Part 6

_Ceylon_.--Here the census is taken decennially, on the same date as in India, in consideration of the constant stream of migration between the two countries. The schedule is much the same as in India with the substitution of race for caste. Until 1901, however, it was not filled in by the enumerator, as in India, but was distributed before and collected after the appointed date as in Great Britain.

_India_.--The population of India is the largest aggregate yet brought within the scope of a synchronous and uniform enumeration. It amounts to three-fourths of that of the British Empire, and but little less than a fifth of the estimated population of the world. Between 1853 and 1881 each province conducted its own census operations independently, with little or no attempt at uniformity in date, schedule or tabulation. In the latter year the operations were placed for the first time under central administration, and the like procedure was adopted in 1891 and 1901, with such modification of detail as was suggested by the experience of the preceding census. On each occasion new areas had to be brought within the sphere of enumeration, whilst the necessity for the use in the wilder tracts of a schedule simpler in its demands than the standard, grew less as the country got more accustomed to the inquiry, and the efficiency of the administrative agency increased. Not more than 5% of the householders in India can read and write, and the proportion capable of fully understanding the schedule and of making the entries in it correctly is still lower. From the literate minority, therefore, agency has to be drawn in sufficient strength to take down every particle of the information dictated by the heads of families. As it would be impossible for an enumerator to get through this task in the course of the census night for more than a comparatively small number of houses, the operation is divided into two processes. First a preliminary record is made a short time before the night in question, of the persons ordinarily residing in each house. Then, on that night, the enumerator, reinforced if necessary by aid drafted from outside, revisits his beat, and brings the record up to date by striking out the absent and entering the new arrivals. The average extent of each beat is arranged to include about 300 persons. Thus, in 1901, not far from a million men were required for enumeration alone. To this army must be added the controlling agency, of at least a tenth of the above number, charged with the instruction of their subordinates, the inspection and correction of the preliminary record, and the transmission of the schedule books to the local centre after the census has been taken. The supply of agency for these duties is, fortunately, not deficient. Irrespective of the large number of clerks, village scribes and state and municipal employés which can be drawn upon with but slight interruption of official routine, there is a fair supply of casual literary labour up to the moderate standard required. The services, too, of the educated public are often voluntarily placed at the disposal of the local authorities for the census night, with no desire for remuneration beyond out-of-pocket expenses, and the addition, perhaps, of a personal letter of thanks from the chief official of the district. By means of a well-organized chain of tabulating centres, the preliminary totals, by sexes, of the 294 millions enumerated in 1901 were given to the public within a fortnight of the census, and differed from the final results by no more than 94,000, or .03%. The schedule adopted contains in addition to the standard subjects of sex, age, civil condition, birthplace, occupation and infirmities, columns for mother-tongue, religion and sect, and caste and sub-caste. It is printed in about 20 languages. The results for each province or large state are tabulated locally, by districts or linguistic divisions. The final compilation is done by a provincial superintendent, who prepares his own report upon the operations and results. This work has usually an interest not found in corresponding reports elsewhere, in the prominent place necessarily occupied in it by the ethnographical variety of the population.

FOREIGN COUNTRIES

Inquiries by local officials in connexion with measures of taxation, such as the hearth-tax in France, were instituted in continental Europe as early as the 14th century; but as the basis of an estimate of population they were intrinsically untrustworthy. Going outside Europe, an extreme instance of the results of combining a census with more definite administrative objects may be found in the census of China in 1711, when the population enumerated in connexion with a poll-tax and liability to military service, was returned as 28 millions; but forty years later, when the question was that of the measures for the relief of widespread distress, the corresponding total rose to 103 millions! The notion of obtaining a periodical record of population and its movement, dissociated from fiscal or other liabilities, originated, as stated above, in Sweden, where, in 1686, the birth and death registers, till then kept voluntarily by the parish clergy, were made compulsory and general, the results for each year being communicated to a central office. A census, as a special undertaking, was not, however, carried out in that country until 1749. The example of Sweden was followed in the next year by Finland, and twenty years later, by Norway, where the parish register was an existing institution, as in the neighbouring state. Several other countries followed suit in the course of the 18th century, though the results were either partial or inaccurate. Amongst them was Spain, though here a trustworthy census was not obtained until 1857, or perhaps 1887. Some of the small states of Italy, too, recorded their population in the middle of the above century, but the first general census of that country took place in 1861, after its unification. In Austria, a census was taken in 1754 by the parish clergy, concurrently with the civil authorities and the military commandants. Hungary was in part enumerated thirty years later. The starting-point of the modern census, however, in either part of the dual monarchy, was not until 1857. Speaking generally, most of the principal countries began the current series of their censuses between 1825 and 1860. The German empire has taken its census quinquennially since its foundation, but long before 1871 a census at short intervals used to be taken in all the states of the Zollverein, for the purpose of ascertaining the contribution to the federal revenue, the amount of which was revisable every three years. The last great country to enter the census field was Russia. From 1721, what are known as revisions of the population were periodically carried out, for military, fiscal and police purposes; but these were conducted by local officials without central direction or systematic organization. In 1897 a general census was taken as synchronously throughout the empire as was found possible. It embraced a population second to that of India alone, as China, probably the most populous country in the world, has not yet been subjected to this test. The inquiry was made in great detail, under central control, and on a plan sufficiently elastic to suit the requirements of so varied a country and population. As in India, the schedules had to be issued in an unusual number of languages, and were dealt with locally in the earlier stages of tabulation. The principal regions of which the population is still a matter of mere conjecture are the Turkish empire, Persia, Afghanistan, China and the Indo-Chinese peninsula, in Asia, nearly nine-tenths of Africa, and a considerable portion of South America. (J. A. B.)

UNITED STATES

Modern census-taking seems to have originated in the United States. Professor von Mayr declares in a recent and authoritative work, "It was no European state, but the United States of America that made a beginning of census-taking in the large and true sense of that word," and Professor H. Wagner, writing of the censuses of Sweden, said to have been taken in the 18th century, uses these words, "Since 1749 careful parish registers have been kept by the clergy and have in general the value of censuses." The same authority, although mentioning a reported census of Norway in 1769, indicates his conviction that the first real census of that country was in 1815. Sweden, Norway and the United States are the only countries with any claim to have taken the first modern census, as distinguished from a register of tax-payers, &c., the lineal descendant of the old Roman census, and the innovation seems to be due to the United States. If so, the first modern census was the American census of 1790. At the present date more than three-fifths of the estimated population of the world has been enumerated in this way. It is of interest accordingly to note how and why the device originated.

The Federal census, which began in 1790 and has been taken every ten years since under a mandate contained in the Constitution of the United States, was the outgrowth of a controversy in the convention which prepared the document. Representatives of the smaller states as a rule claimed that the vote, and so the influence, of the states in the proposed government should be equal. Representatives of the larger states as a rule claimed that their greater population and wealth were entitled to recognition. The controversy ended in the creation of a bicameral legislature in the lower branch of which the claim of the larger states found recognition, while in the upper, the Senate, each state had two votes. In the House of Representatives seats were to be distributed in proportion to the population, and the convention, foreseeing rapid changes of population, ordained an enumeration of the inhabitants and a redistribution or reapportionment of seats in the House of Representatives every ten years.

The provision of the Constitution on the subject is 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."

In 1790 the population was reported classed as slaves and free, the free classed as white and others, the free whites as males and females, and the free white males as under or above sixteen years of age. In 1800 and 1810 the same classification was preserved, except that five age-groups instead of two were given for free white males and the same five were applied also to free white females. In connexion with the census of 1810 an attempt, perhaps the earliest in any country, was made to gather certain industrial statistics showing "the number, nature, extent, situation and value of the arts and manufactures of the United States." In 1820 a sixth age class was introduced for free white males, an age classification of four periods was applied to the free coloured and the slaves of each sex, and the number of aliens and of persons engaged in agriculture, in manufactures and in commerce was called for. The inquiry into industrial statistics begun in 1810 was also repeated and extended.

In 1830 thirteen age classes were employed for free whites of each sex, and six for the free coloured and the slaves of each sex. The number of aliens, of the deaf and dumb and the blind were also gathered.

The law under which the census of 1840 was taken contained a novel provision for the preparation in connexion with the census of statistical tables giving "such information in relation to mines, agriculture, commerce, manufactures and schools as will exhibit a full view of the pursuits, industry, education and resources of the country." This was about the first indication of a tendency, which grew in strength for half a century, to load the Federal census with inquiries having no essential or necessary connexion with its main purpose, which was to secure an accurate enumeration of the population as a basis for a reapportionment of seats in the House of Representatives. This tendency was largely due to a doubt whether the Federal government under the Constitution possessed the power to initiate general statistical inquiries, a doubt well expressed in the 9th edition of the _Encyclopaedia Britannica_ by Francis A. Walker, himself a prominent member of the party whose contention he states:--

"The reservation by the states of all rights not granted to the general government makes it fairly a matter of question whether purely statistical inquiries, other than for the single purpose of apportioning representation, could be initiated by any other authority than that of the states themselves. That large party which advocates a strict and jealous construction of the constitution would certainly oppose any independent legislation by the national Congress for providing a registration of births, marriages and deaths, or for obtaining social and industrial statistics, whether for the satisfaction of the publicist or for the guidance of the legislature. Even though the supreme court should decide such legislation to be within the grant of powers to the general government, the distrust and opposition, on constitutional grounds, of so large a portion of the people, could not but go far to defeat the object sought."

The difficulty stated in the foregoing quotation, although now mainly of historic importance, exerted great influence upon the development of the American census prior to 1900.

The pioneer work of the census of 1840 in the fields of educational statistics, statistics of occupations, of defective classes and of causes of death, suffered from numerous errors and defects. Public discussion of them contributed to secure radical modifications of scope and method at the census of 1850. Before the census law was passed, a census board, consisting of three members of the president's cabinet, was appointed to draft plans for the inquiry, and the essential features of its report prepared after consultation with a number of leading statisticians were embodied in the law.

The census of 1850 was taken on six schedules, one for free inhabitants, one for slaves, one for deaths during the preceding year, one for agriculture, one for manufactures and one for social statistics. The last asked for returns regarding valuation, taxation, educational and religious statistics, pauperism, crime and the prevailing rates of wages in each municipal division. It was also the first American census to give a line of the schedule to each person, death or establishment enumerated, and thus to make the returns in the individual form indispensable for a detailed classification and compilation. The results of this census were tabulated with care and skill, and a preliminary analysis gave the salient results and in some cases compared them with European figures.

The census of 1860 followed the model of its predecessor with slight changes. When the time for the next census approached it was felt that new legislation was needed, and a committee of the House of Representatives, with James A. Garfield, afterwards president of the United States, at its head, made a careful and thorough study of the situation and reported an excellent bill, which passed the House, but was defeated by untoward influences in the Senate. In consequence the census of 1870 was taken with the outgrown machinery established twenty years earlier, a law characterized by Francis A. Walker, the superintendent of the census, who administered it, as "clumsy, antiquated and barbarous." It suffered also from the fact that large parts of the country had not recovered from the ruin wrought by four years of civil war. In consequence this census marks the lowest ebb of American census work. Tie accuracy of the results is generally denied by competent experts. The serious errors were errors of omission, were probably confined in the main to the Southern states, and were especially frequent among the negroes.

Since 1870 the development of census work in the United States has been steady and rapid. The law, which had been prepared for the census of 1870 by the House committee, furnished a basis for greatly improved legislation in 1879, under which the tenth census was taken. By this law the census office for the first time was allowed to call into existence and to control an adequate local staff of supervisors and enumerators. The scope of the work was so extended as to make the twenty-two quarto volumes of the tenth census almost an encyclopaedia, not only of the population, but also of the products and resources of the United States. Probably no other census in the world has ever covered so wide a range of subjects, and perhaps none except that of India and the eleventh American census has extended through so many volumes. The topics usually contained in a census suffered from the great addition of other and less pertinent matter, and the reputation of the work was unfavourably affected by the length of time required to prepare and publish the volumes (the last ones not appearing until near the end of the decade), the original underestimate of the cost of the work, which made frequent supplementary appropriations necessary, the resignation of the superintendent, Francis A. Walker, in 1882, and the disability and death of his successor, Charles W. Seaton. The eleventh census was taken under a law almost identical with that of the tenth, and extended through twenty-five large volumes, presenting a work almost as encyclopaedic, but much more distinctively statistical.

The popular opinion of a census, at least in the United States, depends largely upon the degree to which its figures for the population of the country, of states, and especially of cities, meet or fail to meet the expectations of the interested public. Judged by this standard, the census of 1890 was less favourably received than that of 1880. The enumerated population of the country in 1880 was larger than had been anticipated; and in the face of these figures it was difficult for local complaints, even where they were made, to find hearing and acceptance. But according to the eleventh census the decennial rate of growth of population fell suddenly from over 30%, which the figures had shown between 1870 and 1880, and in every preceding decade of the century, except that of the Civil War, to less than 25%, in spite of an immigration nearly double that of any preceding decade. For this change no adequate explanation was offered by the census office. Hence the protests of those who believed that the figures for population were too small swelled into a general chorus of dissatisfaction. But the census was probably more correct than the critics. Most of the motives influencing popular estimates of population in the United States tend to exaggeration. The convention which drafted the Constitution of the United States attempted to secure a balance of interests by apportioning both representatives in Congress and direct taxes according to population. A passage in _The Federalist_ suggests the motives of the convention as follows:--

"As the accuracy of the census to be obtained by Congress will necessarily depend in a considerable degree on the disposition if not co-operation of the states, it is of great importance that the states should feel as little bias as possible to swell or reduce the amount of their numbers. Were their share of representation alone to be governed by this rule, they would have an interest in exaggerating their inhabitants. Were the rule to decide their share of taxation alone, a contrary temptation would prevail. By extending the rule to both objects the states will have opposite interests, which will control and balance each other, and produce a requisite impartiality."

With the disappearance of direct taxation as a source of federal revenue, the motive mentioned for understating the population disappeared. On the other hand, the desire for many representatives in Congress has been reinforced by the more influential feelings of local pride and of rivalry with other cities of somewhat similar size. Hence a complaint that the population is overstated is seldom heard, and hence, also, popular charges of an under-count afford little evidence that the population was really larger than stated by the census.

After the detailed tabulation had been completed, it was shown that the number of persons under ten years of age in 1890 was surprisingly small, and that this deficiency in children was a leading cause of the slow growth in population. Before the tabulation had been made Francis A. Walker wrote:--"If the birth-rate among the previously existing population did not suffer a sharp decline ... the census of 1890 cannot be vindicated. To ascertain the facts we must await the tabulation of the population by periods of life, and ascertain how many of the inhabitants of the United States of 1890 were under ten years of age." These results thus confirmed the accuracy of the count of 1890. Efforts to invalidate the census returns by comparison with the registration records of Massachusetts cannot be deemed conclusive, since in the United States, as in Great Britain, the census must be deemed more accurate and less subject to error than registration records. A strong argument in favour of the eleventh census, apart from its self-consistency, is that its results as a whole fit in with the subsequent state enumerations. In eleven cases such enumerations have been taken; and on computing from them and the results of the federal census of 1880 what the population at the date of the eleventh census should have been, if the annual rate of increase had been uniform, it appears that in no case, except New York City and Oregon, was the difference between the enumerations and these estimates over 4%. In Oregon about 30,000 more people were found in 1890 than the estimate would lead one to expect; in New York city, about 100,000 less. It seems not improbable that in the latter, where the difficulties incident to a count during the summer are almost insurmountable, serious omissions occurred. Still, such a comparison confirms the accuracy of the eleventh census as a whole.

The results of the twelfth census (1900) further refute the argument that would maintain the eleventh census to be inaccurate because it showed a smaller rate of increase in population during the preceding decade than had been recorded by other censuses during earlier decades. The rate of increase during the decade ending in 1900 was even less than that for the preceding decade; and it is impossible that a falling off so marked could in two successive enumerations be the result of sheer inaccuracy. The rate of increase from 1890 to 1900, eliminating from the computation the population of Alaska, Hawaii, Indian Territory and Indian reservations, was 20.7; the rate of increase if these places are included--in which case the figures of the population of Hawaii in 1890 must be taken from the census of the Hawaiian government in that year--was 21%.