As regards the isolation of species I do not understand why you should suppose that the facts of hybridisation to which you allude should in any way modify my 'belief.' As fully set forth in 'Physiological Selection,' what I maintain is that the origin of species is in _all_ cases due to isolation of _some kind_, but that only in the case of differential fertility can physo. sel. have been the kind of isolation at work. Therefore, it would be fatal to my views if all species were cross-sterile, because this would prove vastly too much. What the theory of phy. sel. requires is exactly what occurs, viz. cross-sterility between allied species in nearly all cases _where species have been differentiated on common areas_ or identical stations, and more or less complete cross-fertility where they have been differentiated on different (discontinuous) areas, or else prevented from intercrossing by yet some other means of isolation. I have collected a quantity of evidence in favour of both these otherwise inexplicable correlations. But I should like to know the species of wild fowl which you have found to be hybridisable or cross-fertile, so that I may ascertain whether their natural breeding areas are, or are not, identical. Of course I should expect them not to be. I have been told to save my eyes as much as possible, and therefore conduct most of my correspondence by dictation. But not being used to this process, I find it even more difficult than before to express my meaning with clearness, so I will tackle with my own hand what you say about Aquilegias. I have looked up the group, and find that, with the exception of vulgaris (common columbine), all the European species seem to occupy restricted areas, or else well-isolated stations. Also, that the same seems to apply as a very general rule to other species all over the world, for, wherever mountains are concerned, stations are apt to be isolated by difference of altitude, &c. Now if such be the case with the group in question, the fact of its constituent species being freely hybridisable when artificially brought together is exactly what my theory requires. For the specific differentiation has presumably been effected by geographical (or topographical) isolation, without physiological having had anything to do with it. In fact, as stated over and over again in my original paper, _this_ correlation between geographical isolation and cross-fertility is _one_ of my lines of verification, the _other_ line being the correlation between identical stations and cross-sterility. Now, as above stated, I have found both these correlations to obtain in a surprisingly general manner.