Japanese vs Chinese stroke order
Which characters Japan and China write in a different order, counted stroke by stroke on KanjiVG and Make Me a Hanzi.
The short answer
KanjiVG (Japanese school order) and Make Me a Hanzi (the mainland order as it records it) share 5,084 characters by code point. Matching their strokes one by one, we find 312 written in a different order, 555 with a different number of strokes and 8 with a stroke drawn differently; 1,300 match stroke for stroke, and for 2,909 the shapes differ too much to match every stroke with confidence, so we make no claim.
Six famous cases
Highlighted numbers are the strokes that move between the two orders. Each pair links to both pages.
The characters with a page here
| Character | JA strokes | ZH strokes | What differs | Pages |
|---|---|---|---|---|
| 主 | 5 | 5 | JA 3 = ZH 4; JA 4 = ZH 3 | kanji · hanzi |
| 以 | 5 | 4 | 5 vs 4 strokes (data sets) | kanji · hanzi |
| 住 | 7 | 7 | JA 5 = ZH 6; JA 6 = ZH 5 | kanji · hanzi |
| 出 | 5 | 5 | JA 1 = ZH 3; JA 2 = ZH 1; JA 3 = ZH 2 | kanji · hanzi |
| 勉 | 10 | 9 | 10 vs 9 strokes | kanji · hanzi |
| 北 | 5 | 5 | JA 1 = ZH 2; JA 2 = ZH 1 | kanji · hanzi |
| 右 | 5 | 5 | JA 1 = ZH 2; JA 2 = ZH 1 | kanji · hanzi |
| 国 | 8 | 8 | JA 4 = ZH 5; JA 5 = ZH 4 | kanji · hanzi |
| 建 | 9 | 8 | 9 vs 8 strokes (data sets) | kanji · hanzi |
| 書 | 10 | 10 | JA 5 = ZH 6; JA 6 = ZH 5 | kanji · hanzi |
| 有 | 6 | 6 | JA 1 = ZH 2; JA 2 = ZH 1 | kanji · hanzi |
| 母 | 5 | 5 | JA 4 = ZH 5; JA 5 = ZH 4 | kanji · hanzi |
| 注 | 8 | 8 | JA 6 = ZH 7; JA 7 = ZH 6 | kanji · hanzi |
| 海 | 9 | 10 | 9 vs 10 strokes | kanji · hanzi |
| 漢 | 13 | 14 | 13 vs 14 strokes | kanji · hanzi |
| 理 | 11 | 11 | JA 2 = ZH 3; JA 3 = ZH 2 | kanji · hanzi |
| 生 | 5 | 5 | JA 3 = ZH 4; JA 4 = ZH 3 | kanji · hanzi |
| 田 | 5 | 5 | JA 3 = ZH 4; JA 4 = ZH 3 | kanji · hanzi |
| 着 | 12 | 11 | 12 vs 11 strokes (data sets) | kanji · hanzi |
| 考 | 6 | 6 | stroke 5 drawn differently | kanji · hanzi |
| 長 | 8 | 8 | JA 1 = ZH 2; JA 2 = ZH 1 | kanji · hanzi |
| 院 | 10 | 9 | 10 vs 9 strokes | kanji · hanzi |
| 青 | 8 | 8 | JA 2 = ZH 3; JA 3 = ZH 2 | kanji · hanzi |
| 准 | 10 | 10 | JA 7 = ZH 9; JA 8 = ZH 7; JA 9 = ZH 8 | kanji · hanzi |
| 叫 | 6 | 5 | 6 vs 5 strokes | kanji · hanzi |
| 喝 | 11 | 12 | 11 vs 12 strokes | kanji · hanzi |
| 坐 | 7 | 7 | JA 5 = ZH 6; JA 6 = ZH 5 | kanji · hanzi |
| 姓 | 8 | 8 | JA 6 = ZH 7; JA 7 = ZH 6 | kanji · hanzi |
| 希 | 7 | 7 | JA 3 = ZH 4; JA 4 = ZH 3 | kanji · hanzi |
| 往 | 8 | 8 | JA 6 = ZH 7; JA 7 = ZH 6 | kanji · hanzi |
| 所 | 8 | 8 | JA 2 = ZH 3; JA 3 = ZH 4; JA 4 = ZH 2 | kanji · hanzi |
| 星 | 9 | 9 | JA 7 = ZH 8; JA 8 = ZH 7 | kanji · hanzi |
| 望 | 11 | 11 | JA 9 = ZH 10; JA 10 = ZH 9 | kanji · hanzi |
| 猫 | 11 | 11 | JA 9 = ZH 10; JA 10 = ZH 9 | kanji · hanzi |
| 玩 | 8 | 8 | JA 2 = ZH 3; JA 3 = ZH 2 | kanji · hanzi |
| 班 | 10 | 10 | JA 2 = ZH 3; JA 3 = ZH 2; JA 8 = ZH 9; JA 9 = ZH 8 | kanji · hanzi |
| 球 | 11 | 11 | JA 2 = ZH 3; JA 3 = ZH 2 | kanji · hanzi |
| 睡 | 13 | 13 | JA 8 = ZH 9; JA 9 = ZH 10; JA 10 = ZH 11; JA 11 = ZH 12; JA 12 = ZH 8 | kanji · hanzi |
| 舞 | 15 | 14 | 15 vs 14 strokes | kanji · hanzi |
| 那 | 7 | 6 | 7 vs 6 strokes (data sets) | kanji · hanzi |
| 都 | 11 | 10 | 11 vs 10 strokes (data sets) | kanji · hanzi |
“Data sets” marks a count that the reference lists (KANJIDIC2 for Japanese, Unihan for Chinese) do not both confirm.
237 order differences in simplified and common forms
75 in traditional forms
For traditional characters the Make Me a Hanzi order is not a mainland standard, and Taiwan and Hong Kong teach their own; read these as a difference between the two data sets.
362 stroke-count differences confirmed by reference lists
KanjiVG's count equals KANJIDIC2's and Make Me a Hanzi's equals Unihan's zh-Hans count for these.
- 似
- 侮
- 偉
- 健
- 傀
- 傑
- 傲
- 像
- 僧
- 免
- 勉
- 勤
- 勵
- 勸
- 卑
- 卿
- 叫
- 呱
- 唳
- 啄
- 喝
- 嗟
- 嗷
- 嘆
- 器
- 嚮
- 嚼
- 堅
- 塚
- 墜
- 墟
- 墨
- 夢
- 姫
- 娜
- 婀
- 孤
- 宦
- 層
- 岌
- 嵯
- 巍
- 幣
- 弊
- 弧
- 悔
- 愧
- 慨
- 慳
- 憎
- 憐
- 懊
- 懲
- 拒
- 挺
- 捩
- 揶
- 搓
- 撻
- 擒
- 擲
- 攬
- 敏
- 敝
- 斃
- 既
- 曠
- 桀
- 梅
- 棄
- 椰
- 楮
- 概
- 槎
- 槐
- 槨
- 橄
- 橡
- 橢
- 檎
- 檻
- 櫚
- 欄
- 欖
- 歡
- 残
- 毒
- 汲
- 浅
- 海
- 涎
- 溲
- 滞
- 漓
- 漢
- 潴
- 澳
- 濫
- 炬
- 熙
- 熬
- 燠
- 燧
- 爨
- 狐
- 猾
- 獲
- 琢
- 瑯
- 瑰
- 瑶
- 璃
- 瓊
- 瓜
- 瓠
- 瓢
- 瓣
- 瓶
- 瓷
- 甄
- 甌
- 甓
- 甕
- 痾
- 瘟
- 監
- 睹
- 瞥
- 瞬
- 瞰
- 碑
- 磋
- 磔
- 祀
- 祕
- 祗
- 祚
- 祠
- 祺
- 禅
- 禊
- 禧
- 禪
- 禳
- 穀
- 突
- 箪
- 箸
- 篳
- 篷
- 簧
- 簪
- 籃
- 籬
- 糾
- 緊
- 緯
- 練
- 繁
- 繭
- 纜
- 羌
- 羞
- 肺
- 腎
- 腱
- 腴
- 臣
- 臥
- 臧
- 臨
- 臭
- 舜
- 舞
- 艇
- 艘
- 艦
- 苡
- 苣
- 茲
- 莊
- 菰
- 萬
- 萸
- 葉
- 蒋
- 蒔
- 蒼
- 蓋
- 蔽
- 薦
- 薺
- 藥
- 蘆
- 蜒
- 蝉
- 螂
- 螯
- 蟒
- 蠶
- 衛
- 衷
- 褐
- 褪
- 襁
- 襤
- 覽
- 觀
- 觚
- 誕
- 諱
- 謁
- 謨
- 謹
- 譴
- 護
- 豎
- 豫
- 賓
- 賢
- 贈
- 贓
- 赧
- 赭
- 赳
- 践
- 蹉
- 躇
- 躑
- 迢
- 迴
- 迸
- 迹
- 逋
- 逍
- 逑
- 逕
- 逖
- 逞
- 逡
- 逶
- 遐
- 遑
- 遒
- 遘
- 遙
- 遞
- 遥
- 遨
- 遲
- 遽
- 邀
- 邂
- 邃
- 邇
- 邊
- 邏
- 邨
- 邯
- 邱
- 邵
- 邸
- 郊
- 郛
- 郡
- 郢
- 部
- 郵
- 鄂
- 鄒
- 鄙
- 鄲
- 鍵
- 鏈
- 鑑
- 鑒
- 閭
- 阡
- 阪
- 阮
- 阻
- 陀
- 陂
- 陌
- 陏
- 陛
- 陜
- 陝
- 陞
- 院
- 陣
- 陪
- 陬
- 陲
- 陳
- 陵
- 陷
- 陸
- 隅
- 隈
- 隍
- 階
- 随
- 隔
- 隕
- 隗
- 隘
- 隙
- 障
- 隨
- 隰
- 隱
- 隴
- 雅
- 離
- 難
- 霆
- 韃
- 韋
- 韓
- 韜
- 頻
- 類
- 飩
- 飫
- 飭
- 餃
- 餉
- 餒
- 餘
- 餞
- 餡
- 餬
- 餾
- 饅
- 饉
- 饋
- 饌
- 饑
- 饒
- 饗
- 騷
- 驚
- 骰
- 骸
- 骼
- 髀
- 髑
- 髓
- 鬮
- 魁
- 魃
- 魅
- 魍
- 魑
- 魔
- 鰹
- 鱗
- 鹽
- 麟
- 麿
- 黛
193 counts that differ between the data sets
Here at least one count is not confirmed by its reference list, so the difference can be a convention of the data (艹 in four strokes, 辶 in three) rather than a difference in teaching.
How the comparison works, and its limits
Both glyphs are centered and scaled to the same box. Each stroke becomes twelve points along its center line, and the Hungarian algorithm pairs the Japanese strokes with the Chinese ones by the smallest total distance. We call the order different only when every moved stroke matches its new partner clearly better than the stroke with the same number; otherwise the pair counts as unsure and no claim is made. Two things the method cannot see: a difference in the shape of a stroke that keeps its place, and a difference that comes from the two fonts rather than from the order. The comparison is ours, released under CC BY 4.0 on the data page.