The result means that the general population is made up of definite kinds of individuals that may have been sorted out.

That his conclusion is correct is shown by rearing a new generation from any plant or indeed from several plants of any one of these lines. Each line repeats the same modal cla.s.s. There is no further breaking up into groups. Within the line it does not matter at all whether one chooses a big bean or a little one--they will give the same result. In a word, the germ plasm in each of these lines is pure, or h.o.m.ozygous, as we say. The differences that we find between the weights (or sizes) of the individual beans are due to external conditions to which they have been subjected.

In a word, Johannsen's work shows that the frequency distribution of a pure line is due to factors that are extrinsic to the germ plasm. It does not matter then which individuals in a pure line are used to breed from, for they all carry the same germ plasm.

We can now understand more clearly how selection acting on a general population brings about results in the direction of selection.

An individual is picked out from the population in order to get a particular kind of germ plasm. Although the different cla.s.ses of individuals may overlap, so that one can not always judge an individual from its appearance, nevertheless on the whole chance favors the picking out of the kind of germ plasm sought.

In species with separate s.e.xes there is the further difficulty that two individuals must be chosen for each mating, and superficial examination of them does not insure that they belong to the same group--their germ plasm cannot be inspected. Hence selection of biparental forms is a precarious process, now going forward, now backwards, now standing still. In time, however, the process forward is almost certain to take place if the selection is from a heterogeneous population. Johannsen's work was simplified because he started with pure lines. In fact, had he not done so his work would not have been essentially different from that of any selection experiment of a pure race of animals or plants. Whether Johannsen realized the importance of the condition or not is uncertain--curiously he laid no emphasis on it in the first edition of his "Elemente der exakten Erblichkeitslehre".

It has since been pointed out by Jennings and by Pearl that a race that reproduces by self-fertilization as does this bean, automatically becomes pure in all of the factors that make up its germ plasm. Since self-fertilization is the normal process in this bean the purity of the germ plasm already existed when Johannsen began to experiment.

HOW HAS SELECTION IN DOMESTICATED ANIMALS AND PLANTS BROUGHT ABOUT ITS RESULTS?

If then selection does not bring about transgressive variation in a general population, how can selection produce anything new? If it can not produce anything new, is there any other way in which selection becomes an agent in evolution?

We can get some light on this question if we turn to what man has done with his domesticated animals and plants. Through selection, i.e., artificial selection, man has undoubtedly brought about changes as remarkable as any shown by wild animals and plants. We know, moreover, a good deal about how these changes have been wrought.

(1) By crossing different wild species or by crossing wild with races already domesticated new combinations have been made. Parts of one individual have been combined with parts of others, creating new combinations. It is possible even that characters that are entirely new may be produced by the interaction of factors brought into recombination.

(2) New characters appear from time to time in domesticated and in wild species. These, like the mutants in Drosophila, are fully equipped at the start. Since they breed true and follow Mendel's laws it is possible to combine them with characters of the wild type or with those of other mutant races.

Amongst the new mutant factors there may be some whose chief effect is on the character that the breeder is already selecting. Such a modification will be likely to attract attention. Superficially it may appear that the factor for the original character has varied, while the truth may be that another factor has appeared that has modified a character already present.

In fact, many or all Mendelian factors that affect the same organ may be said to be modifiers of each other's effects. Thus the factor for vermilion causes the eye to be one color, and the factor for eosin another color, while eosin vermilion is different from both. Eosin may be said to be a modifier of vermilion or vermilion of eosin. In general, however, it is convenient to use the term "modifier" for cases in which the factor causes a detectable change in a character already present or conspicuous.

[Ill.u.s.tration: FIG. 82. Scheme to indicate influence of the modifying factors, cream and whiting. Neither produces any effect alone but they modify other eye colors such as eosin.]

One of the most interesting, and at the same time most treacherous, kinds of modifying factors is that which produces an effect _only_ when some other factor is present. Thus Bridges has shown that there is a factor called "cream" that does not affect the red color of the eye of the wild fly, yet makes "eosin" much paler (fig. 82). Another factor "whiting" which produces no effect on red makes eosin entirely white. Since cream or whiting may be carried by red eyed flies without their presence being seen until eosin is used, the experimenter must be continually on the lookout for such factors which may lead to erroneous conclusions unless detected.

As yet breeders have not realized the important role that modifiers have played in their results, but there are indications at least that the heaping up of modifying factors has been one of the ways in which highly specialized domesticated animals have been produced. Selection has accomplished this result not by changing factors, but by picking up modifying factors. The demonstration of the presence of these factors has already been made in some cases. Their study promises to be one of the most instructive fields for further work bearing on the selection hypothesis.

In addition to these well recognized methods by which artificial selection has produced new things we come now to a question that is the very crux of the selection theory today. Our whole conception of selection turns on the answer that we give to this matter and if I appear insistent and go into some detail it is because I think that the matter is worth very careful consideration.

ARE FACTORS CHANGED THROUGH SELECTION?

As we have seen, the variation that we find from individual to individual is due in part to the environment; this can generally be demonstrated.

Other differences in an ordinary population are recognized as due to different genetic (hereditary) combinations. No one will dispute this statement. But is all the variability accounted for in these two ways? May not a factor itself fluctuate? Is it not _a priori_ probable that factors do fluctuate? Why, in a word, should we regard factors as inviolate when we see that everything else in organisms is more or less in amount? I do not know of any _a priori_ reason why a factor may not fluctuate, unless it is, as I like to think, a chemical molecule. We are, however, dealing here not with generalities but with evidence, and there are three known methods by means of which it has been shown that variability, other than environmental or recombinational, is not due to variability in a factor, nor to various "potencies" possessed by the same factors.

(1) By making the stock uniform for all of its factors--chief factors and modifiers alike. Any change in such a stock produced by selection would then be due to a change in one or more of the factors themselves.

Johannsen's experiment is an example of this sort.

[Ill.u.s.tration: FIG. 83 a. Drosophila ampelophila with truncate wings.]

(2) The second method is one that is capable of _demonstrating_ that the effects of selection are actually due to modifiers. It has been worked out in our laboratory, chiefly by Muller, and used in a particular case to demonstrate that selection produced its effect by isolating modifying factors. For example, a mutant type called truncate appeared, characterized by shorter wings, usually square at the end, (fig. 83a). The wings varied from those of normal length to wings much shorter (fig. 83b). For three years the mutant stock was bred from individuals having the shorter wings until at last a stock was obtained in which some of the individuals had wings much shorter than the body. By means of linkage experiments it was shown that at least three factors were present that modified the wings.

These were isolated by means of their linkage relations, and their mutual influence on the production of truncate wings was shown.

[Ill.u.s.tration: FIG. 83 b. Series of wings of different length shown by truncate stock of D. ampelophila.]

An experiment of this kind can only be carried out in a case where the groups of linked gens are known. At present Drosophila is the only animal (or plant) sufficiently well known to make this test possible, but this does not prove that the method is of no value. On the contrary it shows that any claim that factors can themselves be changed can have no finality until the claim can be tested out by means of the linkage test. For instance, bar eye (fig. 31) arose as a mutation. All our stock has descended from a single original mutant. But Zeleny has shown that selection within our stock will make the bar eye narrower or broader according to the direction of selection. It remains to be shown in this case how selection has produced its effects, and this can be done by utilizing the same process that was used in the case of truncate.

Another mutant stock called beaded (fig. 84), has been bred for five years and selected for wings showing more beading. In extreme cases the wings have been reduced to mere stumps (see stumpy, fig. 5), but the stock shows great variability. It is probable here as Dexter has shown, that a number of mutant factors that act as modifiers have been picked up in the course of the selection, and when it is recalled that during those five years over 125 new characters have appeared elsewhere it does not seem improbable that factors also have appeared that modify the wings of this stock.

[Ill.u.s.tration: FIG. 84. Two flies showing beaded wings.]

(3) The third method is one that has been developed princ.i.p.ally by East for plants; also by MacDowell for rabbits and flies. The method does not claim to prove that modifiers are present, but it shows why certain results are in harmony with that expectation and can not be accounted for on the basis that a factor has changed. Let me give an example. When a Belgian hare with large body was crossed to a common rabbit with a small body the hybrid was intermediate in size. When the hybrid was crossed back to the smaller type it produced rabbits of various sizes in apparently a continuous series.

MacDowell made measurements of the range of variability in the first and in the second generations.

_Cla.s.sification in relation to parents based on skull lengths and ulna lengths, to show the relative variability of two measurements and of the first generation (F_1) and the back cross (B. C.)_

CHARACTER

GENERATION

-13

-12

-11

-10

-9

-8

-7

-6

-5

-4

-3

-2

-1

---------+----------+---+---+---+---+---+---+---+---+---+---+---+---+---+ Length of{ F_1

skull { B.C.

3

Length of{ F_1

ulna { B.C.

1

1

2

3

1

2

4

4

_same table continued_

CHARACTER

GENERATION

0

1

2

3

4

5

6

7

8

9

10

11

12

---------+----------+---+---+---+---+---+---+---+---+---+---+---+---+---+ Length of{ F_1

2

2

8

5

10

7

skull { B.C.

6

4

13

18

42

32

38

34

16

16

8

4

3

Length of{ F_1

1

2

1

1

1

2

2

5

3

ulna { B.C.

12

11

20

26

17

19

18

15

12

13

15

11

5

_same table continued_

CHARACTER

GENERATION

13

14

15

16

17

18

19

20

21

22

23

24

25

---------+----------+---+---+---+---+---+---+---+---+---+---+---+---+---+ Length of{ F_1

3

2

2

skull { B.C.

1

Length of{ F_1

1

7

3

2

1

2

1

1

ulna { B.C.

2

4

2

2

1

1

He found that the variability was smaller in the first generation than in the second generation (back cross). This is what is expected if several factor-differences were involved, because the hybrids of the first generation are expected to be more uniform in factorial composition than are those in the second generation which are produced by recombination of the factors introduced through their grandparents. Excellent ill.u.s.trations of the same kinds of results have been found in Indian corn. As shown in figure 85 the length of the cob in F_1 is intermediate between the parent types while in F_2 the range is wider and both of the original types are recovered. East states that similar relations have been found for 18 characters in corn. Emerson has recently furnished further ill.u.s.trations of the same relations in the length of stalks in beans.

[Ill.u.s.tration: FIG. 85. Cross between two races of Indian corn, one with short cobs and one with long cobs. The range of variability in F_1 is less than that in F_2. (After East.)]

A similar case is shown by a cross between fantail and common pigeons (fig.

86). The latter have twelve feathers in the tail, while the selected race from which the fantails came had between 28 and 38 feathers in the tail.

The F_1 offspring (forty-one individuals) showed (fig. 87) between 12 and 20 tail feathers, while in F_2 the numbers varied between 12 and 25. Here one of the grand-parental types reappears in large numbers, while the extreme of the other grand-parental type did not reappear (in the counts obtained), although the F_2 number would probably overlap the lower limits of the race of fantail grandparents had not a selected (surviving) lot been taken for the figures given in the table.

[Ill.u.s.tration: FIG. 86. Cross of pigeon with normal tail P_1 and fantail P_1; F_1, bird below.]

[Ill.u.s.tration: FIG. 87. Cross of normal and fantail pigeons. (See Fig. 86.) The F_2 range is wider than that of F_1. The normal grand-parental type of 12 feathers was recovered in F_2 but the higher numbers characteristic of fantails were not recovered.]

The preceding account attempts to point out how I should prefer to interpret the problem of selection in the light of the most recent work on breeding. But I would give a very incomplete account of the whole situation if I neglected to include some important work which has led some of my fellow-workers to a very different conclusion.

[Ill.u.s.tration: FIG. 88. Scheme to show cla.s.ses of hooded rats used by Castle. (After Castle.)]

Castle in particular is the champion of a view based on his results with hooded rats. Starting with individuals which have a narrow black stripe down the back he selected for a narrower stripe in one direction and for a broader stripe in the other. As the diagram shows (fig. 88) Castle has succeeded in producing in one direction a race in which the dorsal stripe has disappeared and in the other direction a race in which the black has extended over the back and sides, leaving only a white mark on the belly.

Neither of these extremes occurs, he believes, in the ordinary hooded race of domesticated rats. In other words no matter how many of them came under observation the extreme types of his experiment would not be found.

Castle claims that the factor for hoodedness must be a single Mendelian unit, because if hooded rats are crossed to wild gray rats with uniform coat and their offspring are inbred there are produced in F_2 three uniform rats to one hooded rat. Castle advances the hypothesis that factors--by which he means Mendelian factors--may themselves vary in much the same way as do the characters that they stand for. He argues, in so many words, that since we judge a factor by the kind of character it produces, when the character varies the factor that stands for it may have changed.

As early as 1903 Cuenot had carried out experiments with spotted mice similar to those of Castle with rats. Cuenot found that spotted crossed to uniform coat color gave in F_2 a ratio of three uniform to one spotted, yet selection of those spotted mice with more white in their coat produced mice in successive generations that had more and more white. Conversely Cuenot showed that selection of those spotted mice that had more color in their coat produced mice with more and more color and less white. Cuenot does not however bring up in this connection the question as to how selection in these spotted mice brings about its results.

Without attempting to discuss these results at the length that they deserve let me briefly state why I think Castle's evidence fails to establish his conclusion.

In the first place one of the premises may be wrong. The three to one ratio in F_2 by no means proves that all conditions of hoodedness are due to one factor. The result shows at most that one factor that gives the hooded types is a simple Mendelian factor. The changes in this type may be caused by modifying factors that can show an effect only when hoodedness is itself present. That this is not an imaginary objection but a real one is shown by an experiment that Castle himself made which furnishes the ground for the second objection.

Second. If the factor has really changed its potency, then if a very dark individual from one end of the series is crossed to a wild rat and the second generation raised we should expect that the hooded F_2 rats would all be dark like their dark grandparent. When Castle made this test he found that there were many grades of hooded rats in the F_2 progeny. They were darker, it is true, as a group than were the original hooded group at the beginning of the selection experiment, but they gave many intermediate grades. Castle attempts to explain this by the a.s.sumption that the factor made pure by selection became contaminated by its normal allelomorph in the F_1 parent, but not only does this a.s.sumption appear to beg the whole question, but it is in flat contradiction with what we have observed in hundreds of Mendelian cases where no evidence for such a contamination exists.