I have spoken of the legitimate unions as being fully fertile; and I am fully justified in doing so, for flowers artificially fertilised in this manner yielded rather more seeds than plants naturally fertilised in a state of nature. The excess may be attributed to the plants having been grown separately in good soil. With respect to the illegitimate unions, we shall best appreciate their degree of lessened fertility by the following facts. Gartner estimated the sterility of the unions between distinct species, in a manner which allows of a strict comparison with the results of the legitimate and illegitimate unions of Primula. (1/6. ‘Versuche uber die Bastarderzeugung’ 1849 page 216.) With P. veris, for every 100 seeds yielded by the two legitimate unions, only 64 were yielded by an equal number of good capsules from the two illegitimate unions. With P. Sinensis, as we shall hereafter see, the proportion was nearly the same- -namely, as 100 to 62. Now Gartner has shown that, on the calculation of Verbascum lychnitis yielding with its own pollen 100 seeds, it yielded when fertilised by the pollen of Verbascum Phoeniceum 90 seeds; by the pollen of Verbascum nigrum, 63 seeds; by that of Verbascum blattaria, 62 seeds. So again, Dianthus barbatus fertilised by the pollen of D. superbus yielded 81 seeds, and by the pollen of D. japonicus 66 seeds, relatively to the 100 seeds produced by its own pollen. We thus see--and the fact is highly remarkable--that with Primula the illegitimate unions relatively to the legitimate are more sterile than crosses between distinct species of other genera relatively to their pure unions. Mr. Scott has given a still more striking illustration of the same fact: he crossed Primula auricula with pollen of four other species (P. palinuri, viscosa, hirsuta, and verticillata), and these hybrid unions yielded a larger average number of seeds than did P. auricula when fertilised illegitimately with its own-form pollen. (1/7. ‘Journal of the Linnean Society Botany’ volume 8 1864 page 93.)