[time-nuts] UTC and leap seconds

[Tom Van Baak]

> I think the answer is in your previous post, that is the year is more stable 
> than the day as compared to the same clock.
> 
> And this measurement very likely has been actually made.
> 
> Antonio I8IOV

Right.

Defining a second as 1 / 86400 of the length of a mean
solar day is also problematic because you first have to
calculate the mean. How many days or years do you
take the average for?

For a while the second was re-defined as a fraction of
a year instead of a fraction of a day.

Fortunately the definition quickly evolved into a multiple
of a cesium transition (the 9192631770 number), an
apparatus a national lab could build or buy from hp, etc.

The problem was that telling someone that a second is
1 / 31556925.9747 of the tropical year 1900 is not an
easy thing to calibrate a quartz clock by. Hard to get
that year back in your lab in order to re-measure it.

The goals of a units standard in metrology are not just
extreme accuracy and stability. It is also desirable for
a standard to be reproducible, reliable, and commercially
manufacturable; perhaps also affordable and portable.

/tvb