Journal of Heredity 5: 93-97 (1914)
Practical Idea of Stock Breeder and Theoretical Ideal of Biologist Differ—Evidence Advanced in Support of "Pure Lines" Inadequate—Variation in Asexual Reproduction—Experience With Hooded Rats Shows Selection Possible in Either Direction.
Professor of Zoology in Harvard University and Research Associate of the Carnegie Institution of Washington.

CURRENT discussions among students of heredity about pure lines of animals and plants are likely to be very perplexing to those breeders of live-stock who pride themselves upon the excellence and purity to breed-standards of the animals which they keep. "Selection within the pure line," we are told by the biologist, "is without effect." But selection is the foundation principle of the stock breeder and a statement addressed to him that selection is useless in the breeding of any sort of domesticated animal, however pure, would be received with incredulity.

It may be worth while therefore to point out that the biologist, as quoted above, and the stock breeder are talking about very different things when they speak of "pure lines." The stock breeder has in mind a race of animals bred as closely as possible to a recognized standard. But he realizes that some individuals conform more closely than others to the standard of the breed, and his experience indicates that continual selection of the "some" rather than the "others" is essential to maintain the purity of the breed. He understands the biologist's dictum that "selection within the pure line is without effect" to mean that one animal is as good as another to breed from provided both belong to the same pure race, and his comment is "Nonsense", and he is quite right too.

But that is not the biologist's meaning. The biologist's "pure line" is an imaginary thing. I doubt very much whether it was ever realized in any actual race of animals or plants. It has no more relation to actual animals and plants than a mathematical circle has to the circles described by the most accomplished draftsman. All the circles of the draftsman have wiggles in them, if you look at them carefully enough; only the mathematician's imaginary circle is perfect. Now the biologist undertakes to be the mathematician of breeding and to construct an "exact" system of heredity in which the "pure line" concept holds a conspicuous place. He reasons thus: Individuals of the same parentage differ for two reasons, (1) because of inherited differences and (2) because of differences in environment. If differences of environment (soil and situation of plants, care and food of animals) could be altogether eliminated, then animals of the same heredity should be identical, and should produce only offspring like themselves. They would constitute a pure line.


The biologist reasons further: Although it is impossible to control the environment completely and thus make it uniform and so eliminate its effects upon variation, nevertheless the effects of environment are not inherited. Consequently, if two germ-cells could be brought together in fertilization to form an individual, each of which germ-cells was identical with the other in its heredity, a pure individual would result, so far as heredity is concerned, all of whose germ-cells would transmit the same inherited characteristics, and all would look alike except as they were made to vary by the environment. But as the effects of environment are by hypothesis not inherited, all would breed alike regardless of their appearance, hence it would be useless to make any choice among them.

The pure line theory, as outlined above, like the theorems of the geometer, rests upon the validity of certain postulates, any one of which being disproved, the whole argument fails.

The pure line postulates are: (1) The effects of environment are not inherited: (2) inherited characteristics do not vary. The first of these postulates is a bone of unending contention among biologists, and the last word has not been said upon it yet. Practical breeders believe that good care and feeding are not wasted upon their stock, that not only the individuals so treated are benefited but subsequent generations also. It must be admitted, however, that the evidence is not satisfactory that subsequent generations are given a better endowment by better treatment of the present generation. What is inherited is capacity to utilize opportunities of the environment. It is doubtful whether utilization of the environment by one generation increases the capacity of the next generation so to utilize it, though we all optimistically hope for this outcome. Plant breeders have shown that plump, well-nourished seed produces more vigorous plants and larger harvests than shrunken ill-nourished seed, a seeming inheritance of environmental effects. But in reality the seed is only part embryo; part of it is food stuff, environment of the previous generation carried over bodily to form the early environment of the new generation. So that when we select well nourished seed we select, not only a bundle of inherited plant qualities, but also a good initial environment for the plant. On the whole the first postulate of the pure line theory stands, if not proved, at least not disproved.

The second postulate of the pure line theory is a much shakier one, but I must hasten to qualify the form in which it has been stated before it is disowned by all supporters of the theory. The Hagedoorns, who have championed the theory in The American Breeders' Magazine (Vol. 4, No. 3) [1913], are careful to say that inherited "factors, not characters", are constant. What they mean, I take it, is this. An animal has some visible character, such as black fur. The black color of this fur may, however, result from several independent physiological processes, or agencies, no one of which by itself produces a visible effect. These agencies. known or unknown, are factors in producing a black coat. Some of them may be inherited, others environmental. Although the blackness of the coat may vary (owing to variation in environmental factors) its ultimate inherited factors do not vary.


1) The Macmillan Co., N. Y., 1911.

What evidence is there for the idea that the ultimate factors of inheritance do not vary? First let us inquire how the idea originated. Ten years ago, when the rediscovered Mendel's law was new, it was supposed that characters which conformed with Mendel's law did not vary. They behaved as units in heredity; how could they vary; how could units vary? The gametes were "pure". Each one either contained or did not contain a particular Mendelian unit. Units could not be split or modified. This idea found its fullest development in Punnett's Mendelism.1 Later investigations have shown beyond question that Mendelian characters do vary. Practically everyone has now abandoned the idea of "gametic purity", but the idea of purity has been shifted from the characters which can be seen to vary, to factors which may be imagined to be invariable, though they can not be seen.

2) Vol. 46, No. 551, Nov. 1912.

We might at this point with propriety end the discussion concluding that factors which are imaginary, like the circles of the mathematician, are of necessity perfect and constant. This is practically the burden of East's argument in The American Naturalist2. But as the world has benefited by the speculations of the geometers, so, let us hope, it may profit by the speculations of the biologist, especially if some means can be devised to test their validity by experiment.

Since the supposed "factors" of inheritance are invisible, we can not hope to deal with them directly by experiment, but only indirectly. Our method obviously should be to eliminate all environmental factors so far as possible and also all factors of inheritance except one. If then this one being present gives a uniform result and being absent a result also uniform but different, we can conclude the factor constant. But it is very difficult to apply this method to specific cases, since when variation is observed it is always possible to suppose that all factors but one have not yet been eliminated.

3) Elemente der exakten Erblichkeitslehre, Jena, 1909. Am. Naturalist, Vol. 43, No. 510, June 1909.

From existing experimental work chat evidence have we for the idea of factorial constancy? Perhaps Johannen3 has contributed more than any other one person toward popularizing the idea of factorial constancy. Selecting size variations in bean seeds, he was sometimes successful in modifying the racial mean, sometimes unsuccessful. Whenever he was successful, he attributes the success to variation in genetic factors; whenever unsuccessful he assumes that no variation in genetic factors occurred, i.e.. that a pure line had been established, which could not subsequently vary unless hybridization occurred out of the race or a mutation occurred within it. It would be of much interest to know whether Johannsen's pure lines were as pure for all other characters as for size of seed and equally unresponsive to selection in all particulars of leaf and stem. For the bearing of his observations on seed size upon the question of the constancy of Mendelizing characters is not very obvious, since seed size does not Mendelize.


Although extensive observations upon the subject of size inheritance in both animals and plants have been made, they have resulted in the demonstration as yet of no single clear-cut Mendelizing unit character (or factor either). Dwarf plants, known to Mendelize when crossed with tall ones, form only an apparent exception; in reality they differ in habit rather than in size from tall ones.

The results of all observers, as regards the inheritance of ordinary differences in size, are closely in accord. When two races differing in size are crossed, the immediate offspring are intermediate in size. The next generation of offspring is likewise intermediate but more variable as a rule, and it has been found possible in some cases to select from them forms as extreme in size as the original parents. To interpret such cases as Mendelian requires the assumption that no single unit or factor is concerned in the size difference, but many wholly independent units. For a single Mendelizing unit would produce a wholly different result. But suppose we allow the assumption that many independent Mendelizing units or factors are concerned in the inheritance of size. The pure line hypothesis is not benefited by this assumption, unless we suppose further that these hypothetical factors do not vary. But this is an assumption wholly without warrant. For in all cases studied critically with reference to the constancy of characters demonstrably Mendelian, the characters have been found to be inconstant and subject to modification by selection. What ground is there, then, for supposing that in a case where no factors are demonstrable, such factors are invariable? This is like supposing that the moon is made of cheese and that further this cheese is green. The speculation is harmless, if one chooses to amuse himself that way, but it can scarcely be called a valid scientific conclusion.

4) Am. Naturalist, Vol. 43, No. 510, June 1909.

Aside from the size selection experiments of Johannsen, which were made on a self-fertilizing plant, the experiments of Jennings4 are most often cited in support of the view that selection within the pure line is without effect. In some respects the material used by Jennings was even more favorable than that of Johannsen. It consisted of a species of small animal (paramecium) which ordinarily multiplies by dividing into two parts, each of which becomes an entire individual. Only at rare intervals does it reproduce by a method comparable with sexual reproduction in the higher animals and plants. If variation in genetic factors arises only from sexual reproduction, then asexually produced lines of paramecium should be devoid of genetic variation and so unaffected by selection. Jennings believed this to be true. He regarded the asexually produced descendants of a single paramecium as constituting a pure line, which selection could not change, and recognized several diffferent pure lines isolated from his mixed cultures, each varying about a different average of size. Within each pure line he found selection ineffective.


5) Journal of Experimental Zoology, Vol. 15, No. 4, Nov. 1913.

But considerable doubt is raised concerning the soundness of these conclusions by the subsequent work of Calkins and Gregory5 who find that from a single "pure line" as defined by Jennings subordinate lines may be isolated which differ in mean size from each other more than the pure lines which Jennings supposed to be genetically different. If this is so, it must he possible for genetic differences to arise in the course of asexual reproduction, which on the pure line hypothesis is impossible; and if such genetic differences do arise, they afford material for selection to act upon with effectiveness, even within the supposed pure line.

Further evidence that racial differences may originate within an asexually produced and so genetically "pure line" is afforded by horticultural varieties of plants propagated asexually. Examples are found among apples and potatoes. An orchard variety which has endured for any considerable length of time is apt to develop sub-varieties. The same is true of varieties of potato which have enjoyed wide and long continued popularity. The Williams apple, widely cultivated in the eastern United States, occurs in two distinct strains, one conspicuously striped, the other nearly solid red. The Baldwin apple which has been in cultivation for over a century has developed sub-varieties differing in shape. The Early Rose potato, formerly the favorite early variety in the United States, developed sub-varieties, some of which were more elongated in shape than others.

So common are changes of this sort within asexually propagated varieties that botanists have given them a special name, that of bud variations. East, from his studies of bud variation among potatoes, concluded them to be ordinarily loss variations, or retrogressive variations due to the loss of some normal racial character, as are most of the germinal variations which arise as sports in the course of sexual reproduction among domesticated animals and cultivated plants. If so, we are led to conclude not only that variations may occur within a "pure line" asexually produced, but also that these variations are of the same nature as those which arise in the course of sexual reproduction. There can be no question of the effectiveness of selection in either category of cases.

The Hagedoorns think that evidence in favor of the pure line hypothesis is afforded by Vilmorin's varieties of wheat bred each from a single plant. They figure four heads of wheat selected each as typical of those produced by the ancestral plant 50 years ago, from which a pure variety has descended. Beside these they figure a typical head of each of the derived varieties at the present time. A strong resemblance exists in each case between the head of 50 years ago and the present day representative of the same variety, though this of course does not amount to identity. We are, however, told that "As can easily be seen from the photograph, all these generations of selection have not changed any one of the varieties one little bit."


The photographs afford a rather small basis for so sweeping a conclusion. We are not shown the entire plant of 50 years ago beside that of today; we know nothing about the appearance of its leaves, stem, or root; we know nothing of its physiological behavior under controlled environment; yet we are asked to believe the two plants, of 50 years ago and of today, identical in genetic character, because a selected head of one can be roughly matched with a selected head of the other. Does this prove that selection is without effect? Rather it proves that wheat can be selected to a type. What evidence have we that any attempt has been made in these 50 years to change the type? Not once, I venture, has the shortest headed plant been chosen in the long headed variety, or the least branched head in the branched headed variety. Yet on the pure line hypothesis, this would have made no difference in the result. How do we know that it would have made no difference, the experiment not having been made? Doubtless the pure line advocate would reply, because selection in the opposite direction has effected no change, the long headed race being now no longer and the branched headed race no branchier. But this assertion is not established by the photographs of matched heads, and even granting its correctness, it might well be that a physiological limitation had been reached 50 years ago which it has not been since possible to remove. When selection has attained the ultimate, it can go no farther. If I select animals spotted with white for increasing blackness and obtain an animal black all over, the task is accomplished. Nothing more can be done in that direction. This does not prove selection to increase or to decrease the white spots of animals to be ineffectual; indeed, the contrary has been fully established.

6) A preliminary account of certain of the experiments was given in Am. Breeders' Magazine, Vol. 3, No. 4. [1912]

The Hagedoorns think further to strengthen the argument for the pure line theory by a challenge to its opponents to modify a character within a narrowly inbred line of animals by selection and then to bring the character back again to its original condition. "For," they say, "it is obvious that it may be possible by selection to get rid of such genetic factors as influence the character selected in a direction opposite to that in which we want to modify it, and thus to make great progress. On the contrary, if selection really might change genetic factors, modification of any character would be as easy in one direction as in the other." This test (in all essential features) has been applied repeatedly in a series of selection experiments with the hooded pattern of spotted rats (a Mendelizing unit-character) described in Publication No. 195 of the Carnegie Institution of Washington.6 The experiments show that plus selection and minus selection are both effective, and that selection away from and back to any modal condition of the pattern are equally easy of accomplishment.

If I were disposed to use challenges for argument, I would make a counter challenge in these words—Let him who thinks selection ineffectual try it on any character of any animal. If he is persistent and critical in his experiments, I am confident that he will be convinced of the effectiveness of selection, as all practical breeders are, and will thenceforth regard "pure lines" as purely imaginary.