峰 and 峯 are the same word — "peak" — with the same sound, fēng, and the same meaning. The only difference is where the mountain radical sits: beside the character in 峰, on top of it in 峯. In mainland China 峰 is the standard form and 峯 is a variant, an 异体字. In Taiwan, too, 峰 is standard. To a computer, though, they are two different characters with two different code points, and a search for one will not find the other.
Which form counts as correct, and which forms count as the same character, is decided separately by several authorities, and they do not always agree. This guide sets the standards from the mainland, Taiwan, Hong Kong, Singapore and Unicode side by side, with worked examples, and explains what their differences mean for anyone searching or storing Chinese text. You can look up any character's recorded variants in our variant character lookup.
The mainland: one list, decided by the state
Mainland China's current standard is the 通用规范汉字表, the Table of General Standard Chinese Characters, issued by the State Council in 2013. It lists 8,105 characters and replaced the character tables before it. An appendix pairs each standard character with the forms it replaces: 794 groups, containing 1,023 variants, alongside a separate list of traditional forms.
The appendix distinguishes between two easily blurred ideas. A traditional form (繁体字) is the pre-simplification version of a character, such as 線 for 线. A variant (异体字) is an alternative way of writing the same character that the table has chosen not to use, such as 峯 for 峰. A single character can have both.
The 2013 table also reversed some old decisions. Its notes record that 45 characters that an earlier list, from 1955, had set aside as variants were made standard again — among them 皙, 喆, 淼, 昇 and 邨 — largely because they are used in people's names.
Taiwan: standard forms, and a dictionary of everything else
Taiwan's Ministry of Education sets its standard forms in its own tables of 標準字體, one for commonly used characters and others for less common and rare ones. Those tables are the normative counterpart to the mainland's 2013 list, and in traditional script — 正體字, as Taiwan's official usage calls it — they choose the correct form of each character.
Alongside them sits a much larger reference work, the ministry's 異體字字典, the Dictionary of Chinese Character Variants, whose seventh official edition appeared in 2024. It sets out to record the variant forms of characters found in historical and modern sources, and the ministry describes it as a database for language education and research. It records variants; it does not choose between them.
Hong Kong and Singapore
Hong Kong has its own list of standard traditional forms, the 常用字字形表, compiled for schools, and it does not always choose what Taiwan chooses. Hong Kong writes 衞 and 着 where Taiwan's ministry writes 衛 and 著. Two places using traditional script can still disagree about which traditional form is standard.
Singapore took its own route to simplified characters. Its Ministry of Education issued a first table in 1969, with 498 simplified characters made from 502 traditional ones. A fuller table followed in 1974, with 2,287 characters, of which 49 differed from the mainland's forms or were not simplified there at all. The 1976 revision removed every one of those differences, and since then Singapore's simplified characters have matched the mainland's. When the mainland adjusted a handful of characters in 1986, restoring 叠 and 覆 among others, Singapore adjusted its school character lists to match in 1993. Sources differ on some of these counts; the figures here follow a published history of Singapore's character standards by the scholar Xie Shiya.
Six groups, three authorities
The Unicode standard, which assigns every character its code point, adds a third point of view. Its database records which characters are variants of which, but that record is separate from any national list, and the two do not always line up.
| Characters | Mainland, 2013 | Taiwan standard | Unicode |
|---|---|---|---|
| 峰 / 峯 | 峰 standard, 峯 a variant | 峰 | two code points, linked as variants |
| 群 / 羣 | 群 standard, 羣 a variant | 群 | two code points |
| 够 / 夠 | 够 standard, 夠 a variant | 夠 | 夠 recorded as the traditional form of 够 |
| 里 / 裏 / 裡 | 里 standard; 裏 traditional, 裡 a variant | 裡 standard, 裏 its variant | three code points |
| 凉 / 涼 | 凉 standard, 涼 a variant | 涼 | two code points |
| 秘 / 祕 | 秘 standard, 祕 a variant | 祕 | two code points |
Two rows are worth a closer look. For 够 and 夠, the mainland table and Unicode's database file the relationship under different headings: the table calls 夠 a variant, while Unicode records it as the traditional counterpart. For 里, 裏 and 裡, the mainland table and Taiwan's standard choose opposite primary forms: the mainland treats 裏 as the traditional form and 裡 as a variant, and Taiwan makes 裡 the standard and 裏 its variant.
Unicode itself follows a rule of unification. Where two shapes are judged to be the same character, differing only in style, they share one code point and a font decides how it looks — 骨, "bone", is drawn with a slightly different inner stroke in mainland and Taiwanese typefaces but has a single code point. Where an earlier national standard had already encoded two shapes separately — Japan's character standard gives 峰 and 峯 a code each — Unicode kept them separate. For cases in between, a later mechanism, the ideographic variation sequence, lets text request a particular glyph of one code point without creating a new character.
What this means for names
The rules matter most for names. On the mainland, the national language law revised with effect from 1 January 2026 lists the cases where traditional and variant characters may still be used, and variant characters in surnames are one of them; the list names surnames, not given names. Taiwan's name act accepts a character for registration if it appears in one of several standard dictionaries, including the ministry's own Mandarin dictionary. In both places, the question of whether a name can be registered as written turns on which list the character is on.
Searching when the forms differ
For anyone building a search box or cleaning a database, the main takeaway is this. Unicode defines normalization forms, NFC and NFKC, which make text comparable by folding equivalent sequences together. Neither folds any of the pairs above. Tested on 21 September 2026 against a recent version of Unicode, 峯 stays 峯 and 夠 stays 夠 under both, and so do 裡, 線 and 涼.
Normalization does change a different set of characters: the compatibility ideographs, duplicates kept for round-tripping with older standards. U+F907, for example, becomes 龜 under NFC. Normalization alone is therefore a poor fit: it misses the variants a reader expects to match, and it silently changes compatibility characters the writer did not expect to change. A search that should find 峯 when someone types 峰 needs an explicit mapping — Unicode's database of variant relationships is one starting point — and a decision about which relationships to trust, since the mainland table, Taiwan's standard and Unicode do not always agree.
Where the classifications come from
The mainland figures come from the 2013 table and its notes, read from the government's scanned publication. Taiwan's classifications are from the Ministry of Education's variant dictionary and its editorial notes. For Hong Kong, the forms are taken from its standard-form list and described in reference sources. The history of Singapore's standards is from Xie Shiya's review, which cites the ministry's tables; one widely used reference gives slightly different counts for the 1974 table. Rules for names are from the texts of the mainland language law and Taiwan's name act. Unicode relationships are from its character database, and we ran the normalization test here on the characters shown.
