The Pneumatic Clocks of Paris | Hacker News Reader

The Pneumatic Clocks of Paris

amusingplanet.com

121 pointsakpa139 comments

Meanwhile, in the USA, Western Union used their telegraph network to build out a system of synchronized clocks. A long pulse on the line at the top of the hour engaged a solenoid in the clock that pulled the minute hand back to :00.

https://hackaday.com/2019/09/25/100-year-old-atomic-clock/

And, since this is HN, it's worth mentioning that this was a subscription service.

Since the speed of sound is lower in tubes, I wonder how late the clocks were at the end of the network.

As far as i understand the signal was transmitted through telegraph lines, not tubes.

If you're referring to the pneumatic network, one of the footnotes mentions a latency of 10 seconds, but the author does the math and says that's probably not accurate, more like 58 seconds at the 20km max distance.

http://www.douglas-self.com/MUSEUM/COMMS/airclock/airclock.h...

The article doesn't seem to mention it, but the clocks were each offset to compensate for the propagation delay. It's a static correction, so easily done by just offsetting the minute hand position. (I can't seem to find the citation on this, though.)

It's amazing that Western Union didn't lobby congress to ban free, government operated, time synchronization services.

Probably didn't need to. WU had the monopoly on cross-continent communication lines. The government didn't have the resources to compete.

I was actually referencing: https://qz.com/1913544/telecom-lobbyists-stop-cities-from-bu...

This was not just for clocks. There were pneumatic network in Paris "distributing compressed air at a relatively low pressure of 5-6 bar over a network of (eventually) more than 900 km of mains, serving more than 10,000 customers." That run the industry and lasted over 100 years until 1994.

Also interesting (especially that it lasted until 1994), but the system from this article was just for clocks - I don't think power tools could use power which was only delivered for 20 seconds and then off for 40 seconds. If you translate that to computing, one system supplied the Vcc, the other supplied the (very slow) clock. No data lines unfortunately :)

These were two distinct networks, the other network was run by the PTT : https://fr.m.wikipedia.org/wiki/Poste_pneumatique_de_Paris

It was kept in operation until 2004 for government services!

"Citywide compressed air energy systems have been built since 1870.[23] Cities such as Paris, France; Birmingham, England; Dresden, Rixdorf and Offenbach, Germany and Buenos Aires, Argentina installed such systems. Victor Popp constructed the first systems to power clocks by sending a pulse of air every minute to change their pointer arms. They quickly evolved to deliver power to homes and industries.[24] As of 1896, the Paris system had 2.2 MW of generation distributed at 550 kPa in 50 km of air pipes for motors in light and heavy industry. Usage was measured by cubic meters.[23] The systems were the main source of house-delivered energy in those days and also powered the machines of dentists, seamstresses, printing facilities and bakeries." [1]

[1] https://en.wikipedia.org/wiki/Compressed-air_energy_storage

Apparently, they even had subscribers that paid half a penny a day for a timepiece.

https://www.nature.com/articles/022226a0

Another interesting illustration of innovation displacing a successful and widespread infrastructure. It wasn't better pipes or a better gas in the pipes, it was technology from a different source that displaced these compressed air clocks and obsoleted the hundreds of miles of pipe needed to transport it.

In the instance, the piping can easily be repurposed to run other kinds of lines (for example electric, phone, fibre). Which it certainly was.

I wonder what the human side of maintaining the clock network was like. Did dedicated civic employees walk the clocks each day looking for telltale time differences that would indicate a break in the lines, and then debug with maps? I bet those maps would be interesting too. And how did they keep problems in one section from propagating through the whole system? Buffers and interlocks I suppose.

My impression is that those trade/tech "expos" in the 19th century were really a massive acceleration force of innovation and industrialization. To the point that I could imagine that the lack of such expositions might be an explaining factor why the Roman empire never experienced an industrial revolution. Nowadays they're less important because they've been replaced by conferences (which are the same idea dialed up to 11, I'd argue, and no less important in driving innovation). Maybe I'm going completely off here, but would be interesting to read an analysis.

Yeah why did the Roman empire not advance into electricity/steam power? All the precursor elements were there such as metalworking. Did they just not have a network of smart nobles with leisure time to tinker? Are universities required for that? Referring to likes of Newton, Leibniz, Faraday, Maxwell, etc.

One theory is that the abundance of slave labour disincentivised any real push towards automation. That said, they did have massive water mill complexes for grinding grain.

> "...However, the system was repaired and pneumatic clocks continued to work until 1927, when the invention of accurate mechanical and electric clocks made the need for a central synchronization clock obsolete..."

This is not really true, in that you may remember from your elementary or high school days that many places still have centrally synchronized clocks.

When there are about 100 clocks distributed around a building, and not yet migrated to individually-kept radio/GPS/network time, the only staff-sanity preserving method of keeping them together is to use a central electronic pulse transmitted over the power lines (which the clocks are plugged into).

If I recall right, there is an hourly pulse that tells the clocks to wait/speed up until the top of the hour is reached. Then there's a daily pulse that tells them to wait/speed up to reset to midnight!

I would have loved to see how the world’s technology would be like if electricity was discovered much later and there had been no major wars. Probably closer to Steampunk/Gearpunk.

I wonder what reliability was like... With just one pipe accidentally cut, the pressure across the pipe network might not rise high enough to 'tick' the other clocks.

And if the clocks ever got out of sync for any reason, there doesn't seem to be any way to get them back in sync - for example they could have used a vacuum on the same pipes to 'suck' for a minute before midnight, and use a mechanical system which would fast-forward all clocks to midnight at that point.

An easy way to do 'syncing' is to have a pin stopping the clock passing midnight or passing the 'o'clock' position or something. That pin is only lifted by double the normal pressure.

Then, in the minute before midnight, you tick the clock say 3 minutes, with the final one being double the normal pressure allowing every clock to pass midnight.

That should mean every clock on the network that is up to 3 minutes slow will fastforward to being correct. And any clock that is fast will stop till the midnight sync signal. If a clock is way off (eg. 5 hours), then when it reaches midnight it will halt till it gets the signal to continue, and from tomorrow will be correct.

> The pipes ran through the sewers of the city, and the tunnels of the Metro and the RER, a commuter rail network serving Paris and its suburbs.

> However, the system was repaired and pneumatic clocks continued to work until 1927.

Considering that the RER started to operate in 1977 [0], I find that hard to believe.

[0]: https://en.wikipedia.org/wiki/R%C3%A9seau_Express_R%C3%A9gio...

None of the RER tunnels are shared/repurposed?

No that's definitely a mistake in the article. Not only are the RER tunnels from the 70s, they also run much deeper than the metro tunnels. They didn't exist in any form at the time. The metro tunnels did, though

It’s possible that on construction of RER that previously separated pneumatic lines were then routed through the RER tunnels as well, no?

I know in NYC that water pipes and electric tend to get routed through/around subway tunnels once constructed.

The RER integrates converted sections of existing tunnels, IE the RER B:

The northern terminus of the Ligne de Sceaux opened at Luxembourg in 1895. Between 1973 and 1977 it was converted into RER B.

( https://en.m.wikipedia.org/wiki/Luxembourg_station_(Paris) )

The Wikipedia article https://en.wikipedia.org/wiki/R%C3%A9seau_Express_R%C3%A9gio... doesn't sound like it:

> The RER's tunnels have unusually large cross-sections. This is due to a 1961 decision to build according to a standard set by the Union Internationale des Chemins de Fer, with space for overhead catenary power supply to trains. Single-track tunnels measure 6.30 m across and double-track tunnels up to 8.70 m, meaning a cross-sectional area of up to 50 square metres, larger than that of the stations on many comparable underground rail networks.

Paris has basically three distinct stages in "urban rail" development:

- first, the mainlines were built from stations on the then-outskirts of Paris (Gare du Nord, Gare de L'Est etc.)

- then trams and later metros were built to link these train stations to the city and serve additional areas.

- finally starting from the 1970s they started building rail tunnels so suburban trains could run directly from one train station through the city to another station - similar to the S-Bahn systems in Germany or Crossrail in London. The RER tunnels were all built in this phase.

maybe when using RER, the author meant the short-distance train lines that already existed before the RER network was built up?

I do wonder if there was an (unmentioned in article) synchronization system like a special midnight pulse to get the clocks all back in sync in case there was drift.. or it was all handled by periodic inspections

> when the invention of accurate mechanical and electric clocks made the need for a central synchronization clock obsolete, and the service was discontinued

Most accurate clocks in factories and railway stations are controlled by a central timekeeping source, so it's not quite correct that such a source is obsolete.

(edit: wording)

Yep, in many many cases it's more important to have all the clocks be the same than it is for them to all be correct - it's just with the rise of radio clock signals and NTP it's easiest to make them all be the same by making them all correct.

I also noticed that this statement was a bit weird. Accurate mechanical clocks were already pretty well invented at that point. By the mid-19th century the best regulators were good to about 0.5s/month, and any further development in anything but the most specialized mechanical clocks after that point were to make them cheaper, not more accurate.

As for electrical clocks, indeed rather than replacing central synchronization, the incorporation of electricity spurred the wide adoption of master/slave systems which didn't really subside until the last quarter of the 20th century. And most often the master clocks in those systems were highly accurate mechanical weight driven pendulum clocks, with the slaves being electrically driven. (apologies for the deprecated master/slave terminology, that's how they were referred to at the time)

The source article linked from the submission mentions electrically synced clocks with the master in the Observatoire.

I don't know why pneumatically synced clocks were preferred. Maybe it was cheaper to integrate the clocks themselves into the system? Patents?

<http://www.douglas-self.com/MUSEUM/COMMS/airclock/airclock.h...>

> At the Paris Observatoire a high-standard astronomical regulator clock was kept running on correct mean time by astronomical transit observations, being corrected daily. Pulses of electricity were sent every second to secondary clocks around the city, the wires being run through ducts in the sewers. Two loops starting and ending at the Observatoire carried thirteen clocks between them, the farthest being at a distance of seven and a half kilometres, or nearly four and a half miles from the observatory. The clocks were of a high standard, so they could keep good time even if the synchronising pulses failed. (The pulses synchronised the clocks but did not drive them, they were weight-driven in the conventional way) The secondary clocks were furnished with second-hands, and were placed so that they could be easily seen from the street, usually in prominent positions. They further distributed time by sending electric signals once an hour to synchronise various public clocks. The system came into operation in 1878.

incidentally conedison still operates commercial steam delivery in manhattan to this day.

https://www.coned.com/en/commercial-industrial/steam

pneumatic signaling is cute. it makes me imagine the french mob, running around with black strap eyemasks, stripey shirts, bicycle pumps and hand-drills jacking into the system and running the clocks forward for nefarious purposes.

I wonder if there was a way to 'reset' the clocks remotely, for example to send a 60 second burst of air that resets all the clocks to 12:00.

do the clock still exist?

OMG I misread the title.

Reply on news.ycombinator.com