Skip to main content

Where the steps met water

In a folder marked hs: 78.693, inside the Résidence Supérieure au Tonkin records at the National Archives Center I in Hanoi, there is a photograph a French explorer named Firmin-André Salles took in 1896. It shows a gate the French sometimes mislabeled Grand Bouddha — the temple's own name is Quán Thánh, and the deity behind its doors is Trấn Vũ, not Buddha at all — and it shows, at the base of the gate's stone steps, water. Not approaching. Not visible somewhere past a wall. Touching. The lake's edge sits directly against the lowest stone, the way a threshold sits against a floor.

I was standing at the intersection of Thanh Niên and Thụy Khuê a hundred and thirty years later, four lanes of traffic between me and a lake I could hear but not see, when I understood that the water in that photograph had not receded on its own. It had been asked, for decades, one project at a time, to leave. That asking is most of what this essay means by Quan Thanh Temple's history with West Lake — not a monument and a lake sitting beside each other, but a long negotiation over how close the two were permitted to remain, and who got to decide.

The Distance Now

Four streets meet where the water used to be: Thanh Niên, Quán Thánh, Hùng Vương, Thụy Khuê. A signal box, a planted median, traffic islands — the intersection has the ordinary furniture of any Hanoi crossing, which is exactly what makes it hard to read as a former shoreline. The lake sits tens of meters back now, held there by a concrete embankment that runs, unbroken, for seventeen kilometers around the whole of West Lake — a brick-paved walking path on top of it, an iron railing along the water side, benches spaced out for people who come to watch the sunset over water that no longer touches anything old.

The temple's triple gate still stands in its original form, and its steps still descend from it in the same direction they always did. They used to end in water. Now they end in a curb, then a crosswalk, then the northbound lane toward Thụy Khuê. Nobody redesigned the steps to face somewhere else. They simply kept doing what they had always done, while the ground around them changed what that meant.

The Year of Letters

Between 1902 and 1925, a photographer named Pierre Dieulefils sold postcards of this same shoreline to a steady market of colonial travelers. By the time a flood-dyke proposal reached Hanoi's city council in 1927, that market had already gone quiet — the postcards stop two years before anyone thought there was something here worth fighting over. Whatever kept cameras pointed at this water for two decades had nothing to do with whether the water was in danger.

The 1927 plan, drawn up by the municipal urban works office, called for raising Cổ Ngư — the causeway road between West Lake and Trúc Bạch, running directly past the temple's outer gate — into an earthen dyke two and a half meters high, meant to hold back the Red River if the embankment further north ever failed. It meant dumping tens of thousands of cubic meters of earth and stone along the causeway, felling every old tree that shaded it, and closing the road for years while the work went on. Built as designed, it would have blocked the sightline from the temple gate to the lake entirely and buried the site behind a dry defensive mound.

It did not get built, and the reason is on record in more detail than almost anything else about this shoreline. On December 17, 1929, Nores, president of the Hanoi Geographical Society, wrote to the Résident Supérieur asking that the project be stopped. The next day, L. Finot, acting director of the École française d'Extrême-Orient, filed a formal protest arguing the dyke would destroy the setting of monuments already under official protection. Among what was cited alongside those letters was Salles's photograph of 1896 — proof, in an administration that trusted documents over description, of exactly how close the water had always come. On December 21, Governor-General Pierre Pasquier ordered the project suspended. It took another five weeks to settle for good: a nine-page petition from council member Feysal and his colleague Lê Nguyễn, arguing the dyke also contradicted architect Ernest Hébrard's own plan for the city, reached the Governor-General on January 15, 1930, before Résident Supérieur René Robin cancelled the project outright. On May 26, Mayor Tholance announced that floodwater would be routed instead through the Tứ Châu depression to the north.

This is remembered, correctly, as a victory. The causeway kept its trees. The gate kept its water. But a victory built out of one named threat and a stack of formal letters is a specific kind of protection — it works when there is a single project to stop, a file number to cite, a date by which a governor-general has to decide. No one in 1930 was defending the shoreline against a four-lane intersection that would take the better part of a century, and half a dozen separate road projects, to finish arriving. There was no dyke to write a petition against. There was only pavement, added a little at a time, each stage small enough to look like maintenance rather than loss.

The Slow Version

What the fast version would have destroyed all at once, the slow version took apart piece by piece, and it is worth being precise about what that piece was. It was not only water meeting stone. Around the lake, the hamlets collectively called Kẻ Bưởi worked the shoreline the way they had for generations: paper at Yên Thái, silk at Trích Sài and Bái Ân, bronze at Ngũ Xá, their workshops backing onto reed marsh and mudflat where the lake's water exchanged freely with the ground it touched. Further along, Nghi Tàm and Quảng Bá grew flowers in soil that flooded and drained on the lake's own schedule. None of it depended on the temple gate specifically. All of it depended on the same thing the gate depended on — an open, working edge between water and land, where the two were still allowed to negotiate with each other.

That edge is gone, replaced by concrete on a scale the 1927 proposal never approached: not two and a half meters of earth in one place, but seventeen kilometers of engineered wall around the entire lake, with sluice gates and treatment systems doing on a schedule what tide and marsh used to do on their own. The craft hamlets are dense urban wards now, threaded with hotel signage and serviced apartments. And here is the part the 1930 victory could not have anticipated, because it is the part the victory itself produced: a threat that had won once by being nameable had no defense left the moment the next threat declined to be named. The dyke disappeared from the record as a defeated proposal, filed and closed. In its place, an intersection arrived that nobody ever had to defeat, because nobody ever had to vote on it as one thing.


I don't know who decided, in whatever year it was, that the steps at the temple's gate no longer needed to end in water. There is no file number for what happened instead — no hs: 78.693 for the slow version, because it was never one event a clerk could stamp and store. I have looked for a date the way I found December 17 and May 26, and there isn't one — which is its own kind of answer. Nobody proposed removing the lake from the temple's front door. It happened as a side effect of several other decisions, none of which mentioned the temple at all. This morning, someone climbed those same steps with incense in her hand, and at the bottom of them, without pausing, she stepped onto a curb.

August 2026


 Bi is a Hanoi-based writer documenting Vietnam's urban textures and cultural margins. About the author.

Comments