Rare & Complex Chinese Characters

Ask which Chinese character has the most strokes and you get three answers, because it depends on what counts as a character. In Unicode the record is 𱁬, read taito, daito or otodo in Japanese: 84 strokes, made of 䨺 (three clouds, 36 strokes) stacked on 龘 (three dragons, 48 strokes). It is a kokuji, a character made in Japan, recorded in dictionaries of name characters, and no document has been found in which anyone actually bore it as a name. It entered Unicode in version 13.0 (2020) in Extension G. Next come two 64-stroke characters that are both real dictionary entries: 𪚥 zhé, four dragons 龍 meaning 'verbose, talkative', and 𠔻 zhèng, four 興. Both have been in Extension B since 2001.

The one people actually search for is 𰻞 biáng, the second syllable of biángbiáng miàn, the wide belt noodles of Shaanxi province and the street stalls of Xi'an. Unihan counts 58 strokes for the traditional form 𰻞 and 43 for the simplified 𰻝. It appears in no standard dictionary; noodle shops write it on their signs and teach it with a rhyme that walks you through its parts: 言 speech in the middle, 幺 on each side, 長 long on both flanks, 馬 horse below, 月 moon and 刂 knife, 心 heart at the bottom, all riding on 辶 walk. Unicode encoded both forms in 2020, in Extension G, after years of proposals, which is why fonts older than that show a box.

Then come the characters you can meet in running text. 䨻 bèng (52 strokes, four thunders 雷) is a thunderclap. 龘 dá (48 strokes, three dragons) means 'the look of a dragon walking' and became a national talking point in 2024, when the CCTV Spring Festival Gala chose 龙行龘龘 lóng xíng dá dá as its Year of the Dragon slogan. 靐 bìng is three thunders, 齉 nàng (36 strokes) means a blocked nose and a nasal voice and is the heaviest character in the Xinhua Zidian, 爩 yù is smoke, 鱻 xiān is three fish and the old way to write 'fresh', 麤 cū is three deer and means 'coarse', 爨 cuàn (30 strokes) is a cooking stove and a living surname. The rest of the top table is mostly stacking: 𩇔 nóng is four clouds (48), 䲜 yè four fish, 'lots of fish' (44), 䯂 shēn three horses in a line (34), 龗 líng a dragon under rain (33), 灪 yù and 籲 yù 'to appeal' (32, the traditional form of 吁), 籱 zhuó a bamboo fish trap, 厵 yuán three 原 'a source' (30), 龞 biē a soft-shelled turtle, 蠿 zhuō, 灥 xún three springs 泉, 飍 xiū three winds 風 and 䶵 chí a seven-holed bamboo flute (27). The one you are most likely to need is 鬱 yù: 29 strokes, 'dense, gloomy', the character of 憂鬱 'depression' in Traditional Chinese, simplified on the mainland to 郁.

Two, three and four of a kind: 林 森, 品 晶 鑫, 犇 鱻 and the stacked characters

Chinese has a simple trick for 'more': write the part again. 木 is a tree, 林 lín two trees is a grove, 森 sēn three trees is a forest. 人 is a person, 从 cóng two people is 'to follow', 众 zhòng three is a crowd. 口 is a mouth and 品 pǐn, three mouths, is 'goods, grade, taste'. 日 sun gives 晶 jīng 'crystal, sparkling', 石 stone gives 磊 lěi 'a pile of rocks', the character of 光明磊落 'open and upright', 水 water gives 淼 miǎo 'a vast expanse of water', 火 fire gives 焱 yàn 'flames', 土 earth gives 垚 yáo 'high ground', and three 金 gold make 鑫 xīn, a character invented for names that means nothing but prosperity and is on shop signs all over China. Those five, 鑫 森 淼 焱 垚, are the five elements metal, wood, water, fire and earth, and parents reach for them when a fortune-teller says a child's birth chart is short of one.

Animals triple well. 犇 bēn, three oxen, is a stampede; 猋 biāo, three dogs, is a gale; 羴 shān, three sheep, is the smell of mutton; 驫 biāo is a herd of horses, 鱻 three fish, 麤 three deer, 雥 zá a flock of birds, 蟲 chóng three insects (now simplified to one, 虫), 毳 cuì three hairs for down and fluff. People and tools: 刕 lí three knives, 劦 xié three 力 'joint effort', 叒 ruò three hands, 尛 mó three 小, 孨 zhuǎn three children, 厽 lěi. Three carts 轟 hōng is a boom or explosion, three ears 聶 niè is a whisper and the surname Niè, three tongues 舙 huà is gossip, three 言 譶 tà is talking fast, three 吉 嚞 is an old form of 哲 'wise', three 直 矗 chù is 'towering', and three hands 掱 pá is a pickpocket, a playful respelling of 扒手. Rarer still: 芔 huì three 屮 'plants', 惢 suǒ three hearts, 瞐 mò three eyes, 皛 xiǎo three 白 'bright white', 歮 sè three 止, 贔 bì three 貝 'strong', 畾 léi three fields, and 姦 jiān three women, an old moralising character for 'adultery' that the mainland simplified to 奸.

Four of a kind is rarer. 叕 zhuó, four 又, means 'to connect'; 茻 mǎng is a thicket of grass; 朤 lǎng, four moons, is an old form of 朗 'bright'; 燚 yì, four fires, is a blaze and a name character; 㵘 màn, four waters, is a great flood; 𠈌 is four people and 𣊫 four suns. Doubles range from words you know, 比 'compare', 朋 'friend', 羽 'feather', 哥 'elder brother', 弱 'weak', 絲 'silk', to characters you will meet once: 吅 xuān two mouths, 竝 bìng the old form of 並 'side by side', 兟 shēn 'to advance', 喆 zhé a name form of 哲, 祘 suàn an old 算 'to count', 皕 bì 'two hundred', 誩 jìng 'to quarrel', 孖 mā 'twins' (in Cantonese zi1, as in 孖寶 'twin treasures'), 奻 nuán, 雔 chóu a pair of birds, 聑 two ears, 覞 yào two 見, 玨 jué two pieces of jade, 犾 yín two dogs, 屾 shēn two mountains, 斦 yín two axes and 砳 lè 'rocky'. And 槑 méi, an old form of 梅 'plum', is two 呆 'dumb' side by side, which is exactly why internet users adopted it for 'very dumb'. 囍 xǐ, double happiness, is not a dictionary word at all but the single most pasted character on wedding invitations and red paper cut-outs.

Invented characters: 曌, 龑, 圕, 囧 and the words squeezed into one

A handful of characters have a known author. 曌 zhào, sun and moon over the sky, was coined in 689 for Empress Wu Zetian as her own name, and it has been used for almost nothing else since. 龑 yǎn, a dragon 龍 over the sky 天, was created in 917 for Liu Yan, founder of the Southern Han kingdom, and its simplified form 䶮 sits in Extension A. 圕, read tuān, packs 圖書館 'library' into one square and was coined by a Chinese librarian in the 1920s; it still turns up in library catalogue codes. 圐圙 kūlüè writes a Mongolian word for an enclosed pasture, with 四方 'four sides' and 八面 'eight faces' fenced inside 囗, and survives in place names in Inner Mongolia and Shanxi.

Spoken Chinese contracts some two-syllable phrases into one syllable, and a few of those have their own character. 甭 béng is 不用 'no need' and is in every mainland dictionary; 嫑 biáo is 不要 'don't', 嘦 jiào is 只要 'as long as', 孬 nāo is 不好 'bad, cowardly'. The late Qing and Republican era went further and built characters for Western units that are read as two syllables: 瓩 qiānwǎ 'kilowatt', 兡 bǎikè 'hectogram', 嗧 jiālún 'gallon', 瓱 háokè 'milligram'. They were used in Taiwan and Hong Kong for decades; modern standards spell the words out, so these are now curiosities you will mostly meet in old textbooks.

The Chinese internet revived another set for their shape. 囧 jiǒng was an ancient character for a bright window; around 2008 it became a face with drooping brows and an open mouth, the standard sign of embarrassment, and 囧rz, with r and z as a kneeling body, was its emoticon. 槑 is 'very dumb', 兲 tiān is an old form of 天 'sky' that happens to look like 王八 'turtle', a stock insult, and 烎 yín (開 open over 火 fire) became gamer slang for 'fired up, fierce' in 2009. 巭 stacks 功夫 'kung fu' into one character, 夯 hāng (大 big plus 力 strength) is 'a heavy load' and online 'super', 氼 nì is a person under water, 'to drown', and 氽 tǔn a person over water, 'to float'; 奣 wěng is a clear sky and 嘂 jiào, four mouths around 丩, is 'to yell'. At the other end of the scale sit characters so small they look like typos: 乀 fú, 乁 yí, 丂 kǎo, 丅 (an old 下, also used to write the Latin letter T), 丆 (a Korean reading mark), 乜 miē, in Cantonese mat1 'what?', and 𠀀 hē, the first character of Extension B at U+20000.

Hong Kong, Japanese and simplified characters that live in the Extension blocks

Written Cantonese needs characters that Mandarin never had, and Hong Kong collected them in the Hong Kong Supplementary Character Set. Many landed in Extension B in 2001, so phones and websites that only ship a basic Chinese font turn them into boxes. The ones you see every day in Hong Kong: 𡃁 leng1 'young, a youngster' (𡃁仔, 𡃁模), 𨋢 lip1 'lift, elevator' (from English), 𩠌 sung3 'the dishes eaten with rice', a variant of 餸, 𩬎 'fringe, bangs', 𡆇 wok1 'watt', 𡃈 aak3 'a loop or frame'. Verbs: 𠝹 gaai3 'to cut with a blade', 𠱁 tam3 'to coax, to fool', 𡃶 sek3 'to kiss', 𥄫 gap6 'to peek', 𨈇 laam3 'to step over', 𡄽 soe4 'to slide down', 𦧲 loe1 'to spit; to nag'. Grammar words: 𠻹 tim1 'as well', 𡁵 gan2 the marker of an action in progress, 𠸎 naa1 'with', 𠺝 gaak3 a sentence-final particle and 𠵱 ji1 of 𠵱家 'now'. The full everyday set, including the common ones that older encodings already had (冇 乜 嘢 啲 咁 嘅), is on the Cantonese Characters collection.

Japan has its own residents of the extensions, nearly all of them name forms. 𠮷 is the 吉 with 土 on top that the beef-bowl chain Yoshinoya writes in its name and that many Yoshida families use; 𡈽 is a variant of 土 found in surnames such as 𡈽井, 𫝆 a variant of 今, and 𥝱 jo is the number 10 to the 24th power from the 1627 arithmetic book Jinkōki. 彁 and 妛 are famous for having no meaning: they are 'ghost characters' that slipped into the 1978 Japanese JIS X 0208 standard. 妛 was later traced to a photocopy of 𡚴 (山 over 女) where the paste line between the two halves was read as a stroke; nobody has ever explained 彁. A related trap sits in the CJK Compatibility Ideographs block. 460 of its 472 code points are duplicates kept for old Korean and Japanese encodings; any normalization step (a URL, a search index, a database) silently replaces them with the ordinary character. Twelve are not duplicates at all but real unified characters that happen to sit in that block: 﨎 﨏 﨑 﨓 﨔 﨟 﨡 﨣 﨤 﨧 﨨 﨩. 﨑 is the one you will meet, in surnames like 山﨑 and 﨑山, and it is safe to use in running text.

Some simplified characters are rarer than their traditional originals and ended up far out in the extensions. 驫 has 骉, 龑 has 䶮 (Extension A), 齾 has 𫜰 (Extension C), and Extension G of 2020 holds 𱈜 for 鱻, 𱍂 for 龖, 𱍈 for 龞 and 𱃝 for 飍. The simplified 龘, three 龙, is 𮹝 and only arrived in Extension I in 2023. The most famous case is the anglerfish, 鮟鱇 ānkāng, whose simplified forms 𩽾𩾌 are in Extension B and for years could not be printed on mainland menus, so restaurants wrote the fish's name in the traditional characters or in pinyin. The newest Chinese characters of all are chemical elements: Chinese gives every element one character, metals with the 钅 radical, and when new elements are confirmed China coins new characters. 𬭊 dù dubnium (105), 𬭳 xǐ seaborgium (106), 𬭛 bō bohrium (107), 𬭶 hēi hassium (108) and 𬬻 lú roentgenium (111) were encoded in Extension E in 2015; 鿏 mài meitnerium (109), 𫟼 dá darmstadtium (110), 鿔 gē copernicium (112) and 𫓧 fū flerovium (114) sit where space was found. The four confirmed in 2016, 鿭 nǐ nihonium (113), 镆 mò moscovium (115), 鿬 tián tennessine (117) and 鿫 ào oganesson (118), got their characters in 2017, with 鉝 lì livermorium (116) beside them, so a Chinese periodic table is a sampler of the rarest characters in the language.

What the Extension blocks are, and why your font shows boxes

The ordinary CJK Unified Ideographs block holds about 21,000 characters and covers everything a newspaper in Beijing, Taipei, Hong Kong, Tokyo or Seoul needs. Everything else is in the extensions, named by the letter of the round that added them: Extension A (1999, about 6,600 characters), Extension B (2001, about 42,700, the big one with the historical dictionaries, Hong Kong's Cantonese set and most Vietnamese chữ Nôm), then C (2009), D (2010), E (2015), F (2017), G (2020, where biáng and taito live), H (2022), I (2023) and J, added in Unicode 17.0 in September 2025 with 4,298 more. Together they take the count of encoded Han characters past 100,000. Almost all of them are variant spellings, surname and place-name forms, and characters that appear once in one old book; a literate reader uses perhaps 4,000 and recognises 8,000.

That is also why they show up as boxes. A font is a list of drawings, and drawing 100,000 characters is a decade of work, so most system fonts stop at the basic block plus Extension A. For the extensions you need a dedicated font: the free HanaMin (花園明朝) and BabelStone Han cover Extension B and most later blocks, and recent Noto Sans CJK and Noto Serif CJK releases include the two biáng characters. On a phone, iOS and Android render Extension B characters in their bundled fonts but not the newer Extension G to J, so 𰻞 may still be a box on a device that shows 𪚥 fine. If you copy a character from this page and see a box, the character is intact; paste it where a fuller font is in use and it reappears.

Typing them is the other problem. Pinyin keyboards only offer characters in their dictionaries, so 龘 and 鬱 come up but 𰻞 and 𡃁 usually do not; copying from here is the practical route. If you must type by code point: on Linux hold Ctrl+Shift+U, type 30ede and press Enter; in HTML write 𰻞; on macOS the Unicode Hex Input keyboard needs the two-part surrogate form, D883 DEDE for 𰻞, because these characters are above U+FFFF. That last fact is why rare characters are a favourite test string for programmers: anything that counts 'one character per 16-bit unit' will report 𰻞 as two characters, and a system that truncates on byte length may cut it in half.