A city under a city. Not fibre, not copper, not anything you can see. Just pipe, and air, and a capsule moving through the dark at fifteen miles an hour with somebody's urgent news inside it.
And for about a century, that was how a surprising amount of the world's mail actually moved.
Daniel's been reading about forgotten infrastructure again. His angle is the stuff that looks comically elaborate now but was once an ingenious fix for a very real problem, and pneumatic tubes are his Exhibit A. He wants to know how these networks actually operated. How a capsule was propelled, how fast it went, how anybody routed a message through a network that spanned hundreds of kilometres. Whether the systems could carry real traffic or whether the physics capped them from the start. He wants Paris specifically, the network that ran for over a century, and how London and New York compared. Why some survived so long and what finally killed them.
And he noticed they never actually died.
Right. Hospitals still run them for blood and medication. He wants to know why that makes sense in a world where everything else went digital. And whether anybody seriously proposed going further, into parcels, freight, even passengers.
Passenger pneumatic transit. He's going to get his answer on that one.
Let's start with the obvious question. How does a tube system actually move a message across a city?
Push and pull. That's the whole trick, and it's the part people get wrong. There's this image of a capsule being blown through a pipe like a spitball, and that's half of it. Compressed air goes in behind. But ahead of the capsule there's vacuum, and the vacuum is doing just as much work. You're sucking it forward while you push it from behind. Pressure alone would be inefficient and honestly dangerous, because you're building up a column of air in front of a capsule that has nowhere to go. The moment you pull from the front as well, the thing flies.
How much pressure are we talking?
Less than you'd think. Paris ran at roughly ten psi. London ran ten psi of pressure and six and three quarter psi of vacuum, and the 1911 Encyclopaedia Britannica noted that combination gave approximately the same speed as the Paris system. So both cities had converged on the same answer independently. Ten psi, vacuum ahead, and a capsule that fits the pipe tightly enough to be a piston.
A piston. That's the thing, isn't it. The capsule seals the tube. It's not a package rattling down a chute, it's a working part of the machine.
And it's why the tolerances mattered so much. Paris called the carrier a curseur, which literally means slider. It's a felt-padded cylinder that has to seal against the tube wall without jamming. If it's too loose, you lose pressure and the thing stalls. Too tight, and it seizes.
What sort of speed did that actually produce?
Paris averaged about a kilometre per sixty seconds. So call it twenty-four kilometres an hour, fifteen miles an hour. Trains of up to ten canisters, about four kilos total. Some sources push the peak speed to forty kilometres an hour, and I'd treat that as a best case rather than the norm.
Fifteen miles an hour. My commute.
Your commute is a nap, Corn.
A productive one.
New York got up to thirty-five miles an hour with theirs, and they were moving canisters that held six hundred letters each. That's not a message. That's a mailbag with a felt collar.
So then the obvious follow-up. How does the capsule know where to go? A tube is a pipe. A pipe doesn't have junctions that read an address.
At first it didn't know anything, because a human did the knowing. Every station on the line, the train stops. An operator opens the tube, pulls out the canister addressed to his station, drops in a canister that needs to go onwards, seals it, and sends the train on. That's the entire routing system for the first several decades. A person at every station, opening the pipe, swapping the contents, closing it.
Twelve minutes for the first hexagon loop in Paris, wasn't it? Clockwise, all the way round.
Twelve minutes for a full clockwise circuit of the Paris hexagon, yes. And remember there's a person at every vertex of that hexagon doing the swap. So the speed of the network isn't the speed of the capsule, it's the speed of the capsule plus however fast a man can work a seal.
And that was the bottleneck.
That was the bottleneck for sixty years. Then in 1931 Paris introduced automatic redirection, capsules steered across the network without a human opening the pipe, and Berlin had electric routing by the same year. That's the transition I find remarkable, because what you've just built is a switching network. You have addressed packets, mechanical switches, and a routing decision made by a machine at a junction rather than a person at a counter.
It's packet switching in a metal pipe. That's what that is.
It's packet switching in a metal pipe, decades before anybody said the words. The difference is the packet weighs four kilos and arrives with a felt collar.
Where did the power come from? You're compressing air for a network that spans a city. Somewhere there's a plant, and plants are loud.
This is my favourite detail in the whole subject. Early Paris didn't use engines at all. They compressed the air with water. A fifteen-metre head of water in reservoirs, which generates two atmospheres of pressure, and they chose that method specifically in districts where noisy machinery was not permitted.
They engineered the whole power source around not annoying the neighbours.
They built a water-powered air compressor because a steam engine would have been heard. And it works. You've got reservoirs on high ground, you run the water down, you drive the compressor, you pressurise the network. The system that moves urgent messages across Paris is being driven by gravity and plumbing.
So the tube system is quieter than the machinery that would have run it. I'd have signed off on that.
Now, scale. And this is where the numbers get hard to hold in your head. Paris peaked in 1934 at four hundred and twenty-seven kilometres of pipe and a hundred and thirty offices. That's a network you could drive from Paris to Lyon along.
Four hundred and twenty-seven kilometres. Under one city.
And it averaged around ten million messages a year through the thirties. In 1945, its busiest year, thirty million. That's not a curiosity. That's a utility.
What was New York running?
Twenty-seven miles of tube across twenty-three post offices, moving up to ninety-five thousand letters a day. That was about thirty percent of the city's mail. By the mid-twenties, fifty-five percent of New York's mail went through tubes. More than half the mail in the largest city in the country was travelling through a pipe.
And the canisters held six hundred letters.
Twenty-one to twenty-four inches long, seven to eight inches in diameter, twenty-five pounds of steel, six hundred letters, thirty-five miles an hour. You could send a substantial chunk of a day's correspondence from the General Post Office to Grand Central in a few minutes.
London?
London's first pneumatic post actually opened in 1853. Latimer Clark, from the Stock Exchange to the Electric Telegraph Company, two hundred and twenty yards. That predates Paris by thirteen years. By 1909 London had forty miles of tube, and the Central Telegraph Office alone employed three thousand people.
Three thousand people in one building, moving paper.
And Berlin gets slept on in this story. Two hundred and fifty-four kilometres of tube at its peak, speeds up to ten metres a second, and two point three million items in 1898 alone. That's the nineteenth century.
Here's the question Daniel actually put his finger on. Could these systems carry substantial traffic, or were they capped by the physics from day one?
Both, and the answer is the cross-section. The pipe diameter is the ceiling. Paris ran sixty-five millimetre tubes, later eighty. New York's canisters were seven to eight inches across. Whatever you're sending has to fit inside a tube the width of a coffee cup, or in New York's case, a tube you could just about get your forearm into.
So the system is high throughput for small things and structurally incapable of anything else.
You can move a hundred thousand letters a day. You cannot move a parcel. Not a shoebox, not a bicycle part, not a bag of anything. The tube decides what the network can ever be, and the tube was sized for paper.
Which is fine, until the thing people want to send stops being paper.
And that day arrived. It just took about eighty years.
There's a cultural layer here too, isn't there. The petit bleu.
The petit bleu was the blue telegram card in Paris, and sending one was a statement. It meant this is urgent, and everyone in the recipient's household knew it the moment it arrived. The tubes weren't just a delivery mechanism, they carried a social signal. You sent a petit bleu the way you'd send a courier now.
And the tubes survived in Paris longer than they should have, for a reason that has nothing to do with tubes.
The telephone system. For decades the Paris telephone network was unreliable, and the tubes kept working. So you have a physical messaging network outliving the electronic one that was supposed to replace it, because the electronic one wasn't very good.
The future arrived and it dropped calls.
The future arrived and it dropped calls, and the eighteen-sixties pipe was still the reliable option. There's a lesson in that about how fast old infrastructure actually retires, which is not fast, and usually not on schedule.
Let's get into the decline, because the technology didn't stop working. Nothing about a pneumatic tube wears out conceptually.
Nothing about it fails conceptually, but everything about it wears out physically. By 1966, fifty-seven percent of Paris's eighty millimetre tubes and sixty-five percent of the sixty-five millimetre tubes were worn. That's the pipe itself, underground, degrading. And worn tubes mean the seal doesn't hold, which means capsules jam.
Two hundred and seventy blockages in 1970.
Two hundred and seventy blockages in 1970, on a network that was also being starved of budget from 1945 onwards. And the price was climbing the whole time. In 1902 a pneumatic message cost roughly three times a letter. By 1975 it cost seven point eight times a letter.
So the premium went from three to one to nearly eight to one, and the thing you're paying extra for is decaying.
And the alternative stopped being a boy on a bicycle. The alternative became a van, then a fax machine, then a phone call. Paris closed the public pneumatic post on the thirtieth of March, 1984, at five in the afternoon, after a hundred and eighteen years. And they replaced it with two services, one a fax service and one a parcel delivery. The tube network was retired in favour of the fax machine.
Which lasted, what, fifteen years?
The fax machine didn't outlive the tubes by much, no. And there's a lovely coda, which is that Paris kept one line running. The government network linking the Senate and the National Assembly to the Journal Officiel, the official gazette, stayed in service until 2004. So the last pneumatic tube in Paris was carrying legislation.
The state kept its own private tube for twenty more years because it didn't trust anything else with the paperwork.
Nobody wanted to be the man who replaced the reliable thing with the new thing and then had to explain the outage.
New York's story is blunter.
New York is where the economics argument happens out loud. Postmaster General Albert Burleson in 1917 calls the tubes an incubus and the rental exorbitant, unjustified and an extravagant waste of public funds. That's the head of the postal service describing his own infrastructure in a congressional hearing.
Incubus. That is a strong word for a pipe.
He wanted it gone. It took until 1950 for the first real cut, when the Brooklyn to Manhattan line was severed during the Brooklyn Bridge reconstruction and simply never restored. Three years later, on the thirty-first of December, 1953, Postmaster General Arthur Summerfield shut the whole thing down, and his line is the one to remember.
Go on.
The mail can be satisfactorily handled by the addition of two trucks at a yearly cost of about twenty-five thousand dollars. The tubes were costing about a million dollars a year.
Two trucks. They traded a citywide network for two trucks and a rounding error.
To be fair to Summerfield, the network was expensive to run, the pipes were old, and the trucks did work. But you can hear the entire twentieth century in that sentence. The distributed fixed network loses to the flexible vehicle, every time, on a spreadsheet.
And the pattern repeats. Hamburg's Grossrohrpost closed in 1983. Berlin's two hundred and fifty-four kilometres was gone by the middle of the century. Same shape every time. Pipes wear out, maintenance is a line item that never wins a budget fight, and the van gets cheaper.
Except in one place.
Hospitals.
Hospitals. Pneumatic tubes are basic inventory in any hospital with more than two hundred beds. Stanford runs seven thousand transactions a day through four miles of tubing, a hundred and twenty-four stations, twenty-nine blowers. Cylinders at about twenty-five feet a second, eighteen miles an hour. The longest run is fifteen hundred feet and it completes in under three minutes. Uptime is ninety-eight point eight percent.
Ninety-eight point eight percent uptime on a mechanical system moving seven thousand things a day.
The Hopital Cochin in Paris has fourteen kilometres of tubes and a hundred and fifty collection points. That's a network roughly the size of a small nineteenth-century city post, running inside one hospital.
Why? What does a tube do that a digital system can't?
Nothing digital can carry a blood sample. That's the whole answer and it's worth sitting with. Every other kind of information in a hospital became electronic. The chart, the order, the result, the note, all of it went digital, and that made the physical transport of the specimen the bottleneck. Because the specimen itself never digitises. You can order the test in two seconds and then you have to get the tube of blood from the ward to the lab.
The digital revolution didn't retire the tubes. It promoted them.
It isolated them. Everything around them got fast and free, so the one remaining physical step became the thing everybody waits on. Labs quote you a turnaround time and the turnaround time is now mostly transport.
That's a strange outcome. The hospital is the most technological environment you can walk into, and the most advanced logistics system in it is a hundred-and-fifty-year-old idea moving a cylinder with air.
And the engineering on top of that old idea is modern. Since the late eighties, systems have controlled airflow so the capsule lands softly instead of slamming into the receiving station, because you're not moving letters anymore, you're moving serum and plasma and blood gases. Modern systems add RFID tracking and computer-controlled transfer units. The pipe is Victorian. Everything about what happens at the ends of it is twenty-first century.
What about the damage question? If you're firing a blood sample through a tube at eighteen miles an hour and stopping it, somebody must have asked whether that ruins the sample.
Constantly. There's real literature on it, on whether pneumatic transport causes hemolysis, which is the red cells breaking open and contaminating the sample. A meta-analysis in 2021 and a clinical study in 2016 both looked at it. The concern is genuine. The finding is that modern systems with soft landings have largely managed it, but it isn't nothing.
So the hospital keeps an old technology and then spends real money making it gentle.
Which is the honest picture. It's not that tubes are perfect. It's that the alternative for a stat blood gas at two in the morning is a person walking.
Or running.
The chief of staff at Stanford put it well. We like the human touch, but we don't like the human foot speed. And their lab manager, when asked what happens if the system goes down, said she doesn't know what they'd do without it, especially for the tests that happen during surgery.
That's the sentence that reframes the whole episode, though. During surgery. You're not sending a routine panel, you're sending the sample that decides whether the surgeon closes or keeps going.
And the tube answers in under three minutes, on a fifteen-hundred-foot run. A porter cannot.
Let's get to the part Daniel really wants. The people who looked at this and thought, why stop at letters.
They never stopped at letters. The London Pneumatic Despatch Company in 1859 stated its ambition explicitly, and the phrase is all kinds of general freight, and passengers. That's a company charter. That's not a speculative article. They intended to move people.
How far did that get?
There was a real proposal, the Whitehall Pneumatic Railway, an eighteen-foot-diameter tube under the Thames between Waterloo and Charing Cross. Carriages driven by compressed air and vacuum, a train every two minutes. Eighteen feet across. That's a tunnel you could stand up in, with a train inside it every hundred and twenty seconds.
What killed it?
The financial crisis of 1866. Not physics, not a technical failure. The money ran out before the tunnel got built.
And then Beach, in New York.
Alfred Ely Beach built a working passenger pneumatic subway under Broadway in 1870. Three hundred feet of tunnel, a twenty-two-passenger car, twenty-five cents a ride. Four hundred thousand rides in its single year of operation.
It worked. It actually worked, and people paid to ride it.
It worked. And it got blocked by Boss Tweed, who controlled the politics of the street it was under, and then the Panic of 1873 killed the financing. So the first two serious attempts at passenger pneumatic transit, one in London and one in New York, both die of money, and neither dies of engineering.
Which is the same cause of death as the postal networks, incidentally. Money, not mechanics.
Same disease. And it keeps going. In 1889 the Johnson Pneumatic Tube Company proposes seven-hundred-and-fifty-pound iron capsules, cannonballs, travelling at two hundred miles an hour, and predicts New York to Chicago in five hours. That's 1889. That's a same-day intercity parcel service, proposed before the automobile existed.
Two hundred miles an hour in a cast-iron ball. What could go wrong.
Nothing they wrote down. And the modern version is the Foodtubes Project in 2010, which proposed covering Britain in high-speed pneumatic tubes to replace supermarket lorries entirely. That's the same idea a hundred and twenty years later, aimed at freight instead of mail.
Did any of it scale? Any of the passenger or freight proposals, is there a single one that became real?
No. That's the honest answer, and it's worth saying plainly. Beach's line was a demonstration and closed. The Whitehall railway was never completed. The cannonball scheme and Foodtubes never got past concept. No large-scale pneumatic passenger network has ever been built anywhere.
But the concept isn't dead. There are pneumatic capsule pipelines moving coal and ore and grain over long distances, aren't there? Two hundred kilometres and up.
Industrial bulk transport, yes. Two to three hundred kilometre pipelines moving ore and aggregate. But that's a mine or a port, not a city, and it's carrying rock, not mail. The urban version never found a niche that roads couldn't fill more cheaply.
One more thing Daniel would enjoy. The hoax.
1935, Modern Mechanix runs an article claiming Berlin delivers food to every apartment by pneumatic tube. Dinner arrives up a pipe, to your kitchen, on demand. It's judged an April Fool's hoax, and it's a good one, because it's plausible. Everything in it is a slightly larger version of a system that already existed in that city.
They had two hundred and fifty kilometres of tube under Berlin. Punching a line into an apartment building isn't a stretch, it's an extension.
The idea had captured the imagination so completely that a fake story about it read as ordinary.
Senator Depew in 1897 said the pneumatic tube is a howling success. He was right for about forty years.
And then he was wrong for forty, and then he was right again, in hospitals, for reasons nobody in 1897 could have predicted. That's the shape of this whole thing.
Herman, one question I want to put to you directly, because I don't think the answer is as clean as the numbers make it look. When a city like New York trades a network for two trucks, is that actually a rational decision, or is it just the decision that wins a budget meeting?
It's both, and that's what makes it uncomfortable. On the numbers, the trucks were cheaper, and they were flexible, and you didn't have to dig up the street when the pipe failed. But Summerfield wasn't comparing a network to two trucks. He was comparing the cost of maintaining an aging fixed asset to the cost of a new variable one, and fixed assets almost always lose that comparison, because the maintenance is a known bill every year and the truck looks like a smaller number.
So the tubes were killed by accounting, not physics.
The tubes were killed by accounting, and the accounting wasn't wrong. That's the part I keep turning over. It was a defensible decision that was also a permanent one.
And it's not clear anybody could have chosen differently. Paris spent a hundred and eighteen years running a network it knew was losing money, and it only stopped because the pipes physically wore through.
Which may be the only honest ending to any infrastructure story. It runs until the metal gives out, and then a man in a suit closes it at five in the afternoon and announces a fax service.
We had one. A hospital I worked at, in the early two thousands, had one. And I want to say two things about it, because you've both been talking about the wrong part.
You spent twenty minutes on pressure and pipe diameter and routing, and none of it is why the thing actually survives. It survives because it's fun. That's it. That's the whole reason. You cannot tell me a nurse prefers typing an order into a computer over loading a cylinder and watching it disappear. The tubes make a grim job slightly less grim. I saw a nurse in that hospital send a bag of blood up to the lab with a note taped to the cylinder that said stat, and then a smiley face. Stat with a smiley face. That is a person enjoying a machine.
You think that's why the hospitals keep them?
I think the hospitals keep them for the turnaround time, and I think the turnaround time is why they were never the first thing cut, and I think the smiley face is why nobody ever volunteered to be the one who killed it. But that's not what I came out to say either.
What did you come out to say?
There was a station in the basement nobody used. Off the manifest. The line still went there, the diverter still worked, but nobody had sent anything down it in years, and the receiving end was inside a room that got locked in nineteen eighty-something. So I figured out that if I addressed a cylinder to that station, the diverter would take it, and then I could walk down, open the receiving hatch, take it out, and put it back in the send side. And it would come straight back to me. Addressed to the station it just left, the one that's decommissioned.
So you built a loop.
I built a loop. And I ran that cylinder back and forth for an entire shift. Same cylinder, over and over, in the basement, while the hospital ran on top of me. Nothing in it. Nothing in it at all. It was just going back and forth because I'd worked out how.
How long was a shift?
Twelve hours. I want to be precise about that, it was twelve hours, and I did it for most of them. I'm not proud of it. I'm not ashamed of it either. It was the single most satisfying thing I did at that job.
You spent twelve hours sending an empty cylinder to a room nobody had opened in twenty years.
I spent twelve hours sending an empty cylinder to a room nobody had opened in twenty years and I'd do it again tomorrow. There was a morgue station too. On the manifest. Live. There was a morgue station and I never sent anything to it and I thought about it the entire time I was there.
The morgue had a tube station.
The morgue had a tube station, yes.
And you didn't use it.
I didn't use it. Some lines you don't cross, Herman, and the morgue tube is one of them. But the thing I actually wanted to tell you two. When they renovated that basement in the mid-nineties, they opened up one of the old tubes, and there was a cylinder inside it that had been sealed in there since the original installation in the sixties. Not stuck. Deliberately sealed. The crew that put the system in had left it.
A time capsule.
A cylinder with a note in it and a pack of cigarettes. And the note said, if you're reading this, the tubes still work. Good.
That's it?
That's the whole note. If you're reading this, the tubes still work. Good. Men installed that system in 1963, sealed a cylinder in the wall, and trusted that the next person to open it would be forty years later on a renovation. And they were right.
And the cigarettes.
My brand. Still my brand. I've smoked them since I was fourteen.
Of course you have.
There's a cough at about the nine-minute mark you'll want to take out, by the way, the one before the morgue section. I'll flag it in the edit.
Noted.
The point is the men who installed it did the job properly, and then they wrote a note about it. That's all. That's the only part of any of this I actually care about. The engineers wrote a note.
That's a hell of a story.
Let's land this properly. Here's the question I keep coming back to. If we'd kept building these instead of roads, what would a city look like now. The Foodtubes Project is a reminder that the idea never really went away, that somebody in 2010 was still proposing to cover Britain in tube.
And the plain answer is that the roadway won, and it won for reasons that had nothing to do with pneumatic tubes being bad. Roads carry anything. Cars, buses, vans, ambulances, bicycles, and they can be rerouted in five seconds. A tube carries one size of thing and can't change its mind. Flexibility beat throughput, and it wasn't close.
But the tubes aren't dead, and that's the part Daniel noticed. As hospitals get more digital, the physical transport of the sample gets more of the bottleneck, not less. Every other step in the chain went to zero. The one that didn't is now the whole delay.
The tubes are a solution to a problem that software cannot solve, and that is a very short list. You cannot email a blood sample. You cannot fax a blood gas. So the hospital keeps the pipe, updates everything at the ends of it, and keeps the uptime above ninety-eight percent.
A hundred and fifty years of infrastructure, kept because of a tube of blood at two in the morning.
That's where I'd leave it. The tube is the technology that survived by being the one thing digital couldn't replace.
Thanks, as always, to our producer Hilbert Flumingtop, who has been at the desk this whole time.
If you enjoyed this one, leave us a review on your podcast platform of choice. It helps other listeners find the show.
And send us your own prompt on Telegram at t dot me slash MWP listener bot. If this was your kind of episode, go back for episode ten, How ASR Went From Frustration To ... Whisper Magic; and episode four twenty-five, The Arc of Deprecation. This has been My Weird Prompts.
We'll be back soon.
