[LEAPSECS] Straw men

Ian Batten igb at batten.eu.org
Tue Jan 10 12:06:17 EST 2012



On 10 Jan 2012, at 1528, Tony Finch wrote:


> Rob Seaman <seaman at noao.edu> wrote:

>> Warner Losh wrote:

>>

>>> It is only one possible definition, not the only one. That makes it a

>>> belief, not a mathematical identity.

>>

>> Alternate definition?

>

> It used to be local apparent solar time. Then local mean solar time. Then

> standard mean solar time. Then standard time with daylight saving. The

> definition of civil time is evidently not fixed.


And mostly in relatively recent history, too. The switch from local mean solar to standard (usually "mean solar in the capital city of the political entity we are part of") mean solar is a product of the railways, and daylight savings wasn't implemented until the first world war.

Let's put my cards on the table. UK Civil Time is not Mean Solar Time in any formal way.

You cannot set up a bijection between successive 1s timestamps of UTC and successive valid 1s timestamps of UK Civil Time, because the civil timestamps between 01:00:00 and 02:00:00 on the fourth Sunday in October each map to two distinct UTC timestamps, as they are repeated in the sequence of UK Civil timestamps. Therefore UK Civil time is not mean solar time.

As another way of making the same argument, the set of valid, distinct, 1s resolution UK Civil time stamps between 00:00:00 Year X and 00:00:00 Year X+1 has a different cardinality to the set of UTC timestamps over the same range. That is because 3600 of the timestamps in that range are missing (00:01:00..00:01:59 last Sunday of March don't exist). Therefore UK Civil Time is not mean solar time.

Any Julian date of the form d=[1, 365], h=[0, 23], m=[0, 59], s=[0, 59] is a valid UTC or mean solar time timestamp which uniquely defined a single one-second interval. This is not true for UK Civil Time, where a Julian date of that form defines 0, 1 or 2 one-second intervals. Therefore UK Civil Time is not mean solar time.

Now I'm sure some people will argue that civil time with DST incorporated is in some way "really" derived from civil time without DST, which is somehow more "real". My argument collapses if UK Civil Time is defined to include a zone specifier to indicate whether DST is in force. With that flag (perhaps "S" or "W" appended for Summer and Winter time), there is a bijection between UTC and civil time, and the cardinality of the set of distinct timestamps is the same (the unique Julian date argument needs slightly more complex tweaking, but also fails). But unfortunately, UK civil time does not include a DST indicator. UK civil time has a genuine discontinuity of civil time in the spring, and a genuine repetition of time stamps in the autumn. UK Civil time doesn't have a means to distinguish 01:30:00 from the 01:30:00 that follows an hour later during the autumn DST event. Civil time renders 01:30:00 of the fourth Sunday in March undefined: it just doesn't exist, so that 00:30:00 + 1 hour is 02:30:00 and that particular Sunday only contains 82800 seconds, while the fourth Sunday of October contains 90000 seconds. You just have to know this.

You can argue that this civil timescale has absurdities, ambiguities and inconsistencies. Brother, pass me the hymn sheet and I'll sing descant. You can argue that the only sane way to implement it is to tick UTC and add the zone offset subsequently, storing all timestamps in UTC. Again, you hum the melody, I'll harmonise where I can. You can argue that DST is a con trick on simple folk which has outlived its usefulness, that UK civil time should include a zone specifier in its formal usage, and that although it has interesting facets, the "Redux" cut of Apocalypse Now in fact has some weaknesses and longeurs that are not present in the original theatrical release. All of these things will enjoy my whole-hearted support, or at least a fair hearing. None of this gets away from the fact that lawfully defined UK civil time _does_ contain these ambiguities and inconsistencies and, let's be honest, absurdities, does not contain a zone identifier, and can only be converted into a continuous timescale by knowledge of the DST rules prevailing in the relevant year. It is simply not mean solar time.

ian







More information about the LEAPSECS mailing list