[raattgift]

Latitude, mass concentrations, and climate also messed with the half-period/metre ("seconds pendulum") definition; with increasing frequency precision, one would need an almanac, an accelerometer, and probably other tools. Additionally, stabilizing the length of the pendulum under environmental conditions was already known to be tricky, with materials science unable to produce reasonably low thermal-expansion rods prior to the 20th century.

Consequently, the seconds-pendulum/metre relationship gets in the way as one might want to go to sub-millimetre length precision for parts made in different locations or at different times of the day or year. Precision copies of a prototype was more reliable in practice.

(In practice we mostly still generate precise and accurate physical artifacts and make copies from those, it's just that there one can in principle generate such an artifact just about anywhere and anywhen, calibrating with (for example) interferometry <https://iopscience.iop.org/book/edit/978-0-7503-1578-4/chapt...>)

Finally, the Trinity Clock <https://clock.trin.cam.ac.uk/main.php?menu_option=theory> is a neat examination of a well known pendulum clock that's surprisingly accurate (if not really precise; it's been reliably accurate to within two seconds over the course of a month for a very long time, but it's not going to give you a 10MHz sine-wave, and it's not a good for disciplining an oscillator which does so). Do check out the various plots.

[bonzini]

> Latitude, mass concentrations, and climate also messed with the half-period/metre ("seconds pendulum") definition

Probably not by 15% which was the difference between the traditional second and the decimal second.

[raattgift]

Sure but using a physical pendulum as a frequency standard is unreliable; an unreliable frequency standard is a bad basis for any sort of time-of-travel definition of length.

Many difficulties of using pendulum clocks (and in transporting any sort of chronometer) in real circumstances were also known before the revolution, with French clockmakers competing for the prize money in Britain's Longitude Act 1714 (13 Ann. c. 14) and the ancien regime's various prize offers in the 1740-1770s.

Prior to Harrison's marine chronometers, minimum longitude errors introduced in multi-degree changes of latitudes were indeed on the order of 10% across an oceanic part of a great circle or other more favourable route under cloudy conditions, and sufficient that in the early 18th century it was common for ships to navigate by dead reckoning along a single line of latitude -- a boon to pirates and other enemies, and also often adding many days to the travel time, in an effort to avoid the common problem (eg. HMS Centurion, 1741) of not knowing whether one was west or east of a landmark at a known latitude.

Prominent pre-revolutionary figures also disliked the idea of relying on chronometry for position/length/angle measurements generally -- most notably the excellent geometer and astronomer Pierre Bouguer (after whom the relevant <https://en.wikipedia.org/wiki/Bouguer_anomaly> is named) -- so it's not as if messing up a seconds-pendulum-based definition of a metre (and its consequences for the neat pole-to-equator 1/4 great circle length or mass of a cm^3 of water at STP, both of which now are just approximately round numbers) would have been universally outrageous.

And anyway surely one could consider a solution in which the half-period of the metre pendulum might not be exactly one decimal second. After all, at the time in practice one had to measure across many swings to obtain the effective length with reasonable precision. And Earth's rotation was known to be unstable (Richer, Newton, Maupertuis).