Who are Kenya’s 42(+) tribes? The census and the political utility of magical uncertainty

ABSTRACT

The idea that Kenya is made up of 42(+) tribes is widespread, but the origins, nature and consequences of any list are not well-known. This article compares ethnic classifications in all Kenyan censuses to demonstrate the origins of the ‘42’ in (only) the 1969 census, and the multiple political purposes of classifying and counting. To make sense of why the 42(+) remains significant, I argue a cultivated vagueness provides a sense of consistency, linking a national past to present and future, while providing the basis for both numbers-based competitive politics and more inclusive politics. Moreover, it avoids engaging in politically risky work of making legible sense of shifts in ethnic identities, classifications and numbers, and avoids having to resolve their relation to the nation, which benefits both state and citizens. Extending literature on the political utility of uncertainty, I theorise this cultivated vagueness as magic, backed by opaque forces, potentially dangerous or beneficent, which deters interrogation or certainty on all sides. To further clarify this awkward relationship between vagueness and certainty, I argue ethnic classifications are intelligible via the social imaginary of the 42(+), but not especially legible, contesting the literature on census practices as tools of legibility and governability.

KEYWORDS:

• Ethnicity
• tribe
• Kenya
• census
• magic
• uncertainty
• recognition
• tribalism
• palimpsest

It was a hot and dusty day in January 2019. Sam had been driving me around Nairobi since my first visit, ten years earlier. I often float ideas past him as we endure interminable traffic. ‘Sam, how many tribes are there in Kenya?’ I knew there was no definitive answer but I wanted to know his thoughts. ‘Well, now we are … is it … 46? Or 47? We used to be 42 but some new ones were recently added. Makondes. Asians. Who else was it? Nubians … ’. ‘And where is the list?’ I probed. ‘Oh that one … is it gazetted somewhere? I don’t know.’ Later that day, while he refined my left hook, I asked my boxing trainer, Jeff. Embarrassed, he laughed and said ‘You know … I’ve not brushed up on my tribes lately … .’ ‘Just roughly … how many?’ He replied after some thought, ‘I think … well … I know that we used to be … is it 41? Or 42? 42. We used to be 42. But now, I don’t know.’ In multiple interviews with various government officials, I was repeatedly told there were 42+ tribes, but nobody could tell me the nature or location of the list. ‘Do you know?’ one official asked me. Ten years earlier, I asked members of the minority Nubian community too: ‘Forty-two tribes. And we will be the 43rd’. They even had a letter from a Minister declaring they would, indeed, be counted as such in the 2009 census.¹ But I struggled to find the list. Who is on it? Does it even exist? And if so, who controls it, and how? Why does nobody know? And does it matter?

Part of the answer lies in an empirical history of census classifications. I compare here all ethnic classifications in all Kenyan censuses, from 1948 to 2019, and show the number 42 is found only once, in 1969. Indeed, every census has a different list, with changes revealing the multiple political purposes of classifying and counting: in the colonial period, tax collection and labour control; and in the postcolonial period, both ethnic demographic posturing and recognition of minorities. The census is a palimpsest: the result of layers of bureaucratic and political sediment accumulated over time, bearing markings of colonial and postcolonial administrative need and political desire, but not constituting a wholly deliberate artifice. Mapping and putting these changes in historical context generates important insights into practices of ethnic classification in colonial and postcolonial Kenya. But this is only part of the answer.

To make sense of why the 42(+) remains significant, we need to understand the benefits of retaining a vague but powerful social imaginary not only for the state, but also its citizens. I argue ethnic classifications are rendered intelligible via the social imaginary of the 42(+), but not especially legible because, despite the definitive nature of a list, it shifts over time in ways that are very opaque. This renders the relationship between vagueness and certainty an awkward one. While prevailing studies of uncertainty emphasise state abuse and citizen disadvantage, I argue the cultivated vagueness around ethnic classification is beneficial for both. It simultaneously allows anticipation of demographic dominance by large groups, and conferral of recognition to smaller ones, while limiting attendant risks on all sides. I further theorise this uncertainty by refining Veena Das’ analytical concept of magic to capture such workings of the state, backed as they are by dangerous and beneficent forces, the nature and cause of which are opaque and best left unapproached.² This understanding of the role of the census contradicts the thrust of literature on census enumeration of identity, which emphasises the census’s rendering of populations legible and governable.³

My arguments centre on an analysis of the first compilation of all ethnic codes used in all censuses, and available explanations for them drawn from census reports and records at the Kenya National Archives, including enumeration instruction manuals. The arguments are further informed by media analysis and ethnographic observation in Nairobi of the 2009 and 2019 censuses, including conversations with enumerators and low-level administrators from provincial (2009) and county administrations (2019) overseeing enumeration in the neighbourhood of Kibra. I do not, however, have any hard evidence about why codes were changed over the postcolonial period, and the difficulty of acquiring such evidence is also part of my argument about the opacity of the classification process.

The imaginary of the 42(+)

In Kenya, the imaginary of a nation constituted by 42(+) tribes is widespread and powerful. Charles Taylor defines a social imaginary as ‘[t]he ways people imagine their social existence, how they fit together with others, how things go on between them and their fellows, the expectations that are normally met, and the deeper normative notions and images that underlie these expectations’.⁴ Imaginaries are, then, widespread and powerful by definition. They formulate the common understandings through which members of a society make sense of their own and each-others’ actions, including perceptions of legitimacy, normalcy and moral rectitude. As such, they are both factual and normative, structuring both how things normally go and how they ought to go, and what constitutes an ideal case, a foul, an anomaly or a norm. An imaginary is distinct from a theory insofar as the latter is both more concretely articulated and structured, and more elite. An imaginary is more inarticulate and unstructured, a sense more than a theory.⁵

The ethnic makeup of Kenya is a social imaginary insofar as a reasonably firm – but not solidly fixed – idea of who is in and who is out is commonly shared. It shapes people’s sense of belonging, entitlement, and right and wrong action, yet in ways that tend to be more inarticulate than articulate. As shown above, my friends and colleagues in Kenya have a sense the nation is made up of around 42 tribes. The imaginary is roughly factual and fairly normative, entailing an intuition there must be a list somewhere but no hard knowledge of where, and a conviction that those 42ish ethnicities are rightly and proudly Kenyan, but no rigid sense of who they might be. It also includes the legitimacy of the very concept not just of ethnicity, but of tribe, which is the vocabulary used in all questionnaires and census reports, in both English and Swahili (kabila).⁶

Though few people are able to identify it, the 1969 census is the origin of the figure 42, as the only year in which that number of ethnic groups were counted. Between 1948 and 1989, between 38 and 47 groups were counted, and in 2009 those numbers roughly tripled.⁷ There are 14 ethnic groups reported and classified the same way, as main tribes with no sub-tribes, in all censuses: Kikuyu, Embu, Meru, Kamba, Kisii, Kuria, Luo, Taita, Maasai, Samburu, Turkana, Teso, Rendile, and Gosha.⁸ Three more have been counted every decade, but with variations in their sub-tribes: Luhya, Swahili and Mijikenda. A further five have been reported in every census, but sometimes rendered a sub-tribe: Pokomo, Bajuni, Njemps, Ajuran, and Gurreh. Kalenjin and Somali have been included in every census, sometimes classified and counted as such, sometimes as only sub-tribes and sometimes as both. It is, therefore, impossible to establish meaningful criteria for what counts as consistent classification, even if we can recognise the consistent inclusion of these groups. This consistency can mostly be explained by their large size, almost all having populations very near to or over 100,000 in 2019.⁹

There are only a few exceptions to this numbers-based explanation of consistent inclusion. First, there are two groups – Swahili and Gosha – counted consistently despite being small. Both reside in locations that were contested as British territory: Swahili in the ten-mile coastal strip that remained under the sovereignty of the Sultan of Zanzibar; and Bantu Gosha people in the province of Jubaland, the non-Bantu part of which was ceded to Italian Somaliland. Similarly, Taveta were included in every census from 1962, and in the code list for 1948 though not the final report. Taveta people reside on the border with Tanzania. Enumerating these groups asserted authority over people and territory whose belonging to Kenya was fragile, and whose postcolonial belonging has also been in need of edification to combat possible competing loyalties. Gosha and Taveta have also had districts of the provincial administration in their name at various times, further cementing their belonging to the nation.

Second, there are four ethnic groups that have large numbers (over 100,000 in 2019), but have not been consistently counted. Orma and Gabra, both of the north-eastern region, were excluded in 1948, when authorities had a poor administrative apparatus and only estimated that region’s population. Basuba people, numbering 139,271 in 2019, were excluded in 1962 and 1969, subsumed under Luo. As Bantu people of Lake Victoria, Basuba people have assimilated in many ways into the Luo community, but have also asserted their cultural and identity independence, especially in relation to language.¹⁰ The timing of their re-inclusion in the census, in 1979, may reflect President Moi’s suspicion of the Luo community and efforts to undermine it, but for reasons discussed below it is hard to say definitively. Somalis are discussed further below.

The imaginary of the 42(+) persists, then, not because of official figures but because it serves a different purpose. I suggest it provides a sense of consistency that links a national, united past to both present and future, while providing the basis for both a numbers-based competitive politics and a more inclusive politics. Moreover, it does so without having to engage in the politically risky work of making legible sense of numerous shifts in ethnic identities, categories, classifications and numbers, and without having to resolve the relation of these to the nation. In what follows, I further explain the consistencies and variations through what I characterise as three historical periods: colonial; independent nation building; and inclusive nation building.

Ethnic categorisation in colonial Kenya

British colonial rule in Kenya was based on an imaginary of Africans as characterised primarily by their ethnicity. It was through ethnic custom that the British practiced indirect rule and made it possible to govern ‘on a shoestring’ by co-opting local leaders to implement colonial policy.¹¹ British desire for order through classification has been a recurring theme in postcolonial literature, even as it has increasingly been acknowledged that control was always tenuous and needed to be constantly reproduced in the face of anti-colonial resistance and resource and logistical limitations; and that colonial subjects often identified with the classifications under which they were placed.¹² The census, because of its unique dual functions of classification and nation-wide population quantification, became a key instrument in the cultivation of this ethnic imaginary.

The earliest tools for registration and population counting sought to exploit Africans as a labour force, collect hut and poll taxes, and manage anti-colonial unrest by discouraging pan-African organising. Poll taxes were collected by district commissioners (DCs) who kept district- and location-level tax registers, regularly (more in some districts than others) updated by African ‘hut counters’.¹³ Tax registers were supplemented by labour registration requiring able-bodied men to wear registration tags to move off their home reserve. These tax registers and labour records formed the basis of annual population returns prepared by DCs at the sub-district level, where they indicated the ethnic groups in their district, albeit inconsistently, and under conditions of poor resourcing and varied levels of local knowledge. While other records tracking ethnicity were also taken, for example among the police or armies, it was tax and labour registers that formed the basis of ethnic population figures.

Over time, DCs began to complain about duplication of effort for tax and labour records until it was decided, in the mid-1940s, that a national population and housing census should take place to streamline workload. The desire for a robust census was also fuelled by shifting priorities in the colonial office in London, where a more progressive government was focusing on the civilising mission of Empire, and sought to better understand population pressures and their impact on colonial economies and modernisation efforts.¹⁴ An East African general census took place in 1948, with a detailed schedule for non-natives (meaning non-Africans), and a simpler one for natives.

In his report on the planning and conduct of the 1948 native census, Census Superintendent C. J. Martin speaks of tribal enumeration as self-evidently important. After a survey and discussion with DCs, Martin and his statistical team decided that tribe was as important as age and sex, and that it should be enumerated for every Kenyan in the household survey, whereas other factors, such as fertility, education and occupation, were relegated to the sample census covering only 10% of the population.¹⁵ However, Martin explains it was decided it would be too costly to gather ethnic data beyond ‘main tribes’, as ‘the sub-tribes and clans were known only to persons conversant with the area’.¹⁶ The questionnaire allowed for an open-ended response to ‘Tribe’, yet enumerators were provided a list of main tribes from which to choose, which Martin also describes as self-evident in content.¹⁷

However, the results of the census show some groups appeared in the process of enumeration, in the final report but not the code list, while others on the code list were not included in the results. In the absence of an explanation for how those on the list were chosen, or the inconsistencies pre- and post-enumeration, we can only assume that any range of factors may have shaped the final 1948 list, including self-identification, enumerator or DC initiative or discretion, or political lobbying. In other words, determining the tribes of Kenya was not as self-evident as Martin imagined.

The second census was conducted 14 years later, in 1962, the year before independence. William T. W. Morgan, a colonial official involved in the 1962 census, admits a certain arbitrariness in the definition of ‘tribe’, but maintains the particular tribes named and counted were obvious:

[Tribe is] a unit which evades satisfactory definition but which was widely recognised. It may be said to be a group to which the individual feels a strong sense of belonging and which is usually distinguished by a common language and culture and, since marriages are mostly within it, may have inherited traits. […] For this study we have to accept the classification used in the census, for which no justification was published. The ascriptions were those routinely used by the administration and which appear to have presented few problems to those recording or those being recorded. They were the socio-political groups encountered by the colonial power upon its entry and with which it had to deal. Administrative boundaries were normally constructed to contain them and this probably increased the sense of tribal identity at that level.¹⁸

The main finding of Morgan’s paper is that in 1962, outside the ‘White Highlands’, most locations recorded 90–100% membership of one ethnic group; 85.1% of Africans lived in a location or ward in which their tribe was a majority (50% or more), and average demographic dominance was 94.7%.¹⁹ Indeed, the major administrative boundary redrawing that took place in 1962 heard petitions from many ethnically organised delegations seeking to be grouped together, and the colonial commission concurred this was a good principle for territorial administrative organisation.²⁰

By emphasising the confluence of ethnic identities, territories, administration and classifications, Morgan draws attention to the mutually constitutive relationship between them, not to their inevitability (as he at times claims). He therefore also draws attention to the central role of the colonial desire for control, labour and tax in the constitution and reproduction of ethnicity. Given these classifications are linked to those extractive, dominating and divisive colonial projects, it is incumbent upon us to examine how and to what extent those projects linger in the ongoing use of classifications, and in what ways the intentions and effects of classification have transformed (which they partially have). This is not to hold the census or even colonialism solely responsible for the invention of ethnicity. Though the classifications are unambiguously a colonial invention, the imagining of their meaning is far more complex, as both the continuities and the transformations of the postcolonial era show.²¹ As Glassman amply demonstrates, there is no single source of group identities, and Africans’ allegiance to ethnic groups was undoubtedly strong.²² Rather, it is to acknowledge the problematic origins of the population-level practice of classification and enumeration, including the awkward relationship between its expedient and shifting implementation and its masterful and definitive claims, and to examine its effects.

Nation-building, tribalism and the tyranny of numbers in independent Kenya

Appadurai argues it is not just classification, but quantification, that was central to the role of number in the colonial imagination. Quantification reduced complexity to countable abstractions, making the aberrant, marginal or deviant (the whole native population, from the colonial perspective) legible and governable. It cultivated a statistical thinking that fostered a sense of mastery, and served both justificatory and disciplinary purposes with respect to native populations, and pedagogical ones for colonial authorities, teaching them to master the Other. Cold, hard numbers countered the ‘heat of the novel, the light of the camera, and the colonial realism of administrative ethnographies’.²³ The totalising nature of number helped colonial authorities see their native subjects as constituted by incommensurable group difference.²⁴ As in Appadurai’s India, this ‘idea of politics as the contest of essentialised and “enumerated” communities’ stands out as the most significant ongoing legacy of colonial ethnic enumeration.²⁵

Upon independence, Kenyans inherited a highly centralised, exclusionary and divisive state built on a foundation of imagined essentialised, enumerated and territorially contained ethnic groups.²⁶ Among those big groups consistently enumerated in the census, the principal functions of the social imaginary of the 42(+) is the anticipatory politics of numbers. For them, the postcolonial period has been one of often-intense inter-ethnic competition and repeated conflict. The phrase ‘It’s our turn to eat’ is frequently used to describe the expectation of an ethnic group’s chance to benefit publicly and privately from control of the state.²⁷ The presidency of Jomo Kenyatta (1963–1978) set this tone, as he privileged his Kikuyu ethnic kin in politics, civil service positions and land settlement schemes, provoking a lasting perception under today’s president, his son, Uhuru Kenyatta, that Kikuyu people have benefitted disproportionately from the state’s coffers.²⁸ Lonsdale describes this as political tribalism, evident where ‘groups compete for public resources’ and where ethnic identity is used ‘in political competition with other groups’.²⁹

It is through ethnic demographic posturing that the two most significant changes in ethnic enumeration between 1969 and 1999 can be understood. The first was in 1979, a year after Moi became president. Various sub-tribes were grouped and counted as Kalenjin, meaning Moi now belonged not to a small group (Tugen, numbering 130,249 in 1969), but a big one, Kalenjin (966,548 in 1979, the sixth biggest group that year).³⁰ The second significant change was the non-reporting of ethnic numbers or classifications in 1999, and the permanent end to their reporting at district level after 1989.

Census reports provide no justification for these changes, nor are KNBS or related records publicly available. This culture of silence is partly what makes seemingly highly legible lists actually illegible, as we cannot know how they are formed. In 1979, after consultation with ministries and census users, the questionnaire was developed following a directive from Cabinet, suggesting Moi had significant input. A permanent census office was established that year, and had formal control over the 1989 and 1999 censuses, placing the census officially at arms-length from the Executive, but still relying (as in 1979 and before) on the provincial administration – the eyes, ears and arms of the Executive – for implementation. International development agencies provided technical assistance, but it is not clear if they were involved in decisions about ‘the tribe question’. The relative influence of the Executive, provincial administration, bureau of statistics, and international agencies is, unfortunately, unclear, and so to understand changes made to classification and reporting, we can only speculate.

Both changes are consistent with Moi’s anxieties about ethnic numbers, and the increasingly heightened sensitivity about ethnic demographics and land during the 1990s. As Anderson explains, under domestic and international pressure to return to multi-partyism after decades of authoritarian one-party rule, Moi revived debates about majimboism (majimbo meaning ‘region’) from the negotiation of the independence constitution.³¹ At independence, Moi and other African leaders of smaller ethnic groups (as well as Kenyan Europeans and Asians) advocated majimboism on the basis that small ethnic groups needed protection from domination by large ones, and the best way to achieve appropriate levels of autonomy and resource allocation would be through a federal state structure. Some features of majimboism were indeed included in the constitution, but rapidly dismantled by Kenyatta. As calls for multi-partyism gained momentum, Moi remobilised this idea among Kalenjin and other small groups with whom he had close political associations. He cultivated the notion that multi-party politics could only mean oppression for small groups, and actively supported violent attacks by such groups against ‘foreigners’, ‘aliens’ and ‘madoadoa’ (stains or spots), particularly in the Rift Valley and around the 1992 and 1997 elections. This mobilisation was in part made possible because of the reality of ethnic territorial concentration. By 1989, the mean predominance of a single ethnic group in each district was slightly lower than 1962, but still quite high at 87.98%, information only attainable via the census.³² Thus, though we can’t know how this information was directly linked to the violence, the practice of collecting it is deeply consistent with a venomous autochthony.

The ‘tyranny of numbers’ has since persisted, though it has shape-shifted. Kenyans eventually voted Moi and his majimboist allies out, which Anderson declared the triumph of political inclusion.³³ This expression of national cohesion notwithstanding, the spectre of numbers and their implications lingers. At times, it is violent, as in the 2007 post-election violence.³⁴ Often it is exclusionary, as in the constituency-level politics of Kibra where opposition leader Raila Odinga was MP for more than two decades. His Luo ethnic kin frequently take to the streets to express their dominance, particularly over Kikuyu residents (until 2007 when most left) and Nubians, who have lived there for more than a century and sought their own ethnic dominance in parts of the area, despite neither knowing for sure local ethnic numbers.³⁵ At other times, it can be more hopeful: under the 2010 constitution, devolution has mitigated the zero-sum game of state capture and meant smaller groups can satisfy their ‘turn to eat’ at the county level.³⁶ But numbers are always sensitive.

This point is illustrated by two controversies. The first is that surrounding political analyst Mutahi Ngunyi’s prediction that Uhuru Kenyatta would win the 2013 presidential election based on numbers of registered voters in Kenyatta’s Central Province (assumed Kikuyu and ethnic allies) and Rift Valley (assumed Kalenjin, the ethnic group of vice-presidential candidate William Ruto).³⁷ Kenyans widely rejected the projection on the grounds that location is not an indicator of identity, nor does everyone vote ethnically.³⁸ Indeed, research shows strong ethnic voting patterns are not straightforward because such behaviour is likely defensive, based on mistrust; ethnic alliances shift dramatically; and the weakness of political parties and lack of political programme also play a part.³⁹ As well as rejecting ethnic voting, those opposed to Ngunyi were also calling out the practices of both inferring and using ethnic numbers politically.

The second controversy is the call by some county-level leaders, particularly in Central Kenya, the heartland of the Kikuyu community, for Kenyans to return to their rural homes for the 2019 census. As an editorial in one of the country’s leading newspapers put it, in a formulation typical of such commentary,

We are not sure what the interests of such politicians are. It is common knowledge that some politicians are angling to pounce on the census data for future planning and gerrymandering when the time will come for the delimitation of boundaries […] Kenyans should not be made to believe they are being used as pawns in a dirty political game of numbers.⁴⁰

Neither case draws on hard data about ethnicity, but both draw on the imaginary described by Appadurai, which is supported more by the census, as the only source of ethnic population figures, than any other records. The voter rolls at the heart of these two controversies are prepared by the Independent Electoral and Boundaries Commission and do not record ethnic identity. Electoral boundaries are connected to the census, insofar as they draw on population data, but before the 2010 constitution they bore no official relation to ethnic data, and though the 2010 constitution allows for a possible use of ethnic data, the exact role it plays in boundary drawing is not yet clear.⁴¹ Civil service employment records routinely record and make public ethnic data, but that only captures, of course, civil servants. To establish the ‘fairness’ of each ethnic group’s share of civil service jobs, that data is compared to census data, but only at the national level or by problematically inferring ethnicity by location.⁴² Identification cards do not record ethnicity. Nor, contrary to popular understandings, does the gazette list ethnic groups, though it was used as if it does when Asians were gazetted as the 44^th Tribe of Kenya in 2017, despite no identification of the preceding 43.⁴³

Instead of drawing on hard data, Ngunyi assumes a correlation between location and ethnicity, while county leaders assume the possibility of ethnic territorial concentration, despite a lack of such data since 1989. The logic of the tyranny of numbers is intelligible to all, but it rests on an illegible process of ethnic enumeration (where we don’t know why changes are made) that gains its credibility from its legible nature (a definitive list). In this way, the ethnic classifications in the census, and their apparently definitive association with population and ethnic competition, even though there is nothing definitive about it, is a vital component in tribalism and the tyranny of numbers. As Appadurai notes, it is not the numbers so much as the imagination they cultivate that is politically significant.

Recognition and inclusive nation-building from 2009

Ethnic categorisation introduced by colonialism has been enthusiastically adopted by state and citizens not only for competitive reasons, but also for cohesive ones. While enumeration of ethnic groups can feed tribalism and xenophobia, classification can serve national inclusion. As I have argued elsewhere, ethnicity is not necessarily a barrier to democracy, but can be an important vehicle for civic life.⁴⁴ The dramatic shift in classifications from 2009 reflects this inclusive thrust by recognising numerous groups the state has historically struggled to classify and enumerate. In 2009, the overall number of tribes counted rose to more than 110 (including sub-tribes), and in 2019, more than 120, adding the ‘plus’ to the 42. These new classifications constitute an expansion of recognition that is more attuned to the complexity of ethnic identities, including their changes over time, multi-layered nature, and imbrication with colonial histories of migration and border making. However, because all the newly recognised groups are small, this shift is largely non-threatening. In 2009, the top 10 ethnic groups comprised 88.2% of the population, whereas the bottom 23 (excluding sub-tribes) comprised only 6.4%.⁴⁵ In 2019, these figures were similar: the top 5 comprised 65.3%, and those 17 groups numbering under 100,000 (excluding sub-tribes and ‘stateless’), less than 1%.⁴⁶ As well as recognition of small groups, though, inconsistent counting since 1948 has also involved temporary absences and permanent erasures. These patterns of inclusion and exclusion illuminate the kinds of groups whose belonging the state struggles to determine.

Very small indigenous groups – hunter-gatherers or pastoralists with populations under 20,000 – have been among the main beneficiaries of the recent move toward more recognition. This includes Waat (also known as Aweer and Wayyu), Walwana / Malakote, Ilchamus, Endorois, and Ogiek, added over 2009 and 2019, and Dasenach (added 1989) and sub-tribe Merile (added 2019). Njemps were consistently counted until rendered ‘Ilchamus/Njemps’ in 2019. These groups would have previously been counted as ‘Other’ or identified with an umbrella ethnic group, because of the state’s reluctance to accept the concept of indigeneity. The Endorois community’s case at the African Commission on Human and Peoples’ Rights, which ruled in their favour in 2010, applied significant domestic and international pressure on the state to recognise indigeneity, albeit with largely symbolic outcomes.⁴⁷

A further three groupings have complicated patterns of counting because of doubts around their citizenship associated with colonial migration and border making, to which the state has responded in various ways. While Gosha, Taveta and Swahili are small border populations consistently counted because of the significance of their territory in colonial border making, many other border and migrant communities have gained recognition over time, while others remain invisible or – perhaps worse – defined as stateless.

Four small ethnic groups (populations under 25,000) brought to Kenya by the British from elsewhere in Africa were counted in 1948 as part of the East Africa census, then erased until recently re-added.⁴⁸ This includes, in 2009, Nubians, recruited from Sudan and Egypt as colonial soldiers, and Burji, brought from Ethiopia to northern Kenya for farming expertise;⁴⁹ and in 2019, Makonde, who migrated from southern countries to work on white farms; and Konso, closely related to Burji. Perception that these groups were just visiting, and their collusion with colonial authorities and settlers are – as documented elsewhere in detail about the Nubians – the reason the state had been reluctant to recognise them.⁵⁰

Other groups, mostly residing in what are now neighbouring countries, were counted in 1948, when the exercise was for ‘East Africa’, and never counted again, including Lumbwa, Sageju, Konongo, Pare, Bagishu and Baganda.⁵¹ In 2019, Pare appeared again, but under the problematic category ‘Stateless (Galjeel, Shona, Wapemba, Pare, etc.)’, suggesting they are not Kenyans.⁵² Pare reside mostly in Tanzania; Shona have migrated from Zimbabwe; Wapemba from the island of Pemba (Tanzania); and Galjeel are ethnically Somali and internationally recognised as stateless.⁵³

Lastly, inconsistency, but with a general move toward recognition, has also characterised enumeration and classification of (mostly) small northern border populations near Ethiopia and Somalia, often nomadic or semi-nomadic pastoralists or hunter-gatherers who share ethno-cultural and linguistic characteristics across borders. While Rendile and Borana, the two biggest (2019 populations of 96,313 and 276,236 respectively) have been counted in almost every census,⁵⁴ others have been added over time: Gabbra, Sakuye and Orma in 1962, and Wardei sub-tribe of Orma in 2019. Galla (also known as Oromo) were counted in 1948, but not again until 2009, and were rendered a sub-tribe of Orma in 2019. The shifts in classification, especially of Galla/Oromo, are likely due to difficulty enumerating mobile people in remote border areas, but also ambiguous citizenship status.

The general (though not complete) move toward greater inclusion and recognition of minority groups – indigenous, border and migrant populations – can be interpreted as a product of the politics of the time. The 2009 census took place only 18 months after inter-ethnic post-election violence killed 1,000 people and displaced nearly 700,000.⁵⁵ The nation was charged with tension, and there was a palpable desire to move on from tribalism, and for some – from tribe itself.⁵⁶ This was expressed in the ‘Tribe Kenya’ civil society campaign encouraging citizens to refuse to name their ethnicity, and when asked ‘What tribe are you?’ to answer ‘Kenyan’. This campaign influenced the code list provided to enumerators in advance of the census, where ‘Kenyan – so stated’ appeared for the first time.⁵⁷ For those people who responded ‘Kenyan’, shunning ethnicity was a marker of inclusion, but for small communities, the converse was true and the desire to be counted as their ethnic group was strong. Among Nubians, for example, their historical classification as ‘Other’ led to perceptions they were foreigners. As one Nubian woman explained to me at the time, it made them feel ‘the Nubians are not amongst the tribes in Kenya, the 42, so we’ll be put like Others’. The intimate connection between the census and recognition as Kenyan was exploited in its advertising campaign, which announced ‘Nipo!’ (I am here!) and ‘Natambulika!’ (I am recognized!), alongside photos of 28 faces representing the socio-economic, racial and cultural diversity of the country.

A year later, the 2010 constitution established devolution, aiming, as stated in chapter 11, ‘to foster national unity by recognising diversity’, ‘to protect and promote the interests and rights of minorities and marginalised communities’, and ‘to ensure equitable sharing of national and local resources throughout Kenya’ (among other objectives). Constitutional promises include affirmative action in government, education, the economy, employment and support for cultural and linguistic practices; representation in the public service, parliament and county assemblies; and financial measures to address historical disadvantage. Distributional outcomes, however, are yet to materialise, and by 2019 cynicism had overshadowed the optimism of 2009. In Kibra, for example, the overwhelming subject of conversation and rancour around the census was corruption and nepotism in the employment of enumerators, rather than serious expectations of development or redress. Nonetheless, nobody wanted to be left out – ethnically – just in case.

Despite distributional benefits being elusive thus far, at least constitutionally, they are possible. However, how a group acquires a code, and whether they are officially a minority or marginalised group remains opaque. Enumerators are trained to record a person’s tribe ‘as they state it’, but with a pre-existing code (in 2019, with a drop-down menu on digital tablets). This suggests groups are recognised in advance, but in practice there are – as in the colonial period – discrepancies between code lists and published results. This can include movement between sub and main tribes, and groups appearing on code lists but not in final reports – in 2009, Konso; in 2019, Mkenwa, Tswakans, Garre, Sidam, Ngiteuoso, and Kifundi. This could mean no respondents identified with these tribes, or numbers were too small to report, though that seems unlikely given the smallest group in 2019 (Wafaza, Swahili sub-tribe) had a population of 69. It could mean these groups were deliberately erased, though why they would even be on the code list is unclear. The only obvious explanation for any of these cases is for ‘Dorobo’ being removed from the 2019 report, despite being on the code list and counted for many years. The term is derogatory, referring to having no cattle, and they were likely counted under Kalenjin, of which they were a sub-tribe in 2009. More surprisingly, anomalies include groups appearing in final reports but not on code lists (in 2009 Pokot, Endorois, Galjeel, Isaak and Leysan and in 2019 Lembus (Kalenjin sub-tribe) and Wardei (Orma sub-tribe)), suggesting enumerators somehow recorded alternative answers, or there was retrospective enumeration or estimation.

The opacity of the process and consequences of ethnic classification is further illustrated by both Nubians and Makondes being formally recognised as the 43rd tribe; Nubians by the Minister of State Planning, National Development and Vision 2030, overseeing KNBS, in the lead-up to the 2009 census, and Makondes by President Kenyatta in 2017, when he guaranteed their access to citizenship documentation.⁵⁸ My research with the Nubian community showed inclusion was achieved by lobbying KNBS, but how the decision was made, and what it means to have two 43rd tribes, remains unclear and has not been questioned by either group.⁵⁹

The content, nature and consequences of these lists are opaque both because citizens prefer not to inquire, and because it is practically impossible to find out how they have been compiled since independence. The Kenya National Archives has no records after 1962. Presumably KNBS does, but I was unable to view them or interview staff. Over ten years, both formal and informal, cold and connected emails, texts and phone calls have gone unanswered. On my last visit, in 2019–2020, the mood was so sensitive I opted not to visit their office – often the best way in. Mutual contacts repeatedly told me ‘It’s a very difficult time for them’ or, ‘They will be feeling pressure and might not get back to you.’ Informal conversations with government officials from some commissions who were part of a consultative committee around the 2019 questionnaire were unable to explain decisions. Some reported having made written submissions to KNBS about content or having given feedback in meetings, but they did not receive any explanatory feedback about the final questionnaire. In most cases, they neither checked the final questionnaire nor requested any explanation. Lower-level officials ‘at the coal face’ were similarly tight lipped, albeit for different reasons. In 2009, my research assistant and I observed the census enumeration in Kibra, and he repeated the exercise in 2019. Census officials have been cautious talking with us, having sworn an oath of secrecy regarding the exercise. Enumerators, supervisors, assistant chiefs and chiefs knew very little about ‘the tribe question’. Their training was limited to ‘record the response as it is given’, despite the pre-determined code lists. While information about classification decisions exists somewhere, it is evidently remarkably difficult to access. I suggest this is not a failure of research, but a finding that underscores the opacity and illegibility of classification decisions.

Somalis: the hard case

Classification and enumeration of ethnic Somalis demonstrate both that enumeration plays an important role in tribalism, and that classification confers important forms of recognition. Living predominantly in the north-east, and with a sizeable population in Nairobi, ethnic Somalis are emblematic of the colonial and postcolonial states’ inability to ‘transform its subjects into a countable, traceable population or fully police their mobility’.⁶⁰ North-eastern Somalis have been historically marginalised, with little state investment in resources or administration. The highly mobile people, who live collective lives stretching across borders for commerce and pastoralism, have confounded census makers for decades. The collapse of the Somali state over the 1990s and 2000s led to a significant influx of refugees, and a common perception that Somalis are illegal. The rise in Al-Shabaab terrorism in Kenya has further fuelled suspicions. Many Somalis legally entitled to make Kenya home want to do so, but the Kenyan state has been hostile.⁶¹

Colonial censuses reveal a deep confusion regarding categorisation of Kenyan Somalis. In 1948, they were included in both the non-native census, where they were a special exception to the definition of non-native as non-African;⁶² and the native census, where their population statistics were reported by tribe.⁶³ Their intermediary role in the colonial project, particularly as soldiers, and how to count both ‘Kenyan’ (native) and ‘Somali’ (non-native) Somalis, troubled authorities’ clarity on their status in the racial hierarchy. By 1962, they were moved to the racial category of ‘African’ and were classified by a general Somali category, and/or clan names (interpellated as ‘tribe’) in every census.⁶⁴ Somali as a category of its own has existed in every census. In addition, Ajuran, Gurreh and Gosha were counted from 1948, Hawiyah and Ogaden from 1962, and Degodia from 1969, all separately, until rendered Somali sub-tribes in 2009. Isaak and Leysan were added in 2009 and converted to Somali sub-tribes in 2019, while Galjeel, related to Somalis, was added as a main tribe in 2009 and converted to ‘Stateless’ in 2019. Murile and ‘Corner Tribes’ were added as Somali sub-tribes in 2009 and 2019 respectively. Despite Somalis’ inclusion from the first censuses, they were properly counted for the first time only in 1989, with all prior censuses failing or not trying to enumerate in logistically challenging districts such as Turkana and Marsabit, and North Eastern Province.⁶⁵

In 2009, Somalis sparked a major controversy, unexpectedly emerging as the sixth largest group, with a population of 2,385,572. KNBS nullified results on the basis that it was a statistical impossibility given previous official demography. However, Weitzburg convincingly argues the pre-2009 figures were incorrect, given the difficulties the state had in enumeration.⁶⁶ So, while small, unthreatening groups can be easily recognised, controversies over larger groups who threaten others’ demographic dominance, and who are perceived as interlopers, illuminate the limits of the census as a tool for inclusion, returning us to the spectre of tribalism.

A magical palimpsest

I interpret the imaginary of the 42(+) as a magical palimpsest. In postcolonial theory, the metaphor draws attention to that which was erased by colonialism, the contingency of colonially constructed narratives, and to signal the possibility of future reinscription that reasserts the agency of the colonised. Like animal skin parchments that are reused, but on which prior text can be detected or spontaneously appear, old inscriptions lie below the surface and sometimes bubble to the top.⁶⁷ The fluctuating categories over Kenya’s eight censuses constitute a palimpsest insofar as, when layered upon each other, they bear markings of colonial and postcolonial imagination and inscription, including permanent marking of some identities and erasure or addition of others. Among older inscriptions continually resurfacing we can include the number 42, the apparently self-evident necessity of counting ‘tribes’, certainty of belonging for the groups counted relatively consistently, uncertainty of citizenship status among border and migrant groups, and telescoping of ‘sub’ and ‘main tribes’. Every decade, a new inscription is laid atop the old.

Like the reuse of animal skin parchments, the census is a product of expediency. Census categories reflect not just desire of colonial officials for legibility, order and control, or of postcolonial presidents for numerical dominance, or administrative and political aspirations for national unity, but also the incompleteness of all these projects. While it is not by accident that ‘the list’ is mysterious, I am not suggesting there has been a deliberate obfuscation by a unified state. On the contrary, fluctuations in categories are at least in part attributable to administrative convenience and compromises made in response to impossible political demands. The categories reflect the difficult and never-finished work of bureaucrats, politicians, and even enumerators to manage the impossible tasks of meeting public and political expectation that ethnic groups be counted, responding to a proliferation of demands for recognition, and simultaneously allaying fears of tribalism while conducting the enumeration that feeds it.

This, I suggest, is part of the reason the imaginary of the 42(+) has not been publicly interrogated. Veena Das offers the concept of magic as a way of understanding the illegible nature of the state in India. Das argues it is difficult for citizens and often even state officials to read ‘the rules’ and their sources of legitimacy. Magic and the illegible state share the mobilisation of forces that are opaque, but which have very real consequences. This combination of obscurity and power makes the forces of both magic and the illegible state potentially dangerous, and renders vulnerable anyone approaching them.⁶⁸ I add that magic and the illegible state both also have potential for beneficence. Both the threat of encountering dark arts and risk of bursting a bubble of good fortune make people reluctant to investigate.

The palimpsest that is the Kenyan census and the associated imaginary of the 42(+) is magic in this way. It is opaque insofar as almost nobody understands or asks how it is formed, but its consequences are real. The list can render groups powerful because of numeric dominance (or an imagination of it), and (il)legitimate as citizens. It can feed tribalism via enumeration, and inclusion via classification. Accordingly, there are risks associated with knowing too much. Asking could reveal you are not the only 43rd tribe, or your ethnic group has been outgrown by another. This is very different from, for example, Uganda, where the constitution lists 65 ethnic groups indigenous to the nation and allows open contestation over the list, notably by Asians, and African minorities (even if there is little follow-through on changing it).⁶⁹

In using the metaphor of magic, I am drawing attention to the politics of uncertainty. Other scholarly literature offers similar candidates for making sense of this, but none quite captures what is taking place here. One possibility is arbitrariness, which Tapscott and Fontein argue operates as a state strategy for control and fragmentation of a fearful population.⁷⁰ Elsewhere, Fontein, following Mbembe, similarly argues uncertainty can facilitate intentional and sinister political effects, such as the simultaneous reinforcement of fear and legitimacy of a ruling regime in the face of violence without a clear perpetrator.⁷¹ Ambiguity offers another conceptual possibility, implying multiple meanings are possible, allowing political actors to choose the most convenient.

Magic has some important differences to these concepts. First, magic allows for the strange relationship between certainty and uncertainty that characterises classification. The process and the consequences may be unclear and undecided, but codes and lists are, by definition, certain, in ways that arbitrariness, duplicity and ambiguity cannot account for. It is the way this certainty has been rendered uncertain by broad disinterest in interrogation that is important. Second, these analyses suggest desire or benefit for citizens associated with more certainty. Life would be more predictable and manageable; the state more transparent and democratic. In this case, it is not only the state but also citizens that are reluctant to push for certainty. It is in everyone’s interests to let sleeping dogs lie.

Kenyan census categories therefore also offer an important lesson about the relationship between magic, legibility and intelligibility. Illegibility of the process of classification extends in both directions: citizens do not know with any certainty how to get on the list, and the state, evident from the palimpsest, can never be sure of its categories. The rules cannot be read. Nonetheless, the classifications are not unintelligible because the social imaginary of the 42(+) provides a sticky, if ill-defined, common sense of who Kenya’s tribes are, capable of change without upsetting the overall logic and sense of nationhood. The concept of magic illuminates the nature and effects of illegibility here, as for Das, but when married with a social imaginary, also helps us comprehend a loose intelligibility.

Conclusion

Over coming years, precision and certainty around who is an ethnic group of Kenya may be necessary. As I write, a review of the nation’s electoral and administrative boundaries is underway, which may require clarifying whether ‘community of interest, historical, economic and cultural ties’ means ethnicity. If it does, some serious attention to which ethnic groups are where and in what numbers, and perhaps publication of local-level ethnic statistics may eventuate. Constitutional provisions for minority and marginalised groups similarly may require identification of names and numbers. These matters are politically sensitive, with potential for recognition and distribution toward smaller groups, and demographic posturing by larger ones. The magic of ethnic categories and numbers is that they are pregnant with possibilities, both benevolent and dangerous.

Yet, Kenyans have navigated challenging political questions about ethnic groups, their size and their place in the nation largely through avoiding precision and certainty. This method has not been a legible or deliberate one. Rather, it has emerged as common sense in the social imaginary of 42(+) tribes, and the general aversion to investigating who is on the list, how they got there or what it means to be on or off it. Whether or not this approach can persist – as inter-ethnic competition seems to continually intensify at the same rate as a commitment to peace, justice and inclusion – remains to be seen.

Acknowledgements

Thanks to Laurence Cooley, Victoria Stead, Gabrielle Lynch, Radha Upadhyaya, Kiran Pienaar, my writing group in the School of Humanities and Social Sciences at Deakin University, and two anonymous reviewers for invaluable feedback on this work.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by the Australian Research Council Discovery Indigenous Scheme [IN180100055].

Notes

1 Balaton-Chrimes, “Counting as Citizens.”

2 Das, “The signature of the State.”

3 Kertzer and Arel, “Censuses, Identity Formation.” For an example of historical research that counters this argument see Shellam, “‘Our Natives’ and ‘Wild Blacks’.”

4 Taylor, Modern Social Imaginaries, 23.

5 Ibid., 23–27.

6 The politics of the word ‘tribe’ remains conflicted in Kenya. I use it to invite reflection on its continued use and reproduction through the census, rather than as an endorsement. Colloquially, it is used commonly, and frequently without pejorative connotation, so a simple endorsement or refusal would be inappropriate. For opposing arguments from two anti-tribalist Kenyans, see Wamwere, Towards Genocide in Kenya, 95–7, for a defence of the word, and wa Thiong'o, “The Myth of Tribe,” 16–23, for an argument for its abandonment.

7 See Online Supplementary Appendix for a full table of all ethnic classifications (including sub-tribes), and the shifting types of categories (e.g. sub and main tribe) over time. It is difficult to determine a definitive number of ethnic groups counted in each census because of these different types of categories. Unless otherwise stated: analysis refers to final reports rather than code lists provided to enumerators; different spellings of the same group are not treated as different classifications; spellings used are those from the most recent census; classifications and population figures come from East African Statistical Department, “Geographical and Tribal Studies”; Kenya Population Census 1962; Kenya Population Census 1969; Kenya Population Census 1979; Kenya Population Census 1989; KNBS, Ethnic Affiliation; KNBS, 2019 Kenya Population and Housing Census.

8 This does not necessarily include code lists; and in 1948 some were classified as ‘other’ (reported only at the district level, not the national level), but this does not infer they were considered a sub-tribe. Tharaka were also counted in almost every census – excluded only in the 1948 report, though they were included in the code list that year.

9 Kikuyu was the biggest (8,148,668 or 17.13%) while Luo, Luhya (and sub-tribes), Kalenjin (and sub-tribes) and Kamba were all over 4 million.

10 Mhando, Safeguarding Endangered Oral Traditions.

11 Berry, “Hegemony on a Shoestring.”

12 Hirschman, “The Meaning and Measurement”; Appadurai, “Number in the Colonial Imagination”; Glassman “War of Words.”

13 Martin, “The East African Population Census,” 303.

14 Weitzburg, “The Unaccountable Census,” 422.

15 Martin, “East African Population Census,” 308.

16 Ibid., 312.

17 Ibid.

18 Morgan, Ethnic Geography of Kenya, 77.

19 Ibid., 80.

20 Kenya Regional Boundaries Commission, Kenya. The Commission discouraged the view that people occupying land in a district not ‘their own’ would be evicted, prophetically fearing its proliferation would ‘result in chaos and bloodshed’ (6).

21 Ranger “Invention of Tradition Revisited”; Parsons “Being Kikuyu in Meru.”

22 Glassman “War of Words”, 17. This is evident in submissions to the 1933 Carter Commission on land tenure and the 1962 Boundary Review Commission.

23 Appadurai, “Number in the Colonial Imagination,” 323.

24 Ibid., 331.

25 Ibid.

26 Gertzel, The Politics of Independent Kenya, 9; Branch and Cheeseman “The Politics of Control in Kenya”; Mueller, “The Political Economy”; Tamarkin, “The Roots of Political Stability.”

27 Branch et al., It’s our Turn to Eat.

28 Ibid; Tamarkin, “The Roots of Political Stability,” 307; Rothchild, “Ethnic Inequalities in Kenya”; Owiti, “Political Drivers of Inequality,” 551–2.

29 Lonsdale, “Moral Ethnicity and Political Tribalism,” 131–2.

30 See Lynch, “I Say to You” on Kalenjin identity formation.

31 Anderson, “Yours in Struggle for Majimbo”; Anderson “Majimboism”; Branch et al., It’s Our Turn to Eat.

32 Morgan, Ethnic Geography of Kenya, 86.

33 Anderson, “Majimboism,” 47.

34 Anderson and Lochery, “Violence and Exodus.”

35 Balaton-Chrimes, “Ethnicity, Democracy and Citizenship.”

36 D'Arcy and Cornell, “Devolution and Corruption,” 256.

37 See the presentation here: http://ictville.com/2013/02/video-the-tyranny-of-numbers-hypothetically-mutahi-ngunyi-predicts-a-jubilee-alliance-win/ accessed 27 September 2019.

38 For example, Maina, What Tyranny of Numbers?, 110.

39 Bratton and Kimenyi “Voting in Kenya”; Ferree et al. “Voting Behavior,” 155; Cheeseman et al., “Kenya’s 2017 Elections”; Throup, “Elections and Political Legitimacy.”

40 The Standard, “State Should Protect Sanctity of Census to Ensure Credible Result.” The Standard, 24 August 2019.

41 Under Chapter 7 of the constitution, one consideration for boundary redrawing is ‘community of interest, historical, economic and cultural ties’.

42 Simson, “Ethnic (In)equality.”

43 Gazette notice no. 7245, 4621, Kenya Gazette, Nairobi, 21 July 2017, Vol. CXIX – No.102.

44 Balaton-Chrimes, “Ethnicity, Democracy and Citizenship.”

45 NGEC, Who and Where? 32.

46 See Online Supplementary Appendix for figures.

47 Lynch, “Becoming Indigenous.”

48 Somalis can also be considered groups brought by the British.

49 Boru, “Burji Recognition in Kenya.” ‘Bulji’ were also counted in 1989 for reasons that are unclear.

50 Balaton-Chrimes, “Statelessness, Identity Cards and Citizenship.”

51 Curiously, Baganda were on the 2019 code list, but not in the final report.

52 This category was not on the code list, only the final report.

53 See https://www.unhcr.org/ibelong/statelessness-around-the-world/, 16 June 2020.

54 Boran were on the code list in 1948, but not the final report. This could be because there was no enumeration in their region because of logistical difficulties.

55 Lynch, “Durable Solution, Help or Hindrance?,” 604.

56 See https://www.theelephant.info/features/2018/03/05/what-is-your-tribe-the-invention-of-kenyas-ethnic-communities/ accessed 20 July 2020.

57 In 2019, the code list for enumerators listed ‘Refusal / Kenyan’ while the final report returned to the vocabulary of ‘Kenyan – so stated’. This underscores the authorities’ confusion regarding interpolation of people who don’t identify ethnically. While ‘Kenyan – so stated’ carries a tone of national pride; ‘Refusal’ suggests more a frustration or resentment regarding ethnicity. It could be, however, also simply an instruction to enumerators that if a person refuses to give their ethnicity, they are to be recorded as Kenyan.

58 Tobias Chanji, “President Uhuru Declares Makonde 43rd Tribe of Kenya,” Standard Media, 1 February 2017. https://www.standardmedia.co.ke/article/2001227966/president-uhuru-declares-makonde-43rd-tribe-of-kenya.

59 Balaton-Chrimes, “Counting as Citizens”; and informal conversation with leaders of both groups.

60 Weitzburg, Unaccountable Census, 411.

61 Scharrer, “Ambiguous Citizens.”

62 Colony and Protectorate of Kenya, “Report on the Census of the Non-Native Population,” 7.

63 East African Statistical Department, “Geographical and Tribal Studies,” 58.

64 The 1962 census explains “ … the indigenous African and Somali population of Kenya was grouped into 40 tribes; … The grouping is of course somewhat arbitrary in many cases. … Ajuran, who are part-Galla and part-Somali speaking, have been placed in the Somali group since they are largely Islamized and are now included in the North-Eastern Region.” Kenya Population Census 1962, 34–5.

65 In 1948 and 1962 enumeration in Northern Province was not done by household because of sparse population and logistical difficulties. Instead, a de jure count was undertaken, followed by sampling. Morgan, “Ethnic Geography of Kenya,” 79. Kenya Population Census 1969, 1; Kenya Population Census 1979, 8.

66 Weitzburg, Unaccountable Census.

67 Dillon, “Reinscribing De Quincey’s Palimpsest.”

68 Das, Signature of the State, 226. Das contrasts magic with a Weberian concept of rationality, but she may be overdrawing the distinction given they both share bureaucratic embodiment of complex rules, and differ only on legibility. I am, therefore, essentially delinking her use of ‘magic’ from her use of ‘rationality’ here because I am not making an argument about legitimate authority.

69 Article 10 grants citizenship by birth to anyone with at least one grandparent belonging to one of these groups. One can also be a citizen by birth without belonging to one of these groups, provided one has at least one grandparent or parent who is also a citizen by birth. See https://www.newvision.co.ug/news/1510310/indians-tribe-uganda accessed 8 June 2020; and see https://www.pmldaily.com/features/2018/07/minority-ethnic-groups-want-article-32-in-the-constitution-repealed.html accessed 8 June 2020.

70 Tapscott, “The Government Has Long Hands”; Fontein, “Anticipating the Tsunami.”

71 Fontein, “Remaking the Dead.”