result.

EIGHTHLY.

Experiment repeated, but the stigmas were carefully examined after an interval of only five hours and a half. The two stigmas with pollen from a short-styled flower were penetrated by innumerable tubes, which were as yet short, and the stigmas themselves were not at all discoloured. The three stigmas covered with their own-form pollen were not penetrated by a single pollen-tube.

NINTHLY.

Put pollen of a short-styled flower on a single long-styled stigma, and own-form pollen on the other four stigmas; after twenty-four hours the one stigma was somewhat discoloured and twisted, and penetrated by many long tubes: the other four stigmas were quite straight and fresh; but on dissecting them I found that three pollen-grains had protruded very short tubes into the tissue.

TENTHLY.

Repeated the experiment, with the same result after twenty-four hours, excepting that only two own-form grains had penetrated the stigmatic tissue with their tubes to a very short depth. The one stigma, which was deeply penetrated by a multitude of tubes from the short-styled pollen, presented a conspicuous difference in being much curled, half-shrivelled, and discoloured, in comparison with the other four straight and bright pink stigmas.

I could add other experiments; but those now given amply suffice to show that the pollen-grains of a short-styled flower placed on the stigma of a long-styled flower emit a multitude of tubes after an interval of from five to six hours, and penetrate the tissue ultimately to a great depth; and that after twenty-four hours the stigmas thus penetrated change colour, become twisted, and appear half-withered. On the other hand, pollen-grains from a long-styled flower placed on its own stigmas, do not emit their tubes after an interval of a day, or even three days; or at most only three or four grains out of a multitude emit their tubes, and these apparently never penetrate the stigmatic tissue deeply, and the stigmas themselves do not soon become discoloured and twisted.

This seems to me a remarkable physiological fact. The pollen-grains of the two forms are undistinguishable under the microscope; the stigmas differ only in length, degree of divergence, and in the size, shade of colour, and approximation of their papillae, these latter differences being variable and apparently due merely to the degree of elongation of the stigma. Yet we plainly see that the two kinds of pollen and the two stigmas are widely dissimilar in their mutual reaction—the stigmas of each form being almost powerless on their own pollen, but causing, through some mysterious influence, apparently by simple contact (for I could detect no viscid secretion), the pollen-grains of the opposite form to protrude their tubes. It may be said that the two pollens and the two stigmas mutually recognise each other by some means. Taking fertility as the criterion of distinctness, it is no exaggeration to say that the pollen of the long-styled Linum grandiflorum (and conversely that of the other form) has been brought to a degree of differentiation, with respect to its action on the stigma of the same form, corresponding with that existing between the pollen and stigma of species belonging to distinct genera.

Linum perenne.

This species is conspicuously heterostyled, as has been noticed by several authors. The pistil in the long-styled form is nearly twice as long as that of the short-styled. In the latter the stigmas are smaller and, diverging to a greater degree, pass out low down between the filaments. I could detect no difference in the two forms in the size of the stigmatic papillae. In the long- styled form alone the stigmatic surfaces of the mature pistils twist round, so as to face the circumference of the flower; but to this point I shall presently return. Differently from what occurs in L. grandiflorum, the long-styled flowers have stamens hardly more than half the length of those in the short-styled. The size of the pollen-grains is rather variable; after some doubt, I have come to the conclusion that there is no uniform difference between the grains in the two forms. The long-stamens in the short-styled form project to some height above the corolla, and their filaments are coloured blue apparently from exposure to the light. The anthers of the longer stamens correspond in height with the lower part of the stigmas of the long-styled flowers; and the anthers of the shorter stamens of the latter correspond in the same manner in height with the stigmas of the short-styled flowers.

I raised from seed twenty-six plants, of which twelve proved to be long-styled and fourteen short-styled. They flowered well, but were not large plants. As I did not expect them to flower so soon, I did not transplant them, and they unfortunately grew with their branches closely interlocked. All the plants were covered under the same net, excepting one of each form. Of the flowers on the long-styled plants, twelve were illegitimately fertilised with their own-form pollen, taken in every case from a separate plant; and not one set a seed- capsule: twelve other flowers were legitimately fertilised with pollen from short-styled flowers; and they set nine capsules, each including on an average 7 good seeds, ten being the maximum number ever produced. Of the flowers on the short-styled plants, twelve were illegitimately fertilised with own-form pollen, and they yielded one capsule, including only 3 good seeds; twelve other flowers were legitimately fertilised with pollen from long-styled flowers, and these produced nine capsules, but one was bad; the eight good capsules contained on an average 8 good seeds each. Judging from the number of seeds per capsule, the fertility of the two legitimate to that of the two illegitimate unions is as 100 to 20.

The numerous flowers on the eleven long-styled plants under the net, which were not fertilised, produced only three capsules, including 8, 4, and 1 good seeds. Whether these three capsules were the product of accidental legitimate fertilisation, owing to the branches of the plants of the two forms interlocking, I will not pretend to decide. The single long-styled plant which was left uncovered, and grew close by the uncovered short-styled plant, produced five good pods; but it was a poor and small plant.

The flowers borne on the thirteen short-styled plants under the net, which were not fertilised, produced twelve capsules, containing on an average 5.6 seeds. As some of these capsules were very fine, and as five were borne on one twig, I suspect that some minute insect had accidentally got under the net and had brought pollen from the other form to the flowers which produced this little group of capsules. The one uncovered short-styled plant which grew close to the uncovered long-styled plant yielded twelve capsules.

From these facts we have some reason to believe, as in the case of L. grandiflorum, that the short-styled plants are in a slight degree more fertile with their own pollen than are the long-styled plants. Anyhow we have the clearest evidence, that the stigmas of each form require for full fertility that pollen from the stamens of corresponding height belonging to the opposite form should be brought to them.

Hildebrand, in the paper lately referred to, confirms my results. He placed a short-styled plant in his house, and fertilised about 20 flowers with their own pollen, and about 30 with pollen from another plant belonging to the same form, and these 50 flowers did not set a single capsule. On the other hand he fertilised about 30 flowers with pollen from the long-styled form, and these, with the exception of two, yielded capsules, containing good seeds.

It is a singular fact, in contrast with what occurred in the case of L. grandiflorum, that the pollen-grains of both forms of L. perenne, when placed on their own-form stigmas, emitted their tubes, though this action did not lead to the production of seeds. After an interval of eighteen hours, the tubes penetrated the stigmatic tissue, but to what depth I did not ascertain. In this case the impotence of the pollen-grains on their own stigmas must have been due either to the tubes not reaching the ovules, or to their not acting properly after reaching them.

The plants both of L. perenne and grandiflorum, grew, as already stated, with their branches interlocked, and with scores of flowers of the two forms close together; they were covered by a rather coarse net, through which the wind, when high, passed; and such minute insects as Thrips could not, of course, be excluded; yet we have seen that the utmost possible amount of accidental fertilisation on seventeen long-styled plants in the one case, and on eleven long-styled plants in the other, resulted in the production, in each case, of three poor capsules; so that when the proper insects are excluded, the wind does hardly anything in the way of carrying pollen from plant to plant. I allude to this fact because botanists in speaking of the fertilisation of various flowers, often refer to the wind or to insects as if the alternative were indifferent. This view, according to my experience, is entirely erroneous. When the wind is the agent in carrying pollen, either from one sex to the other, or from hermaphrodite to hermaphrodite, we can recognise structure as manifestly adapted to its action as to that of insects when these are the carriers. We see adaptation to the wind in the incoherence of the pollen,—in the inordinate quantity produced (as in the Coniferae, Spinage, etc.),—in the dangling anthers well fitted to shake out the pollen,—in the absence or small size of the perianth,—in the protrusion of the stigmas at the period of fertilisation,—in the flowers being produced before they are hidden by the leaves,—and in the stigmas being downy or plumose (as in the Gramineae, Docks, etc), so as to secure the chance-blown grains. In plants which are fertilised by the wind, the flowers do not secrete nectar, their pollen is too incoherent to be easily collected by insects, they have not bright-coloured corollas to serve as guides, and they are not, as far as I have seen, visited by insects. When insects are the agents of fertilisation (and this is incomparably the more frequent case with hermaphrodite plants), the wind plays no part, but we see an endless number of adaptations to ensure the safe transport of the pollen by the living workers. These adaptations are most easily recognised in irregular flowers; but they are present in regular flowers, of which those of Linum offer a good instance, as I will now endeavour to show.

I have already alluded to the rotation of each separate stigma in the long- styled form of Linum perenne. In both forms of the other heterostyled species and in the homostyled species of Linum which I have seen, the stigmatic surfaces face the centre of the flower, with the furrowed backs of the stigmas, to which the styles are attached, facing outwards. This is the case with the stigmas of the long-styled flowers of L. perenne whilst in bud. But by the time the flowers have expanded, the five stigmas twist round so as to face the circumference, owing to the torsion of that part of the style which lies beneath the stigma. I should state that the five stigmas do not always turn round completely, two or three sometimes facing only obliquely outwards. My observations were made during October; and it is not improbable that earlier in the season the torsion would have been more complete; for after two or three cold and wet days the movement was very imperfectly performed. The flowers should be examined shortly after their expansion, as their duration is brief; as soon as they begin to wither, the styles become spirally twisted all together, the original position of the parts being thus lost.

He who will compare the structure of the whole flower in both forms of L. perenne and grandiflorum, and, as I may add, of L. flavum, will not doubt about the meaning of this torsion of the styles in the one form alone of L. perenne, as well as the meaning of the divergence of the stigmas in the short-styled form of all three species. It is absolutely necessary as we know, that insects should carry pollen from the flowers of the one form reciprocally to those of the other. Insects are attracted by five drops of nectar, secreted exteriorly at the base of the stamens, so that to reach these drops they must insert their proboscides outside the ring of broad filaments, between them and the petals. In the short-styled form of the above three species, the stigmas face the axis of the flower; and had the styles retained their original upright and central position, not only would the stigmas have presented their backs to the insects which sucked the flowers, but their front and fertile surfaces would have been separated from the entering insects by the ring of broad filaments, and would never have received any pollen. As it is, the styles diverge and pass out between the filaments. After this movement the short stigmas lie within the tube of the corolla; and their papillous surfaces being now turned upwards are necessarily brushed by every entering insect, and thus receive the required pollen.

In the long-styled form of L. grandiflorum, the almost parallel or slightly diverging anthers and stigmas project a little above the tube of the somewhat concave flower; and they stand directly over the open space leading to the drops of nectar. Consequently when insects visit the flowers of either form (for the stamens in this species occupy the same position in both forms), they will get their foreheads or proboscides well dusted with the coherent pollen. As soon as they visit the flowers of the long-styled form they will necessarily leave pollen on the proper surface of the elongated stigmas; and when they visit the short-styled flowers, they will leave pollen on the upturned stigmatic surfaces. Thus the stigmas of both forms will receive indifferently the pollen of both forms; but we know that the pollen alone of the opposite form causes fertilisation.

(Figure 3.5. Long-styled form of L. perenne var. Austriacum in its early condition before the stigmas have rotated. The petals and calyx have been removed on the near side. (3/3. I neglected to get drawings made from fresh flowers of the two forms. But Mr. Fitch has made the above sketch of a long- styled flower from dried specimens and from published engravings. His well-known skill ensures accuracy in the proportional size of the parts.)

In the case of L. perenne, affairs are arranged more perfectly; for the stamens in the two forms stand at different heights, so that pollen from the anthers of the longer stamens will adhere to one part of an insect's body, and will afterwards be brushed off by the rough stigmas of the longer pistils; whilst pollen from the anthers of the shorter stamens will adhere to a different part of the insect's body, and will afterwards be brushed off by the stigmas of the shorter pistils; and this is what is required for the legitimate fertilisation of both forms. The corolla of L. perenne is more expanded than that of L. grandiflorum, and the stigmas of the long-styled form do not diverge greatly from one another; nor do the stamens of either form. Hence insects, especially rather small ones, will not insert their proboscides between the stigmas of the long-styled form, nor between the anthers of either form (Figure 3.5), but will strike against them, at nearly right angles, with the backs of their head or thorax. Now, in the long-styled flowers, if each stigma did not rotate on its axis, insects in visiting them would strike their heads against the backs of the stigmas; as it is, they strike against that surface which is covered with papillae, with their heads already charged with pollen from the stamens of corresponding height borne by the flowers of the other form, and legitimate fertilisation is thus ensured.

Thus we can understand the meaning of the torsion of the styles in the long- styled flowers alone, as well as their divergence in the short-styled flowers.

One other point is worth notice. In botanical works many flowers are said to be fertilised in the bud. This statement generally rests, as far as I can discover, on the anthers opening in the bud; no evidence being adduced that the stigma is at this period mature, or that it is not subsequently acted on by pollen brought from other flowers. In the case of Cephalanthera grandiflora I have shown that precocious and partial self-fertilisation, with subsequent full fertilisation, is the regular course of events. (3/4. 'Fertilisation of Orchids' page 108; 2nd edition 1877 page 84.) The belief that the flowers of many plants are fertilised in the bud, that is, are perpetually self-fertilised, is a most effectual bar to understanding their real structure. I am, however, far from wishing to assert that some flowers, during certain seasons, are not fertilised in the bud; for I have reason to believe that this is the case. A good observer, resting his belief on the usual kind of evidence, states that in Linum Austriacum (which is heterostyled, and is considered by Planchon as a variety of L. perenne) the anthers open the evening before the expansion of the flowers, and that the stigmas are then almost always fertilised. (3/5. H. Lecoq 'Etudes sur la Geogr. Bot.' 1856 tome 5 page 325.) Now we know positively that, so far from Linum perenne being fertilised by its own pollen in the bud, its own pollen is as powerless on the stigma as so much inorganic dust.

Linum flavum.

The pistil of the long-styled form of this species is nearly twice as long as that of the short-styled; the stigmas are longer and the papillae coarser. In the short-styled form the stigmas diverge and pass out between the filaments, as in the previous species. The stamens in the two forms differ in length; and, what is singular, the anthers of the longer stamens are not so long as those of the other form; so that in the short-styled form both the stigmas and the anthers are shorter than in the long-styled form. The pollen-grains of the two forms do not differ in size. As this species is propagated by cuttings, generally all the plants in the same garden belong to the same form. I have inquired, but have never heard of its seeding in this country. Certainly my own plants never produced a single seed as long as I possessed only one of the two forms. After considerable search I procured both forms, but from want of time only a few experiments were made. Two plants of the two forms were planted some way apart in my garden, and were not covered by nets. Three flowers on the long- styled plant were legitimately fertilised with pollen from the short-styled plant, and one of them set a fine capsule. No other capsules were produced by this plant. Three flowers on the short-styled plant were legitimately fertilised with pollen from the long-styled, and all three produced capsules, containing respectively no less than 8, 9, and 10 seeds. Three other flowers on this plant, which had not been artificially fertilised, produced capsules containing 5, 1, and 5 seeds; and it is quite possible that pollen may have been brought to them by insects from the long-styled plant growing in the same garden. Nevertheless, as they did not yield half the number of seeds compared with the other flowers on the same plant which had been artificially and legitimately fertilised, and as the short-styled plants of the two previous species apparently evince some slight capacity for fertilisation with their own-form pollen, these three capsules may have been the product of self-fertilisation.

Besides the three species now described, the yellow-flowered L. corymbiferum is certainly heterostyled, as is, according to Planchon, L. salsoloides. (3/6. Hooker's 'London Journal of Botany' 1848 volume 7 page 174.) This botanist is the only one who seems to have inferred that heterostylism might have some important functional bearing. Dr. Alefeld, who has made a special study of the genus, says that about half of the sixty-five species known to him are heterostyled. (3/7. 'Botanische Zeitung' September 18, 1863 page 281.) This is the case with L. trigynum, which differs so much from the other species that it has been formed by him into a distinct genus. (3/8. It is not improbable that the allied genus, Hugonia, is heterostyled, for one species is said by Planchon (Hooker's 'London Journal of Botany' 1848 volume 7 page 525) to be provided with "staminibus exsertis;" another with "stylis staminibus longioribus," and another has "stamina 5, majora, stylos longe superantia.") According to the same author, none of the species which inhabit America and the Cape of Good Hope are heterostyled.

I have examined only three homostyled species, namely, L. usitatissimum, angustifolium, and catharticum. I raised 111 plants of a variety of the first- named species, and these, when protected under a net, all produced plenty of seed. The flowers, according to H. Muller, are frequented by bees and moths. (3/9. 'Die Befruchtung der Blumen' etc. page 168.) With respect to L. catharticum, the same author shows that the flowers are so constructed that they can freely fertilise themselves; but if visited by insects they might be cross- fertilised. He has, however, only once seen the flowers thus visited during the day; but it may be suspected that they are frequented during the night by small moths for the sake of the five minute drops of nectar secreted. Lastly, L. Lewisii is said by Planchon to bear on the same plant flowers with stamens and pistils of the same height, and others with the pistils either longer or shorter than the stamens. This case formerly appeared to me an extraordinary one; but I am now inclined to believe that it is one merely of great variability. (3/10. Planchon in Hooker's 'London Journal of Botany' 1848 volume 7 page 175. See on this subject Asa Gray in 'American Journal of Science' volume 36 September 1863 page 284.)

PULMONARIA (BORAGINEAE).

Pulmonaria officinalis.

Hildebrand has published a full account of this heterostyled plant. (3/11. 'Botanische Zeitung' 1865 January 13 page 13.) The pistil of the long-styled form is twice as long as that of the short-styled; and the stamens differ in a corresponding, though converse, manner. There is no marked difference in the shape or state of surface of the stigma in the two forms. The pollen-grains of the short-styled form are to those of the long-styled as 9 to 7, or as 100 to 78, in length, and as 7 to 6 in breadth. They do not differ in the appearance of their contents. The corolla of the one form differs in shape from that of the other in nearly the same manner as in Primula; but besides this difference the flowers of the short-styled are generally the larger of the two. Hildebrand collected on the Siebengebirge, ten wild long-styled and ten short-styled plants. The former bore 289 flowers, of which 186 (i.e. 64 per cent) had set fruit, yielding 1.88 seed per fruit. The ten short-styled plants bore 373 flowers, of which 262 (i.e. 70 per cent) had set fruit, yielding 1.86 seed per fruit. So that the short-styled plants produced many more flowers, and these set a rather larger proportion of fruit, but the fruits themselves yielded a slightly lower average number of seeds than did the long-styled plants. The results of Hildebrand's experiments on the fertility of the two forms are given in Table 3.19.

TABLE 3.19. Pulmonaria officinalis (from Hildebrand).

Column 1: Nature of the Union. Column 2: Number of Flowers fertilised. Column 3: Number of Fruits produced. Column 4: Average Number of Seeds per Fruit.

Long-styled by pollen of short-styled. Legitimate union : 14 : 10 : 1.30.

Long-styled 14 by own-pollen, and 16 by pollen of other plant of same form. Illegitimate union : 30 : 0 : 0.

Short-styled by pollen of long-styled. Legitimate union: 16 : 14 : 1.57.

Short-styled 11 by own-pollen, 14 by pollen of other plant of same form. Illegitimate union : 25 : 0 : 0.

In the summer of 1864, before I had heard of Hildebrand's experiments, I noticed some long-styled plants of this species (named for me by Dr. Hooker) growing by themselves in a garden in Surrey; and to my surprise about half the flowers had set fruit, several of which contained 2, and one contained even 3 seeds. These seeds were sown in my garden and eleven seedlings thus raised, all of which proved long-styled, in accordance with the usual rule in such cases. Two years afterwards the plants were left uncovered, no other plant of the same genus growing in my garden, and the flowers were visited by many bees. They set an abundance of seeds: for instance, I gathered from a single plant rather less than half of the seeds which it had produced, and they numbered 47. Therefore this illegitimately fertilised plant must have produced about 100 seeds; that is, thrice as many as one of the wild long-styled plants collected on the Siebengebirge by Hildebrand, and which, no doubt, had been legitimately fertilised. In the following year one of my plants was covered by a net, and even under these unfavourable conditions it produced spontaneously a few seeds. It should be observed that as the flowers stand either almost horizontally or hang considerably downwards, pollen from the short stamens would be likely to fall on the stigma. We thus see that the English long-styled plants when illegitimately fertilised were highly fertile, whilst the German plants similarly treated by Hildebrand were completely sterile. How to account for this wide discordance in our results I know not. Hildebrand cultivated his plants in pots and kept them for a time in the house, whilst mine were grown out of doors; and he thinks that this difference of treatment may have caused the difference in our results. But this does not appear to me nearly a sufficient cause, although his plants were slightly less productive than the wild ones growing on the Siebengbirge. My plants exhibited no tendency to become equal-styled, so as to lose their proper long-styled character, as not rarely happens under cultivation with several heterostyled species of Primula; but it would appear that they had been greatly affected in function, either by long-continued cultivation or by some other cause. We shall see in a future chapter that heterostyled plants illegitimately fertilised during several successive generations sometimes become more self-fertile; and this may have been the case with my stock of the present species of Pulmonaria; but in this case we must assume that the long-styled plants were at first sufficiently fertile to yield some seed, instead of being absolutely self-sterile like the German plants.

Pulmonaria angustifolia.

(FIGURE 3.6. Pulmonaria angustifolia. Left: Long-styled form. Right: Short-styled form.)

Seedlings of this plant, raised from plants growing wild in the Isle of Wight, were named for me by Dr. Hooker. It is so closely allied to the last species, differing chiefly in the shape and spotting of the leaves, that the two have been considered by several eminent botanists—for instance, Bentham—as mere varieties. But, as we shall presently see, good evidence can be assigned for ranking them as distinct. Owing to the doubts on this head, I tried whether the two would mutually fertilise one another. Twelve short-styled flowers of P. angustifolia were legitimately fertilised with pollen from long-styled plants of P. officinalis (which, as we have just seen, are moderately self-fertile), but they did not produce a single fruit. Thirty-six long-styled flowers of P. angustifolia were also illegitimately fertilised during two seasons with pollen from the long-styled P. officinalis, but all these flowers dropped off unimpregnated. Had the plants been mere varieties of the same species these illegitimate crosses would probably have yielded some seeds, judging from my success in illegitimately fertilising the long-styled flowers of P. officinalis; and the twelve legitimate crosses, instead of yielding no fruit, would almost certainly have yielded a considerable number, namely, about nine, judging from the results given in Table 3.20. Therefore P. officinalis and angustifolia appear to be good and distinct species, in conformity with other important functional differences between them, immediately to be described.

TABLE 3.20. Pulmonaria angustifolia.

Column 1: Nature of the Union. Column 2: Number of Flowers fertilised. Column 3: Number of Fruits produced. Column 4: Average Number of Seeds per Fruit.

Long-styled by pollen of short-styled. Legitimate union : 18 : 9 : 2.11.

Long-styled by own-form pollen. Illegitimate union : 18 : 0 : 0.

Short-styled by pollen of long-styled. Legitimate union: 18 : 15 : 2.60.

Short-styled by own-form pollen. Illegitimate union : 12 : 7 : 1.86.

The long-styled and short-styled flowers of P. angustifolia differ from one another in structure in nearly the same manner as those of P. officinalis. But in Figure 3.6 a slight bulging of the corolla in the long-styled form, where the anthers are seated, has been overlooked. My son William, who examined a large number of wild plants in the Isle of Wight, observed that the corolla, though variable in size, was generally larger in the long-styled flowers than in the short-styled; and certainly the largest corollas of all were found on the long- styled plants, and the smallest on the short-styled. Exactly the reverse occurs, according to Hildebrand, with P. officinalis. Both the pistils and stamens of P. angustifolia vary much in length; so that in the short-styled form the distance between the stigma and the anthers varied from 119 to 65 divisions of the micrometer, and in the long-styled from 115 to 112. From an average of seven measurements of each form the distance between these organs in the long-styled is to the same distance in the short-styled form as 100 to 69; so that the stigma in the one form does not stand on a level with the anthers in the other. The long-styled pistil is sometimes thrice as long as that of the short-styled; but from an average of ten measurements of both, its length to that of the short-styled was as 100 to 56. The stigma varies in being more or less, though slightly, lobed. The anthers also vary much in length in both forms, but in a greater degree in the long-styled than in the short-styled-form; many in the former being from 80 to 63, and in the latter from 80 to 70 divisions of the micrometer in length. From an average of seven measurements, the short-styled anthers were to those from the long-styled as 100 to 91 in length. Lastly, the pollen-grains from the long-styled flowers varied between 13 and 11.5 divisions of the micrometer, and those from the short-styled between 15 and 13. The average diameter of 25 grains from the latter, or short-styled form, was to that of 20 grains from the long-styled as 100 to 91. We see, therefore, that the pollen-grains from the smaller anthers of the shorter stamens in the long-styled form are, as usual, of smaller size than those in the other form. But what is remarkable, a larger proportion of the grains were small, shrivelled, and worthless. This could be seen by merely comparing the contents of the anthers from several distinct plants of each form. But in one instance my son found, by counting, that out of 193 grains from a long-styled flower, 53 were bad, or 27 per cent; whilst out of 265 grains from a short-styled flower only 18 were bad, or 7 per cent. From the condition of the pollen in the long-styled form, and from the extreme variability of all the organs in both forms, we may perhaps suspect that the plant is undergoing a change, and tending to become dioecious.

My son collected in the Isle of Wight on two occasions 202 plants, of which 125 were long-styled and 77 short-styled; so that the former were the more numerous. On the other hand, out of 18 plants raised by me from seed, only 4 were long- styled and 14 short-styled. The short-styled plants seemed to my son to produce a greater number of flowers than the long-styled; and he came to this conclusion before a similar statement had been published by Hildebrand with respect to P. officinalis. My son gathered ten branches from ten different plants of both forms, and found the number of flowers of the two forms to be as 100 to 89, 190 being short-styled and 169 long-styled. With P. officinalis the difference, according to Hildebrand, is even greater, namely, as 100 flowers for the short- styled to 77 for the long-styled plants. Table 3.20 shows the results of my experiments.

We see in Table 3.20 that the fertility of the two legitimate unions to that of the two illegitimate together is as 100 to 35, judged by the proportion of flowers which produced fruit; and as 100 to 32, judged by the average number of seeds per fruit. But the small number of fruit yielded by the 18 long-styled flowers in the first line was probably accidental, and if so, the difference in the proportion of legitimately and illegitimately fertilised flowers which yield fruit is really greater than that represented by the ratio of 100 to 35. The 18 long-styled flowers illegitimately fertilised yielded no seeds,—not even a vestige of one. Two long-styled plants which were placed under a net produced 138 flowers, besides those which were artificially fertilised, and none of these set any fruit; nor did some plants of the same form which were protected during the next summer. Two other long-styled plants were left uncovered (all the short-styled plants having been previously covered up), and humble-bees, which had their foreheads white with pollen, incessantly visited the flowers, so that their stigmas must have received an abundance of pollen, yet these flowers did not produce a single fruit. We may therefore conclude that the long-styled plants are absolutely barren with their own-form pollen, though brought from a distinct plant. In this respect they differ greatly from the long-styled English plants of P. officinalis which were found by me to be moderately self-fertile; but they agree in their behaviour with the German plants of P. officinalis experimented on by Hildebrand.

Eighteen short-styled flowers legitimately fertilised yielded, as may be seen in Table 3.20, 15 fruits, each having on an average 2.6 seeds. Four of these fruits contained the highest possible number of seeds, namely 4, and four other fruits contained each 3 seeds. The 12 illegitimately fertilised short-styled flowers yielded 7 fruits, including on an average 1.86 seed; and one of these fruits contained the maximum number of 4 seeds. This result is very surprising in contrast with the absolute barrenness of the long-styled flowers when illegitimately fertilised; and I was thus led to attend carefully to the degree of self-fertility of the short-styled plants. A plant belonging to this form and covered by a net bore 28 flowers besides those which had been artificially fertilised, and of all these only two produced a fruit each including a single seed. This high degree of self-sterility no doubt depended merely on the stigmas not receiving any pollen, or not a sufficient quantity. For after carefully covering all the long-styled plants in my garden, several short-styled plants were left exposed to the visits of humble-bees, and their stigmas will thus have received plenty of short-styled pollen; and now about half the flowers, thus illegitimately fertilised, set fruit. I judge of this proportion partly from estimation and partly from having examined three large branches, which had borne 31 flowers, and these produced 16 fruits. Of the fruits produced 233 were collected (many being left ungathered), and these included on an average 1.82 seed. No less than 16 out of the 233 fruits included the highest possible number of seeds, namely 4, and 31 included 3 seeds. So we see how highly fertile these short-styled plants were when illegitimately fertilised with their own-form pollen by the aid of bees.