From Boiling Lead and Black Art.

Even if you have no particular interest in the history of mathematical typography (I don’t, even though I once wanted to be a mathematician), if you have any interest at all in typography, especially the old-fashioned hot-type kind, you will enjoy Eddie Smith’s From boiling lead and black art: An essay on the history of mathematical typography. Smith writes “To fully appreciate mathematical typography, we have to first appreciate the general history of typography, which is also a history of human civilization,” and he follows through; I’ve never read a compact presentation of that history that was as understandable and enjoyable. It’s worth it just for the image labeled “The complex arrangement of characters and spaces required to compose mathematics with metal type,” which is mind-boggling, and for the anecdotes, like this one:

At Chicago in the 1960’s and 1970’s we had a technical typist who got to the point that he, knowing no mathematics, could and did catch mathematical mistakes just from the look of things. He also considered himself an artist, and it was a real battle to get things the way you and not he wanted them.

And for me it was worth it just for this footnote: “TeX is pronounced ‘tek’ and is an English representation of the Greek letters τεχ, which is an abbreviation of τέχνη (or technē).” All these years I’ve been saying “tex” (and “latex” for LaTeX) like a doofus!


  1. January First-of-May says:

    “TeX is pronounced ‘tek’

    I always thought that TeX was pronounced basically the way it was spelled – that is, [tex] (give or take some allophonic distinctions).

  2. Eli Nelson says:

    (bit of a rant:) To me, the prescribed pronunciation and typographical presentation of “TeX” and “LaTeX” seems a bit irritating, in the sense that it’s not very … ergonomic? I mean, you said “like a doofus” in this post, but I don’t think it’s really the reader’s fault when a neologism has an opaque pronunciation. To me, it seems preferable for the creator of a name or term to avoid, as much as possible, doing things like this that will trip people up.

  3. And LaTeX is pronounced [lɑːtɛk].

    Don’t feel bad, hat. It’s just a distinguished computer nerd (Knuth) making an annoyingly clever-and-a-half inside joke, after the manner of his people in their time; cf. Vaxen as the plural of VAX, an old computer brand, not to mention UNIXen. I say, if you’re not privy to their inside jokes, you have lived a pure and forthright life.

  4. Trond Engen says:

    When my wife was in her late teens she volunteered in a Catholic youth magazine. This was before her church owned a computer with a text editor, so among her tasks was typing texts in Vietnamese and Tagalog. She didn’t learn the languages, but she became able to spot errors and confident enough to correct them.

  5. January First-of-May says:

    To me (and probably to most other Russian programmers) TeX is pronounced тех, and LaTeX is pronounced латех 🙂

    The former, discounting the palatalization, happens to be extremely close to straight IPA [tex]; sadly, as far as I can tell, it doesn’t quite work as cleanly with the latter.

  6. I also really dislike math (and text) typeset by standard TeX: the type is too skinny and low-contrast, and the proportions are weird. In the linked article, TeX is exampl 6. Note the proportions of the Σ, the thin stem of the n, and the 2 being the same size as the n.

    One 1970s text, Joe Roberts’ Elementary Number Theory), is typeset from a nicely calligraphied manuscript.

  7. There’s an anecdote told in (I think) Willis, Latin Textual Criticism.

    Oxford had a typesetter who did sterling work on setting Sanskrit (in devanagari) for many years, until he suddenly started to produce nonsense. On investigation the dons found that he had gotten interested in the gobbledigook he was setting, started to learn Sanskrit, and was useless now that he understood part (but not all) of what he was typesetting.

  8. I believe, there is a tiny bit of TeXnical error in the linked article. Setting off math between single dollar signs makes it so called inline, with fractions somewhat compressed vertically (and some other features). The displayed result is in, well, display style, used on a line specifically dedicated to math. I like (la)tex and used to use it a lot, but reading a paper after paper with the same font and the same arrangements and proportions of various types of titles wears even my relatively robust visual dullness.

  9. In fact, I think Knuth prescribed the pronunciation “tech” (with the “ch” as in “loch”) but I don’t think anyone is so affected as to pronounce it that way. I do hear both “luh-TEK” and “LAY-tek” for LaTeX.

    I also find it funny that the progression of math-typesetting-quality was not monotone in the 20th century. Papers (and some books) from the ’50s, ’60s, and ’70s are often very painful to read. But books from before then usually still look quite nice.

    And monotonous though it may be to see everything in LaTeX-typeset style with Computer Modern font… well, it could be way worse. Even though it’s gotten a *lot* better lately, papers written in Microsoft Word still look horrible. One popular blogger said the non-use of TeX is the #1 sign of mathematical crackpottery:

  10. There are people who pronounce “LaTeX” luh-TEK instead of LAY-tek (or, more correctly [yes, going there], LAY-tex)? Why do they think it’s called “LaTeX” in the first place?

  11. From their official site: “LaTeX, which is pronounced «Lah-tech» or «Lay-tech» (to rhyme with «blech» or «Bertolt Brecht»)…”

  12. So “lah-techt” (sic!) is permissible but “latex” is not? It’s enough to turn a peever descriptivist.

  13. Bah. I like Word’s Cambria Math. And yes, there’re still stuff it can’t do, but it’s good enough for teaching.

  14. Don’t feel bad, hat.

    Oh, I don’t really feel bad, it was just for effect. But I’m delighted to hear so much discussion of the varying pronunciations!

  15. Lars (the original one) says:

    Now this is a point where tastes may differ, but a combination of modern HTML and Unicode will often allow inline/textstyle math to be rendered more pleasingly as opposed to LaTeX bitmap images that are not sensitive to font selection and do not benefit from antialiasing. LaTeX still wins for most display math.

  16. I once encountered a young whippersnapper computer salesman who when I mentioned that I would be using TeX (1) warned me that it’s not wysiwyg (I wasn’t asking his advice on whether I should use TeX, I was just answering his question about what programs I would be running) — I think this might have been the moment when I learned the stupid word “wysiwyg”, and it was annoying to have to learn it from this whippersnapper — and also (2) pronounced “TeX” in the most affected way possible, drawing out the final “ch”-as-in-loch to great length.

  17. Well, I’ll do a rant on one aspect of the article (which is otherwise good): the author has missed an earlier system for math in text, namely troff. The very earliest development of Unix at Bell Labs was unfunded; when the developers needed more support they based it on the usefulness of the system to the legal department in producing documents. By 1975 Unix had a set of programs that would produce high-grade typeset math by driving a C/A/T phototypesetter. And how was inline math indicated? With $ signs. TeX was quite an achievement, but it was not the first of its kind.

    In 1978 my (then) university acquired a minicomputer running Unix, and a phototypesetter; the system could be run remotely, and in 1979 I wrote my thesis using it. Troff eventually was replaced by groff, which writes PostScript and is included in any Mac; so I can take the files from 1979 and process them on my laptop without any alterations needed. I find this quite impressive (and useful, actually).

    My thesis advisor was another early adopter. At the time the lead journal in our field was printed from camera-ready documents that you typed (you could have them typeset for higher page charges). In 1980 he submitted a phototypeset paper. There were instructions on what kinds of typing was acceptable, and he had quite a battle with the copyeditors, who wanted to know what pitch he had used on his Selectric, and didn’t (then) know what to make of his response: “10 point on 12”.

  18. You should write Smith and mention it; he might add it to the essay.

  19. I find this quite impressive

    <jargon>I once built and ran a PDP-11 Unix 7th Edition userland emulator on my Linux box, and typed “cal 9 1752” into its shell. The output was exactly the same, modulo the case of the weekday abbreviations, as what the native cal command output on that same Linux box.</jargon>

  20. Another cutesy computer nerd term: the execrable virii (as a plural of virus).

    Another math typesetting system, after a fashion: the still-not-standard standard, MathML.

  21. Lars (the original one) says:

    1-up on JC: My first UNIX account was on a physical PDP11 (/70 I think) running 7th edition. And I can safely say that I ran lots of roff since the manual system was written in it. Still is on Mac and Linux, so if “man 2 creat” works on your command line you just used roff and you can look in /usr/share/man/man2/creat.2 to see the source.

    virii — either I’ve conditioned myself to ignore it or it’s dying out in favour of viruses. In Danish we used to use the vaguely defensible vira (still de rigeur in medical prose) but virusser is winning here as well.

  22. But Classical Greek chi was [kh] like Scots loch not [k] like kappa, and outside of the IPA [x] is pronounced [ks] in all the languages I know. So unless Knuth learned a different Greek than I did, it does not really work … was he thinking of Cyrillic kha?

  23. David Marjanović says:

    Huh. I had taken for granted that TeX had something to do with text; and now I learn that Knuth himself wrote The TeXbook.

    Actually, my first exposure to it was by learning that .tex files exist. Knowing .txt, I took .tex as some alternative approach to reducing text to the three letters allowed in a DOS filename extension.

    I still don’t use it. Outside of math, physics and computer science, journals actually want .docx files; or at least they say so because they know what to do with those?

    Virus – Viren in German (both biochemical and digital), just like Forum – Foren and Atlas – Atlanten. German is essentially English with the oxen > boxen > vaxen &bt; Unixen joke taken seriously and implemented with German thoroughness (mit deutscher Gründlichkeit).

  24. David Marjanović says:

    Modern Greek χ is very much pronounced [x], except of course when it’s pronounced [xʲ] before a front vowel (or silent ι). Likewise, φ and θ are [f] and [θ], and have been since the second century or so.

    (That’s when the first FILIPPVS appears on a Roman graffito. In Pompeii, there’s only a PILIPPHVS where someone couldn’t remember where to put the aspiration.)

    Scots ch isn’t a Carinthian or Iroquoian [kh] cluster, it’s [x]. Scottish Gaelic ch, interestingly, isn’t [x]; it’s [χ], except when it’s [xʲ] next to e or i.

  25. My PDP-11 days were on RSX/11-M-Plus. I first ran *ix in the form of Eunice, a layer on top of VAX/VMS which was to Unix as Cygwin today is to Linux (and I run Cygwin today on every Windows system I get my hands on).

  26. David: Yes, but Greek chi χ and Latin x are not the same letter. As far back as we have enough texts in the script to reconstruct pronunciation, Latin x is closer in pronunciation to Classical Greek xi ξ than Classical Greek chi χ. So transliterating Greek chi as Latin x is just wrong, in any language which I know.

    I used the wrong kind of brackets around my second example due to the trouble of typing angled brackets in HTML … read them as “orthography brackets” not “IPA brackets.”

  27. In the Western Greek alphabet, χ was used to represent /ks/ rather than /kʰ/ > /x/, which was written ψ. As Wikipedia points out, the phrase “Ἔδοξεν τῇ Βουλῇ καὶ τῷ Δήμῳ” (“The Council and the Citizens have decided”) is typically spelled “Εδοχσεν τει Βολει και τοι Δεμοι” in inscriptions of the Athenian democracy prior to 403 BC, when the Ionian alphabet was adopted in Athens.

  28. David Marjanović says:

    Ah, I got confused by the brackets. I just use italics for orthography, though spelling out the HTML entities for “less than” (&lt;) and “greater than” (&gt;) work. // is for phonemic, [] for phonetic transcription (IPA or otherwise, precise or not).

  29. Squiffy-Marie van 't Blad, Dutchman-at-large says:

    Troff’s math was actually eqn, which compiled down – this is the Unix tradition – to rawer troff markup.

    State of the art math markup on the web is done with MathJax instead of bitmaps (except by persons wishing to make me cry, please stop it) and the Varied Reader will readily infer what syntax it is modeled on.

    MathML has considerable advantages in interesting dimensions, but being actually used on the Web isn’t really one of them. (Even by the standards of SGML derivatives it is pretty verbose. For reasons, but that doesn’t much console the fingers.)

  30. John: Thanks for the explanation of how the Latins gave a different meaning to that sign than the Ionians did! My Latin does not go much further back than Sallust.

    This is just a sore point for me because some of the ancient languages I read use <x> = [ks] and some use <x> = [x].

  31. Lars (the original one) says:

    @Squiffy, if you make MediaWiki put back MathJax, I’ll start using <math> again.

  32. I had a professor, ca. 2000, who would e-mail out his course notes in eqn. He was apparently unaware that nobody used it any more.

  33. Marja Erwin says:

    “This is just a sore point for me because some of the ancient languages I read use = [ks] and some use = [x].”

    :𐌾𐌰𐌷·𐍃𐌿𐌼𐍉𐍃·𐍂𐌰𐌶𐌳𐍉𐍃·𐌽𐌹𐌿𐍄𐌰𐌽𐌳·𐌱𐌰𐌽𐍃: And some languages use both!

    But as noted above, Greek had *already* been using both, before this, just not in the same script.

    But do any languages use = [k] as implied for LaTeX?

  34. Xmas?

  35. Oh, so that’s how [xməs] is spelled!

  36. How else would it be spelled?

    But I say [tɛx] and [ˈlɑtɛx] and [ˈtexnəkəl], and consider [k] a concession to anglophone speech habits, as in loch/lough.

  37. consider [k] a concession to anglophone speech habits

    Yes, that’s got to be right.

  38. ə de vivre says:

    There are even packages that make [lɑtɛx] useful for [ɑɹxətɛkts], or for making musical scores to sing in your [xɔɹəs].

    Incidentally, I first encountered LaTeX in linguistics for doing syntax trees and OT tableau(s/x) (it has a package that made aligning multi-line glosses really easy too), and only ever encountered it pronounced [lɑtɛk].

  39. I noticed in GLS’s Clojure conj talk that he’s in the [lə’tɛk] camp.

  40. In “Counting Potsherds”, Harry Turtledove tells the story of a Persian court eunuch who comes to the land of the Western Yauna about four centuries after Khsrish the Conqueror defeated them and destroyed Athens. He is on an archaeological mission to discover more about Demos, the last king of Athens (who was led into captivity by Khsrish) and his queen Boule. What he finds with the help of a local Yauna who can still read Greek about ostrakhismos and potsherd-counting in general fills both of them with astonishment and horror, but due to a regime change back in Babylon (the Persian capital), his report never makes it back there to threaten the peace of Empire. (The POD is that the silver mines of Laureion aren’t discovered until after the Conquest, and so Athens doesn’t have a serious fleet in being, and takes the “wooden walls” prophecy literally. We are told as a throwaway line that Aramaic functions as a lingua franca from India to the border of the Gallic lands.)

  41. Fun! But is ostrakhismos (for ostrakismos) your typo or Turtledove’s error?

  42. David Marjanović says:

    Likewise, I’d expect Khshrish with one more h. (Interesting, too, how many vowels that guy lost in just four centuries.)

  43. Ostrakhismos is my typo; Khsrish is what Turtledove wrote. The story itself..

Speak Your Mind