2. Causes of social poverty: Isolation; bad roads; absence of large and specialized buildings; individualistic philosophy; depreciation of play and recreation; lack of reading-habit; the work-habit. 3. Means of socialization: Good roads; automobiles; telephones and rural delivery; schools and churches built for social purposes; farmers' organizations such as institute, grange, American Society of Equity, farmers' unions, etc.; mothers' clubs and literary clubs among women; athletic meets and tournaments at school grounds; literary and debating clubs; spelling-matches; lectures and entertainments; moving-picture shows; banquets, feasts, and "socials." References. On social isolation: Butterfield, Chapters in Rural Progress, pp. 17-22; Bailey, Insufficiencies in Country Life, "The Training of the Farmer,' pp. 15-19; Butterfield, "The Country Church and Progress," chap. xii; School Buildings for Social Purposes, U.S. Department of Agriculture, Bulletin No. 232, on consolidated schools; Foght, "Libraries for Rural Communities," chap. xiii, and Kern, chap. vi; Johnson, Education by Play and Games, "Organized Play and Recreation"; Butterfield, "Farmers' Institutes," chap. vii, and Kern, chap. ix; Butterfield, "The Grange," chap. x, "Opportunities for Farm Women," chap. xi; Fiske, chap. v; Gillette, chaps. xiii-xviii; McKeever, chaps. vi-x.

THE LARGER SOCIAL WORLD

The larger side of social life, that which reaches beyond the local community into the nation and world, may be developed by means of a brief discussion once or twice a week of the events which are transpiring in the world at large. This should be done in a manner that would made each event treated mean something for life by showing how it changes conditions and thus makes for improvement or deterioration.

A brief treatment and discussion of certain phases of our industrial history would also be useful to cultivate the idea of the articulation of ourselves with the world and to give an understanding of some of the pressing economic issues. The little weekly paper entitled Current Events, published in Springfield, Massachusetts, is recommended as exceedingly useful to accomplish the former purpose. Thurston's Economic and Industrial History would give the material for the second, and an account of its size is quite usable. Coman's Industrial History of the United States, or Bogart's Economic History of the United States are fuller and more pretentious. It would be sufficient to select only the more recent problems of labor and industry.

VARIABILITY AS RELATED TO SEX DIFFERENCES IN

ACHIEVEMENT

A CRITIQUE

LETA STETTER HOLLINGWORTH
New York City

This paper is the outcome of prolonged reflection on the doctrine of greater male variability. It comprises an attempt to assemble and review briefly data at present accessible as to the comparative variability of the sexes in mental traits, and to discuss critically the hypothesis that the great difference between the sexes in intellectual achievement and eminence is due to the inherently greater variability of the males. This hypothesis is stated clearly and concisely by Thorndike' thus:

The trivial difference between the central tendency of men and that of women which is the common finding of psychological tests and school experience may seem at variance with the patent fact that in the great achievements of the world in science, art, invention, and management, women have been far excelled by men. One who accepts the equality of typical (i.e., modal) representatives of the two sexes, must assume the burden of explaining this great difference in the high ranges of achievement.

The probably true explanation is to be sought in the greater variability within the male sex.

In particular, if men differ in intelligence and energy by wider extremes than do women, eminence in and leadership of the world's affairs of whatever sort will inevitably belong oftener to men. They will oftener deserve it.

It is at once evident how important are the implications here stated for those who hope much from the present tendency to remove all disabilities of law, custom, and prejudice from women. If the explanation of women's failure to achieve significant things in the fields named by Thorndike is really to be found in the inherently greater variability of males, then complete liberation of women from excessive maternity and from all the consequent customs and legal disabilities that have developed, will result E. L. Thorndike, Educational Psychology (1910), p. 35.

only in raising the average intelligence and happiness of the race. We shall not expect any increase from this source in the number of eminent individuals, nor in achievement of that high order which forces knowledge and wisdom farther.

Thorndike' states the implications for pedagogy thus:

....

This one fundamental difference in variability is more important than all the differences between the average male and female capacities. . . . a slight excess of male variability would mean that of the hundred most gifted individuals in this country not two would be women, and of the thousand most gifted, not one in twenty. . . . Women may and doubtless will be scientists and engineers, but the Joseph Henry, the Rowland, and the Edison of the future will be men; even should all women vote, they would play a small part in the Senate. . . . . Not only the probability and the desirability of marriage and the training of children as an essential feature of woman's career, but also the restriction of women to the mediocre grades of ability and achievement should be reckoned with by our educational systems. The education of women for such professions as administration, statesmanship, philosophy, or scientific research, where a very few gifted individuals are what society requires, is far less needed than education for such professions as nursing, teaching, medicine, or architecture, where the average level is the essential. . . . . Postgraduate instruction, to which women are flocking in large numbers is, at least in its higher reaches, a far more remunerative investment in the case of men.'

The first discussion of the comparative variability of the sexes bore on anatomical traits, and began about a century ago. The anatomist Meckel3 concluded on pathological grounds that the human female showed greater variability than the human male, "and he thought that since man is the superior animal and variation a sign of inferiority, the conclusion was justified." Later, when anatomists and naturalists arrived at the conclusion that the male is more variable, variability came to be regarded as an advantage, a characteristic affording the greatest hope for progress, and finally as the probable explanation of the fact that all the world's greatest deeds of intellect have been the deeds of men. This latter view obtains at present among men of science, though not without exceptions, the most notable of whom is Karl Pearson.4 E. L. Thorndike, "Sex in Education," The Bookman, XXIII, 213.

2 The italics here are mine.

3 Meckel, Manual of Descriptive and Pathological Anatomy (see Ellis, Man and Woman [1909], p. 410).

4 Karl Pearson, Chances of Death (1897).

It will be well at this point to consider not only the social and biological significance of variability, but also the connotation of the term itself, and whether every author who discusses variability means the same thing. There is, in fact, complaint among authors that the term is indefinite. Even in their controversial matter,' Ellis and Pearson complain of each other that there is failure to define the word. Theoretically greater variability always implies greater range, if the trait distributed conforms to the Gauss curve of probability. Empirical data, however, are not yet forthcoming to demonstrate that mental traits conform to the theoretical curve; and there is at present no conclusive empirical evidence to show that in cases where the coefficient of variation is greater for one sex than for the other, this greater variability consists in greater range. If we neglect theory and confine ourselves to facts as demonstrated, greater variability is found to consist in any or all of three typical conditions:

1. Greater range (Series B as compared with Series A).

2. Equal range for both groups, but greater frequency at the extremes for one group (Series C as compared with Series A).

3. Smaller range for the more variable group, with slight flattening at the top of the curve of distribution (Series D as compared with Series A).

A fourth condition is found in the work of Bonser, where the males are seen to be more variable than the females, though the range for the sexes is equal, and the frequency at both extremes is nearly twice as great for the females. This case will be taken up later in connection with other results from Bonser.

Let us now consider a hypothetical case. Table I gives four possible distributions of the same trait, including the same number of cases. This trait may be, for example, ability to perform an amount of work in a specified time, this ability being indicated by units varying from 1 to 15. Let Series A be a group of 1,000 women, and let Series B, C, and D be groups of 1,000 men each. It is seen that these Series all show greater variability on the part of the males (reference to Table I will show just how much greater is the A.D. in each case), but the social implications differ widely. 'H. Ellis, Man and Woman (Appendix).

In Series B the greater variability of the males consists in greater range. It is on this Series that we might base the explanation of the fact that all the world's greatest deeds of ntellect have been the deeds of men; for here no women equal the best men.

In Series C the greater variability of the males consists in greater frequency at the extremes, the range being equal. On this Series might be based an explanation of the fact that more men than women have reached the same degree of eminence. It would not explain why no women have reached the greatest eminence.

In Series D the greater variability of the males consists in a flattening at the top of the curve of distribution, the range for the men being actually less than for the women.

Now it is clear that if social significance is to be attached to greater variability, not only the coefficient of variation must be stated, but also what form the distribution takes. Obviously a greater male variability like that shown in Series D would have no validity at all in explaining why the greatest deeds of intellect have been the deeds of men. If greater male variability takes this form, all the greatest deeds will be those of women.

In his discussions of greater male variability and its implications for pedagogy, Thorndike' theoretically means greater range: "Though the central tendencies were the same, there would still be two men of the hundred who were better than the best woman and two men who were worse than the worst woman." This condition would be represented under Series B. But, in discussing certain statistics regarding third-year high-school classes see Table I, p. 514.

This condition would be that of Series C. The range for the sexes is equal, but the frequency at the extremes is greater for males. Such cases of greater variability do not suggest an explanation of the fact that no women have achieved the greatest intellectual eminence. They would only explain the condition in which twice as many men as women achieved the same intellectual eminence. But our chief problem is to explain why no women have equaled the best men.

Havelock Ellis, in a chapter on "The Variational Tendency E. L. Thorndike, op. cit., p. 42. 2 H. Ellis, op. cit., p. 412.