Thus Quercus robur has twenty-eight varieties, all of which, excepting six, are clustered round three sub- species, namely Q.

pedunculata, sessiliflora and pubescens.

The forms which connect these three sub-species are comparatively rare; and, as Asa Gray again remarks, if these connecting forms which are now rare were to become totally extinct the three sub-species would hold exactly the same relation to each other as do the four or five provisionally admitted species which closely surround the typical Quercus robur.

Finally, De Candolle admits that out of the 300 species, which will be enumerated in his Prodromus as belonging to the oak family, at least two-thirds are provisional species, that is, are not known strictly to fulfil the definition above given of a true species.

It should be added that De Candolle no longer believes that species are immutable creations, but concludes that the derivative theory is the most natural one, "and the most accordant with the known facts in palaeontology, geographical botany and zoology, of anatomical structure and classification.