There is a national standard for writing numbers and dates in Chinese publications. It was revised in 2011. A great deal of the advice you will find about it — including the style rule that governs Chinese Wikipedia — is quoting the 1995 edition it replaced.
The tell is one line, and it will resolve most disputes about this in about ten seconds.
The one-line check
Both editions cite a companion standard for date and time formats. The 1995 edition cites the 1994 version of it. The 2011 edition cites the 2005 version.
So if a page presenting itself as the current standard lists its normative reference as GB/T 7408—94, you are reading the 1995 text no matter what number is at the top. If it cites GB/T 7408—2005, you are reading the 2011 text.
That single check catches three separate errors at once, because all three come from the same withdrawn document.
What the 2011 revision actually changed
The first is the big one. The 1995 standard said that wherever Arabic numerals could be used appropriately, they should be. The 2011 foreword explicitly withdraws that: it records that the previous standard leaned toward Arabic numerals and states that the new one no longer emphasises that leaning. Where neither form is clearer, both are available.
The second is the scope of Chinese numerals for non-Gregorian dates. The 1995 text framed this as applying to Chinese historical reckoning "of the Qing and before". The 2011 text keys instead on the reckoning system — sexagenary years, lunar dates, dynastic era names and other traditionally Chinese-numeral non-Gregorian systems — with no date cutoff at all.
The change matters because the old framing was wrong in both directions. A rule for dates "of the Qing and before" does not cover the standard's own examples, which include a Qing date, a Tibetan calendar date, and a Japanese era date.
The third is a clause about whether quotation marks depend on how famous an event is. That was dropped in 2011 and survives only in reprints of the older text.
Where the old text is being served as the new one
The error is not confined to stale blogs. A university language-work page, updated in June 2024, carries the 1995 body under the title and number of the 2011 standard. Its normative reference is GB/T 7408—94, its scope paragraph is the 1995 wording, and its clause numbering follows the 1995 arrangement.
The same 1995 body appears on other university pages, and the mislabelling — the 1995 title glued to the 2011 number — recurs across journal author-guidance pages.
The consequence is a generation of house styles that over-prescribe Arabic numerals and repeat the "before the Qing" rule, sourced in good faith to a standard number that is current and a text that is not.
The interpunct rule, which almost everyone gets wrong in one direction or the other
When an event or festival is named by its month and day, the question is whether a separator goes between them. The standard's answer is conditional, and the condition is unusual.
Written in Chinese numerals, the separator is used only when the month is the first, the eleventh or the twelfth. For every other month there is none. The companion punctuation standard states the same three-month rule independently and gives the September 18th Incident and the May Fourth Movement as its own examples of the no-separator case.
The condition prevents ambiguity. With three characters and no separator, a reader cannot always tell if the first character is a month on its own or the start of a two-character month. The rule applies exactly to the months where that ambiguity exists.
Chinese Wikipedia's manual of style states the rule with four months, adding the tenth. The tenth month is not in either standard. Under Wikipedia's version, the National Day of the first of October acquires a separator that the standard does not give it.
The same page offers a different rationale as well — that single-digit numbers do not need the separator. That is not the rule either, and it gives wrong answers in both directions: it would add a separator to the September 18th Incident, which the punctuation standard prints without one, and to the May Thirtieth Movement, which also takes none.
The separator is used incorrectly for the September 18th Incident widely enough that Chinese state media has published corrective pieces about it in consecutive years. The one legitimate survival is a proper name: the museum commemorating it does carry the separator in its registered title, which is an exception belonging to that institution rather than a general licence.
Writing a year in Chinese numerals
Chinese has two characters for zero. The standard distinguishes them by function: in a number that identifies something, use 〇; in a number that measures something, use 零.
A year identifies rather than measures, so 2012 is written 二〇一二. The standard makes the point by naming the wrong version explicitly — it says not to write 二零一二.
Chinese Wikipedia's style page presents both as acceptable. The standard presents one of them as its counter-example.
Leading zeros: both confident answers are wrong
Is 2026年09月01日 correct, or should it be 2026年9月1日?
The standard does not settle it, and anyone who claims it does is overreading.
The relevant clause says that when the month and day are single digits, one may prefix a zero. The verb is permissive. So padding is never required — that much is clear.
Whether it is permitted in the year-month-day form is where it goes quiet. The rule's wording is general enough to license it, but both of its examples are in the hyphenated form, and every year-month-day example elsewhere in the standard is unpadded. So the padded form is defensible on the wording and unattested in the text.
Both popular absolutes are therefore wrong. "Chinese dates require leading zeros" is false. "The standard forbids them" is also false — the prohibition people are reaching for exists, but it lives in a different standard governing the format of official government documents, not publications generally.
The standard contradicts the standard it cites, in one sentence
The clause permitting hyphenated dates authorises them by reference to the extended format of the date-and-time standard, and then gives its examples as 2008-8-8 and 1997-7-1.
The extended format is fixed width. Two digits for the month, two for the day. Those examples are not valid in the format the same sentence invokes.
So 2008-8-8 is conformant with the publications standard and non-conformant with the format it cites. If you are writing anything a machine will parse, pad it — which is also why that permissive leading-zero rule exists at all.
The problem is deeper. The date-and-time standard cited in the rule was itself withdrawn on 1 April 2024 and replaced. The publications standard's only normative reference now points at a withdrawn document, and the clause number it quotes no longer maps to anything in the replacement. The publications standard passed its periodic review in September 2025 and was confirmed as remaining in force, unamended, after its sole reference had died.
Two standards, two names for the twelfth month
Another disagreement is worth knowing, as it arises between two current national standards.
The publications standard gives 腊月二十三 as a model example of a lunar date. The standard governing the compilation and issue of the lunar calendar prescribes 十二月 as the month's name, and holds that line even where it is awkward — it renders the Laba festival, whose own name contains 腊, as the eighth day of the twelfth month.
Neither is wrong inside its own scope: one governs prose in publications, the other governs published calendars. But citing "the national standard" for 腊月 without saying which one is claiming more than either provides.
A trap in the lunar-date format itself
The calendar standard allows a lunar date to be prefixed with a Gregorian year for disambiguation. The prefix is the year in which that lunar year began, not the year the date falls in.
So a date written as the twenty-ninth day of the eleventh month of the bingshen year, prefixed 2016, actually lands in January 2017. Read the prefix as the Gregorian year of the date and you will be a year out, silently.
If you need to move between the two calendars rather than reason about the notation, our Chinese date format tool handles the conversion directly.
Where this comes from, and what will date it
The clauses quoted here were read from the standards administration's own hosted text of the publications standard and the lunar calendar standard, rather than from a reproduction — which matters more than usual given that misquoted reproductions are the subject. The withdrawal date and replacement of the date-and-time standard are from the national standards registry. The fixed-width requirement is corroborated from a freely published internet specification.
Two limits. The clause numbers inside the replacement date-and-time standard could not be obtained: the administration withholds full text for standards that adopt international ones, on copyright grounds, and the international body's own site refused access. And the ambiguity about padded year-month-day forms is genuine ambiguity in the source, not a gap in our reading.
Published reviewers noticed some of these inconsistencies at the time. A 2012 review in a publishing journal identified the year-abbreviation clauses as inconsistent — one is a flat prohibition, the other a hedged rule — and asked for them to be aligned. It also observed that the old "before the Qing" wording was actively misleading, and that the 2011 fix, while correct, supplied no example for Republican-era dates and so left that question open. Neither point has been addressed since.
What will date this: an amendment to the publications standard. The guide's central observation is that the 2025 review did not make one.
