And dogs may be instanced which have got good stock from all sorts of mares and bitches. Yet in opposition to this may be instanced the numbers which have had great opportunities for snowing their good qualities, but while they have succeeded with one or two they have failed with the larger proportion of their harems. So with mares and bitches, some have produced, every year of their breeding lives, one or more splendid examples of their respective kinds, altogether independent of the horse or dog which may be the other parent, so long as he is of the proper strain. It is usually supposed that the sire impresses his external formation upon his stock, while the bitch's nervous temperament is handed down; and very probably there is some truth in the hypothesis. Yet it is clearer that not only do the sire and dam, but also the grandsires and grand-dams affect the progeny on both sided, and still further than this up to the sixth and perhaps even the seventh generations, but more especially on the dam's side, through the granddam, great granddam, etc.

There is a remarkable fact connected with breeding which should be generally known, viz., that there is a tendency in the produce to a separation between the different strains of which it is composed; so that a puppy composed in four equal proportions of breeds represented by A, b, c, and D, will not represent all in equal proportions, but will resemble one much more than the others. And this is still more clear in relation to the next step backwards, when there are eight progenitors; and the litter which, for argument's sake, we will suppose to be eight in number, may consist of animals each "going back" to one or other of the above eight This accounts for the fact that a smooth terrier bitch put to a smooth terrier will often "throw" one or more rough puppies, though the breed may be traced as purely smooth for two or three generations, beyond which, however, there must have been a cross of the rough dog. In the same way color and particular marks will be changed or obliterated for one, two, or even three generations, and will then reappear.

In most breeds of the dog this is not easily proved, because a record of the various crosses is not kept With any great care; but in the greyhound the breed, with the colors, etc, for twenty generations, is often known, and then the evidence of the truth of these facts is patent to all. Among these dogs there is a well-known strain descended from a greyhound with a peculiar nose, known as the "Parrot-nosed bitch." About the year 1825 she was put to a celebrated dog called " Streamer," and bred a bitch called "Ruby," none of the litter showing thi3 peculiar nose; nor did "Ruby" herself breed any in her first two litters; but in her third, by a dog called "Blackbird," belonging to Mr. Hodgkinson, two puppies showed the nose ("Blackbird" and "Starling"). In the same litter was a most celebrated bitch, known as " Old Linnet" from which are descended a great number of first-rate greyhounds. In these, however, this peculiarity has never appeared, with two exceptions, namely, once in the third generation, and once in the fifth, in a dog called "Lollypop," bred by Mr. Thomas, of Macclesfield, the possessor of the whole strain.

One of the bitches of this breed is also remarkable for having always one blue puppy in each litter, though the color is otherwise absent, never having been seen since the time of the above mentioned "Ruby," who was a blue bitch. These facts are very remarkable as showing the tendency to "throw back" for generations, but, as they are well known and fully recognized by all breeders, it is unnecessary to dilate upon them, and the above instances are only introduced as absolutely proving to the uninitiated what would otherwise depend upon dogmatic assertion.