Subject: Re: Europe/Bucharest and Europe/Chisinau: the post-1980 rules appear to rest on an inference from Bulgaria
Paul, Tim, all,
*The documents for Romania and Moldova, and a patch.* It applies to
tzdata2026c and compiles; it touches only Europe/Bucharest and
Europe/Chisinau (and, through Rule Romania, backzone's Europe/Tiraspol
for 1932-1939). Every line it changes rests on a document named below;
what it leaves is left because the documents agree with the file.
One correction to my note of 11 September first: the Bulgarian acts put
every autumn 1983-1995 on E-Eur's instant, the last Sunday of September
at 00:00. Only the hour differs in 1983-1990, and only the date in 1996.
Nothing below depends on Bulgaria either way.
1. Romania, 1891-1931: the 30E hour, not Bucharest mean time.
- The railways first. On the night of 18/19 September old style,
which is 1 October 1891, "orologiele garilor se vor da inainte cu
16 minute": the hour the Romanian railways would use "va diferi de
meridianul orasului Bucuresti cu 16 minute" (Timpul 1891-09-06 and
09, Lupta 09-08, Vointa Nationala 09-12 and 18, Curierul 09-13, all
old style). Sixteen minutes ahead of 1:44:24 is 2:00:00, the 30E
meridian - the hour of the Stockholm timetable conference of May
1891. Adevarul 1891-10-05 prints "Ora orasului" and "Ora garei, 16
minute inainte" side by side; the Iasi paper keeps "Ora garei este
cu 16 minute inainte" beside its timetable.
- Then the country. Royal Decree 3111 of 15 November 1891 old style,
Monitorul Oficial of 17 November o.s. (Timpul that day: "Monitorul
de azi publica decretul"): "cu incepere de la 1 Decembre acest an,
ca ora oficiala pentru toate serviciile publice din toata tara, ora
adoptata de caile ferate romane, stabilita de ultima conferenta
tinuta la Stockholm" (Timpul and Adevarul 11-17, Curierul and
Tribuna 11-20, Gazeta Transilvaniei 11-19). 1 December o.s. is
13 December 1891. Universul 11-15: the authorities' clocks "vor fi
aranjate dupa ora garei, adica cu 16 minute mai inainte";
Constitutionalul 12-05 o.s.: the clocks "indica de alaltaieri
incepand ora garei". The decree's number and date are also in the
Royal Romanian Geographical Society's bulletin of 1921, which
repeats the clause.
- It stayed. Revista Artileriei, March 1931 - before any 1931 act -
computes with "ora oficiala in Romania este cu 2 ore mai inainte
decat ora oficiala de la Greenwich ... Romania are ora fusului orar
al III-lea, adica ora fusului de 30 longitudine"; Ziarul Stiintelor
1926-03-16 speaks of "ora oficiala a Romaniei Mari"; and the
Chisinau papers of September 1931 call the hour used before the
May 1931 episode the Eastern European one (Raza 09-27: from
1 October "ora oficiala va fi ora intrebuintata inainte";
Basarabia 09-05: "noua ora oficiala obligatorie, aceia a Europei
orientale").
The file's "1891 Oct" line is the railway change under the wrong
name, and its "1:44:24 BMT" then runs to 24 July 1931: fifteen
minutes and thirty-six seconds, every hour, for forty years.
And 1931 itself. By a Council of Ministers decision of 9 May 1931
Romania took Central European time: the railways in the night of
Thursday 14/Friday 15 May (Dreptatea 05-16: "Noaptea trecuta, Joi
spre Vineri, [s-a introdus] la C.F.R. ora Europei Centrale in locul
orei orientale"; Aradul 05-17: "la ora 1 dupa miezul noptii acele
ceasornicelor C.F.R."), the state authorities and the towns the
night after, at 01:00 -> 00:00 (Viitorul 05-16: "autoritatile de
stat vor adopta ora Europei Centrale. In consecinta, la noapte cand
va fi ora 1 toate ceasornicile vor fi date cu o ora inapoi, adica la
12"; Dimineata 05-16; Universul 05-18: "ne-am dat ceasornicele inapoi
ieri"; Dimineata 05-17 on the confusion of that night; the Chisinau
weeklies Raza and Cuvant Moldovenesc 05-17: "toate ceasornicele au
fost date [inapoi]"). Back at 24:00 on 30 September -> 01:00 on
1 October, the instant of the law below (Dimineata, Curentul,
Universul, Cuventul 09-28, Neamul Romanesc, Epoca 09-29, Gazeta
Transilvaniei 09-30: "in noaptea de 30 Septembrie a.c. toate
ceasornicele vor fi date inainte cu [o ora]"; Dreptatea 10-01:
"astazi la orele 12 noaptea cu o ora inainte, revenindu-se astfel la
orariul [oriental]"; Cuventul 10-02: "Asta-noapte la orele 12 toate
ceasornicele au fost date inainte"; Patria, Cluj, 10-01; Opinia,
Iasi, 10-02: "la vechiul orar oriental"). The one witness the other
way is Dimineata's own calendar box, which printed sunrise 4:32 and
sunset 20:03 on 22 June and 6:04 / 18:14 on 23 September - the
Eastern European values to the minute, all summer - an almanac
column never adjusted, in a paper that reports moving its own clock
twice. 24 July is the promulgation of the "Lege pentru stabilirea
orei oficiale" (Royal Decree 2713, Monitorul Oficial 1931-07-27:
"Ora oficiala pentru cuprinsul Romaniei este ora fusului al doilea
orar la est de Greenwich, ora meridianului 30 grade", in force "de la
1 Octombrie a.c."): no clock moved that day, and the file's line is
wrong on both sides of it. The zone becomes
Zone Europe/Bucharest 1:44:24 - LMT 1891 Dec 13
2:00 - EET 1931 May 16 1:00
1:00 - CET 1931 Oct 1
2:00 Romania EE%sT 1997 Oct 26 1:00u
(the pre-1932 line written plain because a Romania-ruled line with
no rule inside its years does not compile). Europe/Chisinau copied
the Bucharest line for 1918-1931 and changed with the country in
1931; Bessarabia was under the decree from 1918, so its "1:44:24 -
BMT 1931 Jul 24" goes, the same two 1931 lines go in, and "2:00
Romania EE%sT 1940 Aug 15" follows them (the file's 1918 Feb 15 is a
date I have not read, kept as it is). Backzone's Europe/Tiraspol is
left alone: the left bank was not under the decree.
2. Romania, 1932-1939: one summer, not eight.
- The Council of Ministers' journal of May 1932, printed by eleven
dailies on 19-22 May: summer time "va curge de la 21 Mai ora 24 si
pana la 2 Octombrie [ora 1]" - from 24:00 on Saturday 21 May to
01:00 on Sunday 2 October. Argus 1932-05-23: "yesterday at 24:00 the
clocks were put forward an hour"; Lumea 1932-10-03: "at 01:00 all
clocks were put back an hour". The file's "May 21 0:00s" is a day
early; its 2 October return is right.
- In February 1933 the Council of Ministers abolished summer time
(Cuventul 1933-02-15, "Desfiintarea orei de vara"; Universul
1933-02-16, "in future the hour will remain the same permanently"),
and on 20 March 1933 the minister of public works and communications
circularised that the clock would not change: "Anul acesta nu se
mai schimba ora" (Dimineata, Curentul 03-22; Lupta, Opinia 03-21;
Erdelyi Hirlap 03-21, in Hungarian). The Bucharest dailies of
1934-1939 print every foreign change - Britain, France, Belgium,
Germany - and no Romanian one; Tempo 1939-04-19 says France's summer
time changed the difference between Bucharest and Paris. The seven
summers of 1933-1939 in the file, 182 days each, did not happen.
The rule lines for 1932-1939 become
Rule Romania 1932 only - May 21 24:00 1:00 S
Rule Romania 1932 only - Oct Sun>=1 0:00s 0 -
and nothing else. Chisinau and Tiraspol, on the Romanian rules until
August 1940, follow.
3. Romania, 1979-1997.
- Decree 190 of 1979-05-18 (Buletinul Oficial 49, 1979-05-19): in
1979 summer time "de la 27 mai ora zero, care devine ora unu, pina
la 30 septembrie ora unu, care devine ora zero"; from 1980 "de la
ora zero a primei zile de duminica din luna aprilie ... pina la ora
unu a ultimei zile de duminica din luna septembrie". Text on the
Chamber of Deputies' repertory:
https://www.cdep.ro/ords/pls/legis/legis_pck.htp_act_text?idt=10771
- Decree 415 of 1983-11-08 (Buletinul Oficial 89, 1983-11-10): from
1984 the start moves to "ora zero a ultimei duminici din luna
martie"; the end is unchanged. Same repertory, idt=10815.
- Ordinance 20 of 1997-08-18 (Monitorul Oficial 200, 1997-08-20,
p. 5): repeals both and sets the EU instants, "ora 3,00, care
devine ora 4,00" and "ora 4,00, care devine ora 3,00". Its first
transition is 26 October 1997. Law 211 of 1997-12-16 (Monitorul
Oficial 366) approves it without amendment. Both issues are on
Wikimedia Commons in the publisher's own PDF.
- The dailies announce every change at those hours (read in the OCR
full text of the Arcanum newspaper index, adt.arcanum.com; the
page images are behind its subscription):
1979 Scinteia Tineretului 05-26 ("ora 24 devine ora 1") and 09-29
("ora unu devine ora zero"); 1980 Flacara Iasului 04-05, Inainte
09-27; 1981 Romania Libera 04-01 ("4/5 aprilie ora zero devine ora
unu - prima duminica a lui aprilie"), Scinteia 09-26; 1982 Romania
Libera 04-03, Cronica Satmareana 09-25; 1983 Romania Libera 04-02
("3 aprilie ora zero devine ora unu"), ten dailies 09-24; 1984
seventeen dailies 03-22/25, Informatia Bucurestiului 09-29; 1985 ten
dailies 09-27/28; 1986 Flacara Iasului 09-27; 1987 Romania Libera
03-28, Magazin 09-26; 1988 Zori Noi 03-26, Magazin 09-24; 1989
Tribuna Sibiului 03-23, twelve dailies 09-22/24; 1990 Cuvintul
Nou 03-23, Romania Libera 09-29; 1991 Gazeta de Nord-Vest 03-28,
Magazin 09-28; 1992 Cuvintul Liber 03-28; 1993 Cuvintul Nou 03-27,
Agenda 09-25; 1994 Evenimentul Zilei 03-26, Romania Libera 09-23;
1995 thirteen dailies 03-24/25, ten dailies 09-22/23; 1996
Evenimentul Zilei 03-30, Ziua 09-28 ("Din 29 septembrie inceteaza
aplicarea orei de vara"; trains "dupa ora 1,00 care devine ora
0,00"); 1997 Evenimentul Zilei 03-28 ("Ora 0,00 devine ora 1,00"),
Ziua 10-24 ("26 octombrie ora 04,00 devine ora 03,00 - Guvernul a
decis"). The papers of 1996-10-28 report the Western European
change of 27 October as foreign news; none announces a Romanian
change that weekend.
The file's C-Eur years (1981-1990), its E-Eur years (1994-1996) and
its EU spring of 1997 have no Romanian source; the decrees give
Rule Romania 1979 only - May 27 0:00 1:00 S
Rule Romania 1979 1996 - Sep lastSun 1:00 0 -
Rule Romania 1980 1983 - Apr Sun>=1 0:00 1:00 S
Rule Romania 1984 1997 - Mar lastSun 0:00 1:00 S
and the one zone line of section 1 in place of the four lines from
1981 to 1997.
Against tzdata2026c this moves 31,810 hours of Bucharest time by an
hour: 24 in 1932, 4,368 in each of 1933-1939, 1 in 1979, 1 in 1980,
168 in each of 1981-1983 (the April springs), 4 in each of
1984-1990, 1 in 1994, 1 in 1995, 671 in 1996 (the October the decree
did not have) and 3 in the spring of 1997 - and every hour from
13 December 1891 to 24 July 1931 by fifteen minutes and thirty-six
seconds (347,208 hours), which I count apart: a quarter-hour is not
an hour.
4. Moldova, 1917, 1941-1944 and 1981-1996.
- Kishinev, 1917. Both Kishinev dailies of 2 July 1917 o.s. report
the clock "vo vsei Rossii ... na chas vpered" from 1 July o.s.
(Svobodnaya Bessarabiya, Bessarabskaya Zhizn); the provincial
gazette, Bessarabskiya Gubernskiya Vedomosti 1917-09-22 o.s.,
prints the Provisional Government's resolution of 17 August;
Svobodnaya Bessarabiya 09-19: extended "s 1-go oktyabrya vpred do
osobogo rasporyazheniya"; and 11-24: "Perevod strelki chasov na
zimnee vremya posleduet lish 1 yanvarya". Europe/Chisinau has no
summer time in 1917. Added: forward at 23:00 on 1917-07-14 (1 July
o.s., the decree's hour as the file types it for Moscow), back at
00:00 on 1918-01-14 (1 January o.s., as announced). The Soviet
decree of 1917-12-22 o.s. set the night of 27/28 December o.s. for
Russia, four days earlier; the Kishinev papers in the index run to
30 December o.s. and carry neither night, so the announced date
stands and the alternative is named in the comment. Written with
%z, no abbreviation being attested:
1:55 - CMT 1917 Jul 14 23:00
1:55 1:00 %z 1918 Jan 14
1:55 - CMT 1918 Feb 15
- Chisinau, 1941-1944. The file puts Bessarabia under Romanian
administration on Central European time with the German rules,
the line it also uses for Minsk and Riga under the German civil
administration - which Bessarabia was not under. The decree-law of
1941-09-03 organising Bessarabia and Bukovina, printed in full by
the Bucharest dailies of 09-05 (Timpul, Viata, Universul,
Curentul), kept in force "legile ce se aplicau la 28 [iunie
1940]" - the 1931 law of the official hour, "obligatorie pentru
tot cuprinsul Romaniei", among them. Romania itself stayed on the
Eastern European hour: Timpul 1942-12-05 on the neighbours "cari
in primavara adoptasera ora Europei orientale, ora noastra"; the
one Central European order it ever issued, for the night of
1943-10-04 (Actiunea, Galati, 10-03: "Luni noaptea intra in
vigoare ora Europei Centrale ... ordinul Ministerului Lucrarilor
Publice si al Comunicatiilor"), cancelled the next day ("S-a
anulat introducerea orei Europei Centrale in Romania"). The
Chisinau weeklies around three of the four nights the C-Eur rule
would have moved the clock are in the index - Raza 1942-11-01 and
11-15; Raza, Cuvant Moldovenesc, Basarabia Literara 1943-03-28 to
04-11 and 1943-10-03 to 10-17 - and carry no clock notice; "ora
Europei" never occurs in a Chisinau title of those years; the same
papers print the railway's times unconverted (Raza 1943-04-18 and
05-16, the Bucharest-Chisinau railcar and the Bessarabian
timetable; Cuvant Moldovenesc 1943-12-19), where in 1931, the one
summer the two hours differed, the press had said "o ora la gara,
alta in oras". So
2:00 - EET 1944 Aug 24
in place of "1:00 C-Eur CE%sT 1944 Aug 24". The dates 1941-07-17
and 1944-08-24 are the file's and stand; the offsets on both sides
of each are unchanged (Moscow time before and after, UT+2 in July
and August either way). What moves is the two winters, 1942-11-02
to 1943-03-29 and 1943-10-04 to 1944-04-03.
- Resolution 132 of the Council of Ministers of the Moldavian SSR,
1990-05-04 (Vestile 1990 no. 5, art. 83) - the act Stepan Golosunov
cited in 2016 - read in the Ministry of Justice register
(lex.justice.md id 298782, as archived on 2021-11-07): clocks back
at 2:00 on 6 May 1990, then "timpul de vara se suspendeaza in
ultima duminica a lunii septembrie la orele 3 ... si se introduce
in ultima duminica a lunii martie la orele 2, ... anual".
- Government Decision 79 of 2026-02-25 (Monitorul Oficial 96-99,
2026-02-26, p. 115), point 7, repeals it. Until then it was the
only Moldovan act on official time, unamended; the ministry's
November 2025 note says so in terms.
- The Chisinau press: Moldova Socialista 1990-05-05 (the resolution
reprinted, "la ora 2 ... cu o ora in urma"); Vecherniy Kishinev
and Sovetskaya Moldaviya 1990-09-28 ("v tri chasa", 3:00, one hour
back); Sovetskaya Moldaviya 1991-03-27 ("v 2 chasa vvoditsya letnee
vremya"); Vecherniy Kishinev 1994-03-26 ("nachinaya s 2.00");
Nezavisimaya Moldova 1996-03-29 ("v 2 chasa nochi"). And
Nezavisimaya Moldova 1996-10-01: the return "this year a month
later ... on the last Sunday of October, by putting the clocks one
hour back", per Infotag - so the file's October 1996 stands.
The Soviet transitions of 1981-1989 are as the same papers
announced them - "at 0:00 on 1 April" (1983, 1984), "winter time
comes at 3:00 at night on 30 September" (Molodezh Moldovy
1984-09-27, the first year of the new Soviet rule), "at 2:00"
(1985-1989) - and hold.
The file applies E-Eur's 0:00 for 1992-1996, between two spans that
already carry the resolution's 2:00 and 3:00 (1990-1991 through Rule
Russia, 1997-2021 through Rule Moldova). The resolution gives
Rule Moldova 1992 2021 - Mar lastSun 2:00 1:00 S
Rule Moldova 1992 1995 - Sep lastSun 3:00 0 -
Rule Moldova 1996 2021 - Oct lastSun 3:00 0 -
in place of the 1997-2021 lines; the zone lines "1:44:24 - BMT 1931
Jul 24" (section 1) and "2:00 E-Eur EE%sT 1997" go. 30,625 hours of
Chisinau time move by an hour: 24 in 1932, 4,368 in each of
1933-1939 through the Romanian rules, and 2 in each spring and 3 in
each autumn of 1992-1996; and every hour from 15 February 1918 to
24 July 1931 by fifteen minutes and thirty-six seconds (117,768
hours), counted apart.
The 1990 spring, which my draft had not read: Sovetskaya Moldaviya
1990-03-10 prints the all-Union notice ("25 marta nachinaetsya
period deystviya letnego vremeni"), Molodezh Moldovy 1990-03-29
reckons "dekretnoe i letnee vremya - doloy dva chasa", and the
resolution's 6 May setback "cu o ora in urma" only makes sense from
Moscow summer time. The file's Russia rule for 25 March 1990 holds.
5. Not changed, and why.
- Chisinau 1940 Aug 15 and 1944 Aug 24: the offsets hold, the days
are Shanks's. For 1940 the press brackets it: on 19 July the
occupied province's clock was still an hour behind Moscow's ("nur
bis 8 Uhr abends Moskauer Zeit, 7 Uhr Ortszeit", Siebenburgisch-
Deutsches Tageblatt 1940-07-19), Estonia, Latvia and Lithuania went
to Moscow time on 6, 9 and about 10 August by Soviet order (Pester
Lloyd 08-04, Wiener Neueste Nachrichten 08-09, Curentul 08-10), and
on 22 October TASS reckons a Kishinev earthquake at "9.36 Uhr
Moskauer Zeit" (Bukarester Tageblatt, Pester Lloyd 10-23) - no
dispatch names the day for Bessarabia. For 1944 nothing: the index
holds no Chisinau paper between 24 August and 31 December 1944.
Both left as they are. Chisinau 1918 Feb 15: the offset is corrected
above, the date is not read; 1880 is a 20-second rounding on a bare
year, left as it is.
- The Romanian springs of 1985 and 1986 and the Moldovan autumns of
1992-1996 have no notice with an hour in the index; they follow
the standing act, as every neighbouring year does.
6. Romania since 1997, checked rather than assumed.
The ordinance is in force unamended: the Chamber of Deputies'
repertory card of O.G. 20/1997, read 2026-09-14, says "In vigoare:
da", approved by Law 211/1997, with no modifying or repealing act;
the art. 2 power to follow a change in EU practice by government
decision was never needed, Directive 2000/84/EC being unchanged
since 2001. And every season is read: a notice naming both "ora
3,00" and "ora 4,00" in at least one daily for each of the 55
seasons from March 1998 to March 2025 in the Arcanum index (3 to 45
hits a season), HotNews 2025-10-26 ("Ora 4.00 a devenit ora 3.00"),
Agerpres 2026-03-28 ("ora 03:00 devine 04:00", and 25 October 2026
for the return). The EU rule the file applies from 1997 Oct 26 is
what happened, every March and October since.
The chronology - every moment in both zones from the first line to
2026, what the documents say and the grade of what we hold - with the
acts and the transcriptions is at https://oco.bg/tz/romania and
https://oco.bg/tz/moldova, and, with the state documents and their
hashes, in the deposit https://doi.org/10.5281/zenodo.22695588
(version 4, beside the Bulgarian plates).
Ivan Petrov Ivanov
Sofia, Bulgaria
https://astrocogy.com, https://oco.bg,
ORCID 0009-0007-6123-351X
On Fri, 11 Sept 2026 at 13:18, Ivan Petrov Ivanov <[email protected]>
wrote:
> Tim, all,
>
> You asked on 1 September whether I would formulate similar archival
> requests for Romania and Moldova. This is my reply, and it is a notice
> of intent rather than a proposal: I have no Romanian evidence yet, and I
> would rather bring documents than speculation.
>
> I am posting it now, separately from the Bulgarian submission, because
> the reason to look at Romania comes directly out of the Bulgarian result
> and I would like that connection on the record before I commit anything
> on it.
>
> Reading the europe file with the Sofia material in hand, the Romanian
> block appears to lean on Bulgaria by name:
>
> # From Paul Eggert (1999-10-07):
> # ... For lack of better info,
> # assume that Romania switched to EU rules in 1997,
> # the same year as Bulgaria.
>
> Two things I want to be careful about, because the inference cuts in a
> narrower way than it first appears.
>
> What the Bulgarian archives do NOT do is move the year. Bulgaria did
> switch to the EU rules in 1997, by Decree 94 of 13 March 1997, which I
> have photographed. So the "same year as Bulgaria" reasoning survives on
> the date of the switch as such.
>
> What they DO undermine is the assumption underneath it. Bulgaria has
> been kept in the database on C-Eur from 1983 to 1991 and on E-Eur from
> 1991 to 1997 - and the primary sources say Bulgaria was on neither. It
> ran its own rule the entire time, changing at midnight, last Sunday of
> March and last Saturday of September, until the 1997 decree. The instant
> the Bulgarian block turns out not to have the shape the file assumes,
> "Romania looks like Bulgaria" stops being evidence for anything about
> Romania's shape.
>
> Two more things I notice in the file, and I would be glad to be told I
> am reading them wrongly:
>
> * Zone Europe/Bucharest runs on C-Eur from 1981 to 1991 and on E-Eur
> from
> 1994 to 1997, and the Rule Romania lines stop at 1993. As far as I can
> tell no Romanian primary source is cited for any transition in that
> period.
>
> * For Moldova the substantive citation for the period is the 1990
> government
> act relayed in 2016; I do not find a Moldovan legal source cited for 1991
> through 1996.
>
> I am not proposing any change to either zone at this point.
>
> What I would like to do, if it is of use to you, is repeat the procedure
> that yielded the Bulgarian material: read the official gazette for the
> period (Monitorul Oficial al Romaniei, and its Moldovan counterpart)
> alongside the national dailies for the transition weekends, photograph
> the pages, read them with two independent readers and a third checking
> against the image, and publish the plates so each transcription can be
> verified against the photograph it came from.
>
> Three questions before I begin:
>
> 1. Is this useful, or is the Romanian record already settled from a
> source I
> have not seen in the file?
>
> 2. You mentioned preferring one holistic patch for the region over
> piecemeal
> fixes. Should the scope be Romania and Moldova, or does it pay to
> include the other countries that stand in the same rule groups for the
> same
> period?
>
> 3. Is there a period or a transition you would want checked ahead of the
> rest? I
> would rather spend the first archive visit on whatever is weakest in
> your
> view than on whatever is easiest for me.
>
> The Bulgarian plates and transcriptions are at
> https://astrocogy.com/tz/bulgaria, and the same package is deposited
> with a DOI, as a specimen of what the output would look like:
>
> https://doi.org/10.5281/zenodo.22695589
>
> Ivan Petrov Ivanov OCO, Sofia https://oco.bg, https://astrocogy.com
>
europe-bucharest-chisinau-1891-1997.patch
Description: Binary data
