female births comes out in the aggregates. The commissioners for 1841 claim to have made some contribution to the theory of that remarkable phenomenon. In the census of 1851 a gloomy feature was the decrease of population by almost 20 per cent, due to the famine of 1847. A decrease of 29 per cent is recorded for directors of labour, the second of the threefold division of occupations above mentioned. The decrease of population which set in at this period has continued ever since; the rate of decrease continually diminishing down to the last decade; when, as shown by the Preliminary Report for 1891, the decrease sprung from 44 per cent in the decade 1871-1881 to 9.1 per cent. Much the same distribution as in 1841 is presented by another classification of occupied persons as those ministering to food, clothing, and six other primary wants. In 1861 this classification is modified by the addition of some new categories, in all twelve. For example, out of 10,000 occupied persons 18 minister to religion, 5 to amusement, 1 to science and art, and 1816 to food. A feature of peculiar interest in the census of 1861 was the inquiry concerning the religious profession of the people. The proportions proved to be, Roman Catholics 77.69 per cent; Protestants 22-23 per cent; a small margin of not quite 1 per cent being left for the Jews and the unspecified. Of the Protestants a little more than half belonged to the Established Church. These proportions have been approximately maintained in succeeding enumerations. The latest returns give Roman Catholics 75.4 per cent, Protestants (Episcopalians, Presbyterians, and Methodists) 23.5 per cent; leaving for Jews and other persuasions, as before, about 1 per cent (census of Ireland for 1891: Preliminary Report). In 1871, as in 1861, the returns as to occupation are arranged both according to the English sixfold classification and the system grounded on the wants to which different occupations minister. The Irish commissioners defend the indigenous system with much warmth. The census of 1881 conforms to the English classification with more docility. The later censuses of Ireland deserve commendation for the completeness of the agricultural returns.

(1)

A general idea of the scope and purpose of a census may be obtained by noticing the salient points in a particular report; the latest completed for England and Wales, namely that for 1881; the returns for 1891 not been fully analysed at the time of writing. The first heading is Number of the population and rates of increase. The number falls short of twenty-six million by only a few thousands. The rate of increase per cent is 14.36 per decade; greater by 1.17 than for the preceding decade. The increase, determined by actual enumeration, differs little from the "natural increment"

from the number of births. Hence two methods of forecasting the future increase of population. Either we may observe the natural increment for the year or two immediately adjacent to 1881, and assume that this rate will remain constant for the following decade; or we may assume that the actual rate of increase for the decade 1871-81 will continue to prevail for the next decade. A variant of the second method takes account of the fact that the decennial rate of increase is itself increasing. The average of the results obtained by these three methods gives, as the probable population in 1891, 29,843,898. This prophecy has not been fulfilled. The "preliminary report" for 1891 records 29,001,018 as the population of England and Wales in this year. Taking account of these circumstances, Dr. Longstaff made a more accurate prophecy: namely that the result of the enumeration in 1891 would be 29,012,776 (Economic Journal, No. 2). It is interesting to observe that the number prophesied in 1871, viz. 26,006,098, corresponds fairly well with the number enumerated in 1881, viz. 25,974,439. The increase of the town population is found to be more rapid than that of the country. (2) Density of the popula tion and habitations is the second heading. The density, whether expressed as the mean distance between individuals, or in more familiar terms, is found to vary from place to place and to increase from age to age. The proportion between the number of persons and the number of habitations is less variable. (3) Sexes. The excess of females over males is more than 5 per cent, and is increasing.

(4) Ages.

At this stage we reach less solid ground, and "must proceed with much care and circumspection.' The returns of age are falsified by various causes. Many are ignorant and put down a round number at random. Then there is the confusion between years of life and years completed, "the twenty-first year of life and twenty-one years old." The very old desire to appear still older. Very young women overstate their age with a view to obtaining employment; women not very young understate their age. From this double cause the category of female ages between fifteen and twenty-five is enormously exaggerated. All that can be done is to distribute the answers in periods of five years; and, assuming that these totals are correct, to determine by interpolation the numbers for each year. (5) Condition as to marriage or civil condition. The immense army of nearly four and a half million wives supplies an effective force of nearly three million "of reproductive age," between fifteen and fortyfive. The number of single women and widows of this age-period almost exactly equals the number of wives. The proportion of married persons to the adult population is declining.

(6) "The most laborious, the most costly, and after all, perhaps, the least satisfactory part of the census" is that which is concerned with the occupations of the people. One difficulty is obscurity of the answers given. What sort of occupation is designated by an "all-rounder," or a "barker"? The same designation may have different meanings. Thus clothier means in some parts cloth-maker, in others clothdealer. There is also the difficulty of multiple occupations. In the census of 1881 the sixfold classification framed by Dr. Farr is adopted, with certain modifications. To reproduce the figures in detail would be out of place here. The following round numbers strike the imagination and impress the memory. The number of persons engaged in transport is a million. The number employed on textile fabrics is a million. The number employed on dress is a million. The number of labourers (agricultural and others) is a million and a half; of miners nearly half a million. The numbers of the professional classes and of those employed in preparing food are less round, each nearly 650,000. (7) Under the head Birthplaces of the population we learn what proportion of the inhabitants have filtered into England from other countries: about 1 per cent from Scotland, 2 per cent from Ireland, and from abroad only 67 per cent. (8) Infirmities comprise blindness, deaf-mutism, and the various forms of mental unsoundness. It is gratifying to find the relative numbers of the blind (about 1 per 1000 of the population) diminishing. That men should become blind oftener than women is not surprising. That congenital defects both of body and mind should be commoner in the stronger sex is less in conformity with preconceptions. (9) The ninth section of the report summarises the results for The United Kingdom. (10) The last section relates to The British Empire. "The territory occupied by the 254,187,630 inhabitants of the British empire is estimated as consisting of slightly over eight millions of English square miles," a territory more than sixty-six times as extensive as the surface of the United Kingdom.

The

A more recent, but less complete, model of a census is presented by the preliminary report of the census of 1891, for England and Wales, which has just been issued. The Report begins by stating the total number of persons returned as living in England and Wales at 12 P. M. on 5th of April 1891: namely 29,001,018. deficiency of this number, as compared with the estimate formed in 1881 (see above), is accounted for partly by a diminution in the natural increase partly by an increase in the net emigration from England and Wales during the decade. A deficiency of 288,782 is ascribed to the diminished natural increase, of 414,568 to the increased excess of emigrants. The decline in the natural increase was not due to increased

241

mortality but a decrease in the birth-rate. The number of families, 6,146,901, has not increased in proportion to the population; the number of inhabited houses, 5,460,976, has increased in a rather larger proportion; but there is some uncertainty about both figures. The rate of increase of population varies in different areas; speaking generally, being highest in the counties adjacent to London and counties in which coal-mining is the predominant industry. Fourteen rural counties show an actual decrease. The general law that a district increases more rapidly the more decided its urban character is not overthrown by the figures which show a smaller percentage increase for districts with populations above 100,000 than for districts with populations between 20,000 and 100,000. For "probably, or at any rate possibly," the contradiction is only apparent, the newcomers in the larger towns having to settle outside the municipal or official boundary. Accordingly the decrease in Liverpool of 6.3 per cent, and the falling off of increase in other larger towns, need not imply a decline in prosperity. The population of London, "the London of the Registrar-General," has increased in a somewhat lower rate than the population of England and Wales as a whole-the first time that such a phenomenon has presented itself.

The report concludes with summary tables relating to the other parts of the United Kingdom and the islands in the British seas. The following table shows the figures for the United Kingdom at the last two censuses.

Enumerated population of the United Kingdom

The census of 1901 is so recent that we can only give the totals.

The rate of increase of the United Kingdom in the decade 1881-1891 was 8.2 per cent; lower than in either of the two preceding decennia, in the first of which, 1861-71, it had been 8.8; in the second, 1871-81, 10-8 per cent. The falling off in the rate of growth was shared by all the divisions of the kingdom, but unequally. The population of the islands in the British sea enumerated in 1891 was 147,870; showing an increase of 4.7 per cent as compared with the enumeration of 1881; and a considerable increase in the rate of growth, there having in fact been a decrease in the period 1881-1891.

Valuable as these results are, and though the British census may be described as a good article at the price (about 1d. per head counted in England and Wales, 1871), there is room for many improvements. The reform most

the scientific world by the celebrated Wargentin, were utilised by Dr. PRICE (q.v.) and MILNE (q.v.)(Annuities, ch. xii.-xiii.) MALTHUS (q.v.) builds upon these statistics (Essay, bk. ii. ch. ii.), noticing the correlation between the lean years and the diminution of marriages. In the recent publications of the Swedish Bureau the study of such correlations is facilitated by affix

The

has now become a fortiori, owing to the imperfections, disclosed by the census for 1891. Dr. Noel Humphreys, in his "Results of the Recent Census and Estimates of Population in the largest English Towns" (Journal of the Statistical Society, June 1891), gives instances of towns in which the hypothetically estimated death-rate differed from the true figure by as much as 20 and even 26 per cent. The estimates based on the decennial census were so worthless that they seemed to evidence a decline in the death-rate of Salford and Liverpool, while in reality there was an increase. Statistical Society has also, through the mouth of various committees, advised that the character, as well as the number, of dwellings should be returned. As Mr. Palgrave shows (Journal Statistical Society, 159, p. 411), information as to the house accommodation of the people is very deficient, and more complete returns would be very valuable. Another recommendation sanctioned by the Statistical Society is that the religious profession of each inhabitant should be obtained. Mr. Longstaff, in his "Suggestions for the Census of 1891" (Journal of the Statistical Society, 1889, p. 437), proposes that the number of rooms in dwelling-houses should be recorded. He would discontinue inquiries as to persons of unsound mind. Mr. Longstaff reinforces the arguments in favour of a quinquennial census. An annual census has been proposed by the late Sir Edwin CHADWICK (Journal of the Statistical Society, 1889, p. 468; Internat. Stat. Congress, The Hague, pt. ii. p. 164). Major Craigie demands more conformity between the census and the returns of the agricultural department. "There must be large omissions in the ranks of land occupiers in our census. (Journal Statistical Society, 1887, pp. 98, 149, and Jubilee vol.)

The glory of instituting the first census does not belong to England. While the Population Bill of 1753 was denounced in the House of Commons as subversive of liberty, Sweden was already enjoying a tolerably perfect census. Sweden had the advantage of very complete parish registers, recording not only births, deaths, and marriages, but also the number of persons migrating to or from each parish and residing in it. By the agency of the clergy these materials were embodied in a census in 1749 and subsequent years, first at intervals of three, subsequently of five years. The results of the early Swedish census, communicated to

cating the quality of the harvest (see Journal of the Statistical Society, xxv. p. 141). Looking at the whole series, "the gradual diminution of mortality since the middle of last century is very striking" now, even more than when Malthus made this observation. The increase of suicides is less gratifying. The unrivalled length of the Swedish records affords particularly striking instances of those uniform relations which constitute what has been called physique sociale. The proportion of male to female births was 1.043 to 1 in 1751-60, 1.050 to 1 in 1851-55.

Spain has some claims to priority. A tolerably complete census of the dominions of Castile was taken as early as 1594. There was a general census of Spain in 1787 and in 1797. But then occurred a pause down to the census of 1857. During the first half of the century Spain lagged behind the statistical movement which was general in Europe.

The age initiated by the French Revolution was truly called by BURKE the " age of statists." In France a census was ordered in 1791 and executed in 1801. In Prussia a statistical bureau was founded in 1805. The impulse was followed by other states, at different times and with varying pace-some with quinquennial, some with triennial intervals. The inquiries have become more fruitful by the addition of new questions, and the fruit has become more accessible through increasing uniformity of the questions in different countries. The branches of the European census, as they lengthen, become not only broader, but more parallel. The possibility of international comparison is partly due to the action of the International Statistical Congress during the quarter century which followed its first meeting in 1853. extent to which its recommendations have been adopted may be expressed by the following statement. At the congress of St. Petersburg (1872) consolidating the results of discussions which had been carried on at several preceding congresses, twelve questions or rubrics (as to name, sex, age, etc.) are proposed as "essential," or obligatory. Of these recommendations the first seven are already adopted in twentyfive continental bureaus, with only one or two exceptions. The remaining five directions are carried out, roughly speaking, in about half of the continental states. Among the headings about which it is difficult to obtain uniformity may be instanced conjugal condition. At the Florence congress (1867) it was proposed to

The

ascertain the relationship of husband and wife. The question is put in Sweden and some other countries. Population, de fait and de droitThe Italian census grapples with the difficult questions connected with domicile by an elaborate arrangement distinguishing the birds of passage from the residents, and the temporary from the permanent absentees. Age-Considering the inaccuracy of returns under this head, is it worth while publishing them in full and without manipulation? Some European bureaus have decided in the affirmative. Language spoken-What does this heading mean? the language most used in good society, or in the family, or in the church? Occupations form a rubric which is the despair of the international statisticians. The diversity of employment in different countries is aggravated by the difficulties of language. What would a Frenchman understand by "clerk of the parish," even when translated (?) into "clerc de paroisse" (sic)? "Carvers,"

one of the designations in the St. Petersburg scheme, should not have been rendered in French as 66 agens des abattoirs." On these and other topics perfect unanimity is still only an ideal. Even when agreement has been obtained as to the questions to be asked, there still remains diversity in the arrangement of the answers. To smooth these remaining differences may, it is hoped, be a function of the International Statistical Institute, founded in 1885. As M. Korösi, one of the ablest advocates of uniformity, points out, the diverse practices mostly involve no question of principle; it would give no additional trouble to act in concert. It is needless to dwell on the advantages of increased uniformity. The census was used by ancient Rome as an engine of government; modern Europe should make it also an instrument of science.

[For the antiquities of the subject Gabaglio, Storia... della Statistica (2d ed., 1888), may be consulted. For the history and scope of the census in the United Kingdom, see the preface or report which is attached to every census except the first. Proposals for reform are expressed in the recommendations made by a committee of the Statistical Society before each census (see Journal of the Statistical Society for 1840, 1850, and subsequent decades). Mr. Palgrave's paper, Journal of the Statistical Society, 1869; and Dr. Longstaff's paper in Ibid. 1889 contain other proposals besides those which have been noticed. A general view of the earlier methods of taking the census on the continent is presented in the Journal of the Statistical Society, vol. iii. Mr. Hendriks' detailed accounts of the Spanish and the Swedish census, in the 23d and 25th volumes of the same journal, are very valuable. As to the history and scope of the census in the different European states, copious information will be found in the records of the International Statistical Congress, especially in the reports made by the members for each nation. In the same publication will be found repeated

243

proposals for securing greater uniformity. A conspectus of these recommendations is presented by M. Korösi in his Projet d'un Recensement du Monde (Paris, 1881). Attention is due to his own recommendations offered in that work and in his address to the Jubilee meeting of the Royal Statistical Society (Jubilee volume, 1885), and in his contribution to the second Bulletin of the International Statistical Institute, p. 200. A comparison of the statistics obtained by the census in different countries is afforded by Confronti Internazionali (published by the Direzione Generale della Statistica), Roma, 1884. Many of the figures there compiled have been extracted by Sir Rawson Rawson, who has added explanatory comments (Presidential address to the Statistical Society, Journal, 1885). The most recent results for several European countries and for the Cape Colony and Victoria are presented together in the Notes on the Preliminary Returns of the Censuses, 1890-91, in the Journal of the Statistical Society for Sept. 1891. Less recent comparisons are Movimento della Popolazione in Italia e in altri Stati d'Europa, by M. Bodio; Archivio di Statistica, 1876; Mouvements de la Population dans les divers états de l'Europe, by M. Bertillon; Annales de la Démographie Internationale, vol. i.; and other articles in that series (extending from 1877 to 1883). See also Quetelet, Physique Sociale, 1869.]

F. Y. E.

CENSUS, UNITED STATES. The mode of conducting the census of the United States is so important as to demand special notice.

History of the Census.-The origin of the census system in the United States lay in the necessity of establishing some basis for the distribution of representatives and direct taxation among the different states. When the colonies associated themselves together to resist the mother country it was necessary to find some method of distributing the expenses of the war. The congress which assembled in 1775 had no power to assess and collect taxes through officers of its own. The draft of the articles of confederation (1776) provided that expenses should be distributed according to the population of the states, which should be ascertained trienni

ally and transmitted to the assembly of the United States. The southern members objected to the slaves being counted with the whites, and in the articles of confederation finally adopted (1777-1781) it was provided that all expenses for war or the common welfare should be apportioned among the states "according to the value of all land within the state granted to or surveyed for any person, as such land, and the buildings and improvements thereon, shall be estimated according to such mode as the United States in congress assembled, shall from time to time direct and appoint" (art. 8). The land force, on the other hand, was to be furnished by the different states according to the number of white inhabitants in each (art. 9). Neither of these provisions ever led to a census; in fact in 1783 congress recommended

the levying of money in proportion to the population rather than according to landvalue, and drew up a table of the estimated population in each state. In this estimate only

three-fifths of the slaves were counted. This amendment did not prevail, but we see here the origin of the three-fifths rule which crept into the constitution of 1787 (art. 1, § 2), which is the legal basis of the census at the present time. That clause reads as follows:

"Representatives and direct taxes shall be apportioned among the several states which may be included within this union, according to their respective numbers, which shall be determined by adding to the whole number of free persons including those bound to service for a term of years, and excluding Indians not taxed, three-fifths of all other persons. The actual enumeration shall be made within three years after the first meeting of the Congress of the United States, and within every subsequent term of ten years, in such manner as they shall by law direct."

To carry out this provision of the constitution, congress passed the first census law, 1st March 1790. The law has since been so extended that from a mere enumeration, the census has become an elaborate description of the social and industrial condition of the people of the United States. This progress has in one sense been gradual, but the census of 1850 and that of 1880 stand out as marking epochs, setting the standard for those that followed.

:

The census of 1790 was a simple enumeration of the people under the following classification: free white males of 16 years and upwards, free white males under 16 years, free white females, all other free persons, and slaves. The census of 1800 was similar in scope except that the age classification was slightly more elaborate, namely under 10 years, from 10 to 16, from 16 to 26, from 26 to 45, over 45, with distinction of sex. The same population schedule was used in 1810; in addition questions were asked in regard to manufactures, but the latter returns were so incomplete as to be of no value. They were equally incomplete in 1820, and in 1830 the inquiry was omitted. In 1820 the age classification was further elaborated; "foreigners not naturalised" distinguished; and the people divided, according to occupation, under agriculture, commerce, and manufactures. The census of 1830 adopted the age classification under 5 years, from 5 to 10, from 10 to 15, from 15 to 20, from 20 to 30, and then by decennial periods. It also distinguished the blind, and deaf and dumb, by colour and certain ages. In 1840 the number of insane and idiotic, white illiterates, and statistics of schools were added. A manufacturers' schedule was again attempted, but without much success. The census of 1850 was more elaborate. included the number of dwelling-houses, of

:

It

families, of persons according to sex, age, colour, free or slave, profession or occupation, value of real estate, number of persons married within the year, attendance at school, illiteracy, the deaf and dumb, blind, insane, idiotic, pauper, and convicts, mortality statistics for the year, products of agriculture and industry, wages, capital, taxes, colleges and schools, newspapers and periodicals, and churches. Most of our comparisons at the present time go back to this census. In 1879 a new act was passed which resulted in the elaborate "tenth census of the United States," a description of which is given below.

Method of taking the Census.-The returns of 1790 were taken under the supervision of United States marshals and sent to the president of the United States. From 1800 to 1840 they were sent to the secretary of state and published by him. In 1850 the census office was created in the newly-established department of the interior; since then a superintendent of the census has been appointed for each census. The law of 1st March 1889, for the taking of the eleventh census (1890), provides for the appointment of (not more than) 175 supervisors who shall appoint enumerators, the latter to visit personally every family in their sub-divisions and by inquiry obtain all particulars required by the law, which also allows schedules to be distributed in advance, to be filled up by householders and others. The schedules to be employed will be (as in 1880) a population schedule, one for farms, for manufactures, for mortality and vital statistics, for social statistics, and for transportation companies. Where there is an official registration of deaths, the superintendent may withhold the mortality schedule from the ordinary enumerators and obtain the statistics through official records. The superintendent may withhold also the schedules for manufacturing, mining, and social statistics, and charge the collection of these statistics upon experts and special agents. He may also employ special agents to investigate the statistics of the manufacturing, railroad, fishing, mining, cattle, and other industries of the country, and of telegraph, express, transportation, and insurance companies as he may require. The census shall also take on a special schedule the names, organisations, and length of service of those who had served in the army, navy, or marine corps of the United States in the war of the rebellion, and who are survivors at the time of the inquiry, and widows of soldiers, sailors, or marines. The population schedule shall include an inquiry as to the number of negroes, mulattoes, quadroons, and octoroons. The superintendent shall also collect the statistics relating to the recorded indebtedness of corporations and individuals; also, information relating to animals not on farms. The cost of the census of 1890 is not to exceed