Thanks very much for your analysis of the history of local time in Sweden. Here are some comments I have in response.
From: Ivan Nilsson [mailto:ivan@spica-it.com] Sent: Friday, April 13, 2001 6:46 PM
I suggest SEST rather than SST or ST because other time standards already use them (SE = the internation code for Sweden).
In similar situations, we've been using "xxT" to represent "xx Time" (where xx is the country code), so "SET" would be the logical abbreviation. This is an after-the-fact abbreviation, as two-letter country codes are a relatively recent invention, and if we ever find out what English-language abbreviation (if any) was used back then, we should replace "SET" with that abbreviation.
"September 30 1916 is considered to end one hour later". This is a most ambiguous statement.
It is unclear, but I take it to mean that 1916-09-30 had 25 hours rather than the usual 24, and the most natural way to model this in the tz database would be to move the clock backwards by one hour at 24:00 that day. It would make less sense to move the clock backwards at 01:00 the next day (unless perhaps the legal day in Sweden back then was not from 00:00 to 24:00 as it is now?).
the time 1:00:00 is in no way involved in the summertime changes in Sweden nowadays.
I think the tz database is OK on this point, as it specifies a transition time of 01:00 UTC for both transition times, as required by the EU. This is equivalent to the CET and CEST transition times that you mentioned. Thanks to your message, here is the revised Zone that I'll suggest in my next proposed patch. If you see any further trouble with it, please let me know. # Zone NAME GMTOFF RULES FORMAT [UNTIL] Zone Europe/Stockholm 1:12:12 - LMT 1879 Jan 1 1:00:14 - SET 1900 Jan 1 # Swedish Time 1:00 - CET 1916 May 14 23:00s 1:00 1:00 CEST 1916 Sep 30 23:00s 1:00 - CET 1980 1:00 EU CE%sT