The row is the city

TL;DR

Bangkok’s inner city is not a collection of temples with leftover housing. It is a row city. The unit is the ตึกแถว — a live-work bay, four-ish metres wide, stacked shop-under-dwelling, repeated until it becomes a street. In July–August 2001 and April–May 2002, Nawit Onsawangchai and Shuji Funo walked Rattanakosin and counted 1,656 remaining shophouse units, grouped into 44 façade/location families, owned chiefly by the Crown Property Bureau (46.56%), private landlords (23.85%), and temples (16.73%). Nearly half (751 units, 45.28%) still used the ground floor as a shop.

That survey is twenty-five years old. It is still the best number we have. This working paper locates the official ONEP category-E ledger against sourced places: named OSM street axes for the ensembles, four commercial buildings as points, and the Santa Cruz pier pavilion as a related structure (46 of 46 records linked; location is not a cadastral footprint or a condition survey). The public map is bkk.nonarkara.org/rowhouses. Fine Arts gazettes monuments. The Building Control Act of 1979 (B.E. 2522) legislates a later, flatter row — width ≥ 4 m, depth 4–24 m, ground floor ≥ 30 m², rear open space ≥ 3 m, a run of not more than ten units or 40 m. Neither instrument sees the thing Onsawangchai actually counted: a bifunctional street wall whose landlord is often the monarchy’s property office.

This paper takes those shelf numbers, the CPB’s own heritage inventory (15 shophouse sites, 1,960 units), the Bangrak “ordinary urban heritage” argument, the Luenrit community corporation, and the atlas’s 9,275 Old Town footprints, and holds them in one place so they cannot be collapsed into each other. Then it tells you where to fly the map.

1. How this bible was built

The commission was blunt: research on Bangkok shophouses already exists; it sits on shelves; use it. A Perplexity Computer thread (c2bde830-d668-4516-be4c-cfe95d49acc3) was the pointer. That page is Cloudflare-gated from this sandbox — I could not scrape the thread. A later roundup from the same thread claimed 55 graduate theses plus 3 papers. I recovered what the named authors and university repositories would actually give me: 42 theses and the 3 KMUTT papers, filed in catalog.json. I did not invent the missing titles to fake 55. The spine of the argument is still the primary papers any serious cluster on this topic returns, redrawn as original figures, locked against the atlas.

The spine is the Kyoto pair from 2004: Onsawangchai & Funo on formation and typology (AIJ 577), and the follow-up on spatial organisation across three landlords (AIJ 586). Around that: Yongtanit Pimonsathean’s 2012 Siam Society paper on the Crown Property Bureau; the Bangrak ordinary-heritage criteria; Sayanee Virochrut’s Charoenkrung conservation thesis (Rama IV–VII, Saphan Mon to Chalermphan 53); KMUTT’s flexibility / quality-design surveys of the modern ตึกแถว; the Building Control Act B.E. 2522; Luenrit as documented by the Siam Society and ICCROM; tourism-use transformation work from Charoenkrung and Chiang Mai’s Chang Moi. Atlas counts are from public/bkk-heritage-detail.geojson as of 2026-08-10: 9,275 footprints, median height 9 m, 1,086 tagged commercial, 286 retail, 154 terrace.

Figures in this paper are redrawn. They are not scans of the AIJ plates. Where a number is a field count, it is cited to the survey. Where a number is an OSM tag, it is cited to the GeoJSON. Mixing those two is the original sin of “heritage dashboards.”

2. Seven meta-arguments

The literature does not disagree on facts so much as it talks past itself. Seven arguments keep recurring. This bible names them so you can see which one a paper is actually making.

1. The import thesis. The Rattanakosin shophouse is not an indigenous type that evolved with the 1782 capital. It is a mid-nineteenth-century import, catalysed by Singapore after the Bowring Treaty, built as roads were cut. Onsawangchai & Funo are explicit: “It was not an indigenous architecture that evolved along with Ratanakosin.”
2. The typology thesis. Façade similarity produces 44 groups. Open space produces five plans (A–E). Style is a grouping device; the plan is the argument. Most “conservation” still photographs the façade and ignores the court.
3. The ordinary-heritage thesis. Fine Arts registers ancient monuments. The row is “everyday-life heritage” (Yongtanit) / “ordinary urban heritage” (Bangrak). A gazetted wat and an ungazetted shophouse row can share a party wall. Only one of them exists in law.
4. The landlord thesis. Almost half the remaining Rattanakosin units in 2001–02 belonged to the Bureau of the Crown Property. Conservation of the row is a tenure problem before it is a style problem. Temple rows, CPB rows, and private rows transform differently because their landlords do.
5. The live-work thesis. The type’s job is bifunctional: shop on the street, dwelling above. 45% still did that in 2001. Tourism conversion (café, gallery, guesthouse) is a use transformation. Façade grants that freeze the front and empty the shop have already lost.
6. The community thesis. Luenrit (~220 bays, ~100 families) and Khao Hong (nine units on Phra Athit, tenants kept) are the counter-models to Talat Ming Mueang → Old Siam Plaza. The row survives when the tenant corporation is treated as a conservation actor, not as a relocation cost.
7. The shelf thesis — this paper’s own. The 2001 survey, the 2004 AIJ plates, the 2012 CPB inventory, the Charoenkrung theses: they are good, and they are inert. A city that cannot click from a number to a camera has not used its research. The atlas is the operationalization, not a replacement for the survey.

3. What a ตึกแถว is

Thai names two things English collapses. ตึกแถว (tuek thaeo) is the masonry or concrete row — the modern shophouse. ห้องแถว (hong thaeo) is the older timber row. The Building Control language treats both as a unit measured column-centre to column-centre. In the inner city the thing you are looking at is almost always a ตึกแถว: a party-wall bay, 3.5–6 m wide in the wild, 4 m minimum in the 1979 code, 12 m or more deep, two to five storeys, ground floor open to the pavement, stair at the back of the shop, airwell punching light into the middle, dwelling (or stock, or grandparents) above.

KMUTT’s design surveys (2011–12) describe the later city-wide type as a 4 × 4 m modular frame, ratio 1:1 to 1:6 toward the back, L-stair pushed rear so the shop keeps the front, party wall in brick because the code wants fire resistance and brick is cheap. The roof is often flat so the next generation can add a floor. That is the Bangkok shophouse as a growth machine, not as a heritage object. The Rattanakosin survey is looking at the ancestors of that machine — the ones built before the code flattened the bay.

4. Imported, not indigenous

Rattanakosin is founded in 1782 as a water city. From Rama I to Rama III (1782–1851) only about twelve small primary roads are cut, to connect palaces and temples. People live on the river. The Bowring Treaty (1855) opens the port. In 1859 King Rama IV sends Minister Srisuriyawong to Singapore. He comes back with two things: the idea of a paved road, and the idea of the linear shophouse. Outer Charoenkrung is built outside the citadel in 1861; Inner Charoenkrung, Bamrungmuang and Fuangnakorn follow in 1862–63. Fuangnakorn’s rows are built on the King’s command. Bamrungmuang’s first rows are private and “primitive.” In 1870 Rama V goes to Singapore and Java, comes home, straightens Bamrungmuang, and orders the Privy Purse Bureau to build shophouses with a uniform front façade and verandah on both sides.

Three city elements, in Onsawangchai’s analysis, sponsor the type:

1. The paved road. Concurrent with the shophouse. Turns a water people into a land people. Roadside land is the new rent machine.
2. The city wall. Once the British have Burma, the wall is obsolete. Demolished wall-and-moat land becomes shophouse sites; the ring road runs beside the ghost of the wall.
3. Palace and temple precincts. After sakdina is abolished and Rama V cuts 38 roads inside the wall, palace owners slice roadside precincts into rental rows. Temples do the same. The row is how a precinct becomes an income.

The Privy Purse Bureau is the protagonist. It already holds large tracts; it hears about road cuttings first; land prices jump when a road opens; PPB builds or takes the rows to collect rent from Chinese immigrants who can pay. The Bureau of the Crown Property is that office after 1936. When you look at a uniform verandah on Bamrungmuang, you are looking at a royal investment product from the 1870s, not at “vernacular Thailand.”

They were freely constructed without the control of any ordinances or building codes but were shaped by demand of the royalty and Privy Purse Bureau. They could enjoy full use of shophouse space without any requirement for the construction of front arcaded walkway. Onsawangchai & Funo, 2004

5. The 2001–02 survey — still the best number

Fieldwork: July–August 2001 (distribution, use, physical features of every remaining unit in Rattanakosin) and April–May 2002 (measured drawings and interviews on three landlord sites). Result: 1,656 units, classified by façade, form and location into 44 groups.

Read the rent line twice. In 2001–02 the biggest landlord in the old city was collecting a few hundred to a couple of thousand baht a month on heritage fabric. That is not a market. That is a holding pattern. The transformation pressure does not come from the existing rent. It comes from the option to redevelop.

Three measured sites in the spatial paper make the landlord difference concrete:

• Temple — Wat Bowonniwet. 46 shophouses, built somewhere between 1851 and 1868 (Rama IV). No back lane. Pre-code.
• BCP — Srasong-Longtha. Temple-adjacent BCP stock; later additions (Suthat, 1952) already negotiating the code.
• Private — Asdang. 108 shophouses, early 20th century into Rama VII (to 1925). Roadside bays pre-1936; inner-block modern bays invent a private “open space” by eating the back lane.

6. Five plans, not forty-four styles

Forty-four groups is a field method: if two rows look alike and sit together, they are one group. The spatial paper then asks a better question — where is the open air? — and the 44 collapse to five:

1. A · Forecourt. A court in front of the shop. The street is already a room.
2. B · Front verandah. The Rama V uniform type. A shaded strip, not quite Singapore’s five-foot way, not quite nothing.
3. C · Enclosed. No court. Maximum rentable floor. The airwell, if it exists, is the only lung.
4. D · Rear court. Service, cooking, washing, later parking. The 1979 code tries to legislate this as a 3 m void.
5. E · Two courts. Front and rear. The most generous, the rarest under land pressure.

If you only restore stucco, you can turn a D into a C without anyone at Fine Arts noticing. The plan is the heritage. The façade is the postcard.

7. The missing five-foot way

This is the Bangkok exception, and it is the point of the 2004 papers. In Singapore, Penang, Malacca, the colonial shophouse is legally required to give the street a covered arcade — the five-foot way. The Rattanakosin row was built without that ordinance. Royalty and the Privy Purse wanted full use of the ground floor. The shop meets the pavement. Shade is optional, a verandah when Rama V copies what he liked in Singapore, not a right-of-way.

So when a conservation drawing copies a George Town section onto Charoenkrung, it is lying. The Bangkok type is a cousin, not a clone. It is also why the later ตึกแถว of the 1970s–90s can be so brutal: there was never a public arcade to defend. The code that finally arrives legislates width, depth, rear void, and fire — not a civic walkway.

8. The 1979 envelope

Building Control Act B.E. 2522 and the ministerial rules that follow it are how the later city got its endless identical rows. Onsawangchai quotes the operative language: each unit of hong thaeo or tuek thaeo must have a width, measured perpendicularly from the centre of one column to the centre of another, of not less than 4 m; a depth from the front wall on the ground floor of not less than 4 m and not more than 24 m; a ground-floor area of not less than 30 m²; entrances from front and rear. Rear open space not less than 3 m. If the row is not on a public road, a front open space of not less than 6 m. KMUTT adds the row-length cap: not more than 10 units, not more than 40 m in one run.

The meta-argument: the code is a hygiene and fire document pretending to be a type definition. It produces a liveable minimum and a city of clones. The 1861–1935 rows are interesting because they predate it. Conserving them by bringing them “up to code” is how you erase them.

9. Ordinary urban heritage

Yongtanit’s opening is the legal fact everyone dances around: authorized conservation in Thailand still means ancient monuments under the Fine Arts Department. The power to register has not been devolved. ICOMOS Venice (1964) and Washington (1987) already asked for modest works and historic towns. Thai law did not follow. So the row survives, when it survives, by accident, by landlord whim, or by a community that can make enough noise.

Bangrak is the test case for calling the row heritage without waiting for FAD. The district urbanizes from 1864 with Charoen Krung — New Road — south of the royal island, between the river, Yaowarat, and the embassy belt. Shophouses are first clear on the 1911 map as linear rectangles along Charoen Krung and at the Silom junction. Indian and South Asian rows to the north, Chinese rows to the south, Western offices and churches in the mix. The selection criterion in that paper is almost embarrassingly empirical: still there since 1911. Continuity of a mixed commercial-residential bay is the value. Not stucco. Not a royal gazette volume.

Sayanee Virochrut’s Chula thesis takes a slice of the same road — Charoenkrung from Saphan Mon to Saphan Chalermphan 53 — and looks at eight sets, about 100 bays, Rama IV through Rama VII, as a conservation problem. That is the right scale: not “Bangkok shophouses,” not one photogenic corner, but a run of road long enough to see reign-by-reign change and short enough to walk.

Fly the atlas to the road that made the type: Charoen Krung / Bang Rak · Yaowarat · Song Wat · Talad Noi.

10. The landlord problem

CPB, early 2000s preliminary inventory: 24 palace / residence sites, 6 office / facility sites, and fifteen shophouse and commercial sites totalling 1,960 units. Named inner-city rows include Na Phralan, Tha Chang, Tha Thian, Praeng Phuthon, Praeng Nara, Loenrit, Songwat, Phra Athit, Dheva Sathan on Bamrungmuang. The Bureau holds about 37,000 rental contracts on 41,000 rai; only 7% of the land is “profitable”; that 7% is mostly inner-city. In 2011 rental income was reported at 2.7 billion baht, plus dividends from SCB and Siam Cement.

The 1936 Crown Property Act is the hinge. Rama V had already split private royal property from the national treasury. After 1932 the split is regularised into public property, the King’s private property, and crown property. CPB is not a ministry. It does not have to follow bureaucratic conservation procedure. That is why it can demolish Talat Ming Mueang and also, later, invent matching funds for Na Phralan. The same landlord, two decades, two theologies.

11. Luenrit, Khao Hong, Old Siam

Three CPB stories. They are the whole argument in miniature.

Talat Ming Mueang → Old Siam Plaza (1990s). Two late-nineteenth-century shophouse blocks and a 1920s–30s neoclassical market, famous for tailors and an early bus depot. Decline, clearance, shopping mall with a historic name. Yongtanit calls it what it is: gentrification with a historic touch. The row is remembered in the brand.

Khao Hong, Phra Athit (from 2005). Nine units, about seventy years old. First idea: boutique hotel, compensate the tenants, Singapore Chinatown script. After planners push back, CPB keeps the tenants, pays for renovation, asks for design guidelines. Two-year job. Yongtanit calls it a blueprint for community-based shophouse conservation by the Bureau. Matching-fund later (CPB 75% / tenant 25%) is applied at Na Phralan (completed 2011) and promised for Tha Chang, Tha Thian, Ban Mo — no gentrification, contracts extended.

Luenrit (Chinatown, completed 2022). About 220 classic bays, late Rama V / early 20th century, ~100 families, Indian, Chinese, Thai, wholesale trade. CPB had a developer deal. The tenants formed Chumchon Luenrit Company, took the lease, renovated themselves. First time the Bureau abandoned “renovate it ourselves or lease to a developer.” ICCROM lists it as a joint-investment, co-management case, not a consultation theatre.

The 1997 Rattanakosin conservation and development master plan — prepared without serious public consultation — wanted Tha Thian and Tha Phrachan communities moved for open space and tourists. Both sit on CPB land. The plan had to be abandoned. Mahakan Fort, next door in spirit, became the international version of the same fight. The row-city is not a façade inventory. It is a tenancy map.

Fly: Na Phra Lan · Tha Tian · Phra Athit / Khao Hong · Luenrit / Song Wat.

12. Tourism is a use change

Charoenkrung as “creative district,” Chang Moi in Chiang Mai as café-and-guesthouse row, the Singapore Chinatown script that almost ate Khao Hong: the literature now has indices for this (function transformation, tourism intensity, commercial displacement). The finding is stable across cities. A conservation framework that watches façades, skylines, and landmark projects will miss the death of the type. The type dies when the ground floor stops being a shop the neighbourhood uses.

751 of 1,656 Rattanakosin units still had a commercial ground floor in 2001. That is the baseline. We do not have a 2026 resurvey. The atlas cannot invent one from OSM tags. What it can do is refuse to colour a café row as if it were still a hardware street. Honesty here is a research program, not a paint rule: walk the 44 groups again.

13. What the atlas can and cannot see

The Old Town layer is 9,275 footprints with baked heights. Median height 9 m — the atlas default when OSM is silent. For untagged commercial / retail / terrace in Rattanakosin we lift that default to 12 m so the inner city reads as two- to three-storey rows rather than a flat plate. Temples go to 18 m unless a real tag wins. Landmarks (Victory 50 m shaft, Golden Mount chedi ~60 m) are a separate GeoJSON of 73 schematic masses. None of this is a shophouse survey.

OSM in the Old Town bbox, by building / building_type:

That last row is the tell. The atlas colour table includes shophouse → #e5e5e5, and the legend says “commercial / shophouse,” but the GeoJSON almost never carries the tag. We are painting a type the volunteers did not name. The 2001 survey named it 1,656 times. The honest next step is not a prettier extrusion. It is to join the 44 groups to footprints.

14. How to use this bible

The point of taking papers off the shelf is that a reader can do something by this afternoon.

1. Open Rattanakosin at street zoom. Turn building paint to type. The paper-coloured commercial bays and the gold wats are the two cities sharing a party wall.
2. Walk Charoenkrung as a formation diagram, not a café crawl: Bang Rak was the international commercial district the road invented; the 1911 rectangles are still the criterion.
3. Treat CPB named rows as a separate layer in your head until we have one in the file: Na Phralan, Tha Tian, Tha Chang, Praeng Nara, Phra Athit, Loenrit, Songwat.
4. If you are a student: resurvey the 44 groups. Use is the variable that went stale. Plan type (A–E) is the variable conservation still ignores. Publish the join as GeoJSON. The atlas will take it.
5. If you are BMA or FAD: stop writing master plans that relocate riverfront rows for “open space.” The 1997 plan already failed. Luenrit already demonstrated the alternative. The papers are on the shelf. This tab is so you cannot say you could not find them.

Working paper 01 was the system. Working paper 02 is the fabric the system is supposed to be able to see. The row is the city. The monument is the exception.

15. The shelf: recovered theses

Perplexity’s roundup organised 55 theses into five sections: 17 Thai dissertations on ตึกแถว (A), 5 on อาคารพาณิชย์ (B), 27 on historic districts (C), 6 overseas (D), and 3 KMUTT papers (E). Six named PhDs: Peeraya Boonprasong (Chula), Narut Kupthanaroj (Chula), Tanat Bha-aryaphatn (Silpakorn), Pheereeya Boonchaiyapruek (UCL), Tiamsoon Sirisrisak (Tokyo), Monsicha Bejrananda (Texas Tech). Thammasat links were supposed to be Direct PDFs; many Chula records are repository landing pages.

This sandbox could not scrape the thread. The recovered catalog is 42 theses + 3 papers, not 55. Section A is 12 of 17. Section C is 18 of 27. Section D includes the six named doctorates plus Wimonrart Issarathumnoon’s Tokyo dissertation and Napat Settachai’s 2005 Hawaiʻi PhD on Tha Tien; the claimed MIT Bangkok thesis was not found. Bejrananda is southern Thailand, not Bangkok — kept because it is the named dissertation. Incomplete authors and years stay marked not fully verified.

The catalog is a first-class page, not a footnote: The shelf (catalog.json). Direct PDF versus Repository is a field, not a vibe. Use it to open the files. Do not cite Perplexity as the source.

16. References

Primary papers first. The Perplexity thread is a pointer, not a source. Atlas files are data, not scholarship.

• Onsawangchai, Nawit, and Shuji Funo. 2004. “Consideration on Formation and Typology of Shophouse in Ratanakosin Area, Bangkok, Thailand.” Journal of Architecture and Planning (AIJ) 69 (577): 9–15. doi:10.3130/aija.69.9_2
• Onsawangchai, Nawit, and Shuji Funo. 2004. “Spatial Organization and Transformation Process of Shophouse in Ratanakosin Area, Bangkok: A comparative study of shophouses from three different landowners.” Journal of Architecture and Planning (AIJ) 69 (586). doi:10.3130/aija.69.1_12
• Yongtanit Pimonsathean. 2012. “The Crown Property Bureau and Heritage Conservation.” Journal of the Siam Society 100: 107–122. PDF
• “Selection Criteria of Ordinary Urban Heritages Through the Case of Bangrak, a Multi-Cultural & Old Commercial District of Bangkok.” Nakhara. doi:10.54028/nj202221209
• Virochrut, Sayanee. The conservation of shophouses on Charoenkrung Road: between the reign of King Rama IV and King Rama VII (การอนุรักษ์สถาปัตยกรรมประเภทตึกแถวริมถนนเจริญกรุง…). Master’s thesis, Chulalongkorn University. Eight sets / ~100 bays, Saphan Mon–Chalermphan 53.
• “Bangkok Shophouse: An Approach for Quality Design Solutions.” KMUTT School of Architecture and Design, 2011. 4×4 m module, 1:1–1:6 depth ratio.
• “Flexibility Survey of Bangkok Shophouses.” KMUTT, 2012. Width 3.5–6 m in the wild; code minima 4 m / 4–24 m / 30 m² / 3 m rear / 10 units / 40 m row.
• Building Control Act B.E. 2522 (1979), and ministerial regulations on tuek thaeo / hong thaeo dimensions.
• Crown Property Act B.E. 2479 (1936); Crown Property Bureau Act B.E. 2491 (1948, as amended). CPB 2004. Tassaneeyakan: Heritage Buildings of the Crown Property Bureau.
• Siam Society Knowledge Hub / ICCROM CH4IG. Luenrit Community Conservation and Rehabilitation Project. ~220 bays, Chumchon Luenrit Co., Ltd., restoration completed 2022.
• “Evaluating use transformation of shophouses: a case study of tourism-oriented change in Chang Moi, Chiang Mai, Thailand.” International Journal of Urban Sustainable Development, 2026. doi:10.1080/19463138.2026.2677384
• “The Forgotten Heritage of the Rattanakosin Area.” Journal of Urban Culture Research. Rattanakosin Island 4.1 km²; CPB assets at risk. doi:10.58837/chula.jucr.9.1.4
• “Changes in the Bangrak Area after the Construction of Charoen Krung Road during the Reigns of Kings Rama IV–VI (1851–1925).” Journal of the Faculty of Arts, Silpakorn University.
• UNESCO. 2011. Recommendation on the Historic Urban Landscape.
• ICOMOS. 1964. Venice Charter; 1987. Washington Charter (historic towns and urban areas).
• BKKx 3D Atlas. 2026. Recovered thesis catalog: The shelf (catalog.json — 42 theses + 3 KMUTT papers; Perplexity claimed 55). Direct PDF vs repository is a field. Incomplete metadata stays marked.
• BKKx 3D Atlas. 2026. public/bkk-heritage-detail.geojson (9,275 footprints); public/bkk-landmarks.geojson (73 schematic masses — not the FAD register); working paper 01.
• Perplexity Computer thread (commission pointer, not a citable authority): c2bde830-d668-4516-be4c-cfe95d49acc3.