Man, I love this stuff. Some choice excerpts:
The SI second is defined based on specific transitions of a cesium atom. If this is measured on the rotating Earth, relativistic effects that depend in part on the earth's rotation will influence a cesium clock's rate relative to a clock in an inertial frame. Thus, even the measurement of the SI second that is used to define Terrestrial Time will be at least slightly dependent on the unpredictable rate of Earth's rotation, although the effect of variations in the Earth's rotation is not currently within the sensitivity of our measurements.And:
Uncorrected Solar Time is computed from Pendulum Time by adding a seasonally varying correction for the equation of time. This varies about half an hour or so over the course of a year. The analemma is a two-dimensional graph of the equation of time. The equation of time also varies from year to year, century to century. This variation is predictable within the uncertainty of the Earth's rotation rate, so it is pre-computed over the clock's 10,000-year operating interval and stored in the equation-of-time cam based on values is derived from an extended form of the JPL DE422 solar system solution. The function encoded in the cam assumes the predicted slowing of the earth's rotational period at the rate of 1.8 milliseconds per day per century. The cam function also encodes the uncertainty of slowing of the Earth's rotational period, by a mechanism that will be described later.And:
All displays on the clock are designed to maintain accuracy to within a five-minute tick over the entire 10,000-year lifetime of the clock, as long as the clock detects solar synchronization at least once a year. It is possible that an unusual event such as a volcanic eruption could prevent the clock from detecting noon for more than a year. In this case, the clock may temporarily drift away from the correct time, but it will eventually resynchronize when clear skies reemerge as long as it has not drifted more than 12 hours.And:
As previously discussed, the Earth's rotation is currently slowing at a rate of about 1.8 milliseconds per day per century. Of course, the trend may not continue, especially if the climate changes. The variation is caused by a variety of effects including tidal drags, shifts in the Earth's crust, changes in ocean levels, and even weather. For example an ice age would put more mass near the poles, making the day shorter. Melting icecaps would make it longer. This creates an uncertainty in the average length of day of about 10 parts per million, an uncertainty of plus or minus 37 solar days over the design lifetime of the clock.
Heh, good point
That is amazing.
This reminds me of probably one of my top five things to read on the internet ever: long term warning sign porn.
http://www.wipp.energy.gov/picsprog/articles/wipp%20exhibit%20message%20to%2012,000%20a_d.htm
Time is an illusion. (Lunchtime doubly so.)
I thought it was purely mechanical. How does it detect sunlight, etc?
Thermally.
Doc Brown should have thought to put an equation-of-time cam in the Delorian. Could have minimized his required gigawatts.
I've never wanted to time travel but now I think it would be awesome to be present when the clock reaches end-of-life. Now I'm going to be awake all night wondering if I'd be alone and if I should have bought my pepper spray.
I liked this bit:
"The equation of time and solar elevation cams are rotated once a “cam year,” which is defined
to be 365.2222 mean solar days [...] This interval was chosen because the ratio is easily computed
by gears and makes a beautifully shaped cam."
Aesthetics and engineering :)
I read the paper and still don't understand how it will remain accurate. If the length of a day can change significantly from year to year, and you're only recalibrating based on day heating, how can the sidereal year stay accurate for the sky map?
See http://www.ucolick.org/~sla/leapsecs/dutc.html for plots of measured length of day over the past 2500 years. The variation is stochastic, but the overall trend of slower rotation is predictable enough over 10000 years. The principal aim of the clock is to count mean solar days and give the time of day from subdividing them; the orrery is a nice added approximation.
But that's only 2500 years, and it's quite variable. As they point out the ice caps make a difference, and it's an area that could change significantly and has a lot of uncertainty over the next 10k.
If an ice age happens they suspect that ice could block the solar synchronizer for long enough to lose phase lock on the sun.
"Neil deGrasse Tyson jested that the Long Now should put some signage on the 10,000 Year Clock so that a post-apocalyptic Earth will not think that the world will end when the clock stops working."
We don't want to be the next Myans...