Ian Batten igb at batten.eu.org
Wed Jan 18 04:42:28 EST 2012

On 18 Jan 2012, at 0805, Clive D.W. Feather wrote:


> I am aware of case law where a difference of 8 seconds between clocks was

> relevant. That's the shortest interval I've seen so far in my searches.

And that's between clocks, not between a clock and some abstract reference. If MSF, GPS user interfaces, public NTP servers and the BBC pips (*) are ticking some timescale, but you have your watch set to some other timescale --- especially one which isn't maintained and you have to construct from assorted bulletins and offsets --- then your attempt to claim that everyone else is wrong and you're right is going to be difficult.

(*) Modulo the whole issue of DAB-delay, which throws another wild card into the proceedings.


> When I gave a paper on this at a law conference, it was suggest that -

> should this ever be a problem - lawyers would simply start adding "choice

> of time" clauses to contracts, just like they add "choice of governing law"

> clauses.

Precisely. It's hardly difficult. Retrospectively, GMT would acquire for practical purposes the meaning "clock on the wall time in the UK", which as an example Microsoft use in meeting notifications: they use "GMT" to include "UK time during DST" . The one in a million chance that a contract will contain a clause which is ambiguous between nu-UTC and old-UTC and GMT, and the difference (a minutes, perhaps, over a lifetime) will arise within the duration of the contract is something that can be litigated far cheaper than fixing all the contracts.

Any contract where a few seconds are going to matter is almost certainly going to need some provision to say which accurate time sources are involved anyway, and if those accurate time sources are now ticking nu-UTC (nuTC?) then the whole issue is moot. I've signed off on big projects that had a close link to UK civil time, and MSF receivers were specified as the reference: whatever they tick _is_ civil time, for the purposes of the contract.


More information about the LEAPSECS mailing list