From Garrisoned District to Chinese Town: Land and Boundaries at the Kowloon Walled City, 1898–1912

The Kowloon Walled City in Hong Kong became a named entity around 1810 and was demolished in 1994, but its architecture had long been unclassified. Not until the years just prior to its demolition did this dense slum of informal multi-story buildings receive sustained attention from architects and architectural historians. However, the architectural nature of the six-acre area predated its late-20th-century state. After its founding as a Qing military outpost, it underwent various structural additions and renovations, including an imperial Chinese administrative complex known as a yamen ( 衙門 ) and an outer wall, after which the Walled City was named. Against the grain of scholarship that has focused on the Walled City’s postwar, informal architecture, this article considers the site’s early years, arguing that the Walled City’s yamen and outer wall played a crucial role in the region’s land management practices. These two architectural structures make legible the Walled City’s evolution from a Qing administrative zone to a crowded slum. The Convention of 1898 ushered in a British-led land surveying effort throughout the New Territories region of Hong Kong, followed by the creation of an intricate bureaucracy for managing land lots. This clash of empires saw the use of two forms of land knowledge, Qing land deeds and British cadastral land surveys. In between these systems existed the Walled City, its inhabitation falling outside the British conception of land division but its historical contours very much shaped by the architectural boundaries that gave it its name.


The Architectural History of the Kowloon Walled City
Situated northeast of the Kowloon Peninsula and five miles north of Hong Kong Island, the Kowloon Walled City has been a cohesive, named entity since at least the early 19th century. 1 A fort was built on the site in 1810, and upon the suggestion of Qiying, the governor-general of Guangdong and Guangxi, an outer wall mounted with cannons was added in 1847, turning the area into an imperial Chinese zhaicheng (寨城 walled city) (Krone 1859: 105).
The Walled City was most densely inhabited from the 1960s to its demolition in 1994. 2 During this time, it was a case study in vernacular building and an enclave of slum tenancy. Crime and hazardous construction practices ran rampant, and its six-acre space became crowded with improvised, multi-story dwellings joined by dimly lit alleyways.
Architects and architectural historians began paying attention to the Walled City in the years immediately prior to and following its 1994 demolition. In particular, monographs produced between 1993 and 1999 documented comprehensive architectural studies of the Walled City (Ho 1993;Terasawa and Kani 1997;Lambot, Girard, and Wilkinson 1999). Meanwhile, the work of photographers such as Miyamoto Ryuji have cast the Walled City as a manifestation of -and sometime antithesis to -the planned, postwar East Asian urbanism of cities like Tokyo (Figure 1) (Cushman 2018). However, the architectural history of the area predated its late-20th-century state.
Since its days as a Qing Dynasty military outpost, the Walled City has sustained various additions and renovations, especially to the outer wall that gave shape to its boundaries.
In other words, the Walled City's earlier history was punctuated by land-use planning, in sharp contrast to its final iteration as an under-planned slum. What specters of these developments remained detectable within its later history? This article is bookended by two events: first, the largest and last extension of Hong Kong's colonial boundaries, as stipulated in the Convention Between the United Kingdom and China, otherwise known as the Convention of 1898, and second, the dissolution of Qing China in 1912. I identify two systems of land management as major turning points that shaped the area's turn-of-the-century architectural history: Qing land deeds, a Ming Dynasty legacy still in place under the Qing, and the colonial British cadastral system, established after comprehensive land surveys. With its focus on the Walled City's late-20th-century history, existing architectural scholarship rarely considers the circumvention of one system by the other. That interplay is precisely my article's focus. Rather than seeing the clash between these systems as an obstacle to understanding, I use it to illuminate how the Walled City became systematically obscured between 1898 and 1912. I apply Bruno Latour's concept of maps as 'immutable mobiles' (1986) to explain the supplantation of Qing land deeds by map-based British cadastral surveys, treating this 14-year period as a time of rapid change in both epistemological and spatial realms. Furthermore, by focusing on two architectural anchors, the Walled City's administrative yamen (衙門) office and its outer wall, I call into question the power and, indeed, the true immutability of maps. This history of overlapping systems most significantly resulted in the outer wall eclipsing the yamen in importance, creating the conditions that would eventually turn the area into an aestheticized slum.

Land Management via the Walled City Yamen
The Walled City stood in what became known after British incursions as the New Territories. Humans had been settling in this vicinity for centuries, including pearl hunters since the Han Dynasty and villagers and long-term residents in the Ming and Qing, and by the mid-18th century, a Qing-imposed system of land deeds administered territorial possessions throughout the area. 3 The Qing history of the Walled City must Containing not only a wall to give it shape and cannons for defense, but also offices and military barracks, the Walled City functioned as a garrisoned district during the Qing, responsible for overseeing villages in its vicinity as well as for keeping watch over the British presence in Hong Kong, which, by 1841, had begun its colonial possession of Hong Kong Island (Krone 1859: 100-101). Before 1898, the Walled City's function was primarily administrative rather than residential. In the 1890s, it housed only about 200 civilians but more than 300 Qing officials, military personnel, and their dependents, including the xunjian (巡檢 deputy magistrate) and the commodore of the Dapeng Brigade, the area's head bureaucrat and military official respectively (Lockhart 1898: 75;Sinn 1987: 32). The former's office was a yamen, which remains standing even today, and which served as a center for local governmental affairs within Xin'an County, a territorial entity in place since the Ming. 4 The Walled City yamen was similar to those in other walled cities throughout Qing China. These administrative complexes operated throughout the Qing period, allowing the dynasty's outsider Manchu rulers to govern rural Chinese society using elite Chinese magistrates as their intermediaries (Watt 1977: 357). In other words, the function-defining feature of imperial walled cities was their yamens, rather than the walls that gave walled cities their names and sense of boundedness.
Graphic representations of Guangdong Province from the Qing reinforce this argument. In a watercolor map of coastal Guangdong, produced around the 19th year of Emperor Daoguang's reign (circa 1839), islands and peninsulas are dotted with greenroofed buildings, indicating yamens (Figure 3). In addition to these yamen icons, larger walled cities are depicted with thick, elevated outer walls to indicate the presence of substantial garrisoned enclosures. These wall icons occur with less frequency, with only ten drawn throughout the entire map. Fourth from left among these ten wall icons is Xin'an County's central walled city, featuring four gates and the label 'Xin'an County Xin'an Barracks (新安縣 新安營)'. The Kowloon Walled City, in contrast, is farther to the right, its modest yamen icon framed by nothing more than islands, waterways and its corresponding label, 'Kowloon Bay Garrison (九龍湾汛)'.
Serving as further evidence that the Walled City's wall was of secondary importance compared to its yamen, few of its specific measurements were recorded until after the Convention of 1898. A reconnaissance mission by JHS Lockhart in 1898 showed that it spanned 6.5 acres and stretched 700 feet by 400 feet in a rectangular formation, with another triangular portion extending up a hill (Lockhart 1898: 74-75;Ho 2021).
Surveys done in the next decade reveal a similar shape (Figure 4) ('1900-1902 Survey',   London, 1924London, . 1902. 5 According to Lockhart, the wall's height measured 13 feet, although later accounts place this figure as high as 23 feet. Encircling the entire perimeter of the Walled City in granite ashlar, it also included six watchtowers and four gates, one of which was a main gate on the southeast corner ( Figure 5) (Sinn 1987: 31). Chief among the administrative concerns of the Walled City yamen was land, with jurisdiction over Xin'an County consisting of three processes: grants, taxes, and reclamations. Under the Qing Imperial Land Law, the emperor held claims over all arable land, with imperial grants allowing for 'perpetual imperial tenancies' (Hase 2013: 25). Proof of these land grants was established through tax collection, in which the yamen played a direct role. Research into the social history of taxation in mid-19thcentury Guangdong Province shows that taxes were paid either to traveling yamen representatives -tucha (圖差), as they were sometimes called, who were hired workers that made house calls to the vicinity -or at the local yamens directly (Matsubara 2011: 179). As for reclamations, they addressed instances in which inhabitants wished to claim new land from the emperor's holdings. Reclamation procedures depended upon the sizes of land plots in question, but were also officiated through tax payments to yamens (Hase 2013: 28). In principle, the system underpinning grants, taxes, and reclamations in Qing-era Guangdong was a Ming-inherited land scheme known as yulin tuce (魚鱗圖 冊), the 'fish-scale register' (Matsubara 2011: 173). Tracking the shapes, place names, tenants, borders, measurements, and areas of taxed land, these registers were arranged in plot-by-plot records and stored at their respective yamens (Figure 6). By the waning years of the Qing, land records held at yamens or by officials did not reflect all-encompassing landholding arrangements. In the absence of registers supplying comprehensive information, deeds exchanged during tax transactions conducted at yamens or by yamen-appointed representatives did so, describing locations and borders of properties (Matsubara 2011: 173, 167). Thus, in one way or another, even when records existed in decentralized deed forms, land transactions always connected back to yamens. For the Walled City yamen, as the political identity of its surrounding area shifted from Xin'an County to the British colony of Hong Kong, the Qing land deed system would also be renegotiated in new terms.

The Convention of 1898, Land, and Architecture
Deed-based, decentralized transactions simultaneously affirmed the administrative importance of the Walled City yamen, minimized the need for comprehensive visual records of land plots, and undermined the importance of pictorial and cartographic accuracy in and of itself. However, as Qing-British conflicts continually redefined Hong Kong's borders in the decades leading up to 1898, maps played an increasingly prominent role in political negotiations and their consequent changes to land. Maps thus highlight a contrast between the yamen's land practices and those later adopted in colonial Hong Kong. As argued by Latour, printed maps are 'immutable mobiles'.
Printing presses achieve immutability because they create identical copies of visual matter -the 'perspective, projection, map, log book' -and ensure their unchanging character. Meanwhile, mass production provides mobility, since it makes replicable objects and guarantees their portability. Latour maintains that maps completely modify 'links between different places in time and space', since these image-based representations 'circulate everywhere and in all directions in Europe ' (1986: 7-8, 11).  (Figure 7). Officials and other individuals residing within the New Territories were allowed to remain, with no risk of 'expropriation or expulsion' (1898: 1). Furthermore, continuity in the activities in and out of the Walled City were also accounted for, with 'the existing landing-place near Kowloon City [being] reserved for the convenience of Chinese men-of-war, merchant and passenger vessels, which may come and go and lie at their pleasure; and for the convenience of movement of the officials and people within the city' (1898: 1).
However, these promises of continuity within the treaty contradict its characterizations of land and architecture, revealing fissures between colonial Hong Kong and that which lies beyond it, between the New Territories and the Walled City.
The treaty references structures and infrastructure sporadically in its three pages of English and Chinese text, but only when discussing land outside the Walled City.  Perhaps no place in the history of ages can boast of such a rapid rise as the town of Hong-Kong. In August, 1841, not one single house was yet built, not a portion of the brush-wood had been cleared away from this desolate spot. By June, 1842, the town was considerably more than two miles long, containing store-houses and shops. … The town already boasted of 15,000 inhabitants, and it was almost impossible to prevent the people from the opposite coast from flocking to us. (Cunynghame 1844: 83) Cunynghame's account sets up a contrast between Victoria's rapid urbanization and its nebulous 'opposite coast', defined not by its buildings but by its émigrés (1844: 83). Similarly, the Convention of 1898 was pointedly vague about the makeup of New Territories land, including the architecture that sat upon it. It mentioned the lack of 'exact boundaries' for the whole New Territories area and for the smaller divisions within it, declaring that further details would be settled 'when proper surveys have been made by officials appointed by the two Governments' (Convention 1898; Ho 2021).

Cadastral Surveys and Land Management in Colonial Hong Kong
The same year that the Convention of 1898 was signed in Beijing, a major British cadastral survey took place in Egypt, eventually producing twenty thousand maps that depicted the country's farmable land foot-by-foot, plot-by-plot, and region-byregion, at a 1: 2,500 scale (Mitchell 2002). Cadastral surveying commenced in Hong To produce these New Territories surveys, colonial authorities recruited New Territories residents to serve as 'demarcators' and to perform roles equal parts manual and translational (Department of Land Surveying and Geo-Informatics). The primary sources of information for these surveyors and their demarcators were the local inhabitants themselves. As a 1904 survey report recounted, the demarcator teams 'marked the various holdings as pointed out by the tenant' (Newland 1904;Leung, Sham, and Tang 2008: 31). Then they would lay bamboo along the edges of individual land lots, assign them lot numbers, and gather details about size and occupancy.
Surveyors would then compile this information into fiscal survey maps. This workflow included direct contact with occupants and their land but entirely bypassed the Qing land deeds system. In fact, the kind of information contained on land deeds and yulin tuce, the documents that were in theory archived at, and in any case in practice dependent upon, the Walled City's yamen, were similar in nature to the content of fiscal surveys conducted in the British-Indian cadastral efforts. Thus, the cadastral system circumvented older structures of organizing land -an epistemological bypass -as well as older sites on which such structures were centered -a spatial bypass.
Maps cemented the cadastral system's epistemological bypass, introducing imagebased records as a replacement for the Qing land deeds system. The maps made during these post-1898 surveying efforts were not only mobile, in Latourian terms, but also aided in the production of additional, equally portable information about New Territories land. On non-fieldwork days, surveyors would compute lot sizes based on newly created maps known as Demarcation District sheets (Figure 9). These calculations would then be transferred out of map form and into tables known as Field Area Statements (Figure 10). After this map-to-    (Liu 1994: 7). The move was swiftly followed by an Order-in-Council that aimed to 'legalize' the Qing removal and announce 'British jurisdiction over the Walled City' (Sinn 1987: 37). It acted as a paper edict meant to override the earlier paper edict of the Convention of 1898, although the Qing government never formally recognized this later Order-in-Council.
Following these events, the Walled City became what Sinn calls a 'near vacuum of administrative function and authority ' (1987: 38). According to the Convention of 1898, the last jointly ratified treaty, it was still under the purview of Qing China, but as per the unilateral Order-in-Council issued a year later, it was under a 99-year lease by the British Crown, like the rest of the New Territories. By the turn of the century, its population had been depleted and its buildings emptied, a stark reality that corroborated its ambiguous status. Contradictions that existed in the legal jurisdiction of the Walled City also played out in its land management. Under the older Qing land deeds system, the During surveying, land tenants were given siuchi (小紙), small papers bearing lot and serial numbers (Hayes 2012: 297). 7 These documents were but another measure enacted to overhaul and replace Qing land deeds, logging additional information regarding the sizes and locations of land plots. In some cases, tenants made appearances at the Land Court with their Qing land deeds and siuchi to dispute claims; in others, bureaucrats known as Assistant Land Officers made house calls, just like in the old yamen system.
As Hayes (2012: 297) describes, siuchi documents were codified as part of Hong Kong's broader cadastral records. One New Territories resident's siuchi helps illustrate the process: Mr. Chan's ten 'Chi Tsai' [a colloquial term for siuchi] related to the land claimed by his father Chan Sai-wing, and all bore his name. No one had contested his claim, and his ownership of these lots was later confirmed to the schedule of Crown lessees to the Block Crown Lease for the relevant survey district (No. 436). (Hayes 2012: 297) After most land issues in the New Territories were resolved in 1905, the Land Court's physical site was no longer necessary, and it closed, just like the yamen before it. Thus concluded the new system's bypass of the old, epistemologically as well as spatially.

From the Yamen to the Wall: Defining the Interior via the Exterior Post-1898
While the new British cadastral system governed leases for the rest of the Hong Kong colony, the tenancy in the Walled City was handled with more ambiguity, owing to the contested jurisdictional authority of the area. With most of the former Walled City organizations and inhabitants forced out, its structures were leased in five-year cycles, made intentionally short as a contingency measure. After the Land Court moved out, the building of the former Lung-chin Communal School was used as a free secondary school as well as a part-time medicine dispensary (Sinn 1987: 38). Similar informal adaptive reuse schemes popped up throughout the area, with the buildings all being put to public-facing purposes: the former San-sheng Temple became an Anglican chapel; the office of the Commodore of the Dapeng Brigade was used as a plague hospital; and Qing administrative buildings, including the yamen, the old bureaucratic heart of the Walled City, housed old people's shelters, almshouses, and orphanages (Sinn 1987: 38-39). Enduring a better fate of preservation than most other structures around it, the yamen, to this day, bears a plaque with the inscription 'ALMSHOUSE', acknowledging both its first and second lives as public institutions (Figure 11) (Hong Kong Leisure and Cultural Services Department).
Throughout this post-1898 architectural repurposing, the Walled City's outer wall remained standing. As mentioned earlier, land management under the British cadastral system excluded the Walled City. However, during the cadastral surveying effort at the  ', 1901). Its outer wall was treated as its 'official boundary', giving identity to the spaces that lay within, and this rectangular-triangular enclosure also rendered its bounded area a 'de facto cadastral unit' (Lai 2016: 96). Furthermore, Lai's work with map comparisons and GIS has shown that after the Hong Kong government's 1899 Order-in-Council, it treated the Walled City like a 'de jure Hong Kong territory' despite the clear boundaries of the outer wall weakening this position (2016: 96). Thus, the clash of land systems can be distilled into a tension between the de jure, the de facto, and the wholly illegitimate. Although the Hong Kong government regarded the Walled City as a de jure part of its colonial territory after 1898, the area's architectural use reflected a more precarious, illegitimate position. Simultaneously, the Walled City was not assigned official lot numbers during cadastral surveying, but the presence of lots everywhere else in the colony reduced the site to a de facto cadastral unit. Amid these contradictions, the outer wall preserved the unity of the Walled City, a unity present since before the New Territories was under British rule.
Within the walls, the area underwent significant changes in the early 20th century.
Qing China ceased to exist after 1912, so its regime no longer laid claim to the Walled City's land. Civil strife during the fall of the Qing empire and the advent of the Republic of China brought waves of refugees to the Hong Kong region, and by the 1930s, settlement in the Walled City numbered over 2,000 residents and 100 households (Lu 1988: 90). At the same time as its population swelled, threats of demolition grew. In 1933, the Hong Kong government, in agreement with Republican China, proposed a plan to clear all residents and establishments from the Walled City and turn its premises into a city garden (Lai 2016: 96). Space-clearing continued into the next decade.
When the Japanese occupation of Hong Kong began in 1941, bricks and the outer wall were repurposed as construction material for the nearby Kai Tak airport (Liu 1994: 15). In the mid-20th century, through structural demolition, residential relocation, and subsequent repopulation, the contestation over the area effectively ended, with scholarship on this era of the Walled City's history emphasizing its 'interdependence' with Hong Kong (Harter 2000: 96).
The Walled City's history of contested jurisdictions -of the Qing land deeds system being replaced by the British cadastral system in the New Territories, as well as of the ambiguity cast upon the former Qing garrison and administrative heart during the transition -centered on the organization and tenancy of land. In turn, the structures that sat upon this land made legible, if not directly shaped, both systems. Under Qing land deeds, the Walled City's yamen regulated taxation operations, housing the offices of and the local government that oversaw them. During the New Territories' incorporation into the British crown colony of Hong Kong, another architectural feature defined the area and gave it order in accordance with the cadaster-divided land around it: the outer wall. The early architecture of the Walled City thus outlived its original uses.
Its ghosts lingered throughout Hong Kong's inhabited areas, throughout the varying densities of the Walled City, throughout, even, its century-and-a-half of residential and institutional life.