[tz] 1948-1951 Asia/Tokyo DST information

Paul Eggert eggert at cs.ucla.edu
Sat Sep 8 15:49:11 UTC 2018

Phake Nick wrote:
> Not really, it sat 25:00:00 at Saturday would become 00:00:00 on Sunday It
> used the tine from Saturday to refer to transition that occur at Sunday
> 00:00.

The cited Japanese law 
says, "第二条 四月の第一土曜日の翌日(日曜日)は二十三時間をもつて一日とし、九月の第二土曜日は二十五時間をもつて一日とする", 
which Google translates as "Article 2. The day following the first Saturday of 
April (Sunday) shall be one day with twenty-three hours and the second Saturday 
of September shall be twenty-five hours a day." This law is compatible with what 
tzdb specifies, which is that the clock jumps back from 24:00 to 23:00 on 
Saturday, which means Saturday indeed has 25 hours, the last of which has 
timestamps duplicated from the next-to-last.

As I understand it, you're saying that the practice in Japan (not specified in 
the law) was to call these timestamps 24:00 through 24:59 instead. A nice 
property of this practice would be that timestamps would not be ambiguous. 
During the period in question (1948-1951) it's unlikely that many physical 
clocks showed the times 24:00 through 24:59, so the question is whether people 
adjusted their 12-hour clocks from 1 o'clock (representing 25:00 Saturday) to 12 
o'clock (representing 00:00 Sunday), or from 12 o'clock (representing 24:00 
Saturday) to 11 o'clock (representing 23:00 Saturday). Both actions are 
consistent with the cited law, and so far we have no evidence what people 
actually did. Since tzdb cannot implement what you're suggesting, the answer to 
this question won't affect what we put into tzdb proper, though of course it 
should affect the commentary.

One other thing. If your suggestion is that the practice was to simply number 
the timestamps in a day from 00:00 to the end of the day regardless of the day's 
length, then for consistency the timestamps in a spring-forward Sunday (i.e., 
the Sunday after a spring-forward transition) should have been 00:00 through 
22:59. That is, in spring people in Japan would have moved their 12-hour clocks 
from 11 o'clock to 12 o'clock at 23:00 Sunday. I'm skeptical that this occurred, 
as it's typical to make the change Sunday morning in order to minimize 
disruption to the workweek starting Monday, so it's far more likely that people 
would have moved their clocks from 12 o'clock to 1 o'clock at 00:00 Sunday, as 
tzdb has it now. Surely a "just number the timestamps in a day consecutively" 
rule cannot be what was used in Japan; if there was a rule about 24:00 through 
24:59 it was a special-case rule that appled only to fall-back Saturdays.

More information about the tz mailing list