Sexes, Ages and Marital Conditions

Next Selection Previous Selection

PART IV — SEXES, AGES AND MARITAL CONDITIONS


1. Sexes.


Of the 39,952,377 persons enumerated in England and Wales in 1931, 19,133,010 were males and 20,819,367 females. The excess of females thus numbers 1,686,357 which amounts to 8.1 per cent, of the female total, or is represented by the ratio of 1,088 females to each 1,000 males. The female excess may be said to be slightly exaggerated in that the comparison takes no account of the officers and men of the defence forces temporarily absent at overseas stations or of members of the mercantile marine and fishermen absent at sea on the night of the Census. From the appendices in the 1931 Census General Tables Volume, the absent males in these classes are assessed at some 191 thousands in respect of Great Britain as a whole and the bulk of them must be assignable to England and Wales; the numbers are not, however, exceptional and do not materially affect comparisons with the position at previous enumerations in this country.

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

Year Females per 1,000 males
1801 1,057
1811 1,054
1821 1,036
1831 1,040
1841 1,046
1851 1,042
1861 1,053
1871 1,054
1881 1,055
1891 1,064
1901 1,068
1911 1,068
1921 1,096
1931 1,088

Of these, the female preponderance is probably overstated in respect of the years 1801, 1811 and again in 1901 owing to the forces serving outside the country in connection with the military campaigns associated with those periods. Apart from these years there was an almost continuous increase in the female excess up to 1911 with a final jump from that point to 1921 which was nearly as great as that of the whole preceding century and which was a direct consequence of the unprecedented loss of men caused by the World War of 1914.18. The 1921 maximum has since been abated slightly owing to the high masculinity of the births of the 1921—31 decennium, but with the sole exception of 1921 the 1931 preponderance of females is outstandingly higher than that of any other period exhibited in the series.

The excess does not, of course, affect all sections of the population alike and a fuller aspect of its incidence and of its predisposing factors will be seen in the comparative population analyses according to age periods and according to geographical situation which are provided in Tables 17.27 of the General Tables Volume. The age proportions for England and Wales as a whole are set out in the following table for the years 1911, 1921 and 1931, while a final column shews the actual numerical excess of females in each age group in 1931.

TABLE XXXVII.— EXCESS OF FEMALES BY AGE GROUPS, ENGLAND AND WALES

The 1911 record broadly illustrates the age aspects of the sex distribution to which the country had become accustomed prior to the war disturbance of 1914—18. Commencing with a small male excess at the youngest ages from the masculinity imparted at birth, equality was very soon produced by the more favourable mortality experienced by females and thereupon converted to an excess of females which grew almost continuously in steadily ascending proportions over successive age periods of life up to the final age group shewn in the table at which females were almost twice as numerous as their male counterparts. The increase in proportions was accelerated at the early adult ages as the result of emigration which operated more amongst males than females and the only significant exception to the continuity of the increase occurred as a result of the complementary element of immigration which had the effect of restoring some of the male deficiency at the middle adult ages.

Since 1911 the record has been influenced by three significant factors, two of them directly associated with the war and to that extent of an irregular nature though their effect is likely to be visible in the population curve until the whole of the war generation has passed away. In the first place the exceptional number of war deaths which were located at the limited section of active service ages resulted in the abnormal augmentation of the female excess at ages 20.45 in 1921 and again in the ten year older group, 30.55, in 1931. The second factor also associated though in an imperfectly understood manner with conditions engendered by the war was the curious disturbance to the previously steady masculinity of the birth record. From a pre-war average of less than 1,040 males per 1,000 females born, the proportion rose in 1919 to 1,060 and though, in immediately succeeding years, there seemed to be a prospect of its early return to former levels, the tendency was reversed about 1926 and the figure continues to be high in relation to pre-war experience. The result of this is that the male excess at the youngest ages in the table is greater in amount and extends to a considerably higher age than it did formerly and the transposition has been further influenced by a third factor in the shape of the very considerable diminution that has taken place in infant and child mortality.

Males are now in excess of females at ages up to 15 by 97 thousands as compared with 70 thousands in 1921 the change being the result of simultaneous reductions of 477 and 504 thousands in the absolute numbers of males and females respectively during the ten years.

One of the most significant changes now shewn is in respect of the ages 20.30, the period at which the bulk of marriages occur. The males at these ages have increased from 2,788 thousands in 1921 to 3,328 thousands in 1931, pari passu with a female increase from 3,323 thousands to 3,523 thousands by which the female excess, which was more than half a million ten years ago has been reduced to 195 thousands in 1931; proportionately, the excess represented by 1,059 females per 1,000 males is now barely half what it was prior to the war.

In the wider age group 20.65 covering the mass of those which form the economically productive element of the population it will be seen that while the males have increased in the ten years from 10,082 thousands to 11,342 thousands (i.e. by 12.5 per cent) the females have increased from 11,510 thousands to 12,692 thousands (i.e. by 10.3 per cent.), the preponderance of the latter being expressed by the proportions of 1,119 females per 1,000 males in 1931 as compared with the higher figure of 1,142 per thousand in 1921.

The mortality of females is generally lighter than that of males but it is of less significance at the younger ages where the mortality of both sexes is low. It exhibits the effect of its differentiating influence most prominently at the more advanced ages where there is a rapid acceleration in the female preponderance, the proportions rising rapidly after age 70 and reaching the ratio of more than two females to one male in the final age group shewn in the, table, the proportion here being even higher than it was in 1921.


2. Regional and County Variations in Sex Proportions.


Sex distribution differs widely in different parts of the country. This, however, is not a novel feature and the relative positions of the several types of area in this respect have not been greatly affected by the change in proportions between 1921 and 1931. The general features of the distribution will be seen in the following arrangement by regions and by density aggregates.

Females per 1,000 Males

  1931 1921 1911
 
England and Wales 1088 1096 1068
Density Aggregates:  
  Greater London 1140 1159 1129
  County Boroughs * 1108 1106 0
  Other Urban Areas * 1082 1093 0
  Rural Areas * 1015 1024 0
 
Geographical Regions: ?  
  S.E. South East 1127 1146 1106
  N.1. Northumberland and Durham 1015 1009 993
  N.2. Northern Rural Belt 1048 1062 1035
  N.3. West Riding 1076 1082 1057
  N.4. Lancs. and Cheshire 1115 1112 1088
  M.1. West Midlands 1074 1081 1062
  M.2. East Midlands 1058 1068 1043
  E. Eastern Counties 1039 1062 1035
  S.W. South West Counties 1108 1129 1099
  W.1. South Wales 982 964 929
  W.2. North and Central Wales 1069 1095 1064

* Excluding Greater London

? For constitution of regions see Appendix B

The influence of social and industrial conditions upon the proportion is visible from the tendencies displayed in this arrangement. Personal service occupations and occupations of sedentary clerical type, recruited by women in large numbers, find their market to a large extent in the towns and the proportion definitely increases with increased urbanization. In Greater London, the most congested of the density aggregates identified, the proportion of 1,140 is the highest shewn for 1931 in the table and this figure advances beyond 1,500 in the wealthier metropolitan boroughs as a result largely of the domestic servants employed there. In the county boroughs as a whole the incidence of females is well above and in the smaller towns just below the national proportion; at the other extreme stands the more sparsely populated rural areas with their heavier demands for physical labour, where the female preponderance is low notwithstanding the countering influence proceeding from the fact that the populations are subject to a greater weight of population at the older ages where the proportion tends to be increasingly in favour of the female sex.

The influence of physically exacting occupations is again seen in the low proportions recorded in the region associated with coal mining and other heavy industries, e.g., Northumberland and Durham (N.I.) and in South Wales (W.I.); contrary to the experience of the country as a whole, the proportions here have increased during the past ten years, a direct consequence, no doubt, of the prevailing depression, the fall in the demand for labour bearing most severely on the male element of the population. In Lancashire and Cheshire (N.4-) the proportion is high as might be expected from the concentration there of the textile industries which above all others have come to be associated with the employment of women. For this reason it is perhaps surprising that the proportion already high in 1921 should have risen still further in 1931 in view of the continued depression in the textile trade.

In regions of more rural type, the Northern Rural Belt (N.2.), Eastern Counties (E.) and North and Central Wales (W.2.), female excess is under average as a rule but the proportions are not so low as in the purely rural districts owing to the presence within the regions of seaside and health resorts which offer particular attractions to the retired and unoccupied classes (preponderantly female), an influence which is probably responsible for the high proportion recorded in the South West Counties (S.W.).

The effect of the several tendencies are more clearly brought out in Table XXXVIII in which metropolitan boroughs, county boroughs and administrative counties with the highest and lowest proportions are set out in order of their female predominance.

TABLE XXXVIII.— HIGHEST AND LOWEST PROPORTIONS OF FEMALES PER 1,000 MALES IM METROPOLITAN BOROUGHS, COUNTY BOROUGHS AND ADMINISTRATIVE COUNTIES


3. Sex Proportions in Other Countries.


A similar series of comparisons with other countries for which recent Census records are available, is given in Table XXXIX.

TABLE XXXIX.— COUNTRIES ARRANGED IN ORDER OF 1931 SEX PROPORTIONS, SHOWING ALSO 1921 COMPARISONS


4. Ages.


Since 1911, when the introduction of mechanical sorting and counting devices superseded the more laborious hand tabulation processes which had been in use up to and including the Census of 1901, any physical obstacle there may have been to the presentation of the analyses of the householders' returns in the utmost desirable degree of detail has largely disappeared. The high speeds with which modern machines can handle the carded records and the reduction of the mental fatigue imposed upon the operator have enormously extended the potential field of tabulation with the advantage that the main consideration governing the design of the tabulation programme is not so much what or how much can be done, but how best shall the results be presented from the point of view both of degree of detail and complexity of combination in order to provide information of maximum value and practical service. On this occasion, however, current economic conditions imposed a certain measure of restriction upon the whole scale of the Census undertaking of 1931 and regard for this over-riding condition necessarily influenced the tabulation processes.

The published age statistics in respect of the 1931 Census are not quite so extensive as those of 1921 though it is believed that the slight curtailment has introduced no loss of any practical significance. Populations classified by individual years of age have been published in Table 17 of the General Tables Volume for the country as a whole, for national density aggregates of areas and for each of the twelve regions adopted for the display of general geographical features. For all lesser areas from county aggregates to individual urban and rural districts, individual age details are restricted to ages below 21 which are published in Table 16 of the County volume, with a summary in the conventional quinquennial groups of age (0.5, 5.10, etc.) covering the whole of life in Tables 14 and 15 of the same volumes.

Apart from the fact that the local area is of less relative importance than the country as a whole and to that extent is appropriately served with a less extensive type of record, a definite disability attaches to the age record derived from Census returns in that it is a record of statements rather than of facts, that is to say, that the age statements are subject to error, either wilful or accidental, to such an extent as to deprive the individual year tabulation of a great deal of its value as an exact measure of the age incidence of the population. The experience of earlier Censuses has been repeated in 1931 and as is described later on in the section in which the types of error are examined in greater detail or as may be observed from the graphs in which the record is shewn in pictorial form, there will be seen to be an unnatural heaping up of the numbers at certain ages, more particularly those ending in the digit o, e.g., 30, 40, 50, etc., from which it can only be inferred that over certain portions of the age field many of the ages have been returned in an approximate form. This preference for certain ages is experienced in all sections of the country and since the form of distortion it introduces is minimised if not eliminated by the aggregation of the individual years records into groups of age, the aggregation of the figures over all adult ages in which the error is most prominent may be regarded as an essential prerequisite to the use of the material either for administrative purposes or for purely statistical comparisons. And apart from the greater reliability of the records in this condensed form the occasions for which individual age records of the adult population may be required are comparatively rare and for all practical purposes the form of the published record will be adequate to the needs.

In the case of the infant and juvenile sections of the population the matter is rather different. Though this section is not wholly free from the type of error which affects the older ages, it is comparatively slight, the ages of children being stated on the whole far more accurately than those of adults. Further, the progression from age o to age 16 is subject to a real irregularity arising directly from the abnormal variations in the number of births during the years of the war and post-war periods and to telescope the records in this section of the age field would result in suppressing the inherent features of the curve, for which there would be no justification. Moreover, in their use and application the age record in the case of the young differs from that of the older sections in that it is here that individual years become critical for the many purposes such as infant and children's welfare schemes, elementary and higher education, legislative enactments relating to health and unemployment insurance, juvenile crime, employment conditions, etc., or all of which the facts derived from the Census Record are of direct significance.

For these reasons the standard age distribution adopted for all areas of the country is one by individual years of age up to 21 and quinquennial groups (20-, 25-, etc.) thereafter up to a final group of 95 and over. The extended individual age tabulation shewn for the whole of life in respect of England and Wales as a whole and the large aggregates is mainly of statistical interest in providing the material for detecting and measuring the mis-statement tendencies and comparing them with the similar experiences of 1921 and 1911.

The form of the question in which the age was asked for at the 1931 Census was precisely similar to that used in 1921. It required that each person's age should be stated in years and months and was thus rather more precise than that adopted prior to 1921 when the statement was limited to years of age only except for children under i. Neither in 1921 nor in 1931 was there any intention of tabulating by months of age—such detail would obviously be valueless having regard to the known tendencies in the matter of inaccurate age statement. The sole object was that of influencing the householder to give a critical instead of an approximate answer so as to avoid, or at any rate reduce, the looseness of statement experienced at preceding enumerations. whether due to this device or to a growing appreciation of the value of these periodical counts, the records are steadily improving but it will obviously be a considerable time yet before the returns will be acceptable at their face value, notwithstanding the fact that the generations born prior to the introduction of compulsory birth registration in 1837 have all but passed away.

Reference must also be made to a small innovation introduced at the 1931 Census not specifically dealing with age, but of direct interest to the age analysis in the form of a general instruction inserted in the Census schedule to the effect that newly-born infants who had not been given a name should be described as "Baby". Examination of the age returns at previous Censuses has always revealed an apparent shortage in the returns of children under one year of age and it had been surmised that since the time lag between birth and registration of birth results in the existence at any time of many thousands of children whose births have not yet been registered, many of these children may possibly have deliberately been omitted in the past from the mistaken notion that they had not been given a name or formally placed upon the official roll. The effect of the special instruction was visible in the returns, for the description "Baby" was inserted on many schedules, but how far the numbers represent infants who would otherwise have been omitted altogether it is not possible to ascertain by any form of measurement; it can be said, however, that the shortage at age o in 1931 is much less than any previously noted, in fact the deficiency is now hardly greater in degree than the error attaching to other ages, so that it seems reasonable to assume that the view regarding the discrepancy in 1921 and earlier Censuses was not altogether unfounded.

The age distribution of the population enumerated in 1931, which may conveniently be visualized from its pictorial representation shewn in Diagrams E, F and G, covers practically 100 years of life and its shape has been determined by and, in fact, represents the resultant product of the combined forces of birth, death and migration which have operated over the whole of the past century. It is not possible to examine the course of these contributing forces in any detail here but attention may be called to some of their outstanding features. In regard to the originating element of birth it may be noted that the oldest members of the 1931 population were born prior to the year (1837) when birth registration was made compulsory in this country; it may reasonably be inferred, however, that the annual numbers of births at that time was in the region of 500 thousands a year from which they grew rapidly to about 900 thousands in the decade 1881-1890 and after a somewhat slower ascent to a maximum of nearly 950 thousands, which was reached a few years after the turn of the century, commenced the decline that, but for some abnormal fluctuation over the war and immediately post-war period, has been virtually continuous and has reduced the annual numbers of births to less than 650 thousands at the end of the 1921-1930 decade. That means, of course, that the population now at the earlier adult years of life is abnormally weighted in that it consists of survivors of births which were 40 to 80 per cent, more numerous than those from which the youngest and oldest sections are sprung, and that to that] extent the age curve must be regarded as an exceptional one in that it is quite different from the position as it was 50 years ago or as it will be 50 years hence or from what it would be in a stationary population.

DIAGRAM E — ENGLAND AND WALES CENSUS 1931, POPULATION BY SINGLE YEARS OF AGE DISTINGUISHING MALES AND FEMALES. DIAGRAM E

DIAGRAM F — ENGLAND AND WALES CENSUS 1931, GRADUATED POPULATIONS AT SINGLE YEARS OF AGE DISTINGUISHING MALES AND FEMALES

DIAGRAM F

DIAGRAM G — ENGLAND AND WALES, CENSUS 1931-AGE DISTRIBUTION (PERSONS)

DIAGRAM G

Variations in the death rate over the period in question have also had an important, if secondary, share in the determination of the numbers existing at different ages today. Mortality, judged by the standardized basis of comparison adopted in the Registrar General's Annual Statistical Review, may be said to have been approximately halved during the past 70 years, But the decline has not been evenly distributed over the whole of life; the bulk of the saving is located in the early years and it may be noted that in the age group 5.10, deaths are occurring today at about one-quarter the frequency they had at the middle of last century while for both younger and immediately older ages the reductions are hardly less striking. With advancing age, however, the fall becomes progressively less significant both in regard to its absolute amount and to the period over which it has operated, until at the oldest ages there has been comparatively-little movement at all. These survivorship changes have tended to favour the later born and therefore the younger sections of the community, and to reduce the population disparity originally imparted by the birth rate between the young and middle aged and to aggravate it between the middle aged and elderly sections of the population.

In addition to the natural forces of birth and death there is the factor representing the miscellaneous movements, usually summed up in the term migration, which occur chiefly during the youthful and early adult age periods. The separate inward and outward components of this factor are not inconsiderable in the aggregate, but on balance its effect and its variations from time to time are small in relation to the births or deaths and being generally outward in character it has tended to reduce the population slightly without materially affecting its age incidence.

Turning to the more detailed features of the 1931 distribution it will be seen from Diagram G that at the youngest ages the population increases with advancing age and is thus in contrast with the position in the adult section where, apart from local irregularities, the numbers continuously decline throughout life. The maximum number reached at any single age is that of 775 thousands enumerated at age 10 but its significance is discounted by the exceptional irregularity in the curve at this period of the age field, a 30 per cent, fall in the succeeding three years reducing the numbers to 526 thousands at age 13 from which an equally abrupt rise restores the numbers to the more stable conditions commencing with 702 thousands at age 16. The population curve at the youngest ages, before the effect of migration begins to make itself felt, is almost precisely conditioned by the course of the successive births from which the survivors are derived. The births in question are shewn as a subsidiary curve in Diagram G and it will be seen that the contours of the birth curve and population curve faithfully correspond with one another at appropriately increasing survivorship distances. The abnormality in the population progression between ages 10 and 16 is thus seen to be wholly attributable to the birth rate fluctuations o the war period and must remain a permanent and inherent feature of the population reappearing at each future Census at appropriately older ages until the whole of the generation has passec away. It occurred at ages under 6 in 1921 and to its transposition from the youngest ages at that Census to the more advanced position in 1931 as well as to the decline in births during the past 10 years are due the reductions of 332 thousands (10.0 per cent.) at ages 0.5, 196 thousands (5.6 per cent.) at ages 5.10 and 453 thousands (12.4 per cent.), at ages 10.15 shewn by the 1931 population in comparison with that of 1921. The figures are shewn in Table XL in which the absolute numbers are also expressed as proportions of the total population and it will be observed that whereas the juvenile population at ages under 15 represented 27.7 per cent, of the whole 10 years ago, the figure has now been reduced to 23.8 per cent.

TABLE XL.—POPULATION OF ENGLAND AND WALES IN AHE GROUPS, 1931 AMD 1931, AND ITS DISTRIBUTION PER 10,000

At the age group 15—20 which covers the years of entry into employment the 3,435 thousands enumerated in 1931 are but slightly fewer than the corresponding numbers of 1921 but allowing for the general increase in the population as a whole the proportion has been reduced from 9 per cent, to 8.6 per cent.

After age 20 the individual age records lose most of the significance they might otherwise possess by reason of the saw toothed irregularities which appear in the curves from here onwards throughout the major portion of life and which are largely attributable to the curious habits individuals exhibit in their statements of age. It will usually be found preferable in examining the remaining sections of the age field to have regard to the graduated record or to quinquennial or decennial groups of ages in which the local disturbances tend to be suppressed by mutual counteraction (Diagrams H. and J.).

DIAGRAM H — ENGLAND AND WALES POPULATION BY QUINQUENNIAL GROUPS OF AGE, 1891-1931. MALES
ENGLAND AND WALES POPULATION BY QUINQUENNIAL  GROUPS OF AGE, 1891-1931. MALES

DIAGRAM J — ENGLAND AND WALES, POPULATION BY QUINQUENNIAL GROUPS OF AGE, 1891-1931. FEMALES
ENGLAND AND WALES POPULATION BY QUINQUENNIAL  GROUPS OF AGE, 1891-1931. FEMALES

Generally speaking, the maximum weight of population in 1931 is to be found in the age poup 20.25. The numbers rise slightly within that group for the first few years but thereafter they commence the decline which, but for the irregularities referred to, is continuous over the remainder of the age field. The outstanding feature of the adult age distribution is the marked depression in the male section between the ages of 30 and 50 giving the impression of a bite :taken out of the curve between these age points (Diagram H.). It is directly attributable to the exceptional loss of men during the war and is a permanent scar which, like the irregularities between 10 and 16, must disfigure the population curve, though at progressively advancing ages for many years to come. The displacement of this depression from its position at ages 10 years younger in 1921 is responsible for most of the disparity between the male and female intercensal movements at the earlier adult ages. Thus, in the age period 20—30, the males have increased by as much as 19.4 per cent, as compared with 6.0 per cent, only in the case of females. Between 30 and 40 the rates of growth are not significantly dissimilar at 6.3 per cent, males and 5.0 per cent, females, but in the two following age decades a complementary tendency is exhibited, males advancing at rates of 1.3 (40.50) per cent, and 20.0 per cent. (50—60) simultaneously with the larger female increments of 6.9 per cent, and 24.0 per cent.

The proportion of the population enumerated in the portion of the age field between 20 and 50 is slightly larger at 44.9 per cent, in 1931 than it was in 1921 (44.0 per cent.) the increase being located mainly in the male section where the numbers have advanced by 9.5 per cent, as compared with 5.9 per cent, in the case of females.

The growth at these ages is however completely overshadowed by that which has occurred at ages over 50. Between 50 and 60 the increase is of the order of 20 per cent, or more and is still greater in the next higher age groups 60.70 and 70.80 where it is nearer 30 per cent, and even reaches 32.1 per cent, in the male group 70.80. The effect of these high rates is discounted by the smaller weight of population in the later years of life but they have resulted nevertheless in raising the proportion of people aged over 50 from 19.0 to 22.7 per cent, between 1921 and 1931.

One hundred and fifty-six persons returned ages of 100 or more; 27 of the total were males and 129 were females, a weight and distribution which may be compared with 30 males and 80 females in 1921 or 36 and 92 in 1911. The general scale of increase in the numbers of centenarians from a total of no in 1921 to 156 in 1931 is not out of character with that which might have been expected having regard to the general shifting of weight from the younger to the older ages; at the same time the irregularity shewn in the separate sex changes and experience generally of past Censuses suggests that there is a measure of unreliability in the statements at these advanced ages which is sufficient to obscure the real movement at this period of life.

In spite of the local efforts made at the enumeration itself to secure complete and intelligible returns, there remained about 10,000 persons in respect of whom the statement of age was either omitted altogether from the Census schedules or given in a form too indefinite for immediate classification. The statistical significance of these numbers, representing but a fraction per 1,000 of the total population, was not considered to be sufficient to justify the complication which would have been involved in their separate identification and record throughout the numerous classifications and sub-classifications in which the age incidence is featured and the procedure followed in the similar circumstances of previous Censuses has again been adopted, ages being assigned in accordance with other information on the schedule so far as was 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 further ageing of the population which continues the gradual transformation that has been taking place since about 1881 coincidently with the fall in the birth rate which dates approximately from that time. This may be illustrated by the following statement which sets out the average age of the population at each Census from 1881 to 1931 inclusive.

AVERAGE AGE

  Persons Males Females
1881 26.2 25.7 26.7
1891 26.6 26.11 27.1
1901 27.4 26.9 27.9
1911 28.6 28 29.1
1921 30.6 29.9 31.2
1931 32.6 31.8 33.4

It would appear from the tread ol the foregoing series ot average ages that the rate of ageing, or increase in the average age is accelerating rather than diminishing and that higher and increasing averages are to be expected at successive future Censuses. That may be said to be inevitable from the structure of the population in which early adult sections are abnormally weighted by reason of the fact that they are the survivors of births greatly in excess of those from which the younger or older sections of the population originated; the moving forward of the weighted section with the progress of time has steadily increased the average age and will continue to do so until it reaches the older ages where the higher mortality will reduce its prominence and so weaken its influence.

TABLE XLI.— POPULATION IN AGE GROUPS AND DISTRIBUTION PER 1,000 TOTAL POPULATION AT SUCCESSIVE CENSUSES, ENGLAND AND WALES, 1881-1931

The general significance of the age trend disclosed by the Census records of the past 50 years will be better seen by the analyses in Table XLI; the upper half showing the actual numbers of population at each Census within a series of comparable age groups and the lower expressing the several numbers in each group as proportions of the total national population involved in each period.

In the employment or productive age field, conventionally taken as between ages 15 and 65 the proportion for both sexes combined which stands at 688 per 1,000 total population is higher than the corresponding proportion in 1921 (663) thus maintaining the rise which has been continuous since 1881, the first year shown in the table. At the same time as an index of production capacity the effect of the total proportion will be governed by its internal age structure and the increase in the total is to that extent partially offset by the ageing which has taken place within the group. This is noticeable in the comparison with 1901 or 1911 since when the proportions in the younger and more active ages 15-39 have declined at the expense of a considerable rise in the older half of the group; on the other hand, the position is no worse and is in some respects better than it was 10 years ago since the proportion of men 15.39 (194) has improved from that of 1921 (187) which had been abnormally depressed as a result of war casualties and thus more than compensates for the reduction in the female proportion at these ages. In the complementary dependant age fields the proportion in the under 15 section shows a substantial fall from 277 in 1921 to 238 per 1,000 in 1931 which is directly due to the fall in the birth rate and thus continues the marked decline observable since 1881; at advanced ages over 65, the rise in the proportion which commenced after 1901 has steepened from 60 in 1921 to 74 in 1931 and will continue to grow as the high numbers at the middle years of life resulting from the maximal births at the beginning of the century pass into the higher age groups.

The reproductive section of the community represented by the proportion of women between the ages of 15 and 45 has diminished slightly from 250 per 1,000 population in 1921 to 246 in 1931. In this form the change is not itself remarkable, for the current proportion does not depart materially from the general level which has been recorded since the beginning of the century and is definitely higher than the corresponding figures of the decades prior to 1901. The significant feature is the change taking place within the reproductive ages, the figures showing that the decline in proportion is located at the younger and more fertile section of the group, the proportion at ages 15.29 having fallen continuously from 1901 and now being increasingly below what it was in 1881 and 1891.

(a) Sections of the Country


Sex and age proportions within separate sections of England and Wales are shown in Table XLII. They present marked contrasts in their comparable features though to a large extent they maintain constitutional characteristics which have been recorded and commented upon at previous enumerations. Two types of divisional analysis are shown; one as a series of Density Aggregates having no particular geographical significance (except the Greater London Division) and the other a division according to the eleven self-contained geographical divisions employed as the major units of areal aggregation throughout the census analysis and presentation. The so called Density Aggregates are divisions according to degree of urbanization and the characteristics displayed are strictly those pertaining to this type of division. Even so the contrasts will tend to be blurred since many towns embrace areas of rural type, while a number of rural districts, particularly in mining areas, contain considerable populations living under urban conditions; again as between the towns, while the largest towns have generally acquired the status of County Boroughs, the latter also include a number of smaller units, less populous and less densely aggregated than many towns included in other urban areas.

As between urban and rural conditions generally the outstanding differences in their age distributions are to be seen in the higher proportions of young children and old people in the rural areas and their complementary lower proportions of adults, particularly females, at the middle years of life. The excess at the young ages, taken in conjunction with the deficiency of females at the reproductive ages, arises from the generally higher fertility in rural areas and is maintained up to adolescence, after which the marked reduction in the proportions, males just after 20 and females somewhat before that age, reflects the considerable migration from country to town which then takes place and which has long been recognised as a feature associated with the ages of entry into employment. The effect of this early migration diminishes with advancing age but it is not until the ages of retirement are reached that the balance is restored after which the proportions are increasingly in favour of the rural areas up to the end of life.

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

As between the grades of urban areas, the proportions of young adults are exceptionally high in the metropolis and are complementary in this respect to the low proportions in rural areas, though there is some evidence in the case of males that some part of the inward migration at the early ages comes from other towns as well as from rural areas; the movement of females at this period of life has a slightly differing incidence, mainly favouring London but in a lesser degree county boroughs as well at the expense both of smaller towns and rural districts.

The position of London is further exceptional in that, in spite of the fact that females at reproductive ages are relatively considerably more numerous than elsewhere, the numbers of young children are abnormally low. On the other hand the outward movement at retiring and later ages does not appear to be significant in London; the complement of the high increases over 60 in the rural areas being mainly found in the decline of the proportion in county boroughs.

Among the geographical regions identified in Table XLII the Northern Rural Belt (N.2.) Eastern Counties (E.) and North and Central Wales (W.2.) are predominantly rural in character and reflect conditions already referred to in respect of rural areas though with some differences. In all three there is evidence of material outward migration at early adult ages with a contrasting reversal at retiring ages resulting in abnormally high proportions after age 60, more especially in the Eastern and North Wales sections. But whereas the proportions of children are high in the Northern belt, they are only modestly so in the Eastern counties and in North Wales are little different from the average. By way of contrast the South West Counties (S.W.) also largely rural in character returns exceptionally low proportions of children with apparently little surplus for migration elsewhere at the employment ages; on the other hand its favourable climatic and residential features offer special attractions to the elderly, and the proportions are as outstandingly high at ages after 60 as they are low at the young ages.

Northumberland and Durham (N.1.) and South Wales (W.1.) have common features in their association with mining and other heavy industry which may not be unconnected with the similarity in the sex age constitutions of their populations. They are each weighted with a high proportion of children, particularly the northern area where the proportions are the highest shown in the table, which is heavily reduced in two successive stages, the first by outward migration at the early employment ages when the earlier supernormal proportions fall to sub-normal levels and again at the later retiring ages when the proportions are again reduced, this time to levels lower than in any other region, the position in this respect being especially marked in respect of respective female components. The scale of the general decline in the population proportions with advancing adult age has no doubt been aggravated by the recent prolonged unemployment and industrial depression which has fallen with special severity on these particular

(b) Mis-statements of Age


The form of the question on the householders' schedule eliciting the age statistics derived from the 1931 enumeration was identical with that introduced at the previous Census of 1921, and consisted of a request for a statement of each person's age in years and months. At Censuses prior to 1921, the age returns were limited to statements in terms of integral ages alone in respect of all persons over the age of i and from the examination of the 1911 record—the first occasion in this country in which the analysis was made by individual years of age—it was obvious that the ages in this form were carelessly or approximately returned by a considerable element of the population. With the object of avoiding or lessening this type of mis-statement the question in 1921 was extended to cover months as well as years of age; it was never intended to make positive use of the more detailed record; it was a device solely intended as an indirect stimulant towards greater precision, to encourage a greater concentration upon the enquiry and avoidance of the looseness of statement evidenced in replies to the earlier enquiry.

How far the device succeeded in its object it is impossible to say. The age statements of 1921 were undoubtedly better than they were ten years before and they have again registered some further improvement in quality at the latest enumeration of 1931, but the defects are still present, and though they may not be of great consequence in the practical use and application of the Census age material, they have a wider significance from the fact that the age question is the most direct and least ambiguous of all Census enquiries and the one therefore in which the degree of statistical error is likely to be less than in other matters dealt with by the Census.

An innovation made in 1931 with the object of securing a greater degree of completeness in the enumeration, but with a direct bearing upon the age distribution, since the class affected was limited to a single year of age, was the general instruction that newly born children who had not yet been given a name should be entered and described as "Baby". It had been repeatedly observed at preceding Censuses that the enumerated numbers of children under i year of age were markedly less than those that might have been expected from the survivors of the births of the preceding twelve months. It was realised, that, with the considerable time lag which was known to exist between the occurrence of a birth and its subsequent registration, there would always be many thousands of recently born children whose births had not yet been registered and in the absence of any better explanation it was conjectured that the apparent Census deficiency at this age might well be due to the fact that many such unregistered and possibly unnamed children were being withheld from the enumeration. The expectation seems to have been confirmed by the 1931 record for, as is seen hereafter, the deficiency has been reduced to small proportions and is only slightly in excess of corresponding deficiencies associated with the ages of slightly older children.

The error examination procedure generally follows that undertaken in 1921 and in so far as the methods and the results are to a large extent similar, the description below will be usefully amplified by reference to the somewhat more detailed account set out in the General Report of the 1921 Census.

The examination was undertaken in two stages, appropriate with two categories into which the errors may be expected to fall, viz., (I) those of a generally local and unbiased character such as may arise from looseness of statement or from ignorance of the precise facts and (2) wilful omissions or deliberate mis-statements. Intermediate between the two may be placed the form of mis-statement suspected at past enumerations and referred to as the "age next birthday" error, referring to the implied tendency on the part of some individuals within a short period of their next birthday, to return the higher age instead of their attained age asked for; the possible Grange of the error is limited to a single year in any individual rase but it has the disadvantage of always operating in an upward direction.

(1) LOCAL ERRORS. These are identifiable from the observations themselves and their correction or elimination can be carried out by reference to the Census record alone, either by a smoothing process or by telescoping the individual age record in quinary or other suitably designed groups.

The nature of the 1931 disturbances of this character is to be seen from Diagrams E and G in which the numbers of the population at successive ages are shewn in graph form.

In considering the behaviour of the record at successive ages, care must be taken to distinguish between apparent irregularities which are essential features of the age distribution and those to which immediate attention is to be directed. For example, the greatest element of discontinuity is the prominent depression which occurs between the ages of n and 16, but from the subsidiary curve A shewn in Diagram G it will at once be appreciated that this is a more or less exact reproduction and result of the violent changes in the birth rate which occurred during the war and immediately succeeding years. If is not until after age 17 is reached that the present population can be regarded as survivors of a relatively even flow of births dating prior to 1915 and that the peculiarities of the curve can be associated witli defects in age statements. From hereon will be observed the tendency toward a heaping up of the numbers at ages ending in the digit 0 and an apparent preference for ages in even, at the expense of odd, digits. From the subsidiary curves B and C in Diagram G it will be seen that the 1931 experience in this respect corresponds with remarkable consistency to that of 1921 and 1911 though with some diminution in the more outstanding prominences and depressions.

For a numerical expression of the cyclical tendencies associated with the digit of age, the method adopted in 1921 has been applied to the 1931 record. The, male and female sections of the population over the 5o-year age period 23 to 72 have each been divided into five decennial sections of age 23.32, 33.42, etc., and the numbers at individual ages aggregated in groups with common unit digits of age so that persons returned at 23,33,43, etc.form the first group, those at 24, 34, 44, etc., the second gnrup and so on, making 10 groups in all. The series for each sex was then rateably modified so as to aggregate to a common total—10,000—and the resulting distribution is shewn in the first column of the subjoined statement.

The numbers so derived represent a progression by single years of age and as such would normally be expected to lie on smooth and continuously falling curves; and for a measure of the extent to which they do or do not comply with this expectation, graduated counterparts1 of the enumerated distributions are shown for comparison in the second column of the statement. The differences between the enumerated and graduated records are shewn in adjoining columns and in further columns the corresponding differences from the 1921 and 1911 experiences are recorded for comparison.

Ages ending in 1931 Experience Corresponding Percentage Difference in
Proportionate Population Excess or Deficiency (-) of Enumerated
Enumerated Graduated Amount Per cent. Of Graduated 1921 1911
 
MALES:            
3 1,087 1,080 7 0.6 0.8 -0.1
4 1,072 1,070 2 0.2 -0.4 0.6
5 1,051 1,058 -7 -0.7 -0.1 1.1
6 1,039 1,042 -3 -0.3 -0.6 0.1
7 1,000 1,024 -24 -2.3 -2.3 -6.1
8 1,018 1,002 16 1.6 1.5 3
9 976 977 -1 -0.1 0.3 -1.5
0 988 948 40 4.2 5.1 11.3
1 887 917 -30 -3.3 -5.3 -10.9
2 882 882 0 0 1 2.5
Total 10,000 10,000 65 0.6 0.8 1.8
 
FEMALES:            
3 1,092 1,089 3 0.3 0.5 0
4 1,076 1,073 3 0.3 0.3 0.9
5 1,051 1,055 -4 -0.4 0.1 0.6
6 1,034 1,037 -3 -0.3 -1 -0.5
7 994 1,016 -22 -2.2 -3.3 -5.7
8 1,017 995 22 2.2 2.5 3.4
9 962 972 -10 -1 -0.5 -1.8
0 991 947 44 4.6 7.3 12.1
1 879 922 -43 -4.7 -7.4 -11.9
2 904 894 10 1.1 1.6 3
Total 10,000 10,000 82 0.8 1.2 1.9

From this test, conducted over the age field in which the bulk of the local type of misstatements occur, it may be inferred that the general quality of the present age statements is somewhat better than it was ten years ago, the aggregate percentage error as shewn by the total lines having declined from 0.8 per cent, to 0.6 in the case of males and from 1.2 per cent. to 0.8 for females. The amount of the improvement was however distinctly greater in the previous decennium and as that synchronised with the change in asking for the age statement in years and months instead of in years alone it will probably be deemed desirable to maintain the question in its enlarged form in future Censuses or other enquiries at which the age is asked for rather than the date of birth. As between the sexes, the female record is somewhat inferior to that of males but the difference is not great and such recent improvement as is now registered is in favour of the female section.

As regards the incidence of the errors it will be seen that for each sex and for each of the three Census records the bulk is consistently associated with two pairs of ages which shew large and complementary variations accounting for between three-quarters and five-sixths of the total errors. The largest disturbance, as is already obvious from the diagrams, occurs at ages ending in o and I where the heavy excess at the former, customarily accounted to a particular attraction of these decennial round numbers, is obtained almost wholly from the ages next above, any evidence of understatement in the age below (ending in 9) being almost negligible in comparison. The other seat of disturbance is located in the 7 and 8 groups, the more outstanding difference here being the deficiency at ages ending in 7, complemented by an excess of nearly the same amount associated with ages ending in 8. The general magnitude of the distortion here is only about half that associated with the 0 and 1 group and is in contrast therewith in that the transfer of population which it reflects is upward from the lower age in the former and downward from the higher age in the latter case. The apparent transfer from 7 to 8 has in the past been ascribed to a preference for ages ending in even numbers but there is no material and consistent excess at other even numbers and it may well be that in this case it arises from the return of the age next birthday with the possible desire of avoiding the digit 7. Outside these major seats of disturbance, the differences are of less consistency as between one experience and another and are hardly of sufficient consequence materially to impugn the validity of the age record.

The main advantage of the test so described is the evidence it provides of the progress recorded the matter of age statement over the decades covered by the comparison. An alternative measure of the local irregularities without this advantage but identifying single years of age and extending to the younger and older sections outside the 23.72 range is that provided by the comparison of the enumerated and graduated populations shewn in Table XLV on page 98.

From this comparison it would seem that the overstatement at ages ending in o, at the expense usually of the next higher age, is a feature extending practically throughout the whole range of ages, being in slight evidence as early as age 10 and at a maximum at age 50 where the excess may be as much as 7 per cent, in the case of each sex; the one exception to this rule worthy of note relates to age 20 where the enumerated numbers appear to be slightly deficient and where the presumably greater attraction of the maturity age of 21 appears to result in some transference from the lower to the higher age instead of vice versa as elsewhere. The apparent preference for ages ending in 8 at the expense of the preceding lower age is also in evidence throughout a large part of the range but at a much lower distortion level than that attaching to the 0 and 1 combination. Differences associated with other unit digits of age1 are not consistent either in sign or magnitude; they are generally no larger than would be expected in a comparison with an artificially smooth series and for that reason they cannot be read as necessarily implying defect in the enumeration record.

(2). OTHER AGE INCONSISTENCIES. In addition to the cyclical or local error it has been inferred in the past that the age records may have also been subject to a series of errors, not discernible, as are the local errors, from the distribution curve itself, but consisting either of the omission of substantial numbers altogether at some ages or of transfers from one part of the age field to another sufficient to lead to a sensible distortion of the true distribution. Such light as is to be thrown on them, if they do exist in any material degree, must be sought from such inference1 as can be drawn from a comparison of the Census record with some alternative distribution of independent merit. The procedure applied below follows that employed in the past and takes the form of comparing the enumerated population of each age group in 1931 with the numbers expected from the births of the years 1921-1931 or from the numbers recorded at appropriately younger ages at the 1921 Census after making such allowance as is possible for the mortality and migration to which they have been subject during the intervening period.

At the same time it must be expected that a test of this nature will be largely inconclusive since the expected numbers with which the enumerated are compared are themselves estimates only and from the mode of their construction will be subject to a degree of error which, in general, must be deemed to be far greater than that likely to attend the Census record. In the matter of the death record, for example, the age statement is supplied by a relative or other informant registering the death who may or may not be fully aware of the circumstances of the deceased, particularly at advanced ages where the majority of deaths occur; added to which is the methodological approximation introduced by the need of apportioning and debiting the recorded numbers at each age to population groups which are continuously and steadily ageing throughout the estimation period.

The greater difficulty which attends the estimation process however is the treatment of the balance of movement covered by the term migration. The component elements of this movement have been referred to on page 26 from which it may be setn to have been comprised approximately of a net loss of some 495 thousands permanent migrants to countries outside Europe offset by an inward civilian net balance from Europe and other parts of the United Kingdom of 240 thousands together with a further gain (on balance) of 55 thousands by a reduction in the numbers of the Armed Forces and merchant seamen serving outside England and Wales. In respect of migrants between England, Wales and countries outside Europe, records distinguishing sex and a series of broad age groups are regularly collected and published by the Board of Trade but in respect of the other elements no information is available beyond the bare numbers involved, and an arbitrary sex-age allocation of the total migration based on the coarse distribution available in respect of part of the movement predominantly outward in character may well introduce material error in respect of the other components of the movement, which on balance occurs in an inward and opposite direction.

For ages under 10 the error in the calculated survivors should be at a minimum since they are derived basically from the births registered in the decennium and they are shewn in comparison with the enumerated at individual ages in the following statement, the separate sex records being amalgamated for simplicity as they shew no characteristic variation at this period of life. The differences revealed are supplemented with comparable figures for 1921 and 1911 so far as they are available.

COMPARISON OF 1931 CENSUS POPULATION AT AGES 0.9 (as projected to 3qth june, 1931) WITH SURVIVORS FROM PREVIOUS BIRTHS (in thousands)

Original Births (July to June) * Estimated net reduction by mortality and migration prior to mid 1931 Age (mid 1931) Expected survivors (mid 1931) 1931 Census population carried forward to mid 1931 Differences as excess or deficiency (-) of Census population 1931 Comparable differences (Number)
Number Per cent. 1921 1911
1930-1931 642 33 0 609 598 -11 -1.8 -24 -50
1929-1930 643 46 1 597 593 -4 -0.7 -22 -54
1928-1929 650 55 2 595 591 -4 -0.7 -3 -1
1927-1928 656 64 3 592 592 0 0 -7 -3
1926-1927 673 66 4 607 605 -2 -0.3 -5 4
1925-1926 701 74 5 627 623 -4 -0.6 -2 0
1924-1925 718 79 6 639 637 -2 -0.3 0 0
1923-1924 740 87 7 653 653 0 0 1 0
1922-1923 760 91 8 669 670 1 0.1 -2 0
1921-1922 816 99 9 717 711 -6 -0.8 0 0

*Births registered August to July, thus allowing for one month's registration delay.

The latest experience is similar to that of 1921 and 1911 in that the calculated survivors at the youngest ages are in excess of the enumerated but with a definite improvement at the ages of o and I where the differences were formerly markedly in excess of those applicable to older children. For 1931 the difference at age I is now less than one per cent, and thus falls more or less into line with later ages and the only remaining outstanding difference is that for age o where it is still nearly two per cent, notwithstanding its material improvement over 1921. It has already been stated that one of the explanations offered for the deficiency of the enumerated at age o in 1921 was the possible omission of very recent births and that to counter this a special instruction was included in the 1931 Census schedule that unnamed infants should be entered and described as "Baby"; as the latest improvement in the record appears to accord with the earlier anticipation it seems reasonable to assume that the view regarding the discrepancy in 1921 was not unfounded though it is not possible to test it in any direct way. At the same time the Census deficiency at ago o appears to be still materially in excess of that at later ages and for tin's the only explanation that suggests itself is one which has been offered before, namely the return of the age next birthday on the part of a section of the community. It may be observed that one per cent, of the population are always within 3 or 4 days of their next birthday or two per cent, within a week, and that to many of them who have no appreciation of the statistical issue involved, the age so nearly attained will appear a truer representation of the position than that of an age apparently out of date; such error would be of the magnitude of that now emerging and it would only be visible at age o since for all later ages, the number wrongly transferred out to the next higher age would be offset by the similar number transferred in from below.

The discrepancy at childhood ages after the first, appears to be consistent and, though not of groat size, is more than would ordinarily be expected to arise from error in the survivorship calculation having regard to the comparatively short intervals elapsing between birth and Census. On the other hand there seems no reason to account for the omission of such children from the Census or for the deliberate mis-statement of ages which must be precisely known by their parents or guardians, and in the absence of such explanation the small differences now disclosed can hardly be accepted as an adequate justification to dispute the validity of the Census record.

For ages over 10, the comparison becomes more involved, for the period over which the survivors are traced is never less than ten years and in addition to the difficulties attending the treatment of deaths and migration of this period, the survivorship construction is based upon the Census record of 1921 in which the errors, if they exist, will be as great if not greater than those attaching to the 1931 record which it is the object of the test to reveal.

The comparison is set out in the following table, the ages being grouped in quinary periods so chosen as to minimise the effect of the cyclical local irregularities associated with the unit digit of age.

TABLE XLIII.— COMPARISON OF 1931 CENSUS POPULATIONS AT AGES OVER 10 (AS PROJECTED TO 30th JUNE, 1931) WITH SURVIVORS FROM THE 1921 CENSUS (AS PROJECTED TO 30TH JUNE, 1921) (IN THOUSANDS)

The differences between the enumerated and the expected are shown both for 1931 and 1921 in the last four columns of the table, the figures being shown as positive or negative according to whether the enumerated or the survivors are the greater. On the whole the 1931 differences are somewhat lower in magnitude than those associated with the 1921 record thereby suggesting that there has been some improvement either in the Census age statement or in the survivorship computation process. Subject to this general improvement the outstanding features of the distributions of the differences are similar for the two Censuses from which it is to be inferred that the causes responsible for them are much the same today as they were in 1921, the change being in their intensity rather than their nature. As between the sexes there is a common likeness in incidence of the differences at the youngest and oldest ages but not at the earlier adult period, thus again repeating the experience of 1921.

For both males and females at ages 10-13 the enumerated is in excess of the calculated survivors; and since the amount of the excess corresponds in magnitude with the accepted deficiency at ages 0-3 at the 1921 Census and would practically disappear if the 1931 survivors were traced from the survivors instead of the enumerated at the younger age group in 1921, the difference now brought out will probably be regarded as confirming the 1921 deficiency rather than throwing any doubt upon the 1931 record.

From age 14 up to the early fifties the male experience differs from the female and is characterised by two well-defined phases consisting of a marked excess of survivors up to age 29 followed by an equally marked excess of enumerated at subsequent ages of which the bulk is concentrated in the 29-34 group. Careful examination of all the relevant features of this section of the male age field indicates a strong probability that this two-phase alternation almost certainly arises from an inadequate allowance in the survivorship construction for the special migration element associated with movements of the Armed Forces. Service establishments abroad, in respect of which the contribution from this country may be. from 150 to 200 thousands at any time, are maintained by series of drafts sent out each year and returned some years later at appropriately older ages. The net migration loss or gain thereby over a decennium may be negligible in numbers but its disregard in the computation of expected survivors would result in precisely the type of age differences now disclosed and in view of their magnitude must be. regarded as completely inhibiting inferences, adverse, or otherwise, regarding the quality of the age statements themselves.

For females over this age period (14 to 53), the amounts of the differences are far less than those associated with males and with a contrary incidence, the enumerated being in excess at first followed by a rather larger deficiency in the latter half of the period. The differences are greater than would be expected from an inadequate treatment of migration in the survivorship construction and on their face, appear to lend some colour to the suggestion, made in the past, that they are evidence of a tendency for females to understate their ages when approaching the middle years of life; but whatever may have been the position in the past it is almost impossible to establish the existence of any significant distortion from this cause in the latest record, since the differences proceed from errors in both the 1921 and 1931 enumeration and on the assumption that age statements have generally improved over the course of time, the bulk of those now disclosed could be assigned to defects in the 1921 record. For example it was explained in the 1921 report that the apparent Census excess of 35 thousands at ages 9—13 in 1921 would have been converted into a deficiency if the then survivors had been traced from births instead of from the Census figures of 1911 and that there were good grounds for thinking that the 1921 population was under-represented at these ages; and if, as is not unlikely, the amount of that under-representation were of an order of magnitude similar to that of the apparent Census excess of 19 thousands at ages 19-23 in 1931 it would largely explain the latter, and would thus tend to support the 1931 Census record at 19-23 rather than to dispute it. At most age groups under 53, a difference in 1931 is associated with a difference of opposite sign at the group 10 years less in 1921, the conjunction going far towards destroying the possibility of identifying the real Census error, if any, concealed within either.

At ages over 54 the male and female experiences are not dissimilar in that the survivors are in excess between 54 and 63 and the enumerated between 64 and 78 after which the differences are reduced to small amounts for each sex. At these, ages the mortality and migration flux used in the survivorship construction becomes increasingly large in relation to the resulting differences and having regard to the practical difficulties in assigning the recorded deaths amongst the steadily ageing population groups, it becomes virtually impossible to arrive at any acceptable conclusion regarding the meaning of the differences or their apportionment as between errors in either of the terminal Census records or in the estimate of the survivorship movement between them. Whatever justification there may have been for suspecting biassed statements of age in the past, the evidence provided by the survivorship comparison would appear to be quite inconclusive regarding omissions or widespread misplacements in the 1931 record with the single exception at age o at which a probable deficiency in the enumerated appears to be sufficiently established to warrant a preference for the slightly larger number of calculated survivors at that age.

(c) Graduation of Ages


The effect of local mis-statements in the ages as given on the Census schedules to which attention has been drawn in the immediately preceding pages, and which are visible over the major portion of the total age field from the pictorial representation of Diagrams E and G, will be largely reduced when the individual age records are telescoped into quinquennial or decennial groups and for many of the purposes for which the Census records are required in practice, where comparatively broad groupings of age are sufficient, their disregard will occasion little difficulty. For some purposes, however, the numbers at individual ages are significant, and for such circumstances a record is required from which the major superficial irregularities have been eliminated by a process of smoothing or graduation. The steps taken towards that end are briefly set out below together with a table shewing the graduated numbers and the differences between them and the corresponding enumerated.

It will be recognised that the true age progression of the population, if it could be established, would not be a perfectly smooth series since it is derived fundamentally from a succession of births of which the numbers are known to vary from year to year; the original unevenness imparted at birth may be modified by subsequent deaths or migration but it will not be eliminated so that a smoothing process applied directly to the enumerated numbers will only be effective in producing a series more closely corresponding to the true underlying numbers if the irregularities to be removed are in excess of those associated with the birth record.

From an inspection of the birth and enumeration records, the latter conditions could be accepted as applying in respect of ages from about 16 upwards in respect of which all the relevant births had occurred prior to the 1914-18 war. But at younger ages the condition was not fulfilled die-re the population variations were obviously dominated by the violent changes in the annual numbers of births over the war and post-war years, the influence of which would need to be eliminated before any lesser unwanted irregularities could be seen and removed. This object was obtained by applying the graduation process, not to the populations direct, but to the ratios of the population at each age to the births with which they were associated and obtaining adjusted populations by multiplying the said births by the graduated survivorship ratios.

In the procedure actually employed, the graduation based upon survivors covered the age field from o to 30 and this was accompanied by a graduation based directly upon the enumerated populations from age 26 to the end of life, the overlap being available to effect a smooth junction between the two.

The Census was taken on the 26/27th April 1931, and the survivorship ratios at successive ages employed as the medium of graduation for the younger section of the experience were accordingly ascertained by reference to the births which occurred during regressively preceding calendar years ending on the 26th April, as estimated from the published Quarterly registration records of births. In this process it was assumed that an average interval of a month elapsed between the occurrence of a birth and its registration except in 1918 when the registration time lag was temporarily reduced (as a result of food rationing) to rather less than 3 weeks.

Since the objective in the case of each of the two graduations was solely that of passing a 20smooth and continuous curve through the ungraduated age series, the type of graduation explained on pages xxxix-xlii of Vol. VII of the 1911 Census was deemed appropriate and was followed in principle on the present occasion; graduated pivotal values being first ascertained at quinary age points from successive sections of the ungraduated data and the intermediate values being thereafter determined and inserted by the method of osculatory interpolation.

The pivotal values were each calculated as the central point of a third difference curve fitted by the method of moments to 2x + 1 consecutive values of the ungraduated functions (i.e. x terms on each side of the pivot age). For the survivorship ratio graduation covering ages 0-30, the ages selected for the pivotal values were those ending in the digits o and 5, the graduated values for ages 5, 10 and 15 being obtained from 7 successive terms of the ungraduated series (i.e., 2x + 1 = 7), and those for ages 20, 25 etc., from 13 terms; for the commencing age o the estimated number of survivors was regarded as a truer representation of the population than the enumerated and the adopted survivorship ratio was based upon such survivors. For the population graduation from age 26 upwards, the pivotal ages chosen were those ending in the digits r and 6 and the graduated functions were derived from 17 terms of the ungraduated.

The pivotal values having been computed at 5 year intervals, a third difference osculatory interpolation formula was employed to determine the values at the four intermediate ages between each pair of pivotal values; except between o and 5 where the method was inapplicable and an ordinary 3rd difference formula was used based upon the previously ascertained values at ages 0, 5, 7 and 9.

The two overlapping series of constructed populations were seen to be almost identical at age 26 in the case of each sex and a simple junction was effected at this age without further refinement, the values for earlier ages thus being derived from the graduation of the survivorship ratios and the values for later ages from the graduation of the population numbers.

Two final adjustments were made to the graduated numbers. The first was applicable to each sex and was of insignificance in itself; when the graduated numbers were aggregated for all ages, the totals were found to be slightly different from the enumerated numbers in respect of each sex arid the differences were in each case eliminated by a rateable modification of the graduated at each age. The second, of more significance, was the correction of an obvious defect in the male graduation; comparison of the enumerated and graduated age by age clearly shewed that the graduation, in smoothing out the local irregularities, was also tending to obliterate an inherent feature of the distribution, viz., that reflecting the loss due to casualties in the 1914-18 war which should have resulted in a somewhat sharp depression in the age curve from about age 31 and have continued with an ultimately diminishing influence for 20 years or more; allowance was made for this by the superimposition of a hand adjustment, smooth in itself but conforming more nearly to the progression of the enumerated. The resulting graduated functions are shewn side by side with the original unadjusted data in Tables XLIV and XLV.

TABLE XLIV.— UNGRADUATED AND GRADUATED SURVIVORSHIP RATIOS-AGES 0-30, 1931 (Ratios of population at each at each age in 1931 to relevant births of earlier years)

TABLE XLV.— ENUMERATED AND GRADUATED POPULATIONS-CENSUS 1931 (The graduated population at ages 0-25 have been obtained via the medium of the survivorship ratios of Table XLIV)


5. Marital Condition.


Of the 39,952,377 persons enumerated in England and Wales in 1931, 20,324,878 or 50-9 per-cent, were returned as single, 17,093,411 (42.8 per cent.) as married, 2,501,373 (6.3 per cent.) as widowed, and 32,715 (0.1 per cent.) as divorced. Among the single of all ages, there Wen: 1,05: female's to 1,000 males, the proportion having declined from 1,065 in 1921, but being still well in excess of the 1911 proportion of 1,029. As usual the number of wives enumerated in the country exceeded the corresponding numbers of husbands, the excess of 113,785 wives representative of 1,013 wives per i,coo husbands comparing with 114,956 (1,015 per 1,000) in 1921 and 134,498 (1,021 per 1,000) in 1911. Widows numbered 1,782,517 against 718,856 widowers, the ratio of 2-48 widows per widower comparing with 2-52 in 1921 and 2-22 in 1911. The divorced, so returned, which according to the instructions governing the returns should have included both parties to a divorce provided that each was still within the country and had not remarried numbered 13,546 males and 19,169 females as compared with the far fewer and more nearly equal numbers of 8,464 males and 8,218 females in 1921.

In the following table, the numbers returned under each condition in 1931 are set out in age groups and the intercensal increases or decreases in the several groups show where the principal changes in the distributions have occurred.

TABLE XLVI.— DISTRIBUTION BY SEX, AGE AND MARITAL CONDITION, 1931 (a) Population in thousands 1931

From the nature of this record it will be apparent that the actual numbers of persons enumerated as single, married, widowed or divorced, or the decennial increases or decreases in the numbers, have not been determined solely or even mainly by the forces governing marriage, widowhood and divorce, but are conditioned to a large extent by the common age factors affecting all marital conditions alike. The latter have already been described in the preceding "Ages" section of the report; originating, as they mainly do, from the steady rise in the annual number of births over the years prior to 1903 and their decline thereafter, coupled at the same time with a generally improving survivorship condition, they are responsible for the recent declines in the population at ages below 20 and the increases at the later years of life; superimposed on these changes in the case of males, is the further and special effect of the incidence of the war losses registered by a serious depression in the 1921 population at ages 20-35 which in 1931, ten years later, has been transferred to ages 30-45, the effect of such transfer being to produce an abnormal increase at ages 20-30 complemented by an equally abnormal decrease in the 35-45 age period. These changes will have affected each of the condition groups though with a varied incidence according to the extent to which the more specific marriage and widowhood factors have operated to modify them.

Though marriages fluctuated during the war and immediately succeeding years, the general effect was to raise the general level of the rate in the 1911-21 decennium; the higher average for that period has not been maintained except at the youngest female ages below 20 and the reduction, though not large, has exercised a depressing effect upon other tendencies affecting the marriage section of the age field. The effect is observable in the figures at male ages 20-29 and female ages 20-24, where the rates of increase are lower for the married than for the single, the reductions in the married at later ages due to fewer 1921—1931 marriages being offset by the increase in the marriages at ages 10 years younger which occurred prior to 1921.

Widowhood rates, like mortality with which they are synonymous, have declined, the effect of the decline being most marked at the older ages where the percentage increases in the numbers married of both sexes are at a maximum and are materially in excess of the corresponding increases in the widowed sections. Further, in the case of younger females, the numbers of widows in 1921 were unduly swollen as a result of the heavy war casualties amongst married men and to this feature is to be accounted the largest percentage decline in the widowed at ages 20—35.

As a result of all the marital condition changes amongst women, the proportion of married women at the reproductive' ages, 15-45 has risen from 121 per 1,000 total population in 1921 to 123 in 1931 while the corresponding proportions of the non-married at the same ages have fallen from 128 to 123; at the same time any increase in fertility potentiality which might be inferred from the rise in the married proportion is more than discounted by the fact that the proportion are probably at a maximum at the present time and are bound to fall in the near future as the present numbers are replaced from the smaller generations now entering the reproductive field.

As regards the sex incidence in what may be called the marriageable population—here regarded as the single, widowed and divorced sections at ages 20 and over—the excess of females is now 1,680,196 as compared with 1,667,846, ten years ago. As will be seen, however, from the subjoined statement, the vast bulk of the excess is located at the older ages where females, through their greater longevity, have always outnumbered males. In the age group 20-30, at which marriages mainly take place, females are fewer than males, the female excess of 1921, which was a direct consequence of the heavy male war casualties of 1914-18, having now given place to a relative deficiency, with existing proportions of 911 and 922 females per 1,000 males at ages 20-24 and 25-29 which are markedly lower even than they were in 1911 (984 and 990). After about age 30, the position is reversed, non-married females outnumbering corresponding males in proportions which quickly rise to the extent of giving the former a 2 to i predominance after about age 45.

EXESS OF FEMALES IN THE MARRIAGEABLE POPULATION (i.e., Single and Widowed and Divorced), AGED 20 YEARS AND OVER.

Ages Ratio of Females to 1,000 Makes Excess of Females over Males
1931 1921 1931 1921 Increase 1921-1931
20-24 911 1,043 -130,199 50,808 -181,007
25-29 922 1,154 -60,272 93,194 -153,466
30-34 1,322 1,470 105,499 147,148 -41,649
35-44 2,005 1,683 367,008 294,577 72,431
45-54 2,094 1,768 384,493 277,132 107,361
55-64 2065 2038 385,036 311,930 73,106
65 and over 2,295 2,263 628,631 493,057 135,574

TABLE XLVII.— MARITAL CONDITIONS: PROPORTIONS BY SEX IN AGE GROUPS, ENGLAND AND WALES (1891-1931) AND GEOGRAPHICAL REGIONS (1931)

At the same time the actual numbers of males and females involved diminish, and when account is taken of the fact that husbands, on average, are somewhat older than their brides, the numerical disparity between the respective elements in the two sexes over their reproductive years of life is probably of little significance, either way.

Persons returned in 1931 as divorced numbered 13,540 males and 19,109 females, representing increases of 5,082 or 60 per cent, in the case of males and 10,951 or 133 per cent, females over the corresponding numbers in 1921, the first occasion in this country in which an attempt was made to obtain separate records of this class. Rather more than half the numbers so enumerated were between the age of 35 and 55 and nearly 90 per cent, between 25 and 65. From contemporary evidence however it is quite obvious that the enumerated returns bear little relation to the facts; the numbers of marriages terminated by dissolution or annulment during the years 1921-1930 numbered 60,914 and during the same period 19,180 divorced men and 15,654 divorced women remarried, so that even when full allowance is made for the further decrement by mortality the true increase in the numbers between 1921 and 1931 must have been several times as great as that recorded and there can be no doubt that disproportionately large numbers of both males and females have failed to return the desired information. In view of the misleading character of the figures, the divorced have been retained as a separate class only in the section of the tables devoted to age and marital condition; in all other sections they have been merged with the widowed with whom from a statistical point of view they may be usually associated.

Table XLVII which has been extracted from Tables 23D and 25 of the General Tables Volume furnishes a comparative summary of the incidence of the several marital condition sections in a series of successive age groups for each sex, both for the country as a whole over the series of Censuses back to 1891, and for the principal divisions of the country in 1931 represented by the density aggregates and main geographical regions adopted for the purpose throughout this report.


1 parabolas of second degree fitted by the method of least squares.

Next Selection Previous Selection