3. V. lychnitis, alba, by pollen V. thapsus, alba, is as 39: 30 of V. lychnitis, lutea, by pollen of V. thapsus, alba.

4. V. lychnitis, alba, by pollen of V. phoeniceum, alla as 56: 42 of V. lychnitis, lutea, by pollen of V. phoeniceum, alba.

5. V. blattaria, alba, by its own pollen, is as 98: 90 of V. blattaria,

lutea by its own pollen.

6. V. blattaria, lutea, by pollen of V. blattaria, alba is as 96: 79 of F. blattaria, alba, by pollen of V. blattaria, lutea.

7. V. blattaria, lutea, by pollen of V. thapsus, alba, as 61: 43 of V. blattaria, alba, by pollen of V. thapsus, alba.

8. V. blattaria, lutea by pollen of V. lychnitis, alba, as 45: 36 of V. blattaria, alba, by pollen of V. lychnitis, alba.

We thus see, from the eight pure, cross, and hybrid unions of V. blattaria alba and lutea given in the above comparative table, that though the white variety exceeds in fertility the yellow variety, when both are fertilised by their own pollen, the yellow variety, in the mixed unions, is in every case more highly fertile than the white. In the different unions of V. lychnitis, alba and lutea, there is some little discordance, this, however, is confined to the hybrid unions which are as yet very insufficiently illustrated, as may be seen by consulting Tables 3 and 4. In the case of the pure and cross unions, we see, as in those of V. blattaria, that in the pure unions the white variety, and in the cross unions the yellow variety is the more fertile.

I know not whether this concordance is casual or otherwise, but I was so forcibly struck with it in the comparative study of my Tables, that I have thus ventured a special statement. I have been more especially induced to notice it also from its evidently bearing and illustrating, as I am inclined to think, that view of Mr. Darwin, (loc. cit.) respecting the good derived from cross fertilisation; inasmuch as we see that the yellow and original, or normally coloured, form of the species is less fertile than the white or derivative form in the pure unions, whereas in general, in the mixed unions, the yellow variety relatively exceeds the white in the degree of fertility. Any how, the mere fact of such variations occurring, whether or not they have any bearing on other points of theoretical natural science, seems to me worth noticing, as affording an additional link to that broken chain of

evidence which is said to disjoin the serial continuity of the phenomena of mongrelism and hybridism.

In Table 7 we have several unions of the yellow variety of F. thapsus. If we compare these results, we see that the fertility of the V. thapsus, lutea, by its cross-unions with the V. thapsus, alba, is decreased in the proportions of 94 relatively to 100, the product of fertilisation by its own pollen. We also see a great difference in the degrees of potency of the two pollens of the white and yellow variety of V. lychnitis on the stigmas of the yellow variety of V. thapsus ; the pollen of V. lychnitis, alba, exceeding in its fertilising influence that of V. lychnitis, lutea, in the proportion of 54: 46. Judging from the results of the seven hybrid unions given in this Table, we also see how little the recognised systematic affinities of species guide us in pronouncing a priori as to the degree of fertility of their several unions. For example V. thapsiforme, V. virgatum and V. blattaria,

though much more closely allied to the V. thapsus than the others given in Table, are nevertheless least effective in their conjunctive fertility with the latter species. Furthermore, we see by those unions of V. thapsus, lutea, as female, with the yellow and white varieties of V. lychnitis, and of V. pyramidatum; that though the pollen of '. pyramidatum is equally potent on the stigma of V. thapsus lutea, as is that of V. lychnitis, alba, there is nevertheless a considerable decrease in the proportionate fertility of the unions with V. lychnitis, lutea. Hence, as we have before shown it to be with the varieties of V. phoniceum, and judging by the physiological test, the V. pyramidatum would interpolate itself between these slightly different and undoubted varieties of a species.

In the foregoing Tables, then, I have given nearly all the results of my experiments in the unions of Verbasca. Before considering the nature of the evidence they afford us as to the relationship of mongrelism and hybridism, I will briefly attempt to show how far these results accord with those of Gartner, who has also largely experimented on these plants. I may premise, however, that as my experiments are much less numerous than Gartner's, comprising some 57 distinct unions, in which 527 flowers were artificially fertilised,-whereas, as will be seen beneath, Gartner subjected no less than 1085 flowers to experiment, they would induce very different conclusions, in certain points, to those arrived at by that careful experimentalist. I readily acknowledge therefore the higher claim of the latter to a provisional acceptance, until further experiments show more conclusively their relative correctness. I have also to notice a cause of some little discordance in such a comparative examination as that which I am about to institute; namely, that I have given in every case the average number of seeds produced both by pure and mixed unions, whereas Gartner gives the average number of seeds in the pure unions only, taking in each case the maximum or highest number produced by a single capsule in the mixed unions. I was not aware of this peculiarity in Gartner's deductions when I counted the seeds in my own experiments, otherwise, I should have drawn them up for the sake of comparison on a similar basis; even though I consider it a less fair method than that which I have adopted, in all such cases as the present, in which the ovaries

contain an indefinite number of ovules. And this the more especially if, as in my own experiments, castration and artificial impregnation be performed in both pure and mixed unions. In drawing comparisons between uncastrated pure unions, and castrated mixed unions, the average of the former, with the maximum of the latter would certainly be the fairer method, as affording a complement for the sterilising influence of castration.

For the following digest of Gartner's experiments I have to thank Mr. Darwin, who kindly sent it to me from his yet unpublished MS. illustrations of these phenomena: "To show the scale on which Gartner worked, I may state that, in the genus Verbascum, he crossed no less than 1085 flowers and counted their seed, and recorded the results. Now in two of his works he distinctly asserts that similarly coloured varieties of V. lychnitis and V. blattaria are more fertile together than when differently coloured varieties of the same species are crossed. But Gartner chiefly relied on the crosses which he made between the yellow and white varieties of these two species and nine other distinct species, and he asserts that the white-flowering species yielded more seed than did the yellow-flowered varieties when crossed with the same white varieties of these two-flowered species, and so conversely with the yellow flowering varieties with the yellow species. The general results may be seen in his Table. In one case he gives the following details; the white Verbascum lychnitis naturally fertilised with its own pollen had on an average in 12 capsules 96 good seeds: 20 flowers artificially fertilised with the pollen of its yellow variety gave as the maximum 89 good seeds. I should have thought that this slight difference might have been wholly due to the evil effects of castration; but Gartner shows that the white variety of V. lychnitis, fertilised by the pollen of the white and yellow varieties of V. blattaria, in both of which cases there must have been previous castration, bore seeds to the white variety in the proportion of 62, to 43 when pollen of the yellow variety was used."

First then, in regard to the greater fertility of the unions of similarily coloured varieties, relatively to that of the unions of dissimilarly coloured varieties of the same species. To these phenomena I will apply in the subsequent parts of this paper the following terms: "Homochromatic" to the unions of similarly coloured varieties, and "hetero

chromatic" to those in which dissimilarly coloured varieties are united In the following table we will at once see the comparative fertility of these different unions given in the previous ones.

RELATIVE FERTILITY OF THE HOMOCHROMATIC AND HETEROCHROMATIC

958

...

867

...

735

680

...

...

563

822

944

792

969

947

...

...

2. V. phoeniceum, rosea, by pollen of V. phoeniceum, 3. V. phoeniceum, alba, by pollen of V. phæniceum, rosea, 4. V. phoeniceum, alba, by pollen of V. phaniceum,. 5. V. phoeniceum, by pollen of V. phoeniceum, rosa, 6. V. phæniceum, by pollen of V. phoeniceum, alba, 7. V. lychnitis, alba, by pollen of V. lychnitis, lutea, 8. V. lychniits, lutea, by pollen of V. lychnitis, alba,. 9. V. blattaria, alba, by pollen of V. blattaria, lutea, 10. V. blattaria, lutea, by pollen of V. blattaria, alba, 11. V. thapsus, lutea, by pollen of V. thapsus, alba, Here the comparative fertility is shown by calculation from the number of seeds produced by 20 assumed capsules of both unions. The various cross-unions of V. phoeniceum and its varieties are in each case to be considered relatively to the assumed results of the pure unions of V. phoeniceum given in Table 2, these plants experimented upon being individually self-sterile as shown in Table 1. The unions, on the other hand, of V. lychnitis, blattaria, and thapsus, with their respective varieties, are each to be considered relatively to the 1000 seeds produced by the pure union of that variety given as female. Now in all the above heterochromatic unions, as compared with the homochromatic, we have the clearest evidence of reduced fertility. Thus, taking the 10 heterochromatic unions given, and comparing them with a similiar number of homochromatic unions, we find that the average proportion in which the former exceeds the latter, is as '05 to 23. On again confining ourselves to those species alone which have the yellow and white varieties, and keeping the unions of white as with yellow 8, distinct from those of yellow as with white as , we find that the cross-unions with white as female are to the pure unions of the latter as 04 to 28; and in those cross-unions with yellow as female,