Archives for category: Typesetting

Lewis Mitchell died earlier this year. He retired from M&H Type (part of the Arion Press set up in the Presidio, San Francisco) in 2014, the year before I visited the Press.

Type casting is the business end of hot metal typesetting: it’s the part where a squirt of molten metal is injected into a mould to create a sort, a piece of type. Mr Mitchell ran the machines that did this at the M&H foundry for many years.

Here Mr Mitchell walks you through the process in a City Exposed video:

If you don’t see a video here. please click on the title of this post in order to view it in your browser.

As a reminder: there are three aspects of hot metal typesetting: keyboarding, creating the type, and arranging the type in the proper order. With hand setting you get lots of little bits of type which someone else has cast for you, and then select in the correct order the letters called for by the copy. Mechanization of hot metal typesetting took two broadly different directions. Monotype chose to keyboard the entire work onto a roll of punched tape, and then rerun it though a casting/setting machine which would output the entire job in individual characters arranged in lines of predetermined length. Linotype integrated things a bit more having the keyboard operator sit at the same machine that cast the letters into lines of characters, not individual letters. You could make a Monotype caster output all the same character, as no doubt Mr Mitchell often did at M&H as part of their foundry business supplying type fonts to other printer customers.

Although they use a “western” QWERTY keyboard for their computers there are people in-putting Chinese characters who are actually able to type much faster than those for whom the keyboard was designed. This represents an incredible turnaround. When computers first arrived in China they appeared to present an overwhelming challenge to the very existence of Chinese script. No way could you get 70,000 characters onto a keyboard with space for about 70. Clearly getting the efficiencies offered by computers would involve adopting an alphabet, wouldn’t it?

To avoid this loss of heritage, the key turned out to be to “spell Chinese characters, not by sound, but by shape”. Professor Wang Yongmin broke the structure of Chinese characters down into 125 elements. Think of early mobile phones with numbers-only keypads on which you could access letters by hitting each number key once, twice, thrice and selecting the letter you wanted when it was displayed — using this technique for all the keys on the QWERTY keyboard Professor Wang managed to create a working computer QWERTY keyboard for his 125 Chinese character elements: select the first element you need for the character you require, then move on to the second element and so on. He demo-ed his keyboard at the UN in 1984, to general incredulity.

One consequence of this method is that different people can use different keys to carry different information based upon their speciality. Chinese QWERTY keyboards, many of which don’t even have any symbols on them, can be and are programmed in a variety of different layouts. Predictive text and auto completion arrived on Chinese computers before we got them — when you type a text message or do a Google search, you get these prompts suggesting to you what word the computer thinks you’re trying to type, and even the next word which you’ll come up with. A bit annoying perhaps, like the related Auto-correct “service”, but an efficient use of artificial intelligence. They were already doing this in China in the 1980s: key in a bit of a character shape, and the machine will suggest how you might want to complete it. Select the correct suggested target and Bob’s your uncle.

By the 1990s the Chinese government had decreed a move to Pinyin transliteration of Chinese, and many computer keyboards now work using Pinyin. However lots of people are still using the Chinese character keyboard — which is more universal than the Pinyin one. Pinyin which is a transliteration of sounds, will look different in different dialects. These dialects/languages use the same script system but pronounce the characters differently, so output from a keyboard with character generation will look the same all across the country, where Pinyin-generated text may be regionally incomprehensible. Another of script input’s big advantages turns out, paradoxically, to be speed. Using the multiple-elements-per-key technology allied to autocompletion and predictive suggestion has resulted in a typist being able to “type” 244 characters/words per minute at a 2016 input contest in Beijing. An extraordinary typist in English can get to 100wpm.

National Public Radio’s Radiolab program tells the story at The Wubi Effect. You can listen to the broadcast there, or, via a tab, go to a transcript of the program.

Do we have to make some allowance in such typing speed trials for the fact that many Chinese words are represented by a single character, whereas the average length of an English word is 4.7 characters? Some maybe, but probably not all that much since the characters are of course the problem: they’re rather complex, and probably more complex than that average 4.7 letter long word.

I wonder if speed is sufficiently important for us to try to emulate the Chinese by coming up with a more efficient method of keyboard entry. We know that when we read rapidly we are tending to recognize word-shapes rather than the individual letters which make up the word. Just because we have an alphabet doesn’t have to mean that the alphabet is the best way to reach any reading or writing destination, does it? But do we need to tread carefully? If you just use predictive text you might go fast, but would you be typing what you wanted or what the cloud thought you should want to say?

See also Setting Chinese, and Chinese typewriter.

The typeface Futura was designed by Paul Renner (1878-1956) at the instigation of Jakob Hegner, a publisher and printer who was looking for “the typeface of our time” to counterbalance the old Fraktur types hitherto prevalent in Germany. The fact that Renner was a painter rather than a type designer was part of his appeal to Hegner. The typeface was designed in 1924-26 and issued by the Bauer Typefoundry. Renner was particularly pleased with the lower case alphabet, but Simon Loxley tells us in Type is Beautiful (Oxford, Bodleian Library, 2016) that modern opinion holds the upper case series in greater regard. The early ad shown above illustrates both.

You are certainly familiar with Futura. According to Fast Company it is the most imitated typeface. You have to remember that in hot metal typesetting days you couldn’t just copy a typeface even if you had permission: you had to make your own matrices in order to cast the hot metal to this design. (It was sometimes possible to negotiate a price to buy mats from the licensing foundry.) Even with the best will in the world it isn’t surprising that variants crept in right from the outset when matrices were being recut. Some foundries, in order to avoid licensing fees, went the whole hog and made their own intentional tiny alterations, and within ten years of its introduction a Futura derivative (often without any reference to that name) appeared in the catalog of every American type foundry.

Here is the cover of Bauer’s 1930 type specimen telling you it was available in Three-quarter-bold, Oblique, Light, Semi-bold, Bold Condensed, and Black.

Germany took the avant garde lead in the arts after its defeat in World War I: think of Bauhaus, modernist movies and so on. World War II resulted in a bit of “backtracking” in this as in many other regards. Here’s a video, forwarded by David Crotty of The Scholarly Kitchen about this.

If you don’t see a video here please click on the title of this post in order to view it in your browser.



We are so used to clip art nowadays that we still call it that even though there’s no longer any need to get out your scissors and clip anything at all. You just copy and paste clip art on your computer now. Clip art itself was anyway just part of an interim phase in print reproduction when people could get a little print job done on an offset press at their corner printshop. Before that there was an extra step in the process of reproducing a little picture.

With letterpress everything has to print from a raised surface, usually metal but also potentially wood, some synthetic substances and even rubber — think of that John Bull printing kit you had as a child or the date stamper in the office. Or notice next time you step in a puddle . . . the next step you take on the dry street will leave a letterpress impression: a rather evanescent one it’s true — for a more lasting impression stir up some mud while you’re in the puddle.

So, when almost everything had to be printed by letterpress, you needed to have access to a raised reversed image to receive the ink and transfer it to the paper. In order to be able to print a little picture you had to go to an engraver and create a block* (called a cut in USA). To get a block engraved cost money, so unless it was totally specific to a particular print job which was never going to reprint, you would carefully wrap each one in paper, label it carefully, ideally with a pull (proof) of the engraving on the outside, and put it into storage in case you ever needed something like that again.

Here’s a piece showing part of one printer’s collection. It is shown in two separate photos, though it is just one 12⅝” x 17¾” sheet. It was printed by The Quarto Press in Coupar Angus in Scotland. You can enlarge the pictures a bit by clicking on them.

I suspect that the main sheet is actually a reprint by offset from an earlier version which was done by letterpress from the original blocks. At the bottom of the main sheet you can see the claim that it was printed in an edition of 75 copies on a Vandercook press by John B. Easson at The Quarto Press in Feltham, quite a long way from Coupar Angus. At the top we are informed to job was done in October 1998. Their website tells us that Mr Easson returned his press from Middlesex to Scotland in 2004.

Behind the main sheet in the cellophane envelope holding it is this insert telling us about the job. Their first line is a little misleading, as it describes the original piece, not the version the purchaser is holding. The 75 copies of the original were apparently mostly supplied to The British Printing History Society,  so in order to be selling copies nowadays in Coupar Angus, Quarto (not the publisher of that name of course) would have had to have reprinted the piece. And I bet they did this by photographing the original piece and printing it by offset lithography.

Some of these ornaments are a bit odd: those aggressive policemen near the top are a bit worrying, though the ballroom dancers to their right seem to be quite unconcerned as do the three kids hiding among the flowers between them. The kid in the middle does appear to be toting a gun and this may be what’s upsetting the cops. And what are those teddy bears up to? The rugby players in the lower portion are rather good, as is the cow being milked with the real business tastefully masked by a milk bottle. And you’ve got to love the pig. Hard to imagine circumstances demanding the reuse of many of these: still some printer had paid for them and thought there might be another use for these each of these blocks, even if some may never have been unwrapped again.

The sheet draws the distinction between borders and ornaments. This seems relatively straightforward to my mind: a border would be available as a font for output by your typesetting machine along with the text, whereas an ornament would have to be created separately from artwork sent to the engraver, and integrated into the typeset page in the composing room. But Mr Easson is the printer and knows better than me.

I have discussed the flag blocks at the bottom right hand corner in a previous post.


* The note at the top tells us that some of these type ornaments precede photo-engraving which is what is shown in the second of these videos. Die sinker describes the process of engraving a block by hand, though the photo-engraving video does show a lot of hand correction work.

A font for “these uncertain times” has been developed by The Third Street Attention Agency. You can download it here at their descriptive justification page. (Link thanks to David Crotty at The Scholarly Kitchen.)

The font is a sort of fractured version of Times New Roman, and rejoices in admirable social distancing in its letter spacing. It includes only two punctuation marks, the question and exclamation marks — which somehow seems especially appropriate in these uncertain times when anxious questioning and raucous assertion have become our main modes of public debate! Despite this brave claim they appear to have allowed themselves to borrow a period for this example sentence.

This video, posted last week by David Crotty, takes us through these sympathetic adverts with which we are now being bombarded. The phrase “these uncertain times” has become a bit of a trope in the surge in new, rely-on-us-we-are-concerned-about-you TV ads. And there are a lot of them: we have a friend whose job is legal vetting of broadcast commercials, and she reports they’ve been really busy since they began working from home.

If you don’t see a video here, please click on the title of this post in order to view it in your browser.

See also Times Roman, old and new.

One of the laws of composition (typesetting) back in the olden days was “follow the copy out of the window”. If the author misspells their own name on the title page of the manuscript you have to assume they meant to and set it that way. There was naturally a strong temptation not to follow that diktat — keyboarding apparent nonsense is not something a craftsman would be eager to become involved in. Still I imagine most experienced keyboarders would be able to protect their sanity by an ability to switch off their critical mode while focussing their attention on accuracy.

John Carter, in ABC for book collectors, describes the situation thus:


“The compositor’s watchword, and his defence against the illegibility, inconsequentiality, impenetrability or errors of the author’s text. But often he failed to follow its dictate, the itch to correct or normalise being too great. In extreme cases, however, he could take refuge in the full text of the adage, and ‘Follow copy out of the window’.”

Now of course there were a few situations where author and compositor almost became collaborators. There were comps at Cambridge University Press who, meeting manuscript or typescript from a local professor for the fifth or sixth time, would take it upon themselves to correct the author’s Greek or mathematics — after all anyone can make a slip of the pen or typewriter key, and even the most eminent professor can have a blind spot. Some even got acknowledged in the front matter of the final book.* But the key feature of this process was the requirement that authorial attention should be drawn to the changes. If the author hits the wrong key and asserts that 2 x 2 makes 5, your duty will tilt to the other side of the scale — but mark the change on a proof so that the Professor of Combinatorics can assess the situation. Of course just because you may not approve of this or that author’s views on what to you is revealed truth, you don’t have a God-given right or duty to silently amend the “blasphemy.” (Go far enough back and I suppose your prejudice would have been supported by force of law.)

This sort of thing, cleaning up errors, you may object, is exactly what editors and copyeditors are meant to be doing: and so it is. The same process of marking the correction in the manuscript, or using computer software which color codes corrections to allow approval by the original author, is pretty universally followed nowadays at this earlier stage of production. We are well organized on this front these days, but there were times, and types of manuscript, which were less thoroughly prepared before the compositor got hold of them, maybe journals and periodicals for example. Go back a century or even less, and no publisher employed people called copyeditor; such nuts and bolts were left to the printers.

When book publishers worked in manuscript or typescript rather than computer files the whole stack of paper would be wrapped up in a paper and string parcel and mailed back to the author after the copyeditor had gone through it and made all their alteration marks. Along the way the copyeditor would have discussed any thorny or contentious matters with the author in a phone call, or more likely in those days, an exchange of letters. When checking the copyedited manuscript, the author it was assumed would carefully consider each change, but of course reality breaks through and many an author, learning to trust the copyeditor after the first few pages, would make a short back and sides job of it.

Following a well-prepared manuscript out of the window would present minimal risk of damage. And there’s always the proofing to come.


* A positive reaction was not however universal. See Jeremy Mynott’s comment to the post Manufacturing defects telling of Professor A. E. Housman’s reaction to the proofs of one of his books.

Russell Maret, a New York artist and designer, has had the idea of creating a new Monotype typeface, the first for 40 years. Monotype went out of business in 1992, so unsurprisingly their output of new designs for hot-metal typesetting has been somewhat interrupted. The typeface he is creating is called Hungry Dutch. (Mr Maret worked by adapting a 17th century font cut by Peter de Walpergen for Bishop Fell, and called his version Hungry Dutch: de Walpergen was Dutch, and the font was originally intended for a book called Hungry Bibliophiles.)  Mr Maret’s website provides an introduction to the development of the typeface, one feature of which is that it’s not as mechanically perfect as Stanley Morison’s Monotype updates of classic type designs tended to be. At first sight it might seem a little odd that “imperfect” alignment might be a desirable feature, but the point being made is that the possibility of perfection does not necessitate our striving to produce perfection. Psychological studies of reading are notoriously thin on the ground, and we don’t really know whether perfect alignment is a good thing or not. The fact that handwriting bobs about quite a bit might be argued to favor a slight irregularity.

The site Books on Books provides a gallery of illustrations of Hungry Dutch, one of which is shown below.

The Type Archive is at the center of this initiative. The Type Archive, housed at 100 Hackford Road, Stockwell, SW9, in a building whose floors had once been reinforced to accommodate a pair of elephants imported from India by The Daily Mirror, holds the National Typefounding Collection consisting (largely) of

  1. the typefounding materials of the Sheffield typefounders, Stephenson Blake, dating from 16th century
  2. the hot-metal archive and plant of the Monotype Corporation from 1897 onwards
  3. the woodletter pattern collection and plant of Robert DeLittle in York from 1888, and in Lambeth from 1996.

Given that The Type Archive holds so much Monotype material, and has associated with it so many experts — including Parminder Kumar Rajput, the only man left in the world qualified to operate all the Monotype machines needed to produce a typeface from scratch — it was the ideal place for the creation of a new hot metal Monotype typeface. (One hopes they are making videos of 71-year old Mr Rajput at work.) One preliminary step in the design process — transferring a drawing onto a metal pattern by means of a glass plate, wax and electrolysis — had been completely forgotten. A 3-D printing solution was invented to get round the loss.

Just when, if ever, Mr Maret expects Hungry Dutch to be completed is not clear. To some extent the exercise is a research exercise and as such doesn’t require a complete font ready for production. We always need to record old technologies. Thatching is undergoing a revival as is house construction using hand cut wooden beams. You don’t know what you’ve got till it’s gone.

Much of my information is gleaned from The Economist‘s Christmas special article about Hungry Dutch.

Standard Generalized Markup Language was what we first became familiar with in the late nineties in order to enable us to “repurpose” our texts. SGML is not a document language, but a description of how to specify one. In the old days typescript would be marked up by the designer or copyeditor, in pretty general terms, for example CT next to a chapter title. The typesetter would mark it up in more detail so that the keyboard operator could fly through it without having to stop and figure out what type size and face was really called for here in the designer’s specifications. Thus we were familiar with the need for markup. But what we were familiar with was markup directed at creating a book, laid out in pages, not markup which would enable the text to be output in multiple ways on different platforms, including as a book. Up until then we were book publishers, and we published books.

In the nineties along came the idea that the text of a book (the content) might in fact need to be used in other ways, and the fact that we already had that content on a digital medium made it obvious that money could be saved by using the same digital storage for all and any reuses. The main difficulty here was the difficulty of getting people’s minds changed so they could countenance the idea. Prior to the invention of computer systems the different ways a book might be used amounted to a paperback edition or a hardback, with the occasional opportunity for an extract to be published in some periodical. Magazines would just reset the extract they were doing, and while we academic publishers would use the same typesetting for hardback and paperback, even if a mass market paperback required resetting, the cost of typesetting was fairly trivial when compared to paper, presswork and binding a huge number of copies being printed. Now we also had the opportunity to allow people to access our content online: this required a severe adjustment of focus.

SGML is ancestral kin to XML (Extensible Markup Language) and HTML (HyperText Markup Language) which are now the primary tools used for text markup. The theoretical background to all markup languages is that before it ever appears to the world the text of any work should have been described in such a non-specific way that any application in whatever form you can imagine can be run off on a computer without any intervention beyond the specification of what medium you are targeting. In the example below you can see the HTML codes enclosed in guillemets < >. Here <h1> denotes a first level heading, <p> a new paragraph, <i> italic, and <em> tells you that this is an emphasized word. (The green color is just there for pedagogical purposes. Markup doesn’t show in green in the real world.)

Given that when you use one of these meta-languages to describe your document you have in theory prepared it for any and all applications, it may be seen as perverse not to use the markup to facilitate certain outcomes. In order to make our ebooks fully accessible to print-disabled people, here’s Bill Kasdorf in Publishers Weekly encouraging us to take advantage of the powers provided by our HTML mark-up. This additional small step is pretty straight-forward if you are doing your markup thoroughly — and if you’re not, why bother?

Almost parenthetically I might note that the transition to digital text processing and SGML markup, like all changes, caused a good deal of low-level turmoil. Once people got on board and accepted that text markup “was a good thing” a kind of enthusiasm gripped those bosses with more power than knowledge. Why couldn’t we take all those digital resources which we’d been holding onto for a few years and magically get them SGML-ed? Well, I can’t imagine that at the end of the last century the digital storage system we had was much different from that at any other book publisher. It consisted of a cardboard box or two into which the disks of any book that had had disks were tossed. Rubber banding together the disks from a particular book was a good idea, but rubber bands give up the struggle after a couple of years. What you had therefore was a mess of disks of various sorts, sizes, and formats, some of which were unreadable because the machines they drove no longer existed, some of which had gotten one of their component disks lost, and all of which required time to assess. Publishers will staff their production departments on the basis of the volume of work going through at any time. The amount of work going though was calculated on the basis of the number of books due to be published in the next 12 months — not with regard to sorting out the disks for every book you’d published over the previous five or so years. Eventually, I suspect, all publishers either threw away their old disks (and tapes), or sent them off to an overseas supplier to sort out, but we all spent a considerable amount of time trying to solve the problem of “looking back”. It’s always easier to implement a new system going forward: you just start doing your new books in the new way. Trying to catch up with the old books which were done differently is a nightmare. (This of course is why lots of older books remain unavailable as ebooks.)

Readers of books should ideally be unaware of the thought processes of designers and layout people, so that nobody has to stop and wonder why this or that decision was made in setting the type in the book they’re reading. The mission of design is to facilitate the smooth transmission of the message from author to reader: not to shout out, look what a beautiful job I’ve done. For design and layout people beautiful ought to be synonymous with invisible. But of course because we remain unaware of these thought processes, when we might wish to consider them we find that we remain unaware of them.

To me, the knowledge contained in the head of a book compositor was amazing. A seven-year apprenticeship can’t have been enough to internalize everything. Hart’s Rules for Compositors and Readers* Oxford University Press’ guide for workers in its printing plant (which closed in 1989) contains the following instructions on word-breaks.


Avoid division if at all possible, having regard for the requirements of good typography. [Which basically means don’t set the line with too much letterspacing — i.e. don’t set the line t o o  l o o s e just to avoid a word-break at the end.] Where word-breaks are necessary, however, the following rules apply:

(a) A minimum of two characters may be left behind and a minimum of three characters carried over at a word-break.

(b) Two successive hyphens only are allowed at the ends of lines.

(c) A divided word should not end a right-hand page.

(d) If the right-hand page is a full-page illustration or table, the facing left-hand page should not end with a hyphen.

And that’s it. Following these simple rules will avoid ugly and confusing word-breaking. Too many hyphens and your eye will begin to pick up the wrong line when flicking back and forth; too few characters and misunderstandings threaten. Turning the page is always an opportunity to loose the place, so don’t make the chances higher by breaking the word. (In this context see also Catchword.) Attention should be given to the structure of the word in making the decision to split it: don’t do pr-oductive or produc-tive. These sorts of rules are now incorporated into software, and will be applied without the benefit of human intervention. But as an overriding rule In Hart states “In borderline cases supervisors are to be consulted for a decision whether an exception is to be made.”

These rules are a distillation of 500 years of trial and error. Printers arrived at this sort of consensus by discovering that doing differently resulted in poor outcomes. Each compositor internalized the rules, which would be drummed into them when they were apprentices. Actually, the thinking about all this goes back more than 500 years, as scribes writing out manuscripts developed rules about all sorts of things, including word division.

There are some word-breaks which although they are by the rule should never be undertaken. The one that sticks in my mind is pre-gnant”


* “Readers” refers here not to you or me snuggling up in an armchair to consume an OUP book — it refers to the proofreaders employed by all printers back then. When the type had been set it would go to an internal proof room where it was read against copy and sent back to the composing room for correction before a proof was ever sent out to the customer. I was at one time involved with the books of W. Edwards Deming, who held the view that Cambridge University Press was the best typesetter in the world because they never made a mistake. For an efficiency expert this was a slightly odd view as the reason for the apparent perfection was that the proof he was seeing had always been read and corrected before it was sent to him — not really the world’s most efficient use of labor. As the Press had closed the proof room by the time we were doing his books, we would send the proof to a freelance proofreader first, get corrections made, and then send the “perfect” proof to the author. I’m sure he went to his grave convinced of Cambridge’s infallibility.

Proof readers and compositors, whether they started out that way or become so as a result of years of experience, were often considerable experts in arcana. From time to time eminent professors of Greek or Mathematics in Cambridge would send notes of thanks to the compositor for saving them from errors in the subject areas in which they were meant to be the experts.

Is this the ultimate in (worthwhile) fame? Uno Cup, Inc. has designed a typeface based upon Ms Thunberg’s handwriting.

Greta Grotesk Regular. © Uno

The font, which is based upon a sign Ms Thunberg deployed in front of the Swedish parliament, will no doubt be appearing in lots of upcoming demos. It’s available for free downloading.

Hyperallergic carries the story, and shows a few words set in Greta Grotesk Regular.

Uno is a startup aiming to provide us all with a single reusable cup which we will carry around with us. “Our vision is simple: we should all have one incredible cup of our own that can be easily filled with all the beverages we love. No more paper cups, plastic bottles or straws. We’re partnering with the most forward-thinking workplaces and merchants to enable Uno anywhere you get beverages.” You can reserve your cup in one of three sizes at their site. Seems to me the biggest problem might be persuading those vendors to accept the cups.