Almost everyone who has studied the structure of papilionaceous flowers has been convinced that they are specially adapted for cross-fertilisation, although many of the species are likewise capable of self-fertilisation.  The case therefore of Lathyrus odoratus or the sweet-pea is curious, for in this country it seems invariably to fertilise itself.  I conclude that this is so, as five varieties, differing greatly in the colour of their flowers but in no other respect, are commonly sold and come true; yet on inquiry from two great raisers of seed for sale, I find that they take no precautions to insure purity—­the five varieties being habitually grown close together. (5/11.  See Mr. W. Earley in ‘Nature’ 1872 page 242, to the same effect.  He once, however, saw bees visiting the flowers, and supposed that on this occasion they would have been intercrossed.) I have myself purposely made similar trials with the same result.  Although the varieties always come true, yet, as we shall presently see, one of the five well-known varieties occasionally gives birth to another, which exhibits all its usual characters.  Owing to this curious fact, and to the darker-coloured varieties being the most productive, these increase, to the exclusion of the others, as I was informed by the late Mr. Masters, if there be no selection.

In order to ascertain what would be the effect of crossing two varieties, some flowers on the Purple sweet-pea, which has a dark reddish-purple standard-petal with violet-coloured wing-petals and keel, were castrated whilst very young, and were fertilised with pollen of the Painted Lady.  This latter variety has a pale cherry-coloured standard, with almost white wings and keel.  On two occasions I raised from a flower thus crossed plants perfectly resembling both parent-forms; but the greater number resembled the paternal variety.  So perfect was the resemblance, that I should have suspected some mistake in the label, had not the plants, which were at first identical in appearance with the father or Painted Lady, later in the season produced flowers blotched and streaked with dark purple.  This is an interesting example of partial reversion in the same individual plant as it grows older.  The purple-flowered plants were thrown away, as they might possibly have been the product of the accidental self-fertilisation of the mother-plant, owing to the castration not having been effectual.  But the plants which resembled in the colour of their flowers the paternal variety or Painted Lady were preserved, and their seeds saved.  Next summer many plants were raised from these seeds, and they generally resembled their grandfather the Painted Lady, but most of them had their wing-petals streaked and stained with dark pink; and a few had pale purple wings with the standard of a darker crimson than is natural to the Painted Lady, so that they formed a new sub-variety.  Amongst these plants a single one appeared having purple flowers like those