[fellerts]

Timekeeping leading up to the marine chronometer is also very interesting. Note that none of the timekeeping devices mentioned in the article would work very well at sea, making it impossible to accurately determine your longitude. Solving that proved to be extremely difficult.

https://www.rmg.co.uk/stories/topics/harrisons-clocks-longit...

[vbezhenar]

> John Harrison was a carpenter by trade who was self-taught in clock making. During the mid-1720s he designed a series of remarkable precision longcase clocks. These clocks achieved an accuracy of one second in a month, far better than any clocks of the time.

Wow. I own Tissot and Casio hand clocks and both are significantly worse precision. Are there any available clocks with that precision? Of course with GPS, WiFi and other wireless ways to access atomic clocks that's not so important, but still interesting.

[dgacmu]

A cheap watch or clock will use an off-the-shelf crystal oscillator, where you'll be lucky if it's accurate to 20 seconds in a month.

A temperature compensated crystal oscillator (TCXO) can do better than one second per month, but probably not better than a second every 2 months.

An oven controlled crystal oscillator (OCXO) or a double ovenized crystal oscillator can get an order of magnitude better than that, But they start costing between $50 and $1,000. Beyond that you get into funky stuff like rubidium or cesium atomic clocks. There are, notably, miniaturized atomic clocks these days - about 2" x 2" - So you could technically put one into a wall clock, although they cost about $2,000.

These days it's usually cheaper to use GPS to control a temperature compensated oscillator, which we call a GPSDO (gps disciplined oscillator).

[russdill]

The mems temperature controlled oscillators are also very good, 0.05ppm range (about a second every 8 months). They are tiny and low power, but still >$79.

[WalterBright]

I used a wall "atomic" clock for a couple decades. It would synchronize itself with a government radio signal.

[dgacmu]

WWVB is great if you can get the signal -- I have two of those clocks in different rooms, and only one of them synchronizes.

[jrockway]

Newer clocks use the phase modulated signal: https://www.nist.gov/system/files/documents/2017/05/09/NIST-...

If you have AM-based clocks, you might find that these work better. However, nobody advertises which module they're using, so it's kind of a crapshoot. I specifically bought a phase modulation WWVB receiver and it reliably receives the signal every night. I also have a random $7 clock from Amazon that pretty much synchronizes every night. Living in NYC, this is truly amazing to me; I never ever got the AM signal here.

[creeble]

Thanks for that! Didn't know there were two kinds of radios.

I have one clock that always works perfectly, syncs after DST 100% of the time. Another that always requires manual reset twice a year, defeating it's purpose.

[cvoss]

It's an incredible feat to me that a single station can broadcast to the almost the entire US, Mexico, and even parts of South America at the right time of day [0]. (I believe this is why many clocks check for the signal overnight.) It's achieved by using an extremely low frequency, at 60 kHz. The antenna is enormous and suspended between 4 towers.

[0] https://tf.nist.gov/stations/wwvbcoverage.htm

[1] https://www.nist.gov/pml/time-and-frequency-division/time-di...

[ghaff]

Communication with subs uses ELF/SLF (3-300 Hz) when they're underwater--in the US case from some big communications stations. The coastal VLF stations, like in Cutler Maine are more in the 24 kHz range. As I recall, at one point, there were ecological concerns with the ELF stations but apparently the projects ended up going ahead.

[eternityforest]

The rubidium standards show up on eBay a lot for much less, but I don't know if they're still accurate and reliable after being used for years, I would imagine they're replaced for a reason

[throw0101d]

> Are there any available clocks with that precision?

Citizens says ±1s/year with their Caliber 0100:

* https://www.citizenwatch-global.com/the-citizen/lineup/cal01...

* https://www.hodinkee.com/articles/citizen-eco-drive-caliber-...

Solar powered as well.

It should be noted that Harrison's watches were low-volume productions, so could be tuned very accurately—and their accuracy was a matter of life and death for ships and their crew. Most mechanical watches 'normal' people can get nowadays are produced at much higher volumes, and so aren't adjusted as much—especially since few people demand it and are willing to pay for it:

* https://www.hodinkee.com/articles/how-accurate-should-your-m...

The Harrison watches/clocks were built for the British government/Navy, so were fairly price insensitive (military procurement and such).

Being 1 minute off in time throws off distance by 15 nautical miles (~28 km), so being off by a second can cause about 500m worth of position inaccuracy.

Also: it may not be a big deal if your clock/watch drifts as long as it does so at a known, consistent rate which you can adjust for.

[SoftTalker]

> Being 1 minute off in time throws off distance by 15 nautical miles

I'm surprised that the accuracy of a hand-held sextant used while standing on a rolling/pitching deck is good enough for that to matter very much.

[storyinmemo]

> Modern sextants can read the angle to a 0.1 minute level of accuracy, i.e. one-600th of a degree or one-tenth of a mile. In practice, actual accuracy to one-half mile is acceptable and quite good. The usual standard is accuracy to within five miles. The sextant (or octant) is meant to get the ship across the ocean. Once near the coast (20-100 miles) the more accurate techniques of piloting are relied upon for a safe landfall. The sextant is still the standard instrument for taking the observations required for celestial navigation.

https://www.ion.org/Museum/item_view.cfm?cid=2&scid=14&iid=2...

One thing that helps with a pitching deck is that the horizon and object remain the same angle. Like a camera following race car, the objects move together.

[throw0101d]

> […] for that to matter very much.

Life and death at times:

> The Scilly naval disaster of 1707 was the loss of four warships of a Royal Navy fleet off the Isles of Scilly in severe weather on 22 October 1707.[a] Between 1,400 and 2,000 sailors lost their lives aboard the wrecked vessels, making the incident one of the worst maritime disasters in British naval history.[2] The disaster has been attributed to a combination of factors, including the navigators' inability to accurately calculate their positions, errors in the available charts and pilot books, and inadequate compasses.[3]

* https://en.wikipedia.org/wiki/Scilly_naval_disaster_of_1707

When you're trying to find a tiny speck of an island (Pitcairn, Saint Helena, etc) in the middle of an ocean it's also important.

[SoftTalker]

Yeah I get that it's important, I just didn't realize that angles could be meaured so accurately by eye at that time (especially on a ship that's moving around on even a moderate sea).

[throw0101c]

> I just didn't realize that angles could be meaured so accurately by eye at that time

Sextants have a bit of magnification (usually 4x, but sometimes 7x or higher). Higher mag allows for better accuracy at the cost of more shaking of the view.

Basic explanation:

* https://www.youtube.com/watch?v=30_wEda3ltM

[dredmorbius]

It was good enough for the Boeing 747:

<https://virginflightdeck.blogspot.com/2010_09_01_archive.htm...>

(Something of a stealth favourite amongst HN commenters: <https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que...>)

And not the only plane to make use of this option: <https://news.ycombinator.com/item?id=9549245>

[ghaff]

Fairly price insensitive.

As I understood the exhibits in Greenwich, the Harrison clocks were largely eclipsed by cheaper versions by the time payment issues were resolved and they made it into widespread production.

[dekhn]

https://en.wikipedia.org/wiki/Shortt%E2%80%93Synchronome_clo...

technically it's "electromechanical" but does not use GPS, WiFi, quartz, or atomic methods. To make thigns this accurate you need extremely good temperature compensation, close-to-frictionless bearings, run in a vacuum, use a pair of pendulums, etc.

When you reach that level of accuracy, you end up basically building a sensitive measurement device that is influenced by second and third order terms like subtle changes in the shape of the earth.

[LeoPanthera]

> Are there any available clocks with that precision?

Yes! The term you need to type into Google is "High Accuracy Quartz". The most easily available wristwatches with this level of accuracy are the Bulova Precisionist line.

Watchuseek has a subforum about HAQ: https://www.watchuseek.com/forums/high-accuracy-quartz-watch...

[Terr_]

If you ever find yourself stranded back in time at that spot, skip over the fancy clocks and invent a radio transmitter that can send out a powerful pulse for "it's noon in $city" once a day.

Then sailing ships can manage with simple receiver plus clock just good enough to measure 12 hours. The time delta between receiving the land-broadcast and their own observed noon will tell them their relative longitude.

Since you aren't trying to encode any other information in this pulse, that makes the job a lot easier than if you were trying to make a wireless telegraph.

If you want to restrict who can benefit (e.g. to support a particular military that happened to recognize your time-traveler genius) plan random offsets for when you will be triggering it, and give authorized users a copy of the offsets.

[pavel_lishin]

How difficult is it to determine local noon at sea? I assume that on a rocking ship, it's probably not a trivial task...

[Terr_]

IANAMariner, but I'm assuming the sextant would already be invented, since it would also be important for determining latitudes.

The goal is to figure out when the sun is as-close-as-possible to the reference vector "straight up". On land, we might determine that vector using gravity and a plumb-bob, which is indeed going to be a lot harder on a rolling ship.

However way out at sea, a new option exists: The horizon is no longer an arbitrary mishmash of mountains and hills, but instead self-leveling water in every direction [0], meaning you can safely assume "straight up" is 90 degrees [1] from all points on the horizon.

So you'd measure the angle between the sun and the nearest horizon, and then noon would be when that angle hits its minimum.

[0] Tides exist, but I assume they aren't likely to cause one direction to be significantly higher than the other.

[1] Unless you're measuring from very high up above sea-level, but if the civilization can make ships that big then you probably don't need much navigational help.

[Terr_]

> and then noon would be when that angle hits its minimum.

*Maximum, but yannowhatimean.