8 February 2026

I was intrigued by a kana I never saw before: . This character supposedly represents visarga, a concept in Indian linguistics. Here are all other non-Unicode kana previously known to exist before my searches:

- pronounced: "E" as opposed to "Ye" (used in contexts where エ was ambiguous between E and Ye) - the letter 𛀀 was also used but it's in Unicode so it's not on my list

- variant of ネ

- variant of ヰ

- pronounced: "U" as opposed to "Wu" (used in contexts where ウ was ambiguous between U and Wu)

- variant of 𛄢, pronounced "Wu"

- variants of 𛄡, pronounced "Ye"

- variant of 𛄠, pronounced "Yi"

In my search for the Visarga's use in Indian languages, and for more Kana characters, I found more characters, that are so obscure, that not only they are not in Unicode, but also ChatGPT doesn't recognize them and makes up random "fluff" details that are not true. Characters I found in my search for the elusive Visarga:

a) characters from the same book that the Visarga came from:

- ChatGPT thinks this is a katakana for /ʔ/, a devoiced vowel mark, or pitch-lowering, but I could not find anything to confirm this meaning it's likely a lie. There are already two characters used informally for glottal stop, ("ッ" - common in comic books, "・" - common in UTAU aliases) so why would you need a third one? In reality, this character, encoded in Unicode as ヽ, is an iteration mark.

- at first I thought this is a variant of レ that has an extra dot on top, and based on this similarity ChatGPT identified it as a member of a group called “dot-katakana”, which supposedly follows a "very specific pattern" used by "Early Meiji phoneticists" to notate "entirely new consonant-vowel symbols for foreign languages." It told me that the group was very common in "1880–1910 books" written by "several Japanese scholars" who created new katakana to represent /l/ vs /r/ distinctions in European languages, /ɹ/ (English “r”), /dʒ/ and /ʒ/, /ʃ/ variants, /ð/ and /θ/, /v/, Arabic/Indic retroflexes, and special vowels. ChatGPT said that nearly all of these used a dot placed near a modified basic katakana skeleton, producing shapes that look like: Katakana レ with a dot, Katakana フ with a dot, Katakana リ with a dot, or new shapes not resembling any existing kana, with a dot. It stated that specifically, was invented specifically for the English language, to mark the sound L or R. I tried to fact check the idea of "dot-katakana", the closest I could find was the 20th century practice to notate "l" as "r+handakuten" for Latin songs, and an old practice using superscript small kanas to represent consonants^[1], but I couldn't find any dotted kanas, making me believe that statement was a complete fabrication. After more closely examining the book, I came to the conclusion that Motoori Norinaga simply had a hard time writing the letter "シ", and because he moved his brush to quickly (perhaps to save time), it came out looking as a weird mutant "レ" glyph.

- at first I thought this is another katakana, as the book doesn't use hiragana, and even uses katakana after kanji for grammar, and was commonly used after kanji. After further checking, the character is likely 〻 - a Kanji iteration mark. ChatGPT claimed this is a character for /ʃ/ or /ʒ/, calling it "the “sh” / “zh” consonants not covered by シ・ジ", on the basis that "S-curves for sibilants are attested in multiple Meiji phonetic charts" (likely a lie, I looked at older Japanese phonetic charts, from the Meiji period and other periods that were close to the Meiji period, and could not find a single one), and that "it has no dot, which indicates a consonant rather than a vowel modifier" (I don't know what does that has to do with anything, a quick Google search for "old japanese vowel modifier" reveals no old kana editions by that name)

- a quick search for "kana ligatures" (an idea I had based on the fact that ligatures are very obscure and look very curly like this letter) reveals that the character is actually 𬻿 - an abbreviation for ナリ, and not a "foreign R/L sound" as ChatGPT thought.

- while ChatGPT claimed the character is "derived from the Latin letter y" and represents /ʎ/, my search for "kana ligatures" came out with a correction: the character, identified as 𬼀 (), is an abbreviation for シテ.

- ChatGPT called it a "19th–early-20th century Japanese phonetic transcription" mark used for aspirations, citing examples like カ + = kʰa, タ + = tʰa, パ + = pʰa, but again, I found no evidence of aspiration being marked in Katakana in any way. A quick Google search for "katakana aspiration mark used by old linguists as constructed kana" proved fruitless. Luckily, an old Katakana chart identified this character as a variant of マ.

- Though I tentatively identified it as a "diacritic that modifies kana sounds", and ChatGPT agreed (even falsely stated that this character is actually attested as a “stop marker” meaning the vowel ends abruptly, giving IPA transcriptions like [V̚] (vowel with no audible release), [Vʔ] (vowel + glottal stop), and [V̆] (very short vowel), saying they are authentic Japanese), when I closer analyzed the context I realized this is simply a quotation mark. For example, "イ" simply means "the letter «イ»".

- ChatGPT said this is an L, stating plainly: "Several foreign-language textbooks printed between 1875–1910 introduced a standalone katakana letter to represent the foreign sound /l/", citing examples like ア - al, イ - il - エ - el, ラ - ral or a syllable with /l/, ヤ - "yal", and even "textbooks wrote sequences like: bal, fel, del, German «L»". I found no examples, meaning this is a lie. Looking for occurences of this kanji with different readings, I came to the conclusion that this is simply sloppy writing, and he intended character was リ.

- Though I first declared it a combination of and a dakuten, and ChatGPT suggested the value /ɛ~æ~ə/ claiming these sounds were commonly introduced “extra e” characters in Meiji manuals (for which I found no evidence for), Wikipedia's iteration mark page gives a Tsurezuregusa example that looked similar to the one I found in my book not only in character use, but also in layout and pattern, meaning this is actually an iteration mark (Expression error: Unrecognized punctuation character "[".). Frustrated with 9 "fails" from ChatGPT, I no longer used ChatGPT to identify more letters I found.

- Based on examining patterns, I believe it was just a sloppy typo for either マ or .

- While I initially believed it to be a constructed kana based on ノ or ミ, the old Katakana chart from 1983 I used earlier to identify , identified it as merely a variant of ク.

- Initially, this was a total mystery. I could not find a character that looks like this in any kana table. I looked at obscure glyph tables, and lists of obscure kana, nothing came up... until I discovered a different copy of the same book, which was in much better quality, and I realized this was not a new character, but simply a scanning artifact that distorted the appearance of an existing character, and the character in question was ヨ (the standard katakana for yo)

- and finally, we have the elusive character itself. On Wikipedia and Wikimedia Commons, it is described that this character was invented by Motoori Norinaga to rerpresent the Sanskrit visarga, and that he used it in a book about Indian orthography. I could not find any evidence for this claim. The book I had, which was the source of the actual character, only claimed that this represents the entering tone of Chinese, and gave the examples of "ジ" and "エ" for the Tang-dynasty Chinese pronunciation of the words "日" and "月".

I also found unencoded kana in an Ainu book. Luckily, the book listed equivalent Japanese kanas, so I know the sounds. Here are all the equivalences the book gives, including ones that already have unicode representations:

ヲ　＝　ウォ

、　ツ゚、　　＝　トゥ

𛄡　＝　イェ

ヱ　＝　ウェ

　＝　チャ

　＝　チュ

　＝　チェ

　＝　チョ

　＝　シャ

　＝　シュ

　＝　シェ

　＝　ショ

As of 8 February, the validity of Wikipedia's claims about remains unknown. It is possible that Motoori Norinaga used the symbol elsewhere to transcribe Visarga, maybe in a different book (the book where the visarga image is sourced from is not a "book about Indian orthography"). If this is true, the book remains lost media (of "Existence Unconfirmed" status) as of 8 February 2026. However, it could just as easily be a misunderstanding of the existing book, or a hoax.

15 February 2026

The AI chatbot Gemini recommended looking for these: 1. The "Mincho" Specific Variations While many Hentaigana made it into Unicode via the 2017 update (v10.0), many specific calligraphic ligatures and cursive abbreviations used in woodblock printing (Edo period) didn't.

The Hunt: Look for Kuzushiji (grass script) versions of specific kanji-kana hybrids often found in Kibyōshi (satirical graphic novels).

2. Specialized Phonetic Symbols (1880s–1920s) During the Meiji era, there was a frantic period of "alphabet invention." Some scholars created hybrid kana to represent English sounds more accurately than standard katakana.

The "Th" and "L" Experiments: Look for the various proposed characters that merged ラ (ra) with other strokes to distinguish "L" from "R."

The "V" Variants: Before ヴ won the war, there were at least three other proposed characters involving modified フ (fu) strokes that never saw a digital font.

3. Regional/Okinawan Variations Early transcriptions of the Ryukyuan languages sometimes used modified kana to account for glottal stops or vowels that standard Japanese lacks.

The Ryūkyū-kana: While some have been standardized, many handwritten variations from 19th-century records remain in "image-only" limbo because they don't appear in enough printed texts to satisfy Unicode's "stability" requirements.

4. Personal "Idiolect" Kana In the diaries of certain Heian-era poets or Kamakura-era monks, you'll find "ghost kana"—characters that appear to be shortcuts or unique blends used only by that specific person or within a specific temple school.

However, I believe these are lies or hallucinations. The "th and l" kana were already found, but only in the form of dakuten and handakuten glyphs, not as new shapes.

Because I have made so many discoveries, I think it's a good idea to keep them in a list. I have already found the Okinawan ones in Nishiki-teki. The Ryūkyū-kana likely refers to superscript kana as seen here. I also came across the information that Palauan used to be written in Katakana, but I was unable to find an example.

1 March

Again, Gemini lies were detected. Annotated quote from Gemini (with my comments formatted like this):

"Before the modern system of using small kana (like フィ for fi) was standardized, many newspapers and dictionaries used unique diacritics that were never encoded. I don't know if it's true but I cannot find any examples.

The "V" Circle: Look for ウ or フ with a handakuten (゜) placed in non-standard positions (like directly inside the curve). I do not believe this exists. The closest I could find is a Reddit post asking about a フ with a dakuten on the bottom (something like フ𓏮 or similar), which was revealed to be a stylized modification for ブ or simply a weird font choice. I couldn't find any mention of a handakuten specifically inserted inside a character (⿴ウ𐩲 or similar) anywhere.

The Aspirated Katakana: In early transcriptions of Chinese or Sanskrit (often by scholars like Bunyiu Nanjio), you may find a small "tick" or "comma" attached to the top-right of カ, タ, パ to represent aspirated consonants ($k^h, t^h, p^h$). Aspirated katakana do exist, they are written using a dot on the bottom of a kana, used for the Taiwanese language. However, I could not find any comma-like variants specifically. Comma like marks on the side of katakana, called side dots or sesame dots (﹅), and are used to indicate emphasis or prosodic stress. They do not represent aspiration or other sound variations.

Target: Early Meiji bilingual dictionaries (English-Japanese or German-Japanese).

[...]

Since you found the unencoded "Tu" (), look for the corresponding "Ti" and "Pi". I checked and all sources I found claim that Ainu has no "ti" sound^[2]. As for Pi sound, standard katakana already has this (ピ), so it would be pointless to make another one.^[3]

In some 19th-century Ainu publications, there was a brief attempt to use a "reversed" Katakana (mirror images) to represent sounds that didn't exist in Japanese. I couldn't find any examples, but I believe it's fake because the letters of katakana are optimized for pen strokes, and a reversed variant of, say, ク, タ, ネ, ヰ, or even ア, would be quite cumbersome to write - while the sounds they were supposedly used for are commonly used Ainu sounds, about as common as the letter "a" in English.

Target: Any scan of the Ainu Shin'yōshū (original manuscripts) or John Batchelor’s An Ainu-English-Japanese Dictionary (1st Edition). lime

[...]

Finding the "Foreign Sound" diacritics requires looking at the specific window of time (roughly 1867–1912) when Japanese scholars were trying to "solve" the problem of Western phonetics using the existing Katakana framework.

Because these characters were never standardized, they don't appear in modern digital dictionaries. You have to look for the "Alphabetized" English-Japanese dictionaries from the early Meiji period.

1. James Curtis Hepburn’s A Japanese-English and English-Japanese Dictionary (3rd Edition, 1886)

While Hepburn is famous for the Romanization system, his 3rd Edition is where he and his Japanese collaborators (like Takahashi Goro) experimented heavily with Katakana to represent English sounds. I checked and this claim is a blatant lie. The book in question does not use katakana to transcribe English sounds.

What to look for: Look for the "Table of Sounds" at the beginning of the dictionary. I checked and there isn't one. There is only a Japanese kana table in iroha order, but no "English sounds" table.

The Diacritics: You will find ラ゚, リ゚, ル゚, レ゚, ロ゚ (Katakana with a circle) used explicitly to distinguish L from R. These characters are a 20th century invention and do not appear in this book, which was written in the 19th century.

V-Sounds: Look for ウ with a dakuten (ヴ) which Hepburn helped popularize, but also look for フ with a handakuten (プ) used as an alternative for "Fu/Hu" vs. "Vu". The プ is already a standard kana, and is sounded as pu. The claim that プ is a nonstandard letter that represents a F or V sound and that it was specifically invented for this one book, is a blatant lie

2. Takahashi Goro’s A New English-Japanese Dictionary (新訳英和辞典, 1887) I did my research and this book appears to be entirely fabricated. Though I did find a few books of a similar title, the author and release date did not match. I did find a book with the same Japanese title, but the author and release date didn't match, and the title appears to simply be a generic book title so it's no surprise I could find a book with a matching title. I was unable to find this specific book, and using quotation marks to narrow down the search turned out fruitless. I don't believe this book exists.

Takahashi Goro was a phonetic maximalist. He found standard Katakana insufficient for English and German. I could not find any information supporting this claim. All I could find is that there exists a dictionary writer by that name, and that he co-edited Hepburn's dictionary which I looked at earlier.^[4]

The Diacritics: He is known for using small ticks or commas next to Katakana to represent specific vowel qualities (like the "a" in cat vs. the "a" in father). I've already debunked the false claim of "commas" being used in Katakana.

The "Th" problem: He often used サ゚ (Sa with a circle) or ス゚ (Su with a circle) to represent the English "th" ($[θ]$ or $[ð]$). It is true that サ゚/ス゚ are used for /θ/ and /ts/ sounds,^[5] I could not find any info about these letters being used for the sound /ð/ (that sound was written using 【ザ゚】/THI【ジ゚】/THU【ズ゚】/THE【ゼ゚】/THO【ゾ゚】instead^[6]), nor could I find any information about Takahashi Goro using these letters.

3. Tetsugaku-jii (哲学字彙, 1881) by Inoue Tetsujirō I checked and the book seems to use only Kanji, not Kana.

This was the first dictionary of philosophical terms.

Because it dealt with complex German and French concepts, the authors had to invent ways to transcribe names like "Kant" or "Hegel" precisely. I only saw Latin letters used for loanwords. Not new kana.

The Diacritics: Look for the "Double Dakuten" (four dots or very thick dots) used to represent highly aspirated or "heavy" foreign consonants. This appears to be a blatant lie.

The "Nasal" Mark: They used カ゚ (Ka with a circle) not just for Japanese nasal-G, but to represent the French nasal "n". This also appears to be a blatant lie.

4. Fukuzawa Yukichi’s Seiyō Jijō (西洋事情, 1866–1870)

While not a dictionary, this book set the standard for Meiji-era foreign transcription.

The Invention of ヴ: This is the primary source for ヴ. Even though this is true, this is not new knowledge and was already known beforehand.

The "W" variants: Before ワ, ヰ, ヱ, ヲ were restricted to their current sounds, Fukuzawa used ウ with various small marks to represent the "W" in "William" vs. the "V" in "Victoria." and this too. "

What is known now