Inside Hiragino: Hiragino Shock and Apple Publishing Glyph Set

The transition to Mac OS X [macOS X] was a very interesting time for the Japanese publishing industry. In his first official appearance in Japan after returning to Apple, Steve Jobs turned the publishing industry on its head when he announced at MacWorld Tokyo 2000 that Apple would bundle professional Japanese fonts licensed from Dainippon Screen with extended character sets: “Capturing the beauty and richness of the Japanese language and kanji characters has always been beyond the capabilities of personal computers. Now, with premium quality fonts and the largest character sets ever, Mac OS X will make high-quality publishing a reality in Japan for all customers—professionals and first-time users alike.” At the time I wrote a report for MacInTouch:

Steve Jobs made a very big deal about kanji fonts and making MacOS X “the best kanji operating system” by licensing 6 ‘Hiragino’ font designs from DaiNippon Screen. It played well in yesterday’s keynote and the demonstrations today. And it is exciting that Apple is addressing some basic problems of Japanese publishing. Apple said these new ‘X’ fonts will use the new Adobe Japan 1-4 and JIS encoding standards unveiled at the last Tokyo Seybold Seminars 1999.

The official unveiling of bundled Hiragino, also used as the Mac OS X Japanese system font, (still in use today as the Japanese system font in macOS and iOS) was the February 2001 MacWorld Tokyo keynote just one month before OS X shipped. Steve:

As you may know, there have never been good Japanese fonts shipped in a personal computer operating system. If you want want to get over 8,400 characters and if you want to get very beautiful fonts you have to pay a fortune to go license these fonts on a per computer basis.

What we have done is we have licensed the most beautiful Japanese fonts around and we are bundling them in every copy of MacOS X. 17,500 characters…characters that just don’t exist in normal computer font sets. Characters that you cannot get on a personal computer without paying a lot of money. All bundled into MacOS X. I don’t think we’ve seen anything like this before.

MacWorld Tokyo 2001 keynote

Indeed the market had never seen anything like it before as the MacOS X Hiragino character set went above and beyond what Adobe offered in their AJ 1-4 character set. The industry quickly coined a new word: ‘Hiragino Shock‘.

The Japanese font problem
To understand Hiragino Shock, it’s important to understand what was happening in the Japanese print market. It was a strange post bubble time when the migration path to the newly announced OS X and ATSUI Carbon framework was not all that clear. Just before OS X launched I wrote about the situation for The Seybold Report.

Japanese DTP arrived with Apple’s NTX-J PostScript printer and Linotype’s first Japanese PostScript imagesetter in 1989. They came at the right time: The early ’90s economy was bubbling, companies had money to burn and Japanese DTP took off. It was a young, booming market and it forgave many mistakes that would haunt the industry later.

By 1996, the go-go days were gone and they would not come back…DTP tools (Quark Xpress, Illustrator, Photoshop and, to a smaller extent, PageMaker) had captured nearly 40 percent of the production process. For a conservative industry like Japanese publishing, this was phenomenal―until compared to the West. There, in the same amount of time, practically the entire industry converted to DTP production. Japan is still about 40 percent and holding.

The Second Wave of Japanese Desktop Publishing, The Seybold Report Volume 30 Number 6, November 2000

DTP growth leveled off because market leader Sha-Ken refused to license their font library for PostScript, keeping it locked to their proprietary hardware, and because the first Japanese PostScript font format had severe limitations: they were a 1 byte hack with horrible performance issues that had to reside on a hard disk attached to the printer, they had small glyph sets and were very, very expensive.

A single unlimited resolution Morisawa PostScript font for an imagesetter cost more than ¥200,000. A basic set of 5 Morisawa fonts was a requirement for every Japanese PostScript licensee. When I was the imagesetter product manager at Heidelberg PrePress (old Linotype-Hell) our #1 customer request was for imagesetters without bundled Morisawa fonts, but our hands were tied by Adobe Morisawa Japanese font duopoly licensing.

Adobe and Morisawa addressed some of the original PostScript problems with CID, the first 2 byte PostScript format as a forced upgrade. The per font CID upgrade was not only expensive and time consuming (the floppy disk era), it also changed font metrics and Kanji designs. At the time Morisawa admitted customer reaction to the CID upgrade was negative: “Our biggest marketing challenge is how to explain all this to customers. We understand how customer feel confused. Part of the problem was that Adobe took so long coming out with CID.” Another part of the problem is Morisawa changed the font spacing data in their CID fonts without really telling anybody about it. Designers and production line operators opened documents and discovered they had to redo everything. The resulting publishing industry uproar forced Adobe and Morisawa to pull the CID upgrade and release an update for the update called ‘New CID’ that incorporated both new and old metrics for compatibility.

Things eventually settled down but the CID upgrade disaster left customers feeling wary about yet another upgrade, this time to OpenType. The uncertainty was palpable when I interviewed font designer Osamu Torinoumi of Jiyukobo for the Hiragino profile piece in mid 2002. The beautiful Yumin font had just been released:

JB: You have a new font, don’t you?

Torinoumi: Yes, the Yumin font. We’re releasing it as an OpenType font. But it doesn’t have the OpenType Pro glyph set. We will add other glyph sets gradually as an upgrade. Adobe Japan 1-4 doesn’t have all of the Japanese Industrial Standard (JIS) X0213 glyphs. JIS X0213 has all the possible characters used in Japan so that is a standard that will probably be used as we go along. Apple’s APGS (Apple Publishing Glyph Set) has all of X0213.

Our Yumin is ¥100,000 for a single license, and that is just for the first year. The second year, it costs ¥70,000. We also have volume licenses. We have to have a business model so we can keep adding more features (glyph sets, etc.). I think Morisawa must be having a tough time with OpenType…We really don’t know how it’s going to pan out.

The OpenType business model
Torinoumi’s comments succinctly outline the challenges facing font vendors at the time. Japanese OpenType wasn’t just a font upgrade for end users, it was a whole new business model for Japanese font vendors.

Adobe’s OpenType Japan announcement at Seybold Seminars Tokyo 1999 was the first official sign that the Japanese DTP industry had to migrate away from the PostScript printer font business model Adobe created 10 years previously. OpenType Japanese fonts downloaded dynamically and did away with printer fonts, they were no longer necessary, Japanese fonts finally worked like Western fonts.

All well and good, but there was the catch: Japanese fonts are expensive to produce, and the market was relatively smaller than the western one. How would a font maker create revenue to pay for all the work involved in making large glyph sets?

Jiyukobo’s Yumin font was the first new OpenType Japanese business model concept. Instead of selling a software package, they sold an annual subscription. In exchange, users got free tech support and upgrades like Adobe Japan 1-4 extended glyph sets, advanced layout and any other features that came along later. In a way it was a return to the traditional business model with printers paying monthly royalty font fees for use on proprietary typesetters…without the typesetter.

It took a long time for the industry to realign. Fontworks Japan launched their LETS program in 2001, which stands for Leading Edge Type Solution. While users could still buy a packaged font for ¥18,000, a single LETS license covering use of the entire FontWorks library cost ¥36,000 a year per client in addition to a one-time LETS “membership” charge of ¥30,000.

Japan’s first and largest PostScript font vendor, Morisawa launched their first OpenType products as OpenType Pro at ¥26,000 per font in 2002. No OpenType Standard, no upgrades for CID or OCF users, take it or leave it. At the time I asked Morisawa product manager Nobuaki Nakamura for a comment and he practically sputtered, “That’s right, no upgrades; and we’re not going to follow anything Apple does because it’s non-standard. Print it.” The remark shows how angry Morisawa was at Apple for bundling the Hiragino font and creating the Apple Publishing Glyph Set. Morisawa eventually came out with a LETS-like subscription service called Morisawa Passport in 2005.

APGS Shock
The Apple Publishing Glyph Set (APGS) was a MacOS Japanese character set initiative, separate from but simultaneous with the bundling of Hiragino. APGS created a stir because it added another set of glyphs above and beyond Adobe’s AJ 1-4 specification. At the time, Japanese font developers grumbled about yet another Apple-created spec that, like QuickDraw GX, would go nowhere. Yet it turned out to be a shrewd move.

APGS was based on the JIS X0213 standard (which was also released in 2000) with some additional glyphs from high-end Sha-Ken systems and the National Language Committee. Any JIS standard is the “final word” for Japanese character sets. By adopting JIS X0213 and releasing it as part of OS X, Apple created an instant “super-font” standard of 20,291 glyphs and upstaged Adobe Japan 1-4. Yasuo Kida was an Apple engineer who worked on APGS. He shared some recollections of the project.

Understanding Hiragino
When I visited Jiyu-kobo’s office for the first time, there was a drawing of the character 「あ」in mincho style on the wall. It drew my eyes and even to my untrained eyes it was obvious that there was nothing like that as a digital typeface. I suddenly realized everything.

I knew people complained that they did not have the digital typefaces that they wanted, but I did not really understand why they complained. Was it just nostalgia, or just an excuse for not changing the way they do their job? As soon as I looked at that character however, I realized that this is THE typeface they wanted but didn’t have.

I do not remember if it was Torinoumi san or (co-designer) Takada san who explained that it was a design called ‘A-Mincho’ and the A stands for “atarimae”. It was a typeface they were developing for printing literature. I fell in love with that typeface. The revelation was not related to Apple’s business decision or anything but it was the moment that I made up my mind. Apple licensed Hiragino, and later on the A-Mincho, or Yu-Mincho typfaces for iBooks.

The challenges
It was the late 90’s when fonts were added to my responsibilities. I saw a few challenges:

1) TrueType was not usable in DTP because of the limitation of the resolution…the bundling TrueType in MacOS created problems for customers.

2) Lack of font varieties: Because you needed to have matching fonts on the printer’s hard disk to print your document, and because printer fonts prices were astronomical, the only reliable fonts that you could use at any service bureau were the few Morisawa fonts bundled with the output device.

3) Sha-Ken fonts were not available: many professionals complained that they can’t do their print job with DTP because Sha-Ken fonts were not available.

4) Lack of characters: professionals also complained that the Kanji glyph sets that they needed were not available in DTP PostScript fonts. These fell in two categories. One is ‘itaiji’ of Kanji that are either in or outside of the JIS X 0208 character specification. Another one is Kanji that are not in JIS X 0208. Interestingly we found many of the latter category were not in JIS X 0212 either.

At that time Mac OS was moving to Unicode. We needed therefore to define the base Japanese character set for the Unicode age. It was also the best timing to resolve the problem #4.

The APGS project was born in this environment. We did two things. One is to define the base character set which resulted in adopting JIS X 0213. The other is to define professional glyph set, APGS. APGS was done independently from choosing the fonts we bundled, however both projects went in parallel for the most part. Adopting APGS was a prerequisite for fonts we bundle.

The Solution
We decided to adopt JIS X 0213 as our new Japanese base character set replacing JIS X 0208 (the previous Mac Japanese encoding). JIS X 0213 was not final at that time but was taking form. We had some doubts if JIS X 0212 would fulfill people’s everyday needs, and we liked the approach that the JIS X 0213 team was taking. Being the base character set means we expanded our Kanji input method to take advantage of the JIS X 0213 character set.

For professional needs, we investigated every single character that phototypesetting systems had, e.g. up to Sha-Ken gaiji set C, and determined if the particular character should be included or left out. We wanted to say we have most if not all characters that are available in legacy systems including their gaiji plates that are not usually available.

In the process we learned that Adobe was also working on defining a new glyph set (AJ 1-4) with similar goals in mind. I wanted to merge the efforts but it did not work out. We could not just use Adobe’s effort as it seemed they were concentrating on itaiji of JIS X 0208, and were not working on adding Kanji that were completely missing. It meant we needed to continue with our own efforts.

At the same time, because the target was the DTP market, we needed to be upward compatible with whatever Adobe has. We made our glyph set upward compatible with theirs, expecting they will add ours to their glyph set later which is what happened.

At the start of the project we had decided to adopt Adobe’s format, i.e. CID, because TrueType was dead in the Japanese DTP market. I then changed the format to OpenType following the industry’s move. Hiragino became the first non-TrueType font bundled in Mac OS, Hiragino also ended up being the first Japanese OpenType font in the market, not our intention but it happened.

Because we did not adopt the Adobe Japan 1-4 glyph set, it forced Adobe to adopt APGS but things worked out nonetheless. Adobe’s development of InDesign J, capable of handing Unicode, and supporting authentic Japanese line layout, was a critical component in re-igniting the DTP market in Japan. Morisawa was upset at the time because Apple’s move changed the structure of the font market, but the market direction were already clear when Adobe announced OpenType J in 1998.

Results and some regrets
Hiragino being OpenType removed the need for printer fonts. Because APGS effectively had all characters found in the largest character sets on phototypesetting machines, it addressed complaints from professionals that there were not enough characters in DTP production.

The Hiragino font has Sha-Ken lineage and was welcomed by people who had complained about the lack of Sha-Ken typefaces. They all come free with Mac OS X. Taken together with InDesign J they removed all excuses for not moving to DTP.

One of our concerns was that killing the printer font business, and increased development costs because of the extended character set would negatively impact the health of the font business. To survive the change we believed the font business needed to transition to the subscription business. And this is what happened. Fontworks started its LETS subscription, and Morisawa followed with Passport. It greatly widened the variety of fonts available for people to use. It was nice.

One small technical regret was that when we defined the properties of each glyph (e.g. fullwidth or proportional), we defined math symbols and some other symbols defined in JIS X 0213 as proportional as they should be in the internationalized system (which caused some turmoils in the Japanese OT world). At that time we decided leave Greek and Cyrillic fullwidth but if I were to decide now I would make them proportional.

Another regret was that we should have created a solid subset (or subsets) of APGS. Applications that really need the whole set of APGS are books. Display typefaces obviously do not need the whole set, nor do magazines and so on. We were concerned that font developers might think it necessary, or be pressured from customers to develop whole AJ15 or whatever whole set for every single font that they have when it is not really necessary.

To demonstrate this was not the case, we intentionally left some of our bundle fonts with a smaller subset. But it was not enough, it did not establish a solid character set category that everyone can follow. I should have worked with Adobe to develop a good standard subset based on X 0213 and give it a name. AJ 1-4 is not a good subset as it contains itaiji that many of applications do not need, and it does not contain important characters from JIS X 0213. AJ 1-4 is effectively used as a fallback subset right now.

In September 2002 the other shoe dropped when Adobe announced Adobe Japan 1-5 that incorporated APGS. A senior Apple engineer who had been part of every Apple OS internationalization project from Pink, to GX, to Taligent and finally MacOS X had this to say about the Apple • Adobe relationship: “If it wasn’t for GX, OpenType would never have happened.” I could not agree more and add that if it was not for GX and AAT tables, we wouldn’t have extended character sets and variable fonts the way we do now. Torinoumi thought that JIS X0213 (AJ 1-5) would become the standard of the high-end market. He was right.

For many developers (Adobe included), creating larger and larger fonts was not the best solution to handle the ever-evolving character standards. Adobe did go on to create more Japanese glyph collections but their ability to rally the industry around them diminished over time. Back in 2002 I thought that most Japanese fonts would probably stop at AJ 1-4, leaving Apple in the enviable position of giving users a industry standard super-font with every copy of Mac OS X…not a bad place to be. It’s pretty much how things panned out.

---

Developers and APGS enhanced Hiragino

I believe it is in everybody’s best interest, Adobe included, for Apple to continue innovating.

Isamu Iwata, Ergosoft marketing director 2002

Ergosoft was one of the very first Japanese Macintosh developers. In the 1980s, their egword text-editing application and egbridge input module were staples for every Japanese Mac user. The arrival of ATOK from JustSystem (for kanji input) and Microsoft Word put an end to Ergosoft’s market dominance in the mid-1990s. At the same time, a foray into cross-platform development turned out to be an expensive disappointment, and Ergosoft dropped Windows development in 1997. OS X gave Ergosoft a new opportunity, and the company took it by adopting Apple innovations that other developers ignored: ATSUI, the extended glyph set for Hiragino and the power of Apple’s AAT (Advanced Typography Tables).

In 2002 Ergosoft product manager Isamu Iwata sat down to discuss their products that used these technologies. It shows a Mac software industry in transition.

JB: One of the big features of egword 12 and egbridge 13 is the ability to use all of the Hiragino APGS extended glyph set. Was it difficult implementing these features?

Iwata: There was some engineering work using ATSUI, but making the decision to use ATSUI or not was a much more difficult decision. It was a little risky. Apple made lots of promises with QuickDraw GX, then completely dropped it. It took us two years to finally decide to support ATSUI. We did surveys of other developers to see if anybody was using ATSUI, but did- n’t find a single one. We were worried that if we were the only ones, it would disappear. I’m fairly certain we are the first developer to use ATSUI in a big way. Not even AppleWorks uses ATSUI.

We put a lot of effort into making a GX version of egword, but ended up throwing it away when Apple killed GX.

JB: Did you see any opportunities in using ATSUI?

Iwata: The deciding factor was the Hiragino fonts becoming part of Mac OS X. Even though Morisawa and the other font makers don’t have APGS [AJ 1-5] extended glyph sets, we still felt it was an opportunity, because it is part of the base system. The Hiragino fonts are high quality; they can satisfy the DTP market but also be of real value to other users as well.

JB: Does egword use any other ATSUI features besides the extended character sets?

Iwata: Not at this time. There was also a bug in ATSUI [prior to 10.2] that prevented us from accessing approximately 300 glyphs. We told Apple engineering when they visited here and they were very surprised. They promised to fix it in the next OS update. Such is the risk of being the only developer to use ATSUI and Hiragino.

JB: How do you maintain font-data compatibility with Adobe’s InDesign [which does not use ATSUI]?

Iwata: Hiragino has CID ID tags so the correct glyphs are imported [and displayed, printed, embedded in PDF documents, etc.] However, we cannot import some glyphs because they do not have CID tags. [Fixed in 10.2.]

JB: All of APGS does not have CID tags?

Iwata: Not all, only some…. egbridge can access all of the characters because of AAT and ATSUI, but applications like Excel and Word cannot accept them.

Ergosoft parent company Koei KK (now Koei Tecmo Holdings) shuttered the subsidiary in late 2007 and exited all Apple related business. The timing was unfortunate. After Apple released APGS in 2001 they did nothing to promote it or ATSUI. iPhone did not go on sale in Japan until 2008, the App Store economy had yet to materialize. If Koei KK had waited a few months things might have turned out very different.

Norihito Hirose and Kenta Arano, the very talented lead programmers of Ergosoft set up with own company Monokakido in early 2008 and created some of the first hit Japanese iPhone apps such as the Japanese dictionary Dajirin. They purchased the assets of egword and egbridge and resurrected them. They are still the best Japanese word processor and Japanese text input module on macOS.

---

Time Line

January 1994 QuickDraw GX launch
After being unveiled at MacWorld Tokyo 1989, QuickDraw GX launched in early 1994. Key font technologies include: extended character support, Apple Advanced Typography (AAT) tables, GX variable fonts, Open Font Architecture.

1995 CID PostScript Japanese Font Upgrade Disaster

August 1996 Apple cancels Copland OS

December 1996 NeXT Purchase (NEXTSTEP)

October 1998 ATSUI (Apple Type Services for Unicode Imaging)
Key GX text technologies brought forward for MacOS 8 and MacOS X: extended character sets, Apple Advanced Typography (AAT) tables, GX variable fonts.

November 1999 Adobe Japan 1-4
Adobe announces the first ‘Pro’ extended character collection specification and OpenType Japanese fonts with 15,444 glyphs at Seybold Seminars Tokyo.

February 2000 MacWorld Tokyo ‘Hiragino Shock’
Steve Jobs announces high end Hiragino fonts will be bundled with Mac OS X

March 2001 MacOS X 10.0 Hiragino
MacOS X 10.0 ships with Hiragino, the very first OpenType Japanese font.

September 2001 MacOS X 10.1 Apple Publishing Glyph Set
Apple Publishing Glyph Set enhanced Hiragino with 20,291 glyphs ships in OS X 10.1.

September 2002 Adobe Japan 1-5
Adobe announces a new OpenType Japanese font extended character set with 20,317 glyphs that incorporate the APGS extended character set.

---

Looking back
Hiragino Shock and Apple Publishing Glyph Set marked the end of an era of bold new typography developments in the Japanese market. Adobe went on to create other Japanese glyph collections but their ability to rally the industry around them diminished over time. The last time that Adobe and Apple cooperated was the OpenType Variable Font specification in 2016 which incorporated the Apple TrueType GX model and AAT tables, though these have yet to see any major release in the Japanese market.

It’s a shame the QuickDraw GX vision of international savvy, insanely great typography and layout as standard universal features all developers implement easily in high-level OS frameworks and standard app feature sets never survived. In today’s modern OS landscape I find it depressing that OpenType advanced typography features only live in full glory in Adobe Creative Suite apps. Although it is almost forgotten history, Apple’s font efforts for OS X raised the bar for standard OS typography features in Japan and moved the industry forward, for that users can be grateful.

Thanks to Yasuo Kida for sharing details behind the Apple Publishing Glyph Set initiative.

---