Chess symbols, ZWJ, Opentype and holly type ornaments.

From: William Overington (
Date: Mon Jun 17 2002 - 06:17:28 EDT

I am interested in the possibility of making a TrueType fount so that an end
user may set chess positions.

I am aware that there exist the twelve Chess symbols U+2654 through to
U+265F in the Miscellaneous Symbols section.

However, in order to set chess positions one needs to have some pieces upon
a white square and some pieces upon a black square: a black square usually
being denoted by diagonal lines from top left to bottom right surrounding
the piece.

I seem to remember reading in relation to the musical notes that the Unicode
characters are not intended to enable the setting of music but only to refer
to a type of note within a text string. Is the reason why there are no
chess symbols "upon a black square" for a similar reason please or is there
some other reason in the sense that, in a parallel with the Unicode position
on ligatures where a ct ligature is not specified in regular Unicode but the
idea is that the ct ligature may be expressed by c ZWJ t, that a chess piece
on a black square is produced with a sequence of codes in some way? For
example, would U+2658 ZWJ U+25A0 be the way to produce a white knight upon a
black square?

I have in mind the possibility that I could prepare and publish a list of
Private Use Area code points for a chess fount suitable for producing a
diagram of a chess position, with all symbols being specified as being upon
a square body, as with a metal type chess fount. I have in mind using
U+E5B0 through to U+E5FF for these symbols, which provides me with the
opportunity to use some code points in the U+E5B0 through to U+E5BF range to
provide various edge pieces for the chess board and to use some code points
in the U+E5C0 through to U+E5FF range to provide a comprehensive set of
symbols so that, as well as both black and white pieces upon both black and
white squares together with an empty black square and an empty white square,
there are also alternative characters for rooks, knights, bishops and queens
and second alternative characters for queens. This will enable founts for
ornate "Isle of Lewis" style chess pieces to be produced where the knights
are men upon horseback and there are various designs. These various designs
do not affect the play of the game at all, they are just decorative
distinctions based upon the variations in the original solid chess pieces.

The ZWJ code can go between characters to indicate a ligature and cursive
joining. Is there a regular Unicode way of nominating that the next
character should be "swash if possible", so that an OpenType fount could
receive a chess position and an alternative knight symbol could be set as
"swash if possible" followed by a code point for an ordinary knight, so that
if the OpenType fount were for Isle of Lewis pieces then an alternative
knight glyph could be used, yet if the fount were for Staunton pieces the
ordinary glyph would be used for the piece. Or does that need to be done by
a sequence such as U+2658 ZWJ U+2766 as if the alternative knight were a
ligatured character. This would then mean that upon a black square that
alternative knight might be U+2658 ZWJ U+2766 ZWJ U+25A0.

So, it would appear that it would be possible to express the pieces for a
chess position in regular Unicode with just a few conventions, such as using
ZWJ U+25A0 and ZWJ U+25A1 following a chess symbol so as to indicate that a
square piece for setting in a chess diagram were needed and using ZWJ U+2766
so as to use an alternate glyph if one is available.

This would mean that the special glyphs would be internal to an OpenType
fount, yet could perhaps also have the Private Use Area code points assigned
to them as well so that people using older equipment could also use the
founts directly.

This raises the interesting procedural matter over the matter of publishing
lists of ligating pairs for such purposes, for, whilst publishing Private
Use Area code points is formally permitted, what is the status of publishing
suggested ligature style usage of two characters such that the combined
meaning is not immediately obvious please? For, although the sequence "c
ZWJ t" might reasonably be taken to mean a ct ligature by inspection, the
sequence U+2658 ZWJ U+2766 ZWJ U+25A0 is not obvious at all, though a
sequence such as U+2658 ZWJ U+25A0 might be deciphered by a reasonable
guess. This technique is potentially very powerful, for it automatically
defaults to ordinary Unicode, yet, in its way, adds a new type of Private
Use Area capability which can be used with regular Unicode codes yet the
combined characters could be anything! Yet the defaulting to Unicode could
look wrong upon a display without the special fount, whereas a sequence such
as c ZWJ t would not look wrong upon a display without the special fount.

Ideally, I would like to be able to publish a list of Private Use Area code
points and, for each, a meaning and also a corresponding suggested sequence
of regular Unicode points for a ZWJ sequence to produce the same result. I
feel, however, that that would not be a thing to do without a permission in
the rules. I wonder whether a reasonable thing would be to say that
publishing a Private Use definition of a ZWJ sequence involving a regular
Unicode character where the default of just the characters is not a correct
display is permitted provided that at least one of the characters in the ZWJ
sequence is itself a Private Use Area character. This would mean that the
Private Use nature of the usage in a plain text file would be apparent.

A related matter is holly type ornaments. In metal type one could get,
maybe one can still obtain, holly type ornaments where there are two items
of type, one which prints the leaves and one which prints the berries, the
idea being that one prints the leaves in green and the berries in red. Can
this be done with OpenType, in having a code such that a two colour
rendering takes place upon a colour display if so requested using a ZWJ
character? I have not managed to find any holly in Unicode, so perhaps
expressing U+274A as a red centre and green leaves could be used as a symbol
for discussion of the matter of the possibility of producing two colour
pairs in OpenType founts please. Indeed U+274A would be suitable for
discussing three colour triples.

A published code point set for chess might be useful for depicting chess
games in various styles of fount, from plain Staunton style to various
ornate designs, not just "Isle of Lewis" style.

Anyone who would like to make suggestions over which symbols to include is
welcome to send such suggestions to me by email please.

William Overington

17 June 2002

This archive was generated by hypermail 2.1.2 : Mon Jun 17 2002 - 04:48:54 EDT