Dependency, Orphanhood and Fertility

Next Selection Previous Selection


Under the title "Dependency, Orphanhood and Fertility" a volume has been issued as part of the series of 1921 Census publications introducing for the first time a series of statistics designed primarily with the object of providing data on a national scale for the service of the increasingly numerous and important problems relating to Pensions, Invalidity Allowances, Workmen's Compensation and such like matters in Which a liability in respect of an individual is to some extent determined by and measurable in terms of the numbers and ages of the children or others dependent upon him. The absence of such records from all earlier official enquiries had been a matter of comment on more than one occasion, and when it was seen that the provision of widows' and orphans' pensions or allowances was likely to be a feature in the social legislation of the post-war period, and that much of the information necessary to envisage the scope of the scheme and for framing the requisite financial estimates, could only be obtained through the medium of the census, it was decided that the appropriate enquiries should be incorporated in the 1921 census questionnaire. Two separate and independent questions were accordingly inserted in the householders' schedule, one relating to orphanhood to be answered in respect of each child under 15 years of age enumerated on the schedule, and the other relating to dependency, to be answered in respect of each married man, widower or widow enumerated on the schedule giving particulars of his or her children under 16 years of age. They are dealt with more fully in the sections of the report which follow. The tabulation and classification of the replies to the two questions have been kept distinct throughout under the subject headings "Orphanhood" and "Dependency," and though in a sense, the two sets of statistics may be regarded as complementary to one another—children with both parents living corresponding to dependents of married men or married women, orphans whose fathers have died to dependents of widows, and orphans whose mothers have died to dependents of widowers—no attempt has been made to reconcile the corresponding groups one with another, nor would it have been possible to do so owing to certain fundamental differences in the constitution of the populations brought under review.

The purpose of the enquiries was primarily an economic one—this is expressed in the use of the term "dependency", and the classification of the data has been designed mainly with this object in mind. Its scope, however, was so wide and the circumstances of collection and tabulation so general in character that it will be found to embrace more than might be implied by a limited use of the term "dependency", and it should provide valuable information of a general sociological nature. It is from the "dependency" tabulation that such fertility statistics as it has been found possible to incorporate in the Census reports of 1921 have been derived.

1. Dependency.

The question dealing with the subject of dependency is a new one and it will be desirable, therefore, to set it out in full as it appeared on the Census schedule. A reply was asked for in respect of every married man, widower and widow enumerated on each schedule.

The immediate object of the enquiry was to obtain as complete a survey as possible of the sizes, constitution and distribution of the families which could be regarded as economically dependent upon the married or widowed sections of the population in England and Wales and in the tables in which the results have been published, occupying 225 pages of closely printed figures, statistics Will be found of the numbers of families of each size ranging from the childless family to families of 13 in number (below the age of 18), each group of families being separately analysed so as to show the number of children at each age (under 16) and the distribution by age of the children comprising the youngest of each family. The detailed family classification is given for a complete range of parents' ages and in respect of each separate type of parent (le whether married man, married woman, widower or widow), and in subsidiary tables the statistics in a less intensive form have been extended to show the variations incident to locality (geographical regions and counties) and, in respect of the families of married men and widowers to the personal occupation of the parent.

Information required only in respect of Married Men, Widowers and Widows .
Number and ages of all living children and step-children under 16 years of age, whether enumerated
on this Schedule or not, i.e., whether residing as members of this household or elsewhere.
Total number
under sixteen
years of age.
If none write
For each child place a X in the column corresponding to its age.

The number of crosses should be the same as the number
shown in Column (n) .
(n) (o)
  Age last birthday.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

It will be observed that while the terms of the census question include step children with ordinary children, no reference is made to the presence of a condition of dependency as between child and parent. The condition may be normally implied as arising out of the natural relationship, and in a very large majority of the cases the rule will hold good, but there are bound to be some instances where parents have ceased to be responsible for some or all of the children properly returned by them in answer to the census enquiry. It must be borne in mind therefore that the statistics strictly represent those of families of defined limits and that they will fulfil the object for which they were primarily designed only to the extent to which the dependency relation may be ordinarily presumed under existing social conditions in respect of children under a certain age.

The use of the age limit—16—for this purpose does not imply that the condition of dependency is necessarily maintained up to the attainment of that age and thereafter abruptly ceases. For some purposes this age may be too low but most of the uses to which the figures are likely to be applied will be sufficiently served by the range of age adopted and the analysis of the data by individual years of age will facilitate the modification which would be necessary if for any particular purpose it was desired to restrict the limit to a lower age than that of 16.

In regard to the data on which the present study is based, two factors, incident to the 1921 enumeration, must be referred to as detracting somewhat from the general application of the statistics as they stand. The first relates to the abnormality of the numbers and age distribution of the children included within the dependency age field. The children at ages below 16 in 1921 were the survivors from births occurring between the middle of 1905 and the date of the census (19th June, 1921), a period embracing the whole of the war years during which the fluctuation in the birth rates was far greater than any recorded since the commencement of registration. Prior to the war the successive years' births were exhibiting a declining tendency and between 1905 and 1914 the numbers had dropped by irregular stages from 929,293 to 879,096. Thereafter the fall was at first suddenly accelerated, a minimum of 662,661 being registered in 1918 and then even more suddenly reversed, the numbers rising from 692,438 in 1919 to 957,782 in 1920, the latter being the highest figure for a single year hitherto recorded in this country. The survivors of these numbers shewn as "enumerated" in Table LXVIII. or their counterparts as "dependents" naturally follow these variations. The total numbers are undoubtedly smaller than they would have been but for the war, and the irregularity in the age series is one which is inconceivably likely to be present in any other set of circumstances.

The other reservation has regard to the quality of the data. The dependency question as framed and printed on the schedule was, in relation to census standards, undoubtedly a complex one. Experience shews that, in census taking, even the simplest and most direct personal enquiries are subject to an appreciable degree of error either through ignorance or carelessness, notwithstanding the forethought displayed to make their meaning clear and free from all possible ambiguity. The wider the scope of the question the less easy it is to ensure these conditions, and, in the present instance, the question, including consideration of family relationship and extending to persons not necessarily present at the time of enumeration, gave rise to an amount of difficulty which was bound to detract from the completeness and accuracy of the returns. In a number of cases a certain amount of editing was possible by reference to other information given on the schedule. Types of error so encountered were (1) the omission of the dependency return altogether when from the face of the schedule there were children who should have been so returned, (2) the inclusion of grandchildren or children other than those covered by the instructions, (3) the mis-statement of the dependents' ages through the placing of crosses in the wrong columns, and whenever it was possible to do so, corrections of obvious errors were freely made. Omissions and mis-statements not inconsistent with the rest of the information returned on the schedules must, however, have remained undetected, and the only evidence of them is such as can be obtained by comparison with other available data.

In the following table the "dependent" children are compared, age by age, with the total children enumerated in the country.


From the nature of the enquiry the "dependent" children should as they broadly do in fact, correspond with the enumerated. The classes however, are not identical, and it is for consideration whether the divergency of nearly 900,000 in the total, representing an apparent deficiency of 8 per cent. in the dependents can be satisfactorily accounted for. In the first place, the dependency question was only asked of married men widowers and widows. No reference was made tangle men or women and illegitimate children were, unless their parents had subsequently married, excluded from the return of dependents though included in the total enumerated. Their numbers are not precisely known, but from the births in the sixteen years preceding the census they may probably be estimated at about 4 per cent. of the total, or about one-half of the apparent total discrepancy. Another source of difference exists in the presence among the enumerated of children both of whose parents had died. Their numbers may be estimated from the orphanhood statistics, and can be regarded as accounting for at least another 1 per cent of the full difference of 8 per cent. Finally, though children outside the country of parents within the country should have appeared amongst the dependents and not in the enumerated, they will have been more than balanced by the children within the country of parents outside. That the balance in favour of the enumerated as compared with the dependents may be considerable is evidenced by the fact that 114,956 more married women were enumerated in the country than married men, and the difference in question may easily account for another 1 to 1 per cent. of the total.

Altogether these and other minor sources of difference, though not wholly mutually exclusive, suffice to explain a large proportion of the 8 per cent. difference between enumerated and dependent children and to show that the unexplained portion which may be attributed to omission is more probably between 2 and 3 per cent.

If the comparisons are taken age by age a tendency will be seen for the percentage difference to decrease slightly down to about age 5 and thereafter to increase, slowly at first and rather more rapidly towards the upper limit of age. It has been suggested earlier in this report (see page 77) that the enumerated children at ages 0 and 1 last birthday are themselves deficient to the extent of about 3 per cent. and if this is so, the causes responsible for omissions in the enumerated at these ages would appear also to have operated in respect of the returns of dependents, such deficiency being additional to that which may be inferred from the discrepancy in the table. The increasing nature of the discrepancy at the later juvenile ages is occasioned mainly by the relatively greater defects in the dependency returns of widowers and widows (see Table LXIX), the proportion of dependent children in the widowed classes becoming greater and greater as the age of the child advances, but it also accords with the suggestion that part of the difference is due to the legitimate omission of orphans and of children whose parents are outside the country, for each of these classes will tend to grow with advancing age, particularly during school years. The marked increase at ages 14 and 15 would appear to suggest that a further factor enters into operation after the attainment of the maximum elementary school age. An increasing separation between parent and child no doubt occurs as the latter enters the economically independent stage, and from the point of view of a dependency study this feature may perhaps be regarded as favourable rather than otherwise.

Testing the three dependency classes against the "orphanhood" returns, and the comparison is now limited to children under 15 since that was the limit of age to which the orphanhood question applied, we have:

Children enumerated as having Dependents under 15 of Excess of enumerated.
Number. Per
Both parents alive 9,315,060 Married men 8,964,070 350,990 3.8
Mother dead 261,094 Widowers 182,384 78,710 30.2
Father dead 730,845 Widows 549,370 181,475 24.8
Both dead 55,245        
Not stated 138,211        

By ages, the percentage differences shewn for each class as a whole in the last column above are distributed as follows:—


From the percentage differences thus shewn, it would appear that the returns in respect of married men, accounting as they do for more than 90 per cent. of the children, may be accepted with a good deal more confidence than those of widowers or widows. This is probably true, since there tends to be a greater degree of separation of child from parent among widowers and widows than amongst married couples, which is bound to be reflected in some omission of children from the widowed classes. But it must be remembered that, for all classes together, of a total discrepancy of 8 per cent., nearly 6 per cent. could be explained, leaving little more than 2 per cent. to be attributed to omission, and it is not unreasonable to suppose, therefore, that the discrepancies shewn for the three parent classes separately can be similarly scaled down and the actual error regarded as nearer one quarter of the differences shewn. The comparison in respect of the widowed sections is probably affected very greatly by the inclusion of illegitimate children amongst the enumerated. The total number of illegitimates at these ages may be estimated in round numbers as 400,000, and, if, as is not unlikely, an undue proportion of them have been returned as having father or mother dead, the explanation will account for a large part of the discrepancy of 260,185 in the relative comparison.

Another feature of the returns which may occasion disquiet to a user of the figures is the apparently large number of married men, widowers and widows who failed to make a dependency return. They are identified throughout the tables and number 1,566,260 or 16.1 per cent of the total, the proportions for the separate classes being as follows:—

  Married Men. Widowers. Widows.
Total enumerated 7,475,051 642,311 1,621,758
Numbers not making the return 1,031,661 232,948 301,651
Percentage 13.80% 36.30% 18.60%

From other evidence on the schedules, particularly that of the age of the individuals concerned, it is apparent that in many instances the existence of children under 16 years of age was extremely improbable. This is generally confirmed by the foregoing examination as to the completeness of the total numbers of children, and in treating the absence of statement as merely indicative of a "nil" return there seems no reason to suppose that the resulting degree of error will be materially worse than that indicated by the deficiency of children estimated above at between 2 and 3 per cent. for the three classes together.

It is to be noted that the dependency question was put to married men, widowers and widows and not to married women. To have asked for a statement in respect of each married woman as well as each married man would have involved the duplication of a large proportion of the returns, and would have introduced a further element of confusion in an already complex question. Moreover where the husband and wife were enumerated on the same schedule—a circumstance embracing nearly 92 per cent. of the total married women enumerated in the country—the husband's dependency return could be associated with his wife and it has been possible by so doing to obtain dependency statistics in respect of married women They have been made available as a fourth class in the published tables and the only reservation to be made in respect of them is that they represent a sample only though a very large one, of the total married women in the country.

The sizes of the various dependent families encountered in the enumeration i.e., the numbers of children and step-children under 16 years of age, varied from 0 to a maximum of 13 and were distributed as follows:—


For each class of parent, families of no dependent children—with which may, for reasons suggested earlier, be included the cases where the return was omitted— are far more numerous than those of any other size identified They comprise childless: families and families in which the children have all passed out of the dependency age period and account for more than 40 per cent. of the total families of married men or married women and more than 80 per cent. in each of the widowed classes. Of the balance, viz.:—families in which dependent children were returned, the frequency is greatest in respect of the one-child family for each class of parent, and diminishes rapidly with every succeeding increase in size of family, the families of married men or women with 9 or more children and the families of widowers or widows with 7 or more children numbering less than 1 per 1,000 of their respective totals. The largest families encountered were of 13 children, of which 8 were returned in respect of married men and married women. The largest widower's family was one of 11 children, while 13 widows returned families of 10 children each. The preponderance of childless families and families of very small dimensions is shown by the aggregate average sizes of families amounting to 1.27 and 1.32 children in the case of married men and married Women and 0.32 and 0.37 in the case of widowers and widows respectively.

As already explained, the married women's figures are derived from those of their husbands with whom they were enumerated, and the difference between the families of married men and married women is due solely to the inclusion amongst the former of a number of married men who were not enumerated with their wives. If the figures for this class betaken out separately it will be found that in 39.8 per cent. of the cases the return was omitted—a percentage greater than in either of the widowed classes—and that the families with no children (inclusive of the "no statement" cases) amount to 67.8 per cent. of the whole. These, in conjunction with the remaining 32.2 per cent. of families in which one or more children were returned result in an average size of family of 0.64 children, suggesting that, as a class, married men not enumerated with their wives are rather more akin to the " widower" group than to the married class in which they are incorporated.

The age distribution of the children in the several classes of families is given in detail in the published tables. Those of married parents—married men or married women—comprise more than 90 per cent. of the total children, and their distribution over the years 0 to 15 last birthday generally corresponds, as it must since it is so large a proportion of the whole, to the incidence of the total enumerated at these ages, with all the abnormality and irregularity introduced by the events of the war period. In the case of widowers and widows, the duration of widowhood has a limiting effect upon the ages of the children in each family. Only in respect of the most recent widowings will the ages of the children range over the whole field, while in others the dependent children will generally have been returned at ages between the age corresponding to the duration of widowhood and the limiting dependency age of 16, so that when the parents are aggregated irrespective of the date of widowhood the numbers of children ascend continuously from a minimum at age 0 to a maximum at age 15 last birthday. The influence of the war birth irregularities are, of course, inherent in the widowed classes just as they are in the children of the married, but they affect only the younger ages where the numbers are relatively fewer owing to the heaping up of the widower's and widow's children at the older dependency ages, and they are to a great extent obscured by the effect of the varied durations of widowhood attaching to the several parents.

A not unusual feature of schemes providing for the grant of allowances to children is the incorporation of a special benefit in respect of the single child or of the first child in a family containing more than one child. To meet the circumstances of such a provision, the youngest child in each family for whom one or more children were returned, was segregated and a complete analysis of the "youngest" children made as a separate section of the dependency tabulation (see Table 1 of the Dependency, Orphanhood, and Fertility Volume). Their ages as a class are naturally lower than those of all dependent children taken together the difference being more pronounced amongst dependents of married men and married women than in the relatively smaller families of widowers and widows.

The characteristic differences in the age distributions of all dependent children and of youngest dependent children for the several parent sections and for various ages of parents is expressed by the average ages shewn in the preceding Table LXX and in Table LXXI following.

From Table LXXI it will be seen that for each parent class, male or female, married or widowed, the average number of dependent children increases with the advancing age of the parent to a maximum which is reached between the ages of 35 and 40 after which the averages as consistently decline. As between male and female parents, both the initial rise and the subsequent decline is steeper in the case of females, the higher maxima of 2.31 and 1.80 children per family being reached at the age of 38 and 37 for married women and widows respectively as compared with the somewhat lower maxima of 2.18 and 1.53 children per family at the age of 39 in the case of married men and widowers.


The age distribution of the children, whether of total children or youngest children, is naturally dependent upon the age of the parents. The gradual shifting of the weight from the younger to the older dependency ages with the increasing age of the parent is shown by the rise in the average ages of children in the above Table. It will be observed, however, that the rise is not continued throughout the whole range of parents' ages. A maximum occurs at about the age period of 55-65 in each parent category and thereafter the children appear to become slightly, but progressively younger It is not clear why there should be a fall, but it is probably due to the inclusion of step-children (and possibly of adopted or other children who should not theoretically have been included) whose frequency would naturally be greater m comparison with the frequency of natural children at the later ages of life.

In Table LXXII the families of married men and married women enumerated on the same schedule are analysed according to the joint ages of the parents, arranged in quinquennial groups.

The relation between the ages of husbands and wives is dealt with more fully on page 174 and 175, but it will be observed from sections, (a) and (b) of Table LXXII that the joint frequencies have a strongly marked maximum, at all but the extreme ages in the Table, when the age of husband and wife fall within the same quinquennial age group. The greatest number of families classified in this way occurs in the group of husbands and wives each of whose ages is from 35 to 39 years last birthday where they represent 6.7 per cent of the total married couples. This is followed by the groups at the common age periods of 30-40 (6.6 per cent), 40-44 (6.4 per cent) and 45-49 (5.8 per cent).


The variations in the size of their respective families shewn in section (c) of the above table follow much the same course as that shewn, in Table LXXI in respect of married men and married women in the aggregate. The range is more restricted when the wife's age is the variable. The initial rise and subsequent fall are both steeper and the peak attained is higher for a fixed husband's age than it is for the corresponding wife's age, the husband's age in the latter case being the variable. The maximum family shown in the table is that of 2.45 children in respect of the combination of husbands aged 40-44 and wives aged 35-39 and this is followed by 2.40 (for husbands 45-49 and wives 35-39), 2.31 (for husbands 35-39 and wives 35-39) and 2.15 (for the two groups, husbands 50-54, wives 35-39 and husbands 35-39, wives 30-34).

Among the broad geographical sections of the country shewn in Table LXXIII. Wales (including Monmouth) returns the highest average families. At all except the extreme parent ages where the numbers of families are relatively small, the average family is considerably larger than elsewhere and this advantage is accentuated in the married men and widower sections by an age distribution of the parents unduly favourable to the larger families so that the aggregate average in the case of married men is 1.48 children per family or more than 16 per cent. in excess of the corresponding figure for the country at large, while in the case of widowers it is 25 per cent. in excess. Widows are not so favourably disposed in the matter of their ages and their average family is shown as 0.41, just over 10 per cent. above the normal. Following Wales, the Central and Northern sections rank next, both being above normal in respect of their average families. At individual age groups the families of the Central Counties are more frequently the larger, but here again, the aggregate averages of Table LXXIII. fail to reveal the whole story owing to differences in the relative numbers of parents at each age and reference must be made to the detailed tables for a more exact statement of the relative positions. In a similar way the Eastern Counties, where the families at comparable parents' ages are generally above the general averages in point size, are shewn in the Table not only as subnormal in relation to England and Wales as a whole, but as ranking below the London region where the individual families are distinctly lower on the whole in each of the parent categories. The smallest families are returned by the Southern region where the average sizes in the aggregate are approximately 13, 25, and 19 per cent respectively below those in respect of married men, widowers and widows in the country at large.

With a reduction in the size of the areal unit the range of average families widens.



From the summary of individual county records given in Table LXXIV, it will be seen that the families of married men are highest in Monmouth where they reach an average figure of 1.60 children and are accordingly nearly 26 per cent. in excess of the general average. Durham comes next with 1.59 children per family followed by Glamorgan (1.56), Carmarthen (1.52), Brecknock (1.48), Northumberland (1.47), and Stafford (1.47); the lowest averages are returned by Sussex (1.03), Devon (1.05), Carnarvon (1.07), Westmorland (1.09), Cornwall (1.10), Northampton (1.11), Dorset (1.12) and Merioneth (1.12). Widowers' families range from an average of 0.49 in Yorkshire to 0.20 in Sussex and those of widows from 0.55 in Durham to 0.23 in Cardigan.

The occupational association vaguely suggested by the order of the county averages is more clearly brought out in Table LXXV. which shows the average age of the married men and the size of their families in each of the principal divisions of the occupational classification. The unoccupied and retired class is also included in this Table for the sake of completeness. By its very nature, however, it must be composed preponderantly of men with no dependent children and the very low average of 0.39 children per family is, accordingly, well below the range of families in the "occupied" groups and is less than 30 per cent. of the average of 1.32 children per family returned in respect of the occupied in the aggregate.


Of the occupied, the largest families are found amongst miners and quarry workers where the average reaches 1.82 dependent children per family, nearly 38 per cent. in excess of the average for all occupied married men. Workers in non-metalliferous mines and quarry products, lime, coke, cement, etc., come next (1.69 children per family), and following these in order are stationary engine drivers (1.59), fishermen (1.56), makers of bricks, pottery and glass (1.54), gas, water and electricity supply workers (1.52) and chemical workers (1.51), in respect of all of whom the families are more than 10 per cent. above normal. The smallest families, those of professional men, viz., 0.90 per family, are 32 per cent. below the average for all occupations and are rather less than half of the maximum figure returned for miners and quarry workers, while clerks, draughtsmen, typists, etc., at 0.94, textile workers {1.01), makers of watches, clocks, etc. (1.04), persons employed in personal service (1.05), persons employed in entertainment, sport etc (1.07), commercial and financial occupations (1.10), return families more than 15 per cent below the general average. Not included in either of the extreme groups but important from the fact that they involve large numbers of workers are, on the one hand, builders. bricklayers etc., with 1.44 children per family, transport workers (1.42) and metal workers (1.41) with which may be contrasted agricultural occupations (1.25), workers in wood and furniture (1.23), and persons employed in public administration and defence (1.5).

2. Orphanhood.

The "Orphanhood" question of the 1921 census, asked in respect of each child under 15 years of age enumerated on the schedule, was as follows:—

For children aged under 15 write "Both alive" if both parents be alive; "Father dead" if father be dead; "Mother dead" if mother be dead; or "Both dead" if both parents be dead.

The only cases where it was thought that some difficulty might be experienced in securing the desired information were those of illegitimate children. The question was framed in such a way as not to require special interpretation in their case and enumerators were instructed that the answers in respect of illegitimate, as of legitimate, children should be determined solely by the known facts in regard to the existence of the father or mother, i.e., according to whether he or she was known to be alive or dead.

Statistics dealing with this subject are presented in tables 8 and 9 of the Dependency volume and in table 24 of the county series of publications.

The total children enumerated under 15 years of age in England and Wales numbered 10,500,455, and in respect of 10,362,244 of these or 98.7 per cent. of the whole, the answers were such as to enable them to be classified within the four categories, the information being omitted or not known in only 138,211 or 1.3 per cent. of the cases.

Of the total children, 9,315,060 or 88.7 per cent. were returned as having both parents alive, 730,845 or 7.0 per cent as having lost their fathers, 261,094 or 2.5 per cent. as having lost their mothers, and 55,245 or 0.5 per cent. as having lost both parents.

The most striking feature of these proportions is the large excess of fatherless as compared with motherless children, the former outnumbering the latter in a ratio of nearly 3 to 1. Male mortality is markedly higher than the corresponding female mortality at all adult ages, and to this, aggravated in general by the higher average age of the fathers and in respect of this particular experience by the loss of men during the war, part of the excess must be due. Part is possibly connected with the returns in respect of illegitimate children whose numbers, though not known precisely, may be put in the neighbourhood of 400,000 under age 15, and in a number of these cases where the parents were not living together, the absent one, commonly the father, may in ignorance of the facts have been described as dead.

The frequency of each type of orphan increases with the advancing age of the child as shown in the Table LXXVI.

For each class the increases are continuous, rising from an approximate zero point at the youngest possible age to a maximum at age 14. This is necessarily so since the rate of orphanhood is an expression, in terms of the children, of the mortality affecting their parents, from the date of birth in the case of the mother and from a date on the average nine months earlier in the case of the father, with a cumulative effect, therefore, as the age of the child increases and the date of birth becomes more remote.

Thus the proportion of fatherless children commences at about 9 per 1 000 at age 0; representing the father's mortality during an average period of 15 months, 9 months prior to and 6 months subsequent to date of birth. The proportion is nearly doubled for children aged 1 last birthday, where the period of mortality exposure is 27 months. At succeeding ages the effect of war mortality amongst male parents is seen in an increasing rise in proportions for a few years and the maintenance higher scale up to the final age of 14 where it reaches 104 per thousand children. In respect of motherless children the commencing proportion at age 0 is 4 per 1,000, and though at this age it represents only 6 months' mortality exposure to the mother, it must be remembered that the period is a special one in that it includes losses due to the risk of maternity itself. At age 1, where the mother's mortality exposure is 18 months on the average, the proportion only increases to 6 per 1,000 and thereafter the rise is more or less steady up to 44 per 1,000 at age 14, The loss of both parents is characterised by an extremely low initial proportion of 0.4 per 1,000 and a much steeper rise to 11.5 per 1,000 at the age of 14.


In regard to sex distribution, orphanhood appears to be rather more frequent in the case of girls than boys. The ratio of boys to girls in the total enumerated under 14 years of age is 1,013 per 1,000, the corresponding ratio in the case of children having both parents alive is 1,016 while, for the several classes of orphan, the ratios are "father dead" 1,002, "mother dead" 999, "Both dead," 984.


It is difficult to assign a reason for this slight sex differentiation, but it appears to be generally characteristic of orphanhood whatever the age of the child (up to 14) and in the several portions of the country as well as for the country as a whole. It might be read as evidence of the association of the sex ratio at birth—in the production of an abnormal proportion of females—with conations predisposing either of the parents to the risk of early mortality, in which case the conditions would appear to be of a general rather than specific nature, since, in the case of the oldest children in the study, the birth of the child and the death of the parent may have been separated by as long an interval as 15 years. On the other hand such tendency as there may be for orphans to be brought up in institutions or with families other than their own may operate more strongly in the case of boys than of girls, conditions leading possibly to some mis-statement of facts in the understatement of the proportion of boys in the orphan categories. This suggestion, however, is not supported by the evidence in the case of children for whom no information as to orphanhood was forthcoming, for in this class the preponderance of females is greater than in either of the orphan classes, viz.:—933 males to 1.000 females.

The incidence of orphanhood may be expected to vary throughout the country according to the relative ages of parents and the mortality affecting them, while the presence of resident institutions provided for orphan and destitute children will tend to inflate the orphanhood proportions in the areas in which they are situated. The distribution of the several classes by age in six broad geographical regions are given in the following table.


Census Ages of Husbands and Wives.

The ages of the 6,971,234 married couples who were enumerated on the same Schedule have been published by individual ages of husband and wife in combination in Table 5 of the Dependency volume. Tables 6 and 7 give a summary of the same facts, under a quinquennial grouping of ages, and these have been repeated as sections (a) and (b) of Table LXXII in this volume. Tables similar to the latter were published in the reports on the censuses of 1871, 1901 and 1911, but 1911 was the first occasion on which it was possible to publish the facts in detail for each year of age separately (Table 9 of the Fertility Report, Part I, of the 1911 Census).

The relation between the ages of husbands and their wives as returned in 1921 is, on the whole, very similar to that recorded in 1911, and as the analysis by single years of age which was introduced on that occasion led to its fairly exhaustive treatment in the report (pp. xi-xiv, Fertility Report, Part II, 1911) the following observations should be supplemented by a reference thereto for a full examination of the subject.

From the subjoined statement it will be observed that husbands are of the same integral ages as their respective wives in one eighth of the total couples examined. In one third of the total cases the difference is not greater than 1 year and in more than a half of the total their ages do not differ by more than 2 years. As the difference in age increases, the number of couples diminishes, the drop being very much steeper on the side representing the cases in which the wife is the older partner. Of the couples whose respective ages differed by 5 years or more, the proportions representing "husband older" and "wife older" cases amount to 23.74 per cent. and 4.78 per cent. of the total couples enumerated.

Difference between Age (last birthday) of Husband and
Age (last birthday) of Wife.
Proportion of
Total Couples.
{ 5 or more years 23.74 22.75
4 years 6.83 6.86
3 8.68 8.65
2 11.24 11.72
1 12.98 12.95
    0 12.51 13.09
{ 1 8.75 8.50
2 5.17 5.40
3 3.18 3.17
4 2.15 2.17
5 or more years 4.77 4.74
  100.00 100.00

Altogether in 63.47 per cent. of the total, the husbands of 1921 are of higher integral ages than their wives and in 24.02 per cent. the wives are the older, whereas in 1911 the corresponding proportions were 62.93 per cent. and 23.98 per cent. These changes combined with the somewhat flatter difference distribution now shewn suggest that the average excess of husband's over wife's age in the general population is slightly greater than it was in 1911. A similar inference might have been deduced from a study of the changes in the married section of the community which have taken place during the decennium. The addition to the class in the shape of new marriages has been characterized by an increasing difference between the age of bridegroom and bride, particularly during the later portion of the period, while among the marriages terminated during the decennium the reduction due to war losses in respect of comparatively young married men has occurred among a section in which the difference in age between husband and wife is less than the normal, thus tending to widen the corresponding difference in respect of surviving couples.

Comparing the proportions for 1911 and 1921 in respect of the several age differences shewn in the above statement, it will be observed that the most frequent combination in 1921 is the one in which the husband is one year older than the wife and that the position in this respect appears to have changed since 1911 when the greatest frequency was recorded for couples of the same integral age. At neither date are the frequencies of these two combinations very different from one another and any significance in the reversal of their respective positions is diminished by the possibility that it is due in part, at any rate, to the improvement in age statement which is recorded in 1921. This has been more fully referred to on page 72. With this improvement there has been, as one would expect, a corresponding modification of the tendency, clearly observable from the 1911 returns, for the age of one partner to be fixed by reference to that of the other with its consequent opportunity for the preference of a difference in round numbers. The tendency may still be seen in the 1921 figures, but it is less than it was 10 years ago, and to this change may probably be ascribed the consistency of the increases in the frequencies at odd differences of age in the above statement and the equally consistent decreases in the frequencies at the even differences.

The following table shows how the relative age position of husband and wife varies with the attained ages of either party to the marriage.


The earlier ages are representative of the shorter duration marriages and the proportions are influenced primarily by the ages at marriage. At the youngest age period identified in the table both husbands and wives are more frequently found married to partners older than themselves, but even here the proportion of young husbands married to younger partners is very much greater than the proportion young wives married to still younger husbands, so that the tendency for the difference in age to be in favour of an older husband may be said to be manifested from the earliest marrying ages. The early adult years following age 20, the period at which the majority of marriages takes place, are characterized by a large excess in the "husband older" proportions in both sections of the table, but later in life mortality becomes an increasingly important factor so that at the advanced ages of wives as well as husbands their partners are predominantly younger than themselves, the surviving marriages having been exposed to less risk of dissolution by death of the partner the lower the partner's age. Re-marriages, so far as their conditions differ from those of first marriages, may also influence the proportions at the later ages but their effect must be small compared with that of mortality.

3. Fertility.

The subject of fertility has been included with Dependency and Orphanhood in the volume devoted to those subjects, not because it bears any necessary relationship to that of dependency or orphanhood, but because the statistics relating thereto and shown in Tables 10-13 of the volume have all been derived returns.

The express fertility enquiry of the 1911 Census was omitted in 1921, notwithstanding its importance, in view of the long range covered by the 1911 investigation and of the fact that the full examination of the material then collected was suspended during the war period and had not been completed when the 1921 Census questionnaire was prepared. Though it has not been possible, in any sense, to repeat the earlier investigation, the form of the dependency return was such as to provide within limits a series of comparative fertility ratios; and in view of the inadequacy of the records available under our present system of birth registration in regard to several aspects of fertility, it was felt that a contribution based upon the extensive data obtained at the census, even if subject to important limitations, cannot but form a useful addition to the meagre statistics available in respect of fertility in this country.

The function on which the tables have throughout been based is the ratio:—

Number of children under one year of age at date of census
Number of married men (or married women) at date of census

as given by the dependency figures. This has been calculated and is shown in respect of various categories of married men and women differentiated according to age, size of existing family, place of enumeration, and, in the case of married men, occupation.

Comparing this function with the fertility ratio more commonly used in current vital statistics, viz., the ratio:—

Number of children born in a year
Average number of married men (or women) exposed to risk during the year

it will be seen that both the numerator and denominator of the former may be regarded as corresponding to the similar functions composing the latter fraction, but at a point of time six months later on the average. In each case the children under one year of age at the date of the census will all have been born within the preceding 12 months and will be approximately six months old on the average at the date of enumeration, but their numbers will be less than the full number born by the losses due to the comparatively heavy mortality in the first few months of life, and possibly to migration, though this factor cannot be of significance at this period of life. The parents, male and female, will similarly be six months older on the average than they were at the births of their children and will similarly have been reduced in numbers by mortality and migration. Against these reductions, however, in the parent classes must be placed the gains by reinforcement in the shape of new marriages and since the balance of losses and gains tends in general to be towards a steady increase in the numbers of married men and women, the denominator of the fertility ratio now in question will tend to be overstated though not to the same relative extent as the understatement in the numerator caused by infant mortality in the case of the children. So far as the resulting fertility ratio is concerned, the two defects, a deficiency in the numerator and an excess in the denominator, operate in the same direction and lead to an understatement of the fertility rate throughout.

Further, since the fertility tables are by the nature of their origin conditioned by the form of the dependency return, not only are they subject to the errors of the data, but they contain an element foreign to their subject owing to the fact that parents were asked to include step-children among their dependent children. These will accordingly be shown in association with parents possibly exhibiting different characteristics (e.g. regarding age, occupation, etc.) from those which were to be expected or had actually been present in the case of their natural parents. Evidence of _this is forthcoming from the face of the tables themselves, where a number of children under one are credited to parents of ages quite outside the normal fertility period.

The extent to which the observations may be said to be biassed by reason of the facts that they have been obtained as an indirect product of an enquiry directed to other purposes and that the number of children under one year of age is less than the full number which would have been enumerated had the returns been complete (as to which reference may be made to the dependency section of this report) may be approximately measured by multiplying the fertility ratios shown in Table LXXX by the numbers of married men or married women of the appropriate ages in the total enumerated population and comparing the sum of the products so obtained with the number of legitimate births registered between the middle of 1920 and the middle of 1921. A test made in this way discloses a deficiency in the number of constructed births of about 14 per cent. of its number, about half of which may be ascribed to the infant mortality referred to above and the balance to other causes, the chief of which is the omission of infants from the census returns. The fertility rates presented are thus relative and not absolute; and even in their relative character they will tend to be affected by differences in the incidence of infant and parent mortality, migration, etc., in the various sections of the population distinguished, sometimes seriously but generally not to such an extent as to destroy their fundamental relativity or to impair it in a high degree. Moreover, the nature of the disturbing influences can be ascertained from other sources such as the Registrar General's Annual Review and compensation introduced, if necessary.


The defects described, part of them inevitable in the adaptation of the material for fertility purposes, undoubtedly detract somewhat from the full value of the analysis, but it is believed nevertheless, that the tables will, with liberal and intelligent treatment, provide a useful study of the comparative incidence of fertility among various sections of the population of England and Wales during the year preceding the 1921 Census.

In Table LXXX the relation of fertility to the age of the parent is shown both for married men and married women, the experience of the former embracing the whole of the married men in the country and that of the latter, those married women—93 per cent. of the total in the country—who were enumerated on the same schedule as their husbands.

In respect of each sex the ratios rise sharply to a maximum which is recorded in the table at age 19, after which there is a decline, remarkably even and continuous to about age 45, at which age the figures have been reduced to comparatively unimportant dimensions for both males and females, but much more so in the case of females. After 45, female fertility rapidly becomes insignificant and disappears, but in the case of males the earlier decline in the ratios slackens somewhat and the figures, though of a small order, are sufficiently uniform to suggest that for men a degree of fertility is maintained even up to comparatively advanced ages. So large a proportion of the total births are provided by parents between the ages of 20 and 45 and so few outside that range of years (in the present experience, children associated with fathers below 20 or over 45 form 7.9 per cent. of the total and with mothers of similar ages 2.6 per cent. of the total) that particular interest lies in the progression of the ratios over the reproductive period 20-45. Within these ages the male figures, as recorded above, vary from .339 to .053 and the female within the slightly wider limits of .356 and .019 and it will be observed that the decline in each case is so regular that if the changes were expressed in diagram form, the curves would, apart from slight local irregularities, take the form of approximately straight lines, indicating a simple and close relation between the age of the parent and the fertility as expressed by the present experience of married men and women in this country.

No account has been or could have been taken, in this experience, of effect of the duration of marriage upon the incidence of reproduction, since the date of marriage was not asked for and the duration was not available directly or indirectly from the census schedules. Marriages of all durations are represented in the tables and such differentiation as exists in this respect must be inferred from the ages themselves or from the record of marriages published each year in the Statistical Review of the Registrar General, In a broad sense, the younger men and women will have been subject to a shorter duration of marriage on the average than those in older classes, and such selective effect in the matter of fertility as may be associated with the initiation of marriage will be more strongly represented in the former, so that the correspondence of the fertility actually recorded, and what may be termed the full physiological fertility, may be expected to be closest at the youngest ages, where marriages must of necessity be comparatively fresh, and to diminish as age advances, and, with it, an increase in the proportion of long duration marriages in which families have attained the desired dimensions and their further extension subjected to more definite restraint.

It may be noted that, in relation to the duration of marriage, fertility, as measured by births, accrues from a date later than the actual date of marriage, corresponding to the time necessarily absorbed in reproduction, and that the births, measured at annual intervals from the date of marriage, will be apparently deficient in the first year owing to the limitation of the opportunity of reproduction to the latter end of that year. Whatever its ultimate effect, the immediate result of the introduction of new marriages in a sample of population will, therefore, tend to check the ensuing fertility for a while. This effect is of importance in the rates shown in the table for the lowest parent ages for, from the following statement comparing the census returns of married men and women and the numbers of males and females married at corresponding ages in 1920 or 1921, it may be gathered that the bulk of the marriages at ages below 22 have subsisted for less than a year and that insufficient time has elapsed, therefore, for a full expression of the fertility characteristics.

Age. Men. Women.
as married
at the Census.
Married during Enumerated
as married
at the Census.
Married during
1920. 1921. 1920. 1921.
Under 16 46 1 77 34 29
16 84 10 15 323 256 247
17 240 207 173 1,867 1,801 1,737
18 1,391 1,535 1,304 7,807 8,344 7,844
19 5,188 5,067 4,307 21,071 17,547 15,880
20 15,425 10,973 9,653 41,866 26,299 22,123
21 31,644 25,853 22,650 67,547 38,772 33,785
22 49,159 28,441 24,537 90,934 35,187 29,669
23 69,080 32,327 27,017 117,609 35,156 28,995

The rise in the curves between the ages of 15 and 19 is at first sight suggestive of a gradual attainment of maturity in the case of the very young of both sexes, but it must be borne in mind that the section of population examined is the married section, amongst whom there must be a very strong degree of selection against immaturity, and it is probable, therefore, that the recorded fertility is influenced far more by the infusion of new marriages, which checks the curve without altering its direction at ages over 19, but below that age depresses the ratios below the maximum ultimately recorded.

Though Table LXXX discloses a close relation between the fertility experienced and the age of both male and female parent, it is not possible to determine therefrom the relative influence of these two factors separately, because of the strong tendency for marriage unions to be composed of partners of approximately similar age and for the resulting fertility records shown in relation to each of the parents singly to be subject to the mutual reflection arising from their common age association. The analysis is extended, therefore, in Table LXXXI, with restriction of the age detail to quinquennial periods of age, to show the variations in relation to each of the variables independently of the other, the ratios being shown in respect of each combination of husband's and wife's age. The experience of married men, as well as married women, in this instance is limited to that of husbands and wives enumerated together at the census.


It is possible to observe from this table the effect of a change in either the age of the husband or wife irrespectively of the other, the figures when read vertically down ward recording variations with increasing age of husband and when read horizontally from left to right, the corresponding variations due to changes in the wife's age.

For a given age of wife, the frequencies in each column are in variably at a maximum for the youngest -and at a minimum for the oldest group of husbands represented. The intermediate frequencies, however, do not exhibit the same regularity in decline as that shown for husbands tabulated irrespectively of the age of their wives in Table LXXX. Following an initial fall which is usually greater than at any subsequent age, the decline exhibited in the several columns appears to be interrupted by a period of stability over which the ratios vary but slightly and even occasionally increase and only at a later period is the fall again resumed, this time with gradually increasing effect down to the oldest age. The married women's curves—reading the horizontal rows from left to right—show no such systematic interruption in the decline of the ratio with advancing age. In association with husbands between 20 and 30 years of age the fall in the wife's fertility with her advancing age is rather greater at the beginning and the end of the curves than it is over the middle of the range, but both here and in the curves associated with older husbands, the age variations, though not so uniform as they are in the aggregate curve of Table LXXX, are constant in direction, with a pronounced fall at each succeeding age interval after the first. The latter exception is associated with wives of the youngest ages in combination with husbands aged 35 and over, and as the marriages represented by this portion of the experience will be of the shortest durations, the explanation of the initial rise in fertility is probably similar to that suggested for the movement shewn for individual ages under 19 in Table LXXX.

Such conclusions, regarding the influence of the age of each parent upon ensuing fertility, as can be drawn from an experience which is based upon attained ages of married men and women and in which no account can be taken of the important factors of duration of marriage or ages at marriage, appear, so far as they go, to coincide with the different, but more detailed, investigation of 1911, in showing that, in respect of women, the association between age and fertility is well marked over the whole of the child-bearing period, and is of high significance throughout, whereas in the case of men, the fertility variation is slight over the period during which the bulk of the births are produced and exhibits a uniformity in significance only at the later years of life.

In Table LXXXII an attempt has been made to show, for parents of different ages, how fertility is influenced by the existence of earlier born children who may be regarded as not having attained the age of independence.


The figures have been derived from the dependency returns by classifying the families according to the number of children between 1 and 16 years of age and obtaining the ratio of the number of children under 1 year of age to the total for each size identified the assumption—sufficiently accurate for the purpose—being made that the numbers between age 1 and age 16 at the date of the census were the same as the numbers under age 15 a year earlier.

The characteristic feature of this analysis is the somewhat surprising tendency for the ratios to increase as the families become larger. The rule is almost invariable among families of which the existing children number 2 or more, and applies to smaller families also in the case of married men and women at ages over 35 though the increases in the ratios for sizes of family 0 to 2 are less than they are for those of larger dimensions. For men and women of the younger ages—age groups up to 35— the fertility ratio declines until the 2-children families are reached, a decline which is greatest in respect of the youngest parents and which diminishes with the advancing age of the parent until, as already stated, the decline is ultimately converted to an increase at ages 35 and over. It may be that the analysis, by size of family itself involves a grading in respect of past fertility, the 3-child families automatically securing the representation of a more fertile stock than that of 2-child families, and so on, and that the continuing fertility indicated by the probability of a further addition to the existing family does little more than express, in terms of the future, the relation inherent in the analysis by reason of its statement in relation to past fertility.

The size of the family in the case of married men and women is, however, influenced not only by the fertility of the parents, but also by the duration of marriage and where, as must be the case with parents of the youngest ages, the latter is the predominant factor in limiting the size of family, the curve expressing the relative probabilities of further additions to existing families may be expected to follow a different course from that based upon families where the effect of duration is small and the grading is more definitely influenced by fertility itself. This may afford an explanation of the gradual change in the curves from a steeply declining fertility with increasing size of family in respect of the youngest parents of each sex, first to a lesser decline and ultimately to an increase when the parents are aged 35 and over, the influence of the duration of marriage as a factor affecting the size of family, presumably diminishing with the increase in the ages of the parents examined.

But it is to be borne in mind that the size of family distinguished here, refers only to surviving children under the age of 15 and takes no account of the existence of older children who, if they could be brought into the picture, as they should be, if the grading by size is to be assumed to be indicative of past fertility, might be expected to modify materially the distribution of families on which the ratios of Table LXXXII are based. The families of younger parents will be least affected by the exclusion of the older children since, in their case, the number of children over 15 years of age must, in any event, be insignificant, but at the middle years of life the modification may be sensible, and in respect of the oldest parents the majority of the children will have been excluded by reason of their being beyond the limiting age, so that in their case the classification of families by the number of children under 15 may bear no resemblance whatever to the history of their past fertility and will not suffice, therefore, to explain the consistent rise in the current fertility ratios as the families become larger.

Another differentiating characteristic attending the grouping of families by the number of children under 15 will be found in the age constitution of children brought within the several groups. The material providing the necessary information will be found in Table 1 of the Dependency Volume from which it may be deduced that the average age of the youngest child in each family is highest in families comprising the smallest total number of children and that this average declines as the families become larger, reaching a minimum in respect of the largest families. At the same time, the disparity between the ages of the youngest in small and large families respectively increases with the age of the parent, the ultimate position of respect of the oldest parent being one in which the average age of the youngest child in 1-child families approaches the upper limit of the children's age field and in families of 10 or more children each, remains little above the lower limit. The grouping of families by size, therefore, introduces indirectly a grading by age of youngest child, or, in other words, by the period elapsed since the birth of the last child, and since from this point of view fertility is relatively high when that period is short, and low when it is long, an alternative explanation of the trend of the curves appears to be forthcoming in the association of a progressive development of sterility with the continued non-exercise of the reproductive function.

The material on which the present study is based is not capable of the analysis necessary if definite conclusions are to be formed regarding the incidence of the several factors to which fertility may be regarded as subject and the above examination is mainly instrumental, therefore, in indicating the probable importance of some of them, among which may be placed, the age at marriage, the duration of marriage, and the interval between succeeding births, and in pointing the desirability, in future enquiries devoted to the subject, of obtaining information which will enable the facts to be classified in such a way as to permit of the isolation and separate examination of each of the several variables independently.

Tables 12 and 13 of the "Dependency" Volume showing the relative Incidence of fertility in the case of each parent by area of enumeration (counties) and by occupation in respect of male parents are summarised in Tables LXXXIII and LXXXIV.


The age rates in respect of the several sections of the population identified are supplemented in the last two columns of each table by percentages comparing in an integral form the local experience with that of total population in all sections combined. The first compares the crude fertility ratios, making no allowance for variations due to differences in the age constitution of the parents in the several groups, while in the second, this variable is eliminated by a method of standardisation, familiar to those acquainted with the publications of this Department, which consists of adopting, as a standard, the age rates in respect of the population as a whole, applying the standard rates, age by age, to the local population and so obtaining the number of births which would have occurred had the standard rates been operating and with which the actual recorded births can be and are compared. The occupation table, it will be observed, is arranged in the order given by the standardised comparison.


Next Selection Previous Selection