Who’s Counting? Ethnicity, Belonging, and the National Census in Burma/Myanmar

Partially rooted in British colonial ethnology and administration, the ethnic categories of the 2014 Myanmar census have attracted controversy, particularly from repre-sentatives of non-Burman political organizations. The categories themselves, as well as the bureaucratic exercise of the census, have a complex genealogy which offers considerable insight into understanding the contemporary situation. Drawing from Hirschman’s theory that the study of measurements of ethnicity is a unique resource for understanding the meaning of ethnicity in a society, this article discusses the controversy surrounding the 2014 census, and how some census-related issues have been crucially framed by bureaucratic structures that came before.

the history of the census as a bureaucratic instrument offers a unique resource to understand the meaning of ethnicity in relation to the state (Hirschman 1987:557). The state of Burma/Myanmar has a decades-long history of internal conflict, with numerous armed groups organized along ethnic lines. Many studies of the ongoing ethnic conflict, however, fail to question the ontologies of ethnicity in the area, or how the machinations of state bureaucracies have sought to name, measure, and understand them. Thus, to examine the contemporary controversies ignited by the census in Myanmar, it is also useful to identify and describe the kinds of classification that came before, and in so doing illuminate the networks of power in which they are embedded (Bowker and Star 1999:42).
This article will discuss the historical role of the Burma/Myanmar census enumeration as part of a process for organizing-and creating-ideas about ethnicity, citizenship, and belonging. It will also reveal some of the difficulties that enumerators faced in making use of categories that did not neatly fit the ethnographic realities they sought to describe and quantify. By highlighting the relationship between political regimes and how they understand their sovereignties through census data, as well as the problems encountered in the data-gathering process, this article will also show how some of these very problems resonate with the contemporary criticism of the 2014 census project. To that end, this article is organized in four sections: First, it will present a historical overview of the British colonial censuses and discuss some of the issues that confronted colonial enumerators as they tried to categorize and understand language and race in Burma; second, it provides a brief discussion of the ethno-national character of the post-colonial regime and of how regime and bureaucratic changes framed the 1973 and 1983 censuses; third, it will analyse the implementation of the 2014 census, paying special attention to the criticism it has attracted; lastly, it will show how these contemporary census debates serve to demonstrate the fractal nature of the political exercise in general, and, in unfortunate irony, that those most critical of the ethnic categories of the census still depend on the categories for their political survival.

Counting on the Census
The main objective of the earliest censuses in ancient Greece was to assess military capability, and later states made use of them for purposes of taxation and resource allocation (Missiakoulis 2010:413). At the most basic level, the act of counting people intrinsically requires eligibility of those people in a certain class or group, but it also 'involves reciprocal performances in which the Downloaded from Brill.com04/27/2019 12:33:14AM via free access counted objects are complicit in, or resistive to, the social production of counts' (Martin and Lynch 2009:243). These acts, in their construction and manipulation, serve as a method for measuring phenomena which otherwise might not be directly measurable (Cobb and Rixford 1998:1). Census data have been used to create social statistics, which have also been referred to as indicators.
Although they were developed as instruments of government rule, in recent decades indicators have also been used by international groups concerned with measuring and comparing international aid, non-governmental organizations, and human-rights issues across national boundaries (Merry 2011:s83), a point that resonates particularly with the development goals as discussed by unfpa. The British colonial censuses introduced in India in the nineteenth and twentieth centuries effectively re-arranged local difference into figures that were legible to the colonial empire, providing an 'essential abstraction from social reality' (Kumar 2006:377), and later serving to frame ethnic problems and issues (Cohn 1996;Dirks 2001). Censuses and population statistics, as governing tools, are used to define minority populations, and, reflexively, majorities (Appadurai 2005:41), although the process is not necessarily unidirectional. Anti-colonial nationalist movements have made use of colonial map and census data to define national territory and population as a basis for their independence (Anderson 1991). Martin and Lynch (2009) coin the term 'numero-politics' as a moniker to describe the disputes regarding counts, estimates, and measures. Numeropolitical disputes can result in modifications of census categories and procedures. For example, changes in questions regarding race in the United States census have been informed by scholarly conceptualizations shifting from biological determinism to socially constructed models, as well as the efforts of political interest groups (Terry and Fond 2013:525). Great Britain, for example, started to collect information on ethnic groups in the 1991 census (Aspinall 2000:110), and community-based organizations in Ireland were able to gain recognition and legitimacy through the categorization of Travellers within the scheme of ethnicity (King-O'Riain 2007:518).
With the state creating a regime of classification, indigeneity also becomes an entity which is both bureaucratically and socially constructed, and several groups within a nation-state may lay claim to this status (Nah 2006:285). The use of categories involving race/ethnicity in a modern census is now internationally common: a study of 138 national-census questionnaires found that 87 of them (or 63%) used some kind of ethnic classification (Morning 2006:4). Therefore, the examination of the history of those categories in each of those nations can reveal key information not only about how a bureaucracy can make sense of, and quantify, the diversity of its subjects, but also how it responds to international ideas about modernity and human development. Finally, the study of census genealogies can give some idea (albeit incomplete and not always deterministic) as to how political organizations and social movements might frame their contestations of that bureaucratic process.

The Colonial Census Takes Form(s)
This 2014 exercise will be Myanmar's twelfth Western-style census. Some scholars of the ongoing ethnic conflict in the country have even argued that there has not been an 'accurate' census since the British colonial2 census of 1931 (Smith 1994:17;South 2011:10). Whereas the British began mapping the territory following the end of the first Anglo-Burmese war in 1826, the first3 colonial census was carried out in 1872, albeit of a significantly smaller territory than what the country is now (Bennison 1933:vii). It was not until 1891 that Upper Burma was enumerated, as those territories were annexed five years prior (India Census Commissioner 1904:58). From 1881, the colonial census repeated once every decade, and censuses from 1901 onwards were more uniform and consistent in their coverage (Maung 1997:9). The final colonial census was in 1941, but results were never completed; during the British evacuation of Burma in 1942, the 1941 census officer fled, leaving behind all of the data save two pages (Andrus 1947:24).
During the colonial years, the over-stretched British administration's goal was to collect all their data synchronously, and thus create a freeze-frame impression of the population. Generally, enumeration would take place over one week. Where officials were sparse, or terrain particularly difficult, the enumerators would make their best estimation (Richell 2006:26), and compilers did make it clear which parts of the country were surveyed 'synchronously' , 'non-synchronously' , and 'estimated' (Bennison 1933:vii;Lowis 1902:3).
Racial classification and exclusion constituted a major trope within the colonial state's empirical understanding of its subjects (Lee 2009:38) and this bureaucratic vision and practice would create very real consequences. Using contemporary ideas of ethnological and geographical mapping, the British Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 1-28 bureaucratically solidified the idea of an intrinsic, mappable connection between a 'people' and a 'territory' . They established 'Burma Proper/Ministerial Burma' as the lowlands which were annexed as part of British India, and the 'Frontier/Scheduled Areas' , where Kachin Duwas and Shan Saophas were given semi-autonomous authority over their subjects. As the historian Thant Myint-U writes, 'the territorial limits of the country, the notion of who is Burmese and who is not, key social and political structures, all find their origins in this period surrounding the fall of Mandalay' (Thant Myint-U 2001:10). Administrative divisions thus exacerbated the problem of ethnic definitions (Callahan 1996:71).
When they carried out the census, the British-trained enumerators required that respondents give single, discrete, unqualified answers for their race, and these would be fit into a prescribed categorical scheme (Charney 2009:9). The Burmese term for race/ethnicity used in the census is luumyo, literally, 'type of person' . Amyo, type, or sort, in reference to people is often translated as 'race' , though the understanding of it being biologically immutable (ascribed), or socially constructed (achieved), is ambiguous, and thus situational. The connection of one's amyo to one's language is a concept brought by colonial ideas of race and ethnicity. Prior to colonialism, in the Dhammasat legal texts, amyo implied class distinction (Ikeya 2011:25). In the pre-British Restored Toungoo and Konboung eras, particularly after 1630, Burmese regimes made use of censuses for purposes of taxation as well as for military recruitment (Lieberman 1991:21). They would not necessarily be concerned with race or language as a phenomenon which required counting or mapping; indeed, at the onset of colonialism the locals subscribed to a very different concept of ethnicity than the colonialists (Gravers 2007:13;Lee 2009:37). Their situation offers further evidence to the position that the very methods of counting themselves constitute social problems (Martin and Lynch 2009:244).
Regardless, it was census enumerators who had to solicit this information from respondents. The problems and confusing situations they encountered, given the tremendous ethnographic, cultural, and linguistic diversity of the terrain, particularly in Upper Burma, are veritably limitless. Some of these problems were discussed in the colonial census report volumes.
The system of caste, or a hierarchical social system based on occupation, prestige, and ideas about pollution, had been considered a master trope in the framing of the imperial census (Kumar 2006:387), and one which colonial officers had grown accustomed to in British India. Census officials even noted that caste could be tested by 'actual existing facts and beliefs to a far greater extent than is possible with respect to race' (Webb 1912:250). However, the categories of caste simply did not gain traction for census officials working in Burma; thus they struggled to find another way to approach human difference and quantify race4 in Burma. According to C.C. Lowis (1902), the Burman is 'so absolutely enamored of freedom that he cannot abide the bonds that caste demands' (Lowis 1902:107). Without the category of caste, and where religion 'indicates but little' , colonial surveyors concluded that language would therefore be the most 'obvious and surest criterion of difference' (Lowis 1902:112): If the speech of a particular community cannot be assigned to a particular group, that community is, ipso facto, isolated, whatever similarity its customs, dress and physical traits may have with the customs, dress and physical traits of any other community, neighbouring or otherwise.
Although there was such demonstrated reliance on language being the best indication of difference, the data did not always readily make sense. H.L. Eales noted that the number of Karen people increased an astounding 56% between 1872 and 1881 (Richell 2006:28), though later C. Morgan Webb made the observation that missionary influence allowed Sgaw and Pwo Karen languages to hold their own, and avoid any 'tendency whatever to succumb to the Burmese influence that surrounds them' (Webb 1912:206). That still does not explain a 56% population rise in less than a decade, though. Where some populations grew by unbelievable proportions, others, logically, shrank. The Tavoyans (those of modern-day Dawei) in 1901 had a total population of five, but according to the census report, the fault was theirs for returning their language as Burmese, rather than Dawei (Lowis 1902:76). It was later noted that the Shan peoples were so 'peculiarly distributed' that logical divisions were impossible (Webb 1912:237). Finally, the Turungs offered an even more perplexing case: they were from Assam, but had been captured by Singphos for five years during a journey through what is now Kachin State. During those five years, they claimed to have learned Singpho, but reported to enumerators that they did not speak Turung anymore (Lowis 1902:78).
Enumerators were sensitive to the fact that they would need to report immigrant populations, given the scale of immigration to the country-Rangoon being the world's number one port of entry for immigrants (Charney 2009:19)and quantifying this was no small task. When enumerators met with populations who spoke languages far removed from their own, for many cases they would report them as kala saga (foreign language, though the term kala is often applied pejoratively to denote Indians). Sometimes they would attempt to be more precise, but this would engender other problems. C.C. Lowis noticed that 'Bingala/Bengala was a term applied indiscriminately to the speech of persons from every portion of the empire' (Lowis 1902:92). This is prescient, considering the seriousness of the situation today, and the violent exclusion of the Rohingyas, who some insist are 'really' Bengali.5 Because of noted discrepancies, the officials for the 1931 census changed the way in which enumerators would solicit information regarding language. They were required to ask respondents for 'mother tongue as spoken from the cradle' , for column 14 of the form, and for 'subsidiary language […] [or] any other language commonly used by the speaker'6 to be written in column 15 (Bennison 1933:173). From 1921 to 1931, the Mon-speaking population increased 61 %, which was attributed to 'the fact that Burmese was returned at the 1921 census by many Mons as the language used in the home, whereas in the 1931 census they probably returned Mon as their mother tongue' (Bennison 1933:174). The colonial authorities have commented on the 'non-coincidence of census and linguistic boundaries' (Webb 1912:212) as well as the 'extreme instability of language and racial distinctions' in the country (Bennison 1933:174). Another noted that 'there is no insuperable boundary between the members of separate races, and still less between the members of separate tribes' (Webb 1912:250). In a special appendix to the 1931 Census, Captain J.H. Green writes: Some of the races or tribes in Burma change their language almost as often as they change their clothes. Languages are changed by conquest, by absorption, by isolation and by a general tendency to adopt the language of a neighbour who is considered to belong to a more powerful, more numerous, or more advanced race or tribe. green 1933:245 5 Rohingya Muslims' estimated population in Myanmar is approximately 800,000 (Kipgen 2014:236). They have also been at the receiving end of communal violence, and of claims that they are illegal Bengali immigrants. Following two outbreaks of violence in 2012, where thousands of homes were burnt and over a hundred thousand people were displaced to brutally inadequate shelters, the un described the Rohingyas as 'the world's most persecuted minority' . See Peter Popham, 'No end in sight to the sufferings of "the world's most persecuted minority"-Burma's Rohingya Muslims' , The Independent, 8-10-2012. http://www .independent.co.uk/news/world/asia/no-end-in-sight-to-the-sufferings-of-the-worlds-most -persecuted-minority-burmas-rohingya-muslims-8202784.html (accessed 7-7-2014). 6 Even though, as Leach noted, 'unity of speech' does not denote 'any deep subjective feeling of social solidarity' (Leach 2004:48).
This cues the well-known work of anthropologist Edmund Leach, who observed that more prosperous groups in the uplands demonstrated a tendency to adopt a Shan or Burmese style of living, and convert to Buddhism. As he famously noted, 'any individual can start as a member of one category and end up in another' (Leach 1960:62). But to what extent have these categories carried a legacy, both legal and social, beyond the colonial period?

Independent Enumeration: The Post-Colonial Years
Although many discussions of the decades-long internal conflict will chalk these problems up to British colonial 'divide and rule' practices, it is essential to take into consideration the ways in which ideas about ethnicity, nation, and political power changed following independence. On the brink of independence, British authorities demanded that Burmese politicians demonstrate cooperation with the Frontier Areas before granting full autonomy to the former colony. As such, an accord was hastily put together and signed at the southern Shan State town of Panglong, and later called the Panglong Agreement.
In this agreement, politicians representing the former Frontier Area states did agree for their constituencies to join the Union of Burma on a conditional basis for ten years, whereupon they would have the option to secede (and it was part of the 1948 Constitution that the Shan and Karenni states had this right) if it was the state majority's elected will to do so. One of the oft-repeated phrases used to characterize the pluralist vision of the Union of Burma was General Aung San's famous, Bama ta kyat, Shan ta kyat, or 'if Burma receives one kyat, Shan will receive one kyat' . In its third article, The Union Citizenship Act of 1948 was explicit regarding which groups would be full citizens: For the purposes of section 11 of the Constitution the expression 'any of the indigenous races of Burma' shall mean the Arakanese, Burmese, Chin, Kachin, Karen, Kayah, Mon or Shan race and such racial group as has settled in any of the territories included within the Union as their permanent home from a period anterior to 1823 a.d. (1185 b.e.).
Therefore, for the newly independent government, to be indigenous meant to exist in the territory demarcated as the Union of Burma before the first Anglo-Burmese War. It was not completely exclusive, though: Section 11, part 2, would allow anyone whose parents and grandparents had lived on the territories to be granted citizenship as well.
Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 1-28 The map for the newly independent Union of Burma was drafted such that there would be seven predominantly Burman taing (divisions), in the central lowlands and upper Dry Zone, and seven pyinay (states), largely populated by non-Burmans and named according to the presumed majority group in each state (Map 1). Such a division, with the seven Burman taing divisions and the seven non-Burman pyinay states, further reinforced the idea of official races, and the basis for bureaucratic representation. The 1948 Constitution designated that the Chamber of Nationalities, totalling 125 seats, would be comprised by 25 Shan, 12 Kachin, 8 Chin, 3 Karenni, 24 Karen, and 53 Burman seats (Constituent Assembly of Burma 1948:67). Even though the major political players were Burman, it is worth noting that there was not any 'special indigenous' status ascribed to one group, unlike the British privileging of Malays over orang asli in post-colonial Malaysia (Nah 2006:285). This is not to say, however, that Burmans necessarily thought of their race as equal in status to that of their non-Burman counterparts; part of the rhetoric which supports Burman ethnic chauvinism is the idea that Burmans led the anti-colonial struggle, and that they are the only 'major group whose total population resides within the territorial bounds of the state' (Steinberg 2001:190). By this standard, while other groups might be indigenous, only the Burmans are exclusively indigenous.
Even so, the vision for a newly independent Burma crucially depended on an image that included these eight official races, and this independent nation posited its indigeneity in opposition to foreign control, and more specifically, against non-indigenous capitalists such as the Chinese, Indians, and Europeans; similar to that of the Dobama Asi-ayone, this ideology was formed in contrast to a colonizing other.
Following independence, the Union of Burma government did not conduct a census for two and a half decades.7 The ongoing insurgencies, and especially the Kuomintang incursion, facilitated the growth of the army, and eventually led to Ne Win's 1962 military coup, the establishment of the Revolutionary Council government and the Burmese Way to Socialism. It was frequently articulated that the socialist economic system would be the basis for ethnic unity, and that building such a society would be the work of all people of Myanmar, not just the Burman majority (ေမာင္ ခ်စ္ လႈ ိ င္ 1969:32). Through the 1963 establishment of the Enterprises Nationalization Law, then the Socialist Economy Protection Law, the Burmese government took commerce out of the hands of private business owners, especially non-Burmese.8 The impetus for this nation-Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 1-28 alization was principally to oust foreign dominance and eliminate farmer tenancy (Steinberg 2001:140;ဦ းျမသိ န္ း 2012:47). However, the Burmese government gradually recognized the shortcomings of its isolationist economic policies, and in the 1970s began a series of state development plans and administrative reforms. In July 1971, the Party Central Committee of the Revolutionary Council moved to draft a state constitution for the establishment of a socialist democratic state (Moscotti 1977:6). At the same time, the government began to seek foreign aid.
International donor organizations, such as the World Bank, the United Nations and the Asian Development Bank, as well as a Japanese foreign assistance programme, responded enthusiastically to what they saw as the biggest liberalization move by the Burmese government in over a decade (Steinberg 2001:255). unfpa agreed to support a Population and Housing Census (unfpa 2013a), which took place in April 1973, and the population data enumerated by the census were used to compile constituency electoral rolls for People's Councils (Moscotti 1977:143). The government presented the census as being important for the development of the economy, the social welfare of the people, and for making new laws (ျပန္ ၾကားေရးႏွ င္ ့ အသံ လႊ င္ ့ ဦ းစီ းဌာန 1973:206).
During the colonial period, the British had made use of substantial numbers of civil servants trained in India; in fact, much of the mid-level colonial administration in Burma was Indian, a significant portion of whom settled in Burma after World War ii. But with the increasing ethno-nationalism and eventual expulsion of foreigners in the 1960s, the Burmese bureaucracy found itself lacking experienced professional administrators to carry out such a largescale bureaucratic endeavour as the census (Maung 1986:1). In spite of this, with unfpa support, they carried out the 1973 census, though they did not include significant swaths of territory that were controlled by separatist armies. For the enumeration process, the categories of race/ethnicity were condensed as 1. indigenous races (the eight major races); 2. non-indigenous or foreign races; and 3. mixed Burmese and foreign. Unsurprisingly, the returns suggested a largely indigenous population, mostly chalked up to the mass emigration of foreigners in the prior decade (Maung 1981:17). Furthermore, it is probable that individuals of mixed parentage would be inclined to self-report as one of the indigenous races, given the anti-foreign political climate at the time. one-party states in Eastern Europe and China (Steinberg 1987:400). This system also served as a mechanism for moving military officers into bureaucratic positions, thus parliament consisted largely of retired generals (Nakanishi 2013:167).
The above-mentioned regime changes, together with the census, sufficed to give donor organizations the green light to provide development assistance to Burma, and the 1970s saw a tremendous expansion in aid money, especially compared with the isolated 1960s. For example, the World Bank allocated 30 ida credits, worth us$804 million, and Japanese support increased to us$ 178 million by the end of the decade (Steinberg 2001:255). All of this development support came only after a census, significant bureaucratic restructuring, and suggestions of return to civilian rule on the part of the Burmese government. It is particularly illuminating to consider these issues in light of the debates regarding the consequences of the 2014 census, as will be illustrated later in this article.
Burmese Socialist Programme Party rule continued, and in 1982, just before the next scheduled census, they revised the 1948 Citizenship Law, and made it more stringent regarding who would be genuine citizens of the country. They used ideas of jus sanguinis, as detailed in the first part of the former citizenship law: only those belonging to the eight national races would qualify for citizenship, and those ethnic groups who had settled within the state prior to 1823 would be allowed Burmese citizenship (1982 Pyithu Hluttaw Law No. 4); they deleted the second part of the citizenship law whereby people whose grandparents had settled in the territory could register as full citizens.
The 1983 census, again supported by unfpa (unfpa 2013a) is widely criticized for its lack of independent supervision, and the fact that there were substantial populations in conflict areas which did not enter the survey.10 Although the 1983 census, like its 1973 counterpart, failed to define the term 'race' for its enumerators, specifying it merely as the 'ethnic origin of the person being enumerated' (Maung 1997:4), the 1983 exercise did introduce a new scheme for categorizing ethnicity: a list of 135 sub-categories within the eight national races. How the government arrived at the number 135 is a bit of a mystery.11 One argument suggests that they adopted the number from the 1931 British Census, which lists 135 languages. Informants in Myanmar have suggested that the number came from Senior General Saw Maung, founder of the State Law and Order Restoration Council (slorc). When he was once questioned as to why the military was having difficulty with 'ethnic' insurgencies, he responded that the problem was partly due to the large number of ethnic groups: 135. Following this proclamation, the government institutionalized the number 135 as an expansion of the eight official races. Another, even more arbitrary, suggestion is that because General Ne Win's favourite number is nine, the government devised 135 sub-categories, as one plus three plus five is nine.
Scrutinizing the list, one can find numerous inconsistencies. For example, some groups are listed according to exogamous names, whereas others are listed by endogamous names. Many are listed according to the name of their language, while others are named based on their location or principal town. One group, the Mro/Wakim, is listed under both Chin and Rakhine. The preponderance of group names, arguably, has unnecessarily subdivided some races. For example, Kachin political leaders have argued that six of the named 10 Tun Tun Thein, 'Census begins in Burma's Kachin state' , Democratic Voice of Burma, 6-3-2014. http://www.dvb.no/uncategorized/census-begins-in-burmas-kachin-state -myanmar/38125 (accessed 7-3-2014). 11 To be fair, the bspp government had previously acknowledged 144 groups in the country, but they were not micro-managed in the form of government id cards.
Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 1-28 twelve sub-groups are but clans within other groups; therefore, there are six Kachin groups, not twelve. The same case has been made for the Karen/Kayin race, with the government scheme placing the number of sub-groups as nine, but Karen leaders saying that there are five. Within the total Chin population of an estimated one million people, the government list specifies 53 sub-groups, but these listed groups are, in some cases, merely alternative spellings of the same name, the name of a variant dialect of one language, or again, clans within another sub-group.12 In addition to recognized groups finding themselves incorrectly identified or subdivided, the list of 135 excludes a number of groups. A few of the groups not included in the list are: Panthay Chinese Muslims, Overseas Chinese (speakers of Hokkien and Cantonese), Anglo-Burmese (Eurasians of mixed Burmese and European background), Burmese Indians, Gurkha, Pakistanis, and Rohingyas. The issue of the exclusion of the Rohingyas has become hotly contested in recent years; how this affects census enumeration will be discussed later in this article.
Following the 1988 protests and the military overthrow of the bspp regime, the new ruling regime effected a controversial name-change. Part of the justification for changing the country's name from Burma to Myanmar was the idea that the new name would be more inclusive of all of the people in the country, not specifically the Burman majority, an idea which had been articulated decades ago (ေမာင္ ခ်စ္ လႈ ိ င္ 1969:31). It would result in a cognitive separation of ethnic identity and the name of the state, in other words, a Bamar or a Shan could be a full citizen of Myanmar Naingan (Taylor 2008:221). This is debatable, however, as both names are derived from the same word (Burma simply being the colloquial version of Myanmar) and have the same meaning, and thus the name change would be unlikely to create a more inclusive nation (Lintner 2003:174).
The State Law and Order Restoration Council regime, specifically Khin Nyunt, was able to convince a number of ethnic insurgent groups to consent to ceasefire agreements, which resulted in several of them gaining significant legitimate economic autonomy and sovereignty. However, the slorc regime (with its more sonorous endonym, State Peace and Development Council, or spdc) did not carry out a census in 1993, in spite of plans to continue the decennial census (Maung 1997:9). It was only following the new 2008 Constitution and the 2010 reforms that the need for a new census was brought to the table.

The New Myanmar and the 2014 Census
The United Nations Population Fund, in consultation with an international technical advisory group and the Myanmar Ministry of Immigration and Population, spent 18 months developing the format and implementation plan for the 2014 census.13 The official enumeration period was scheduled to take place from 30 March to 10 April 2014, and the enumeration form itself consisted of 41 questions. According to preliminary tests, the form required about 25 minutes for each person to complete with the assistance of a trained enumerator. Some reports suggest that the individual enumeration process actually took around 45 minutes.14 Agreements with the ministry and unfpa stipulate that enumerators are required to write down exactly what their respondents say.
Intentionally dishonest reporting attracts a 50,000 Kyat fine, and/or a prison sentence of one month. The survey calls for participants to identify themselves with one major race (one of the eight national races) as well as one sub-group (of the official 135 ethnic groups), though it does have an 'other' option (code number 914), where the respondent can write in the name of her or his ethnic group. In an informational pamphlet issued by the government about the upcoming census, the reason stated for collecting information on ethnicity is explained as follows: Ethnicity of an individual can define culture and behavior of a person and can play differing roles in how some people are shaped and influenced in society. To be able to understand why in some areas the demographic indicators are higher or lower than another area, one has to understand the norms of the people based on ethnicity. Information on ethnicity can also be used in research to further understand how different groups of people are evolving, especially where marriages, migration, education, work and other factors bring different ethnicities into close proximity and may lead The brochure states why ethnicity is important to the census project, but fails to define what exactly ethnicity is, nor does it provide any specific instruction on how to differentiate between ethnicities-though the last sentence seems to suggest that there is potential flexibility in the category due to political and socio-economic circumstances. In February 2014, the director-general of the Ministry of Immigration and Population, U Myint Khaing, claimed that the project had the agreement of all the national races.15 For minority groups, passing a certain minimum representation, as evidenced by the new census, would mean ministerial representation in local governments.16 However, those places would need to be surveyed first. Mountainous geography and poor roads, coupled with seasonal rains, make places in Kachin State and Shan State (the northern and eastern areas) difficult to access and survey efficiently. Some groups petitioned the government to commence the census project in those areas one month earlier (လူ မြ န္ 2014), a request which does have precedence in the 1931 colonial census: the British gave enumerators three months in advance to survey 'sparsely populated, difficult areas' as part of the 'non-synchronous' sections of the census (Bennison 1933:vii). This request was not granted for the 2014 census, as unfpa standards for a modern census require a narrow window for enumeration.
The other geographic issue for enumeration is the fact that areas of the Shan and the Kachin states remain controlled by armed separatist groups. U Aung Myint, a spokesperson for the Wa National Unity Party, said in an interview that an accurate census of Wa areas would help to build peace and economic development, ultimately contributing to a higher standard of living for people in the areas (ခ်ိ ဳဇင္ မာ၀င္ ့ ၊ ေမစု ု ၾကည္ 2014). The Wa areas of Shan State were part of the territories not included in the 1983 census, for reasons of 'security restrictions' , and it had been estimated that those areas contain a population of 1.2 million people (Spoorenberg 2013:310). Another article suggested those Wa areas are home to 450,000 people of 17 or 18 nationalities (ခ်ိ ဳဇင္ မာ၀င္ ့ ၊ ေမစု ု ၾကည္ 2014). Approximately 80,000 residents of Kachin State were estimated to have been excluded 15 Nirmal Ghosh, 'Myanmar census risks stirring ethnic tensions: Zooming in on race and religion can aggravate ongoing sectarian violence' , The Straits Times, 20-2-2014. 16 Samantha Michaels and Yen Snaing, 'Burma urged to drop census questions on ethnicity, religion' , The Irrawaddy, 13-2-2014. http://www.irrawaddy.org/burma/burma-urged-drop -census-questions-ethnicity-religion.html (accessed 10-3-2014).
from the 2014 census because those areas are controlled by the Kachin Independence Organization (Yen Snaing 2014).17 Most of the controversy with the 2014 enumeration process, though, is directed at the interpretation and application of the scheme delineating 135 ethnic sub-groups. It was noted that people in Dawei have tended to identify with the 'Tavoyan' category (number 505), whereas the large population from Dawei that is now in Yangon has chosen 'Bamar' .18 To drum up Mon political numbers, an mp from Chaung Sone Township in Mon State campaigned to encourage more people to self-identify as Mon (number 601), regardless of whether they could speak or read the Mon language, if they believed they were descended from Mon people.19 Do recall that Mon populations expanded greatly between 1921 and 1931, largely due to the rephrasing of the colonial census (as detailed above).
According to mp Aung Naing Oo, many ethnic Mon have already been classified as Bamar, as noted on their national identity cards or household registration cards. For the census, some of these people fear that if they were to try to respond as 'Mon' to census enumerators, they might be accused of false reporting.20 In this case, it is bureaucratic precedent that would be used as a litmus test to determine the 'authentic' response.
Depending on which category is ticked by the census-taker amongst Kachinidentified populations, the Kachin National Council, for example, sees the large number of sub-groups as potentially 'alienating and breaking up ethnic national identity' (Yen Snaing 2014).21 This identity as Kachin partially comes from Jinghpaw emerging as a lingua franca amongst various tribes, but then became necessary for political solidarity (Maran 2007:43). The distinction between clan versus sub-group is blurred, too, as inevitably in such ethno-nationalist movements there is the push for minorities to surrender their language and culture for the purpose of national unity (Lee 2009:40). The more categories, the lesser the likelihood that the minimum population threshold required for ministerial representation will be achieved. There is also the bureaucratic reality that the greater the number of minority groups, the lesser the likelihood for political solidarity of non-Burman political organizations (Gravers 2007:4). This resonates with the resistance to the Statistical Directive 15 in the United States in 1997, whereby census respondents would be allowed Bijdragen tot de Taal-, Land-en Volkenkunde 171 (2015) 1-28 to choose multiple categories. African-American and Latino rights leaders saw the move as a 'whitewash' that would result in a loss of ethnic and policy-related distinctions for their groups (Bowker and Star 1999:4).
A group of Chin activists petitioned President Thein Sein as well as the immigration minister for a re-drafting of the census, as they noted that the form had gotten a number of tribal group's names wrong.22 Whereas the Chin activists, with their greater proximity to such populations, would be in the position to assist the state in adjusting nomenclature, the problem fails to acknowledge the fact that the bureaucratic exercise is in a constant relationship of compromise between the exigencies of the state and the intrinsically un-classifiable and unbounded nature of ethnic categories and identities. For example, the Palaung (Ta-ang) of Namhsan in the northern Shan State speak a Mon-Khmer language, and have long interacted with other Shan Saophas over the centuries. The Myanmar census categories place them as one of the 33 sub-groups of Shan, but some have argued that they should have a separate category (Yen Snaing 2014).23 Specific coding potentially leads to further volatility in the case of excluded categories of people. It has been argued that the percentage of Muslims in the general population was purposely underestimated in the 1983 census; the International Crisis Group (icg) claims that Myanmar is 10 % Muslim, though previous censuses put the figure at 4%. They fear that the collection of data amongst Muslim groups could potentially fuel additional sectarian violence; as the population of Muslims in the previous census was quite small, if a greater number was recorded in a contemporary census it might suggest to anti-Muslim activists that Rohingyas are not indigenous, but rather were recent migrants to the Rakhine State.
Do recall, though, that the 2014 census is designed to offer an 'other' (number 914) write-in category for respondents to self-identify should their ethnic category not appear on the list. Prior to the enumeration, there were reports of some census-takers in Rakhine State being concerned about this. 'What do we do if Bengalis ask us to use the term "Rohingya"? We have to go to each and every house. We are afraid of being attacked if we don't fulfil their demands. There have been problems whenever we are taking information from Bengalis' , claimed one of the enumerators, a female schoolteacher.24  '[u]sing that term (Rohingya) seems to suggest that we accept it as a new ethnic category. It would lead to problems again. We should take action against those who answer with inaccurate or non-existent ethnic names.'25 The issue of false reporting harks back to the lessons learned by the British colonial enumerators in that there is a presumed single, discrete ethnic identity immutable to an individual, and to suggest otherwise would be a lie. One city elder from Sittwe, Than Htun, is quoted as saying, 'If we accept the use of "Rohingya" in the census, then Myanmar will become the destination country for all Bengalis migrating around the world who call themselves "Rohingya". Bengalis do not want to obey the 1982 Citizenship Law and they want to exploit the census to become citizens of Myanmar.'26 Where many residents of Myanmar see the census as a mechanism for identifying citizens for future allotments of social services, this xenophobia suggests fear that 'aliens' seek to legitimize themselves in the eyes of the state, and therefore avail themselves of these resources. The volatility and hotly contested nature of the 'Rohingya' issue, though, raises questions as to whether it would even be within the bounds of strategic essentialism to call oneself Rohingya in such a politically charged climate.
Shortly after the census-taking commenced on 30 March 2014, news of difficulties emerged from two areas in the Rakhine State-Ten Nga Village, Kyauktaw Township, and Tha Win Chaung (called Bassara in the Rohingya language, located just south of Maungdaw)-and enumerators stopped conducting the census. According to reports, enumerators in these areas were not allowing respondents to claim Rohingya ethnicity, and instead forced them to claim Bengali.27 In a last-minute decision, the Myanmar government announced that it would not allow members of the Muslim minority in Rakhine State to selfreport their ethnicity to enumerators as 'Rohingya' . Presidential spokesperson Ye Htut is quoted as saying, 'If we ask a family about their ethnicity and they say Rohingya, we will not carry out [or] accept it.'28 This sudden move was met with mar, 13-3-2014. http://www.elevenmyanmar.com/index.php?option=com_content&view =article&id=5362:don-t-mention-rohingya-in-national-census-local-rakhine-residents& catid=44:national&Itemid=384 (accessed 15-3-2014 ctually there are some Muslims identifying their ethnicity in the census as "Bengali" and some as "Kaman". However, others are bitter and defiant, and insist on calling themselves "Rohingyas", a term that is recognized by neither the Arakan (Rakhine) State government nor the central government.'30 A few weeks later, in a report released following the prescribed census finish date, the Immigration and Population Ministry claimed that over 6,000 families, mainly from the Buthidaung, Maungtaw, and Sittwe townships 'accepted themselves as Bengalis' , and therefore were registered on the census (New Light of Myanmar 25 April 2014).31 The degree of 'acceptance' (or coercion) these people experienced is beyond the scope of this article, but the controversy itself shows that authentic self-reporting, as prescribed by the United Nations, is hardly a reality in a politically charged climate. The unfpa published provisional results of the 2014 census in August of that year, and some were surprised by the population total: 51.4 million peo-ple32 (unfpa 2014b). Prior to enumeration, the government had estimated that the population of the country was approximately 60 million. The provisional census results published in August 2014 included information regarding the population and household size, broken down according to state/region, dis--rejects-rohingya-census-classification-causing-problems-among-muslim-communities .html (accessed 31-3-2014 trict, and sex. The provisional results did not include data on ethnic group, and U Myint Kyaing, the secretary of the Central Census Commission, was quoted as saying that the complete census data would be announced after the elections the following year.33 In October of 2014, U Myint Kyaing announced that the public release of the census data in May 2015 would not include data on ethnicity, religion, or occupation.34

Conclusion
After reviewing these issues with the 2014 census, along with the ways in which the operation has attracted controversy and how the Myanmar government has (or has not) responded to the controversies, it now probably seems ironic that numbers, with their supposedly objective, rational basis promised a 'depoliticization' of politics (Rose 1991:674); by their very nature, indicators project an aura of objective, rational truth, and facilitate broad comparisons (Merry 2011:s84). It is these 'purely technical' ideologies that function to mask the politically and socially charged agendas behind such bureaucratic exercises, and their repetition serves to further institutionalize, and thus conceal, those charged agendas (Bowker and Star 1999:139). Mimiko (2006) argues that it is the politicized and contested state structure and its poor handling of issues of ethnicity that have made the census in Nigeria problematic; if the governance were decentralized with genuine federalism, the census would become depoliticized (Mimiko 2006:1). Ethnic nationalities in Myanmar have designed their own censuses to verify the data of the official, national one.35 But would international aid organizations accept these alternative censuses?
In light of the history of the census in Myanmar, genuine federalism is a project, and one which political leaders, themselves empowered along ethnic lines, have often advocated. The level of politicization surrounding the datagathering project is certainly in stark contrast to that of public 'indifference' as noted by the 1931 census (Bennison 1933:viii). In the case of Myanmar, both unfpa and the government are continually assuring the public that the census exercise is 'purely statistical' , despite sceptics such as Saw Kyaw Swar, secretary 33 Hein Ko Soe, 'Religion joins ethnicity on list of Myanmar census results to be delayed' , Mizzima News, 2-10-2014. http://www.mizzima.com/mizzima-news/myanmar/item/13349 -religion-joins-ethnicity-on-list-of-myanmar-census-results-to-be-delayed (accessed 2-10-2014). 34 Soe, 'Religion joins ethnicity on list of Myanmar census results to be delayed' . 35 Ghosh, 'Myanmar census risks stirring ethnic tensions' . of the Karen Affairs Committee, thinking the results will be used for political purposes.36 Those political purposes (and political outcomes) have already been foreshadowed by the kinds of controversy the operation itself has ignited. Leaders of ethnically based parties and ceasefire groups are vocal in their resistance to the 2014 census's use of ethnic categories, but often insist that the categories be used differently, or that there are more accurate categories which should be used, rather than abandon the use of ethnic categories altogether. Because the results of this 2014 census will surely be taken seriously, it is important to consider them in light of the history of the census-taking exercises in both colonial and independent Burma, as well as the pragmatic histories behind those exercises. Whereas the routinized methods of data-collecting regarding ethnicity in many contexts are more likely to follow precedence rather than create new categories, the existence of numerous separationist armed groups creating fissures surrounding these categories throws their validity into question in other ways. This proves two major points: that studying the history of the census in a country such as Myanmar offers an important repository for understanding the ways in which the state understands human diversity and seeks to manage it; and secondly, although the empirical operation is relied upon to divest the aura of a 'political agenda' from the state through seemingly apolitical numbers, the very process of counting (and being counted)-or not-is highly political indeed.