2004 New Naming Kanji Variants (Jinmeiyou Itaiji)

In addition to the new 488 naming kanjis added in September 2004, 209 variant kanji of existing naming kanjis were also authorized. So far so good, the bad news is that many of the new kanjis are not in the JIS 0208 character set (the standard set of around 6000 kanji that should be more than you'd ever need.) And none of them are in the set of rare kanji - the JIS 0212 character set, which means getting hold of the glyphs is quite a problem. The reason for this seems to be that the kanji variants were previously considered to be the SAME kanji so there was no character code assigned to the variant kanji. All the new naming kanji variants can be found in the new JIS 0213 character set, but unfortunately fonts which support this standard fully are still pretty hard to come by. (Try Sunsale or WadaLabMaruGo2000P4)

Tracking Down the Glyphs

I found information on the new jinmeiyou ("person name use") itaiji ("variant kanji") in various places:

WIKI - the free online encyclopedia
- a personal website belonging to Michio Tomita (who has an interest in this sort of thing apparently). Download his data here. You will need the QANO JIS 0213 font to view this file properly.
The JISC web site - The Japanese Industrial Standards web site

The following sites also have some data but resort to using image files for the more obscure kanjis:

Kanji Bukuro
Seimei Handan

The various sites take different approaches to displaying the more obscure kanjis. WIKI uses a Korean font to display the correct glyph where not present in standard Japanese fonts. In many cases the original kanji displayed in the Korean font gives the required glyph (refer to what I was saying earlier that the kanji and the "variant" previously being considered to be the same kanji.)
Aozora's data requires a JIS 0213 font (QANO) - a sizable proportion the of variant naming kanji glyphs are contained in the QANO font (161 out of 209). It is not clear how the character code mapping in the QANO font works - the character codes do not correspond at all to those in the JISC document.
The JISC data comes in a PDF file which appears to embed a font inside the file. Although I was not able to use the embedded font, the file also contains the Unicode character codes of the all the variant naming kanjis which I was able to use to reveal some additional correct glyphs from the MS UI Gothic font.

During my analysis I also found that in some cases where Wiki had used a Korean font to display a glyph, that Chinese fonts actually contained a more accurate version of the glyph.

Putting it all together

I compiled the data as best I could into an Excel spread-sheet which I invite everyone interested to take a look at. The 5 main columns to note in the spreadsheet are C-G contain the following:

C - the variant kanji given by WIKI displayed with MS UI Gothic. NOTE: Windows does some tricks to display kanjis not in the chosen font, so some of the glyphs may be taken from an alternative font. On my machine it appears that the alternative font being used is the Korean BatungChe font (also being used to display column F)
D - the variant kanji according to Michio Tomita displayed in QANO font
E - the variant kanji according to Unicode (based on the information in the JISC Jinmeicode PDF file) displayed in MS UI Gothic font
F - the kanjis that WIKI displays in a Korean font displayed in the BatungChe font (except row 165 which shows the original kanji displayed in BatungChe font)
G - the kanjis that WIKI displays in a Korean font but which actually display better in a Chinese font (displayed in SimSun font)

I have colour coded "correct" glyphs as follows:

Light Blue - the variant kanji is a JIS 0208 "common" kanji and all 3 sources display a correct glyph
Green - WIKI displays the original kanji in a Korean font to get a correct glyph
Orange - WIKI uses a non JIS 0208 kanji which displays correctly with a Korean font
Yellow - the WIKI kanji in Chinese font displays a "better" glyph than the Korean font glyph
Pink - WIKI uses a non JIS 0208 kanji that displays correctly in MS UI Gothic
Red - WIKI displays the original kanji with a Korean font however the Unicode kanji displays correctly in MS UI Gothic
Blue - the QANO JIS 0213 font contains a matching glyph for a non JIS 0208 kanji

Who Cares Anyway?

After starting to look at kanji variants I was sort of wondering whether or not it was really such a big deal and apparently it is. A colleague at work is called Hi-daka written with two kanjis: the kanji for sun (the first kanji in Nihon) and the kanji for tall/expensive (takai). The official kanji for taka is refered to as kuchi-taka (kuchi=mouth) and looks like this:
There is however a variant refered to hashigo-taka (hashigo=ladder) that looks like this:
Hashigo-taka although it is in Unicode is not an official naming kanji (even with the new additions) and so officially Hidaka is written with kuchi-taka. Ask her to write to write her name freehand though and you'll get hashigo-taka. So yes, kanji variants are a big deal, especially if your name contains one.

My guess is the main motivation for the introduction of the new naming kanji variants is the business of choosing names with "lucky" stroke counts. A lot of the new naming kanji variants have slightly higher or lower stroke counts than their original kanji which makes all the difference when choosing names in Japan.

And the future?

It is rumoured that Windows Vista will contain fonts with all the new kanji variants making most of this discussion redundant. In the mean time, enjoy ...