The Origin of Hiragana /wu/ Kana 平仮名のわ行うの字源

Introduction 紹介
When I decided to learn Japanese, the first thing I did was to type random kana characters using the kana keyboard to try to understand how the keyboard works. This brought me to the gojūon chart and from there I learned what kana is. So naturally, the gojūon chart and I go way back. But I believe it was when I started doing research on Kobun Tomodachi that I first encountered the Internet famous complete gojūon chart.


wikipedia image hiraganawikipedia image katakana

Personally, I think this image is so intriguing on the internet because it’s aesthetically pleasing to have a complete chart, and at such a great quality. But of course, this begs the question, where did the hiragana /wu/ character come from? It’s pretty well-known that kana are derived from kanji but looking at the also internet famous Kana Development Chart, hiragana /wu/ is the only character whose origin is completely unknown.


kana development chart

So naturally, I became intrigued in this one character whose origin was unknown. And now, I am pretty confident that I have found its origin.


So that’s what I’m going to explain to you today. But before we begin, I would like to specify that this is not an academic article. Some of the background information I will accept as canon, even if I cannot find a scholarly source backing it up. I write this more as a keystone investigation to encourage future more serious study.


Phonetic Background 音声的背景
In the Nara period, the /e/ and /ye/, /i/ and /wi/, and /o/ and /wo/ sounds were distinct. /ye/ merged with /e/ before the creation of hiragana and katakana. /wi/ and /we/ merged with /i/ and /e/ respectively after the creation of kana, and thus have kana representations (ゐヰ and ゑヱ). However, the /yi/ and /wu/ sounds have never existed in the Japanese phonemic inventory.


Internet Investigation on Hiragana /wu/ 平仮名「wu」に関するインターネットの調査
In Japanese, the term for /wu/ is W-Row U or わ行う. In the time of writing, there only exist Wikipedia pages on the topic in Japanese, Korean, and Chinese. The Japanese Wikipedia page of /wu/ in spring 2018, when I started researching for Kobun Tomodachi, had much less information than today. The relevant aspect is that Japanese page had the gojūon chart (though the source was not listed) and the page stated the origin of the katakana /wu/ is 宇 and the origin of hiragana /wu/ is 㐵. But besides the Chinese Wikipedia page, I could not find any other sources with the origin of hiragana /wu/ as 㐵. So ultimately, I just decided to put 宇 as its origin in my app (a decision I now think was very foolish). But fast forward to March 2019 and while I’m re-checking the information in my app, I come across an updated Wikipedia page. Now, the listed origin for hiragana /wu/ is 汙, with a source of a GlyphWiki character page, which claims they are related. A little Google search and I come across a tweet from @Umihotarus, explaining the decision. 汙 is a variant form of 汚, which was used as a man’yōgana for /u/ in the Nara period. Okay case closed, but not yet. The Wikipedia page, thanks to the research by Dr. Kazuo Mabuchi (馬渕和夫) in A Discussion on the Fifty Sound Chart (五十音図の話), has finally named the source of the famed complete gojūon chart, 綴字篇 by Junkichi Katayama (片山淳吉). Also, the page now mentions that the distinct characters for /wu/ and /yi/ were first created in 音韻仮字用例 by Hirokage Shirai (白井寛蔭). Now, I had a place to start, and my own research could commence…


The Origin of the Distinction 区別の原始
So, as I mentioned before, the characters for /wu/ and /yi/ were first created in 音韻仮字用例 by Hirokage Shirai (白井寛蔭). However, this isn’t exactly true in that he didn’t invent the kana, he just designated specific ones used for /i/, /e/, and /u/ to /yi/, /ye/, and /wu/. I provided the kana he designated below with the kanji he suspects to be the origin next to it. The kanji in parenthesis is the more generally accepted origin kanji.


/i/い 伊 (以) イ 伊
/yi/𛀆 以Ya-yihira.jpg 以
/u/𛀍 有 ウ 宥(宇)
/wu/う 于 (宇) 于 宇
/e/え 衣 〈  衣
/ye/𛀏 盈 ye katakana 廷

Real origin 2Real origin 1

Obviously, this system did not catch on entirely.


Various Meiji Era Gojūon Systems 明治時代の様々な五十音システム
A difficulty in studying the origin of kana is that different sources say different information. Below I’ve listed some Meiji era systems.


Here is the system in かなづかひ教科書 (1886).



Here is one from 小学教授書 (~1873).



Here is one from 小学日本文典 (~1874).



And finally, one from 仮名遣 (1891).



We’ll circle back to this chart, because it has an interesting feature I did not notice on first glance…


As none of these differing systems seem to have the hiragana /wu/ found in the textbook, I had no other choice but to turn my attention to the textbook with the complete chart itself.


Analysis on 綴字篇 綴字篇の分析
I have divided the textbook into the sections below.
1. Introduction
2. Vowels
3. Consonants
4. Voiced Consonants
5. Katakana Gojūon Chart
6. Hiragana Gojūon Chart
7. Hentaigana Gojūon Chart
8. Correct Kana vs Hentaigana
9. Kanji numerals
10. Monosyllabic words
11. Disyllabic words
12. Trisyllabic words
13. Quadrisyllabic words
14. Pentasyllabic and greater words
15. Kan vs Go vs Tō readings of kanji
16. Geminatd Words
17. Handakuten Words

The famous chart is in the Hiragana Gojūon Chart section.

1. 紹介
2. 母字
3. 子音
4. 濁音の父音
5. 片仮名五十聯韻音の図
6. 平仮名五十音字正体の經
7. 平仮名五十音字変体の緯
8. 平仮名正変両体
9. 数目
10. 一綴の語
11. 二綴の語
12. 三綴の語
13. 四綴の語
14. 五綴以上の語
15. 漢呉唐音
16. 促音
17. 半濁音



Now, Junkichi Katayama published 綴字篇 in 1873 and is the only chart I found to use this specific hiragana /wu/ (though I believe it is mentioned in a later work). We saw other charts use a /u/ hentaigana for /wu/, but none used this specific /wu/ character. Just by looking at this kana, the right radical appears to be 于, so we can assume that if a kanji inspired it, it would have that as its right radical. Looking at the hentaigana chart in my app Kobun Tomodachi, we see that this character does not match any of the Unicode encoded hentaigana /u/ characters.



Now when I first saw this kana, I just assumed that it took its right radical from its katakana counterpart, and the left stroke was added to match the likes of け, に, は, and ほ (hence why I put 宇 as its origin in Kobun Tomodachi But of course, this isn’t evidence, and the theories on 汙 and 㐵 being the origin make equal sense. So I decided to look a little deeper to see if I could find clues to the origin of this mysterious character.


Out of these sections, the Hentaigana Gojūon Chart, and Correct Kana vs Hentaigana sections appear to be the most helpful. As for the Hentaigana Gojūon Chart I have listed the kana and the kanji that inspired it next to it in Unicode.



𛄋𛃰や 也𛃅 滿𛂦𛂁𛁠𛀾𛀚𛀄
𛄚𛄆𛃫𛃙𛃀𛂜と 止そ 曾𛀸お 於

Now intuitively, if we just created a kana character based off a kanji, rather than creating a hentaigana for it, it would be easier to just make a sloppier version of it, closer to its original kanji. We also see a similar looking character in the Correct Kana vs Hentaigana section.



So, let’s examine this character more closely.


grass script wu1grass script wu 2

If we juxtapose it with the “real” kana character, we see that the “hentaigana” variant appears to just be a sloppier version of the “real” character.


real wu 1real wu 2

But what does this mean? This implies that the hiragana /wu/ character was inspired by an actual character, and the “hentaigana” variant is just a sloppier version of that kanji!


Which Character Inspired Hiragana Wu? どの漢字が平仮名「wu」の字源?
Looking at the right radical of the “hentaigana” variant, I think it’s obvious that it’s 于. It wouldn’t be 干 because that is read as /kan/ and most characters with it as the phonetic component are also read as /kan/. Characters with 于, on the other hand, are read as /u/. But the left side was a little trickier to me. While I’m not an expert in reading different Chinese script styles, it looked to me to be Cursive Script (草書). Intuitively this makes sense as writing man’yōgana in Cursive Script created the intermediary between man’yōgana and kana, sōgana. I looked through Marcus Sesko’s Identifying Japanese Cursive Script for all the characters with the 于 radical on the right. There are two. They are presented below.

変体仮名の旁を見れば、于だということが明らかだ。干の漢字の音読みが「かん」であるから、変体仮名の旁は干ではない。一方、于が旁の漢字は音読みが「う」だ。でも偏の方が私にとってはトリッキーだ。私が字体風格を読む専門家でなくても、それらは草書のように思える。直感的に、万葉仮名を草書で書いた結果、草仮名と呼ばれる仮名と万葉仮名の中間の文字が作られたように思える。私は旁が于である漢字をマルクス・セス子の草書の識別(Identifying Japanese Cursive Script)を調べた。そしてそのような感じを二字見つけた。下に書いてあるものがそれだ。


That’s it! 紆 is the origin kanji! Case closed! Well, not exactly. I would like to present some reasons on why I think 紆 is our best theory on the origin of hiragana /wu/ in this textbook.


First, this character written in Cursive Script does look extremely similar to the one in the textbook. Second, this textbook was written during the Meiji period, so if one was to pick a character to represent /wu/, one would most likely use a common character then, which is likely still common or at least somewhat common now. Now I’m not saying that 紆 is common at all, but compared to our other options, it appears to be the best. The characters I found with a right radical of 于 are 迂汙吁盱訏扜旴紆虶杅圩玗骬釪㐵㚥䄨趶衧䏏酑䩒. Only 迂 is a jinmeiyō kanji; the rest are hyōgai. However, 迂, 吁, and 紆 are JIS X 0208, so they are not too obscure. Third, let’s look at the other proposed origin kanji, 汙 and 㐵. Here are Cursive Script characters with those radicals. Clearly they could not have produced that “hentaigana” variation.

先ず、草書で書かれた紆という漢字は教科書で見た、変体仮名に似ている。また、この教科書が明治時代に書かれたから、「wu」を表すための漢字を選ぶとしたら、おそらく、その時代によく使われたの漢字を選ぶだろう。そしてそれらの漢字は、今でもよく使われる、また知られているものだ。また紆がよく知られた漢字だとを言っているのではなく、他の漢字に比べて、紆が一番有力である。私が見つけた旁が于の漢字は「迂汙吁盱訏扜旴紆虶杅圩玗骬釪㐵㚥䄨趶衧䏏酑䩒」だ。迂だけが人名用漢字で他は表外漢字だ。然し、迂と吁と紆はJIS X 0208だから、レアではない。最後に、字源として可能性のある汙と㐵を見よう。下記は、その部首を含む漢字の草書体がある。それらが変体仮名の字源でないことは明らかだ。


On the other hand, look at the 糸 radical Cursive Script characters. All of those radicals look the same as the “hentaigana” one.



However, I won’t hide that other radicals can produce a Cursive Script radical that looks similar to the one in question. If we look at the hentaigana chart below in Kobun Tomodachi, we can see some examples. However, these examples are outliers and usually those radicals are written differently in Cursive Script.



Another Reference 他の参照
Originally, I thought that there was only one instance of this hiragana /wu/ character, but in doing research for other systems at the time, I think I found another example. Let’s look back to 仮名遣 (1891), which came out after 綴字篇 (1873).


wagyou u kanadukai 2wagyou u kanadukai 1

If we look at the hentaigana section for /wu/, we see a character that looks eerily similar to the Cursive Script form of 紆. This implies that the system of 綴字篇 (1873) wasn’t entirely isolated, and it may have possibly been adapted in other works I have yet to read.


Future Research 将来の研究
Looking back, I am pretty convinced that 紆 is the origin kanji for that textbook, but I still think there is room to do more research. I have yet to find any contemporary academic article in Japanese or English purely on this topic, and I question if there even is such a study. I aimed to only start the conversation, not to have all of the definitive answers. I only looked at a handful of systems for /wu/ that appeared during the Meiji period, there are probably many more out there, and possibly even earlier examples. It would be interesting to measure “all” systems and see if they differ based on region or time or something else. Are there other examples of this hiragana /wu/ kana being used? What was the distribution of 紆 during the Meiji period? Why was 紆 selected for /wu/ rather than a man’yōgana for /u/? If you are going to take this upon yourself and do some research, I highly recommend A Discussion on the Fifty Sound Chart (五十音図の話) by Dr. Kazuo Mabuchi (馬渕和夫). He thoroughly explains the history of the gojūon chart and provides many examples of it in the Meiji period.


I hope you enjoyed reading this, I’m glad to have a much more definitive answer to a question that I have had for over a year. I also hope that this is well-crafted enough to be a source on Wikipedia, lol.


Post a comment

Private comment

Only the administrator may view.

Only the administrator may read this comment.