Ligatured characters

From: William Overington (
Date: Thu Sep 07 2000 - 11:47:21 EDT

Having had some experience with handset metal type, years ago, I remember
that many founts had ligatured characters. Most founts had fi ff fl ffi ffl
each provided as one piece of type. Some founts had ct and st ligatures as
well, with a thin ornamental line connecting the top of the c or s and the
top of the t. Some founts such as Garamond Italic had in the specification
sheet a good collection of ligatured characters such as ct and st and
variant forms such as a swash lowercase e with a tail out to its right and
various swash versions of capital letters, such as N. Some founts had long
s characters and ligatures in the specification sheet. For example,
Baskerville and Goudy Text. Unfortunately the typefounders we found never
bought the matrices for the swash characters and the long s character and
the long s ligatures from the matrix makers so one could not actually buy
the ligatured characters (other than fi ff fl ffi ffl) as metal type for
hand typesetting. Certainly, more recently I have seen advertised newly
cast Baskerville metal type for sale in small quantities in hobbyist
letterpress printing circles with all the ligatured characters, including
the long s ligatures, as well as the long s character on its own.

Suppose that one is producing a program, such as a Java applet, to display
pages of printed text, and one wishes to encode text that contains
ligatures. Certainly this could be for original work where the decision as
to whether to use a ligature is made here and now. However, suppose that
one is wishing to transcribe an eighteenth century printed book and one
wishes to preserve the information as to when a long s was used and when a
ligature such as "long s and t" were used. How should one encode the text
in unicode please? For example, have unicode characters been defined for
ligatured characters? Are there special control characters to mean TURN
LIGATURING ON and TURN LIGATURING OFF so that one uses single letters to
code the text but can use control characters, which may be optionally
ignored, to preserve the extra information? Is there a control character to
that? What about the long s character, how is it represented please?

I am aware that there is a private use area available in unicode so that I
could, if there is no other method, define for local use, individual type
sorts to mean ct and st and so on, using software to provide equivalent
strings so that any text thus encoded can be converted to individual letter
form for such purposes as index forming. Am I right in thinking that it is
permissible to use the private usage area for control commands such as TURN
LIGATURING ON and TURN LIGATURING OFF rather than as codes of actual
characters if I so choose?

I would appreciate guidance please as to whether such an approach is
regarded as the elegant solution or whether there is some other better
regarded method. Also, if using the private use area are there any working
conventions that have arisen, perhaps, for example, that characters are
defined starting from the low address end of the private use area and that
control characters are defined starting from the high address end of the
private use area?

Also, does anyone know if one can obtain founts for a PC of Garamond and
Goudy Text that have all of the ligatured characters please? I have
Garamond but it only has fi and fl. Goudy Text is a face from the 1930s
based on the typeface of the Bible printed by Gutenberg with 42 lines to the
page. There was also a fount of alternative capitals for Goudy Text called
Lombardic Initials available in metal type. Is this fount available for a
PC as well please? The question arises that, when one has a fount such as
the Garamond fount where there are fi and fl ligatures included as to how
one accesses them when asking, say, a Java applet to drawString a string of
text. That is, if one has a fount file that has, say, a letter E in it,
then one knows that one can use the E character from a keyboard to mean the
letter. Given that one has a fount file available, such as a True Type
fount, how does one find out which character code accesses a character such
as the fi character that it contains please?

Do there happen to be any metal type cognoscenti in this group who happen to
know if one can purchase metal type of the ct and st and long s and long s
ligatures for Goudy Text in a variety of sizes please?

William Overington

7 September 2000

This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:13 EDT