Next Selection Previous Selection

PART V.—SEXES, AGES AND MARITAL CONDITIONS.

1. Sexes.

Sex Proportions.—Of the 37,886,699 persons enumerated in England and Wales in 1921. 18,075,239 were males and 19,811,460 were females, giving an excess of females of 1,736,221 or a proportion of 1,096 females to each 1,000 males. The female excess would of course be somewhat reduced if we could include in the reckoning the members of the fighting forces and mercantile marine temporarily abroad and also the numbers of fishermen absent at sea on the night of the census.

The number of females to 1,000 males enumerated at each successive census was as follows—

 Year.  Females
per 1,000
Males.
 Year.  Females
per 1,000
Males.
1801 1,057 1871 1,054
1811 1,054 1881 1,055
1821 1,036 1891 1,064
1831 1,040 1901 1,068
1841 1,046 1911 1,068
1851 1,042 1921 1,096
1861 1,053    

Of these the preponderance of females is probably overstated in the years 1801 and 1811 owing to the forces serving outside the country in the Napoleonic wars and again in 1901 when a large number of men were absent on military service in South Africa. Apart from these years there has been an almost continuous increase in the female preponderance up to 1911 with a final jump in 1921 nearly as great as that of the whole of the preceding century. In the sudden acceleration of the female excess between 1911 and 1921 will probably be found the most direct evidence of the effects of the war and this will be even more definitely seen if the excess, which is not spread evenly over the entire population, is analysed into a series of age groups. This is done for a number of census years and for a variety of areas in Table 35 of the General Tables Volume and it will be sufficient here to set out the figures for England and Wales as a whole in the years 1901 1911 and 1921

TABLE XXX.—RATIO OF FEMALES TO 1,000 MALES. ENGLAND AND WALES .

The recent increase in masculinity, the proportion of male to female births, which has been observed in the successive reports of the Registrar General since 1915 and which appears to be associated in an imperfectly understood manner with the conditions engendered by the war, is demonstrated in the heavy fall between 1911 and 1921 in the female proportion in the age group 0.5. The initial preponderance of the male sex, however, diminishes rapidly and the relative proportions of the sexes become transposed after about the age of 15 owing to the heavier mortality amongst males. After the earliest years migration also affects the sex balance, still further reducing the male proportion.

But the greatest changes of all are those shown for the age groups between 20 and 45 in 1921, the years covering the ages of the majority of those who fell in the war. The males at these ages declined in numbers from 6,671,000 in 1911 to 6,566,000 in 1921, while the females increased from 7,307,000 to 7,693,000 so that the female excess at these ages which was decreasing slightly before 1911, rose from 636,000 in that year to 1,127,000 ten years later, corresponding to an increase in proportion from 1,095 to 1,172 females for each 1,000 males.

If the age group is extended to cover the years from 20 to 64, as including the mass of those who form the economic and reproductive elements of the population, it will be seen that while the males have increased in the ten years from 9,450,000 to 10,082,000 (i.e. by 6.7 per cent.) the females increased from 10,354,000 to 11,510,000 (i.e. by 11.2 per cent.) the preponderance of the latter being expressed by the proportion of 1,142 females to 1,000 males in 1921 as compared with 1,096 in 1911.

The mortality of females at adult ages is generally below that of males. It is, however, a subsidiary factor in the earlier years where the mortality of both sexes is low, attaining its importance gradually as the more advanced ages are reached where the excess of females, already high in 1911, has risen to still higher proportions.

Local Variations in Sex Proportions.— Sex proportions vary widely in different parts of the country, local variations being determined in the main by social and industrial conditions.

In London, with its concentration of clerical and domestic occupations, recruited by women in large numbers, the female proportion stands at the high figure of 1,165 per 1,000 males, a figure which is exceeded in the western residential suburbs of Surrey and Middlesex as the result, no doubt, of the proportion of domestic servants employed there. Again in the southern counties, containing numerous seaside resorts which offer attractions to the elderly and unoccupied classes (preponderantly female), the female excess is well above the average. On the other hand in mining areas and in rural areas generally where the heavier physical occupations do not make great demands upon female labour, the sex distribution is more equal and in some counties is in favour of the male.

The general tendencies are illustrated in the following table in which the counties are arranged in the order of female predominance.

TABLE XXXI.—PROPORTIONS OF FEMALES PER 1,000 MALES. ENGLAND AND WALES AND COUNTIES .

Sex Proportions in Other Countries.— In the next table is given a similar series of comparisons with other countries of which recent census records are available.

TABLE XXXII.—C0UNTRIES ARRANGED IN ORDER OF 1921 SEX PROPORTIONS, SHEWING ALSO 1911 COMPARISONS .

2. Ages.

Tabulation.—Now that the use of mechanical sorting and counting devices has become established as an indispensable portion of the census tabulation procedure, the classification of the population of the whole country and of its subdivisions by individual years of age presents little difficulty. The published statistics in respect of the 1921 census will be found to be rather more extensive than those of 1911, Populations, classified by individual years of age, have been published in the county volumes, in respect of each county borough and of the county aggregates of urban and rural districts, while for every individual urban and rural district a tabulation by the conventional quinquennial grouping of age (0-5, 5-10, etc.) has been provided. In Table 32 of the General Tables Volume the full detail is published in respect of England and Wales as a whole and of London, and the aggregates of all urban and all rural districts.

It is doubtful whether the individual year tabulation, as a measure of the incidence of age in a given population, presents any marked advantage over the shorter and more easily studied series of 5-year groups. As is described in greater detail later on, the age returns are not free from error, and, so far as the errors are casual or accidental in character, they may be expected to tend to neutralise one another when the observations of successive years are amalgamated and if, in the absence of contrary evidence, the character of the mis-statements of age can be regarded as similar in different populations it will usually be preferable to base a comparative study of their ages upon a series of age groups rather than upon the records of individual years.

The individual age tabulation, however, has a definite statistical value in that it provides material for detecting and measuring some of the mis-statement tendencies. Even when its abridgement forms a preliminary stage to subsequent operations, the age groups can be selected in a variety of ways, either for the particular purpose of the object in question or merely with the view of reducing to a minimum the inherent errors in the age statements themselves, advantages which were denied by the older form of statement which limited the classification to a fixed and predetermined series of groups between points corresponding to the ages terminating with the digits 0 or 5.

The examination of the 1911 census age tabulation—the first in this country to identify single years throughout the whole of life in respect of the total population —indicated that, in giving their ages on the census schedule, a considerable number of persons appeared to have stated their ages in an approximate form. With the view of reducing these mis-statements and of impressing upon the public that an exact statement was required, the age question on the 1911 schedule was modified in 1921 by asking for each age hi years and months, a degree of detail which was only required in respect of children under 1 year of age in 1911. There was never any intention of tabulating the ages by months. The whole object was that of securing a greater concentration upon the enquiry so as to avoid the looseness of statement which was evident at the previous enumeration. The device appears to have been partially successful as judged by the result, for, though the preference for particular ages still exists, it is sensibly less than it was 10 years ago. If as a matter of fact there exists an appreciable number of people who are in real ignorance of their true age, the adoption of an approximate figure is to be expected and the heaping of the population at the "round number" ages will recur so long as such ignorance remains, whatever precautions are taken during the enumeration.

Comparison with previous censuses.— The age incidence of the population of England and Wales in 1921 and its relation to that of past censuses and to that of other countries is shown in the various diagrams and tables attached to this Report.

DIAGRAM E.
ENGLAND & WALES.—TOTAL POPULATION BY INDIVIDUAL YEARS OF AGE, 1921 & 1911.

Age structure for England and Wales in 1921 and 1911

DIAGRAM F.
ENGLAND & WALES.—TOTAL POPULATION BY INDIVIDUAL YEARS OF AGE, DISTINGUISHING MALES AND FEMALES, 1921.

Age structure of males and females for England and Wales in 1921

DIAGRAM G.
TOTAL POPULATION OF ENGLAND & WALES BY QUINQUENNIAL GROUPS OF AGE, 1921-1881.
PERSONS.

Age structure in five year bands 1921 to 1881 for all persons

DIAGRAM H.
TOTAL POPULATION OF ENGLAND & WALES BY QUINQUENNIAL GROUPS OF AGE, 1921-1881.
MALES.

Age structure in five year bands 1921 to 1881 for males

DIAGRAM J.
TOTAL POPULATION OF ENGLAND & WALES BY QUINQUENNIAL GROUPS OF AGE, 1921-1881.
FEMALES.

Age structure in five year bands 1921 to 1881 for males

Diagram (E) shows the total population by individual years of age in 1921 and 1911.

Diagram (F) shows the 1921 population by individual years of age, distinguishing males and females.

Diagrams (G), (H) and (J) show the 1921 population in groups of ages in Relation to the populations of past censuses.

In considering the age constitution of the population either of to-day or of any other period it must be borne in mind that in itself it is nothing more than a resultant distribution from the effect of factors operating over practically the whole of the preceding century. The population ranges from age 0 to ages beyond 100, so that the numbers in the several age groups in 1921 are survivors of the births which have occurred since before the year 1821 and it is apparent, therefore, that to the variations in the numbers born in successive years and in their subsequent rates of survival is mainly due the moulding of the present age distribution. It is not proposed here to examine the birth and survival rates in any detail, but attention may be drawn to the fact that in England and Wales as a whole the number of births each year was increasing at a rapid rate up to about 1875, after which the pace diminished until about 1903 when the maximum (with the exception of the record number of 957,782 in the rather special circumstances of 1920) was attained, and a definite if somewhat unsteady decline set in. This factor alone would lead us to expect each age group over 45 to show a considerable increase over the same age group of ten years ago while between 20 and 45 a smaller increase, and below 20 a decrease might be looked for. The general improvement in survival rates which has taken place over the same period would ordinarily tend to be in favour of the later population, with the result that increases due to higher numbers of births in earlier years would be still further increased by the lightening of the mortality affecting them and the decreases (below age 20) correspondingly diminished.

In addition to the natural forces of birth and death there are large and varying movements of population, usually referred to as emigration or immigration, which occur chiefly during youthful and young adult age periods.

During the past decennium, the heavy war mortality and the striking variations in the birth rate associated with the war years have been superimposed upon the slower and more gradual changes which might otherwise have been expected from past tendencies. But though the war has left prominent scars upon the population curve the extra-war factors have also had their share in the shaping of the present population and the identification of the former cannot be obtained "with out careful statistical analysis. In this connection reference may be made to the section dealing with mis-statements of age. in which an approximate age analysis of the civilian intercensal movement has been given in a table comparing the survivors from the 1911 census with the numbers actually enumerated at several age periods in 1921.

TABLE XXXIII.—POPULATION OF ENGLAND AND WALES IN AGE GROUPS, 1921 AND 1911 .

TABLE XXXIV.—PROPORTIONATE POPULATION OF ENGLAND AND WALES IN AGE GROUPS 1921 AND 1911 .

The most striking feature to be observed in regard to the 1921 age distribution is illustrated in the curiously irregular shape of the curve in diagram (G). Instead of commencing with a maximum for the youngest age group (0.5) and decreasing more or less steadily with advancing age as hitherto, the curve rises for the first two quinquennial periods to a maximum in the age group 10.14, after which the subsequent decline is characterised by two distinct waves, the first terminating at about ages 40 to 45 which gives the impression of a bite taken out of the curve and which is much more pronounced than the slight flattening observable at younger ages in the curves for preceding censuses given in the diagram, and the second, occurring after age 45, which is similar in general respects with those of earlier censuses.

The two new features—the irregularity at the earliest ages and the depression at early adult ages—are, of course, the result of war conditions, and though the influence of time may be to reduce their prominence by a tendency to nil up the depressions and to fine down the excrescences they will inevitably recur at every fere enumeration, though at correspondingly later ages, until the whole of the existing population has passed away.

From the detailed curves of diagrams H and J in which the special features can be more definitely located it will be seen that the apparent deficiency in children which affects both sexes alike, is greatest amongst those aged 2 and 3 at the date of the census and continues with diminishing weight right up to age 12. The children enumerated at ages 2.6 are the survivors of the births of the period between the middle of 1914 and the middle of 1919, and their presentation in age categories merely reproduces the sequence of the birth rates of the war years which has been fully discussed in the Annual Reports of the Registrar General. But the annual numbers of births had begun to show a definite decline for several years before the war, and the rise in population with advancing age which is just discernible between ages 0 and 2 in 1911 is now very marked between the ages of 6 and 12. By a projection of the curve, as it appears between these age points, back to age 0 it may be inferred that, had circumstances been perfectly normal during the whole of the decennium, though there would have been no depression corresponding to that now found between ages 2 and 6, the starting point of the curve at age 0 would have commenced at a much lower point than it actually does and that the numbers enumerated at ages 0 and 1 must be regarded as abnormally high just as those of the succeeding years are abnormally low. As we know, the former arise from the sudden and temporary increase in the birth rates of the years which immediately followed the armistice and which have now given place to rates which are lower than any recorded in the peace time history of the country.

From the separate sex curves it will be seen that the abnormal depression at early adult ages is confined mainly to the males and must almost precisely correspond to the direct losses which occurred amongst the respective fighting forces after the middle of 1914.

Compared with 1911 the population below 5 years of age shows a decline of nearly 14 per cent., which is reduced to 4.8 per cent. in the next five years and is followed by an increase of 4.8 per cent. for ages 10.19 in the two succeeding decennial age periods (i.e. 20.29 and 30.39) the males have declined by 5.7 and 3.1 per cent. respectively, while small increases of 0.8 and 4.9 per cent. are recorded for the corresponding female ages.

The combined effect of these decreases and meagre increases results in a net intercensal loss of 1.8 per cent. in the total population under 40 years of age (or a loss of 3.3 per cent. of males and 0.3 per cent. of females), and with these may be contrasted the very large increases recorded in the population of 40 years and over. In each of the age and sex groups at the later period the increases are consistently high, rarely falling below 20 per cent., and amount in the aggregate to 23.0 per cent. altogether, the female increase being slightly in excess of the male at percentages of 23.3 and 22.7 respectively.

War losses affect the age group above 40 to some extent and the maximum male rate of increase is not attained till at least 50 is reached. In the 50.59 and the 60.69 groups the rate of growth of the male population is slightly above that of the female, but at the last two age periods of the table the disparity between the sexes, which was heavily in favour of the females in 1911, has been still further extended by the higher rates of increase recorded in respect of the female population.

One hundred and ten persons declared their ages as 100 or over, 30 of them being males and 80 females. The corresponding numbers in 1911 were 128, 36 and 92 respectively, so that at these extreme ages there would appear to have been a decrease in each class. Past experience, however, suggests that there is a measure of unreliability in the statements at these ages due to overstatement, conscious or otherwise, and this probably obscures any real movement at this period of life.

In 14,395 cases (6,826 males and 7,569 females) the statement of age was omitted altogether from the census schedules or was given in a form too indefinite for classification. The procedure followed in 1911 in respect of the 13,167 similar cases then reported has again been adopted, ages being assigned in accordance with other information on the schedule so far as possible and the remainder distributed proportionately among the stated ages.

The combined effect of the several increases and decreases over the whole of the age field has resulted in a general ageing of the population which continues, in an exaggerated form, owing to the special features which have been described, the transformation that has been gradually taking place since about 1881 after the decline in the birth rate which set in at about that period.

The average age of the population at each of the censuses between 1881 and 1921 inclusive are as follows:—

AVERAGE AGE.

   Persons.  Males.  Females.
1881 26.2 25.7 26.7
1891 26.6 26.1 27.1
1901 27.4 26.9 27.9
1911 28.6 28 29.1
1921 30.6 29.9 31.2

Within any range of age less than the whole of life, the population in the older portion will often be found to have increased at the expense of that of the younger, and in questions relating to selected sections of the community the effect of the alteration in age constitution may be as significant as a change in mere numbers.

Thus, if attention be directed to the economically productive section of the population, as distinct from what may be called the dependent section, it will be seen that between the ages of 15 and 65 the proportion of males to the total population of all ages has increased from 30.8 per cent. to 31.2 per cent. At the same time it is clear from the tables that the increase occurs only in the older half—from 40 onwards—where the proportion has risen from 10.7 per cent. to 12.5 per cent., while below 40 the proportions have decreased. Males and females together between the ages of 15 and 65 account for 66.2 per cent. of the total population now as compared with 64.2 per cent. ten years ago, but, as in the case of males alone, the average increase of 2.0 per cent. for the whole group may be resolved into a decrease of 1.8 per cent. below age 40 combined with an increase of 3.9 per cent. above that age, so that in terms of the physical qualities demanded by the majority of industrial occupations, the improvement in the standard indicated by the increase in the proportions at the working ages may well be neutralised if not reversed by the increase in the average age.

Or, to take another example, one of the principal factors determining the birth rate in the country is the proportion of women of child-bearing ages, and it will be seen from Table 35 of General Tables that in the past intercensal period the proportion of women between 15 and 45 has increased from 24.9 to 25.0 per cent. The change, which is apparently in favour of a slightly increased birth rate, is again offset by a shifting of the average age of the group. Between 20 and 35, the ages at which the majority of births occur in this country, the proportion has declined (13.3 per cent. in 1911 to 12.8 per cent. in 1921), and it is mainly in the older and less fertile section that the increase is significant.

Comparison with Other Countries.— In the following tables the age and sex constitution of England and Wales is compared with those of a number of other countries for which recent census statistics are available. Unfortunately the post war figures in respect of France and Italy have not so far come to hand, and as there would be little purpose in inserting figures relating to an earlier census which did not reflect the changes which have occurred since 1914 they have been omitted from the list. Similarly as regards Ireland, a census has not been taken since 1911, and no recent figures are therefore available.

TABLE XXXV.—NUMBER OF MALES AND FEMALES IN VARIOUS AGE GROUPS IN 1,000 PERSONS AT ALL AGES IN SEVERAL COUNTRIES .

TABLE XXXVI.—AGE DISTRIBUTION OF THE POPULATION IN VARIOUS COUNTRIES PER CENT. OF THAT IN ENGLAND AND WALES, 1921 .

As might have been expected, the most prominent feature in the comparison is the sharp contrast between the chief European countries which participated in the war and those which remained neutral, and this is emphasized most strongly in the proportions of children under 5 years of age. In Germany and Austria particularly, and less so in Belgium, Czecho-Slovakia and Hungary, the proportions are markedly lower than in this country, while in all the neutral countries, except Switzerland, the proportions are substantially higher. The differences in the dates at which the respective censuses were taken is not without significance here, however, for the fall in the birth rates, which was serious and progressive between 1914 and 1918, was generally reversed in the post-war years, so that whereas at the census of England and Wales, which did not take place till the middle of 1921, the age group 0.5 includes survivors of a relatively high birth rate as well as the years of the greatest depression in the rates, at a census taken in 1919—that of Germany is an example—the survivors relate to the worst of the war years without any compensation from the succeeding period.

Between the ages of 5 and 20 the proportionate population is lower in this country than in any other of the European countries shown in the table with the exception of Belgium, while at the adult ages from 20 to 60 the proportion is generally higher. The effect of war losses in men is observable in the differences between the male and female proportions at ages 20.40 in the case of Germany, Austria and Hungary.

Compared with the proportions in this and other European countries, the Dominions and the U.S.A. show large excesses at the younger ages combined with a deficiency in later life, and in particular a markedly lower proportion of females at all adult ages. In spite of the relatively fewer females at child-bearing ages, the excesses at the infantile ages are among the highest shown in the table, and while in South Africa, the proportionate excess thereafter diminishes with advancing age, in Australia, New Zealand, Canada and the U.S.A. the progression is less regular, there being a notable rise in the male columns at ages 20.40—presumably due to migration —and usually a further slight rise at the most advanced ages separately identified.

Sections of the Country.— The age and sex constitution of separate sections of England and Wales, shown in Table XXXVII present marked contrasts in their characteristic features. In the analysis according to degree of urbanisation, the local government unit has been adopted as the basis of distribution with the result that the contrasts will tend to be understated to the extent to which such distribution differs from that indicated by a more scientific criterion of population aggregation. It will be understood that many rural districts include considerable populations living under urban conditions, particularly in mining areas, while many towns embrace areas of distinctly rural type. In the classification of towns also, while the county boroughs are as a whole more populous than other urban areas, a considerable number of towns which have not acquired county borough status have a much larger population than the smaller county boroughs.

From the left-hand section of the table, which distinguishes rural districts and certain urban aggregates, it will be observed that the rural areas start with a relatively high proportion of young children, the deficiency of potential parents being more than counteracted by their relatively higher rate of fertility and the lower mortality affecting their children. The latter factor appears to operate in favour of the rural districts up to the age of independence at which the attraction of the town begins to be felt, resulting in the transfer of a large number of young adults to the urban areas. The effect of this emigration, commencing rather before the attainment of age 20 in the case of females and a few years later in the case of males, depresses the rural population curve up to about the age of 50, after which the more favourable mortality experienced in middle and late life, coupled possibly with some return movement from the towns, raises the population, the proportions becoming abnormally high at the advanced ages, particularly amongst males.

As between the urban types of population, the smaller towns present a distribution which but for a slight loss of males at the oldest ages is very similar to that of the country as a whole, whereas the county boroughs evidence more definite signs of the migration movement shown by the rural areas, though in a converse manner. Here the proportions are increased as a result of inward migration at early adult ages with a reversal at about age 50 and a relatively large decrease in the final age groups

In London, where urban conditions may be said to exist in their most extreme form, the excess of females is higher than in any of the. other areas distinguished. This excess, which arises originally from a very heavy influx of young women between the ages of 20 and 30 raises the proportions at these and subsequent ages to a high figure and by reflex action depresses the proportions at other ages shown in the table since they are designed to aggregate to a fixed sum, viz., 10,000 persons in each division. In spite of the large numbers of women of fertile age the proportion of children is extremely low, while at the old ages the London distribution appears to be less urban in character than that of the county boroughs, for the incidence of the proportions of the former are here much more akin to that of the smaller towns and are not very dissimilar from that of the country at large.

The classification of the general population by geographical divisions, shown in the last six columns of the table, presents types of distribution different from those produced by the aggregation according to degree of urbanisation, but types which are equally marked and characteristic in themselves. The most complete contrast will be found as between the population of Wales on the one hand and of the southern counties on the other. In the former, the only section with an actual excess of males, the incidence of population is greatest at the youngest age group of the table—birth-rates being at a maximum in Wales—from which, in comparison with the rest of the country, it falls almost continuously throughout the whole period of life, the fall being consistent among both males and females. In this section, the distribution, instead of conforming to the rural type as might have been expected from the high proportion. of rural population, is completely dominated by the industrial element connected with mining and associated industries where early and fertile marriage is the rule and where the labour demand is mainly for young and vigorous manhood.

The eastern group almost exactly conforms to the rural type of distribution, but in the southern group, which possesses, in addition to a large proportion of rural area, a position and a climate especially attractive to the elderly and infirm, the influence of the latter is seen in an accentuation of the proportions in old age and a relative deficiency in children, so that the incidence here is almost as consistently in favour of advancing age as it is with diminishing age in the case of Wales.

In the central and northern counties the distributions are not very dissimilar from that of the country as a whole, except at the most advanced ages in the northern area. The slight initial advantage in the proportions of young children is maintained rather longer and falls more definitely and consistently in the northern area, giving an impress suggestive of the Welsh type which may reasonably be ascribed to their common association with mining and similar heavy industries.

TABLE XXXVII.—SEX AND AGE CONSTITUTION OF SECTIONS OF THE POPULATION .

Mis-statements of Age.— Reference has already been made to the change introduced in 1921, in the form of the question on the householder's return under which the age statistics of the population have been obtained. In 1911 the inquiry asked for a statement of the age last birthday of each person enumerated, with the exception of infants under one year of age in respect of whom the age in months was required. The variety of inconsistencies then observable from the returns themselves, or disclosed by comparison with other sources of information, showed that the same tendencies to mis-statement which had been experienced at previous enumerations were still at work, and that even if, with the spread of education and the inclusion of a growing proportion of the population within the period of birth registration instituted in 1837, the errors appeared to be decreasing, they were still considerable at certain years of age.

Accordingly, in 1921, an attempt was made to accelerate the improvement in the age statement by extending somewhat the scope of the question and asking all persons of whatever age to return their age in years and months. The detail required, though not intended for or used to elaborate the subsequent tabulation, required more concentration on the information sought, and would, it was hoped, lead to a substantially higher degree of precision in the replies. The customary examination of the age curve furnished by the census returns of 1921 has therefore a special interest, even though it may not be possible definitely to associate any change in the manner of record with its predisposing cause and to say how much of it may have been due to the modification of the terms in which the inquiry was made, how much to the greater familiarity with official forms, or how much to a growing appreciation of the utility of the census and the many objects it is designed to serve.

The errors in the age statements, as in any similar observations, may be divided into two main categories (1) those which are generally local and unbiassed in character such as may arise from a looseness of statement or from ignorance of the precise facts, and (2) wilful omissions or deliberate mis-statements imparting a definite distortion to the curve sometimes over a considerable portion of the whole age field. Intermediate between these two may be placed the form of mis-statement suspected in earlier enumerations and referred to as the "age next birthday" error, implying, as its title suggests., a tendency to give the next higher age instead of the attained age asked for; the possible range of this error is only one year in an individual case, but it has the disadvantage that it always operates in the same direction.

The types of error are examined in turn and, as between them, any improvement arising from the change in the form of the question must be expected to be limited to the first category, though the "age next birthday" error should also be reduced —if it has been of any moment in the past—by the opportunity now given of stating the age in years and months.

(1) Local Errors.— The extent of the mis-statements covered by this designation is generally identifiable from the observations themselves, and, partly because they are so easily recognised, they can be readily removed by graduation or, owing to their local and neutralising character, by the amalgamation of successive years' figures in groups centrally disposed about points of greatest inflexion.

Referring to the preceding diagrams it will be observed that below age 15 the curves, while extremely irregular, do not in themselves give rise to any apprehension as regards the accuracy of the records at successive years of age. Similarly at the latter end of life the fall in population from age 72 onwards is continuous and consistent and arouses no obvious criticism. But the same cannot be said of the ages intermediate between 15 and 72. Both the male and female curves in this section exhibit a regularly serrated edge with the upper points of the teeth occurring, with few exceptions, at the even years of age and in a particularly aggravated form at the years of age 30, 40, 50 and 60. The exceptions to the even number rule, worthy of note and common to both sexes, occur between 20 and 22 where the attraction of the age of maturity appears to elevate the curve slightly at the 21 point at the expense both of the 20 and 22 observations and at the years round the ages 35, 45 and 65. The modification of the alternations at these ages may suggest that a preference exists for stating an age ending in the digit 5, but confirmation of this is not very strong, for an overstatement is only marked at the age 45. At 35 and 65 the curve is only slightly irregular and at 25 and 55 the evidence appears to be of under instead of overstatement. If both the preference for even ages and also for the quinquennial points of age ending in 0 and 5 were established their synchronisation at ages ending in 0 and their opposition at ages ending in 5 would produce a series of irregularities like those present in the distributions under examination, namely, marked elevations at the 0 points and generally feeble and hardly discernible distortion at the 5 points, and the curves may therefore be interpreted as favouring these explanations regarding the error tendencies. The tests described below suggest, however, that the tendencies may have a somewhat different origin or perhaps may be better described in another way.

With the exception of the special preference for the age 21 now observable, the form of the 1921 variations correspond almost precisely with those of 1911 (the comparison may also be extended to the curves of deaths by ages discussed in the Registrar General's Statistical Review for 1923), but the curves clearly show that these variations are less in extent now than they were 10 years ago, indicating that some progress has been achieved in the reduction of this casual type of mis-statement.

The errors vary in degree at different periods of life, being rather more pronounced between the ages of 30 and 50 than elsewhere, but they are definitely cyclical in character and they seem to be so closely associated with the unit digit of the age that an attempt has been made to obtain an approximate measure, in an aggregate numerical form, of this association and to examine its incidence as between males and females and as between the present census and that of 1911.

For this purpose, the male and female populations of 1911 and 1921 between the ages of 23 and 72 were each divided into five decennial sections of age, 23.32, 33.42, etc., and the numbers at individual years of age aggregated in groups differentiated according to the units digit of the age so that persons returned as 23, 33, 43, 53 and 63 formed the first group, those as 24, 34, etc., the second group, and so on, making 10 groups in all.

Ages
ending
in
Proportionate
Population.
Difference. Ages
ending
in
Proportionate
Population.
Difference.
Actual. Graduated. Amount. Per cent.
of
Graduated.
Actual. Graduated. Amount. Per cent.
of
Graduated.
  MALES.—1921.     FEMALES.—1921.  
3 1,083 1,074 9 0.8 3 1,090 1,085 5 0.5
4 1,064 1,068 -4 -0.4 4 1,077 1,074 3 0.3
5 1,056 1,057 -1 -0.1 5 1,061 1,060 1 0.1
6 1,038 1,044 -6 -0.6 6 1,033 1,043 -10 -1.0
7 1,002 1,026 -24 -2.3 7 989 1,023 -34 -3.3
8 1,020 1,005 15 1.5 8 1,026 1,001 25 2.5
9 982 979 3 0.3 9 970 975 -5 -0.5
0 998 950 48 5.1 0 1,015 946 69 7.3
1 868 917 -49 -5.3 1 846 914 -68 -7.4
2 889 880 9 1.0 2 893 879 14 1.6
                   
Total 10,000 10,000 84 0.8 Total 10,000 10,000 117 1.2
                   
  MALES.—1911.     FEMALES.—1911.  
3 1,078 1,079 -1 -0.1 3 1,088 1,088
4 1,080 1,074 6 0.6 4 1,090 1,080 10 0.9
5 1,076 1,064 12 1.1 5 1,073 1,067 6 0.6
6 1,050 1,049 1 0.1 6 1,045 1,050 -5 -0.5
7 967 1,030 -63 -6.1 7 970 1,029 -59 -5.7
8 1,037 1,007 30 3.0 8 1,038 1,004 34 3.4
9 963 978 -15 -1.5 9 957 975 -18 -1.8
0 1,052 945 107 11.3 0 1,055 941 114 12.1
1 809 908 -99 -10.9 1 796 904 -108 -11.9
2 888 866 22 2.5 2 888 862 26 3.0
                   
Total 10,000 10,000 178 1.8 Total 10,000 10,000 190 1.9

Frequency of ages ending in different numbers, males and females, 1921 and 1911

To make the statements comparable with one another each series was then modified so as to aggregate to a common total—10,000—and the figures thus obtained are shown in the first column of each statement and are also displayed by the irregular curves in the adjoining diagrams: In the absence of any distorting factors, each series of 10 records might normally be expected to lie on a smooth and regular curve, but that they do not do so is obvious from the diagrams, and the next step was to pass a smooth curve through each of the series. The nature of the curve is not of particular moment; a parabola of the second degree fitted by the method of least squares was deemed sufficiently suitable, and the only points to be noticed are that its form rendered it free from any personal bias of the operator and that the same type of curve was applied to each of the population distributions.

The graduated values thus obtained are shown by the dotted curves in the graphs and in numerical form in the second column of each statement, while the third and fourth columns record the differences between the original data and the smoothed series.

Three interesting inferences emerge from the difference series measured in this way.

The first concerns the general incidence of the errors in the four sections distinguished. From the total lines, which aggregate the several age group differences, it may be inferred that the local age mis-statements at the census of 1921 are rather less than half in the case of males and less than two-thirds in the case of females than they were at the census of 1911.

Further, while the males showed only a very slight superiority in the matter of accuracy of statement in 1911, the greater improvement which is recorded for them places them now in a more definitely favourable position. The frequently expressed view that the ages given by females are more suspect than those of males apparently did not apply in 1911 to the local, or cyclical, type of error here discussed, and the 1921 figures show that there is still room for improvement in the male section even though females appear to be the greater offenders at the present time.

The second point of interest concerns the relation of the error to age. In each table two pairs—and the same two pairs—of adjacent ages stand out prominently from the rest showing large and complementary variations which account for between three-quarters and five-sixths of the aggregated errors of each of the four distributions. The greatest disturbance occurs at ages ending in 0 and 1 where a heavy excess at the former, usually and probably rightly accounted to a particular attraction in this decennial round number, is obtained almost exclusively at the expense of the next higher age. There is little evidence that any of the excess at the 0 point is derived .from the preceding ages ending in the digit 9, for this group is only slightly below its graduated counterpart in three of the curves and in the fourth—the males of 1921—it is slightly above. The other principal seat of disturbance lies between the 7 and 8 groups. In this case the direction of the movement appears to be reversed for the deficiency is in the first and the movement would appear to be generally one of transfer from the 7 to the 8 group. Mr. George King, F.I.A., in his remarks on the graduation of the 1911 populations, called attention to what he described as an apparent preference for ages ending in the digit 8 and the same feature is again repeated in the 1921 experience. It will be seen, however, that the deficiency in the 7 group is always much greater than the excess in the succeeding group and it is suggested, therefore, that the movement may be more fairly associated with an avoidance of the digit 7 than with any special respect for ages ending in 8, the 8 presumably being given in preference as representing the next birthday figure. The remaining age points, viz., 2, 3, 4, 5, 6 and 9 show no consistent and significant disagreement between the crude and graduated figures and it may be inferred, therefore, that statements at these ages are fairly accurate on the whole. Distortion at ages ending in 5 was perceptible in 1911 but it is, in neither of the sections, of general importance, and there is similarly little support for the theory of a common preference for even as opposed to odd years of age. Ages ending in 0 and 8 are seriously inflated as already described, but the impression of exaggeration at the even age points of 2 and 6—gained from the individual age curve in diagrams E. & F. is a spurious one due to the deficiencies at the 1 and 7 points and not to an inherent excess at the even ages themselves.

The third point of interest in this analysis lies in the light it throws on the result of compressing the individual age records into groups of adjacent ages. The opposing errors at the 0 and 1 and at the 7 and 8 points of age will obviously tend to neutralise one another when the populations are amalgamated and a series of age groups so arranged that these pairs remain unbroken will clearly be preferable to one which involves a split in either of them. From the differences between the graduated and ungraduated values shown in the tables, the best grouping, that is, the one involving the least resultant error, can be readily determined. Thus for the male population of 1921 the quinquennial grouping least distorted by the cyclical error would be one composed of a series of five year groups centred round the ages ending in digits 4 and 9, e.g. 22-26, 27-31, etc., while in the other three sections, central ages ending in 1 and 6 are superior, more so for both the 1911 populations than the 1921 female population for which last, central ages ending in 4 and 9 are almost as good. In the conventional series 0-4, 5-9, 10-14, etc., the resultant error is not as low as it might be, but on the other hand it is far preferable to the worst combinations associated with the central ages ending in 3 and 8, e.g. groups 1-5, 6-10. 11-15. etc

(2) Deliberate or Biased Mis-statements and Omissions.— The more fundamental displacements caused by the deliberate omission of large numbers at particular ages or by the tendency at certain periods of life to return an age several years greater or less than the true age cannot be identified from the age curve itself, and for that reason will not be corrected, as local or unbiased errors may be, by passing a graduated curve through the crude frequencies given by the enumeration returns. Even if the returns were complete and accurate, the resulting age sequence would not be perfectly regular. The existing population has been derived in the first instance from a series of births, which itself shows considerable variations from year to year. Subsequent reductions by death and migration, operating with different degrees of force at different ages, while having little effect on the original irregularities/will tend to alter the shape of the curve and to impart to it characteristics representative of their own special incidence. The marked depressions, for example, in the 1921 curves at the ages of infancy and in the male curve at the early adult ages, are significant impressions of the events of the past intercensal period and must remain permanent features of the population of this country, though at advancing ages, for practically the whole of the next century. Care must, therefore, be taken not to confuse the real features of the distribution with what, on a cursory survey, may appear to indicate over or understatement in the returns.

The following examination takes the form of comparing the enumerated population at the several age periods in 1921 with the numbers expected by survivorship either from the births of the period 1911-1921 or from the numbers recorded at the 1911 census, precautions being taken to keep the errors in the method of tracing survivors down to a minimum so far as is possible. Advantage has been taken of the fact that since 1911 the deaths of each quarter of the calendar year have been recorded in sixteen age divisions below age one and by individual years thereafter. Allowance has also been made for migration in respect of which the Board of Trade have, since 1912, obtained returns, by age, of persons leaving or coming to this country from places out of Europe with the intention of permanently changing their residence. Further, the age groups forming the basis of comparison do not follow the conventional series, 0-4, 5-9, etc., but have been selected so as to minimise the effect of the cyclical errors already discussed which might otherwise tend to obscure other features in the comparisons.

Some difficulty arises, in estimating the expected survivors in 1921, in the distribution of migrants over the intercensal years and by groups of age. The records obtained from time to time from the Board of Trade indicate a civil outward migration balance of approximately 870,000 for the decennium as compared with 590,000 (270,000 males and 320,000 females) deduced from the actual intercensal increase after allowance for births and deaths, the difference possibly being accounted for by an inward balance of migration from other countries of the United Kingdom (no records of which are obtained) and possibly by some imperfection in the returns of the war years. The Board of Trade returns for individual years were reduced rateably so as to total 590,000 and were distributed by age in accordance with such age statistics as were available in respect of them.

The lack of detailed information concerning the deaths in the male population of military age in the war years unfortunately renders it impossible to obtain satisfactory comparisons at the ages affected by that class of the population. Below age 9 the expected number of survivors, as at the 19th June, 1921, has been calculated for each individual year of age, and these are shown in relation to the enumerated in the statement on page 77. Corresponding 1911 figures, so far as they are available, are also given side by side the separate sex figures being amalgamated for simplicity in each case as they show no characteristic differences.

The experience of 1911 which disclosed a serious discrepancy between the survivors at ages 0 and 1 and the population enumerated at those ages, with a fairly close agreement at succeeding ages, appears from the following to have been repeated in 1921. At each of the ages below 2 the numbers returned at the 1921 census are more than 20,000 less than the comparable numbers calculated from the births of the preceding two years after allowing for the deaths and migration among them. The deaths at these ages, it may be noted, are recorded and tabulated in considerable detail as regards age, and migration is almost negligible, so that the conditions for an accurate computation of the survivors are more favourable here than at any subsequent period of life. In attempting to account for the discrepancies it may be well to consider which of the two figures, that for survivors or for enumerated more probably represents the facts. Preference has, in the past, been given to the survivors at ages under 5 on the ground that while the machinery of registration left little room for error, either of commission or omission, in the tracing of survivors the conditions under which the census returns are taken are much less rigid, and that mis-statements of age or even exclusion altogether from the returns are quite possible The fact that the survivorship calculation does reproduce the enumerated almost exactly at each of the seven ages 2 and 9, and that in doing so it accounts satisfactorily in respect of them for the whole period between birth and census (a period which in respect of each age class includes the first two years of life) supports a preference for the "survivors" at the first two years of life, particularly as it follows the very similar experience in respect of the 1911 population. Further confirmation of the preference for "survivors" over "enumerated" at early ages in 1911 is forthcoming from the 1921 returns in respect of later ages, for, if reference be made to Table XXXVIII, it will be seen that there is an apparently large excess of enumerated over survivors in the male and female age groups 9-14, which would be converted into a deficiency of smaller magnitude if the 1921 survivors were calculated from births instead of from the enumerated in 1911.

Age. 1921. 1911.
Enumerated.

(Thousands)
Survivors.

(Thousands)
Survivors,
less
enumerated.

(Thousands)
Enumerated.

(Thousands)
Survivors.

(Thousands)
Survivors,
less
enumerated.

(Thousands)
0— 795 819 24 782 832 50
1— 826 848 22 743 797 54
2— 552 555 3 789 790 1
3— 537 544 7 777 780 3
4— 611 616 5 763 759 -4
5— 655 657 2      
6— 707 707 0      
7— 725 724 -1      
8— 721 723 2      

The evidence at hand undoubtedly points to the superiority of the "survivors" figures and to a real deficiency in the numbers returned as aged 0 or 1, a deficiency which though apparently only half as great as it was in 1911 is still large in contrast to the differences at succeeding ages. The two suggestions usually put forward in explanation of the deficiency at this period of life are that it is due either to the omission of very young children from the enumeration or to the statement of age next birthday in a proportion of the cases.

The age next birthday theory is not, from general reasoning, without plausibility. At whatever date the census is taken, about 4 per cent. of the population at any age are within a fortnight of their next birthdays, or 8 per cent. within a month, and to a large number of these the statement of next birthday may not appear a mis-statement at all. In fact, to the majority who are ignorant of statistical methods an age nearly attained will seem a truer representation of the fact than one apparently out of date, though the error caused thereby should naturally be less when the opportunity is given of stating the age in years and months than it was, e.g. in 1911, when the returns were confined mainly to integral years of age. A comparison made by the Registrar General for Scotland1 between the true ages of a sample of the Scottish population aged 0.5 as given by the birth registers, and the enumerated ages as returned by the same individuals at the census, showed that the tendency to overstate the age in the returns, though much less than in 1911, still exists, and that in from 2 to 3 per cent. of the cases examined the age next birthday was returned at the census instead of the true age. The census of this country was taken on the same day and in generally similar circumstances and without assuming that the experience of the small Scottish sample corresponds precisely to that of the whole child population of this country, it does serve to confirm the assumption that the age returns continue to be affected by the age next birthday error. There is no reason, however, to suppose that it is limited to the ages of infancy. It occurs presumably at all periods of life though it should only be prominent in respect of children under 1 year of age, since at any subsequent age the loss by transfer to a higher age will tend to be compensated by a similar transfer from the age next below. But the discrepancy in the above table occurs in respect of the second year of life as well as in the first and the 'age next birthday' theory, which may explain the latter cannot with the same confidence be applied to the former. In fact if the whole of the discrepancies shown in the table were attributable to this cause alone, the proportions of children in respect of whom ages next birthday were given, would be as follows:—

Percentage of Children whose Age was given as next birthday on the assumption that
the whole of the discrepancy is due to this form of mis-statement.


True Age.
1911.
percent.
1921.
percent.
0— 6.0 2.9
1— 13.0 5.4
2— 13.3 8.8
3— 13.8 10.3
4— 13.7 9.9
5— 9.6
6— 8.9
7— 8.6
8— 8.9

As regards 1911 the result is in accord with anticipation. The error is smallest below age 1 where the age was asked for in months and is rather more than double at each succeeding year where a statement in years only was required. Unfortunately the correspondence is destroyed in 1921, for though the discrepancy—on the hypothesis adopted—is least at age 0, it increases continuously to age 3, and this notwithstanding that ages in years and months were asked for throughout. It seems probable therefore that the age next birthday mis-statement which; of necessity, involves ages where the survivors and enumerated are in approximate agreement, as well as those at which serious differences occur, is not the only factor in the deficiency at the earliest years. This, in the absence of a better explanation, must be due to the omission of young children from the returns.

It may be observed that the population always comprises a large number of newly born infants whose births have not yet been registered, and that this number may easily have been as high as 50,000 or 60,000 at the date of the census. The attitude of mind which regards such children as not having been placed upon the official roll and not subject, therefore, to the census procedure is an intelligible one and may have been partly responsible for the loss at age 0, though not at age 1.

A contributory cause may also exist in respect of children born within a short time of the marriage of their parents, in whose case some reluctance may have been felt to throwing the fact into relief by a statement of their true ages. The influence of this would be felt mainly while the marriage was comparatively fresh, and might, therefore, have contributed to some suppression of the record of recently born children and possibly to some understatement of the numbers aged 1, not by entire omission in the latter case, but by their record at the younger age.

On the whole the numbers obtained by tracing the survivors from the appropriate births are probably to be preferred to the numbers enumerated at the youngest ages, and the deficiency in the latter attributed either to the statement of age next birthday in a number of cases, or to more indefinite causes of which the most likely appears to be the definite omission of children of the youngest ages from the returns.

Above age 9 the comparison is made in quinquennial groups of age, the 1921 census population, carried to 30th June, 1921, being compared with survivors from the numbers recorded at an age 10 years younger at the date of the 1911 census (2nd April, 1911) after allowing for civilian deaths and civilian migration. The record of non-civilian movement (army, navy, and air force) during the latter part of the decennium is unfortunately not available in sufficient detail to permit of its being included in the analysis as a recognised intercensal movement, and the differences between the survivors and enumerated at the male military ages must rather be regarded as an expression of the omitted movements themselves. These are so large as to conceal effectively any irregularities in age statement which might otherwise have been observed.

But, apart from this group of ages, the identification of the sources of the differences between the survivors and enumerated at ages above 9 becomes very involved, owing to the fact that survivors are traced from the enumerated in 1911 and that any mis-statements in the latter will be passed on to recur with an opposite sign at an age ten years older in 1921. So that the apparent discrepancy in 1921 may be compounded of errors in both 1911 and 1921 census figures — errors unrelated to one another — and in addition it will also include any error contained in the estimated intercensal movement used in tracing the survivors from 1911. This movement, it may be noted from the following table, is small in relation to either of the corresponding census totals until the later ages are reached and the errors therein would have to be of a very large order in the earlier years to account for any considerable proportion of the resultant census discrepancy.

TABLE XXXVIII.—COMPARISON BY AGE GROUPS OF ENUMERATED POPULATION IN 1921, WITH THE NUMBERS EXPECTED BY SURVIVORSHIP FROM THE 1911 ENUMERATED POPULATION AFTER ALLOWING FOR THE BIRTHS, CIVILIAN DEATHS AND CIVILIAN MIGRATION OF THE INTERVENING PERIOD. (ALL FIGURES IN THOUSANDS.)

It will be observed from this table that the discrepancies between enumerated and survivors are roughly similar for each sex at ages 9-13 and from 59 onwards (ages as in 1921)

The apparent excess in the enumerated at the age groups 9-13 has already been referred to and, as it would be converted into a much smaller deficiency if the survivors in 1921 were traced from the survivors instead of from the enumerated in 1911, the bulk of the differences can probably be ascribed to the 1911 population which formed the basis of the survivorship calculation rather than to an inherent defect in the 1921 figures. After age 14, the female table exhibits a very marked excess hi the enumerated at ages 19 to 28, accompanied by deficiencies both above and below, the loss being very pronounced at the period 34 to 38. These discrepancies might be accounted for if there had been any considerable migration during the decennium for which insufficient allowance had been made in the survivorship calculation. This, as a matter of fact, is possibly responsible for a portion of the excesses, for, in the Board of Trade age statistics of migrants, movements within the United Kingdom and between this country and Europe are not included, and examination of the Scottish census returns (no census was taken in Ireland in 1921) suggests that there is a depression in the number of females between 20 and 29 which may quite well have been occasioned by transfer to this side of the border. But even allowing that some immigration of this nature occurs both from Scotland and other countries outside the scope of the official migration returns, it is not judged nearly large enough to account for the whole of the excess disclosed in the table, nor is there evidence of the contrary movement, which, if mis-statements of age were rejected, would be involved by the reversal of the discrepancy, especially at ages 34.38. It is therefore difficult to escape the conclusion that the experience of earlier censuses has been repeated in 1921 in the deliberate mis-statement of age by a large number of females.

The age period 19-28, and especially its latter half, appears to possess an attraction to those immediately younger as well as to older sections, for the deficiency in the enumerated at ages 14-18, though not relatively large, appears to be well marked and is more probably due to transfers to the succeeding ages in 1921 than to overstatements at ages 4-8 in 1911 or to defects in the intercensal records of deaths and migration, which are comparatively small at this period of life. And it is by no means clear that the overstatements are limited to the 19-28 period. The small apparent excess of survivors at the next group (29-33) is not inconsistent with a real excess in the enumerated there, if, as is not improbable, such excess is over-weighted by a larger overstatement of survivors arising from an inflated 19-23 group in 1911.

The similarity of the deficiency in the enumerated at ages 34-38 to the excess in the 24-28 group might at first suggest that the bulk of the mis-statements of age was attributable to females approaching 40, understating their true age by 10 years or so, but, oh further consideration, it seems much more likely that the understatements cover a more extended age field and that the 34-38 deficiency in 1921 is due in the main to an excess in the expected survivors, deduced as they are from the enumerated at ages 24-28 in 1911, in which age group there is reason to suppose that a large overstatement occurred as it has in 1921.

Above the age of 59 the discrepancies in the female table parallel those in the male and suggest that each is subject to a common type of misstatement at this period of life. The fact that the enumerated, hitherto less than the estimate of survivors, becomes in excess after age 64 has been interpreted, as indicating that a number of persons of both sexes tend hereabouts to give their ages as somewhat in excess of their true value leading to an overstatement of population. The inference is not conclusive, however, for if it be assumed that both the 1911 and 1921 populations are affected by the same kind of error an overstatement of the enumerated aged 64-69 in 1911 would by itself lead to an overstatement of survivors at 74-79 in 1921, whereas, in fact, it is the enumerated which are in excess at this point, a condition which could only arise if the overstatement at 74-79 was in amount very much larger than the overstatement at 64-69 in 1911. While the discrepancies from ages 64 upwards could be attributed to a series of overstatements of population at each census, progressing with advancing age right up to the oldest ages in the table, they could equally be ascribed to a series of understatements diminishing with advancing age and terminating at about age 70 and since the period from 70 onwards is that covered by the Old Age Pension Scheme under which grantees are required to produce evidence of their ages, an explanation which is consistent with the assumption that the population at ages over 70 is approximately correct appears to be preferable to one which assumes that very large overstatements have occurred. At the same time the intercensal deaths become an increasingly important factor in the calculation of the survivors at the old ages, and the possibility of considerable errors in the ages returned in the death Registers in respect of old people must not be disregarded in an attempt to account for the discrepancies between enumerated and survivors. But, if it is not possible to apportion the errors with any finality, it can be definitely stated that they are smaller in amount than those hitherto recorded, and that the improvement in the accuracy of age statements shown in the 1901 Census report to have been continuous over the latter half of the last century has been well maintained up to the present time.

3.—Marital Condition.

Of the 37,886,699 persons enumerated in England and Wales in 1921, 20,540,890 were returned as single, 15,065,058 as married, 2,264,069 as widowed, and 16,682 as divorced. Among the single of all ages there were 1,065 females to 1,000 males; the number of wives enumerated in the country exceeded the number of husbands by 114,956 and were in the proportion of 1,015 wives to 1,000 husbands the number of widows was 1,621,758, as compared with 642,311 widowers; the divorced, so returned, which, according to the instructions governing the returns, should have included both parties to each divorce, provided, of course, that neither had remarried, were approximately evenly divided between the sexes, there being 8,464 males and 8,218 females. In 1911 the ratio of single females to single males of all ages was 1,029 to 1,000 the wives exceeded the husbands by 134,498 and were in the proportion of 1,021 to 1,000 and the widows numbered 1,364,804 against 615,811 widowers. Divorced persons were not identified in 1911.

In the following table the numbers returned under each condition in 1921 are set out in age groups and the intercensal increases or decreases in the several groups show where the principal changes in the distribution have occurred.

TABLE XXXIX.—AGE, SEX AND MARITAL CONDITION DISTRIBUTION, 1921 .

Apart from the features of the past decennium to which reference has already been made (viz., the heavy decline in the numbers of births and the male war losses, both of which have registered their own peculiar influence on the condition proportions), a further factor enters into this aspect of the census classification in the change which has taken place in the marriage rates. The pre-war rates of marriage, within the intercensal period, were slightly higher than those recorded at the end of the preceding period, and these were followed successively by a sudden rise in 1915, a depression which took them just below the pre-war rates in the years 1916-1918, and a subsequent rise thereafter to a point higher than any previously recorded. The average annual rate over the 10 years 1911-1920 was in consequence higher than any similar average for several decades. In addition, although the average age at marriage has increased owing to the general ageing of the population, the increase in the marriage frequency in relation to the marriageable population has been most marked at the younger ages, so that the considerable decline recorded at past censuses in the proportion married, at the earlier adult ages has now given place to an equally notable rise.

In regard to the single, both males and females show actual decreases in population as compared with 1911. The reduction in the number of births which is common to both sexes is naturally limited to this section and is shown in the decline at ages under 15. The loss to the single owing to the increase in the marriage rates is also common to both, though it can only be inferred from the female table by the reductions at ages between 20 and 35, the corresponding male populations at these ages being those in which the incidence of war losses was heaviest.

The married sections both show large intercensal increments. Numerically the additions are greatest at ages over 45, as they are in the case of the single, but at most of the earlier ages an increase has taken place showing that the new marriages in the case of males have generally been more than sufficient to counteract the war losses in this class.

These war losses are more directly reflected in the increase in the numbers of widows, in respect of whom the percentage additions at ages under 35 are the heaviest recorded in the table. Large numerical increases, however, are also shown at the later ages where the rather greater age of the husband and the lighter mortality experienced by females are jointly responsible for the more frequent termination of married life by the decease of the husband than by that of the wife.

As a result of the changes in the condition proportions, the incidence of the sex proportions in what may be termed the marriageable population—here regarded as the single, widowed and divorced sections at ages 20 and over—is now very different, especially at the earlier adult ages, from what it was in 1911. The female excess in this class has increased from 1,046,077 to 1,667,846 in the ten years, a rise of about 60 per cent. From the subjoined analysis it will be observed that the increase is common to all the age periods identified. The heaviest increments are located at the earliest years where the marriage tendency may be said to be at its highest and where therefore the reduced opportunity for marriage in the case of a large number of women will be felt in an extreme degree. Thus, at ages 20-24 and 25-29, there was an actual deficiency of females in 1911 represented by proportions of 984 and 990 females per 1,000 males which have now been converted into surpluses of 1,043 and 1,154 respectively; at ages 30.34 and 35-44 excesses already considerable in 1911 (1,111 and 1,366 per 1,000 males) have been greatly increased and now amount to 1,470 and 1,683 females per 1,000 males.

TABLE XL.—EXCESS OF FEMALES IN THE MARRIAGEABLE POPULATION (i.e., SINGLE AND WIDOWED AND DIVORCED), AGED 20 YEARS AND OVER .

It should perhaps be pointed out that the contrasts in this form are between men and women of similar age. Husbands, however, are generally older than their wives (see page 174 dealing with the relative ages of husbands and wives) and this should be taken into account in a strict examination into the effect of the changes of sex ratio upon the marriage conditions of the future.

The following summary, extracted from Table 36 of the General Tables Volume, provides a general survey of the incidence at various groups of ages of the several marital conditions in the total population of the country at each census back to 1881, and shows also how the 1921 population is distributed in the same respect throughout the urban and rural districts and the geographical sections referred to in earlier portions of this report.

TABLE XLI.—MARITAL CONDITIONS: PROPORTIONS BY SEXES IN AGE GROUPS. ENGLAND AND WALES (1881-1921), LARGE AGGREGATES, AND GEOGRAPHICAL DIVISIONS (1921) .

At the 1921 census an attempt was made for the first time in this country to ascertain the number of divorced persons in the population, that is to say, the number of persons in the enumerated population whose marriage had been terminated in this manner and who had not subsequently remarried up to the date of the census.

The term divorced was intended to apply to each of the parties to a separation irrespectively of whether he or she had been the applicant or respondent in the divorce proceedings, and instructions were given to this effect at the time of the enumeration. The total number returned in this category amounted to 16,682 in all, of which 8,464 were males and 8,218 females, considerably more than half of each of them being within the age period 25-44. It is greatly to be feared/however, that doubts as to the value of such returns/which were felt and expressed when it was first decided to include the inquiry in the general census questionnaire, have proved only too well founded, for from an examination of the records of the divorces which have been granted year by year, after making full allowance for reductions in the numbers by mortality and by a very high remarriage rate, the expected numbers might well be put at a figure twice as large as the total recorded above, and it appears more than probable therefore that a large number of persons failed to return the desired information. In view of the probably misleading character of the enumerated figures, the divorced have been retained as a separate class only in the section of the tables devoted to age and marital conditions in all other sections they have been included with the widowed, with whom, from a sociological point of view they may usually be associated.


1 Mis-statement of Age in the Returns of the Census of Scotland by J. C. Dunlop, M.D., F.R.C.P., Edinburgh.—Journal of the Royal Statistical Society , 1923.

Next Selection Previous Selection