[LEAPSECS] What timekeeping system should the Terra Nova settlers use?

Zefram zefram at fysh.org
Mon Oct 3 11:14:59 EDT 2011

Rob Seaman wrote:

>If not sexagesimal then what? Could humans actually use a decimal

>counter productively?

The decimal day is quite practical. It doesn't catch on on Earth, of
course, because of the network effect favouring interoperability with the
existing sexegesimal division. On a new planet, if you're deliberately
constructing a new timekeeping system, you'd logically end up with
decimal subdivision, because in our present technical culture decimal
is the default way to subdivide anything. Sexegesimal subdivision is
a relic of an older technical culture.

I think the question here is which approach would dominate in practice:
copy the old mechanism or devise a new one. Pure sexegesimal subdivision
*is* applied to Mars at present, but you can attribute this to the fact
that the time-of-Martian-day users are also time-of-Terran-day users.
The situation might be different for one-way colonists, and especially
if there's no communication with (present day) Earth.

>Would the counter have to reset once a day?

Not sure what you mean here. I think if you do tabula-rasa time-of-day,
without reference to a local atomic time scale, then you get pure decimal
subdivision with no irregularity. The count gets to .9999 just before
wrapping to .0000 at midnight. Are you imagining something like counting
seconds within the day, denoted in decimal, and resetting to zero at
midnight? So on Mars, with about 88775 s per day, the count would go

:88773, :88774, :00000 (midnight). I think such an awkward radix would

be avoided at this level. You might get that in a lower-order component
of time-of-day (see my last suggestion below).

>Is there some way to indicate time without numbers?

Sure, but numbers are awfully convenient.

>Would they of necessity try to reconcile ship time with planet time

>(whatever either of these are)?

I think if the tech level is high enough to use atomic clocks on the
planet's surface, and the diurnal cycle is of great practical importance,
then there would naturally be an attempt to reconcile the local atomic
time scale with planetary rotation. That's what UTC does for us.

>There are a lot of reasons that the Martian equivalent of TAI will

>be TAI.

It can't be, once we've got precise synchronisation requirements
for Mars-local applications. Relativistic effects mean that TAI
(or TT) as perceived on Mars is substantially non-uniform. It's a
practical necessity for the base technical time scale, reference point
for radio frequencies and the like, to be uniform as perceived in a
planet-centred-planet-fixed reference frame. So I foresee us relatively
soon defining a Martian equivalent of TAI based on atomic clocks located
on the Martian surface.

>First off, all the equipment will be manufactured on Earth for the

>first few generations.

Sure, but location of manufacture doesn't dictate which reference
frame it'll tick with. Location of operation mainly determines that.
Are you using this as an argument that the clocks will tick SI seconds
rather than some Mars-sensitive unit? I'm expecting the Martian TAI to
use SI seconds, but in any case it's a trivial transformation to read
a different unit from the same equipment.

>I'm taken by your mention of Hertz as the unit of TAI. That's an

>interesting way to break the artificial symmetry between the two meanings

>of the word "second".

There's some resemblance in your suggestion to the old controversy from
the early days of atomic clocks. I recall reading that some astronomers
denied that the atomic clock was a timepiece at all, claiming that it was
merely a frequency standard and that Earth rotation was the true time.
Relatedly, there was the proposal that the atomic second should be named
"essen" rather than "second". If you're building a new timekeeping
culture from scratch, I guess you get to make those decisions differently
if you like.

So, you can use unambiguous terms "essen" and "hertz" to refer to physical
time. Then continuous time-of-day uses the unit "day", corresponding
directly to the angular unit "circle". On Mars you'd plan your schedule
in centidays, during each of which the mean sun moves one centicircle
(3.6 arcdegrees) around the sky. A native would know that the duration
of the centiday is about 887.75 essens, in the same way that ey'd know
that the duration of the year is about 668.6 days.

Working from this relatively fresh start, free of sexegesimals, I think
you might still end up with a UTC-alike, but with a looser resemblance.
You could divide the calendar day into calendar centidays, and the
calendar centiday into (an integral number of) essens. Some centidays
are 887 essens long, and others are 888 essens long. There's an
obvious regular pattern: three centidays of 888 essens followed by one
of 887 essens, repeating throughout the day. The last centiday of the
day, exceptionally, would vary between 887 and 888 essens day by day.
Anti-leap-second activists would rail against the irregular ".99:887"


More information about the LEAPSECS mailing list