ΩTimes and ΩHelvetica Fonts Under Development: Step One

Full text

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Yannis Haralambous
Atelier Fluxus Virus, 187, rue Nationale, F-59 800 Lille, France
Yannis.Haralambous@univ-lille1.fr
John Plaice
D´epartement d’informatique, Universit´e Laval, Ste-Foy (Qu´ebec) Canada G1K 7P4
John.Plaice@ift.ulaval.ca
TheTruthIsOutThere
—Chris CARTER, The X-Files (1993)
Introduction
ΩTimes and ΩHelvetica will be public domain
virtual Times- and Helvetica-like fonts based upon
real PostScript fonts, which we call “Glyph Con-
tainers”. They will contain all necessary characters
for typesetting efficiently (that is, with T
E
X quality)
in all languages and systems using the Latin, Greek,
Cyrillic, Arabic, Hebrew and Tifinagh alphabets and
their derivatives. All Unicode characters will be
covered, although the set of glyphs of our Ω fonts
will not be limited to these; after all, our goal is high-
quality typography, which requires more glyphs than
would be required for mere information interchange.
Other alphabets will follow (the obvious first
candidates are Coptic, Armenian and Georgian) as
well as mathematical symbols, dingbats, etc.
Why PostScript instead of ?-
is the ultimate tool for font development.
Extending Computer Modern fonts to the Unicode
encoding is still one of our goals. We have started
developing a set of fonts we call “Unicode Com-
puter Modern” fonts, using techniques such as Vir-
tual (virtual fonts are created directly
by using the gftotxt utility for text
output). Nevertheless, we realized that the task
of writing code for some thousands of
characters (including code for typewriter style) is a
tremendous task, which will take several years.
So we have decided to take a small break
from ing, and to develop in a limited
time period PostScript fonts that will cover a
maximum number of languages and will give the
T
E
X community a good reason to switch to Ω.
Why Times and Helvetica? First of all because,
after Computer Modern, they are the most widely
used fonts in the T
E
X community. Many journals
and publishers request that their texts be typeset
in Times; Helvetica (especially the bold series) is
often used as a titling font. Like Computer Modern,
Times is a very neutral font that can be used in a
wide range of documents, ranging from poetry to
technical documentation. . .
It would surely be more fun to prepare a
Bembo- or Stempel Garamond-like font for the serifs
part and a Gill Sans- or Univers-like one for the
sans-serifs part; but these can hardly be used in the
scientific/technical area, and that’s perhaps where
T
EX (and hence potentially Ω) is used the most.
When will the Ωfonts be finished? The
development of ΩTimes and ΩHelvetica fonts is
divided into four steps:
1. Drawing of PostScript outlines and packaging
of Glyph Container fonts.
2. Development of virtual code, based on the real
fonts of step one.
3. Kerning of virtual fonts.
4. Development of L
A
T
E
X code and Ω Translation
Processes necessary for the use of these fonts.
For the time being (June 1996) we have done
the biggest part of step one, and this is what we
present in this paper. We hope to have finished with
steps two, three and four before the next teT
E
XCD-
ROM in December 1996.
We want your support! Please keep in mind:
The choice and shapes of glyphs presented in
this paper are only a first attempt. We need
your feedback to improve them, so that you
can use them efficiently.
In the tables we present only Times family and
medium series fonts (except in the case of Tifinagh,
which is also presented in the Helvetica family).
Up-to-date tables of the remaining fonts can be
consulted on our Ω WWW server
http://www.ens.fr/omega
126 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Figure 1: Italic-style letters with ogonek (Polish
and Lithuanian)
You can also retrieve the PostScript code from the
same address, or from
ftp://ftp.ens.fr/pub/tex/yannis/omega
General remarks on the fonts
To prevent confusion, the word “font” in this section
is meant in the sense of the PostScript Type 1 font
structure; and not of T
E
X text or math fonts: the
fonts we describe in this paper will never be used
directly for typesetting. Their raison d’ˆetre is to
provide glyphs for the virtual Unicode+Typography
Ω fonts which we will develop in steps two–four.
Hence, there is no need to look in the tables for
a‘´e’: this character will be assembled by the virtual
font, using the glyphs of letter ‘e’ and of the acute
accent.
The same stands for the letter ‘c’ with cedilla:
it can be assembled out of the two corresponding
glyphs; however, this is not true for letters with
ogonek: the shape of the ogonek changes while it
gets attached to the letter; that is why you find
letters with ogonek in the Glyph Containers and
not letters with cedilla. In Fig. 1 the reader can
see examples of letters in italic style, carrying an
ogonek accent.
In Fig. 2 the reader can find the general
structure of the fonts:1On the left, the 16-bit
Unicode+Typography virtual font, on the right a
certain number of Glyph Containers, that is 8-bit
PostScript fonts.
The reader will notice that a certain number of
glyphs are repeated in the different Glyph Contain-
ers. This is because we want to minimize the number
of Glyph Containers used for a single-alphabet text.
For example, accents for all fonts are stored in Glyph
1In the figure, the reader will notice real font “Adobe
Zapf Dingbats”. In fact, the glyphs of this font have become
Unicode characters (0x2701–0x27be) and we see no reason to
redraw them since this font is widely available. Hence the
virtual 16-bit font will also point to the standard Adobe Zapf
Dingbats font.
Container “Common”; theoretically, to produce an
acute-accented Latin letter and an acute-accented
Cyrillic letter, one would use three Glyph Contain-
ers: one for the accent, and one for each alphabet.
To avoid this, we store all accents relevant to a given
alphabet, in the alphabet’s Glyph Container. The
same method is used for shapes that are similar in
the different alphabets: Latin, Greek and Cyrillic al-
phabets share the letter ‘A’, Latin, IPA and Cyrillic
alphabets share the letter ‘a’.2
The “Common” Glyph Container
The “Common” Glyph Container, shown in Table 1,
contains glyphs that will be used potentially in
conjunction with all alphabets. These are described
in the following subsections.
Punctuation, digits, editorial marks Special
care has been taken to distinguish between “typo-
graphical” punctuation and “typewriter/computer
terminal-derived” one: compare the typographical
double quotes lm and the straight ‘ASCII’ones".
This table covers all Unicode punctuation
marks from the ASCII and ISO 8859-1 tables as
well as from the general punctuation table
(0x2010–0x2046). We have not included a few
punctuation marks specific to a single alphabet:
Arabic asterisk, inverted comma and semicolon,
Hebrew colon, Greek upper dot. These will be found
in the corresponding Glyph Containers.
In Fig. 3 the reader can see how regular
curly braces have been transformed into square
brackets with quill, by simply reflecting the
central part of the brace.
Commonly adopted Latin alphabet derived
symbols Symbols like Ause Latin alphabet letters
but are used in many non-Latin-alphabet based
languages. Symbol zis an even stranger example:
although the glyph ‘N’ does not exist in Cyrillic,
this symbol is used mainly in Cyrillic-alphabet
languages.
2We will use a totally different approach when dvips
and Adobe Acrobat are able to use 16-bit PostScript fonts.
Instead of having many ‘small’ PostScript fonts and one
‘big’ virtual font, we will use a single ‘big’ PostScript font,
in Unicode+Typography encoding. In that font, accented
letters and repeated identical shapes will be obtained by the
PostScript font technique of composite characters. This will
allow Acrobat users to select and copy Unicode-encoded text
directly from the document window.
Since the two techniques (16-bit + 8-bit real, vs. 16-
bit composite real) will share the same .tfm metrics for
characters, it will be possible to convert .dvi files from one
format to the other, so that files obtained today by the
first method will be “Acrobat-ready” later, whenever Acrobat
switches to Unicode (hopefully soon!).
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 127

Yannis Haralambous and John Plaice
OmegaTimesCommon
OmegaTimesLatin
OmegaTimesIPA
OmegaTimesGreek
OmegaTimesCyrillic
OmegaTimesHebrew
OmegaTimesTifinagh
Adobe™ ZapfDingbats
…
ΩTimes
virtual font
(Unicode
encoding +
characters
needed for
typography)
PostScript fonts
16-bit
8-bit
8-bit
8-bit
8-bit
8-bit
8-bit
8-bit
8-bit
8-bit
OmegaTimesArabicOne, …
Figure 2: General structure of the ΩTimes fonts (idem for ΩHelvetica)
Figure 3:Theleft and right square
brackets with quill were drawn by reflecting
the central part of regular curly braces
In Fig. 4 the reader can see our small tribute
to the GNU foundation: the “copyleft” symbol. We
hope that this character will soon be included in the
letter-like symbols section of Unicode.
—Q. Why not use the \reflectbox macro to
reflect the “copyright” glyph into a “copyleft” one?
— A. Because we want to treat “copyleft” as a
separate character in the virtual font, which may
be searched inside a .dvi or PDF file. In other
cases, such as
B
,usedintheA
B
BA logo and the
cobar construction in Algebraic Topology, one can
128 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Figure 4: The GNU “copyleft” symbol
Figure 5:Theestimated symbol was drawn
inside a perfect circle
easily use PostScript-manipulation macros without
damage.
Finally, in Fig. 5 the reader can see how the
estimated symbol fits inside a perfect circle (in
red, for readers of a color version of this paper).
Currency symbols Among currencies having
proprietary symbols, there are strong and weaker
ones. The strong ones have made it into ISO 8859-
1 (you-know-who made it even into ASCII itself,
and is used by a well-known typesetting language
Letter ‘O’
Letter ‘C’
Letter ‘e’
C
e
Figure 6: How shapes ‘C’, ‘O’ and ‘e’ were used
for the design of the latin and cyrillic capital
letter shwa
to enter math mode...), the remaining ones have
found their home in the currency symbols section
of Unicode.
We have included them all (even the Thai
currency symbol è, which looks suspiciously Latin)
in the “Common” Glyph Container. Note that the
symbol Ñfor French Franc is virtually unknown in
France. . .
Diacritics The zone 0xa0–0xe2 of the “Common”
Glyph Container is dedicated to combining diacrit-
ics. These diacritics are supposed to be useful
for more than one alphabet; whenever a diacritic
belongs specifically to one alphabet, it has been
included only in the corresponding Glyph Container
(this is the case, for example, of Vietnamese dou-
ble accents, Greek spirit+accent combinations, and
Slavonic accents). Thanks to the diacritics in the
“Common” Glyph Container we will be able to con-
struct all latin extended additional Unicode
characters (0x1e00–0x1ef9) virtually, by combining
them with letters from the “Latin” Glyph Con-
tainer. This Unicode region covers Welsh, Viet-
namese and transcriptions of Indic and other lan-
guages.
The “Latin” Glyph Container
The “Latin” Glyph Container, shown in Table 2,
contains glyphs of letters for Latin alphabet lan-
guages. These are described in the following sub-
sections.
Letters for Latin alphabet languages All
glyphs necessary to typeset Western and Central
European, Nordic, Baltic, African languages, Viet-
namese and Zhuang.
Some African characters are derived from the
International Phonetic Alphabet. It is a fascinating
challenge to design uppercase and italic-style forms
for these characters (for example, see in Fig. 6 the
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 129

Yannis Haralambous and John Plaice
Figure 7: The African letter f with hook in
roman upper- and lowercase, as well as in italic
lowercase form
Figure 8:Capital letter b with topbar
(which shares the same glyph as cyrillic
capital letter be)small letter b with
topbar and cyrillic small letter be
uppercase version of letter ©letter which derives
from the phonetic shwa, and in Fig. 7 the straight
uppercase and straight lowercase versions of an
African letter that becomes a standard fin italic
style).
Sometimes, although a Greek form (or a form
derived from Greek) is used for the lowercase, the
uppercase does not follow the Greek model: for
example, the uppercase of African letters ™and ≠
(the former is 100% Greek, while the latter looks
more like a “phonetic gamma”) are äand ç.
Inversely, sometimes a Greek form is used for
the uppercase only: ôis the uppercase form of the
integral-like π, a character derived from the IPA,for
which one can hardly imagine an obvious uppercase
form — taking a Greek letter for that purpose is the
easiest solution.
It is quite interesting how the lowercase of the
Latin letter Ådiffers from that of the Cyrillic letter
sharing the same glyph (see Fig. 8).
Special attention has been paid to the notorious
Dutch ligature ‘ij’, in italic Times lowercase form
and in Helvetica uppercase form (see Fig. 9).
Figure 9: Italic letter ‘y’ with umlaut accent
followed by the Dutch ligature ‘ij’ in Times italic,
Helvetica medium uppercase and Helvetica bold
uppercase form
Figure 10: The Vietnamese small letter o
with horn was designed using the glyphs of the
ring accent, the apostrophe and letter ‘o’
In the former case, we have connected the two
letters, creating an intentional confusion with a ‘y’
letter with umlaut accent. In Fig. 9 the reader
can compare these two constructions. On the other
hand, the shape of the uppercase sans serif ‘IJ’ is
derived from that of the letter ‘U’: this is common
Dutch titling practice.
The characters ^and ~are used in Breton.
They are not (yet) included in Unicode, and have
been brought to our attention by Jacques Andr´e.
Finally, in Fig. 10 the reader can see how the
“horn” of Vietnamese letters has been designed,
using graphical elements from the font: in this case,
the ring accent and the apostrophe.
Ligatures Besides the “standard five” ligatures Ã,
Õ,Œ,œ,–, we have included ligatures for the case
where:
•the second or third letter is an ‘i’ with ogonek:
—,“, useful in Lithuanian;
•the second or third letter is a stroked ‘l’ : ”,‘,
useful in Polish;
•the second or third letter is a ‘j’: ’,÷;
•instead of an ‘i’ one has an ‘ij’ ligature: ◊,ÿ,
useful in Dutch;
•instead of ‘f’ one has a long ‘s’: Ÿ,⁄,¤,‹,
›. We haven’t (yet) included any ‘f’ + ‘long s’
combinations.
130 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Figure 11: French ligatures ‘st’ and ‘ct’ in Times
and Helvetica fonts
Finally we couldn’t resist the temptation of
making “French” ligatures fi,fl, of course, both
in Times and Helvetica (!) font families. These
ligatures are well-known because of their use (in the
Garamond typeface) in the Pl´eiade book collection.
One can argue about their reason for being in
Times (and especially in Helvetica) style, which
has absolutely no historical background. . . Consider
them an experiment, and trust us for not making
them automatic in default text mode!
Diacritics Accents 0xf8–0xff are used for Viet-
namese only. Diacritics occupying positions 0xe8–
0xf6 are shared with the “Common” Glyph Con-
tainer.
The “IPA” Glyph Container
The “IPA” Glyph Container, shown in Table 3,
contains a collection of glyphs needed for the
International Phonetic Alphabet. They have been
found in different sources: Unicode encoding,
literature on phonetics (in particular we covered the
complete table of characters of the French classic
Initiation `a la phon´etique, by J.-M.-C. Thomas, L.
Bouquiaux and F. Cloarec-Heiss, PUF, 1976). Only
a small number of these characters are contained in
the Unicode encoding. We would be grateful for
any feedback from scholars on additional characters
or corrections of the existing ones.
Since IPA is so. . . international, we have as-
sumed that one can encounter phonetic insertions
in text written in any language. Hence, we have in-
cluded all glyphs needed, including lowercase Latin
alphabet letters, and small capitals— the latter be-
ing included only whenever these are significantly
different from their lowercase counterparts: includ-
ing, for example, small caps ‘s’, ‘x’, ‘z’ would be
Figure 13: Three closely-related glyphs: German
‘sz’, IPA ‘beta’ (not a Unicode character), Greek
lowercase ‘beta’
Figure 14: Different types of ‘gamma’: the first
one from the Greek alphabet, the others from the
IPA
useless since they are indistinguishable from lower-
case ‘s’, ‘x’, ‘z’.
This point deserves some explanation: one
should not confuse small capitals used for text and
IPA small capitals. The former are a stylistic
enhancement of text; they appear only in words
entirely in small capitals style; their height is not
equal to the x-height of the font, generally they are
slightly higher. The latter are phonetic characters
used together with authentic lowercase letters: they
must have exactly the same height.
In Fig. 12 the reader can see some small caps we
have designed for the ΩTimes IPA Glyph Container.
It may not be obvious from the figure, but stroke
widths of the small caps are exactly the same as
the ones of lowercase letters (see lowercase letter ‘a’
to compare, as well as glyph ‘v’, which is used to
represent both a lowercase and a small caps ‘v’). On
the second line of the figure, one can see the same
letters obtained as ‘fake’ small capitals.
We had a lot of fun designing these characters.
In some cases, they just had to look a little different
from their Greek counterparts: for example, in
Fig. 13 the reader can see how the ‘phonetic beta’
has been inspired by the German ‘sz’, and not by the
‘real’ Greek beta; in Fig. 14 we present a collection
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 131

Yannis Haralambous and John Plaice
Figure 12: On the first line, lowercase letters ‘a’ and ‘v’ and specially designed IPA small capitals; on the
second line, uppercase letters reduced 68%
Figure 15:TwoIPA symbols of different origin:
inverted letter ‘f’ and dotless ‘j’ with stroke
of gamma-like glyphs, as well as the ‘real’ Greek
letter gamma.
In some cases we were not very sure about
the origin of some IPA characters and made several
attempts: for example, in Fig. 15 one can see two
symbols with a superficial resemblance: an inverted
‘f’ and a dotless stroked ‘j’. Which one is used in
phonetics? The choice is left to the user.
Finally there was one case where we had to solve
areal design problem: the one of the ‘l with retroflex
hook’ and ‘ezh’ ligature (not a Unicode character).
In most real-life examples we had the opportunity
to see, the letter ‘ezh’ was sadly squeezed so that its
tail remains higher than the retroflex hook of the ‘l’;
we find it bad typographical practice to squeeze the
‘ezh’ and propose three solutions (only one of which
Figure 16: Which one is the best ‘l with retroflex
hook+ezh’ ligature? (Three proposals)
of course will survive in the final Glyph Container).
In Fig. 16 the reader can see: (a) the tail of ‘ezh’
merged with the retroflex hook of the ‘l’, (b) the
retroflex hook of the ‘l’ continuing deep enough for
it to be seen under the tail of ‘ezh’, (c) the retroflex
hook rising higher so that it fits between the tail of
‘ezh’ and the baseline. In cases (a) and (c) the risk
may be that the ‘l with retroflex hook’ be taken for
an ‘ordinary l’ (compare ù,¯and ö); in case (b) we
are going too deep under the baseline. . .
Anyhow, we expect your feedback to resolve
this issue.
The “Greek” Glyph Container
The “Greek” Glyph Container, shown in Table 4,
contains glyphs needed for the Greek language,
ancient and modern. The problem with most
132 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Figure 17: Two forms of Greek circumflex accent:
the first used in modern Greek typesetting, the
second one used in a number of scholarly typefaces
commercial Greek fonts is that they are either made
for modern Greek use, or for ancient Greek use by
non-Greek scholars. There is a third possibility:
fonts made for ancient Greek use by Greek scholars.
In this Glyph Container we have tried to satisfy
all three categories of users. Of course, this font
can be used both for monotonic and polytonic text
(there is a straight monotonic accent, while the
acute one can be used as well). But we have
gone even farther, by including two versions of the
circumflex accent, shown in Fig. 17: the tilde-like
one, used in Greece, and the cap-like one used in
scholarly Western editions.3
Faithful to the first TUGboat paper by one of
the authors (Haralambous and Thull, 1989), we have
included the inverted iota with circumflex accent,
found in certain 19th-century editions of modern
Greek (see Fig. 18).
Version 1.0 of the Unicode standard contained
uppercase versions of Greek numerals, just as roman
numerals were provided in upper- and lowercase;
in ISO 10646 these characters were removed, ap-
parently by decision of the Greek delegation. We
find this action absurd (there are many examples of
uppercase numerals in literature) and have of course
included the relevant glyphs in the Glyph Container.
3This brings us to a nice typographical joke: the Greek
“circumflex” accent is either “tilde”-like or “cap”-like, but
never actually. . . “circumflex”-like!
Figure 18: The inverted ‘iota with circumflex’
(not a Unicode character) used in 19th-century
modern Greek printing
Figure 20: The different variant forms of Greek
letters: beta, theta, phi, rho, kappa
In fact, we have included all known variants of Greek
numerals: stigma, digamma, qoppa and sampi (see
Fig. 19, next page), and the upper and lower nu-
meral signs.
Greek letters are also very common in math-
ematics and physics. Some variant forms are used
for different purposes in these fields (rho with curved
or straight tail, open/closed phi, open/closed theta,
curly or straight kappa). To these we add a variant
form used in regular text: the initial and medial
beta (bvs. 1). All variant forms of lowercase letters
are shown in Fig. 20.
There is also a variant form of the letter sigma:
the so-called “lunate” sigma. This character is used
in some scholarly editions to avoid the distinction
between ordinary medial and final sigmas. In Fig. 21
we compare it with Latin letter ‘c’: the lowercase of
lunate sigma has no bulb and the upper one, no serif
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 133

Yannis Haralambous and John Plaice
Figure 19: Collection of glyphs used for Greek numerals; on the first line: lowercase stigma, digamma (2
variants), qoppa (2 variants), sampi; on the second line: uppercase stigma, digamma, qoppa, sampi
Figure 21: The glyphs of Greek lunate sigma and
Latin letter ‘c’ (in upper- and lowercase)
(and hence the Greek and Latin letters are identical
in the Helvetica family).
The reader may ask why on positions 0x80–
0x8d and 0xa0–0xad of the Glyph Container table
we have accented letters, while on positions 0x90–
0x9d there are only accents and no letters. The
answer is very simple: accents on Greek letters
sometimes change shape according to the width of
the letter. Greek vowels can be roughly divided
into three width classes: narrow ones (the iota),
wide ones (the omega) and medium ones (all the
remaining). On row 8and awe have placed accents
for narrow and wide letters, respectively; on row
9we have placed the ordinary accents. And since
rows 8and ahave been made for unique vowels,
we have prefered to include complete accent+letter
combinations (letters being aliases, of course) so
that the virtual font doesn’t need to make the
construction.
The same reason justifies positions 0x5e and
0x5f: the dieresis (dialytika, in Greek) does not
have the same width, depending on whether it is
placed on an iota or an upsilon.
Finally, in the table there are also the specifi-
cally Greek guillemets (round ones), the upper dot,
and the two forms of “subscript” iota: for low-
ercase letters (ypogegrammeni) and for uppercase
ones (prosgegrammeni).
The “Cyrillic” Glyph Container
The “Cyrillic” Glyph Container, shown in Table 5,
contains glyphs needed for all languages using
the Cyrillic alphabet, whether European or Asian.
Characters for pre-Lenin Russian (fita, izhitsa, yat)
have also been included, as well as “modernized”
versions of Slavonic characters. As in the case of the
Latin ‘st’, ‘ct’ ligatures, one may argue the necessity
of modernizing Slavonic script, especially when it
comes to drawing the Helvetica version. It happens
that these characters have been included in Unicode
(like the Coptic ones), and this is already reason
enough to draw the font; it is up to the user to
actually adopt the “modern” font, or use beautiful
traditional Slavonic typefaces.
What was fascinating when designing Slavonic
letters was their relation to Greek ones. In Fig. 22
we compare Slavonic and Greek Xi and Psi: at
a first glance, Slavonic Xi bears a resemblance
to Greek lowercase xi, it is quite surprising that
134 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
Figure 22: Comparing Slavonic letters Xi and Psi
with the corresponding Greek ones
the Slavonic xi is reflected with respect to the
Greek one. Uppercase Psi letters are identical,
and lowercase Slavonic psi has a heavier stem
(with serif, while Greek lowercase letters never have
serifs). Uppercase Slavonic Omega +is a magnified
lowercase omega (where we have placed a serif on
the central stem). For the lower part of the Slavonic
Yus /we have used an inverted Psi: this is a small
designer’s secret allowing better integration of the
letter in the Times (resp. Helvetica) font family.
Another design initiative of ours was to use
graphic elements from the Serbian letter !1,for
the Asian Cyrillic åú,§¥,Øø. We expect user
feedback to validate or deny this choice.
The “Arabic” Glyph Container
The “Arabic” Glyph Container, shown in Tables 6–
8 contains the glyphs that are necessary to typeset
in any language using the Arabic alphabet.4The
design of these Ω Arabic fonts doesn’t actually
have much in common the Times and Helvetica
families; in fact, we have used popular modern
designs, which can be used both for technical and
literary text, and which allow easy readability in
small sizes. Fat and thin stroke width, ascender
height and descender depth have all been calculated
with respect to the corresponding parameters of the
Latin/Greek/Cyrillic fonts.
In Fig. 23 the reader can see the metrical
relationship between the Latin, Arabic and Hebrew
fonts: we want these characters to fit with one other,
4We have also included undotted versions of letters ba,
fa, qaf, for typesetting of old manuscripts.
so that multilingual texts using the three scripts will
produce typographically acceptable results.
Concerning diacritics we have included all
vowels, and combinations with shadda and hamza,
as well as some special cases: madda, wesla, vertical
fatha, vertical fatha + shadda, and other diacritics
used in Arabic spellings of African languages,
Kurdish, Baluchi, Kashmiri, Uigur and Kazakh.5
Esthetic Ligatures We have included a very small
number of esthetic ligatures (fewer than 150): ba-
like letters followed by a final noon-like, initial fa-
like letters followed by a ya-like, an initial lam-
meem ligature, and the llah ligature with and
without vertical fatha + shadda. We are not
convinced that heavy ligaturing of these fonts, in
the manner of traditional Naskhi fonts, would be
esthetically judicious. Nevertheless, we are open to
any suggestion on possible ligatures that might be
added.
The “Tifinagh” Glyph Container
The “Tifinagh” Glyph Container, shown in Tables 9
and 10, contains the glyphs needed for typesetting
the Tamazight (alias Berber) language. A complete
description of the Berber T
E
X system developed by
Haralambous [2].
The glyphs shown on the tables warrant some
explanation. Tifinagh script has always been
written in a non-serif style. On table 9 we show
a “Helvetica” version of the script, in the sense
that everything has been done to bring these glyphs
closer to the Latin/Cyrillic/Greek Helvetica types.
This approach is quite safe, and — in all modesty —
the result should not surprise any speaker of Berber.
On the other hand, table 10 shows a 100%
experimental font! The main idea was to say:
“What would Tifinagh letters look like today if
they had followed the same evolution as Latin
ones?” See Fig. 24 for a closer comparison
of some Tifinagh letters in both Helvetica and
Times styles. Admittedly, the result seems weird,
and a lot of corrections have to be made before
these drawings actually can be called Tifinagh
glyphs. . . Nevertheless, the Berber language is
being typeset more and more in its traditional
script, often together with other scripts. As a
result, the problem of homogenization with Western
typographical traditions must be faced; the style of
the Times typeface is one of the first challenges.
5We have not included diacritics used for the Qur’¯an,
as we believe that these should be printed using very specific
traditional typefaces; nevertheless, the diacritics provided are
sufficient for excerpts from the Qur’¯an.
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 135

Yannis Haralambous and John Plaice
Figure 23: Capital height, x-height, baseline and descender depth of Latin, Arabic and Hebrew letters
Figure 24: On line 1: Tifinagh letters in Helvetica style, with shapes identical to Greek and Latin ones;
on line 2: idem. but shapes become stranger; on lines 3 and 4: the same letters, in Times style, using
“evolutional logic”
The “Hebrew” Glyph Container
The “Hebrew” Glyph Container, shown in Table 11,
contains the glyphs needed for Hebrew, Yiddish
(both in the Russian or American YIVO spelling)
and Ladino. The font design is based on a —
very popular in Israel — modern Hebrew typeface.
Once again we have adapted the stroke widths
and character dimensions of Latin/Greek/Cyrillic
glyphs. In Fig. 23, the reader can compare the size
and weight of Hebrew letters with those of Latin and
Arabic.6
We have not included Masoretic signs in the
font, because we believe that the Bible should be
typeset in traditional “square” fonts. Nevertheless,
we have included letters with dagesh which appear
only in a few isolated positions in the Bible, as well
as the inverted nun. We have also included two
6It should be noted that the height of Hebrew letters is
exactly equal to the half distance between that of upper- and
lowercase Latin letters.
136 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
different forms of the lamed-aleph ligature (with and
without left stroke), used in older texts.
References
[1] Haralambous, Yannis and Klaus Thull. “Type-
setting modern Greek with 128-character codes.”
TUGboat 10,3 (1989), pages 354 — 359.
[2] Haralambous, Yannis. Un syst`eme T
E
X
berb`ere. Actes de la table ronde internationale
pPhonologie et notation usuelle dans le domaine
berb`ere q. Paris: Institut National des Langues
et Civilisations Orientales, 1993. [Forthcoming
in the Cahiers GUTenberg thematic issue on
Semitic scripts.]
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 137

Yannis Haralambous and John Plaice
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
» …  À Ã Õ Œ œ
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ Ÿ ⁄ ¤ ‹ › fi fl
"Ex ‡ ·
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 1: Tentative OmegaTimesCommon Glyph Container Table (June 10, 1996).
138 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
ò ô ö õ ú ù û ü
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
» Ã Õ Œ œ
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ Ÿ ⁄ ¤ ‹ › fi fl
"Ex ‡ · ‚ „ ‰ Â Ê Á
Ë È Í Î Ï Ì Ó Ô
"Fx  Ò Ú Û Ù ı ˆ ˜
¯ ˘ ˙ ˚ ¸ ˝ ˛ ˇ
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 2: Tentative OmegaTimesLatin Glyph Container Table (June 10, 1996).
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 139

Yannis Haralambous and John Plaice
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
ò ô ö õ ú ù û ü
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
» …  À Ã Õ Œ œ
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ Ÿ ⁄ ¤ ‹ › fi fl
"Ex ‡ · ‚ „ ‰ Â Ê Á
Ë È Í Î Ï Ì Ó Ô
"Fx  Ò Ú Û Ù ı ˆ ˜
¯
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 3: Tentative OmegaTimesIPA Glyph Container Table (June 10, 1996).
140 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"3x 0 1 2 3 4 5 6 7
8 9 ; < = > ?
"4x A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
òôöõúùû
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 4: Tentative OmegaTimesGreek Glyph Container Table (June 10, 1996).
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 141

Yannis Haralambous and John Plaice
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
ò ô ö õ ú ù û ü
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
»
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ
"Ex ‡ · ‚ „ ‰ Â Ê Á
ËÈÍ
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 5: Tentative OmegaTimesCyrillic Glyph Container Table (June 10, 1996).
142 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
"3x 0 1 2 3 4 5 6 7
8 9 > ?
"4x @ABCDEFG
HIJKLMNO
"5x PQRSTUVW
XYZ[\]^_
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
ò ô ö õ ú ù û ü
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
» …  À Ã Õ Œ œ
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ Ÿ ⁄ ¤ ‹ › fi fl
"Ex ‡ · ‚ „ ‰ Â Ê Á
Ë È Í Î Ï Ì Ó Ô
"Fx  Ò Ú Û Ù ı ˆ ˜
¯ ˘ ˙ ˚ ¸ ˝ ˛ ˇ
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 6: Tentative OmegaTimesArabicOne Glyph Container Table (June 10, 1996).
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 143

Yannis Haralambous and John Plaice
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x ` a b c d e f g
h i j k l m n o
"7x p q r s t u v w
x y z { | } ~
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
ò ô ö õ ú ù û ü
"Ax † ° ¢ £ § • ¶ ß
® © ™ ´ ¨ ≠ Æ Ø
"Bx ∞ ± ≤ ≥ ¥ µ ∂ ∑
∏ π ∫ ª º Ω æ ø
"Cx ¿ ¡ ¬ √ ƒ ≈ ∆ «
» …  À Ã Õ Œ œ
"Dx – — “ ” ‘ ’ ÷ ◊
ÿ Ÿ ⁄ ¤ ‹ › fi fl
"Ex ‡ · ‚ „ ‰ Â Ê Á
Ë È Í Î Ï Ì Ó Ô
"Fx  Ò Ú Û Ù ı ˆ ˜
¯ ˘ ˙ ˚ ¸ ˝ ˛ ˇ
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 7: Tentative OmegaTimesArabicTwo Glyph Container Table (June 10, 1996).
144 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting

ΩTimes and ΩHelvetica Fonts Under Development: Step One
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * +
"8x Ä Å Ç É Ñ Ö Ü á
à â ä ã å ç é è
"9x ê ë í ì î ï ñ ó
òôöõú
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 8: Tentative OmegaTimesArabicThree Glyph Container Table (June 10, 1996).
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x ` a b c d e f g
h i j k l m
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 9: Tentative OmegaHelveticaTifinagh Glyph Container Table (June 10, 1996).
TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting 145

Yannis Haralambous and John Plaice
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " # $ % & '
( ) * + , - . /
"3x 0 1 2 3 4 5 6 7
8 9 : ; < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \ ] ^ _
"6x ` a b c d e f g
h i j k l m
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 10: Tentative OmegaTimesTifinagh Glyph Container Table (June 10, 1996).
"x0 "x1 "x2 "x3 "x4 "x5 "x6 "x7
"2x ! " $ % & '
( ) * + , - .
"3x 0 1 2 3 4 5 6 7
8 9 : < = > ?
"4x @ A B C D E F G
H I J K L M N O
"5x P Q R S T U V W
X Y Z [ \
"6x ` a b c d e f g
h i j k l m n o
"7x p q r t v w
x y z { | } ~
"x8 "x9 "xA "xB "xC "xD "xE "xF
Table 11: Tentative OmegaTimesHebrew Glyph Container Table (June 10, 1996).
146 TUGboat, 17, Number 2 — Proceedings of the 1996 Annual Meeting