Content area
All administrative departments had to provide figures to the Central Secretariat. [...]the Trade Department had to provide returns of import and export, specified by commodity group and by country of provenance or destination. [...]of the direct interaction with the enumerator, a person counted in the township had more influence over the way in which he or she was recorded than a person in the village. [...]of these "corrections" the usefulness to historians of the published population census data for the provinces must be regarded as extremely limited. [...]this seems untenable in view of the unsystematic way in which the initial census data were "corrected" on the basis of Colonial Officers' guesswork. [...]the claim that the 1931 census is less reliable because its population statistics are largely based on tax returns, records, and estimates by Colonial Officers is based on an incomplete understanding of the process by which the 1921 population data were produced.
Full text
I
Most historians writing about twentieth-century Africa have, at one time or other, used colonial statistical data. When we do this, we normally add a disclaimer, pointing out that these statistics are likely to be unreliable, and then proceed to use them anyway. But surely, we should be able to say something more definite about the reliability of these data? If we know more about the process by which these statistics were collected, for which aims, and with what preconceived ideas in mind, we should be able to establish, if not a margin of error, then at least some idea of which aspects of colonial statistics are more reliable than others. Furthermore, the process of colonial data-collecting was linked to establishing ethnic and other categories, which have since become generally accepted. This paper addresses these questions in an analysis of the context and contents of the published report of the 1921 Census of Southern Nigeria, and discusses its usefulness as a source for historians. The issues I discuss here with specific reference to this Nigerian census are characteristic for colonial censuses in general and should therefore be of relevance to all historians using colonial census data, and also-more generally-help us to understand how some of the most basic categories describing African societies have been constructed in the process of the acquisition of information by colonial governments.
The 1921 Census of Southern Nigeria was part of the first comprehensive census of Nigeria. Until now I have not succeeded in locating the raw census data in either Nigeria or the UK.1 However, the abstract of the census, included in the four volumes of P. A. Talbot, The Peoples of Southern Nigeria. A Sketch of their History, Ethnology and Languages, with an Account of the 1921 Census (London, 1926) is generally available and commonly used as a source by historians of Southern Nigeria. A companion publication for Northern Nigeria also exists.2 Percy Amaury Talbot, of the rank of Resident, was the Census Officer for Southern Nigeria. Talbot was an obvious choice for Census Officer, being an experienced Colonial Officer with a longstanding interest in anthropology.
Before he was asked to conduct the Census in the Southern Provinces, Talbot had already been conducting anthropological research during intervals of official work for some years, which had resulted in books on the Ekoi and on the Ibibio.3 He was not, however, a statistician. It took Talbot years to mould the raw census data into the published report, much to the annoyance of his superiors (for comparison, the report of the admittedly much more restricted census of 1911 had been published in 1913, and the Census of Nigeria, 1931 would be published in 1933). In 1923 Nigeria's Chief secretary wrote that "our experience of Mr. Talbot with the Census does not inspire His Excellency with any kind of confidence in his methods of work or habits of industry."4 This was unkind and perhaps unfair: Talbot had to spend a long time negotiating with the colonial Officers on the spot' to adjust the original census data. Furthermore, in his foreword to The Peoples of Southern Nigeria, Talbot claimed to have submitted five volumes to the Government in October 1923, three of which were devoted to the census data (published in the final report as Vol. I Historical Notes, and Vol. IV Linguistics and Statistics). He also claimed to have completed the ethnological sections (Vols. II and III) in July 1924.5
I did not find any evidence for the first date mentioned by Talbot, but I did come across a letter from November 1924 in which Talbot's census report, and especially its anthropological sections, were criticized.6 This criticism might explain a further delay in the production of the report. At any rate, the report published in 1926 is very comprehensive, and contains an impressive and wide-ranging overview of Southern Nigeria, collated from a vast array of sources. The larger part of the report is devoted to detailed discussions of history, ethnography, and language of the various groups, covering more than 1300 pages and illustrated with 251 drawings and photographs. The wide range of statistical figures (mission, education, etc.), complete with interpretations and even policy recommendations, make up less than a quarter of the publication.7 According to Talbot, that was merely one-fifth of the total of statistical data that he had supplied to the Government.8 It is not clear why at least these census statistics had not been made available at an earlier date, as one might have doubts about the relevance for the administration of publishing figures that were five years out of date. In subsequent years the census data were used in a number of other official government publications such as the annual colonial reports and the Blue Books of statistics.
The long time it took Talbot to come to a publication of the census results already indicates how problematic colonial census data are. In this light it is surprising how little has been written about this subject. There are some publications on Nigerian censuses that mainly address the political use that has been made of census figures since the 1962 census (which was highly disputed and eventually nullified), but their discussion of the colonial censuses is limited.9 Casting our net wider, we find Bruce Fetter's discussion of the use of colonial censuses for the historical demography of Central Africa, and Peter Uvin's study of Burundi and Rwanda.10 Cohn discusses problems with concepts and census categories for colonial censuses in India,11 while PuruShotam does the same for Singapore.12
In this paper, therefore, I aim to introduce and explore a colonial activity that has not yet received much attention, speculating about the ways in which this activity has contributed to the social construction of colonial African societies, and indicating some potentially fruitful ways of approaching colonial census data as a historical source.
II
The 1921 census of Nigeria was part of the census of the Empire, which in principle was held every 10 years and included every British colony and dominion, as well as the United Kingdom itself. The British administration started census-taking in Lagos in 1871, and in 1901 began to make estimates for Nigeria as a whole.13 In 1911 a house-tohouse enumeration was made in the ports and a few more townships, but for most of the country, the census was based on estimates prepared by District Officers.14 The census of 1921 stands out as the first attempt at a systematic count of the entire population. Subsequent colonial censuses were held in 1931 and 1950-53, but the 1931 census is generally regarded as far less reliable than that of 1921, because arrangements had been interrupted by the riots and disturbances that took place in Southeast Nigeria during 1929-30, and as a consequence, the census results for many areas in Southern Nigeria were based on tax returns and records.15 The 1921 census is also of particular relevance because its detailed listings as to which villages belonged to which "tribe" and "sub-tribe" or "clan" constituted a first complete ethnic map of Nigeria.
In Nigeria the census took place on the same day, 24 April 1921, as in the United Kingdom and in other parts of the Empire.16 This fact draws our attention to an essential characteristic of British colonial census-taking: the desire to generate data that can be compared across the empire. Attempts to collate the census data from all the different colonies and dominions into a coherent overview were hampered when data were incomplete, when differing definitions were used in the different territories, or when the data related to different periods.17 Indeed, in 1920 a conference of colonial statisticians attempted to devise a common scheme for the creation of Imperial statistics. While the need to take into account the differences between the territories in certain cases was recognized, it was also maintained that "[t]o attain a satisfactory degree of uniformity in the statistical data for various countries, standard schemes of nomenclature and classification must be adopted."18
For the 1921 census the Government of each individual colony determined the exact categories on the census forms. However, this was done on the basis of models from Britain, and the forms used in the various colonies appear to be broadly similar. One example is the choice of the age of 15 as the age at which girls became classified as adult women. In Southern Nigeria the District Officers were informed that the age of 15 had been decided on to conform to the practice in Northern Nigeria, and indeed this age was normally chosen for colonial censuses across the Empire.19 However, it did not constitute a relevant age in the societies in which the census took place. It did not coincide with the age of marriage (which could be younger), or with any relevant indigenous transition in the individual's life. This will have limited the usefulness of the data for the Nigerian government somewhat. It also made it rather difficult to determine who were adults and who were not.
In his report Talbot observed, rather unhelpfully, that "natives have but the vaguest ideas on this subject."20 He had indeed found that the number of non-adults in the census returns was almost certainly smaller than in reality and put this down to "the feeling that ill-luck, or death will befall any family, of which the exact number is given, is widespread, and is especially strong in regard to children, and perhaps this is more particularly the case with those of the female sex."21 Fetter, in his discussion of census data for Central Africa, noticed a similar trend, but explained that this was a consequence of the early age at which African women married: he argued that, in order to avoid condemnation from administrators and missionaries, Africans systematically inflated the age of these adolescents.22
Imperial census-taking was part of a much larger project of acquiring information about the colonies. This project has been analyzed in discourse studies as part of the colonial discourse. The connection is made clearly by Said, when he writes that "Eurocentric culture relentlessly codified and observed everything about the non-European or presumably peripheral world, in so thorough and detailed a manner as to leave no item untouched, no culture unstudied."23 However, while clearly Eurocentric, this activity was not restricted to colonies or the "non-Western" world. Bernard Cohn has shown the connections between the role of knowledge creation in the process of state formation in Europe and that in British India. From the eighteenth century onwards, European states made their power increasingly visible through "the gradual extension of "officializing" procedures that established and extended their capacity in many areas."24 They defined and classified space, counted and classified their populations, standardized languages, and codified and represented their past. There was widespread agreement that the society to be governed could be known and represented as a series of facts.
Whether in Britain, in India, or in Africa, these facts had to be collected in systematic ways. Cohn calls such a system to collect facts an "investigative modality:" "the definition of a body of information that is needed, the procedures by which appropriate knowledge is gathered, its ordering and classification, and then how it is transformed into useable forms such as published reports, statistical returns, histories, gazetteers, legal codes, and encyclopaedias."25 The modalities could be the same in the colonies as in Britain, or different. The acquisition of knowledge of indigenous languages was important in India and Africa, as here, the British were ruling peoples who spoke different languages. Indeed, linguistics was considered the prerequisite form of knowledge for all others. In his census report Talbot used language as the main feature for distinguishing between and classifying the different "tribes."26 While an emphasis on language was specific for colonial contexts, other aspects, such as surveillance, were a concern everywhere. Furthermore, the investigative modalities in the different territories influenced each other. Cohn emphasised the extent to which developments in India influenced the developments in Britain.
Investigative modalities could be general, such as history, or highly-defined, such as the enumerative modality, of which census-taking is an aspect. In Nigeria, as elsewhere, the census-the enumerative modality-was only one among a whole range of modalities of colonial knowledge that emerged over time. Most pervasive and powerful was the ethnographic modality, which aimed to describe and classify precolonial society. The prominence of this type of knowledge reflected the intention to graft colonial administration onto existing traditional authorities. The colonial administration needed to know the extent and boundaries of the traditional political units (the "tribes"), the political organization of these units and who were the legitimate leaders, and the influence of traditional religion. In India the administration assumed the requisite facts could be discovered through history.27 In Africa the same function was provided by ethnography.
During the twentieth century the defining method for ethnography has come to be that of fieldwork. However, around 1900, at the beginning of the British presence in Nigeria, the standard method for generating ethnographical data was still the circulation of a questionnaire.28 This was the method adopted by the Colonial Office in 1906, when it decided to compile a systematic account of the population of British West Africa. A questionnaire focusing primarily on the laws of the different groups was produced and circulated to the District Officers. When the answers came in, the Colonial Office decided to ask a professional anthropologist to analyze the data and organize the information into a book. They employed Northcote W. Thomas, who had recently published an armchair study of Australian kinship.29 Thomas, however, decided that the data produced by the questionnaire were of inferior quality, and subsequently managed to convince the civil servants in the Colonial Office to send him to Nigeria as the first Government Anthropologist.30
On arriving in Nigeria, Thomas followed established practice and distributed his own questionnaire to the Colonial Officers together with the then-current edition of Notes and Queries on Anthropology. Kuklick has noted that this was the same style of research as that practiced by the census commissioner for India.31 Thomas' relations with the Colonial Officers were not very cordial. Although he did produce some reports that were useful for the administration, he spent much of his time in Nigeria doing research that the Colonial Administration considered impractical. Although the Colonial Office had funded his research, Thomas did not bother to indicate how the information he had accumulated could benefit the administration of the colony. The administrators considered this to be a major shortcoming. Their doubts about Thomas were increased by the fact that the District Officers in the area where Thomas conducted his research disagreed with most of his observations. They reached the conclusion that an anthropologist was not always of service from an administrative point of view.32 Thomas' appointment did not last long; in 1912 the Protectorate's administration (apparently at the instigation of Frederick Lugard, then the Governor of the Protectorates of Northern and Southern Nigeria) had him removed to Sierra Leone, where his work proved equally controversial.33 The experience with Thomas served as a deterrent to any temptations the Nigerian administration might have had of hiring another professional anthropologist.
However, Lugard had his own ideas about ethnographic research-totally unrelated to the experience with Thomas-which did not include professional anthropologists. According to Lugard, Colonial Officers were in a position to conduct ethnographical studies that were of greater relevance to the administration than those of professional anthropologists, as their results would be more practical. This was an extension of Lugard's experience in Northern Nigeria, where Colonial Officers managed to combine their duties with ethnographical researches.34 From now on the District Officers were expected to make their own notes about local African culture, and indeed every District Office contained an Intelligence Book in which relevant observations were recorded. Furthermore, they were expected to regularly write Intelligence Reports containing practical ethnographical and local political information collected by the District Officer.
By 1920 the ethnographic modality had become the most important aspect of colonial knowledge acquisition, and the District Officer, the "man on the spot," had become recognized as the person with the most intimate and most reliable knowledge of local African society.35 In 1920 another attempt was made to collect information on the "customs and superstitions" of the peoples of Nigeria using an ethnographic questionnaire. This project was a rather straightforward affair. Questionnaires were sent out to the District Officers, containing questions relating to a number of aspects of traditional culture. Once the answers had been collected, they were analyzed, not by a professional anthropologist, but by R. Hargrove, the Resident of Calabar Province.36 Hargrove was careful to edit the original answers as little as possible, although he occasionally changed the writing, and he also felt obliged to delete what he considered to be "undue criticism of education, missionary work, government regulations, etc."37 He was under the impression that his compilation of the District Officers' reports would eventually be published.38 This never happened, but the material was made available to Talbot for use in preparing the census report.39
Where the importance of the ethnographic modality was predominantly political, the survey modality was largely about uncovering the economic potential of the colony. The Survey Department had a broad array of duties, including cadastral surveys (there was a Cadastral Branch), charting the course of rivers, trigonometrical and topographical work, and surveying for minerals.40
In considering the enumerative modality we must note that, from the beginning of British rule, this modality included much more than just the census. All administrative departments had to provide figures to the Central Secretariat. Thus the Trade Department had to provide returns of import and export, specified by commodity group and by country of provenance or destination. The Prisons had to report on how many inmates were in the various prisons in the colony, as well as the mortality rates among prisoners. Township authorities had to report how many dog licenses, vehicle licenses, drumming licenses, etc., they granted. The Education Department had to report how many schools were supported through Government grants-in-aid, how many schools were operated by the Department, how many pupils were in the schools, how many of them passed which standard, and so on.
Quite apart from the decennial census data, the amount of enumerative data produced by the colonial apparatus was considerable. These data formed the basis for the annual Blue Books of Statistics. The Blue Books had been published since the creation of the Southern and Northern Protectorates in 1900 (and been in existence before that for Lagos and for the Niger Coast Protectorate) and contained data mainly relating to trade and government activities. They also contained useful information for trading companies, such as the rates for import and export duties applicable in the specific year. Population and vital statistics included in the Blue Books were usually extrapolated on the basis of census data.
In all, it is clear that the 1921 census was part of a much broader project of gaining colonial knowledge, and some of these other activities might have appeared to be of more direct use to the Government than the population census, for example the mineral survey or the ethnographic intelligence reports. In this light it might seem surprising that the published census reports did not limit themselves to an analysis of the census data, and contained much information derived from other investigative modalities. This was not unique to Nigeria. Bernard Cohn, writing about India, observed that "[t]he census represents a model of the Victorian encyclopaedic quest for total knowledge."41 This assertion seems certainly applicable to Talbot's Peoples of Southern Nigeria, which included discussions of "geological history" and on "the origin of man." As part of the census report he had also submitted a volume on natural features, geology, climate, meteorology, and flora and fauna of the country, which had not been published.42
I think we need to regard the 1921 census as in many ways a transitional enterprise. The inclusion in the report of a majority of information that had not been collected as part of the population census and belonged to other modalities of investigation not only places the census in the much broader production of colonial knowledge in Nigeria, but also refers to the much older British Indian tradition of census reports. The fact that the majority of this additional information was in the ethnographic modality follows from the extent to which ethnography had become the most important form of colonial knowledge in Africa. Subsequent census reports Would be less elaborate, focusing on an analysis of the statistical material, and did not claim to offer encyclopedic knowledge.
The 1921 census was also transitional in marking the first attempt to count the population of Nigeria systematically rather than to use estimates from Colonial Officers. At the same time though-and in line with the tradition of generating colonial knowledge that had emerged during the preceding years-when the data were all collected together, Talbot tended to attach greater weight to the impressions of the District Officers than to the census returns provided by his enumerators. A final aspect in which the 1921 census was transitional is that it fixed the ethnographic categories for Nigeria. Until 1921, for large parts of the area the administration had only a very general notion of which "tribes" inhabited it. However, as the colonial census forms treated each rural village as belonging to one "tribe" or "sub-tribe," the census enumerators and the District Officers had to decide which group belonged to which. Out of a combination of the reports of the census enumerators, the opinions of the District Officers, and linguistic information Talbot managed to arrive at a first complete ethnic classification of Southern Nigeria. Even though he indicated that this classification was merely provisional, and that he expected that further research would prove that certain groups belonged to different tribes altogether, in practice the 1921 classification was long considered definitive.43
III
The 1921 census was held at a time when the British colonial administrations were still recovering from the loss of funding and manpower that had occurred during World War I.44 In fact the administration of the nearby Gold Coast had written to the Colonial Office asking for the census to be postponed. This request was ignored and a despatch dated 25 May 1920 informed the colonial administrations that the Census of the United Kingdom would be taken on 24 April 1921, and that it was desirable that a Census should be taken in all parts of the Empire on that same date, giving the colonies slightly less than eleven months to prepare for their censuses.45 The budget available also limited what could be done. In all, the census of the Gold Coast colony cost £4,118.46 The census in Northern Nigeria cost about £1,000, a very small amount, considering that the total expenditure for the Nigerian administration in 1921 was £7,171,497.471 could not find a figure for the Southern Nigerian census, but every indication is that it must have been equally modest.48
While standard census forms would be used throughout Nigeria, many of the further practical arrangements would be organized separately by the Census Officers for Northern Nigeria and Southern Nigeria, Charles Kingsley Meek and Percy Amaury Talbot, respectively. In each District it was the task of the District Officer to organize the census. To assist the District Officers the administration enlisted the help of large numbers of census enumerators. These could be teachers, or Africans working with European commercial enterprises as clerks, but many of the enumerators were already working with the administration as court clerks or messengers, and were therefore familiar with the area (if not necessarily popular with the local population).49 Talbot nevertheless complained about the scarcity of educated and trustworthy clerks. In his report he describes his enumerators as "comparatively illiterate men."50
In preparing for the census, the District Officers had to instruct the various Native Court Clerks to prepare lists of all the towns in the District and of all the principal quarters in each town. These lists then had to be checked by the District Officers, who had been instructed that "special care should be taken that the spelling is correct."51 After this it was again the District Office that assigned towns and quarters to the census enumerators, distributed the census forms to the enumerators, ensured that the completed forms were received back, and subsequently collated the census data on separate sheets headed "Division-District-Tribe-Sub-Division of Tribe." The District Officer then had to forward the resulting data to the Census Officer, and report on the actual taking of the census, including an assessment of the likely reliability of the District's data.
In addition to the population statistics, the District Officer had to provide returns of the number of commercial firms and how many branches each had; the number of schools, pupils and teachers; the number of missions and how many Europeans were employed by each; the estimated number of livestock; and a rough sketch of each Division showing the boundaries of tribes and sub-tribes.52 It appears that some of this information would also have been available from other sources, such as the colonial Blue Books of statistics. For example, Talbot mentioned explicitly that his data for education were collected separately from, and differed from, education data included in the Blue Books.53 The decision to collect these data again must have added to the District Officers' burden. Indeed the census resulted in a considerable addition to the ordinary duties of the Colonial Officers at a time when the Administration was already stretched thin. The following comment, made in relation to the Gold Coast, probably holds equally true for Nigeria: "It would be idle to deny the unpopularity of the Census amongst all officers who had to carry it through."54 However, the central role of the District Officers also meant, as I will discuss later, that their opinions and hunches had a significant impact on the final shape of the census data.
In line with procedures elsewhere in British West Africa,the Nigerian census had different methods for collecting data in the larger towns (the "township census") and for collecting data in the smaller towns and villages (the "provincial census").55 The aim for the township census was the direct enumeration of every individual. There was a separate copy of the census form for every house. An enumerator visited each house to collect the forms or to help complete the forms if the occupants were too illiterate to do so. The latter must have occurred in most cases, as the census gives the overall level of illiteracy in the townships as 72%.56 The census in the townships was held on one single day, 24 April. Talbot considered the statistics thus produced as nearly correct, unlike those produced by the provincial census, the procedures of which had been much less precise. The provincial census was conducted over a period of two months, allowing for population movement to skew the data. Also the census forms were not distributed to every house-instead a census form was completed for the entire village or for part of a village or town. While the enumerators had been encouraged to visit every house if possible, this did not usually happen. Another difference was that it was much easier to escape enumeration in the provinces, and there is every indication that many successfully evaded the count.57
In addition to these differences in approach, there was also a more subtle difference between census-taking in the colonial townships and in the provinces. As a result of the direct interaction with the enumerator, a person counted in the township had more influence over the way in which he or she was recorded than a person in the village. Take the returns for "occupation," for instance. In the townships enumerators could give the occupation of each adult in detail. In the provincial census, however, a simple classification had to be followed. This classification had eight categories and a ninth category "miscellaneous" ("domestic servants, native doctors, juju men, professional dancers") for the men, while the women had to be fitted into only three categories: "craftswomen," "traders," and "domestic."58 Another example would be that of ethnicity. While in the townships each individual counted could describe themselves, in the provinces it was assumed that all indigenous inhabitants of a village belonged to the same "tribe" and "sub-tribe."
I will show later that in many cases, the decision as to what the "tribe" and "sub-tribe" were was taken by the District Officers in disregard of what people themselves might think. This difference is reflected in the layout of the census forms. On the form used in the provinces, the entries for "tribe" and "sub-tribe" are required first, followed by questions about age, gender, language, religion, etc. In contrast, the form used in the township starts by categories for "town," "quarter," "owner of the house," age, and sex. "Tribe" is only the sixth item on the schedule.59 This is not to say, though, that people in the townships were entirely free to decide on their ethnic group. Talbot gives the example of non-Hausa northern Nigerians who were classified as Hausa "partly because they talked that language and partly because their real tribes were unknown to the enumerators."60 Then there are also the limitations resulting from the circumstance that the census enumerators were males, as this is likely to have rendered the ethnicities of females less visible in those cases when these differed from those of the males in the household.61
The issue of ethnicity, or at least the attempt to draw up an "ethnic map" for the colony, was one of the primary aims of the enumeration. Not only did the provincial census forms place "tribe" as the first category, the Nigerian forms also contain the additional category of "subtribe," absent on the census forms used in nearby Gold Coast.62 In the published census report Talbot states that "[t]he chief work has been an attempt to classify the tribes and sub-tribes and to define their boundaries."63 This was mainly for administrative reasons, "since it is difficult to govern a people efficiently unless the political divisions are known and taken into consideration."64
In 1921 many parts of Southern Nigeria had only relatively recently been effectively occupied by the British, and there was still a lot of uncertainty in official circles about the political makeup of the area. At the same time, the guiding principle of British colonial rule was to leave intact as much as possible of what they considered to be the "traditional political structure." The intention was that each African community continued to be governed by its own traditional rulers, who were in turn controlled by the colonial government. This in itself created a need to establish definite boundaries between the various African polities. However, the British were balancing their aim of maintaining the traditional African political structures with that of administrative efficiency. To this extent a hierarchical structure of tribes and sub-tribes or clans was desirable, as this would create an easily understandable ethnic map. It would also allow a rationale for the merging of the "native courts" of various "sub-tribes" into more efficient units.
Maps are policy tools and as such do not necessarily correspond to reality. They not only fix what is fluid and changing, they also simplify, since they are based on only those aspects considered relevant to the administration. Talbot explained that in the 1921 census he used language as the sole means of classification (see map 1).65 It may seem strange that an exercise that claimed to help uncover the existing African political divisions did not take into account how the people in question would classify themselves. However, this can be seen as an attempt to arrive at an objective definition of the ethnic groups in the area, and a perfectly plausible one at a time when it was commonly assumed that "tribes" and nations were defined by their respective languages.66
Further evidence for an attempt at holding out for "scientific objectivity" can be found in Talbot's apologizing for his failure to use the result of an anthropometrical survey due to the illness of the mathematician.67 Of course we nowadays frown on anthropometry, but this "science" was perfectly respectable in the early 1920s, and was certainly considered more objective and reliable than the "evidence of similar customs" on which Talbot felt he had to rely "to a small extent" in addition to language. The Census Officer's commitment to objectivity over and above the notions African people might have held themselves, can also be seen in the attempt he made "to give the true designations of the tribe or people, instead of the often inaccurate names in present use, and in some cases, where there was no tribal or sub-tribal appellation, it has been necessary to extend the content of a name in order to cover the additional members."68 For us nowadays, however, this statement reveals the extent to which the colonial census and the resulting ethnic maps had the ability actually to create the reality they claimed to describe, as the categories used did not correspond to locally-held notions, but to the perspective of the colonial administration.69
IV
The 1921 census arrived at a total population figure for Southern Nigeria of 8,371,459.70 Talbot felt that this was an underestimation: "on the whole, in my opinion, the population is considerably larger than that given in this report, and it would appear safe to say that, if 10% were added to the native tribes, the results would be more accurate."71 Yet even this underestimation was already significantly higher than the original census returns, as percentages had been added on the advice of the District Officers. Talbot similarly qualifies the reliability of the ethnic classification provided, emphasising it was "purely provisional."72 However, in this case, too, the data presented had already been substantially revised from the original census returns. This brings us to the question of how these original census returns were achieved, and what were the processes subsequently employed to correct them. I will focus on the provincial census, as the results of the township census were considered to be "nearly correct" and were therefore not altered.
I have already discussed how the census was organized. Here I will have a look at what happened when census enumerators visited local communities to conduct the count. While the enumerators had been instructed to try and visit every house, this was rarely possible in practice. According to one enumerator, "only some compounds in Owerri & Nekede I counted every person with difficulty. In other towns I found it absolutely difficult to do so - as they see me some ran and hid themselves."73 Census clerk M. Tasie described the procedure he followed thus:
When I got to a town, my first duty is send for the headmen to be assembled in the head chief house. When they come, I then asked them to call the sub-headmen under them. After each and every one of them are assembled, I asked them to give me the Names of each house that each of them representing. After this is done then each headman began to give the No. of Men, Women, Boys & Girls in each house. Before this a previous notice has been given to the particular town to get ready.74
Another system used was that of counting with grains: "the chief and headmen drop a grain of corn as they call each man's name and then women, boys and girls. I took the total of each".75 The District Officer for Owerri noted that "much depended on the personality of the chiefs," and that Court Clerks acting as census clerks obtained more accurate returns than the enumerators especially employed. He explained that "[t]his is not surprising as the official position of a court clerk enables him to exert greater pressure to obtain information."76 While coercion might have proved effective, it is of course unlikely to have added to the popularity of the census.
Indeed, almost all involved in the enumeration mentioned protests, attempts at evading the census, and a profound suspicion among the local population about the government's motives for the census. The annual Nigeria Report for 1921 described it as "a measure which was generally disliked".77 Talbot blamed this partly on what he regarded as a widespread feeling amongst Nigerian peoples against giving the number of their family, especially of the children, as "a person in possession of this gains power over them and can do them harm."78 According to the District Officer, Owerri, "apart from local superstitions of misfortune following enumeration, explanations by Political Officers failed to remove suspicion that the real motive of the Government was to obtain information for a levy for war purposes and taxation."79 One census clerk reported: "The people dislike to be counted-they inclined to believe the rumour that it was for taxation purpose or coming of American people."80
Of course this suspicion of the motives of the colonial administration was not unreasonable. During the period 1915-18 direct taxation had been introduced in all areas west of the river Niger, and serious plans existed to extent direct taxation to the rest of the territory. In those areas where taxation had been introduced already, this introduction had been preceded by an enumeration that provided a tax assessment.81 Thus people, and especially their properties, were counted more often than just for the decennial census, and at times specifically for taxation purposes. Not surprisingly therefore, Talbot noted that the strongest opposition to the taking of the 1921 census was found in areas already under direct taxation.82
When the results of the enumeration came in, they were immediately dismissed on the District level as unreliable.83 The District Officer for Owerri reported that women and children ran away on the days of the enumeration, and that in total only about 80% of the population in his District had been included.84 The other District Officers had similar reservations about the raw census data, generally regarding them as being too low. Earlier I argued that by 1921 the District Officers' knowledge, as that of the practical men on the spot, had come to be regarded as the most accurate and valuable data available, and certainly more useful than that offered by anthropologists. In fact the District Officers had been explicitly instructed to supervise the Census Clerks and to check the figures produced by them.85 Therefore, now that the District Officers proved in disagreement with the original census returns, Talbot-himself formerly a District Officer and as Resident merely seconded to do the job of Census Officer-would have to take their assessments into account.
But how do you weigh up the results of the enumeration-evidently flawed due to people evading the count, but at least based upon a reasonably consistent methodology-with what were essentially the educated guesses of the District Officers? (And how would they have been able to tell the difference between, say, 8000 and 9000 people living in a sprawling set of villages called a "native town" just from their experience?) How can the Census Officer take observations such as the following into account: "Owerri-Nguru-Okpata-Oguta areas are fairly correct-though some towns are obviously underestimated. Ngor and Umuakpo on the other hand have I think a much larger populationespecially Ngor."86 Eventually Talbot invited the District Officers to supply the percentages to be added to the original results, and these percentages ranged from 5% for Okpala to 75% for Umuakpo, and 100% for Ngor.87 Corrections to the census figures continued to be made up to at least twelve months after the date of the census.88 From the correspondence on this it is clear that these corrections were not based on any statistical evidence, but merely reflected the intuition of the District Officers.
As a result of these "corrections" the usefulness to historians of the published population census data for the provinces must be regarded as extremely limited. To the best of my knowledge, the original census returns no longer exist, and extant correspondence on the corrections proposed by the District Officers is very incomplete. As a result we can establish the general process through which these data were produced, but we cannot reliably check individual figures. The way these data were presented in the report suggests a level of reliability that obscures the guesswork underlying these figures. For example, the figure given for Owerri Division-which includes Okpala, Umuakpo, and Ngor-is simply given as 615,557. The population data from the provincial census therefore provide us only a very rough approximation of population levels. We also have to be very careful about comparing the data for the various districts or towns, as these have been corrected using differing percentages. The population data included from the township census, and figures relating to foreign residents, appear to be more directly useful. Of course, considering the fluidity of the urban population at the time, there can be no doubt that these figures are also underestimates, but at least they are underestimates arrived at through a consistent methodology.
Turning to the issue of the ethnic classification of the population, we find a similar process of negotiation between District Officers and the Census Officer to arrive at the final results. Again the results for the townships are relatively unproblematic, in the sense that the returns here were by and large based on individuals' own claims as to where they were from. There were problems in the interaction between enumerator and the person being counted, as I mentioned earlier, and it is not clear what labels such as "Ibo unclassified" or "Edo unclassified" mean, but it seems that at least the data on the census returns were accepted without further modifications. The real interest in drawing up the ethnic map lay in defining the groups in the provinces, as at the time it was assumed that the ethnic identities of migrants in the cities would merely reflect those. The entire exercise was based on the assumption that all inhabitants of a village (excluding obvious "outsiders" such as resident Hausa traders) would belong to the same tribe and sub-tribe or clan.89 The meaning of terms such as "clan" and "subtribe" was not clearly defined in this context, and Talbot uses the terms very loosely to indicate "divisions" of larger peoples.90 He also assumed that the villages belonging to a sub-tribe would be covering a contiguous area, and that all the sub-tribes of a tribe would similarly inhabit one contiguous area with a clear boundary. This reflected the colonial administration's policy that the boundaries of its administrative divisions should follow the boundaries between the various tribes and sub-tribes. On this basis the ethnic map could be translated into a simple map of administrative divisions.
The other key assumption, as I mentioned earlier, was that each tribe was defined by its own language. As the census did not actually collect linguistic data, much of the ethnic classification was based on pre-existing linguistic studies of the area. On the one hand the categorization of the census followed the main language families as distinguished at the time, and the general location of some of the main languages, such as Igbo, Yoruba, Ibibio, and Ijo, did not pose problems either (see map 1). What needed clarification were the exact locations of the boundaries of languages and the question of the classification of possibly different languages or dialects: was such a "dialect" merely a branch of a larger dialect or sub-tribe, did it constitute its own sub-tribal language while part of a larger language, or was it a language of its own? In other words, the census tried to establish the exact location of tribal boundaries and also to which "main tribe" the different groups and "sub-tribes" belonged.
In terms of ethnic categories the raw data collected by the census clerks was organized according to the lists of the names of towns and tribes and sub-tribes, which had been drawn up under the supervision of the District Officers as part of the preparation of the census. As part of the process of evaluating the raw data, a blank map of the area was circulated among District Officers, who then had to mark on it the "tribal sub-divisions" in their respective districts.91 Various versions of these maps in progress were circulated, as the information on them needed further and further fine-tuning. In his report Talbot drew attention to the labor spent on this and remarked that "[s]ub-tribe and clan (or branch and dialectical) boundaries were in many cases very difficult to define, while the general map of the Southern Provinces is so inaccurate in parts that it was often difficult to insert them."92
Another issue that he had to resolve was that the lists drawn up by District Officers from different Districts contained many names that were very similar. To establish whether groups with roughly similar names belonged to one and the same "tribe" or "sub-tribe" or not, the Census Officer sent telex messages to the different District Offices to ask for clarification. For example, Talbot would send a telegram asking: "Are Isu the same tribe as Isu in Okigwi division please," to which the reply would follow: "yes same tribe-but different branch".93 An added complication was that if there was a possibility that a sub-tribe stretched across administrative District boundaries or Provincial boundaries, then the assumption that administrative divisions should reflect traditional divisions provided an added incentive for District Officers to describe similar groups as distinct rather than the same.94 The District Officers must have been aware that any change in classification could have direct implications for native court arrangements and other aspects of administration.
A now familiar pattern is visible: after an initial classification of groups according to language, corrections and clarifications were sought from the District Officers, who, as men on the spot, were assumed to have the most reliable knowledge of the area. This procedure of course directly determined the outcome of the exercise: as the colonial administration aimed at administrative divisions that followed those of pre-existing African groups, and as District Officers were expected to comment on their own District (which already had its boundaries), more often than not the obvious assumption would be that tribal boundaries coincidentally followed that of the colonial administrative divisions. Throughout the process all those involved expressed doubts about the accuracy of the boundaries and groupings. However, the way in which the final result is presented, in tables for each "tribe" specifying the various sub-tribes, and the various clans and towns within the sub-tribe, complete with specific population figures for men and women, renders all these doubts invisible.
It would be wrong therefore to conclude that the entire exercise did not really clarify anything. Circular though the process might have been, it was important in fixing the boundaries of ethnic groups formally and in providing legitimacy for the existing administrative divisions. The bureaucratic exercise, through the process of designing and completing census forms, the rounds of corrections by the District Officers, and the systematic presentation of the data in many pages of tables, provided the first detailed listing of local groups, and the subtribe and tribe to which each of them belonged, thereby completing the ethnic map of the territory. For an example of this, see Map 2, showing branches of the Edo language, and Table 1, showing "sub-tribes" and "clans" in the Edo "tribe" (note how each branch of the Edo language corresponds with a sub-tribe or clan). Although Talbot made it clear that his classification was merely provisional, and that he expected that further research would prove that certain "sub-tribes" belonged to a different tribe altogether, in practice his classification was long considered definitive.95
V
We have seen that it took a long time to get the census data right, as the correspondence between Census Officer and the District Officers continued for more than a year after the actual census-taking. Nevertheless, as we have seen, the preparation of the report by Talbot took years longer. This was partly the result of the inclusion in the report of large amounts of material from other investigative modalities, more specifically the volumes on history and anthropology, which make up more than three-quarters of the publication.96 It is to these volumes that we should now turn our attention.
The volume on "history" reflects what Cohn described as the colonial censuses' "Victorian encyclopaedic quest for total knowledge."97 This volume starts with "general historical notes," some of which are very general indeed.98 This section contains six pages of geological history and includes a discussion of the "probable" course of the formation of the African continent. This is followed by speculations about the origin of man and the subsequent peopling of the African continent, including some reference to the then-current theory of the "Bantu migrations." Much space was given to a discussion of the various "foreign influences" on Nigeria's peoples throughout history, which include-according to Talbot-Hamitic influences, cultural influences from Egypt, the introduction of mining and iron-working, and the arrival of new crops. This discussion of foreign influences then leads into a discussion of the English, Portuguese, and other traders on the coasts, the slave trade and its abolition, and the eventual establishment of Nigeria as a colony. The final pages of the general historical notes include a detailed year-by-year listing of various achievements of the colonial state, including railways, roads, and telegraphic communication. Talbot's general notes reflect what was the orthodoxy in the colonialist historiography on Africa during the 1920s, focusing on "foreign influences" as the origin of change and progress throughout Nigeria's history, thereby providing justification for British colonial rule.99
The "general historical notes" are followed by historical overviews of each of the provinces. The lengths of these overviews vary. There are 50 pages on Lagos and Owerri Province is covered in 38 pages. However, Onitsha Province has only four pages and Ondo Province only three. These differences appear to reflect the different amounts of data available to Talbot, and hot a ranking in terms of relative importance. The format is that of a chronicle: the entries are organized by year and presented as facts. Some of the entries are only one sentence ("1907. A severe epidemic of small-pox broke out."), while entries for other years can run into several pages.100 There is no obvious attempt at explaining the course of events, and Talbot did not attempt to create flowing narratives. There are no obvious selection criteria for inclusion of information; it seems as if Talbot included everything that he could find on each province.
Trade and politics feature prominently, probably reflecting the preoccupations of Talbot's sources. As was to be expected, there is relatively much detail on the more recent colonial activities such as treaties and conquest, the opening of hospitals, and the completion of railway lines. The chronological range is equally generous: the chronicles start early: often around 1500 AD, but sometimes much earlier (2000-1000 BC for Oyo Province, for instance), and continue until the census year of 1921. Talbot tended to quote his sources verbatim, sometimes for pages long, but was inconsistent in attributing them. In some instances he introduced the name of his source in the text, while at other times a name appeared in brackets after a paragraph or statement. In many other instances, however, there is no indication at all as to the source of a statement. Most of Talbot's sources appear to have been European sources: (slave) traders and travelers for the earlier centuries, and colonial sources for the more recent period. It is nevertheless clear from the text that he has also used some African oral traditions, especially for the earlier periods.
It would be easy to dismiss the historical section of the report as reflecting colonial prejudice and as lacking rigorous historical method. However, while the "general historical notes" fail to move beyond generally held colonial assumptions as to the importance of "foreign influences" in Africa's past, the sections on the provinces bring together a lot of material from diverse sources, including African oral traditions, some of which is unlikely to be available elsewhere. It is also worth noting how much space is devoted to the historical sections at a time when anthropology was the dominant colonial investigative modality. All the same, the ethnological sections still take up more than half of the report. This information suffers from the same defect as the historical data: there is insufficient detail about the sources on which the sections are based.
In his introduction to the ethnological sections Talbot claimed that "[t]he whole of the information, except for some details about the Yoruba, has been obtained personally during my various tours ... and . . . has been corroborated by several informants."101 However, it is clear from the correspondence on the census that Talbot made extensive use of the multi-volume report on the "Southern Provinces tribal customs and superstitions" compiled between 1920 and 1922 by R. Hargrove from the reports of District Officers.102 These reports had been prepared in response to an ethnographical questionnaire drawn up by Hargrove, and circulated to the District Officers in autumn 1920.103 These reports were also made available to Talbot for use in compiling the census.104 This is relevant, as Hargrove noted that many of the reports he received had been written by African interpreters and "semi-educated chiefs."105 This indicates that, just as African enumerators had collected much of the raw statistical information, part of the ethnological information was also derived from African sources.
The ethnological sections do not attempt to give ethnographies of the various African groups. Rather, each of the chapters focuses on one topic deemed to be characteristic for "primitive peoples." The volumes start with chapters on ceremonies and religion, including chapters on "Minor deities or jujus," "Magic," "Witchcraft," and "Totemism." Next are chapters on politics and social organization with titles such as "Government," "Law," "Slavery." "Tabus (a) sex and birth," and "War and cannibalism." The final topics tackled are "Houses, etc.," "Occupations and industries," and "Rain, etc. (rain, lightning, celestial luminaries and folklore)." Each chapter begins with a general introduction to the topic, which provides a definition of the custom or belief, and makes general statements about its prevalence and functioning in Southern Nigeria. This general introduction is followed by specific descriptions of the topic for ethnic groups in the census. The length of these descriptions varies from group to group and here again this appears to reflect the differing amounts of data available to Talbot rather than any notion of relative importance of the groups. The chapters usually end with a table indicating prevalence, type, and names used to describe the phenomenon in the various "tribes" and "subtribes" of Southern Nigeria, and some chapters also contain maps indicating the geographical spread of the phenomenon discussed.
Let us look at the chapter on "witchcraft" as an example of this. The chapter starts with a general definition of witchcraft, noting that:
The fear of witchcraft is strong throughout nearly the whole of the country, and among some peoples - especially the more ancient, such as the Ijaw and Semi-Bantu - exercises an almost paralysing effect and colours their whole life. The bigger the Hamitic strain in a tribe the less is it addicted to this practice.106
To illustrate this, the chapter includes a map of Southern Nigeria, indicating the distribution of the extent of the belief in witchcraft, categorized as either "excessive," "very strong," "much," "moderate," or "slight." Talbot then proceeded to list essential elements of the local form of witchcraft, and discusses the differences between European and Nigerian witchcraft. This is followed by descriptions of the specific witchcraft beliefs among the various ethnic groups. There is a page on the Yoruba, and nearly five pages on the Igbo (indicating how practices of "sub-tribes" differ), but only very few lines on groups such as Edo or Popo:
Diarrhoea, pains in the abdomen and dizziness were usually ascribed by Popo to witchcraft. Witches are supposed to send their Aza-he (Witchbird), the owl, to sit on an enemy's roof at night-time, and by means of this they can see through the roof and into the body of the sick man beneath. The latter feels something moving within him whenever the owl hoots.107
From these examples, it becomes clear that the ethnological sections show a similar pattern to the volume on history: where Talbot attempted a general overview his observations tend to reflect fairly standard colonial assumptions about "primitive" peoples, as well as racist theories, such as the supposedly Hamitic origins of some west African peoples, and that these peoples of Hamitic descent showed a higher level of development than the other groups. Here we also find a wealth of attempts at comparisons across time and cultures, some of which appears to us now as rather farfetched. The specific local material, however, even though largely anecdotal and unsystematic, contains a wealth of detailed descriptions and cases.108
The general organization of the ethnographic material reflects an older colonial pattern of listing the variations on a set of "laws and customs." This survey approach to ethnography had grown out of the ethnographical questionnaire as the main tool for collecting ethnographic data. Therefore, in a way the existing system of questionnaires determined the outcome and organisation of Talbot's work. Here we encounter again the key role of the District Officers as suppliers of information about local circumstances; even in cases where the answers to questionnaires had actually been written by African clerks or traditional leaders, they were forwarded as the work of the District Officer. During the course of the twentieth century, the ethnographic survey became less and less common, as the staple of anthropology became the ethnographical monograph on a single people or community. The survey did not disappear altogether, and during the 1940s and 1950s the International African Institute initiated and published a multi-volume ethnographic survey of Africa.109 Nevertheless, Talbot's report seems to represent a mode of representation that had started to go out of fashion among anthropologists by the time it was published.
The last volume of the published report contains an overview and discussion of the census data, a topic I have already discussed in some detail earlier in this article. The chapters focus on the categories of the census form, dealing with "The population," "The tribes," "Religions," "Education," "Civil condition," "Age and sex," and "Occupations." It also contains a chapter on languages, which is on the whole not based on the census, but is important because of the extent to which the ethnic map was based on a linguistic classification. The text introduced the data, offered definitions of the terms used, and indicated points of comparison or divergence between the various groups. The data themselves are given in the form of tables. As I mentioned above, the text of the census report contains health warnings about the trustworthiness of the data collected, while the way in which the statistics are presented at the same time implies great reliability and detail.
When Talbot's report finally appeared, it was welcomed as a "monumental work" and a "valuable contribution to African ethnology," but also rather severely criticized. In a review in Man, Emil Torday accused Talbot of reading "his own conception into the native mind."110 He also criticized Talbot's comparisons as too liberal, giving the example of his discussion of a taboo on eating oxen, described as "a remnant perhaps of the ancient bull worship which prevailed in Mediterranean and Central American areas." Torday wonders: "Does he connect Sudanese natives with Central Americans? By the way, how could the latter worship bulls without possessing cattle?"111
While Talbot considered the anthropological sections of the report the most important and valuable, it appears that the impact of this part of the report was limited. Created firmly within a colonial administrative tradition, the form in which the information had been collected and presented soon came to be perceived as somewhat old-fashioned. Talbot was not an academic anthropologist, but a political officer with an interest in ethnography. While this placed him in a good position to do ethnographical research in the context of a colonial administration suspicious of academics, it also placed him outside the developments in anthropology as an academic discipline. In his introduction Talbot claims (deceptively, as we have seen) that the work was based on his own observations, and that he had also avoided theoretical influences, asserting that, "[t]hough Sir James Eraser's [sic] works form no mean proportion of my very exiguous library out here, I have refrained from consulting them in order that my judgment should be rendered, as far as possible, independent."112 There is of course more than a hint of Frazer in Talbot's work, and in this it is rather far removed from the functionalist anthropology that had just started to emerge.113 Functionalist anthropology was not yet very visible when Talbot was preparing his manuscript, but it would soon take off, which meant that the published report quickly seemed outdated.
However, the ethnographic traditions that emerged in the colonies did not necessarily mirror the developments in anthropology as an academic discipline. Ethnographic traditions in the colonies were localized, reflecting the concerns of specific colonial administrations, as well as practices of data collecting that had grown over decades.114 In contrast to the rise of functionalist anthropology in Britain, in the field in Nigeria, the collecting of random information on customs continued, and remained well respected. In this context Talbot's report certainly did not look out of place or out of date and copies of the report were distributed to the District Officers for practical use.115 However, the District intelligence books, the regular intelligence reports and handing-over notes offered more detailed information on the local level. Talbot's work also suffered from the competition of ethnographical monographs on the area, such as G. T. Basden's and C. K. Meek's works on the Igbo, and Samuel Johnson's book on the Yoruba.116 These monographs provided more coherent and more detailed discussions than Talbot's report, even though Talbot was perhaps stronger in pointing out the many differences and variations within "tribes."
I have found evidence of Talbot's use of the work of Basden and other monographs, and also of the use of the Intelligence Reports, jboth by Colonial Officers and by Africans.117 I have not come across many references to the use of the census report's ethnographical data.118 However, the statistics from the report have often been quoted and repeated. In particular the population statistics were frequently reproduced. They were included in official publications such as the Blue Books of statistics and Annual Reports and were also referred to during debates on future policy, and in the course of day-to-day administrative duties. Being the first comprehensive set of data for Nigeria, the statistics have frequently been referred to by historians and other scholars, who often use them as a baseline for looking at subsequent developments in Nigeria. One effect of this was that for the next decades the ethnic map drawn up by Talbot came to be generally accepted among scholars, colonial administrators, and African subjects alike. Talbot, as we have seen, regarded his classification as merely provisional, and he believed that further research would prove that certain sub-tribes had been incorrectly classified and belonged to different tribes altogether. However, the practical needs of administration, the presentation of the data in detailed tables, and commonly-shared assumptions about the nature of "tribes" helped to render a provisional classification definitive.
This ethnic classification affected the lives of Africans: the tribes and clans thus defined became the basis for the local justice systems (the Native Courts) and administrative arrangements (the Native Authorities), and political representation came to be channeled through the medium of the ethnic group. However, the census also had an impact on Africans in more direct ways: the census was to some extent an exercise in demonstrating the existence of the colonial state at a time when some communities still refused to acknowledge its authority, bringing almost everybody in contact with an aspect of the colonial administration.119 The project had required the involvement of many Africans as clerks and census enumerators. Many of these had already been employed by the administration, but others came from commercial firms, or were schoolteachers or came from other backgrounds. Those involved had to engage with the way colonial knowledge was created through a bureaucratic process, involving specific definitions of categories, the counting of people, and then the presentation of these data in a specific format.
One result of their involvement in this process was that these Africans started to take colonial categories more seriously. At the same time this also empowered them: it gave them access to a language in which to communicate with colonial authorities and defend their own claims. For the majority of Africans who were not involved in the census-taking, the census was anything from a minor nuisance to a potential threat (considering the fears of a link to taxation), and hence they often attempted to avoid enumeration. However, the insights gained by those involved with the enumeration as to the relevance of the count and of the categories employed, contributed to an emerging censusawareness among the population. In future censuses, a pattern of avoiding the count would give way to attempts to manipulate resource allocation or political representation by inflating the numbers for specific groups or communities.120
VI
In this paper I have characterized the 1921 census as transitional. It was the first attempt at a comprehensive census based on the systematic enumeration of every house or compound in the colony and therefore the first modern census of Nigeria. At the same time, however, it also took the form of a much older tradition of imperial census-taking and, in line with an existing Nigerian colonial tradition, the Census Officer rated the experience and knowledge of the Colonial Officers higher than the completed enumeration forms. It was also transitional in that it produced the first complete ethnic map of the colony.
The 1921 census is often regarded as one of the most reliable in Nigeria's history. However, this seems untenable in view of the unsystematic way in which the initial census data were "corrected" on the basis of Colonial Officers' guesswork. Therefore the claim that the 1931 census is less reliable because its population statistics are largely based on tax returns, records, and estimates by Colonial Officers is based on an incomplete understanding of the process by which the 1921 population data were produced. However, it may be important to note that in 1921 and 1931 census-taking had not yet become the political battlefield that later censuses would become.
The detailed discussions of the way in which the 1921 census data were produced and corrected, and of the form that the published report, took help historians to understand the type of source the colonial census constitutes, and which aspects are likely to be more or less reliable than others. It does not give us any clear pointers to correct the data, but it does give an idea of where the problems are likely to be: the general historical and ethnographic sections and the population statistics from the provincial census are clearly problematic, while the data from the township census and the specific historical and ethnographic comments about particular groups seem more useful. More importantly, however, the analysis of this census proves to be a lens through which to understand the creation of colonial knowledge more generally. It offers us an insight in how categories that we still use were constructed, and it also allows us to speculate about how the bureaucratic process of census-taking helped to convince Africans of the relevance of these categories to their own lives. In this sense this analysis of the 1921 census may function as a detailed case study of the colonial process of objectifying cultural and linguistic differences.
DMITRI VAN DEN BERSSELAAR
UNIVERSITY OF LIVERPOOL
Copyright African Studies Association 2004