Hello, I would like to report a likely error in Europe/Sofia's 1996 data, found while verifying tzdb's Bulgarian history against primary legal sources, and to flag a related open question for 1983-1990. This is a preliminary report: we have initiated formal archive requests to establish what clocks actually did, and will follow up with documentary evidence either way.
## Autumn 1996 tzdb ends Bulgarian DST on 1996-10-27 (Europe/Sofia is on the shared E-Eur rules for this period, with no Bulgaria-specific citation in the europe file). We can find no Bulgarian legal basis for an October end in 1996: * The DST regime in force through 1996 was established by Council of Ministers Decree No. 6 of 26 January 1979 (State Gazette / Държавен вестник No. 12/1979), which provides for summer time "until 30 September" each year. * A last-Sunday-of-October end appears in Bulgarian law for the first time in Decree No. 94 of 13 March 1997 (State Gazette No. 24/1997). Its Article 2 also fixes the autumn change at 04:00 local time. The full text is in the government's legal-information system PRIS (pris.government.bg). The decree's repeal clause enumerates the entire prior regime: Decree 6/1979 and item 12 of Directive (Разпореждане) No. 15 of 11 March 1980, the latter marked as never published. * We find no 1996 act changing the end date. PRIS records Council of Ministers acts since 1990 and contains none on clock time before the 1997 decree. If the law was followed, summer time in Bulgaria ended in late September 1996, and Europe/Sofia is one hour ahead of reality (+03 instead of +02) for roughly 1996-09-29 through 1996-10-27. The report tzdb cites for Bulgaria (Plamen Simenov via Steffen Thorsen, 1999-09-09) describes the 1997 decree's rules and does not address autumn 1996. Since law and practice can differ, before proposing a patch we have initiated: - a request to the Bulgarian National Library for scans of five national dailies for 27-30 September and 25-28 October 1996 (the clock-change notices published to the public); - freedom-of-information requests to the Council of Ministers and the State Archives Agency for the unpublished Directive 15/1980 and all clock-time acts 1979-1997; - inquiries with the Bulgarian Institute of Metrology and the BTA news-agency archive. We will report what the archives show, in either direction. ## Why this matters to us We build astrological software, which makes a historical UTC offset an input to a computation rather than a display detail. The quantity most sensitive to it is the ascendant, the ecliptic degree rising at the given moment, which completes a full circuit every day. Measured at Sofia's latitude (42.698 N) for 1 October 1996, sampled every minute of the day: a one-hour error in the offset moves the ascendant by between 11.6 and 27.1 degrees, and it lands in a different one of the twelve 30-degree sectors in exactly half of the minutes of that day. The 50% is not a coincidence; a one-hour arc averages 15 degrees against 30-degree sectors. So for a Bulgarian birth between 29 September and 27 October 1996, a wrong offset does not produce a slightly wrong result. Half the time it produces a categorically different one, and there is nothing in the output to indicate which. That is also why we are asking about 1983-1990, where the published record looks incomplete rather than wrong, and why we would rather see a documented gap than a confident value. The same sensitivity applies well outside our field: anything that reconstructs past local civil time from records inherits whatever the database says about a date nobody can re-observe. We are not asking for a change that happens to suit us. If the archives show that clocks in practice followed the October date despite the decree, we will report that, with the same evidence and in the same detail. ## 1983-1990 transition times For 1983-1990 tzdb has Sofia on C-Eur rules (02:00s/03:00d transitions), apparently ex-Shanks and uncited; Bulgarian retrospective sources consistently describe midnight transitions for the whole 1979-1996 period, and tzdb itself moves Sofia to midnight rules (E-Eur) in 1991 with no documented reason. Our archive requests cover all sixteen transition nights of 1983-1990 (the last Sundays of March and September) in the national dailies of the period, asking specifically for the announced hour. If any list members have contemporaneous Bulgarian sources (newspapers, broadcast listings) for these transitions, we would be glad to hear of them. For completeness: the same review confirmed tzdb is correct on the 04:00 autumn hour for 1997-1998, the decree text matches tzdb exactly, against several Bulgarian secondary sources that claim 03:00. Best regards, Ivan Petrov Ivanov Sofia, Bulgaria
