about that each type maintains a high degree of integrity in its own region we can only imagine. Each has in actual fact a stability which the intermediate forms have not, but we cannot yet analyse the nature of that stability. Mendelian conceptions show us how by segregation the integrity of the factors can be in some degree maintained, but not why certain combinations of factors should be exceptionally stable. All that is left us to fall back on is the old unsatisfying suggestions that some combinations _may_ have greater viability than others, that there _may_ be a tendency for like to mate with like, and so forth. These difficulties acquire more than ordinary force in those cases in which the two fixed types inhabit regions differing in some respect so obvious and definite that we are compelled to regard each type as climatic and as specially adapted to the conditions. When for example an animal has a distinct type never met with except in Arctic or Alpine conditions, and another type proper to the plains and temperate regions, what are the characteristics of the population of intermediate latitudes or at intermediate levels? Some of the examples discussed in the last chapter may be instances of this very nature, but even if they are not, others are forthcoming which certainly are. The evidence of these cases leads to the suspicion that with further knowledge they will be found to consist of two classes, some in which the observer as he passes from the one climate to the other will find the intermediate area actually occupied by a population of intermediate character, and others in which, though we may presume the maintenance of intermediate conditions in the transitional area, there is no definite transitional population. This interrupted or discontinuous distribution seems, so far as I have means of judging, to be by far the more common of the two. I do not doubt that by sufficient search individuals representing every or almost every transitional form can be found, but it is apparently rare that _populations_ corresponding to these several grades can be seen. The question has in few if any cases been studied with precision sufficient to provide a positive answer; but I suspect that real and complete continuity, in the sense thus defined, will only be found where the character of the local populations depends _directly_ on the conditions of life, and shows an immediate response to changes in them apart from that postponed response which we suppose to be achieved by selection. Obviously the character must be one, like size for instance, capable of sensibly complete gradation.