Correct. Accurate pendulum clocks were (reasonably) easy. But a pedulum isn’t useful on board a ship. That introduces a host of difficulties that, with much work, John Harrison was able to solve.
Bumped.
Harrison has a Google Doodle today: Google.com
Not a bad WAG since that precisely the problem. The chronometer needs to be fairly accurate over the length of a sea voyage: longitude changes at the rate of about a quarter mile per second, so fewer than four seconds of error if you want be accurate to within a mile, Which is good enough for most navigational problems.
Yeah, I saw that.
There was an interesting exhibit put together by the Royal Observatory on the quest to find an accurate means of measuring longitude at sea. There were other ideas besides Harrison’s, including a suggestion to use the moons of Jupiter as a means of timekeeping. They had replicas of Harrison’s first three marine chronometers, and a genuine case (but without the internal works) from one of his sea watches.
I saw the exhibit at the Mystic Seaport in Connecticut. Don’t know if it was shown anywhere else in the States.
That idea was actually proposed by Galileo in 1612 soon after he discovered the Galilean moons of Jupiter and determined their periods. It was hard to implement at sea due to the difficulty of accurately pointing a telescope on the deck of a moving ship. But it was actually used for surveying on land.
I have not tried it, but supposedly if you are good with a sextant you can in practice measure lunar distances within a quarter of a minute, which gives an accuracy of about 30 seconds of time, or 7.5 minutes of longitude (i.e., about 7.5 miles; not too great but it will get you across the ocean).
And of course, navigators in the early days of sail knew that their longitude measurements were lousy, and so developed navigational techniques to compensate. For instance, if you’re trying to reach a particular island that you know the latitude of, you first sail to that latitude, at a point sufficiently far to the east or west of that island that you know that you’re east or west of it (the Principle of the Deliberate Error), and then sail due west or east until you bump into it.
Nobody mentioned the good, though very long, 2000 film Longitude with 7.9 IMDb stars.
~ ~ ~
I became aware of this interesting fact years ago … perhaps by browsing this very thread? 
~ ~ ~
Rock Hudson explains why he needs a clock to navigate. In case anybody wants to watch this 1958 film, I’ll spoiler the scene I could NOT find on Youtube:
Rock Hudson gets drunk and forgets to wind his clock.
With all of the talk of latitude and Polaris, wouldn’t any star at its zenith work for finding latitude as long as you knew it’s declination that night?
Or at its nadir, for circumpolar stars. And in principle, you could use it at any time of night, if you know the local sidereal time. Or use any two or more stars at any time, even if you don’t know local sidereal time, and you’ll then get the sidereal time as a bonus. Of course, in each of these cases the math gets more complicated.
And “declination that night” is easy, since it (for practical purposes, over human timescales) doesn’t ever change.
IIRC, the difference between the Paris and Greenwich longitude plays a role in the treasure hunt in a Tintin adventure story. Basically, we forget that things were not always standardized in the Good Old Days.
Keep in mind that a method is not really practical if it cannot be computed in several minutes using an almanac and some tables of reasonable length. Even the determination of latitude by observation of Polaris requires taking into account its circular motion around the pole (the current almanac has a table of correction terms based on the local sidereal time, time of year, and approximate latitude, plus a real sailor should be able to figure it out based on the relative position of other stars like η Ursae Majoris). The star itself may be reasonably steady over human timescales, but the pole isn’t (it precesses approximately one circle every 26000 years), and that definitely needs to be taken into account (not every single night, but over years).
That depends. Are you trying to determine the exact location of your ship under full sail, so you know what direction to head for your destination? Are you trying to determine the exact location the ship was at, when some event happened? Or are you trying to determine the exact location of an island you’ve stopped at, or of a city? In the first case, you want to complete the whole process quickly. In the second, you want to complete the observations quickly, but can take your time on the calculations. In the third, you can take your time on both the observations and the calculations, and might even be able to wait for a convenient eclipse, and send your observations home to Greenwich (or wherever your home base is) for the data processing.
In actual practice, sailors pretty much universally found latitude by using the nearest, brightest, star’s elevation at its zenith. As a bonus, that also gave them the exact moment of local noon.
Well, maybe, depending on what you already know and what you’re willing to do to measure it. You can find the Sun’s zenith (and hence local noon) by repeatedly measuring its altitude and using the highest one, but that depends on making many measurements over the course of perhaps an hour. You can get local noon by splitting the time difference between sunrise and sunset (if you have a flat horizon like the ocean), but then you have to make observations over the course of twelve hours, and you need a clock that’ll stay good for that timespan. You can use the moment when the Sun is due south (or due north, if you’re in the south), but then you need some way of knowing true North. You can get true North by splitting the difference in angle between sunrise and sunset, but then you’re back to twelve hours of observation. You can also get true North from the night sky, by sighting the North Star (and correcting, if you care about that level of precision), but if you’re doing that, you can just get latitude directly from that, too. You can get north from your magnetic compass, but then you need to know the magnetic deviation for your location. Any of these methods could be used, depending on exactly what your needs were.
Bowditch (1861) gives algorithms for finding the latitude by the meridian altitude of any object (the sun, a fixed star, the moon, or a planet), by double altitudes of any of the above (including observing two different objects), by one altitude of the sun and a previously-regulated watch (of course many methods are described for determining the time), by the mean of several altitudes of the sun near noon, by means of an artificial horizon on shore, by the polar star, etc. He writes,
In the chapter on keeping a ship’s log, he basically suggests taking celestial observations whenever it is possible to do so satisfactorily (he gives detailed examples, such as finding the altitude of the sun (occasionally the moon or a planet/star), and measuring lunar distances. I don’t know whether he made the example data up or whether the voyage was real.
The fancy lunar/chronometer stuff is obviously modern, although determining latitude from the altitude of the sun is an ancient method. Not every navigator is a Bowditch, and presumably a simple and reliable method is best.
Oh, I should also mention that not all measurements are equal, where precision is concerned. For instance, multiple observations of the Sun for a little while before and after local noon will give you a pretty good measurement of your latitude, but a not-so-good measure of the time of local noon, because noon is when the altitude of the Sun is changing most slowly. Similarly, if you take just one measurement at what your clock says is noon, or when your compass-plus-correction says it’s due south, it won’t matter much if your clock or compass is a moderate amount off. If you want a really good measurement of the time of noon or of true north, then you want to use the sunrise-and-sunset method.
In 1609 King Phillip wanted to know the extent of his kingdom. That required knowing the longitude of his various territories.
It was done in each territory using the date of a predicted lunar eclipse and measuring the time between sunset and the eclipse using hour glasses.
The results were sent back to Spain by ship. They were surprisingly accurate.
Crane
Could you not also use the time that’s halfway between two times at which the sun’s altitude is equal? (The sunrise/sunset method being a special case of this.)