In order that plants and animals may be improved and modified to any considerable extent, it is of course essential that suitable variations should occur with tolerable frequency. There seem to be three conditions which are especially favourable to the production of variations: (1) That the particular species or variety should be kept in very large numbers; (2) that it should be spread over a wide area and thus subjected to a considerable diversity of physical conditions; and (3) that it should be occasionally crossed with some distinct but closely allied race. The first of these conditions is perhaps the most important, the chance of variations of any particular kind being increased in proportion to the quantity of the original stock and of its annual offspring. It has been remarked that only those breeders who keep large flocks can effect much improvement; and it is for the same reason that pigeons and fowls, which can be so easily and rapidly increased, and which have been kept in such large numbers by so great a number of persons, have produced such strange and numerous varieties. In like manner, nurserymen who grow fruit and flowers in large quantities have a great advantage over private amateurs in the production of new varieties. Although I believe, for reasons which will be given further on, that some amount of variability is a constant and necessary property of all organisms, yet there appears to be good evidence to show that changed conditions of life tend to increase it, both by a direct action on the organisation and by indirectly affecting the reproductive system. Hence the extension of civilisation, by favouring domestication under altered conditions, facilitates the process of modification. Yet this change does not seem to be an essential condition, for nowhere has the production of extreme varieties of plants and flowers been carried farther than in Japan, where careful selection continued for many generations must have been the chief factor. The effect of occasional crosses often results in a great amount of variation, but it also leads to instability of character, and is therefore very little employed in the production of fixed and well-marked races. For this purpose, in fact, it has to be carefully avoided, as it is only by isolation and pure breeding that any specially desired qualities can be increased by selection. It is for this reason that among savage peoples, whose animals run half wild, little improvement takes place; and the difficulty of isolation also explains why distinct and pure breeds of cats are so rarely met with. The wide distribution of useful animals and plants from a very remote epoch has, no doubt, been a powerful cause of modification, because the particular breed first introduced into each country has often been kept pure for many years, and has also been subjected to slight differences of conditions. It will also usually have been selected for a somewhat different purpose in each locality, and