facts by an observer thoroughly conversant with the particular plant or animal, its habits and properties, checked by the test of crucial experiment, can disentangle the truth.
To prove the reality of selection as a factor in evolution is, as I have said, a work of supererogation. With more profit may experiments be employed in defining the limits of what selection can accomplish. For whenever we can advance no further by selection, we strike that hard outline fixed by the natural properties of organisms. We come upon these limits in various unexpected places, and to the naturalist ignorant of breeding nothing can be more surprising or instructive.
Whatever be the mode of origin of new types, no theoretical evolutionist doubts that selection will enable him to fix his character when obtained. Let him put his faith into practise. Let him set about breeding canaries to win in the class for Clear Yellow Norwich at the Crystal Palace Show. Being a selectionist, his plan will be to pick up winning yellow cocks and hens at shows and breed them together. The results will be disappointing. Not getting what he wants, he may buy still better clear yellows and work them in, and so on till his funds are exhausted, but he will pretty certainly breed no winner, be he never so skilful. For no selection of winning yellows will make them into a breed. They must be formed afresh by various combinations of colors appropriately crossed and worked up. Though breeders differ as to the system of combinations to be followed, all would agree that selection of birds representing the winning type was a sure way to fail. The same is true for nearly all canary colors except in lizards, and, I believe, for some pigeon and poultry colors also.
Let this scientific fancier now go to the Palace Poultry Show and buy the winning brown leghorn cock and hen, breed from them, and send up the result of such a mating year after year. His chance of a winner is not quite, but almost nil. For in its wisdom the fancy has chosen one type for the cock and another for the hen. They belong to distinct strains. The hen corresponding to the winning cock is too bright, and the cock corresponding to the winning hen is too dull for the judge's taste. The same is the case in nearly every breed where the sex-colors differ markedly. Rarely winners of both sexes have come in one strain—a phenomenon I can not now discuss—but the contrary is the rule. Does any one suppose that this system of 'double mating' would be followed, with all the cost and trouble it involves, if selection could compress the two strains into one? Yet current theory makes demands on selection to which this is nothing.
The tyro has confidence in the power of selection to fix type, but he never stops to consider what fixation precisely means. Yet a simple experiment will tell him. He may go to a great show and claim the