On 7/28/22 16:07, Steve Allen via tz wrote:
That comes from the tzdata asia file for Phu Lien Observatory. That entry gives Paris Meridian as 2 deg 20' 14.03" E with no citation.
Every historical value of longitude needs an accompanying citation.
Thanks. Sorry to have induced you to go to all that work in digging up historical measurements of the Paris Meridian, as the current datum doesn't depend on the exact value. It's simpler, I think, to remove that longitude from the comments as that will save us the trouble of picking and citing a longitude. Done in the attached patch.
Please do not attempt sub-millisecond time offsets for astronomical time.
Sure, but if our understanding is correct, the cool thing about the GMT offset in Vietnam from 1906 to 1911 is that it wasn't astronomical time. That is, what we're calling Paris Mean Time was defined by French law to be 9 minutes 21 seconds east of GMT; and what we're calling Phù Liễn Mean Time was defined by law to be 104° 17′ 17″ east of Paris Mean Time. Neither of these numbers are astronomical; they're both legal numbers, and so their sum is a legal number, and has infinite precision (since French lawyers don't admit to any imprecision :-). That being said, this is currently academic since tzcode isn't using the subsecond values, and I take your point that the actual offset was not accurate to 9 places even with the best technology of the time. So I installed the attached further proposed patch, which rounds to the nearest centisecond (and fixes one other, more important thing), and I hope that is good enough.