Paris, 24/06/2001

This distribution contains a new set of TrueType fonts for assyriologists. They are based on the Palatino font family that came with MacOS 7. This document contains a number of reflections on the ideas behind and the use of these fonts. At the end is an ad-hoc disclaimer, and a glossary with attempts at explication of some technical jargon.

The underlying idea/concept for the use of these fonts is that there are two subfamilies, one for "normal" text in a modern, western language, the other for transcriptions. The first of these, Palatino Nisaba, can be used for normal text, but lacks all special transcription symbols (this subfamily is named after the Sumerian goddess of writing, as the most profound ideas, insights, etc., will probably be typeset in this typeface :^)). The second, Palatino Cuneitype, contains the symbols necessary for writing transcriptions, but misses some of the characters that may be needed in normal text (this subfamily is named after the CunéiType1 font that set my whole effort in motion… ).

Français:

Voici une distribution contenant des nouvelles polices pour l'Assyriology, basées sur le fameux Palatino. Ce document donne un certain nombre d'informations nécessaires aux bon usage de ces polices. Des informations plus détaillées (et techniques) sont données en Anglais uniquement.

Il y a 2 familles: Palatino Nisaba (PN) et Palatino Cuneitype (PC). La première est la police à utiliser pour le texte "normal"; elle contient toutes les caractères nécessaires dans la plupart des langages de publication. La deuxième famille (PC) est à utiliser pour les transcriptions: elle contient les diactriques, mais aussi suffisamment de caractères "normaux" pour éviter des allers/retours trop fréquents entre PC et PN. Les 2 familles sont parfaitement identique en ce qui concerne la taille, aspect, etc: les changements entre les membres ne sont pas visibles. Elles partagent un autre atout: il y a une parfaite compatibilité possible entre le pc et le macintosh. Ceci est assuré par le fait que pour chaque caractère dans la police pc, il existe la même caractère dans la police macintosh, et vice versa.

Why 2 families, when other authors have provided single families containing all the necessary symbols? I started working on my fonts when it became clear that documents set in the aforementioned CunéiType1 font for Macintosh can not in any easy way be transported to a PC. A lot of conversion is necessary, including of that font family itself. Anyone familiar with both Mac and PC knows that the order of the letters in the respective "alphabets" is not identical between these platforms. In addition, each of these alphabets (encodings: each letter/symbol has its own code, from 0 to 255, with the first 30 being so-called control characters) contains unique symbols that don't exist in the other. When a document from one platform is opened on the other (say, a Mac document on a PC), programmes like MS Word will examine each character code encountered, look it up in an internal table of Macintosh characters, and exchange it for the code that will should give the same character on the PC. Platform-unique characters can not be translated, and for there, the result is undefined (often an open square, the "whatever" symbol).

The main reason for using a 2-family principle is that it allows documents to be exchanged between Mac and PC with the least amount of risk, because no attempt has to be made to squeeze all symbols for both regular text and transcriptions into a single font. To ensure maximum exchangeability, both subfamilies contain only those characters that exist on the two different platforms (Mac and PC). And all "normal" characters are at their normal position within the font. In the Palatino Cuneitype fonts, several characters have been changed for a transcription character. This has been done in such a way that a maximum of "normal text" characters have been preserved, for use in English, French, and (to a lesser extent) German. Also, where possible, lower~ and uppercase versions of a character (say, a sjhin), are stored at positions that originally stored a lower/uppercase letter pair. This is to make it possible to convert a typed lowercase letter to its uppercase variant with one of the various wordprocessor functions that do this (e.g. enabling smallcaps mode). NB: this is often not possible with CunéiType1, not even for some often used characters (é vs. É…)

Trouble shooting:
20000808: Their may be a compatibility issue with the 2subscript character, that occupies the position normally reserved for the currency symbol (¤); several PC programmes do not correctly import this character from a Macintosh MS Word 5 document. I am investigating this, and may eventually change the encoding of the 2subscript. 20000809: reading a Macintosh RTF document (with {\rtf1\mac ... header, and all characters available in the Mac encoding) poses problems in Word97 viewer and StarOffice5.2 . Ragtime5 reads it correctly, except for the 2subscript character. StarOffice 5.1a reads it correctly.

20000918: Mystery solved. The aforementioned currency symbol was substituted in Mac OS 8.5 and higher by the Euro symbol. Recent PC software "knowing" this would thus convert this Macintosh ASCII coding by a Unicode Euro code ($20AC). The effect of this is that those who used the currency symbol on the Mac (and continue to use it with/in their pre-OS8.5 fonts) have a problem converting their documents to PC. Apart from that, I finally decided to change the position for the 2subscript character, and to remove the currency symbol from the compatibility character set. The 2subscript has taken the place of the mu (µ) character. The currency symbol has disappeared from the PC fonts, and has been replaced by a Euro symbol in the Mac fonts. The PC fonts have the Euro symbol at the location 128, with the (additional) Unicode $20AC. The PC fonts may receive an additional copy of the Euro at the "currency position", and the Mac fonts may get the Unicode code assigned to it when this proves necessary.

20010128: There are some mysteries associated with installation under Windows. The Palatino Nisaba font was initially called Palatino Latin. I am pretty sure that this caused problems with several versions of Word, esp. under Windows 95 – "Latin" may have been interpreted as a codepage selection, i.e. it may be a "reserved" word. Palatino Nisaba has also caused some similar problems under Windows 98, where it seems that just copying the fontfiles into the fonts folder did not properly install the fonts; use of the "Install new font" menu item (in the File menu) was necessary to get correct behaviour with notably the Word97 programme (wordprocessor and viewer). A telltale symptom of improper installation is when either of the plain-style fonts appears as bold.

20010624 important update!! The above-mentioned problem is maybe partly explained by the following. The Palatino Cuneitype (Regular) (PCR) font contains a few glyphs that cause problems with certain postscript printers when printing from certain applications (StarOffice) under Windows 9x. I will spare you the details: printing a document once renders the PCR font useless until it is reinstalled or Windows has been restarted. If you are using a PostScript printer or Acrobat Distiller, verify the postscript driver's version. This can be done by opening the printers' properties windows, and clicking on the [about] button on the Paper settings tag. The problem arises with adobeps 4.4.1 (and maybe with earlier versions), but not with the Microsoft pscript.drv driver (older than adobeps 4.41). If you are using a driver that does not identify itself as pscript.drv, go into the Fonts tag. There, you will find a button [Send fonts as...]. Click this button, and select to send TrueType fonts as Type 42, and click OK. I will try to find out exactly what goes wrong, and why it does not when sending fonts as Type 42; in the mean time, this is the setting to use.

The design rules for the transcription characters are taken from the design of D. Charpin's CunéiType1 font. Thus, the relative size and relative placement of the superscript and subscript characters are largely identical to those in the aforementioned font. Notably, the alphanumerical subscript characters and the superscript ? and ! are scaled at 70.7% of their regular forms. The other (textual, superscript) symbols are scaled at 90%. Superscript characters are placed at 5/9 times the x-height. The subscript are placed at 150/217 times the descender under the font's baseline (this is the baseline of the subscript 1). For Palatino Cuneitype (regular/plain), these "superscript baseline" and "subscript baseline" are at 536 and -390, respectively. The corresponding, scaled "x-heights" are at 1404 (i.e. a height of 868, ) and 292 (i.e. a height of 682), respectively. The x-height of the regular symbols is (at) 965. NB: in CunéiType1 italic, the subscript x is mistakenly italicised also. This has not been retained in Palatino Cuneitype. Pal. Nisiba and Pal. Cuneitype are visually identical so that they can be mixed freely, without the font changes being visible. There are some differences with the original Palatino fonts (from MacOS 7); notably, screen legibility has suffered a little. This is mostly an issue on (older) Macs. Under Windows (98 and higher?), text is still legible at 9pt, accents become easily distinguishable at somewhat bigger sizes (this also depends on screen resolution!). Output on a printer with a good, modern-day resolution (> 300DPI) is impeccable.

Organisation and Installation:

The distribution is organised in the following fashion. There are two directories (folders), containing the two subfamilies, Palatino Nisaba and Palatino Cuneitype. In each of these folders, there are separate compressed archives for PC (PC.zip) and Mac (Mac.sit) fonts.

La distribution est organisée ainsi. Les 2 familles se trouvent dans des répertoires distinctes, portant leur nom. Dans chaque répertoire, un fichier archive (zip pour pc, stuffit pour macintosh) contient les polices de la famille en question.

PC. Each archive contains the 4 files that "are" the 4 faces of a subfamily (plain, italic, bold and bold-italic). The 4 fonts are to be copied to the c:\windows\fonts directory (or wherever you keep your installed fonts; you can get there via the Control Panel). It may be necessary to take the long route... If in Word, Nisaba and Cuneitype have different appearances (esp. weight) on the screen (and in print...), do the following. Remove the offending fontfiles from the fonts folder. Then, under the File menu of that folder, choose "Install new font". This pops up a window that allows you to choose a directory with fonts; go to the one where the originals are stored. If you don't see anything in the list window, this is normal; just cancel, and recall the dialog with "Install new font". You will probably now see a list of fonts (if not, find the AUTODESTRUCT\B.GATES option in the registry and activate that one. No, the B is not for Baldur :^). I mean, this is a joke… ). Select the fonts you want to install, and hit the OK button. For me, Palatino Nisaba was the offending font, and this procedure helped...

Installation sur pc. Dézippez le fichier pc.zip . Ça donne 4 fichier .ttf: les polices. Maintenant, ouvrez le répertoire c:\windows\polices (ou windows\fonts), ou bien ouvrez le gestionnaire de polices dans le panneau de configuration. Dans le menu Fichier de la fenêtre qui s'ouvre, il y a une option "Installer la police...". Sélectionner cette option donne un dialogue qui permet d'aller chercher les polices: cherchez et sélectionnez le répertoire où se trouvent les polices à installer. Une liste avec les noms des polices trouvées est donnée. Cliquez sur le bouton "Toutes", puis sur OK ou Installer. C'est tout – mais il faudra le faire pour les 2 familles.

Note importante pour ceux qui utilisent une imprimante PostScript, ou Acrobat Distiller. Il faudra vérifier la version du pilote postscript dans les Propriétés des imprimantes. Ouvrez la fenêtre avec les imprimantes, et pour chaque imprimante en question, demandez les propriétés. Dans l'option Papier, sélectionnez le bouton [A propos] (ou [About]): ça donnera la version du pilote. Si c'est un pilote qui s'appelle adobeps.drv avec version 4.2 ou plus récent (4.4.1 est le dernier du moment), allez dans l'option "Polices" (ou Fonts), et cliquez sur le bouton [Envoyer en tant que] (ou [Send font as]). Dans la nouvelle fenêtre, choisissez l'option "Type 42" dans la liste avec les possibilités pour les polices TrueType. Cf. la rubrique "Trouble Shooting" pour plus de détails.

MAC. The Mac archives all contain a single file with the extension .mcb. These are so-called macbinary files. They can be converted with e.g. Stuffit, something that will likely be done automatically when you unstuff the archives. This will give a suitcase containing the full font subfamily (plain, bold, italic, bold italic). These suitcases of course have to go into the Fonts folder within the current System folder. These are "old" Stuffit archives, that can without problems be opened with version 4 or even 3 of Stuffit. Fonts and distribution were tested on a system 7.1 LC-III and system 7.5.3 running on the BasiliskII Mac emulator for Windows and Linux.

Installation sur macintosh. Desemballez les fichiers mac.sit. Si Stuffit (Expander) est bien configuré, ça donnera toute de suite une valise de polices portant le nom de la famille en question. Sinon (ça donnera un fichier palcun.mcb ou palnis.mcb), il vous faudra le "passer par" stuffit encore une fois: ce sont des fichiers dite MacBinary. L'installation est ensuite facile: il suffit de glisser les valises sur le dossier système: le système s'occupera du reste.

Conversion.

This effort would not be complete without a means of converting documents to make use of the new fonts. This is all the more necessary for those documents written in CunéiType1. I wrote a little programme that does just this, pazuzu. It will convert a document saved as RTF on the Macintosh, into an RTF (Rich Text Format) file where each CunéiType1 character is substituted by a Palatino Nisiba/Cuneitype equivalent, or a character from the Symbol font where necessary. It will do each conversion on a character basis, but will try to limit the number of fontchanges. I can currently provide binaries that run under Windows (actually, in a DOS window) and Linux-x86. Of course, similar conversions can also done in e.g. Word (a recent version), but only on the Mac! Once the document has been opened on a PC, irreversible conversions will have been done that render the document (almost) useless. The same programme can also be used to translate between Macintosh RTF and PC RTF – it turns out that there are cases where the regular wordprocessor programmes refuse to do this (notably when Macintosh Word 5 is involved). For an overview of the options, and/or an explanation, launch the programme as pazuzu –help or pazuzu –explain.

Selon les explications que j'ai données ci-dessus, il n'est donc pas possible de ouvrir/convertir automatiquement un document écrit avec CunéiType1 sur PC. Pour cette raison, j'ai écrit un petit programme, pazuzu, qui se charge de la conversion de documents sauvegardés en tant que RTF (Rich Text Format). Ce logiciel traduit chaque caractère en CunéiType1 dans un caractère en Palatino Nisaba ou Cuneitype, ou encore dans la police Symbol si cela est nécessaire. Il est également de faire les conversions Mac/PC ou PC/Mac – avec ou sans la traduction "CunéiTypique" (ce qui s'avère parfois nécessaire, surtout quand Macintosh Word 5x est "impliqué"). Ce logiciel existe pour l'instant uniquement en version "ligne de command" pour DOS/Windows et pour Unix/Linux. Pour une explication des options, veuillez taper pazuzu –help ou pazuzu –explain.

Disclaimer.

Some legal notes. I did not find any license information within the original fonts, but there was probably a sort of general license that came with the system. I did not change a single character. All "cuneitype" specific characters are composed of characters that were already contained in the font (the only exception are the demicrack symbols, that are based on a combination of the square brackets with the bottom of the dagger symbol). For this reason, I decided not to change the name, but rather add a denominating word to distinguish the various subfamilies (Nisaba, Cuneitype). The other operations I made on the font were 1) recalculating the global metrics (typographical dimensions); 2) adding kerning information; 3) hinting on some cuneitype glyphs; 4) adding some leading space to the colon (:) and semicolon (;) characters (so that it is not necessary to put a space before them); 5) I shifted the acute (accent grave) on a number of characters to obtain a nicer aspect and match with the Adobe version installed in (most?) PostScript printers; 6) I renamed the middot character to periodcentered (this is the encoding used on PC to "show" spaces when the "show invisible characters" option is activated); 7) the Euro glyphs come from the Book Antiqua family (this is a Palatino "clone" that is bundled with Windows 95 and higher). I hope that in this way I did not infringe too much on the copyrights detained by Apple and AG-Linotype (who own Palatino). I do not claim any copyright, nor intellectual rights on these fonts. I do claim a copyright on the encoding. I composed this distribution for personal use and the advancement of a particular field of the historical/linguistic sciences, without aim for or need of any financial benefit.

Please keep this distribution in the form that you got it, including this note, and pass it on as such if you decide to do so.

File list:

Documentation

Note.*: this file

cuntable.*: Rich Text Format version of a table listing all available characters,

and comparing them to the original Palatino. You can use this file as

an immediate verification of the installation.

encodings: text files with the various encodings ("character tables") used.

Palatino Cuneitype:

PC:

PC.zip archive containing the PC TrueType fonts.

Mac:

Mac.sit archive containing the Macintosh suitcase as MacBinary

file (PalCun.mcb).

Palatino Nisaba:

PC:

PC TrueType fonts, exchangeable characters only. NB: several characters

normally available are thus missing, notably the fractions and the superscript

1, 2 and 3. This is because these characters would be lost when the document is

converted to Macintosh format.

Mac:

Macintosh suitcase as MacBinary file (PalNis.mcb); exchangeable characters

only.

Glossary:

ASCII: standard computer alphabet containing the first 127 characters of every font. Contains various so-called control characters in the first 31 entries; among others the Tab, new line and carriage return "characters". The "text (printable) characters" proper start at position 32 with the Space character; the remainder are the standard alphanumerical characters used in e.g. American English (this coding is also known as US-ASCII). Various extensions up to 255 exist, that contain the accented characters and a collection of special symbols. The problem of transporting text documents between e.g. Mac and PC finds its origin in this multitude of standards. So-called ASCII text files, usually referred to as text files, are nothing more or less than a sequence of these ASCII values. Each letter, digit, punctuation, space, etc. occupies 1 byte in the file, containing the corresponding ASCII value. The characters shown when viewing such a file depend on the ASCII "dialect" in use on the computer. Hence, displaying a PC ASCII file on a Macintosh without any conversions will correctly show the "standard", non-accented characters, but not so the accented characters, and other special symbols. See also Unicode.

Font: a set of characters (glyphs, letter forms) that share a similar design. Contained in a single file.

Font family/Typeface: a set of fonts containing the plain font, and one or more of its italic, bold and bold-italic versions. Under Windows, each member is contained in a single font file (this can also be the case in the Mac OS 8.5 and higher). Under the Mac OS, all members are typically contained in a single suitcase file, that can also contain bitmap versions, and even other fonts or font families (e.g. all fonts used for a given task, like writing assyriology papers). When not all members of a family are present, the OS (operating system) will construct the missing ones from the regular (non-bold, non-italic) font, by slanting and/or smearing. Obviously, this does not always give the same result (compare the 'a' to the 'a' in this font: the 2^nd is not a slanted version of the 1^st!).

Global metrics: each font contains some information about the overall dimensions, shared by all characters. Like the height of an 'x', the width of an 'M', etc. This information is used by the operating system to determine how much to scale a font (all characters) at a certain requested size (say 12points).

Kerning: The distance between characters is normally determined from the width of the first character, and the "leading whitespace" contained in the second. But not all characters are equally wide, and some may have parts that could well overlap a following character. Like a capital 'T', or an italic 'é'. A font can store a list of such two-character combinations, and how much to decrease or increase the "standard" distance to obtain a more satisfying result.

TrueType: a font technology originally developed by Apple. In scalable (vector) fonts, characters (glyphs) are defined as scalable (vector) outlines. To be used, whether on screen or to print, these outlines must be filled in with pixels (screen "points") or ink/toner dots. This is done with a rasteriser, a specialised programme that determines how to fill in the outline with dots to obtain the best representation of the original character. Needless to say that this is easier at larger point sizes and higher resolutions: the smaller the dots with respect to the character's size, the more precise its outline can be restituted. At sizes below 12 points, it becomes very difficult to make a faithful representation on a computer screen (which has a much lower resolution than printers do). TrueType fonts contain hints that help maintain a readable result down to approximately 8 points, but small details (like accents) can loose definition or disappear altogether. This applies only to the screen display (as can be seen by selecting a larger-than-100% zoom).

Type1: a scalable font technology from Adobe. These are so-called PostScript fonts (PostScript is a multi-purpose-with-accent-on-graphics language that is used in many high-end (laser)printers). These fonts also need a rasteriser. This rasteriser is built into postscript printers, but for use on screen or with non-postscript printers, Adobe TypeManager (ATM) is necessary. There is no quality difference between the 2 technologies, at least not as long as a good rasteriser and a good printer is used. ATM is good, and the MacOS and Windows contain good TrueType rasterisers. I opted for TrueType since this is probably the format that the most people can work with, without the need for 3^rd-party software (ATM currently exists in a free "lite" version, but this has not always been the case).

Unicode: an attempt to finish with multiple standards. Unicode assigns 2 bytes to each character. That extends the possible range from 256 to 16384 characters that can be represented with a unique number (and thus in a text using a single font). In this standard, all characters have a canonical name (like Eacute for a capital E with an acute accent, or delta for a small greek delta) and a 2byte number. In addition, there are scripts, that group a selection of maximally 256 of such characters; there exist such scripts for almost all currently existing languages. Ligatures can also be defined as a combination of 2 or more simple characters. Thus, the gisjh would not need to exist as a separate glyph (character) in a Unicode version of Palatino Cuneitype. For a complete overview, see www.unicode.org. Unicode is definetely the future, although there is little chance that the Unicode organisation will officially recognise the transcription characterset used for transcribing cuneiform texts. This means that the "community" will have to decide internally how to code the different glyphs in the so-called Unicode user-ranges (basically parts of the 0...16383 range that will not be used by Unicode, but remain free for individual use). Unicode may recognise a script for cuneiform characters: a project to this end is currently being undertaken by Dean Snyder and co-workers (in 4 years, we'll know the result). The same people may afterwards propose a transcription standard, if (or whether-or-not...) nobody else does it before (Bendt Alster is one person currently considering this, and there is something going on in Frankfurt too). The drawback of Unicode is that you need an operating system and applications that understand Unicode. Most Unicode fonts existing nowadays also contain "old-fashioned" code tables, making it possible to use them on older systems, without off course being able to profit of their additional features. Windows 98 and newer supports Unicode, as do the recent versions of the Mac OS. Older versions of that OS that are to my knowledge still widely in use do not, and even in the Windows universe, not all programmes understand Unicode. This is one of the main reasons why I did not immediately include Unicode support. This may well change in a far or near future, when there is more clarity on a standard.