[LEAPSECS] The definition of a day

Rob Seaman seaman at noao.edu
Thu Jan 29 17:21:13 EST 2015

Yes, both solar time and atomic time are based in physical reality.  That’s the point as made innumerable times on this list.  Civil time is derived from two distinct clocks.  Solar time defines the day.  Atomic time defines the second (fundamentally it is a frequency standard).  There are inherent complexities in allowing the two to live together in harmony.

Consider King Canute trying to command the tide to halt:


By all means discuss different ways to accommodate both.*  It does not seem radical to suggest that a systems engineering process will produce better results than politics.  But seeking to advance one clock at the expense of the other will fail just like Canute, and we should be as graceful to accept this fact as Henry of Huntingdon’s description of Canute.

Rob Seaman
National Optical Astronomy Observatory
* For instance, define a new leap-less timescale with a different name (as discussed in Torino in 2003), or projects with requirements for unsegmented timescales should just use TAI as has also been recommended, or use GPS time.  Redefining UTC will not provide additional access to atomic time, rather it simply removes access to solar time.

> On Jan 29, 2015, at 2:47 PM, Warner Losh <imp at bsdimp.com> wrote:
>> On Jan 29, 2015, at 7:25 AM, Peter Vince <petervince1952 at gmail.com> wrote:
>>     There have been some strong views expressed that a "day" has to mean a solar day, i.e. midnight-to-midnight (or midday-to-midday).  That is fine, but we seem to be arguing about what precision that is defined to.  One camp is arguing for a smooth atomic time that slowly drifts away from mean solar time by minutes over a period of centuries, while the other wants us to stick rigidly to mean solar time, inserting an extra second (no, sorry - "resetting the clocks by a second") every so often, which currently causes some disruption and confusion.
> The debate is should time be uniform in duration, or track the spinning globe. Both have antecedents to the modern age, and the difference is small enough that we weren’t able to measure it until the past generation or two.
> So the contention is should we follow the more precise time that we’ve discovered, or should we follow the age old custom of days. The delta between the two is tiny, and not likely to make much of a difference to people’s day-to-day lives for hundreds, if not thousands of years. Do we accumulate the error, refine our laws and adjust the folks that depend on broadcast time being an approximation of UT1 and go to a uniform accumulation of seconds? Or do we keep the status quo and deal with an increasingly difficult insertion of seconds to keep them in lock step?
> These are actually both quite reasonable points of views, with different costs associated with them. Some of these costs are actual dollars for people dealing with leap second planning and leap second execution errors vs costs of adding in additional correction factors to existing systems (some of which may be quite expensive or difficult to do). Some of these costs are philosophical about which “physical reality” you think is primal: the spinning earth, or the uniform duration of a second with regular minutes, hours, etc from there. Both have basis in physical reality, and each has its own set of logical step to go from physical reality to what the best approach to time is.
> One may argue about which one is best, and why (I present the archives of this list for proof of this statement). Both both are based in physical reality and differ only in philosophy and where the burden of costs are placed.
> Warner
> _______________________________________________
> LEAPSECS mailing list
> LEAPSECS at leapsecond.com
> https://pairlist6.pair.net/mailman/listinfo/leapsecs

More information about the LEAPSECS mailing list