of the grandmother, but with the petals slightly streaked with a paler tint:  this was thrown away.  Seeds were again saved from the foregoing plants, and the seedlings thus raised still resembled the Painted Lady, or great-grandfather; but they now varied much, the standard petal varying from pale to dark red, in a few instances with blotches of white; and the wing-petals varied from nearly white to purple, the keel being in all nearly white.

As no variability of this kind can be detected in plants raised from seeds, the parents of which have grown during many successive generations in close proximity, we may infer that they cannot have intercrossed.  What does occasionally occur is that in a row of plants raised from seeds of one variety, another variety true of its kind appears; for instance, in a long row of Scarlets (the seeds of which had been carefully gathered from Scarlets for the sake of this experiment) two Purples and one Painted Lady appeared.  Seeds from these three aberrant plants were saved and sown in separate beds.  The seedlings from both the Purples were chiefly Purples, but with some Painted Ladies and some Scarlets.  The seedlings from the aberrant Painted Lady were chiefly Painted Ladies with some Scarlets.  Each variety, whatever its parentage may have been, retained all its characters perfect, and there was no streaking or blotching of the colours, as in the foregoing plants of crossed origin.  Another variety, however, is often sold, which is striped and blotched with dark purple; and this is probably of crossed origin, for I found, as well as Mr. Masters, that it did not transmit its characters at all truly.

From the evidence now given, we may conclude that the varieties of the sweet-pea rarely or never intercross in this country; and this is a highly remarkable fact, considering, firstly, the general structure of the flowers; secondly, the large quantity of pollen produced, far more than is requisite for self-fertilisation; and thirdly, the occasional visit of insects.  That insects should sometimes fail to cross-fertilise the flowers is intelligible, for I have thrice seen humble-bees of two kinds, as well as hive-bees, sucking the nectar, and they did not depress the keel-petals so as to expose the anthers and stigma; they were therefore quite inefficient for fertilising the flowers.  One of these bees, namely, Bombus lapidarius, stood on one side at the base of the standard and inserted its proboscis beneath the single separate stamen, as I afterwards ascertained by opening the flower and finding this stamen prised up.  Bees are forced to act in this manner from the slit in the staminal tube being closely covered by the broad membranous margin of the single stamen, and from the tube not being perforated by nectar-passages.  On the other hand, in the three British species of Lathyrus which I have examined, and in the allied genus Vicia, two nectar-passages are present.  Therefore British bees might well be puzzled how to act in the case