Frequently Asked Questions About Fonts ---------------------------------------------------------------------- Compiled by Norman Walsh Copyright (C) 1992, 93 by Norman Walsh . Subject: Table of Contents 1. General Information 1.1. Font Houses 1.2. What's the difference between all these font formats? 1.3. What about "Multiple Master" fonts? 1.4. Is there a methodology to describe and classify typefaces? 1.5. What is the "f" shaped "s" called? 1.6. What about "Colonial" Typefaces? 1.7. Where can I get ... fonts. 1.8. Where can I get fonts for non-Roman alphabets? 1.10. How can I convert my ... font to ... format? Subject: 1. General Information Many FAQs, including this one, are available by anonymous ftp from rtfm.mit.edu in the directory pub/usenet/news.answers. Each posted section of the FAQ is archived under the name that appears in the "Archive-name" header at the top of the article. This FAQ is a work in progress. If you have any suggestions, I would be delighted to hear them. This FAQ is maintained in TeXinfo format. A Perl script constructs the postable FAQ from the TeXinfo sources. TeX DVI, PostScript, and Info versions of this FAQ are available from ftp.shsu.edu in /tex-archive/help/comp-fonts-FAQ. A "Gopher" server is also maintained at shsu.edu which can provide interactive access to the FAQ. Finally, an online, hypertext version of the FAQ is maintained (experimentally) on jasper.ora.com where an HTTP server runs. For example, point XMosaic (or a similar WWW browser) to http://jasper.ora.com/. The posted version of the FAQ is organized in a quasi-digest format so that it is easy to find the questions you are interested in. All questions that appear in the table of contents can be found by searching for the word "Subject:" followed by the question number. The "TeXinfo" distribution from the Free Software Foundation contains a program called "Info" that can be used to read the Info version of the FAQ in a hypertext manner. The "TeXinfo" distribution can be obtained from prep.ai.mit.edu in the /pub/gnu directory. At the time of this writing, texinfo-2.16.tar.gz is the most recent version. Info files can also be read in hypertext form by GNU Emacs. Future versions of the FAQ will make more use of the hypertext capabilities provided by the Info format. At present, the FAQ is organized as a simple tree. A plain ASCII, postable version of the FAQ will always be maintained. All trademarks used in this document are the trademarks of their respective owners. Standard disclaimers apply. Subject: 1.1. Font Houses This section will be expanded on in the future. It contains notes about various commercial font houses. Compugraphic ============ See "Miles, Agfa Division" Miles, Agfa Division ==================== Compugraphic which was for a while the Compugraphic division of Agfa, is now calling itself "Miles, Agfa Division" (yes, the Miles drug company), since CG's off-shore parent Agfa has been absorbed by Miles. So typographically speaking, Compugraphic, CG, Agfa, A-G ag, and Miles all refer to the same company and font library. Their proprietary fonts are still CG Xyz, but the name is Miles Agfa. Subject: 1.2. What's the difference between all these font formats? This question is not trivial to answer. It's analogous to asking what the difference is between various graphics image file formats. The short, somewhat pragmatic answer, is simply that they are different ways of representing the same "information" and some of them will work with your software/printer and others won't. At one level, there are two major sorts of fonts: bitmapped and outline (scalable). Bitmapped fonts are falling out of fashion as various outline technologies grow in popularity and support. Bitmapped fonts represent each character as a rectangular grid of pixels. The bitmap for each character indicates precisely what pixels should be on and off. Printing a bitmapped character is simply a matter of blasting the right bits out to the printer. There are a number of disadvantages to this approach. The bitmap represents a particular instance of the character at a particular size and resolution. It is very difficult to change the size, shape, or resolution of a bitmapped character without significant loss of quality in the image. On the other hand, it's easy to do things like shading and filling with bitmapped characters. Outline fonts represent each character mathematically as a series of lines, curves, and 'hints'. When a character from an outline font is to be printed, it must be 'rasterized' into a bitmap "on the fly". PostScript printers, for example, do this in the print engine. If the "engine" in the output device cannot do the rasterizing, some front end has to do it first. Many of the disadvantages that are inherent in the bitmapped format are not present in outline fonts at all. Because an outline font is represented mathematically, it can be drawn at any reasonable size. At small sizes, the font renderer is guided by the 'hints' in the font; at very small sizes, particularly on low-resolution output devices such as screens, automatically scaled fonts become unreadable, and hand-tuned bitmaps are a better choice (if they are available). Additionally, because it is rasterized "on demand," the font can be adjusted for different resolutions and 'aspect ratios'. Werenfried Spit adds the following remark: Well designed fonts are not scalable. I.e. a well designed 5pt font is not simply its 10pt counterpart 50% scaled down. (One can verify this by blowing up some small print in a copier and compare it with large print; or see the example for computer modern in D.E. Knuth's TeXbook.) Although this fact has no direct implications for any of the two methods of font representation it has an indirect one: users and word processor designers tend to blow up their 10pt fonts to 20pt or scale them down to 5pt given this possibility. Subtle details, but well... LaserJet .SFP and .SFL files, TeX PK, PXL, and GF files, Macintosh Screen Fonts, and GEM .GFX files are all examples of bitmapped font formats. PostScript Type 1, Type 3, and Type 5 fonts, Nimbus Q fonts, TrueType fonts, Sun F3, MetaFont .mf files, and LaserJet .SFS files are all examples of outline font formats. Neither of these lists is even close to being exhaustive. To complicate the issue further, identical formats on different platforms are not necessarily the same. For example Type 1 fonts on the Macintosh are not directly usable under MS-DOS or Unix, and vice-versa. It has been pointed out that the following description shows signs of its age (for example, the eexec encryption has been thoroughly hacked). I don't dispute the observation and I encourage anyone with the knowledge and time to submit a more up to date description. It has further been suggested that this commentary is biased toward Kingsley/ATF. The omission of details about Bitstream (and possibly Bauer) may be considered serious since their software lies inside many 3rd-party PostScript interpreters. The moderators of this FAQ would gladly accept other descriptions/ explanations/viewpoints on the issues discussed in this (and every other) section. [Ed Note: Liam R. E. Quin supplied many changes to the following section in an attempt to bring it up to date. Hopefully it is a better reflection of the state of the world today (12/07/92) than it was in earlier FAQs] Henry Schneiker wrote the following description of the differences between several scalable font technologies: ((( semi-quote ))) There has been a lot of confusion about font technologies in recent times, especially when it comes to Type 1 versus Type 3 fonts, "hints," PostScript compatibility, encryption, character regularizing, kerning, and the like. * Encryption (eexec) All fonts produced with Adobe's font technology are protected through data encryption. The decryption is provided by the `eexec' (encrypted execute) PostScript operator and, until recently, was only present in Adobe's licensed PostScript. Adobe has published the details of the Type 1 font format in the `Black Book', Adobe Type 1 Font Format (version 1.1), Adobe Systems Inc., 1990. The encryption was mainly used because of font copyright problems; unencrypted fonts can also be used, but these tend to use an efficient binary encoding, also in documented the Type 1 book, and so are still not readable PostScript. * Type 1, Type 3, and Type 5 font formats There are generally three font formats used in Adobe PostScript printers: Type 1, Type 3, and Type 5. Type 1 fonts are Adobe's downloadable format. Type 3 fonts are third-party downloadable format. Type 5 fonts are the ROM-based fonts that are part of your printer. There is no functional difference between a Type 1, Type 3, or Type 5 font. A Type 3 font can do anything a Type 1 or Type 5 font can do. The only real difference between them is where the `BuildChar' routine comes from. For Type 1 and Type 5 fonts it's built into the printer. For Type 3 fonts it's built into the font. In other words, anything a Type 1 font can do a Type 3 font can also do. [Ed note: the reverse is not true. Type3 fonts can do things that Type1 fonts cannot. But they aren't hinted...] When PostScript is asked to generate a character, PostScript looks in the font's dictionary for FontType. If FontType is 1 or 5 PostScript executes an internal routine that knows how to interpret the font data stored in CharStrings. If FontType is 3 PostScript executes the routine BuildChar from the font's dictionary to interpret the font data (often stored in CharStrings). However, each BuildChar routine is written to read data formatted in a method convenient to the vendor. Adobe, Altsys, Bitstream, and Kingsley/ATF all format their font data differently and, hence, have different BuildChar routines. [Ed note: relative hard disk efficiency of Kingsley vs. Adobe fonts deleted on 12/07/92] Type 5 fonts are special in that they often include hand-tuned bitmaps for the commonly used sizes, such as 10- and 12-point. Other sizes are generated from the outlines in normal fashion. Don't confuse Type 1, Type 3, and Type 5 fonts with Bitstream's Type A, Type B, Type C, and Type F. They are not the same and serve only to confuse the issue. * Resolution `hints' When a character is described in outline format the outline has unlimited resolution. If you make it ten times as big, it is just as accurate as if it were ten times as small. However, to be of use, we must transfer the character outline to a sheet of paper through a device called a raster image processor (RIP). The RIP builds the image of the character out of lots of little squares called picture elements (pixels). The problem is, a pixel has physical size and can be printed only as either black or white. Look at a sheet of graph paper. Rows and columns of little squares (think: pixels). Draw a large `O' in the middle of the graph paper. Darken in all the squares touched by the O. Do the darkened squares form a letter that looks like the O you drew? This is the problem with low resolution (300 dpi). Which pixels do you turn on and which do you leave off to most accurately reproduce the character? All methods of hinting strive to fit (map) the outline of a character onto the pixel grid and produce the most pleasing/recognizable character no matter how coarse the grid is. [Ed note: deleted some paragraphs that are no longer true. Times change...] * Optical Scaling Optical Scaling modifies the relative shape of a character to compensate for the visual effects of changing a character's size. As a character gets smaller, the relative thickness of strokes, the size of serifs, the width of the character, the inter-character spacing, and inter-line spacing should increase. Conversely, as a character gets larger, the relative thickness, widths, and spacing should decrease. Contrast this with linear scaling, in which all parts of a character get larger or smaller at the same rate, making large characters look wide and heavy (strokes are too thick, serifs are too big) while small characters look thin and weak. * Kerning As applied to PostScript fonts, kerning refers to kern pairs. A kern pair specifies two characters (e.g., A and V) and the distance to move the second character relative to the first. The typical use of a kern pair is to remove excessive space between a pair of characters. However, it may also be used to add space. * PostScript clones There are currently several printer manufacturers on the market with PostScript clones. To be viable, a PostScript clone must comply with the `red book' (PS Language Reference Manual). In order to avoid paying royalties to Adobe, and because Adobe's Type 1 font format was originally proprietary, many PostScript interpreters use some other font format. Sun uses F3, and some other vendors use Bitstream's Speedo format, for example. The only real problem this causes is that the widths of characters (the `font metrics') may vary from Adobe's, so that programs that assume the Adobe character widths will produce poor quality output. Bitstream fonts used to be particularly bad in the early days, but they and most or all of the other vendors have solved those problems. * Apple TrueType [Ed note: formerly "Royal (`sfnt')"] format and System 7 Apple's new System 7.0 supports a new format of outline font that will allow high-quality characters of any size to be displayed on the screen. TrueType stores font outlines as B-spline curves along with programmed resolution hints. B-spline curves are faster to compute and easier to manipulate than the Bezier curves used in PostScript. Adobe is not going to support Apple's new format by converting the Adobe/Linotype library to B-spline format. There are two reasons for this: First, there is no support for font encryption (yes, the hooks are there, but nothing is implemented). Second, Adobe does not want to dilute PostScript and its font library. However, the Macintosh is too big a market to simply turn away from. Therefore, Adobe will provide its Font Manager to display its own fonts on the Mac screen. Apple ships Adobe's ATM for this purpose. ((( unquote ))) Subject: 1.3. What about "Multiple Master" fonts? Multiple Master Fonts are an extension to the Adobe font format. providing the ability to interpolate smoothly between several "design axes" from a single font. Design axes can include weight, size, and even some whacko notions like serif to sans serif. Adobes' first Multiple Master Font was Myriad - a two-axis font with WEIGHT (light to black) on one axis, and WIDTH (condensed to expanded) along the other axis. In the case of Myriad, there are four "polar" designs at the "corners" of the design space. The four designs are light condensed, black condensed, light expanded, and black expanded. Given polar designs, you can set up a "weight vector" which interpolates to any point within the design space to produce a unique font for a specific purpose. So you can get a "more or less condensed, somewhat black face". Multiple Master Fonts can be used on any PostScript printer. Multiple Master Fonts need a new PostScript operator known as makeblendedfont. The current crop of Multiple Master Fonts supply an emulation of this operator so the printer doesn't need this operator. A short tutorial on Multiple Master Fonts and makeblendedfont appears in PostScript by Example, by Henry McGilton and Mary Campione, published by Addison-Wesley. Danny Thomas contributes that there are a few PostScript interpreter (version)s which have bugs that appear with the emulation of the makeblendedfont operator used to support Multiple Master fonts. There weren't many exhibiting this problem, though it may have happened even with one Adobe interpreter. Subject: 1.4. Is there a methodology to describe and classify typefaces? There is a standard, Panose, but it is mostly ignored by typographers (not because it's bad, just because they don't need it). The Panose system is documented, among other places, in the Microsoft Windows 3.1 Programmer's Reference from Microsoft Press. The ISO also has a scheme, but it is not Panose. At least one book by a respected authority, Alexander Lawson, Printing Types: An Introduction, describes another, less rigorous system [ed: of his own], which is exposited in "An Introduction" and used without exposition in his later "Anatomy of a Typeface". There is another book, Rookledges International Typefinder, which has a very complete system that uses tell-tales of individual glyphs as well as overall style to index most known faces right in the book. J. Ben Leiberman has another book on type face description. Terry O'Donnell adds the following comments: The current ISO system was initiated (I believe) by Archie Provan of RIT--a successor to Mr. Lawson. Whereas in typographic practice or teaching--only a high level classification is necessary - times have changed and the current ISO system aims to accomplish something beyond the high level. A major goal is to aid software to help users make selections. For example, a naive user might ask for all fonts on a font server which have a Roman old style appearance. Another goal would be to help users with multi-lingual text: a user creating a document in English using e.g. Baskerville wants to know what Arabic or Japanese language font on his system/file server would harmonize well with the Baskerville. It is not all in place yet--but the more detailed ISO classes--and the current addition of non-latin typefaces--are an attempt to address this issue. A second goal is to help with the font substitution problem. Neither ISO or Panose address the metrics issues in font substitution--but both might aid software in picking the nearest style of available available fonts. Subject: 1.5. What is the "f" shaped "s" called? Both the "f" with half a crosbar (roman) and the integral sign (italic) are called long-S. Subject: 1.6. What about "Colonial" Typefaces? Why does colonial printing have that "Colonial" feel? ===================================================== Colonial type was either very roughly treated by moist salt air on the crossing and in colonial port cities, or was copied locally by tacky techniques (such as driving used foundry type into soft lead to make very soft deformable matrices), and the paper was very rough, which abrades both the serifs and the hairlines. So except for the best work done with new, european types, the serifs were much smaller, even broken off, than the original founder/punchcutter intended. Thins could be abraded by rough paper to nothingness, esp after humid salt air had leached the hardener out of the alloy. What fonts are good for mock-colonial uses? =========================================== For example, what fonts have the following features: old-style figures (non-lining numbers), the long s character, slightly irregular shapes (a la type produced by colonial printers), and a decent complement of ligatures. And what about free or cheap faces like this? I don't know if any exist with all of 1-5. As I believe you get what you pay for, especially in fonts, I haven't looked at free and cheap-copy fonts. Microsoft's expansion set for their Win3.1 optional fonts has Garamond Expert and I think I remember it having the long ess too. I forget about OSFigs; it should tho'. Monotype's metal faces "16th Century Roman" and "Poliphilus" may be available in digital; if so, they imitate early presswork with early and are very close to what one wants. "A commercial supplier [not yet sampled] is Image Club Graphics in Calgary (1-800-661-9410). It is called Caslon Antique. It is supplied as both roman and italic, together, for $25. They advertise in MacWorld/MacUser/MacBlah. I am unable to tell from abcDEF123 if the numerals are old-style, but I think not. Ligatures? long-S? Not yet known. Guillemots, though, are there. ... Letraset, circa 1977, showing a Caslon Antique with modern numerals, no ligatures, and only UKPounds and German ss extensions." [Ike Stoddard] NB: Caslon Antique is not a Caslon per se: "The last Caslon to mention is that ubiquitous but unrelated Caslon Antique, which possesses no similarity whatsoever to the original. This old reprobate was introduced by Barnhart Brothers of Chicago under the name Fifteenth Century. Its negative reception lasted until about 1918, when, with a simple name change to Caslon Antique, it became the most commonly selected type for reproductions of colonial American printing. It is now seen in everything from liquor advertisments to furniture commercials" [Lawson, 1990,Anatomy] Miles Agfa (Compugraphic) has always had a Caslon Antique; I don't know if it is available for TrueType or Type 1, but Agfa has been doing TrueType bundles at reasonable prices. [wdr] What fonts could a colonial printer have had? ============================================= According to D.B.Updike in the classic reference "Printing Types: Their History, Forms types of the Caslon Old Style fonts and cheap copies of same in the 18th C. Before that, it would have been the older Dutch faces, almost always lagging English tastes. If you can find the Oxford Fell types, they are classic Dutch-as-used-by-englishmen. Anything with a Dutch moniker and the Oldstyle adjective is probably ok; Van Dijck if you find it, say (died 1673). Ben Franklin recommended Caslon faces. But these were not available in England before 1720, first full broadside in 1734. Lawson declares that the first printing of the Declaration of Independance was in Caslon. Wilson's Scotch Modern was the "modern" font that surfaced in quantity in america. If the Scotch Roman your vendor has is sort-of like-Bodoni but nicer than his Bodoni, that's it. It wasn't available until late 1700s, though. Subject: 1.7. Where can I get ... fonts. Before I go any farther, let me extol the virtues of the Archie servers. If you need to find something on the net, and you have any idea what it might be called, Archie is the place to go. In North America, telnet to "archie.rutgers.edu" and login as "archie". There are many other servers around the world, any Archie server can give you a list of other servers. There are better documents than this to describe Archie and you should be able to find them from the above starting point. If you have trouble, feel free to ask norm (via Email please, no need to clutter comp.fonts with a query about Archie ;-). In addition to the telnet option, several archie clients exist including a very nice X11 implementation (Xarchie). * Adobe Type 1 Fonts in MS-DOS/Unix Format: ftp.cica.indiana.edu:/pub/pc/win3/fonts ftp.cica.indiana.edu:/pub/pc/win3/fonts/atm archive.umich.edu:/msdos/mswindows/fonts * Adobe Type 1 Fonts in Mac Format: mac.archive.umich.edu:/mac/system.extensions/font/type1 sumex-aim.stanford.edu:/info-mac/font * Adobe Type 3 Fonts in Mac Format: mac.archive.umich.edu:/mac/system.extensions/font/type3 * TrueType fonts in MS-DOS Format: ftp.cica.indiana.edu:/pub/pc/win3/fonts/truetype * TrueType fonts in Mac Format: mac.archive.umich.edu:/mac/system.extensions/font/truetype * TeX PK/PXL/GF fonts: The TeX community has its own support groups that can provide better answers to this question. The canonical list of MetaFont fonts is posted occasionally to comp.text.tex. The comp.text.tex newsgroup (or the Info-TeX mailing list, if you do not have access to news) are good places to start. Email norm if you need more specific information. * LaserJet bitmap fonts: wuarchive.wustl.edu:/mirrors/msdos/laser Also on other simtel20 mirrors... If you know of other archive sites (the above list is no where near complete) or other formats that are available on the net, please let us know. The sites above represent places where shareware and public domain fonts are available. Many, many typefaces are not available in shareware form. And many shareware faces are less than adequate for a variety of reasons, particularly at small sizes. It seems to be the consensus of the comp.fonts community that "you get what you pay for." If you need a professional quality font, you should probably buy it from a professional. The list of font vendors in Appendix A (annotated with information about non-Roman alphabets) was contributed by Masumi Abe. Masumi was Adobe's Manager of Typographic Marketing for Asia. He has since left Adobe. Many font CDs are now available which offer many fonts for a low cost/font. Subject: 1.8. Where can I get fonts for non-Roman alphabets? As mentioned above, the list of font vendors is annotated with information about non-Roman alphabets. Commercially, Masumi suggests that Linguists' Software is the current [ed: as of 7/92] leading supplier of non-Roman fonts. Subject: 1.10. How can I convert my ... font to ... format? Conversion from one bitmapped format to another is not generally too difficult. Conversion from one scalable format to another is very difficult. Several commercial software packages claim to perform these tasks, but none has been favorably reviewed by the comp.fonts community. ATech's AllType program, in particular, has had poor reviews [ed: as of 7/92]. In an effort to settle a long-running and oft-asked question, I'll be blunt: as of today [6/93], THERE ARE NO NON-COMMERCIAL PROGRAMS WHICH WILL CONVERT FROM ONE SCALABLE FORMAT TO ANOTHER. Not from TrueType to PostScript Type 1, Type 3, Type 5, or any other scalable PostScript format. Not from PostScript Type 1 to TrueType. Not to or from Intellifont. Not to or from Sun F3 format. There are some commercial programs: AllType, Metamorphosis, Font Monger, and even MoreFonts (to/from some proprietary format, I believe). And there are probably other commercial programs as well. However, as several people have noted, conversion from one scalable format to another is a bad idea. If the original font was well hinted, the converted font will not be. Of course, if the original was poorly hinted, maybe it won't matter much. For specific conversions, check the platform specific parts of the FAQ. Most of the conversions discussed require platform specific tools. Here is a summary of the conversions discussed (and the section in which they appear): Mac Type 1 PostScript To PC Type 1 PostScript (MS-DOS). To TrueType (commercial). PC Type 1 PostScript To Mac Type 1 PostScript (Mac, commercial). To TrueType (commercial). To TeX PK (MS-DOS). TrueType To Type 1 PostScript (Mac and MS-DOS, commercial). To HP LaserJet bitmaps (MS-DOS, hack!). TeX PK To HP LaserJet bitmap softfonts (MS-DOS). HP LaserJet bitmap softfonts To TeX PK (MS-DOS). In addition, Adobe ships a copy of Adobe Font Foundry with all of its fonts which can convert Type 1 fonts into HP LaserJet softfonts. Subject: 1.11. Are fonts copyrightable? This topic is hotly debated at regular intervals on comp.fonts. Terry Carroll. provides the following analysis of current [ed: as of 6/92] legislation and regulation regarding fonts and copyrights in the United States. Terry is "Editor in Chief" of Volume 10 of the Santa Clara Computer and High Technology Law Journal. Members of the comp.fonts community are encouraged to submit other materials that add clarity to the issue. It has been pointed out that this section deals primarily font copyright issues relevant to the United States and that this situation is not universal. For example, in many parts of Europe typeface designs are protectable. "First, the short answer in the USA: Typefaces are not copyrightable; bitmapped fonts are not copyrightable, but scalable fonts are copyrightable. Authorities for these conclusions follow. Before we get started, let's get some terminology down: A typeface is a set of letters, numbers, or other symbolic characters, whose forms are related by repeating design elements consistently applied in a notational system and are intended to be embodied in articles whose intrinsic utilitarian function is for use in composing text or other cognizable combinations of characters. A font is the computer file or program that is used to represent or create the typeface. Now, on to the legal authorities: Volume 37 of the Code of Federal Regulations specifies this about the copyrightability of typefaces: "The following are examples of works not subject to copyright and applications for registration of such works cannot be entertained: . . . typeface as typeface" 37 CFR 202.1(e). The regulation is in accordance with the House of Representatives report that accompanied the new copyright law, when it was passed in 1976: "The Committee has considered, but chosen to defer, the possibility of protecting the design of typefaces. A 'typeface' can be defined as a set of letters, numbers, or other symbolic characters, whose forms are related by repeating design elements consistently applied in a notational system and are intended to be embodied in articles whose intrinsic utilitarian function is for use in composing text or other cognizable combinations of characters. The Committee does not regard the design of typeface, as thus defined, to be a copyrightable 'pictorial, graphic, or sculptural work' within the meaning of this bill and the application of the dividing line in section 101." H. R. Rep. No. 94-1476, 94th Congress, 2d Session at 55 (1976), reprinted in 1978 U.S. Cong. and Admin. News 5659, 5668. It's also in accordance with the one court case I know of that has considered the matter: Eltra Corp. V. Ringer, 579 F.2d 294, 208 USPQ 1 (1978, C.A. 4, Va.). The U.S. Copyright Office holds that a bitmapped font is nothing more than a computerized representation of a typeface, and as such is not copyrightable: "The [September 29, 1988] Policy Decision [published at 53 FR 38110] based on the [October 10,] 1986 Notice of Inquiry [published at 51 FR 36410] reiterated a number of previous registration decisions made by the [Copyright] Office. First, under existing law, typeface as such is not registerable. The Policy Decision then went on to state the Office's position that 'data that merely represents an electronic depiction of a particular typeface or individual letterform' [that is, a bitmapped font] is also not registerable." 57 FR 6201. However, scalable fonts are, in the opinion of the Copyright Office, computer programs, and as such are copyrightable: "... the Copyright Office is persuaded that creating scalable typefonts using already-digitized typeface represents a significant change in the industry since our previous [September 29, 1988] Policy Decision. We are also persuaded that computer programs designed for generating typeface in conjunction with low resolution and other printing devices may involve original computer instructions entitled protection under the Copyright Act. For example, the creation of scalable font output programs to produce harmonious fonts consisting of hundreds of characters typically involves many decisions in drafting the instructions that drive the printer. The expression of these decisions is neither limited by the unprotectable shape of the letters nor functionally mandated. This expression, assuming it meets the usual standard of authorship, is thus registerable as a computer program." 57 FR 6202."