The Unicode Consortium Discussion Forum

Regarding U+2016 and U+2225, it seems there used to be some inconsistency between implementations of different operating systems in the age of Shift-JIS. As Unicode (and JIS X 0213) have been widely accepted, U+2016's mapping to the DOUBLE VERTICAL LINE glyph also has become stable. On the other hand, the DOUBLE VERTICAL LINE glyph itself has long been implemented in most Japanese fonts, since the days when neither JIS X 0213 nor Unicode were widely accepted, with the 'vert' table including its vertical glyph using the －90 degrees rotated shape.

Also, it is known that the DOUBLE VERTICAL LINE glyph is used to represent the norm of a matrix in mathematics.

On the other hand, because JIS X 4051 defines a set of note "reference marks" including the DOUBLE VERTICAL LINE, the JLREQ document seems to regard this character to be one of the note reference marks used in Japanese typography.

However, as far as I suspect, the reason why JIS X 4051 defines this character to be a note reference mark is not very clear. (By the way, this character is usually called 'parallels' in Western typography, according to Robert Bringhurst's book mentioned below).

As Ishii-san mentioned, Wikipedia also has a page introducing the character as a note reference mark:
http://en.wikipedia.org/wiki/Footnotes

However, look at the sentence written by Robert Bringhurst quoted in the Wikipedia article. According to Bringhurst's The Elements of Typographic Style, the following marks: section, parallels, and pilcrow should not be used as note reference marks today, because use of them is obsolete and can confuse the reader.

Similarly, Chicago Manual of Style, 16th Edition, allows all such reference marks to be used for tables, but it says only asterisk, dagger, double dagger, and section should be used for note references in the body text.（p. 666, p. 148）。

Hence, it can be said that the "parallels", viz. the DOUBLE VERTICAL LINE glyph is not used widely for publications today.

In addition, as I mentioned in the beginning, many Japanese fonts have used the －90 degrees rotated shape for the vertical writing mode, with the help of the 'vert' table, and this makes it clear that using the －90 degrees rotated shape for the vertical writing mode is the most practical and reasonable usage of the U+2016 character.

For this reason, I suspect that JLREQ's assumption that the DOUBLE VERTICAL LINE glyph corresponding to U+2016 should be regarded as a note reference mark, is incorrect in practical publishing and printing contexts in Japan.

Also, for the reasons mentioned above, I think the vertical posture of the U+2016 character should not be "Upright", and it should be the －90 degrees rotated shape, as represented by using the 'vert' table in many Japanese fonts available today.

Regards,

Taro Yamamoto

Bibliography:

1. Robert Bringhurst, The Elements of Typographic Style, Second Ed., Hartley & Marks, Point Roberts and Vancouver 1996.
2. Chicago Manual of Style, 16th Edition, The University of Chicago Press, Chicago and London 2010.

I'm sorry to say that reasons you raised didn't convince me enough.

First of all, saying "U+2016 has long been implemented in most fonts using the -90 degree rotated shape" is too much, given the fact that MS Gothic and MS Mincho do not implement them.

Second, it looks like you think U+2016 is not used in Japanese but only in Latin context, but the code was defined in JIS X0213, and JIS X 0213 defines it to be vertical lines (upright) even in vertical flow.

Third. I agree that some authors use U+2016 for math. But we haven't got agreement whether math should be sideways or upright. Laurentiu mentioned Unicode math (Sm) code points are mostly snippets and we should not assume these code points being used to express real math expressions, and I think that makes sense. If they're snippets, upright makes more sense to me.

So, while I'm still not 100% sure that making them upright is correct thing, I do not see good enough reasons to make them rotated either.

You wrote:

I know there were (and are) these two exceptions. Before Vista, they did not have the U+2016 mapping at all. And there seem to have been inconsistency in their interpretation of the U+2016 and U+2225 characters. Each font has its own history, and I understand it's not always possible to eliminate every existing inconsistency.


No. It seems you misunderstood what my comments meant. It is JIS X 4051 that mentions the Western typographic (viz. bibliographic) usage of the character, while there was the fact the character has long been implemented by Japanese fonts for Japanese usages, as I introduced first in my previous comment. JLREQ also seems to implicitly reference JIS X 4051's interpretation and support the reasoning based on the usage in Western typography.


Understood. This is one way of thinking. It's understandable that the glyph shape may be useful for various purposes including math, but not limited to it. I don't think it is used only for mathematical purposes. It's one of many usages of the character.


I think I could succeed in refuting the assertion that it must not be rotated because of the usage in bibliographic purposes (as a note reference mark). The usage of the character as one of the note reference marks in Western typography, as mentioned in JIS X 4052 no longer exists in today's typography in the real-world publishing in the West.

Now, I do not find any reasons why the character cannot be rotated in the vertical writing mode, because the fact that many Japanese fonts have implemented it with the rotated vertical glyph still remains unchanged.

Regards,

--Taro

Please give the evidence that JIS X 0213 defines it to be upright in vertical flow. I could not find such definition in JIS X 0213.

JIS X 0213 gives vertical variant shape examples for certain characters but it is stated that the shape examples (字形例) is informative and not a part of the standard.

(JIS X 0213:2000 附属書4（規定）仮名，特殊文字及びけい線素片 says:
字形例　当該面区点位置で表現される図形文字の字体で，参考となるその他の字形がある場合の字形を例示したもの。この例示は，一部の面区点位置についてだけ行った。また，縦書き用の字形が異なることがある場合も字形例で示した。この欄は，参考であって規定の一部ではない。)

I understand that informative examples may be useful information. However the fact that JIS X 0213 does not give a vertical variant shape example cannot be a clue that it should be upright in MVO, because JIS X 0213 gives such informative shape examples for only limited characters that the vertical variant is essential for Japanese vertical text and does not give shape examples for characters not usually used in Japanese vertical text. In fact, there are many characters that are MVO=R but JIS X 0213 does not give a vertical variant, for example, the following characters:

COMMA,
FULL STOP,
HYPHEN,
HYPHEN-MINUS,
VERTICAL LINE,
Arrows,
Box Drawings,
Math Symbols,
Latin, Greek, and Cyrillic Letters,
etc.

Regards,

Shinyu Murakami
Antenna House