Popular Science Monthly/Volume 63/May 1903/The Decrease in the Size of American Families
|THE DECREASE IN THE SIZE OF AMERICAN FAMILIES.|
By Professor EDWARD L. THORNDIKE,
TEACHERS COLLEGE, COLUMBIA UNIVERSITY, NEW YORK.
THE vital statistics of three other eastern colleges show the failure of Harvard graduates to produce their share of the present generation to be but a single example of a widespread condition. They further prove that the common discussions of the theoretical and practical questions which this failure suggests are superficial and misleading. In reality its explanation leads us directly to the fundamental problem of evolution. The facts are best seen in tabular form.
Size of Families of American College Graduates. 
The first number in each column gives the average number of children; the number in parenthesis gives the number of cases on which the average id based.
|Middlebury.||Wesleyan.||New York Univ.||Harvard.|
|1830-1839||3.9 (189)||4.5 (110)||(35-44) 4.0 (110)|
|1840-1849||3.4 (83)||4.5 (110)||(45-54) 3.2 (83)|
|1850-1859||2.9 (90)||3.2 (227)||2.9(90)|
|1860-1869||2.8 (114)||2.6 (250)||2.5(66)|
|1875-1879||1.8 (32)||1.99 (1872 inclusive) (634)|
|Total||946||807||349||634.In all 2,736|
These figures are from a sufficient number of cases to be substantially reliable. For instance, there is not one chance in a thousand that the Harvard average is 10 per cent. too low. The existence and approximate amount of the decrease in the size of family is thus certain. Its substantial identity in Middlebury, a country college in Vermont with a local attendance, in New York University, a city college, and in Wesleyan University, a strongly sectarian college with an attendance drawn from the northeastern states, makes it probable that it has prevailed throughout the college population of the north Atlantic states. It must depend upon some fundamental cause.
City life and advanced age at marriage are out of question. The former cause would work to a far greater extent upon New York University or Harvard graduates than upon Middlebury graduates, all of whom come from and most of whom go back to life in small towns. Yet in the statistics there is little difference. An increase in the age at marriage can not have been the cause for the simple reason that such increase, as I have elsewhere shown, amounts only to a very few months. An increase in the age at marriage of the wives of our group of men would be a more efficient cause. I know of no available statistics to decide the question, but it would seem extremely unlikely that the age of wives should have increased much when the age of husbands has increased so little.
The most plausible explanation attributes the change to the custom of conscious restriction of offspring. Greater prudence, higher ideals of education for children, more interest in the health of women, interests of women in affairs outside the home, the increased knowledge of certain fields of physiology and medicine, a decline in the religious sense of the impiety of interference with things in general, the longing for freedom from household cares—any or all of these may be assigned as the motive for the restriction. The only other explanation which to the present writer seems adequate assigns the decreased productivity of college men to real physiological infertility of the social and perhaps of the racial group to which college men and their wives belong.
It is possible to do more than speculate about the relative shares of unwillingness and incapacity. The figures themselves tell a plain story to the student who examines them in the light of recent knowledge of the variability of physical traits.
If we tabulate the records by decades so as to show the percentages that families of 2, 3, 4, etc., children were of the total number of families, we can see just how the decrease in the averages has been brought about. Suppose for instance that we had in 1803-1814 and in 1865-1874 the following percentages:
It would be clear that the change was due to the substitution of families of 0, 1, 2 and to a slight extent of 3, for fifty per cent. of the families over 3, that all these groups of larger families had given up the same proportion to swell the groups of small families. This would point clearly toward restriction as a cause.
Suppose that the following were the facts:
In this case it is clear that the change was due to the substitution throughout of families less in each case by three children. There is no cutting off equally from all the higher groups. Families of 4 and 5 for instance increase in number. There is no special increase of the 0, 1, 2 families. The movement has been simply a general decrease in size, a moving backward of the general tendency to produce. Such an appearance in the statistics would point toward decreased reproductive capacity.
In our second illustration there would probably be in connection with the lowered average tendency a reduction of the variability. That is the range or spread from the common occurrence (a four-children family) would be less, and our figures would be something like the following:
Generalizing the argument we may say:
In so far as conscious restriction is the cause of the lesser fertility of the late decades it will show itself by a disturbance of the form of distribution of the different-sized families.
1. Restriction as commonly considered would increase the 0, 1 and 2 and to some extent 3 children families at the expense of all larger families. For according to the common view there would be no influence of restriction in a family which had already five or six children.
2. Each group of large-sized families would then lose in proportion to the number of families in it, the psychological and social sources of the custom being in no way correlated with fertility.
3. The result will be the appearance in the statistics of late decades of two species of families, one showing the natural tendency and in every way comparable to the species shown in the first decades, the other a species of restricted families with a range from 0 to 3 or 4 and a preponderance of 2's and 0's.
In so far as growing incapacity is the cause, it will show itself not by a disturbance of the form of the distribution of the different-sized families, but by a shifting of the whole distribution back toward a lower point, with probably a reduction of its variability or spread.
If now we turn to the actual facts we shall see that restriction of this type is utterly inadequate to explain them while a growing incapacity would explain them very well.
The comparison of what has actually occurred with what would have occurred as a result first of growing restriction and second of decreased fertility may be more conveniently made by the use of graphic representations than by the numbers.' There are thus presented: (A) the changes that would have occurred if the real fertility of this species of individuals had decreased to a bit less than one half what it was in 1803-1835, the variability being reduced in proportion to the square root of their average; (B) the actual changes in the size of families of college graduates from 1803-1874, and (C) the changes that would have occurred if the reduction in the average size of families had been due to an increase in the number of families in which the natural fertility had been restricted to from to 4 children. In the last case I have calculated the result upon the hypotheses that 2 would be favored by forty per cent., and 3 by twenty per cent, each, and 1 and 4 by ten per cent. each. But any other distribution of the restrictions would lead just as emphatically to the same general conclusions. Still more so would a restriction to families of from to 3 children.
This conclusion is that the changes in distribution actually found decade by decade have far more likeness to those that would result from a decrease in fertility, than to those that would result from restriction. Indeed, the likenesses in the first instance are so close as to force upon us the conviction that the causes are identical. If one forgot the common opinions about the prevalence of restriction and looked directly at the facts he would say: The general fertility sinks from 5 to 2-3; the very large families become impossibilities, the range of possibility which was from 12 to—2 has changed to from 8 to—3 or—4; this species, whatever it is, is dying out. The facts are surely sufficient to rule out restriction of the type described, but before jumping to the conclusion that the obvious explanation of the statistics by a steady decrease in fertility is the true one we must seek other possible explanations of them.Among such explanations that have been suggested to the writer none seems satisfactory. It might be thought that restriction was to 3, 4 and 5 in the early decades, to 2, 3 and 4 in 1835-55, and finally
to 1, 2 and 3. But we can not then account for the great number of zeros in the early decades, nor for the way in which the reduction of the variability occurs. Again it might be thought that there has been a growing reluctance to have families over a certain size, a reluctance that becomes more and more intense in the case of large sizes. But it is impossible to find any scale for the increase of this reluctance such that by assigning more and more individuals to the reluctant class we can derive a series of distributions by decades at all like those actually found.
Of course if we postulate both a lowering with time of the size to which families are restricted and a sliding scale of reluctance that also varies with time we can account for the observed facts. Such a hypothesis is, however, suspicious because of its complexity and apparent artificiality. I do not deny that it may be true, but until we find some further support for it, we are bound so far as the observed facts go to prefer the vera causa which explains the observations with perfect simplicity, and to attribute the numerical degeneration of our group to a real decrease in fertility.
So far as our general mental prepossessions go, however, a real decrease in fertility seems at first sight a preposterous doctrine. One can well imagine the sneer of the physician whose experience emphasizes the frequency of restriction and the pitying smile of the biologist who discerns that a progressive decrease in fertility of a species is a flat contradiction of the doctrine of natural selection. 'Play on with your statistical hair-splitting,' they would say, 'Nothing that you find will disturb our beliefs. We know better.'
But I venture to assert that the experiences of metropolitan physicians will not serve to prophecy the social psychology of the species we have studied, that their opinions may here be as wide of the mark as the common belief that unwillingness is the main cause of the failure of the women of the better classes to nurse their children. As to the contradiction of natural selection, I may suggest that the existence, amount and results of the elimination of types by their failure to produce their kind is after all a problem which only statistical inquiries can settle and that if the doctrine is to be used as an excuse for evading certain obvious facts in human history it is perhaps time that it should be questioned.
The issue is clear. The more fertile members of a race produce of course a larger measure of the next generation than do the less fertile. So also do their children, if fertility is inherited. There should then, according to present-day biology, be a quantitative evolution of fertility. Absolute sterility would needs be the first trait to be eliminated from a species. It should have disappeared from the human stock æons agro. And so long as there are variations in fertility and a transmission of these variations the fertility of a race must keep up to the racial type and ought to increase. It makes no difference whether the type can change only by sudden extreme variations or by a gradual change of its center of gravity. Of whatever sort the effective variations are, the ones that must needs win in the case of fertility are variations on the plus side. But what we actually find is good evidence of a decrease.
Although such emphatic facts as those reported here have never previously been at hand, the question has been clearly seen. In 'A Statistical Study of Eminent Men' in the February number of this Monthly, Professor Cattell called attention to the apparent inadequacy of natural selection to account for the rise and fall of nations. A note in the April number referring to the Harvard statistics also suggests the dilemma of the doctrine. The question is there raised whether even if the failure to produce were due to a psychic epidemic of restriction, there should not be on current biological theory a natural selection for certain inheritable mental traits of those individuals who resisted the epidemic and consequently a maintenance of race productivity. Our returns give support to this claim since the three generations involved should give nature a fair amount of time. I shall not, however, make any use at this time of this argument.
The decision of the question is equally clear. In so far as the decrease in the size of families is due to a real decrease in fertility, we have an absolute disproof of racial progress by the perpetuation of the characteristics of those who survive and reproduce. It is a simple question of fact. A comparison of families of different epochs, all of which are known to be unrestricted, would give an indubitable answer, and the argument here must not be a flourish of vague generalities.
So far as present facts go the probability is against natural selection in the case of fertility in man. The contrary hypothesis, that a stock like an individual has a birth, growth, senescence and death; that, apart from the onslaughts of rivals or the privations of a hard environment or the suicide of universal debauchery, races die a natural death of old age, lends itself very well to the interpretation of human history and perhaps to the history of animal forms as well. It leaves the causation of this race life and death as a mystery. But a mystery is less objectionable than a contradiction.
- These figures come in the case of Middlebury and New York University from the alumni catalogues, which give the number of children living and dead, from the answers to questions collected by Professor Nicholson in the case of Wesleyan University (both living and dead children ate included), and from President Eliot's report, in which case only living children were counted. There are doubtless inaccuracies in the records, but the tendency of these would be to make our figures relatively too small for the earlier decades, and consequently truer records would only emphasize the decrease in productivity upon which all the arguments of this discussion will be based.
In the case of the Middlebury and New York University records, I have used only those families where the husband had been married at least ten years before he died. In the case of the Wesleyan records all married graduates have been included as the data required to make a selection on the basis of length of married life were lacking. I have to thank Professor F. W. Nicholson, Secretary of Wesleyan University, for the use of his records and Mrs. E. B. Brown for the report of the New York University graduates.
- The Wesleyan averages include cases from 41 through 50, etc.
- The New York University averages include cases from 35-44, 45-54, etc.
- This average would be slightly raised by children to be born after th time of record.