Old Phototypesetter Tales home

        Credits -

This document was created by Clark E. Coffee, who founded and operated Graphion, a phototypesetting company in San Francisco, that pioneered many computerized typesetting techniques. I was involved in some of these developments. Clark sold the business in 2000 and retired.  I have included his document with a few minor updates in my web site because of its historical interest.


The past thirty years have seen the rise and fall of an industry and a trade, called "phototypesetting". As it grew, it replaced the lead type that had been around since the Renaissance with generated automatic systems. It now has virtually died, as small inexpensive computers and "desktop publishing" have replaced the old skills.

Having ridden this technological wave this far, it seems worthwhile to preserve a little of the history, gossip, and individuals who made it happen. Most of what we'll write is from imperfect recollections, which we'll be happy to correct when errors are called to our attention. We'll check our stories, when we can, but in this case it's true that "history is old gossip," so don't take this work as a final authority.

Our knowledge is limited to the developments in the United States, though we've heard of a machine developed by Nobuo Morisawa, in Japan that used many of the features later used in Europe and the US as early as the mid 1920s. The Morisawa Co., and an off-shoot Sha-Ken, still manufacture typesetting equipment in that country. We are accumulating information regarding early developments in France and Germany which we'll add when we can.

There has always been a problem with the words "typesetter" and "phototypesetter" which can mean either a machine, or a person. We'll capitalize them when we mean a person in this work.

First a Word about the People We Replaced

Typesetting as a skilled trade originated in the renaissance. The Typesetter was solely responsible for the appearance of every page. The wonderful vagaries of hyphenation, particularly in the English language, were entirely in the Typesetter's control (for example, the word "present" as a noun hyphenates differently than the same word as a verb). Every special feature: dropped capitals, hyphenation, accented characters, mathematical formulas and equations, rules, tables, indents, footnotes, running heads, ligatures, etc. depended on the skill and esthetic judgment of the Typesetter.

Such was the attention to detail and pride in the appearance of a well composed page that Typesetters would occasionally rewrite bits of text to improve the appearance of the page. This greatly annoyed Mark Twain (who began his own career as a Typesetter) and encouraged him to invest heavily in an early, and unsuccessful, attempt to produce a keyboard-driven typesetting machine that wouldn't edit his words.

There was a romantic tradition, in this country at least, of the drifter Typesetters, who were good enough at the craft to find work wherever they traveled. They'd work in one town until they wanted a change and then drift on. They had a reputation for being well read, occasionally hard drinking, strong union men who enjoyed an independence particularly rare in the 19th century.

Typesetting was a skilled and respected trade even after the keyboard-driven typesetting machines were introduced, around 1900. These machines typically produced lead strips for each line of type, which were stacked in a frame, proofed (the type was of course backward), and clamped into columns or pages. Extra space between lines was supplied with thin strips of lead, inserted between lines. Pages such as price lists and directories would be kept in "standing type" and edited by adding and removing individual lines of type. Large type in headings, etc., was likely to be set by hand and combined with the machine set lines. The International Printing Museum in Carson, CA. https://www.printmuseum.org/ has some of the antique hot lead machines on display.

The I.T.U.

The International Typographical Union, was described as, "the oldest union in America, and organized to prevent the use of labor saving improvements." The union fought hard for its members and when times were hard would send money and train fare to unemployed Typesetters, and direct them to places where prospects were better.

When preset advertising copy began to be provided by advertisers, in the late nineteenth century, the union required that this type could be used as received only if a union Typesetter was employed to reset, print, proof, and throw away the same copy. The union leader who negotiated this requirement is reported to have been a Mr. Bogus, and this redundant make-work typesetting was called "bogus" type and added a word to the language. (There are other explanations for the word, but none that we know of contradicts this one).

Even as late as the 1980s, most type was set on lead casting machines, and the production manager at the San Jose News complained that his reporters' stories were being retyped by "400-dollar a month secretaries who type 80 words a minute and don't make mistakes, and then retyped at 40 words a minute on Linotype machines by 800-dollar a month Typesetters who do make mistakes."

In the 1970s when the machines that set type began to use low cost mini, and later microcomputers that automated the old typesetting skills, the need for the ITU members began to decline. One after another, newspapers that were already losing advertising dollars to the new upstart television were hit by ITU strikes called to prevent the loss of jobs for Typesetters. One by one these papers closed their doors forever, and Typesetters were really put out of work. Finally the union had to settle for agreements that said basically, "you can't fire our people, but you can give them any kind of honest work you have available." Since these Typesetters had an average age of over 50 years, the papers could use them for anything from driving trucks to managing the paper warehouse, and they'd all be gone, replaced by people with lower wages (if inflation didn't make the wages equal) within 10 or 15 years.

A sad 49 year old Typesetter told me in 1978, "My Daddy always told me 'get a trade', so I did my apprenticeship and became a Typesetter! Now I'm unemployable!"

The ITU no longer exists as an independent union. It had a long proud history,protecting and getting good wages for its members through some very hard times for trade workers. We'd be well advised to realize that most of the jobs we do so well now will probably go away or change completely in a single life time, and when you reach the age of the Typesetter quoted above, you probably won't have a union working to protect your right to work. So stay up to date!

Lewis Mumford tells us that the guild of scribes and copyist, delayed the introduction of printing presses into Paris for as much as twenty years. In this century people and machines become obsolete almost over night. Absitomen

See the story of one paper that survived.

The Rise and Fall of Phototypesetting Machines

How It Came About

There were phototypesetting devices of one kind or another, including one patented in Japan by Nobuo Morisawa in 1925, but the revolution, or evolution, that would change the entire process of getting words into print twice in the last quarter of the twentieth century, was brought about by four forces:

  1. The improvement in Lithographic printing presses, and photographically produced litho plates, and the introduction of photographically produced plastic plates replacing lead for letter-press printing equipment.
  2. The development of phototypesetting machines.
  3. The computer automation of typesetting and the whole page make-up process.
  4. Increasing environmental sensitivity, that frowns on dumping metal etching solutions or even photographic solutions into the public drains.

Photon Begins It All

In this country it really began shortly after the end of World War II when two Frenchmen, Higonnet and Moyrou, developed a viable phototypesetter that used a strobe light and a series of optics to project characters from a spinning disk onto photographic paper. They licensed their patents to a Massachusetts firm called Photon, which began producing a series of very expensive phototypesetting machines in 1945. Photon grew to be a major firm, with sales and service offices all over the country.

The machines made by Photon, and the competitors who began to appear, were operated by punched paper tape produced on special "perforators," that had been used for some lead casting machines since 1932. The paper tape used a 6-level code called TTS. Because you can only represent about 36 different characters in six bits, it had to use shift, super-shift, and "bell" characters to get upper and lower case alphabets, numerics, punctuation and special characters. They had large keyboards with all of the special typesetting commands like "quads", em-space, en-space, en-dash, em-dash, open and close double and single quotes, and some pi characters like bullets and stars. At first the skill required to prepare these tapes was little changed from the lead casting machines; the operator controlled the line breaks, based on line lengths shown on a line width counter and his knowledge of hyphenation rules and conventions.

The First Automated Typesetting

In the 1960s when the term "computer" described a large air-conditioned room full of big machines, probably from IBM, RCA, Burroughs, Univac, or DEC (where have all the flowers gone?), typesetting began to be automated. RCA offered a CRT based machine, called a "VideoComp," made in Germany by Hell (now a merged with Linotype), which inspired IBM to fund the development of the IBM 2680, a CRT-based machine, made for IBM by a Long Island company called Alphanumeric Inc. (Alphanumeric had pioneered a digital phototypsetter operated by a small DEC computer/controller). Both the RCA and IBM typesetters were remarkable machines, but the computers necessary to run them leased for thousands of dollars per month, which kept their market relatively small. These machines disappeared because small cheap computers to drive them didn't begin to appear until the early 1970s. The 2680 was the basis for the first Autologic, which contained a minicomputer, replacing the big IBM equipment, but it didn't appear until much later.


In the late 1960s Mergenthaler was purchasing perforator terminals to drive its lead casting Linotype machines, from a Massachusetts company named Compugraphic. This led to a new development... At this time Photon was the leading, if not the only, phototypesetter manufacturer; its president was a man named Bill Garth. He wanted to produce a small, inexpensive typesetting machine, but his board of directors preferred more glamorous large, expensive machines. Garth left Photon and moved to Compugraphic, a much smaller company only a short distance down the road. At Compugraphic he arranged to buy back from Mergenthaler the rights to the machine that Compugraphic had been making for them, which had been developed at Mergenthaler's expense. Then he used the keyboards, logic, and hardware to directly drive a small integral phototypesetter. These machines took up less room than the old hot type machines and were, I believe, the first inexpensive typesetting machines that could be said to run "on line," allowing justified type with several different fonts, and special characters without any paper tape punching. They weren't very versatile, didn't allow a lot of fonts at one time, and didn't set very wide columns, but then the hot type machines they replaced were even less clever. Best of all, they didn't worry the ITU because they simply replaced the hot type machines without reducing the number of operators.

In only a few years Compugraphic was doubling in size every year. It is said that at a large typesetting show, Garth pulled a chair over across from the Photon booth, where the salesmen were standing around watching the stream of people pouring in and out of the Compugraphic booth. He sat there smiling, entertaining his friends, and making Photon's management miserable.

This was of course noticed by Harris Intertype, Mergenthaler, and Monotype, who had been the foremost makers of the lead casting type machines for 60 or 70 years. They set about making their own phototypesetting machines, when Photon showed the way.

Harris Intertype

The Harris phototypesetter, called a TXT, was about eight feet long, four feet wide, and six feet tall. The fonts were on a half dozen platter-sized glass plates spinning at high speed at the ends of radial arms that rotated around when a type face change was called for. These monsters were used by newspapers well into the 1980s.

Mergenthaler Linotype

Mergenthaler's first phototypesetter, the Linofilm, came out in 1954, but by 1970 they produced a phototypesetter called the VIP which held six fonts at a time (one reserved for punctuation and special characters like the asterisk, etc.), and selected sizes from 5 to 35 points with a moving zoom lens. (Special font strips could be used to set type from 35 to 72 points.) It would set a page of type in about four minutes. The type fonts for the VIP were on film strips a little larger than a business card, and cost hundreds of dollars each. They were not priced as families of light, medium, italic, bold, & bold italic; such a list would require five separate purchases. Phototypesetters nearly always had well equipped dark rooms, and a special punch was soon on the market to duplicate the registration holes in the original font strips, so that anyone could make copies of these fonts. Since they could be easily damaged in cleaning, and frequent changing, it was a good idea to keep a copy or two as backup even if you had only one VIP.

But then a company named "Storch," started offering font strips for the VIP at very low prices. Mergenthaler of course filed suit, and dropped its prices. The font prices went from $100 or $200 to $30 or $40. These affordable fonts made the VIP the dominant electromechanical phototypesetter, and was probably the best thing that ever happened to Mergenthaler. The VIP was followed by a hybrid machine, the 408, which used an image tube to scan the characters on a film master, and a CRT to place them on the output copy. It found considerable use in newspapers. The 202 family (about 1980) were truly digital typesetters with fonts stored as digital data in memory, in its minicomputer controller (marketed as the Naked Mini) or on a disk drive. The firm introduced the concept of the "imagesetter" with its 300 in about 1985, recognizing the change from "typesetting" to "graphics." That machine's descendants are still a dominant output source in 1995.

Having carried its founder's name for about 80 years, Mergenthaler changed its name briefly to "Allied," which most people thought was a moving company, then quickly changed it again to the name of Ottmar Mergenthaler's 19th century type casting machine, the "Linotype." More recently it has combined with the German firm Hell, as Linotype Hell.


Others had noticed the sudden growth in the typesetting equipment market. One was Singer Corp. which set up a national sales force for typesetting and graphic arts equipment, including the phototypesetting machines made by another Massachusetts company, Graphic Systems Inc. These machines had a brief but spectacular run in the 1970s until someone at Singer asked what they were doing in the typesetting equipment business (which they knew nothing about), so little GSI was left to its own devices with no marketing organization. Singer had been responsible for providing the type font masters and the machines were getting a reputation for poor base line regularity. This may have been a hardware problem because strobe timing errors showed up as base line errors on the GSI machines, while on the Mergenthaler VIP such errors showed up between the characters, and unless fairly serious, escaped notice. GSI began producing excellent font masters and the machines continued to find a market. They built a minicomputer into the machine (a Nova), making it capable of setting justified, hyphenated type from a pure text input using a particularly convenient coding system that owed nothing to old Linotype practices. It allowed stored formats, or in-line codes. These machines held four fonts at a time, but had more characters per font than the VIP, so that the asterisks and other common characters were included. They used a rotating turret of lenses and a second "doubler lens" to offer a range of sizes from 5 to 72 points. The GSI machines were particularly popular with the in-house typesetting departments that began to grow in most larger companies. (Graphion's first typesetter was a GSI we called Beverly, in honor of Beverly Sills, because it was a Singer. We used Beverly for several years before moving on to the next generation machines, and parted with her sadly).

Who Owns the Typefaces

There was a wonderful period in the late 1970s when most of the important typefaces where owned by, or under exclusive license to, Harris Intertype, Mergenthaler, and Monotype, so the growing new companies simply copied these fonts and renamed them. Phototypesetters like ourselves had to keep conversion lists so they'd know when a customer asked for "Toms Roman," he'd been dealing with a Compugraphic system. The same type would be called "London" on a GSI or Singer machine, or "Times Roman" on a Mergenthaler. The fonts were about as similar as anyone could make them, but the serif on the descender of a letter like Q might be longer or shorter, to justify this outright theft of type styles by claiming that they were unique new faces.

The End of Photon

By the late 1970s, Photon was seeing its markets erode. According to the book, Words Into Type, (Prentice Hall 1974) "by 1972 over 100 machines to set type photographically were on the market". Photon responded by launching a crash program to develop a low cost machine to be called the Pacesetter and by entering into an agreement to market as a Photon 7000, the new CRT machine being developed by Autologic, in California.

The Pacesetter was a very fine electro-mechanical machine, potentially a major threat to the Mergenthaler and Singer machines. But its crash program was a very expensive way to develop anything. The machines were late coming out. In desperation, some were shipped as late as the end of January 1974 with December packing slips so the sales could be credited to the last quarter of 1973. One technician described them as "going out with wires hanging out of them." The local sales and servicing offices were promptly swamped with problems that should have been solved at the factory. There weren't any spare parts, so they'd cannibalize the first machines rejected by angry buyers, to try to keep, or get, the others running. Within the year Photon had closed its doors. But some of the Pacesetters had long, useful lives.


It had become obvious to Alphanumeric, in 1969, that the IBM mainframe hardware requirements to operate their typesetter, called the IBM 2680, were going to keep their sales volume very low. Alphanumeric had become its own best customer, operating its machines on a service bureau basis in a number of major cities. The company decided to spin off the typesetter development to a new corporate shell, and replace the big IBM systems with a minicomputer controller. To do this they bought a small San Francisco company called Autologic, which was in the process of developing a Computer Aided Design system having nothing in particular to do with typesetters but offered a clean corporate shell. After long and arduous negotiations with a variety of potential investors, a major share of the company was sold to Volt Inc. The typesetter development was moved from Long Island to Southern California, and the CAD activities, after a few sales, were phased out.

Which brings us to the Autologic machines. These fast, wide-output machines were first offered through Photon as the Photon 7000, the numbers being the last digits of Photon's office phone number. The combination of new product start-up problems, and the growing corporate sickness at Photon doomed that agreement. The Autologic machines continued after the demise of Photon and carved a major place in the market for newspaper production, a field it dominated. From the beginning these have been 100-pica wide machines (16 and 2/3 inches), allowing them to set an entire newspaper page in one pass with no need for paste-up, and the type was of very high quality. It is reported that, by mid 1995, Autologic no longer produces any hardware, but continues to market its newspaper pagination software.  In 2002 Autologic was bought by Agfa Gevaert.

There were Others

As we've mentioned, "by 1972 over 100 machines to set type photographically were on the market," including Harris Intertype, GSI, Monotype, Friden, Mergenthaler, and Photon, so we've only skimmed the surface here. We need stories about some of the other players, and would welcome input from anyone interested.  Please direct emails to dan@haagens.com

Tapelim -- A Footnote

An interface device called a Tapelim was produced by Graphion in 1974 (yes, that's us, we built it for our own use, then continued to make and ship them in various forms for about 15 years). The Tapelim looked to a computer like a serial device such as a modem, and replaced the paper tape reader in the typesetter. This allowed the GSI, Mergenthaler, and Harris Intertype machines to be operated directly by the minicomputers that had started appearing about 1971, and later by a variety of microcomputers from CPM to PCs. Interestingly enough, the Mergenthaler 202 series of machines which introduced true digital CRT typesetting were first marketed as paper tape-driven machines, although a number of on-line interfaces were later added. The Tapelim was being sold for use on these machines as late as 1990, in conjunction with a program, created by Jeremy Griffith, which would convert Ventura Publisher page descriptions to the 202's language, called CORA.

Front End Systems

Without a computer to drive them, phototypesetters are just like the old linotype machines except that they produce paper instead of lead. But, with a computer, all of the old Typesetters' decisions can be programmed. We can kern characters with abandon, dictionaries and programs can make nearly all hyphenations correctly, lines and columns can be justified, and special effects like dropped capitals become routine. The type data itself can be part of or used by: billing programs (classified ads), membership records (directories), hardcover books and pocket books, etc. It was the development of these programs and the appearance of low cost computers that began the change to today's desktop systems.

By the early 1970s, there were systems being developed by firms like DEC, Bedford, Information Systems Inc., and in-house developers like ourselves, that would provide automatic hyphenation and pagination using the minicomputers that had just become available. At the time they seemed cheap compared to renting time on IBM machines at between $50.00 and $150.00 dollars per hour. However, the first Graphion computer, a superior 16-bit machine which we called RALPH, cost about $200,000 in 1970, with the total software development costing as much again. We acquired it at a bargain price of about $180,000 in 1972. (By way of comparison, this was about the time we bought our house in San Francisco for $25,000). We now have a room full of PCs, any one of which is more powerful than that first machine, and they cost about $2,000 each. We were fond of Ralph, but sadly sent him to the landfill many years ago.

Phototypesetting R.I.P.

The development of WYSIWYG systems (What You See Is What You Get) began in the 1980s with a variety of proprietary systems, most of which settled for as little as two screen fonts, one serif and one sans-serif. It wasn't until the Adobe output description language called PostScript became an industry standard, that output machines and input systems could begin to offer the generalized "Desktop Publishing Systems." These were first offered for in-house newsletters etc., but have continued to mature with many of the automated typesetting features that had only been available in such systems as the Bedford, Information Systems (developed by Fred Rose), and our own.

Photographic output for type is now challenged by high resolution laser devices that double in resolution every year or two. Reasonable type has been available at 1000 dpi for several years, but some firms offer up to 2400 dpi in mid 1995. Some use laser printer engines like the Cannon, or Toshiba, that began life as low resolution computer print out devices, but we will surely see this technology making printing plates in the near future. The prospect of getting away from the whole silver halide process has to appeal to anyone who has maintained photo processing machines, and contributed a dollar per page to Kodak profits for any time at all. The prices for these machines is so low we can afford to buy them expecting to replace them with better machines in one or two years, just like the computers we use to drive them.

We purchased a LaserMaster 1000 dpi machine in 1990, when it offered PostScript compatibility, and wrote our own PostScript driver program to utilize our system's output. When we started producing copy, however, we found voids appearing in the type. After rejecting every toner offered by LaserMaster, we learned that HP model III toner produced type on the LM 1000 without voids. The output compared with our three Linotype 202s. As a result, those great expensive machines were given away, or sent to the land-fill, along with our second Oscar Fisher processor. Our customers prefer the plain paper output, particularly for proofing, because it is much easier to write on than RC paper phototypesetter output.

We have more recently tried an 1800 dpi Laser Master machine, and a 1200 dpi Xante machine. Both left voids in the type when using LaserPlus paper from Hammermill, which was delivered with the machines, and had worked for two years with our LM1000-HPIII-toner combination. These machines use Toshiba and Cannon engines respectively. We have found no other source for toner for these machines. At the present state of the art, it is possible to reduce to insignificance the tendency to leave voids by using a very good bond paper, which has a slightly rougher surface. We are presently putting out most of our work on the Xante with Neutech 25 (25% cotton) laser paper. No one has asked us for film, but we know that it's possible.

These machines seem to offer reasonably good half-tone capabilities, and LaserMaster is offering products for color separations and proofing, but our particular business niche has little requirement for graphics. Other high resolution lasers are offered by LexMark, NewGen, XLI, Dataproducts and probably others we don't know about.

The toner/paper problem does not appear to have been solved in a very satisfactory way on the two machines we've tried, but those of us who remember how long it took the photographic paper suppliers to achieve real consistency, or had to live with "stabilization paper" that faded within days, are not discouraged. We can produce acceptable type, and we've been able to give up our downtown office and move into our own "electronic cottage," while producing more pages than was ever before possible.

Where Do We Go from Here?

The Graphion system was begun in 1972, on a minicomputer driving an electro-mechanical typesetter made by GSI. Much has changed since then. We moved to a system called TurboDos, sold a few systems to friends, and drove a whole series of different typesetters, including our last, a series of three Mergenthaler/Linotype 202s. The Graphion system is still alive and moving with the times, running on IBM clones since 1993 and outputting PostScript code to drive imagesetters, laser printers, or platemakers. Since we're dealing in text rather than advertising we still don't offer color, but will when it's required.

We entered the phototypesetting field when it was just beginning, and we've seen it through to the end. We can see no reason to expect that it won't end for everyone very soon. As laser printers continue to increase their resolution there will be no need for photographic processing at all. In fact, we will increasingly be producing only data to run plate-making, or direct input printing machines. Proofing will be on CRTs and our output will be sent out by modem. In some cases the copy will appear on paper only when it's downloaded by a user for some reason, which may save a lot of trees. Unless this "post literate society" becomes an "illiterate society" there will continue to be a market for our services for a long while to come.

The new desktop systems have put great power in the hands of some aesthetically challenged operators who can and sometimes do produce type that is just plain ugly, but it isn't the fault of the computer systems. Even as the phototypesetters, after a slow start, ultimately produced better type than the lead type systems, we expect continuing improvement. For now, we miss the aesthetic sense that professional Typesetters brought to their craft, but it's progress that no one wants to stop.

New Information

We've heard recently (July 18, 1995) from Brian W. Case (bcase@mail.best.com) that he is using a DEC 5100 laser printer with an upgrade board for 1200 DPI. It's a letter/legal size machine available for $999 for the machine, and $350 for the upgrade. As with most other machines full page graphics may require additional memory, but text does not. He sounds quite knowledgable and very pleased with the output which he feels is better than the Laser Master or the Xante. It's giving him good solid type without voids or pinholes, on Hammermills Laser-Plus.

We've seen sample copy from this machine (August 18, 1995) and its output is definitely superior to our 1200 DPI Xante, no voids, and on Laser-Print paper, which we can't use on the Xante or the 1800 DPI LaserMaster. His DEC costs about $5,000 less than either of them!!!!

 "Oh well...It's only money!"

A Successful Change to Phototypesetting

A small suburban newspaper called The San Rafael Independent Journal, generally known as the IJ, under the management of Norman Jaffe, was a pioneer in the change over from the "hot-type" typesetters and into the new automation. They made the decision to move to a Mergenthaler 408 phototypesetter to be driven by a small IBM computer, in about 1969. They expected to retrain their ITU typesetters, some of whom had been with them for many years. Unfortunately, they got into a dispute with the union, which called a strike. The IJ was far enough along, when the strike began, to bring in what they called "ex-legal secretaries, the best typists in the world" and teach them to run their terminals, which by this time were actually modified electric typewriters.

The strike was bitter, tires were slashed, trucks were sabotaged, and finally one IJ man was killed. The murder caused everyone to fall back, and the union eventually dropped the strike, with no concessions from management. The IJ, because they were ahead of most firms, could boast that they went to phototypesetting when their hot type machines could still be sold for money.

At one time visitors from all over the world would come to San Rafael (just north of San Francisco) to monitor the progress at the IJ. With the leadership of Jerry Fingerlos, the technical manager, the IJ developed or helped fund the development of software that would set entire pages of classified advertising automatically, and also would prepare the billing for each advertiser. The input was provided by the person who took the order. The IJ continued using their letter press printing presses, but switched to new photographically produced plastic printing plates.