Built on the Last One: Why No City Is Ever Really Rebuilt

There is a doorway in the middle of Rome that opens onto nothing.

It belongs to the church of San Lorenzo in Miranda, which was built inside the shell of a Roman temple raised in 141 AD. The door is a perfectly ordinary church door, and it sits about seven metres up a blank wall with no steps beneath it and no floor behind it. Nobody designed it that way. It was at street level when it was installed, because the street was seven metres higher then. Archaeologists later dug the Forum back down to the Roman floor, and the door was left hanging in the air with nowhere to go.

Cities do that. They rise. Not as a figure of speech but as a physical fact, because the cheapest place to put the remains of the last building is directly under the next one. Rome's ancient ground level is metres below its modern one. So is London's. So, in the parts that were filled rather than reclaimed, is Mumbai's.

That is a curiosity. What follows from it is not.

A city replaces its buildings constantly. Very little standing in Mumbai or London today is a hundred and fifty years old, and almost nothing is used for what it was built for. What a city almost never replaces is the thing underneath the buildings: the street lines, the plot boundaries, the drains, the sewers, the made ground itself. Those get inherited. They get patched, extended, tunnelled through and worked around, but they are hardly ever taken out and done again.

So every large city is running a modern load on an old substructure. And when a city fails, which it does most visibly when it rains, the failure is almost always in the oldest layer.

The city is made out of the city before it

Mumbai was seven islands. Mahim, Worli, Parel, Mazagaon, Bombay, Colaba and Little Colaba, separated by creeks and tidal marsh that filled and emptied twice a day. Between them lay the low ground, and the low ground was the problem, because at high tide the sea came into it.

The fix was to fill it in. In 1769 the hill at Dongri was demolished and the material carted off to raise the swampy land nearby. In 1782 the Governor, William Hornby, began the causeway now called the Hornby Vellard, finished in 1784, which blocked the Worli creek and stopped the tide flooding the middle of the settlement. More reclamations followed, and by 1845 the seven islands had become one.

Read that sequence again, because it is doing something specific. A hill was knocked down to raise a marsh. The tide was walled out of a floodplain. The mangroves and wetlands that had spent the monsoon absorbing water and letting it back out slowly were replaced by ground that does neither. Every one of those decisions was sensible at the time and each one was permanent, in the sense that no subsequent generation was ever going to un-fill the land or put the creek back.

What Mumbai sits on today is therefore not ground in the ordinary sense. It is a nineteenth-century engineering decision with a twenty-first-century city on top of it. The water that used to have somewhere to go still arrives on schedule every June. It simply has nowhere to go now, and the only route out is whatever drainage the city has built.

The report that sat on a shelf for twelve years

Mumbai's storm water drains were sized to carry 25 millimetres of rain an hour, and only at low tide. At high tide the outfalls are under water and the system backs up, which is a design constraint the drains have had since the British laid them.

The city knew this. In 1993, after earlier flooding, a study called BRIMSTOWAD reported on the drainage network and recommended raising its capacity to 50 millimetres an hour. That is the whole story of the recommendation: the engineers looked at the system, did the arithmetic, said it needed to be twice as big, and wrote it down.

Then nothing happened for twelve years, because the work was expensive and because doing it means opening up the roads of a working city for years on end.

On 26 July 2005, Mumbai received 944 millimetres of rain in twenty-four hours. Six hundred and forty-four of those millimetres fell between eight in the morning and eight at night. Against a system built for 25 an hour, that is not a near miss. Hundreds of people died and the city stopped.

After that, the 1993 report came off the shelf and money was found. Work has been going on ever since, in pieces, between monsoons, and it is still not finished.

The useful thing here is not that a government was slow, which is the way this story is usually told and is the least interesting thing about it. The useful thing is the twelve-year gap between knowing and doing, and what closed it. What closed it was not the report. The report had said everything necessary in 1993. What closed it was a catastrophe large enough to make the disruption of fixing the drains cheaper than the disruption of not fixing them.

That is the pattern, and it is not confined to Mumbai.

What actually survives a fire

In September 1666 the City of London burned. Four days of fire destroyed something like thirteen thousand houses and left the medieval city as an open field of ash. If a city is ever going to be rebuilt properly, that is the moment.

Christopher Wren had a plan in front of the King within about a week: broad straight avenues, open quays along the river, the tangle of medieval lanes replaced with something a cart could turn around in. Other men submitted plans of their own. Every one of them was set aside.

The reason is the point of this article. The fire had destroyed the buildings. It had not destroyed the deeds. Property owners came back to their plots, and their plots were exactly where they had always been, defined by boundaries recorded on paper that had not been in the fire. People needed to trade and to live, and every week spent arguing about a master plan was a week of no income. So they rebuilt, quickly, on their own land, along the old lines.

London did get rules out of it. Streets were widened to standard widths, brick was required instead of timber. But the shape of the City of London today, the reason its lanes run where they do, is the medieval plan, preserved through a fire that removed every physical trace of it.

This is worth sitting with. The most durable thing in a city is not its stone. Stone burns, or gets demolished, or falls down. The most durable thing is the ownership map, because that is an idea held in a register, and fire does not reach it.

The price of doing it properly

It can be done. Ninety years after London, on 1 November 1755, an earthquake and the tsunami and fires that followed destroyed central Lisbon.

The Marquis of Pombal had his chief engineer, Manuel da Maia, draw up options. They ranged from rebuilding what was there to demolishing what was left of the Baixa and laying out something entirely new. The king chose the most radical one. Central Lisbon was cleared and rebuilt between roughly 1756 and 1775 on a grid, with wide streets, a sewer system, and buildings framed around an internal timber cage, the gaiola pombalina, designed to shake without collapsing. Thousands of piles were driven to stabilise the ground. It is arguably the first deliberately earthquake-resistant urban district in Europe, and much of it is still standing and still in use.

The academic literature on that reconstruction has a phrase for how it was achieved. One well-known study of it is titled a study in despotic planning.

That is not a slur, it is a description of the mechanism. To rebuild a city on a new plan you must override every existing property claim at once, and there are only two conditions under which that has ever happened. The claims must be temporarily weak, which requires a catastrophe. And there must be someone with the authority to ignore them anyway, which requires a Pombal. Lisbon had both in the same year. London had only the first, and got a fire-resistant medieval city.

When the ground itself is what you inherited

There is a version of this problem with no available fix at all.

Mexico City stands where Tenochtitlan stood, in the middle of what was Lake Texcoco. The Spanish drained it, progressively, and by 1900 the valley was emptied by canal and tunnel. The city therefore rests on roughly a hundred metres of fine clay lake bed.

Then it pumped its water out of the ground beneath itself. Drained of water, the clay has been repacking itself tighter ever since, and the city has been sinking into it for over a century. Parts of the historic core have gone down by something in the region of forty to fifty centimetres a year, which is fast enough to see in a human lifetime, and the sinking is uneven, so buildings tilt and pipes shear and the drainage that was supposed to run downhill no longer does.

You cannot re-lay a street plan here and solve it. The inherited layer is not the drains or the plot lines. It is the ground. There is no version of the repair that does not involve the lake.

The thing nobody wants to say

The way this is usually written up, in Mumbai and everywhere else, is as a story about corruption and incompetence. Somebody did not do their job, somebody took money, somebody should be sacked. Sometimes all of that is true and it should be reported when it is.

But it does not explain the pattern, because the pattern shows up in cities with entirely different politics, different levels of wealth and different degrees of honesty. London in 1666 was not stopped by graft. It was stopped by property law and by people needing to earn a living that winter.

The real constraint is simpler and much harder to get around. A city is a going concern. It has to keep working while you repair it. You can replace a building because the rest of the city carries on around the hoarding. You cannot replace a drainage network, a street grid or the ground itself without switching off the district that sits on top, and no city can afford to switch off. The cost of the disruption is paid immediately, by everyone, in a form they will notice on the way to work. The cost of not doing it is paid later, occasionally, in a form that can be blamed on the weather.

So the fix loses. It loses in 1993 and it loses again in 1994, and it keeps losing until the arithmetic changes, and the only thing that reliably changes the arithmetic is a disaster.

Which brings it back to the door in Rome. Somebody installed that door at what was obviously, at the time, the correct height. They were building on eight or nine centuries of accumulated city and they had no particular reason to think about it, because the ground they were standing on felt exactly like ground. It always does. Every city is standing on a decision somebody else made, usually for a good reason, usually permanently, and mostly the residents do not know what the decision was until the water arrives and finds that the way out was filled in a long time ago.

Sources

• Wikipedia, "Temple of Antoninus and Faustina" (the 141 AD temple, its conversion into the church of San Lorenzo in Miranda, and the raised doorway left above the excavated Roman level).
• Wikipedia, "Hornby Vellard" (the causeway begun in 1782 and completed in 1784, the blocking of the Worli creek, and the joining of the seven islands by 1845).
• The Better India, "How the Hornby Vellard Helped Turn Bombay's Seven Islands Into Mumbai" (the 1769 demolition of Dongri hill to fill low ground, and the loss of the mangroves and wetlands that absorbed monsoon water).
• Scroll.in, "Another monsoon, another deluge: why Mumbai gets inundated nearly every year" (the 25 mm per hour drainage capacity at low tide, the 1993 BRIMSTOWAD recommendation to raise it to 50 mm, and the delay in funding it).
• Free Press Journal, "Mumbai 26 July 2005 Floods" (944 mm of rain in 24 hours, 644 mm of it between 8 am and 8 pm).
• Apollo Magazine, "Post-Fire London was a magnificent, beautiful compromise" (Wren's rejected plan, the return of property owners to existing plots, and the rebuilding on medieval street lines with standardised street widths).
• UNESCO World Heritage Centre, "Pombaline Lisbon" (the grid rebuilding of the Baixa, the piling of the foundations, and the anti-seismic construction).
• Wikipedia, "Gaiola (construction)" (the Pombaline timber cage developed for the reconstruction after 1755).
• Chaussard et al., "Over a Century of Sinking in Mexico City", Journal of Geophysical Research: Solid Earth, 2021 (the compaction of the drained lake bed, subsidence rates in the historic centre, and the absence of any prospect of recovery).