When Laxton speaks of selection and the need for it, he means, what the raiser of new varieties almost always means, the selection of _definite_ forms, not impalpable fluctuations. When he says that without selection there will be utter confusion, he means--to use Mendelian terms--that the plant which shows the desired combination of characters must be chosen and bred from, and that if this be not done the grower will have endless combinations mixed together in his stock. If however such a selection be made in the fourth or fifth generation the breeder may very possibly have got a fixed form--namely, one that will breed true[146]. On the other hand he may light on one that does not breed true, and in the latter case it may be that the particular type he has chosen is not represented in the gametes and will _never_ breed true, though selected to the end of time. Of all this Mendel has given us the simple and final account. [146] Apart from fresh original variations, and perhaps in some cases imperfect homozygosis of some hypallelomorphs. At Messrs Sutton and Sons, to whom I am most grateful for unlimited opportunities of study, I have seen exactly such a case as this. For many years Messrs Sutton have been engaged in developing new strains of the Chinese Primrose (_Primula sinensis_, hort.). Some thirty thoroughly distinct and striking varieties (not counting the _Stellata_ or “Star” section) have already been produced which breed true or very nearly so. In 1899 Messrs Sutton called my attention to a strain known as “Giant Lavender,” a particularly fine form with pale magenta or lavender flowers, telling me that it had never become fixed. On examination it appeared that self-fertilised seed saved from this variety gave some magenta-reds, some lavenders, and some which are white on opening but tinge with very faint pink as the flower matures. On counting these three forms in two successive years the following figures appeared. Two separately bred batches raised from “Giant Lavender” were counted in each year. Magenta Lavender White red faintly tinged 1901 1st batch 19 27 14 " 2nd " 9 20 9 1902 1st " 12 23 11 " 2nd " 14 26 11 -- -- -- 54 96 45 The numbers 54 : 96 : 45 approach the ratio 1 : 2 : 1 so nearly that there can be no doubt we have here a simple case of Mendelian laws, operating without definite dominance, but rather with blending. When Laxton speaks of the “remarkably fine but unfixable pea _Evolution_” we now know for the first time exactly what the phenomenon meant. It, like the “Giant Lavender,” was a “mule” form, not represented by germ-cells, and in each year arose by “self-crossing.”