Read on Twitter
In about 1900, the American astronomer Newcombe published his “tables of the sun”, a very good model of the solar system based on centuries of observations, that was used to predict orbital motions for several decades https://twitter.com/TartanLlama/status/1354203466021928964
In the 1930s, very precise pendulum clocks and the earliest quartz clocks became about as good or better at keeping time as the earth itself; by the 1940s it became clear that the earth does not spin at a perfectly even rate
The keepers of time were all astronomers, so their solution for creating a time standard that could keep up with technological progress was astronomical: instead of using transit telescopes to measure the rotation of the earth, measure the orbit of the moon
Observe when stars are occluded by the moon and use Newcombe’s tables of the sun to relate those observation times to a precise timescale
So in the 1950s the second was redefined in terms of the ephemeris year, via Newcombe’s tables and the moon
So in the 1950s the second was redefined in terms of the ephemeris year, via Newcombe’s tables and the moon
But lunar observations are very inconvenient compared to transit observations
And at the same time the atomic clock was being invented
And at the same time the atomic clock was being invented
Atomic clocks promised to make precise timekeeping much more convenient
Very quickly the early caesium clocks were calibrated against Newcombe’s tables and lunar observations to match the 1950s definition of the second
Very quickly the early caesium clocks were calibrated against Newcombe’s tables and lunar observations to match the 1950s definition of the second
In the 1960s atomic clocks were commercialized, miniaturized, and ruggedized, so it became possible to directly compare national timescales by transporting a primary time standard from one time lab to another https://www.hpmemoryproject.org/news/flying_clock/celebration_01.htm
The second was redefined again, based on caesium clocks, which had been calibrated to match Newcombe’s tables of the sun
But there was a problem
The rotation of the earth did not match this definition of the second - there are not 24x60x60 atomic seconds in a day
But there was a problem
The rotation of the earth did not match this definition of the second - there are not 24x60x60 atomic seconds in a day
In the 1960s, national time laws were still based on 24h days, not atomic seconds
The workaround was that national time broadcast signals used “rubber seconds” - they adjusted the frequency of the broadcasts slightly, to match the earth’s irregularly changing speed of rotation
The workaround was that national time broadcast signals used “rubber seconds” - they adjusted the frequency of the broadcasts slightly, to match the earth’s irregularly changing speed of rotation
Rubber seconds were kind of annoying, because they had to re-tune the high power electronics of their broadcast antennas - the carrier frequency of broadcast time signals was supposed to match the length of the seconds they were broadcasting
So they wanted to gat broadcast time signals using atomic seconds, to match the official SI standard, and so they didn’t need retuning twice a year
But what to do about the mismatch between earth rotation and atomic seconds?
But what to do about the mismatch between earth rotation and atomic seconds?
One millisecond per day doesn’t seem like much, but it adds up to a second every few years or less (at this time, around 1970, it was more like a second per year)
And celestial navigation is still a thing, both on sea and in the air
And celestial navigation is still a thing, both on sea and in the air
Celestial navigation, since Harrison invented the marine chronometer in the 1700s, measures longitude using clocks and the position of the sun
And one second of difference in time is a quarter of a nautical mile of distance at the equator
And one second of difference in time is a quarter of a nautical mile of distance at the equator
The mismatch between the need for atomic time standards and fixed frequency standards on the one hand, and the wobbly earth on the other hand, was bridged using the disgusting hack called
LEAP SECONDS
which were pushed through the standards committees with a back-room deal
LEAP SECONDS
which were pushed through the standards committees with a back-room deal
But my story is not quite done!
It’s the start of leap seconds in 1972. Computers are rare and expensive, timesharing is new
Batch jobs are run by hand, not from cron
Computers usually don’t have real time clock hardware; their time is set by hand at every boot
It’s the start of leap seconds in 1972. Computers are rare and expensive, timesharing is new
Batch jobs are run by hand, not from cron
Computers usually don’t have real time clock hardware; their time is set by hand at every boot
In the 1970s very few programmers knew about leap seconds, and even if they did know, it wasn’t practically possible to keep time accurately enough on a computer for leap seconds to matter
