Could it be good reference?
1953/1/1 United Daily
> Martin J. Dürst <duerst@it.aoyama.ac.jp> 於 2019年2月13日 下午6:57 寫道：
>
> Hello Richard, others,
>
> On 2019/02/08 00:37, r12a wrote:
>> https://www.w3.org/2019/02/07-i18n-minutes.html
>
>> richard: actually vertical text with 1 character per column is
>> really an urban myth
>
> I'm not sure about this. When you see 門生羅 (and not 羅生門, as it would be
> with LTR) at the top of a gate, the easiest way to explain why the
> characters are placed the way they are is that whoever wanted to write
> them was used to writing vertical lines (i.e. columns), with the columns
> going from right to left, but only was able to fit one character per
> 'line' (i.e. column).
>
> Of course, this was done rather implicitly and without calling each
> character a column. And we cannot go back to the people who wrote the
> name of a door on the top of a door in ancient times, but it's the best
> explanation we have so far. If you have a better one, I'd like to hear
> about it.
>
>> ... it's nonsensical
>
> It may be nonsensical (or not) as an approach to how to format it with
> CSS. It may also not apply to more 'modern' Chinese examples, in
> particular from Taiwan, where I think multiple-line RTL headings can
> also appear.
>
> Regards, Martin.
>
>> ... horizontal-tbrl was suggested
>> ... but that is inconsistent with writing modes
>> ... key issue is that japanese/chinese are strongly LTR
>> ... UBA relies on character properties
>> ... thinking is to use bidi override with direction=rtl
>> ... embedded numbers/text in <bdo>
>> ... but very similar to "close wrapping" recommendation we do
>> for normal bidi text
>
