[time-nuts] Re: Primary frequency standard

[Lux, Jim]

On 3/27/21 1:43 AM, Poul-Henning Kamp wrote:
> --------
> Lux, Jim writes:
>
>> Would a sundial  (or zenith meridian crossing sensor) be a primary
>> standard for measurement of a solar day?
> Using Corbys "desert island" criteria, I think the answer is "yes(footnote)".
>
> Imagine you wake up, teleported to a different planet, about which you know nothing.
>
> First you need to establish the orientation of the planets rotational axis, because your sundials'gnomon' must be aligned parallel to it.
>
> You can do this by measuring the minimal and maximal noon elevation of the appearant sun over approx 60% of the planets "year".  As a side effect you get a pretty good idea how long a "year" is in "days".
>
> Note that unlike this planet, there is no guarantee that these two numbers "mesh" well, or even which is the longer time interval[1].
>
> Next you need to partition the scale in however many "hours" you want, but that is simply an invariant exercise of constructive geometry.
>
> So I think the answer is yes.
>
> However, the footnote is that your sundial is only primary on that particular latitude, and it is debatable if it is even that, if the planets rotation has any chaotic or higher order components, such as nutation and all that stuff.
>
> [1] See for instance Venus:  A venus-day = 243 earth-days, but a venus-year = 225 earth-days.
>
Well, and there is the problem of "day" continuously changing length, so 
you'd need a correction chart for the "equation of time" - and I think 
that's where you start to philosophically get away from "primary 
standard". Although I suppose it is a primary standard (without 
correction) for "this day", recognizing that tomorrow is a "different day".

A cesium standard does require arithmetic to convert from "vibration 
period" to "useful sized unit" but that ratio is always the same.

Is the equation of time always the same? Other than the .2422 day 
difference between integer number of days and length of a year. The EoT 
has a periodicity of a year, so EoT for 00:00 1 Jan this year is more 
like 06:00 1 Jan next year, etc.  (or maybe 18:00 31 Dec)