Or, again, suppose that a breeder has a chestnut mare and wishes to make certain of a bay foal from her. We know that bay is dominant to chestnut, and that if a homozygous bay stallion is used a bay foal must result. In his choice of a sire, therefore, the breeder must be guided by the previous record of the animal, and select one that has never given anything but bays when put to either bay or chestnut mares. In this way he will assure himself of a bay foal from his chestnut mare, whereas if the record of the sire shows that he has given chestnuts he will be heterozygous, and the chances of his getting a bay or a chestnut out of a chestnut mare are equal. It is not impossible that the breeder may be unwilling to test his animals by crossing them with a different breed through fear that their purity may be thereby impaired, and that the influence of the previous cross may show itself in succeeding generations. He might hesitate, for instance, to test his polled cows by crossing them with a horned bull for fear of getting horned calves when the cows were afterwards put to a polled bull of their own breed. The belief in the power of a sire to influence subsequent generations, or telegony as it is sometimes called, is not uncommon even to-day. Nevertheless, carefully conducted experiments by more than one competent observer have failed to elicit a single shred of unequivocal evidence in favour of the view. Until we have evidence based upon experiments which are capable of {168} repetition, we may safely ignore telegony as a factor in heredity. Heterozygous forms play a greater part in the breeding of animals than of plants, for many of the qualities sought after by the breeder are of this nature. Such is the blue of the Andalusian fowl, and, according to Professor Wilson, the roan of the Shorthorn is similar, being the heterozygous form produced by mating red with white. The characters of certain breeds of canaries and pigeons again appear to depend upon their heterozygous nature. Such forms cannot, of course, ever be bred true, and where several factors are concerned they may when bred together produce but a small proportion of offspring like themselves. As soon, however, as their constitution has been analysed and expressed in terms of Mendelian factors, pure strains can be built up which when crossed will give nothing but offspring of the desired heterozygous form.