According to the de facto standard book for typography, The Elements of Typographic Style, when we write abbreviations with 2 or more letters like SEM (Scanning Electron Microscope), small cap should be used and in LaTeX this would be \textsc{sem}.

Another case are old style numerals. When we write Arabic numerals in the passage, according to the book, old style numerals should be used and in LaTeX this would be \oldstylenums{1234567890}. I know mathematical equations or numerals in a table should not use old style numerals as they are considered standalone and do not disturb the flow of passage.

For me, I have never seen any publications with small cap abbreviations and old style numerals, and this seems to go against the typographic style suggested. The author of The Elements of Typographic Style said actually there is no rigid rule for typography when we aware we are breaking it. So my question is why are small cap abbreviations and old style numerals not used in technical writing when they appear on the passage? Why are these rules are broken in technical writing?

EDIT: The papers I usually read are from natural science like physics, chemistry and biology without computational nature. It seems to me the use of small cap and old style numerals is very rare in these fields.

  • 5
    "never" ? I use small caps all the time in my papers, for precisely the use case you mention. Maybe you're thinking of a particular discipline ? – Suresh Jun 3 '14 at 10:11
  • 1
    May I ask which journals you are talking about? – bingung Jun 3 '14 at 10:27
  • I publish in database/data mining/theoretical computer science venues, and use small caps especially in the first two. – Suresh Jun 3 '14 at 11:26
  • 3
    If you're going to quibble about typography -- when's the last time you saw a ligature other than &? When's the last time you saw REAL kerning? Fact is, fine typography just isn't a high priority in publications whose goal is to communicate factual content rather than to be exemplars of the graphic arts. – keshlam Jun 3 '14 at 23:54
  • 6
    @keshlam: It will be very difficult to find a paper without the ligature w. Nitpicking aside, all journals using LaTeX and most other ones I know use ligatures. Whether you consider their kerning “REAL”, is something only you can tell. – Wrzlprmft Jun 4 '14 at 0:02

The answer lies in a combination of technology and efficiency. In the old days when all was typeset "by hand" and not electronically, switching between Roman numerals and old style ones was not difficult. When electronic publishing became available, typography as a whole was ruined because everyone thought they now could make professional looking typography on their own personal computer and laserwriter. For a long time the digital typefaces did not include old style numbers because the character sets were too small to include those (E.g. PostScript Type 1) but now with the unicode standard, much more is included in a single type face. Thus, for a long time one had to switch from one type face to another to include these effects. In LaTeX, this was a little easier as you point out than in, say, Word.

The use of old style numerals was thus partly lost through the "digital revolution" but I also think it was lost because they simply went out of fashion. As with everything else typography and ideas within typography changes over time. I am not sure that the loss is entirely due to the introduction of digital tools but this introduction was a sort of final "stab in the back" from which it is hard to recover. In fact we are still seeing the effects of the PC in many publications since everyone is now the typographer without knowing much about the trade.

So to cap off, in the scientific publishing world, where speed is a key ingredient, attention to these details have simply not been considered as worth the time. One therefore must look at publications with more aesthetic goals to find fine typography.

  • 2
    Good answer, but I suspect that there may be more. For instance, I think that en- and em-dashes have encountered the exact same problems in the history of computer typesetting, but they are now in wider use in both Word and LaTeX. – Federico Poloni Jun 3 '14 at 12:06
  • Yes, you are quite right and the list can be made much longer since, for example, spacing is a key issue as well. The problem, in my opinion, is that typography, an art and a profession, cannot be replaced by mechanics alone. Deeper understanding is always necessary to get the most out of the existing technology. – Peter Jansson Jun 3 '14 at 12:26
  • I suspect this has as much to do with (late 19th century) typewriters and thence (mid-20th century) character encodings as with modern (late 20th century) typesetting. – JeffE Jun 3 '14 at 22:07
  • 3
    Old-style numbers being included in modern typefaces is not thanks to Unicode but to intelligent font standards such as OpenType, AAT and Graphite. Old-style numbers are not (specifically) encoded in Unicode and doing so would arguably be a very bad idea. – Wrzlprmft Jun 3 '14 at 23:38
  • 2
    As my edit seems to have got rejected: You seem to have mixed up Roman numerals with lining numerals in the first paragraph. – Wrzlprmft Jun 4 '14 at 11:49

There are actually good reasons to break the general typographical rules and use uppercase numerals in technical writing.

In usual texts, uppercase numerals are considered ugly or negatively affecting the readability as they form one block without ascenders or decenders and as they stand out from the text and are more difficult to read (for the same reasons that all-caps are more difficult to read). However, the properties of uppercase numbers are often desirable in technical writing:

  • You may actually want numbers to stand out.

  • In a formula, you usually want a number consisting of more than one numeral to be perceived as one element at first. For example:

    x = 1234 + 5678

    Here the first thing you want the reader to see is that x is the sum of two numbers and not the exact values of these numbers (which would be more emphasised with lowercase numerals). Also, from a readability’s point of view, most formulas are a chaotic mess – using lowercase numerals would only add to this.

  • There are several mathematical notations that do not mix well with lowercase numerals, such as indicating repeating digits in a decimal fraction with a bar above the number.

  • You usually would not want to use lowercase numerals for super- and subscripts, as it makes it more difficult to recognise whether something is a super- or supscript or not. (Note that this is not the same as for lowercase letters, as they are easier to distinguish from their uppercase counterparts and exceptions, such as the letter o, are usually not used as variables for that very reason.)

  • Using lowercase numerals for axis tics would make them optically less regular and thus be more emphasised, which is usually not what you want.

  • In tables, uppercase numerals help outlining the rows, while lowercase numerals obfuscate the structure. (On the other hand, lowercase numerals may make some aspects of the data easier to recognise.)

While you could use uppercase numerals for tables and figure legends specifically, there is a gray zone between

  • formulas in which lowercase numerals do not work well
  • formulas in which lowercase numerals are no problem
  • just numbers.

Therefore using lowercase numbers not at all or only in a few special cases like page numbers or affiliations is arguably the only way to achieve consistency.

A similar thing applies to super- and subscripts: While in normal writing no consistency issues arise, as a super- and subscripted numbers is always super- or subscripted, the semantically same number often appears in normal size as well as super- or subcripted in technical writing. Therefore using uppercase numerals only has a larger impact on consistency than in normal writing.

  • 2
    It's consistency really - in theory you could use old style numerals for page numbers, chapter numbers etc., but then they would clash with numbers in inline maths, or be inconsistent with e.g. equation numbers. There's no clean place to draw the line, and it would have a similar awkwardness to using similar but not quite the same fonts. – Chris H Jun 3 '14 at 16:10
  • 3
    Actually, there is a clean place to draw the line. If a number is used as a convenient label (pages, chapters, equations, figures), use lower case. If it's used as an actual numerical value, use upper case. Or in TeX terms: text mode=lower case, math mode=upper case (in parallel with text mode = italic, math mode = roman for Latin letters). – JeffE Jun 3 '14 at 22:12
  • @JeffE: True, though this usually makes uppercase numbers the majority. I added a remark regarding this. – Wrzlprmft Jun 3 '14 at 23:11
  • I am just finishing 250-pages scientific-technical book. I wrote it in Latex and made sure that every proper physical/mathematical number is in math mode, leaving only years and label numbers out. So, what would you advise, mode=osf and math mode=lining, or all numbers=lining? – Pygmalion Feb 26 '17 at 14:18
  • @Pygmalion: If you really used this consistently (and do not perform any calculations involving years), I would recommend lowercase numbers for non-math numbers. (In fact, I did the same thing for my thesis.) – Wrzlprmft Feb 26 '17 at 14:26

Without any evidence to back it up, my guess is that people don't know it, so they don't use it. As a consequence, it becomes a de facto standard of the field not to use them.

One would expect editors to at least be aware of this rules, but if they are not experts in typesetting, or don't personally like some of the rules, they may not be interested in enforcing them.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.