Stay informed: Sign up for eNews Subscribe

Government of Paper The Materiality of Bureaucracy in Urban Pakistan

Read Chapter 1



Islamabad is linked to Rawalpindi, a city in the Potohar region of Pakistan in Punjab province, by several miles of the Islamabad Highway, a divided four-lane road from which autorickshaws, ubiquitous in Rawalpindi, are banned. The car-driving classes frequently quip, "Islamabad is five minutes from Pakistan," with Rawalpindi standing in for Pakistan. Poorer residents, often relegated to slower modes of intercity transport such as buses and minivans, more commonly joke, "Islamabad is ten minutes from Pakistan." The joke is a comic recognition of one of the original goals of the Master Plan: to distance the government from the society it was to govern. In its initial conception, Islamabad can be seen as giant anticorruption machine, an effort to use spatial isolation to engineer a social isolation of government servants from the wider populace. Unlike most modernist projects that aimed to make aspects of a wider society legible to the government (Scott 1998), Islamabad was designed to make the government legible to itself, partly through isolation from the wider society and partly through its own internal order.

Timothy Mitchell argues that the distinction between state and society "is a defining characteristic of the modern political order" (1999:184), even though "the edges of the state are uncertain; societal elements seem to penetrate it on all sides, and the resulting boundary between state and society is difficult to determine" (88). He argues that an "apparent boundary" (176) between state and society is produced by Foucauldian "disciplines," practices of "spatial organization, temporal arrangement, functional specification, supervision and surveillance, and representation" (185). As an effect of such practices, the line between state and society is not the perimeter of a separate entity, but "a line drawn internally, within the network of institutional mechanisms through which a certain social and political order is maintained" (175, emphasis added). Seen in this light, the spatial organization of Islamabad was not the project of an autonomous state actor, but an attempt to create such an actor by cutting the entanglements of the state bureaucracy with Pakistani society. It was an effort to make officials of a modern state who are, in James Scott's words, "at least one step-and often several steps removed from the society they are charged with governing" (1998:76). The city thus takes its place in the traditions of South Asian dual cities and cordons sanitaire.

Islamabad also drew on much older practices of separation: those effected by documents. Scholarship on colonial India has emphasized the objectifying knowledge practices of documentation: surveys and ethnography, mapping, and enumeration. However, the most fundamental function of documentation, evident in the earliest practices of the English East India Company as in all bureaucratic organizations, is to constitute the organization by distinguishing actions done for the organization from all others. In contrast to the gross social sorting of physical planning, documents differentiate among the actions of individuals, distinguishing official from unofficial, or public from private, actions. This problematic of separation was especially acute in the early Company because of its corporate character and the organization of its trade.

Although the Letters Patent (charter) specified who would be a member in the early East India Company, the Company needed to determine when its members were acting on their own and when on behalf of the Company. It was the documentary practices of accountability more than the Letters Patent that constituted the Company as an organization rather than simply a society of individuals with trading privileges. Company documentary practices exploded with the consolidation of trade in joint stock, constituting the "jointness" of joint stock trading by defining when a purchase, payment, or shipment was a Company action. Important in London, such documentation was even more central in India, where the so-called private trade, that is, the trading of Company servants on their own accounts, constantly threatened the integrity of the organization. These concerns about differentiating the private actions of Company servants from their Company actions were later expanded and racialized in the mature colonial state as a problem of native corruption. I'll argue that this problematic partly accounts for the pervasiveness of writing within the Company, within the British colonial governments, and within contemporary Pakistani governance.

As Akhil Gupta observes, discourses about bureaucratic corruption in India portray the actions of low-level officials as "thoroughly blurring the boundaries between 'state' and 'civil society'" (1995:384). But rather than focusing on the blurring of a boundary between two independently constituted domains, it is more productive to follow the practices that make, remake, and undermine the difference between the actions of government and all others. As we would expect, the Master Plan has been only partially successful in establishing a sociospatial order liberated from spatial practices prevailing throughout urban Pakistan. Islamabad may be five minutes from Pakistan, but banned horse carts (tongas) still ply the shoulder of the Islamabad highway. Neighborhood and kinship relations pulse within the bureaucratic procedures designed to ensure the correspondence of residential and bureaucratic hierarchies. "Private" work is still done through government offices. Documents, the very mechanisms for protecting the integrity of government, are often the means through which it is undermined.

Splendid Isolation

The establishment of a capital city for the new state of Pakistan was born in the turbulent politics of the state's first decade. The issues that came to the fore in the debates about a new capital were those that dominated national politics more generally, especially the fundamental question of the relation of the government to the populace. The halting failure of electoral politics in the 1950s, undermined by the maneuvering of an alliance of the civil bureaucracy and the military, eventually brought a martial-law government to power in 1958. As Ayesha Jalal argues, the new civil and military officials embraced "a policy aimed at depoliticizing Pakistani society before it slipped into the era of mass mobilization" (1995:55; see also Alavi 1983; Burki 1986; Sayeed 1980). The establishment of Islamabad was an expression of this program.

The founding of Islamabad followed years of debate on the construction of a new capital area in Karachi. Following Partition in 1947, civil servants of the government of Pakistan were housed in evacuated buildings, tents, and temporary quarters in Karachi. Two alternatives for a permanent government seat in Karachi were considered. The first plan was hastily produced by Lt. Col. G. Swayne Thomas, an Australian who was a town planning consultant to the Government of Sind. The plan advocated a new administrative satellite city of sixty-five thousand inhabitants twenty to thirty miles northeast or east of Karachi. This city would have contained the offices as well as the residential areas of civil servants, creating a postcolonial version of the "civil lines," the orderly civilian settlements of British colonial government servants. Objections to the plan were raised, however, because it isolated the government from the rest of Karachi and, symbolically, the Pakistani people.

A second plan completed in 1952 by the Swedish firm Merz Rendel Vatten responded to this criticism:

The authors of the Plan have, at an early stage, emphasized the desirability of promoting as close a contact as possible between the state administration and the economic and cultural functions. Specifically this means that the Capital and its administrative buildings should be located near the old town, with its business life and its cultural institutions.... The desire to isolate the Capital in a new and separate town, or section of the town, has appeared to the authors to be an echo of ideas from a past era during which the functions of the state were confined merely to the responsibility for a certain degree of order and a certain disposition of justice. In such a community the state system could be segregated and could, in magnificent surroundings, manifest its supremacy in splendid isolation. (Lindstrom and Ostnas [1952] 1967:36)

The new plan placed the federal government enclave in the middle of Karachi, on an extension of the existing central business artery. The report stated, "The new capital and the existing business section should be given the possibility of growing together into one common core, built around one axis only" (Lindstrom and Ostnas [1952] 1967:2). The large avenue, running from the central business district to the capital area, would open up onto a large open space framed by public buildings dominated on one side by the parliament building and on the other by a mosque. This space, a hexagon with sides 1,400 feet in length, would be large enough for a million people to gather for political meetings, public assemblies, and Eid prayers. Housing for government servants was to be distributed throughout the city in new finger-shaped districts extending from existing boundaries of Karachi.

Criticism was leveled at the Merz Rendel Vatten plan for its costly traffic plan and its extravagant central plaza, but this second capital plan was undone not by the expense; rather, it was soon overrun by the tumultuous political events of the 1950s. In 1951, before the report had even been published, Liaquat Ali Khan, the prime minister who had initiated the project, was assassinated. He had succeeded Mohammad Ali Jinnah as head of the Muslim League, the party that had lead the movement for the creation of Pakistan. The Muslim League was the only national party at the time, but its support came mainly from Urdu-speaking immigrants from north India (so-called Muhajirs). Ironically, the Muslim League had never enjoyed strong support in the areas that came to make up Pakistan. Liaquat Ali Khan's death severely weakened the party and initiated an "institutional shift from elected to non-elected institutions" (Jalal 1995:55).

Years of deeply contentious politics followed, in which various prominent political leaders from Karachi attempted to govern at the center through opportunistic and quickly shifting alliances with political leaders from East Pakistan and the West Pakistan regions of Punjab, Northwest Frontier Province (NWFP), Sind, and Baluchistan (see map 1). [PLACE MAP 1 HERE]A more stable alliance between the civil bureaucracy and the military shaped the political intrigues in an effort to consolidate state power itself. Simultaneously, the prominence of Muhajirs within the upper echelons of these two institutions was gradually replaced by that of ethnic Punjabis. In 1956, the Constituent Assembly ratified a constitution, but the document generated strong opposition from provincial leaders outraged by its centralizing provisions and from religious leaders decrying its tepid Islam. Infighting among National Assembly members, challenges from the provinces, monstrous inflation, and wheat shortages combined to generate massive street protests by public and private labor unions, opposition parties, Karachi businessmen, and students. The army presented itself as the only guarantor of stability and openly confirmed its dominance over elective politics. In October 1958, Field Marshal Mohammad Ayub Khan, soon after becoming chief martial law administrator, deposed President Iskander Mirza and installed himself as president as well.

Shortly after assuming office, Ayub Khan formed the Special Commission for the Location of the Federal Capital to reexamine the establishment of the capital in Karachi and, if necessary, recommend another location. While most of the commission members were technical professionals, it was headed by Major General Yahya Khan, a close associate of the new president. The Final Report of the Special Commission has never been made public and, as Frank Spaulding (2003:369-70) observes, while Ayub Khan later claimed that the commission recommended the Rawalpindi region for the new capital, there are suggestions the commission called for further study. At the time, with the Final Report remaining secret, officials spoke for it through the press. Newspapers reported that the Special Commission report found Karachi unsuitable and recommended locating the capital at its present site in northern Punjab, on agricultural land north of Rawalpindi, near Ayub's native village of Haripur. Newspaper accounts, at least one of which claims to quote the report, state that the report of the commission articulated a range of reasons to move the capital to the north. The region's moderate climate and changing seasons would prevent boredom and lassitude and promote health and administrative efficiency. Lying on the Grand Trunk road, the site offered advantages as a center for the region's economic development. The availability of cheap rural land would decrease development costs. The geography of the region and the military base in nearby Rawalpindi gave the site greater strategic security than the seaside Karachi. Planning from scratch would allow greater order and beauty; the Commission reportedly observed that "although Karachi is a relatively modern city, its development has been unplanned and grotesque. It cannot be converted into a city of aesthetic beauty-an essential requirement of a capital." In addition to these advantages of the northern location, Constantinos Doxiadis, the Greek planner appointed to design the new city, would soon praise the "typical, characteristic architecture of the [Rawalpindi] area, growing out of the land, the people and the climate," while finding "that of Karachi area ... strongly influenced by Hindu principles" (Doxiadis Associates 1960b:68).

However, although most of the commission's purported justifications for shifting the capital were technical or aesthetic, the strongest reason was openly political. The commission reportedly argued that Karachi was unacceptable because commercial development rendered it "unwholesome from the point of view of administrative integrity." Having just moved against the commercial groups in Karachi with the support of the military and civil bureaucracy, General Ayub was anxious to protect government servants from what he considered the corrupting influence of Karachi businessmen, most of whom were Muhajirs (Ayub Khan 1967; Jalal 1990; Sayeed 1980). A newspaper quoted the commission report declaring that:

Close contact between the business community and personnel of the Administration at Karachi has not done any good to either. Too much of social contact between those who want things to be done to suit them and the officials cannot lead to healthy results. It is desirable both for the business community and the administration that the capital should be away from the commercial center of the country.

The commission was quoted arguing that not only the location of the capital but also its layout "should be designed to restrict contact between Government servants and business circles." While the report emphasized the distance from Karachi, many observers suspected that the proximity to Rawalpindi was equally important to the new bureaucratic military government. From the new site, Ayub Khan could keep a close eye on the army, which was in Rawalpindi, to prevent moves against him. In fact, Rawalpindi was immediately made the interim capital, even though most of the ministries were expected to remain in Karachi until the new capital city had been established.

Fifty years earlier, the British had similarly opted to escape the political agitation in Calcutta and the city's "foreign" associations (stemming from its origin as a British trading settlement) and located the new imperial capital outside Delhi. Following a major uprising throughout north India in 1857, the British had established a major military center southwest of Delhi and developed the city as a transportation center. Delhi had been the seat of the last major indigenous political power, the Mughals, and the British attempted to portray themselves as the successors of the Mughals in the subcontinent's grand history of imperial rule (Cohn 1987; Metcalf 1989). The British imperial capital been completed only three decades before Partition and had been home to many of the officials who left to form the new government of Pakistan. The Merz Rendel Vatten plan made indirect reference to Delhi in its criticism of the proposal for an isolated administrative satellite city. The significance of Ayub Khan's decision, therefore, was obvious. This was a move from a city founded by English traders, still strongly identified with British colonialism and dominated by in-migrants from north India, to a city in the Punjab near Taxila, seat of the Mauryan Empire, now claimed to be the forerunner of modern Pakistan. The move took the government away from a contentious political environment in a geographically peripheral coastal city and placed it alongside the country's military center in the geographic center of the dominant West Pakistan territory.

Under martial law, public opposition to the move was muted. However, many considered the move disrespectful to Jinnah, head of the Muslim League party, who had selected his new political base in Pakistan, Karachi, as the site of the new capital. Many businesspeople thought it would stunt economic growth. Opinion in East Pakistan was even more sharply opposed, suspecting that the shift would destroy the precarious balance between the two distantly separated and culturally different wings of the new country. East Pakistanis, recognizing the political impossibility of locating the capital in Dacca (now Dhaka), were satisfied with the cosmopolitan Karachi, which was dominated by migrants and accessible by a relatively inexpensive sea route. While Karachi was obviously much harder to reach from East Pakistan, it was at least geographically peripheral to West Pakistan and could therefore represent a meeting of the east and west wings of the country on something approaching equal terms. The move from Karachi to a city in northern Punjab confirmed the suspicions of East Pakistanis that West Pakistan, and in particular Punjabis, intended to dominate East Pakistan. East Pakistanis were not mollified by the plan for two parliamentary buildings, one in each wing, in which parliament would sit in alternate sessions. However, with the bureaucratic military government dominated by Punjabis in place, opposition to the move was ineffective.

Construction on Islamabad began in short order, but the location of Pakistan's capital was not settled finally until 1970. Following a short and indecisive war with India in 1965, deteriorating economic conditions, labor militancy, student radicalism, and opposition from provincial leaders led by East Pakistan generated demands for a restoration of parliamentary government. By the end of 1968, Ayub was also facing street protests from Islamists and even from usually quiescent groups such as teachers, doctors, and low-ranking government employees. In March 1969, Ayub ceded power to general Yahya Khan. By the end of that year, a wide range of political leaders were calling for the capital to be returned to Karachi. Once again, the main issue was the relation of the government and the "people."

Both the English and Urdu press covered the issue with enthusiasm, giving extensive coverage to arguments by political leaders in East Pakistan and Karachi from the Muslim League, Awami League, and the United Front. "Both provinces have a claim on this city [Karachi]," declared the Karachi United Front president. A veteran Muslim Leaguer from Karachi argued that east-wing political leaders considered it their "second home" and many owned property there; furthermore, they thought that "if Karachi became again the capital of Pakistan Quaid-i-Azam's [the Great Founder Jinnah's] wish would be fulfilled." Newspaper articles argued there were no advantages to shifting government employees from Karachi to Islamabad and confidently discussed the future of Karachi as a capital. Soon east-wing politicians demanded that the capital be moved to Dacca. One leading East Pakistan Muslim League officer declared that a dictator had moved the capital to Islamabad "purely in his own interests" and that the "sufferings of East Pakistanis will not come to an end until the central capital is shifted to this Wing of the country. ... [I]f the capital is shifted here the integrity, sovereignty, existence, and ideology of Pakistan will be free from challenge." While many in the civil service supported the return to Karachi, the military was set against it. President Yahya Khan, the chief martial-law administrator, argued such a transfer would be too expensive and difficult. Talking with reporters, he quipped, "you don't want us to be a mill ox [kolhu ka bail, which goe], do you?" likening moving the capital back to Karachi to a mill ox going round and round.

Facing a complete refusal by the central regime to consider moving from Islamabad, east-wing leaders had accepted by the fall of 1970 that the capital would not be shifted but insisted that the "Second Capital at Dacca" would not be allowed to be a "Consolation Capital." The position of Islamabad was confirmed the next year when East Pakistan broke away after a brief war to become the independent state of Bangladesh. The war also brought the populist bureaucrat and landowner Zulfiqar Ali Bhutto into power. Bhutto himself was invested in Islamabad, having been involved in its design and construction as minister of works under Ayub. His populism led him not to return the capital to Karachi, but merely to add to Islamabad a grand prime minister's residence.

The Dynapolis and the Colonial City

Following the report of the commission, Ayub Khan put the planning of the city in the hands of Constantinos Doxiadis, a Greek modernist architect and planner. Doxiadis had founded one of the largest architectural firms in Europe and was given the title of "chief consultant" on the Islamabad project. He had proved himself to the new regime by quickly completing Korangi, a housing development in Karachi for the resettlement of seventy-five thousand refugees. Even so large a project as a new capital city was hardly a match for Doxiadis's ambitions. His grand vision transcended monumentality, focusing on the universal requirements of a whole society, even a civilization: "I have an obligation to follow only that road ahead of me that is not obstructed and cluttered up with monuments, a road whose largest shadows will be cast by simple plain human buildings" (Doxiadis 1963:195). Following this road required nothing less than the foundation of a new practical science, which he called "ekistics" (from oikos, the ancient Greek word for "house" or "household"): the study of human settlements to discover the relations among nature, man, society, "shells" (buildings), and "networks" (communications) (Doxiadis 1968). Such a broad approach demanded an all-encompassing role for the planner: "He must become a scientist, carry out research, create a system of thought, devise a programme of action and carry out proper schemes of organization in government, in industry, in production, in design" (Doxiadis 1963:9).

For Doxiadis, egalitarian concern for the life of the common citizen required a scientific organization of cities as a solution to the contemporary "urban nightmare" of unregulated growth (1963:19). However, the political role entailed by this technical program converged with the authoritarian political program of Ayub Khan's bureaucratic military regime. James Holston has observed that the modernist architects maintained affiliations across the political spectrum, aligning themselves with "whichever authority, on the Left or Right, seemed capable of implementing total planning (1989:42). Le Corbusier, the leading figure of modernist architecture, dedicated his major publication, The Radiant City, to "AUTHORITY" (Holston 1989:42). Doxiadis too welcomed the establishment of political power strong enough to carry through planning programs of massive scope. In his first major report on the Master Plan for Islamabad, he defined his own political role and that of his successors using the rhetoric of technical necessity: "It is imperative to create the master builders, the people who are going to be in charge of the overall city, from its conception to the implementation of every detail. There is a necessity for a conductor of the whole orchestra which is to create the symphony. He must be a strong conductor, for he will be responsible for everything within Islamabad" (Doxiadis Associates 1960b:435). This conductor was institutionally realized in the Capital Development Authority, established in 1960 and granted comprehensive planning and administrative powers, which I discuss in detail later.

While Ayub Khan saw the isolation of the bureaucracy as a means to control the country, Doxiadis saw it as means to control the city. Doxiadis declared that the city of the future must include "all social, all income groups and all types of functions" (Doxiadis Associates 1960b:108). However, he embraced the views of the Federal Capital Commission:

The influence of the diverse in origin and cosmopolitan population of Karachi on government administration would be eliminated, if the Capital were to be a capital only without non-official civilian population located in it and pulling it in different directions.... The capital should be in a place where the business community does not come into contact with administration on a social level. (Doxiadis Associates 1960b:54)

The unity of the Master Plan itself was an icon of the single-mindedness of a modern administration and its insulation from the social influences of the present as well as the tenacious grip of the traditional past.

Not only the provisions of the Master Plan but also the process of its production and authorization were covered extensively in the English and Urdu press, fed liberally by government press releases. Newspapers carried articles about the preparation of various reports, Doxiadis's comings and goings, and his meetings with senior government officials, including Ayub Khan. The Master Plan itself was the object of well-publicized state rituals in the 1960s and 1970s. Foreign dignitaries received tours of a museum enshrining the plan, materialized through city maps, photos, models of buildings, and charts of statistics. Afterward, such dignitaries were taken to Shakarparian Park, where they would ceremonially plant a tree beside the concrete garden grid plan of the city. Echoing the future-oriented nationalism of the new government, Doxiadis proclaimed that Islamabad was "to be created without any commitments to the past" (1965:6). The development of the metropolitan complex in Islamabad and Rawalpindi, however, lay in the tradition of British colonial city building. The political program of the bureaucratic military government and the spatial organization of the new metropolitan complex parallel the British colonial government's political objectives and the spatial orders through which it attempted to realize them. The division of the metropolitan complex in Rawalpindi and Islamabad into areas for the bureaucracy, military, and "non-official civilian population" was a perfect spatial expression of the new regime's vision of the polity. The absence of a spatialized colonial racial division is certainly significant. However, this tripartite organization can be placed in the context of colonial patterns of bureaucratic and military control and urban configuration, well exemplified by Rawalpindi itself.

Located on the roads linking the Khyber Pass and the Kashmir Valley with the cities of Lahore and Delhi, Rawalpindi was for centuries a minor trading city and district administrative center under the successive regimes of the Mughals, Sikhs, and British. The city was surrounded by hundreds of agricultural villages ranging in size from a few dozen to several thousand inhabitants. Rawalpindi became a major city after 1880 when the British selected the site as a military center for the forces defending the northern "frontier" of the empire. From a population of only 52,000 in 1880, Rawalpindi grew into the largest city of the region after Karachi and Lahore, with a population of 185,000 in 1940 (Specht 1983:37). Since the British recruited their military forces heavily from the Punjab, the ethnic composition of the city remained relatively similar despite the population growth. In the tradition of British colonial city building (King 1976), spacious areas for civilian authorities ("civil lines") and military forces ("cantonment") were added to the densely settled area of the indigenous city. The distinctness of each of these three zones is easily exaggerated, since each of them included human and nonhuman elements of the other two within them (Glover 2008; Hosagrahar 2005). This was even more the case in Rawalpindi, where the areas allocated to the civil lines and cantonment virtually surrounded the indigenous city to the south, east, and west.

The Indian areas of Rawalpindi, east of the Leh nala (seasonal river) and north of Murree Road, were structured by mohallas (neighborhoods) generally divided from (and joined with) one another by bazaars along wide transversal roads. Bazaars were not considered part of any individual neighborhood and had only the broadest social affiliation, permitting the passage of any person for any purpose any time. Bazaars thus served as gathering places of socially diverse people for trading and small-scale manufacturing. Mohallas were usually structured by galis (narrow branching lanes) usually stemming from single points of access that terminated in cul-de-sacs. The entrances to mohallas were generally from bazaars and were architecturally marked by wooden doors or gates, or simply by the narrowness of the lane. The branching and sharp turning of lanes limited physical and visual access, providing gradient degrees of seclusion to residents, shielding them from nonresidents and from residents of other parts of the mohalla. Small internal chowks (squares or intersections), usually at junctions of lanes, provided collective space for weddings, holiday celebrations, and political meetings.

In the architecture and use of domestic space, there was a similar progression from common to possessed areas continuous with that of the mohalla as a whole. Most houses were made of brick, stood one or two stories, and had a central courtyard with rooms opening onto it on the first story. In multistory dwellings, upper-story rooms opened onto an internal veranda that overlooked the courtyard. Many of the structures, often grouped around small open spaces, were occupied by several families. Although most of the buildings were occupied by a single household, in many cases a single family would have exclusive use of only one room and a small storage area; all the residents of the dwelling would share the internal veranda, courtyard, roof, and toilets. These shared spaces mediated between the common space of the lane outside the dwelling and the spaces reserved for the exclusive use of the family.

In Rawalpindi as elsewhere, the social and spatial organization of the older urban areas frustrated the attempts of authorities to make them legible and governable. As is often remarked, British colonials found Indian areas of cities to be disorderly and confusing, by nature inaccessible to colonial modes of surveillance and control. The streets weren't numbered and seemed innumerable, in both senses. The definition of the house was complicated by the variance in the number of physical structures and social groups found behind a single door or mohalla gate. The complexity of the ownership of structures in Indian areas made the purchase of land for reconstruction projects costly and complicated. Furthermore, government intrusions in these areas frequently provoked political unrest. For these reasons, the British usually resigned themselves to fragmentary or small-scale interventions, such as regulating building heights and setbacks and widening existing roads. Aside from the major restructurings for strategic reasons following the uprising in northern India in 1857 (Gupta 1981; Oldenburg 1989), the only large-scale government undertakings in Indian areas were utilities infrastructure projects: drainage, water supply, and later electricity (Glover 2008).

The expansion of the British presence in Rawalpindi coincided with the admission of the first generation of Indians into the civil service. In Rawalpindi, many of these civil servants moved into newly developed areas outside both the indigenous city and the developed areas of the civil lines. The settlements of these better-off Indians were often followed by those of low-ranking government employees and household servants, who gardened, cooked, and cleaned for the nearby British residents. Colonial officials sometimes surveyed the land and shaped the overall layouts of these new settlements through the construction of road networks, but often the land was sold on the open market and houses were constructed according to the owner's preference. The result, here as in many north Indian cities, was a new urban form that Mohammad Qadeer has termed the "new indigenous community" (1983:180-84). These indigenous enclaves near and in civil lines, never mastered by the colonial administration, featured many of the elements of the indigenous city.

Covering entire lots, the two- and three-story houses of these neighborhoods created a street facade of continuous high walls broken only by screened windows and balconies. These houses overlooked linear streets usually about twelve to fifteen feet wide, wide enough to permit vehicular traffic, but far narrower than the avenues of civil lines. The interiors of houses were generally built around atriums or enclosed balconies, allowing for sunlight and air circulation, but at the same time structuring the interior spaces to face inward and away from the public spaces of the street. While the blocks and streets were laid out in discernible geometric patterns, rectangular or semicircular, land uses were not segregated by function. Artisan workshops, bakeries, firewood stalls, and warehouses were interspersed with homes, and the continuous buildings accommodated shops at ground level. In new indigenous communities, as in precolonial urban cores, linearity marked market sites. The main thoroughfares of new indigenous communities quickly developed into markets.

In Qadeer's view, this new form of community combined modern conceptions of public health, ease of access, and land subdivision with the more traditional preference for spatial proximity and an interweaving of living areas and commercial sectors. The symbolism of these modern "utilitarian" aspects of the new communities was also important for the new class of educated Indians. In the 1930s, with the growing salience of communal identities, corporate groups began to develop exclusive new indigenous communities for Sikhs, Hindus, Christians, and Muslims. Following Partition in 1947, Rawalpindi grew rapidly as immigrants from eastern Punjab, now under Indian control, settled there, making up 40 percent of the population of 340,000 in 1959 (Ahmed 1960). From 1947 on, the salience of religious identities to the differentiation of neighborhoods was eliminated by the departure of Sikhs and Hindus, and more residences of civil lines itself became available to the now-Pakistani elite as the British exited. The organizational basis of new and existing indigenous communities also shifted at this time. From this period, the construction of new indigenous communities was led by extremely varied types of associations and corporations (sometimes including a range of classes), such as government departments, military divisions, private corporations, and ethnic or regional associations.

Indian enclaves in civil station-never mastered by colonial administration-had many of the elements of the indigenous city itself. These neighborhoods filled in the areas between civil lines and the indigenous city. The civil lines of Rawalpindi, like that of Lahore as described by Glover, "never attained the grid-like, tree-lined, bungalow-dotted clarity" often described for civil lines (1999:100). This blurring of the indigenous city and civil lines was greatly accelerated by the arrival of immigrants following Partition.

It was precisely this gradual intermixing that Doxiadis sought to prevent through his plan for the region. In line with his science of ekistics, Doxiadis, unlike more conventional modernists, insisted that planners needed to understand local conditions to plan effectively. He little comprehended the sociocultural dimensions of urban environments in the region; nevertheless, through careful studies of Rawalpindi and the surrounding rural areas, he came to understand at least the physical structures of dwellings and some of the environmental factors that contributed to them. One senior planner told me that, having conducted systematic studies, Doxiadis was better informed about traditional rural and urban architecture than today's planners. While Doxiadis was fond of the humble adequacy of Punjab rural houses (he did not examine villages as wholes), his understanding of local urban areas only increased his revulsion for them. Like most planners, he disliked the haphazard development of Rawalpindi, the lack of distinction among transportation arteries, the intermixing of functions, the congestion, and the lack of simple overall unity apparent from maps (Doxiadis Associates 1960a).

Doxiadis bluntly argued that "Rawalpindi should not have any role [in the capital]. It should remain the regional center ... [and] the servicing center of the capital" (Doxiadis Associates 1960b:244). Doxiadis wrote, "a green belt is provided between Islamabad and Rawalpindi in order to form a physical barrier between them" (Doxiadis Associates 1960a:54). The Master Plan for Islamabad called for Islamabad and Rawalpindi to expand indefinitely on parallel rays out from their nuclei, forever divided by a green belt, a major transportation artery, and a linear industrial zone (see fig. 0.1 in the introduction). Initial plans even provided for a wide military zone running parallel to the industrial zone (Yakas 2001:82), highway, and green belt, but this was eliminated in later plans. This division loudly echoed the one between New Delhi and Old Delhi and similar divisions in many other colonial cities, which represented the relation of the imperial government to its subject population and functioned to ensure or reinforce the social division between rulers and the ruled. There was, however, one major difference: New Delhi's hexagonal Beaux Arts plan was a complete whole-a symbol of imperial order and permanence-and no provision was made for the city's expansion or articulation with peripheral growth. By the 1950s, however, large-scale urbanization made the management of expansion a major priority of urban planning throughout South Asia (Hull 2011). Doxiadis planned for growth. His plan provided for the parallel, unidirectional expansion of Islamabad and Rawalpindi to ensure that they would never grow together and that concentric growth would not choke the "center" of Islamabad (Federal Capital Commission 1960). This program has indeed succeeded in keeping the two cities at a distance.

The monumental "national administrative" area was sited at the highest point of Islamabad, against the Margalla Hills, at the origin point of the city's ray of extension. The President's House, sometimes called the "Presidential Palace" in early newspapers, was placed on a small hill at the focal point of the city. The Federal Secretariat and the National Assembly buildings were located directly in front of the President's House, at a lower level. As one senior CDA official put it to me, "The president is on the hill and the parliament is under his feet."

The main axis of Islamabad, a large avenue officially named with the honorific form of reference for Jinnah, the Khayaban-i-Quaid-i-Azam (Avenue of the Great Leader), begins at this point. The land on both sides of this avenue was zoned as the central business district and early on became known as the "Blue Area" after its color designation on initial zoning maps.

The basic spatial structure of Islamabad is a grid of 11/4-mile square sectors that extends from northeast to southwest down a gentle slope, articulated by roads cutting through hillocks and spanning shallow gullies. Each sector was given an alphanumeric designation for precise location within the overall plan. Doxiadis called Islamabad a "dynapolis" (from "dynamic" and "metropolis") and praised the grid because it "can develop dynamically, unhindered into the future, into space and time" (1965:26). In Doxiadis's vision, the nucleus of Islamabad would remain serene, secure, and static while the city expanded indefinitely, eventually to be absorbed by what he called an "ecumenopolis," an urban blur smeared from Brussels to Beijing. For this reason, the limited set of alphabetic designations was assigned to the southeast axis, while the northeast to southwest axis of expansion was given the unlimited set of numerical designations. The grid of the Federal Capital Area, approximately 3,626 square kilometers annexed from border areas of the Punjab and North West Frontier Province (now Khyber Pakhtunkhwa), surrounds Rawalpindi on three sides, much as the British cantonment surrounded the indigenous city. While publicity focused on plans for the civilian population of Islamabad, as Frank Spaulding (1994:193) points out, the army was the quiet recipient of the largest amount of land allocated to any single institution or to any single function, save that of civilian residence.

Communities of All Classes and Categories

Doxiadis conceptualized Islamabad as a hierarchy of "communities," from the smallest gathering to the city as a whole. In a rationalized form of the neighborhood unit originally developed by American planners (Hull 2011), each sector was functionally differentiated and subdivided into a five-level hierarchy. The principle of subdivision was division into four parts, to preserve the geometry of the square, though irregularities of terrain often impeded the realization of this ideal order. The sector as a whole, a class V community, was to contain between 30,000 to 40,000 people. The sector was subdivided into four quadrants (subsectors), class IV communities, to house roughly 10,000 people. Four class III communities (sub-subsectors), with populations of around 2,500, were to make up each class IV community. Each class III community contained a number of class II communities, a block with a population of 100 or more. The lowest level, class I, was to consist of a family or any gathering of two or more people (Doxiadis 1964:332). Subsectors and sub-subsectors were numbered from one to four in a clockwise direction beginning with the southernmost division. So a particular area might be designated with a letter and up to three numbers, for example, G-6/4-1. Doxiadis observed that geometrical orders, in particular that of the square, were prevalent in the architecture of the region from Mohenjo Daro to the Lahore Fort (Doxiadis Associates 1961c:139)-leaving aside the aberrant "old cities" like Lahore and Rawalpindi. His real source, however, was the rationalism of European modernist planning, so striking in its simplicity.

Good architecture demands rationalism in the city plan, and this rationalism in turn requires consistency in the conception of all spaces forming the city. The room, the smallest nucleus of a house, must have straight walls, and these must be at right angles to each other so that they can be connected with the other rooms; otherwise there is no house. The house and its plots should have straight walls at right angles to each other so that they can be connected with other houses and other plots. The plots as a whole form a block, and the blocks, too, should have straight walls at right angles to each other so that they can be connected in a rational way to the other blocks. More blocks form a neighborhood, more neighborhoods form a city (Doxiadis Associates 1961c:12).

Schools, markets, mosques, medical institutions, and recreational facilities commensurate with their populations are focal at every level of community (fig. 1.1). In contrast to the mohallas, these functions are located in the center of each community. Thus, at the center of a class V community is a post office, a fire and a police station, a large mosque, clothing and food markets, and so forth. A class III center includes a primary school, a teahouse, a few shops, and sometimes a small mosque. Internal streets are numbered but not named. Doxiadis intended that each level of community be laid out so that its boundary could be seen from its center and the whole of its area easily imaged. Each community was to be iconic of the whole. The size, location, and quality of materials of every structure and street were planned so that it would be possible to see their replication of and incorporation into a higher order (Doxiadis Associates 1961c:18).

This ordering of space was consistent with a totalizing and external perspective on the city (de Certeau 1984), which could not grasp the mohallas of Rawalpindi. William Hanks (1990:296) contrasts two kinds of spatial systems: "centered" systems, in which spaces are defined relative to activities and speech events; and "co-ordinate" or "grid" systems, in which relations among objects are computed relative to fixed axes and dimensions. Hanks discusses these systems in relation to the social actors' spatiotemporal orientations and representations. Such orientations and representations may, in addition, provide bases for the material organization of a built environment. Islamabad is literally organized by a grid down to a small scale, making any location in the city theoretically specifiable in coordinate terms. By contrast, the mohalla, originally organized to control visual access and movement, is a centered system. While structurally the area of a mohalla is precisely (if irregularly) defined perimetrically, the perimetric definition is not directly perceptible. In contrast to Islamabad, it would seem that a location within a mohalla is almost unspecifiable in coordinate terms.

This spatial and functional order was to be the foundation of the social order as well: "The structure of a residential community is that its physical pattern should be in complete accord with the social organization of the human group which is to settle there" (Doxiadis Associates 1961a:6). The future inhabitants of residential sectors were conceptualized as a population organized by the national bureaucratic hierarchy, rather than as groups formed around family, religion, tribe or ethnicity, regional affiliation, or wealth, all of which are significant bases of social order in Pakistani society. These forms of sociality largely shaped the settlement patterns of Rawalpindi's mohallas. The social homogeneity of squatter settlements testifies to the continuing salience of such regional and religious affiliations even in urban Islamabad. As Spaulding (1994) has shown in his study of ethnic Gujars in Islamabad, the bureaucratic process through which government housing and plots are distributed makes it impossible for any group to establish itself in a particular area.

Continuing a British colonial practice, planners prescribed a hierarchy of lot sizes and house designs corresponding to residents' salary level and position in the government bureaucracy (King 1976; Nilsson 1973). The initial Master Plan provided for the construction of nine types of government houses (see table 1.1).

The correlation of house types with income, however, is misleading. The Socio-Economic Survey of Rawalpindi, while hastily completed in 1960, gave the planners an impressive range of information on housing and income in urban Punjab. However, the income divisions employed in the plan were not derived from this study but from the salary scale corresponding to the rank hierarchy of Pakistan's National Civil Service. Furthermore, government houses were to be allocated on the basis of the occupant's government rank, not income derived from any source. Doxiadis's criticism of variety within house types makes clear the status basis of the housing policy.

Civil servants who have more or less the same income and belong to the same class of civil service should be allocated to similar units. They should all be treated in a similar manner and on an equal basis as far as possible. Houses given, for example, to peons should all be of the same nature, of the same design, and the same accommodation capacity. Otherwise, bad feelings would be created among civil servants belonging to one and the same class (Doxiadis Associates 1961b:24).

Even by the mid-1960s, on Doxiadis's recommendation, the direct reference to income found in early reports had been replaced by correlations with civil service rank, as in table 1.1. [PLACE TABLE 1.1 HERE] The rank hierarchy of the civil service is officially called the "Basic Pay Scale" (BPS), a financial designation that is somewhat misleading, since the more important role of the scale is status rather than income differentiation. "Grade," the term for the rank of a government employee used in common conversation, better captures its social significance.

In general, higher-rank houses and larger plots for private houses were distributed within the northern sectors, with particularly high-status plots found in the easternmost areas near the administrative area (fig 1.2). The F-series of sectors, on the whole, contains government and private houses that are much larger than those of G-series sectors. However, Doxiadis's "neighborhood unit" plans did provide for a degree of integration of income groups.

The neighborhood unit was a notion developed in the United States in the early twentieth century as a means to generate and represent the egalitarian and democratic community. In laying out the basic objectives of Islamabad, the Federal Capital Commission had proposed "to create neighborhoods that will foster a sense of belonging to a community and promote social cohesion" (1960:19). Drawing an implicit contrast between the modern, nationalist social divisions of government rank and the socially retrograde hierarchy of traditional Hindu India, the Federal Capital Commission feared what it called a "caste system" based on income groups (19). This goal limited slightly the commission's emphasis on the isolation of the bureaucracy. The plan for the new city, declared the commission, would not "segregate completely for the distribution of incomes government servants from other groups" (19, emphasis added).

In the Master Plan, however, the egalitarian planning notion of the neighborhood unit was given a hierarchical and paternalist cast. It was feared that an attempt to integrate too broad a spectrum of residents would lead to social conflict. Rather, a principle of a "gradual integration" of a narrow range of income groups was adopted "to help lower income people mature and assure comfort to the higher income classes" (Doxiadis 1964:333). Accordingly, the range of house types in any given area was limited to two or three steps on the schedule. In a review of the Master Plan, a newspaper article enthusiastically praised this aspect of the plan, echoing the Federal Capital Commission's invocation of caste:

Another notable aspect of the Capital's "personality" would be the absence of a "caste system" among its inhabitants. Efforts will be made to secure an integration of the various strata of society. The higher classes will jostle with their fellow citizens in the lower income brackets. It is fully realised that this cohesion is possible only up to a point if embarrassments are not to mar the atmosphere. Anyhow, the planners appear keen that Islamabad should come as near to the ideal of a "casteless" city as humanly possible.

Over the last four decades, housing for lower-income classes has continued to be built, but Doxiadis's original plan for a gradually integrated city stretched no further than his designs for F-6 and G-6. Even F-7 contains few smaller plots and no government housing. As a senior CDA planner, observed, "Over the last twenty years there has been a complete separation of income groups." According to him, mixed-income projects simply cannot be pushed through nowadays. Why could it be done before? "Because," he said puckishly, "the gora [white] sahib had the clout" to push it through. While the disappearance of this objective is clear, accounting for it is more difficult. His explanation points to how much planning, conceived as the realization of rational intention, is shaped by wider sociocultural trends: "I can't point to a specific place in the record that the decision was made. It was not a conscious decision. It was a matter of a change in the culture."

As many have observed (King 1976), government grades constituted a kind of social template through which all members of British colonial society, not only government servants, could be placed in a social hierarchy. The Pakistan scale of government grades performs a similar function in contemporary Islamabad, where even positions for private teaching and business jobs are given a BPS equivalence in advertisements. The tendency to use the government scale as a representation of the general order of social status, while a national phenomenon to some extent, is especially strong in Islamabad. In its early years, the macrosocial order of the city was roughly co-extensive with the organization of the government, leaving aside the limited population of traders and workers. An article from the early 1970s describes Islamabad as "still a government city predominantly populated by government employees neatly divided into so many classes inherited from the good old colonial days." Today, class or status (rarely distinguished with Weberian precision) is still commonly identified with government rank. In discussions with me in Urdu about the social order of the city, residents moved fluidly among the terms haisiyat (roughly, "status" with an emphasis on capacities and resources), category, and scale.

From Separation to Participation

In October 1963, the CDA terminated its contract with Doxiadis Associates, two years before the end of the scheduled five years. The reasons for the early termination are unclear, though it appears that two of the central factors were disputes over the pace of Doxiadis's work and the desire of Pakistani officials to play a greater role in the development of the city (Yakas 2001:147-51). Several important CDA officials had been influenced by Doxiadis's approach through training in the School of Ekistics in Athens and through their work on the Master Plan itself. Eminent international architects such as Gio Ponti, Edward Stone, and Derek Lovejoy continued to design major buildings and landscapes for Islamabad. Lesser-known architects, mostly English, labored on designs for houses, mosques, and markets. Nevertheless, after the charismatic Doxiadis departed and the senior political and military officers withdrew from direct involvement in the planning, long-standing South Asian bureaucratic practices began to assert themselves in the development of the city.

One CDA planner told me, "The Master Plan is our constitution, the constitution for the city, for the country." Doxiadis's Report No. 32 has continued to be the basic reference for the Master Plan, though slightly revised in 1985 and again in 2006. However, few CDA officers I talked with have a copy. When I asked about Report No. 32, I was referred to the CDA library in a building away from the main CDA compound. As with legal practice, the day-to-day tasks of the CDA are not constitutional matters. Although the basic provisions of the report are rarely violated, the planning and regulation of the city are carried out through millions of more humble documents such as files, house plans, and forms.

As with the spatial plan of the city itself, the politics of reports produced by technocratic planners and high-level committees effectively isolated broad public participation in plans for the new city. As we've seen, even the most modest documents of the South Asian bureaucratic traditions similarly aimed to make a boundary between the government and its subjects, a division between public and private. In practice, such documents are often mediators that incorporate aspects of the people, things, and processes they are designed to control from a distance. This was brought home to me one day when I saw a CDA draftsman storm down a hallway and explode into a heated argument with a building control officer who had rejected a house plan on a technicality. The clerk I was with explained that the draftsman himself had made the drawings for a private client, which his superior knew. The plan had transformed them from superior and subordinate into bureaucrat and client. According to the norms of this latter relationship, the superior was reportedly hitting up the subordinate for Rs. 3000 to pass his plan. The house plan articulated conflicting market and bureaucratic relations. The two men simultaneously argued about the technical application of rules as subordinate and superior and the just price for approval as bureaucrat and client.

It is not so much that documents like this house plan move through distinct "regimes of value" (Appadurai 1986), in and out of commodity phases, but that they can be at once a thing paid for and an object of bureaucratic practice, mediators of practices that are at once bureaucratic work and a paid service. The paradigmatic case of corruption understood as the marketization of government is when a functionary profits purely from his control over a government resource, such as when an officer accepts a bribe in exchange for a signature approving an application, or when a clerk demands a hundred rupees to find a file or to register receipt of a letter. In many other cases, however, the functionary does a substantial amount of work for the client in producing required documents.

One town planner who worked in the government overseeing the projects of housing societies, for example, ran what he called "private work" from his office with the help of another town planner who was not a government servant. The government planner described himself as a "silent partner." According to this planner, housing societies must in any case "go to the market to find a planner, so I tell them, 'I am a planner, and it is better that I do your work.'" With his partner, he drew up plans and advised the societies on regulation issues. He saw the money these societies pay him-which always went to his partner-as fees rather than bribes, just compensation for the integrated services of town planning and help in satisfying regulatory demands. He admitted the conflict of interest, since he was the one who scrutinizes the plans and makes the recommendations for approval, but stressed that he could do nothing illegal since his superiors reviewed all the proposals. His signature of approval was the essential attraction for clients ("If I leave the seat, they will leave me," he admitted to me), but it is not the only service sold. His services included a considerable amount of work beyond the "resource" he can deliver as an officer in the review process. Even what seems to be the clearest expression of the marketization of government, the sale of official approval, almost always involves more than a signature. Officials accepting bribes for a contract or for the approval of a project usually serve the functions of lawyer and lobbyist for those who have paid them, shepherding documents through bureaucratic processes over which they, like their clients, have limited control.

As in the case of the architect, the product or-in the language of management consulting-the "deliverable" of private work is a document. When accompanying this planner on some of his private work, I observed one of his clients when she suddenly realized how a plan of her plot of land was to figure in his service. The planner had met the woman when she had come to the CDA on some unrelated business several years earlier. She had contacted him again regarding the subdivision of her plot in the elite E-7 sector, inherited from her father, on which she planned to build a second house. After the three of us enjoyed a pleasant lunch at her house, the planner unfolded the site plan he had brought, and we went upstairs to a little balcony to inspect the plot lines. She noted immediately that the subdivision map did not show the portion of the existing house that sat on the second lot. She was very concerned about this. She said that there was a legal case proceeding on the property, and that such an irregularity might cause some problem if it fell into the adversary's hands and was subsequently presented in court. The planner acknowledged her concern, but without elaborating he said that since the CDA had done it this way it was best just to leave it the way it was. When she insisted, the planner told her that he had had quite a difficult time getting the CDA draftsmen to make the site map without showing the projecting structure, precisely what she was objecting to. He explained that if the projecting part of the house had been shown, the land covered by it could not be included in the new plot under subdivision rules, so the plot would not reach the minimum area required for subdivision. If the map were "corrected," the subdivision could be revoked. She readily agreed to leave the issue but complained about regulations regarding subdivision, saying, "Why isn't a father's daughter able to do this?" (Bap ki beti aise kyon nahi ker sakti?). The planner smiled and replied in English, "I'm am sorry, but CDA doesn't care who you are; whether you are a brother, a son, a mother or a daughter, the rules are the same."

The way that documents draw together bureaucratic and other kinds of relations can be seen in how people morally evaluate actions like the planner's production of an inaccurate site map. Since the postcolonial Pakistan state is often not regarded as a moral actor, standards of public service do not play as great a role in evaluations of bureaucratic activities as prevailing Pakistani public discourses on corruption might lead us to expect. Most people consider actions corrupt when they transform administrative relations into another sort, as in marketization or "personalization." The transformation itself, however, is often not the crux of moral condemnation. Often more important is how the action is evaluated in light of the moral norms invoked by the transformation. That is, actions within the bureaucracy are translated into analogous contexts of social life outside the bureaucracy and evaluated in light of the moral norms reigning there. If a bribe was taken, did the service provided make it a fair exchange? If merit was disregarded in hiring, was it for money or to help a needy relative? Actions are strongly condemned as corrupt only when they violate both the norms of public service and the morality governing the analogous social domain. Many violations of norms of public service are excused on account of their conformity to other moral norms. Thus, the moral category of corruption, while only applied to activities within the politico-bureaucratic realm, nevertheless converges with conceptions such as greed (lalach), selfishness (matlabi), dishonor (beizzati), dishonesty (beimandari), injustice (beinsafi), oppression (zulum), and impiety (fisq), which are grounded in moralities prevailing in other social domains.

This was the sort of thinking that a major client of this same planner used to evaluate the planner's activities. The client distinguished him from other bureaucrats who demand bribes, telling me, "It is not like he is charging for just a signature." By contrast, the planner's partner, who in his own opinion did most of the planning work, was much more ambivalent in his evaluation of the town planner's practice.

When others evaluated the town planner's activities, they translated his actions into the market realm, and the question of corruption became one of a just price for services. Similarly, the woman for whom the planner had made the inaccurate map, while seeming uncomfortable with the matter, was willing to go along with it since she considered it her right to modify her own father's house as she saw fit.

CDA record-keeping practices can also facilitate the involvement of those outside the bureaucracy, promoting precisely the practices that the records were intended to curtail. Blueprints for private houses provide an enlightening example of this phenomenon. Until the 1960s, most urban houses in the region were built by contractors who worked from rough sketches and the regular input of owners. In contrast, the Islamabad building codes required that plans drawn up by a licensed architect be submitted to the CDA. One goal of this code was to compel compliance with regulations; another objective was to ensure the uniqueness of each house. Ironically, the CDA architecture record storeroom has provided a supply of ready-made plans for reproduction. At the prompting of clients who have admired a particular house, architects would buy these blueprints from CDA staff in control of the record room for Rs. 500-1000, change the names on them, and sell them to clients for Rs. 7000-10,000. In other cases, CDA architects, like the planner discussed earlier, offer the same service. One private architect I spoke with sees himself as a competitor of the CDA architects and draftsmen, who try to "steal" his clients when they file their application forms. There are now at least nine reproductions of one large, white, pillared house placed on a prominent road that was featured in a popular television serial drama as the residence of a very rich family. (Architects told me people would often come in and tell them they want to live in a "White House.") The trade in blueprints is such a common practice that, in a conversation with me, one architect marked it as a point of pride that his firm usually designs houses from scratch. One rough measure of the scale of traffic in house plans is provided by statistics compiled by an officer in the architecture directorate showing that an architectural firm with just one architect produced 396 plans in 1997.

In the cases I've discussed so far, the nonsanctioned production and use of documents draw other sorts of relationships into bureaucratic practices. In other practices, formal bureaucratic procedures themselves blur the boundary between the bureaucracy and the wider society of the city. Such is the case with the allotment of housing to government employees.

In principle, the original plan to make government house type correspond to service rank could not have been simpler. In practice, it has been more difficult to realize. There are no statistics that I know of, but it is common for government employees to live in houses of types far below or above that to which the allocation rules entitle them. For example, one resident of the oldest sector of Islamabad, G-6, whom I'll call Ahmed, has lived in a type A house for nineteen years, though in this time he has moved from grade 4 to grade 15 and is now entitled to a type D house. While many of the original allottees of these houses have retired and are entitled to retain their quarters only for six months after their retirement, few quarters have ever left the family. Statistics on house transfers are not available, but good estimates suggest that in G-6/1, for example, since the departure of Bengali government servants during the war in 1971 that resulted in the creation of Bangladesh, less than 10 percent of the houses have left the families of their original occupants.

Although some residents like Ahmed have been trying for years to obtain the housing to which he is entitled, others have been pressing the government to award them ownership of their current unit. Residents of this area refer to themselves as "owners" (maliks) of their houses rather than "allotees" or "renters" (karaedars). Most occupants have invested substantial resources into the modification of their government dwellings, modifying the front purdah (seclusion) wall and interior layouts and adding typically two more rooms at the back of the large rear courtyard. Ahmed told me he waited five or six years before realizing that he might remain in the house another fifteen years, at which point he began modifying and adding to it. While he vows that he will not try to retain the house for his children after his retirement-"They should go out and get something better"-he recognizes that they may have no alternative. Residents of G-6 have organized and agitated for decades to have what they see as their customary rights to their quarters confirmed through the grant of legal ownership. During the 1980s and 1990s, Pakistan presidents and the heads of both leading political parties have promised to grant ownership rights to occupants of those houses, but the awards have never come. While electoral politics has continued to fail these residents, they have been better served by bureaucratic practice.

Though legal ownership has not been granted, individual and family rights to their quarters, irrespective of government service, are recognized in both the legally sanctioned and informal aspects of the allocation process. The "Out of Turn Allotment" (OOTA), an ad hoc rationalization in both the Weberian and moral senses of the term, has been codified in the allocation rules in order to regularize the ad hoc allocation of quarters to those who are not entitled to them according to service qualifications.

If a son enters government service before his father's retirement, for instance, the allotment of the family quarter may be transferred to the son, as long as the son is entitled to it by service rank, a qualification that is rarely enforced. In practice, if the father occupies a house commensurate with his service rank, the son, rarely having a rank equal to his father, is allocated his father's house, one of a higher classification than that to which he is entitled. Ironically, in cases of a retiring father living below his entitlement who irregularly transfers his house to his son, these two factors combine to realize something approximating the planned relation between the rank of occupant and the house classification.

OOTA forms (fig. 1.3) also figure in the cases in which houses have been transferred outside the family. The way records on government quarters are maintained plays an important role in such transfers. At least through the late 1990s, the CDA office that allocates housing, the Administration Directorate, did not have synoptic documentation of the units under its authority, that is, a single register of all of its housing units. Information on each unit was contained in a file of its own, stacked one upon another on the shelves of a large number of cabinets. As I will discuss in the next chapter, files are network documents that move along narrow paths, greatly restricting the range of people who have access to their contents. The file-based organization of government housing records, by limiting the knowledge of housing units and their occupants, enabled the allocation of housing units to run almost entirely on the local knowledge of those with access to the records, coworkers, and neighbors who come to know of a retirement, transfer, or illegal sublet.

Generally, when a government servant with no children eligible to take over the unit is retiring, he (and occasionally she) is approached by new prospective occupants, before official procedures of reallotment have even begun. We might call such transfers friendly takeovers. In recognition of the occupant's right to the house, the prospective occupants pay him or her (Rs. 2000-10,000 in the late 1990s) to move out and let them take possession (qabza). Only after settling in do the new occupants submit a letter to the government stating that they have taken possession of the quarter and are petitioning to have it officially allotted to them through the OOTA process. Possession is nine-tenths of the allocation, so such petitions are usually accepted.

Hostile takeovers take place when an individual comes to know that the allottee of a quarter either has illegally sublet it or, when the allottee is retiring, is unwilling to accept a financial offer and vacate. In yet another example of the explicit codification of a practice originating outside the organization, the allocation rules stipulate that the "first informer" to bring a violation to the notice of the authorities may claim the quarter through an OOTA application. Retiring officers, whose ineligibility is easily verified, often accept the offer of the "informer" as an alternative to official action.

In contrast to retirement, the case of subletting is difficult to prove. The CDA officer in charge of investigating such charges told me that if his investigators ask about the clothes and other furnishing in the quarter, the allottee will simply claim they are his. If the renters are present, the allottee will say they are his good friends and they are staying free. Inquiries from neighbors are no more help; they usually lie to protect their neighbor, especially since they themselves are often subletting a portion of their quarter. Moreover, in most cases, the officer told me, the allottee learns well in advance that an investigation is imminent and will be expecting the arrival of inspectors. This observation was borne out by the experience of a friend of mine who was occupying a quarter allocated to his uncle. He and his family were summarily hustled out of the quarter when the uncle got wind of someone filing a claim against him. Even when such claims against illegal occupation are successful, though, the claimant will usually end up paying a substantial sum to the occupant to dissuade him from tying up the case in court for years.

Clearly, the file organization of housing documents is an important factor in the divergence of the occupancy of government housing from the planned (and legally stipulated) correspondence between house type and the service rank. The allocation office acts more like a registrar of market transfers than an agent of those transfers in accordance with the allocation rules. The records of CDA government quarters are another example of how a documentary infrastructure facilitates practices it was produced to prevent. The documentary infrastructure has enabled other practices to overwhelm house allocation rules.