En dash = width of n?

Nick Shinn

“What I'm referring to aren't just matters of some ancient world of print, but are as much a part of literary and business communications as ever before.”

I’m afraid you’re mistaken. Just look at the curly quote fiasco and the smart quote screw-up of abbreviated numbers and rock ‘n’ roll etc.

What changed things was removing the type house from the workflow, putting typography in the hands of graphic designers and art directors with no grammatical inclination.

Scott-Martin Kosofsky

Martin, I'm afraid that you are seeing this as a pure design issue when, in fact, it is about clarity in literature and information. Type design is not an end in itself and it's sad to think that it is now being taught as an independent discipline without much relationship to the purposes that it serves. If you start with an understanding of what the em dash and en dash do—that is, what function they have in the languages and literatures in which they are used—you might come to a different opinion than the one you now have, which is based on your own sense of visual harmony. (I wish to be clear that I am a talking here about text type, not type used in advertising or for other kinds of display, in which the grammatical aspects of typography may or may not apply.)

Nick, I don't disagree with you for a minute. The reality has fallen far the ideal amongst many users of type, but I hope that doesn't mean that you are no longer including curly quotes and apostrophes in your fonts just because some ignoramuses don't use them.

I don't teach, but I know many people who do, quite a few of them at major institutions. What I learned from them is that instruction in the editorial aspects of design is not part of the curriculum, or at least not a mandatory part. I don't mention this to sound superior or snobbish (like Bringhurst), but out of real concern that people are now designing typefaces without an understanding of what certain important parts of the character set actually do, and how what they do may differ in the many languages that use the Latin alphabet.

Martin Silvertant

I don't think I am. It IS about clarity in literature and information, but there needs to be a balance between functionality and visual harmony.

If you start with an understanding of what the em dash and en dash do—that is, what function they have in the languages and literatures in which they are used—you might come to a different opinion than the one you now have

You're implying I'm missing something. What am I missing? In what way am I failing to acknowledge their function in the languages and literature they're used in?

but out of real concern that people are now designing typefaces without an understanding of what certain important parts of the character set actually do

I have to agree there. I'm currently studying graphic design and there is no editorial design practice to speak of. In fact, we barely do and learn anything with typography at all. I've been researching typography for years and that's the only reason I'm quite adequate. I'm likely the only one from my class who understands what dashes are and knows not to use hyphens. In my first year I often knew better than my teacher and even had to correct a guest teacher on his lecture three times. That shouldn't be.

The one typeface my fellow students use almost by default is DIN. Now this has nothing to do with editorial design, but what's striking is that graphic designers aren't actually thinking about what typeface might be best for their project; they just select one of their favorites which they've seen other academy students use. I submit that has nothing to do with graphic design. The whole area of typography deserves a lot more focus, and yet the opposite is going on in my country. At every company I worked at so far I greatly improved their use of typography and even replaced their logos. I think that's odd. Why don't they have this knowledge?

Ben Blom

For those people who think that the relationship of names and the things to which they are applied don’t matter

Sometimes, people overestimate the meaning (implication) of the names of things, in relationship to what those names are applied to. An example. Some people tend to believe that if a font contains a “double-storey a”, it should also contain a “double-storey g”; or if a font contains a “single-storey g”, it should also contain a “single-storey a”. Another example. If something is called “italic” (no uppercase letter), some people believe that it should be related (in some way, other than just this name) to Italy.

Scott-Martin Kosofsky

Martin, what I thought you might be missing—and perhaps you're not—is that maintaining easily distinguished proportions of the dashes and hyphen are essential to their correct use. Yes, there's some latitude, but one must be cautious. (Note the correct use of em dashes above.)

I realize that some of these things may be cultural. As I mentioned earlier, U.S. literary and scholarly publishing tend to be quite strict about such things, whereas elsewhere, other traditions may have emerged or fallen away and other standards may be in force. In American literature, for example, there were uses of the em dash which haven't been seen for over a century, such as an em dash followed by a comma then a word space. (This was a favorite of Henry James, who was American by parentage though not by residence.)

Type designers would do well to consult with expert users. Lately I've seen a number of text fonts with small caps at x-height, the same as lowercase. Who can make use of them and for what purpose? It's a design decision that seems detached from typographic reality.

In another thread, I suggested that young type designers looking for work might do well to spend some time as typesetters for as large a variety of purposes as they can. It would give them a much better idea of the kinds of types that might be needed in the marketplace. Perhaps it doesn't sound as glamorous as DESIGN, but I would remind them that some very successful people in type design started there. One name that comes to mind is Erik Spiekermann.

Spending more time as serious readers would help, too. Read old books and new ones, even on subjects you don't (or think you don't) care about. Read like a professional.

Rainer Erich Scheichelbauer

Wherever possible, I try to make the emdash twice the width of the tabular figures. And whenever that looks wrong, I make an /emdash.tf with that width, and include it in the tnum feature. Because in price lists with tabular figures, I want the em-dash to be useable as a substitute for two zeros.

Luckily, in German, it is not called en- or em-dash, so I have an excuse to make them the size that looks right to the eye. :-) Seriously, I take the names as a reference to the history, and thus with a grain of salt. Words can change their meaning or parts of their meaning, and what once was true (or made sense) in lead type can be a different thing in digital type.

Martin Silvertant

Scott-Martin (or do you prefer Scott?), thanks a lot for the information. I thought I knew exactly how the em-dash is used but I have to admit I had no idea in America there is such broad use of the glyph which demands different requirements. In fact, when it comes to the use of the em-dash I find Americans tend to abuse the hyphen more than anyone else, but then I also have to admit they're quicker to want to use the em-dash, but I don't find it acceptable to replace an em-dash with a double hyphen either way.

From now on I think I will include a stylistic set in my typefaces so you can choose whether you want traditional or optical lengths when it comes to dashes, because I have very strong opinions on their length in everyday use which deviate from historical convention. I think a typeface like Lexicon shows that at least in the Netherlands there seems to be a preference to the rhythm and harmony of the text whereas I suspect particularly Americans will value functionality more in this case due to their broad use of the glyph with specific requirements regarding its length.

Since this discussion I've been paying extra attention to the use of the em-dash in books and I have yet to find the historical em-dash. This might be due to ignorance or simply preference to optical dashes. I'm reading "Reading Letters" now, which uses shorter em-dashes but I also noticed initially they use spaces around the dash and halfway through the book start using the em-dash without spaces. The inconsistency can't be justified.

Lately I've seen a number of text fonts with small caps at x-height, the same as lowercase. Who can make use of them and for what purpose?

I can, though its application is very narrow. It's a feature in Vesper and Mota Italic calls them "petite caps" (motaitalic.com/typefoundry/fonts/vesper/opentype-features). They're a contemporary addition and shouldn't really be mixed up with small-caps as they have a different application. Rather than using it for emphasis or contrast, these petite caps are for more creative uses, similar to what you might use a monocase typeface. It's a stylistic choice rather than a functional one.

In another thread, I suggested that young type designers looking for work might do well to spend some time as typesetters for as large a variety of purposes as they can. It would give them a much better idea of the kinds of types that might be needed in the marketplace.

I'm all for that, but at the same time in this specific case you seem to be urging everyone to conform to the US standard. Yes, it's also the historical standard but I think in Europe we have different considerations regarding the length of the dashes so I feel it's a conscious decision to deviate from the norm rather than ignorance. If this was a practice due to ignorance, it could easily be solved with education. As it seems to be a conscious choice, I think we as type designers have to have an increased awareness of our target audience; do we want to cater to the US or the EU or do we keep both of them happy? I'm now convinced that I should be giving the US what they want as well, so a stylistic set seems to be a good choice. I'm curious to know if there are objections to that idea.

but I would remind them that some very successful people in type design started there. One name that comes to mind is Erik Spiekermann.

Yes, but so did Bram de Does and he designed his typefaces specifically to accommodate for what he learned about the requirements in type setting and yet his dashes deviate from historical convention as well.

Nick Shinn

I had a lot of experience in commercial design; perhaps because of that my inclination as a type designer is to question rules and conventions, and trust my eye as to what looks right.

Karen Cheng

Thomas Phinney said:

No, the author is wrong.

For a long time, the en dash was a sort: it came from a standard set of extra bits that were shared across typefaces. It was half an em in width, a.k.a. half the current point size.

When it started getting built into individual fonts, that tradition mostly continued. But not everybody follows it. In particular, some type designers note that if the typeface is particularly condensed (or expanded) it may not make sense to keep the en and em dashes at their traditional widths.

Hi Thomas,
I know this is an old thread, but was wondering if you could explain this a bit further. That is, in Glyphs, the em is set to 1000 units. Therefore, is the en dash just 500 units? I'm sorry if this is a dumb question. I actually also thought the en dash should be the width of the lower case n (including sidebearings). 

Thomas Phinney

Short answer: Yes.

Slightly Longer: 500 units out of 1000 is a good default for the en dash.

More:

I was originally responding to a question from James Puckett, who was citing (usage in? specification in writing?) a British book:

James Puckett said:

Is there a British tradition of making the en dash the width of n? I am reviewing a British book and I don’t want to assume the author is wrong if this is just a British thing.

Anyhow, the usual standard in metal type was that the em dash was the same width as an em quad (the same as the point size), which is 1000 units in a 1000-unit em space. The en dash was the same as the en quad (half the point size), or 500 units in a 1000-unit em space. John Hudson expresses some strong opinions in support of this standard. While I agree that the standards can change, I do not believe this one has, particularly.

 That said, particularly condensed or extended fonts just look wacky with standard-width em and en dashes.

 I am currently working on a variable font with a major width axis. I will make sure the regular-width masters use a standard-width em and en dash (em-size and half that). But I am going to let those two glyphs vary with the font width axis. Otherwise they would be unusable in the most condensed and extended widths.

Karen Cheng

Thanks Thomas!

John Savard

John Hudson said:

There’s nothing wrong with that

Except that what you've made are not em and en dashes. I really believe that fonts should contain proportional dashes of, at least, narrow, medium and wide widths. But calling any of these em and en dashes when they are no actually em- or en-width isn't the solution.

It should be noted that American Type Founders made 3-to-em spaces, 4-to-em spaces, 5-to-em spaces, and so on, that, for some sizes of type, didn't actually go exactly that many to the square of the body size.

Instead, they rounded them so that the width of every space, like the body width of every character, was an exact multiple of 1/4 point. This, among other things, means that you didn't need to insert one space made of rubber into every line of type to obtain justification.

So there is precedent for "em-" things and "en-" things not being exactly as wide as they're supposed to be officially.

ClintGoss

Is there a correspondingly wide range of opinions ― and respectful disagreement ― about the width of space characters? I'm thinking U+2000 through U+200A.

The Unicode Standard as well as the Microsoft OpenType Font Development document (https://docs.microsoft.com/en-us/typography/develop/character-design-standards/whitespace) do seem to be somewhat more specific about the width of spaces compared with dashes.

I have been:

• setting dashes based on best look, with alternates glyphs that use the UPM-based widths, and
  
• setting the widths of these space characters based on the UPM-based recommendations.

Is there a similar argument for altering the width of, for example, FOUR PER EM SPACE for typographic reasons?

Thomas Phinney

John Savard said:

It should be noted that American Type Founders made 3-to-em spaces, 4-to-em spaces, 5-to-em spaces, and so on, that, for some sizes of type, didn't actually go exactly that many to the square of the body size.

Instead, they rounded them so that the width of every space, like the body width of every character, was an exact multiple of 1/4 point. This, among other things, means that you didn't need to insert one space made of rubber into every line of type to obtain justification.

So there is precedent for "em-" things and "en-" things not being exactly as wide as they're supposed to be officially.

That's just rounding to the grid. We aren’t going to necessarily set our UPM to some common multiple of the space fractionals to make them perfect, either. But rounding to the nearest quarter-point, or rounding 1/3 of an em to 333/1000 (or 683/2048) is not at all the same as saying it can be something else entirely for purely aesthetic reasons.

ClintGoss

Thanks Thomas!

So you feel it is generally safe to set widths of dashes "to taste" and keep widths of spaces "as per the spec" (i.e. bound to the "real" em)?

i.e It is OK that EM DASH and EM SPACE have different widths? 

John Savard

Thomas Phinney said:

John Savard said:

It should be noted that American Type Founders made 3-to-em spaces, 4-to-em spaces, 5-to-em spaces, and so on, that, for some sizes of type, didn't actually go exactly that many to the square of the body size.

Instead, they rounded them so that the width of every space, like the body width of every character, was an exact multiple of 1/4 point. This, among other things, means that you didn't need to insert one space made of rubber into every line of type to obtain justification.

So there is precedent for "em-" things and "en-" things not being exactly as wide as they're supposed to be officially.

That's just rounding to the grid. We aren’t going to necessarily set our UPM to some common multiple of the space fractionals to make them perfect, either. But rounding to the nearest quarter-point, or rounding 1/3 of an em to 333/1000 (or 683/2048) is not at all the same as saying it can be something else entirely for purely aesthetic reasons.

Yes, I totally agree.

With one caveat.

I agree it would be silly to change the width of an en space, or a 3-to-em space, and so on, for any reason other than "rounding to the grid".

However, when it comes to an em-dash or an en-dash, these are printable characters serving a typographic function, and so adjusting their widths for aesthetic reasons seems to me no more objectionable than adjusting the width of the hyphen to be whatever suits the typeface.

However, then

ClintGoss said:

So you feel it is generally safe to set widths of dashes "to taste" and keep widths of spaces "as per the spec" (i.e. bound to the "real" em)?

i.e It is OK that EM DASH and EM SPACE have different widths? 

raises another question, since having an en-space and an en-dash that have different widths does have the potential to be confusing. While I would be OK with it, your answer, as that of a knowledgeable typographic professional, is rather more valuable than my amateur guessing.

Basically, the situation we have is that there are the following three plausible statements:

• The en-space and em-space are spaces which are equal in width to the height of the point size and half that point size, respectively, and they should always have that width;
• The en-dash and em-dash are characters that occur within text, and thus, like the hyphen, their width should be adjusted to suit the appearance of the typeface;
• As en- and em- are prefixes that denote width, the en-space and the en-dash should have the same width, as should the em-space and em-dash

which, unfortunately, can't be true simultaneously.

But I see that we have discussion previously in this thread which addresses that point:

John Hudson said:

There’s nothing wrong with that

Except that what you've made are not em and en dashes. I really believe that fonts should contain proportional dashes of, at least, narrow, medium and wide widths. But calling any of these em and en dashes when they are no actually em- or en-width isn't the solution.

to which the reply

Nick Shinn said:

John, you are proposing to shut the stable door after the horse has bolted. What is to become of all the fonts with “proportional” em and en dashes? Pablo: http://typophile.com/node/27742

was made. So apparently the historical convention is indeed to ignore the third apparent desideratum, and allow en-dashes that are not the same width as en-spaces. (Of course, instead of historical convention, it could be just a bunch of modern-day typefaces in which mistakes were made, but I suspect that Nick Shinn isn't going to count random junk fonts as being significant examples.)

On the  other hand, I thought that I would check out the ultimate authority on this matter. The best and most perfect example of a typeface (or font for a typeface) designed by people who knew better.

Examining layout 1216 with unit shift for Times 327, I find that there is an 18-unit long dash, a 9-unit short dash, and a 7-unit hyphen. If the set width of Times Roman were always exactly the same as the point size, which it is not, those would indeed be true em-dashes and en-dashes.

So that's an argument in favor of keeping the third point, and tossing out the second point, subject to a somewhat liberal interpretation of "rounding to the grid", although a single typeface, however exemplary, is not a compelling argument that everyone else must do things the same way.

Thomas Phinney

John Savard said:

Examining layout 1216 with unit shift for Times 327, I find that there is an 18-unit long dash, a 9-unit short dash, and a 7-unit hyphen. If the set width of Times Roman were always exactly the same as the point size, which it is not, those would indeed be true em-dashes and en-dashes.

A font does not have to be monospaced for these to be “true em-dashes and en-dashes.” I am not sure where you are getting all that.

John Savard

Thomas Phinney said:

John Savard said:

Examining layout 1216 with unit shift for Times 327, I find that there is an 18-unit long dash, a 9-unit short dash, and a 7-unit hyphen. If the set width of Times Roman were always exactly the same as the point size, which it is not, those would indeed be true em-dashes and en-dashes.

A font does not have to be monospaced for these to be “true em-dashes and en-dashes.” I am not sure where you are getting all that.

Monospaced? Where did I say anything claiming that Times Roman should be monospaced? (I suspect the source of the confusion is that you weren't familiar with the meaning of the technical term "set width" in Monotype caster composition, and thought it just meant "character width"; as I will need to precisely define it in order clarify the point at issue, in what follows I am going to have to explictly explain certain things that you, as a typographic professional, already know quite well, so that the definition will be generally intelligible, and I apologize for this in advance.)

Monotype machines work on an 18-unit system; different letters occupy different numbers of units; thus, M is 18 units wide, and i is 5 units wide. I did not suggest in any way whatsoever that this should be changed, say to make every character 9 units wide, to make it a monospaced typeface.

As you are well aware, at least in the era of cold type, most typefaces were optically scaled. The Monotype system also provided for optical scaling. The matrices in the matrix case were drawn with optical scaling, and the machine had a setting, called the "set width", that ensured the type slugs it created would correspond correctly in width to what the characters needed.

9 point Times Roman had a set width of 9 points, so an 18-unit wide M would be on a slug that was square: 9 points high and 1 1/2 pica ems (9 points, but in printers' language) wide.

But 7 point Times Roman had a set width of 7 3/4 points, so an 18-unit wide M would be on a slug that was 7 points high and 7 3/4 points wide.

And 11 point Times Roman had a set width of 10 1/2 points, so an 18-unit wide M would be on a slug that was 11 points high and 10 1/2 units wide.

So if you had said "A font does not have to be without optical scaling for these to be..." you would have at least come closer to what I was talking about.

But I'm not saying that either.

If one is allowed to make "em spaces" and "em dashes" that are modified slightly in width in order to fit on the grid...

then if one wants by that definition a "true em-dash" and a "true en-dash" in Times Roman, then while they would be 18 units and 9 units wide respectively in 9 point Times Roman...

clearly, then em-dash would have to be 19 units wide in 11 point Times Roman,

and in 7 point Times Roman, the em-dash would have to be 16 units wide, and the en-dash would have to be 8 units wide.

So not only does Times Roman not have to be monospaced, it also does not have to be not optically scaled.

But the number of units that describes the width of the em-dash and the en-dash has to change, depending on the point size, if the only deviation in their width from "a square" and "half a square" respectively that is allowed is to snap them to the nearest available grid point.

That is the point I was making. Something more than "snap to grid" but less than "change the widths of the em-dash and the en-dash to harmonize with the design of the typeface" is happening. One is also allowed to apply optical scaling to the width of the em and the en, at least within the Monotype system (Foundry typefaces, like those from ATF, did no such thing, of course).

John Savard

On further reflection, though, I should also have stepped away from the precise world of physics and mathematics into the more subjective world of the arts for a moment...

and noted that, since it would be a real pain to re-organize the matrix-case arrangement for a typeface in different point sizes,

while this does involve changing the widths of the em-dash and the en-dash away from the closest physically realizable approximations of the square and the half-square, the reason for it is still a bona fide technical consideration as valid as the need to round to the grid, and so it still is not really an argument in favor of letting the widths of those dashes be flexible.

So instead of just noting "Statement X is wrong", I should have also applied some perspective to see whether its deviation from precise accuracy was of any significance with respect to the issue being considered, or whether it was just a tiny footnote to be added for completeness.

Thomas Phinney

Thanks for the further explanation.

John Savard

Scott-Martin Kosofsky said:

The 3-em dash is still the standard to indicate repeated names in bibliographies.

I didn't even know there was such a thing as a 3-em dash.

I know that one convention for indented paragraphs was to indent them with an em-space.

On the one hand, if certain kinds of dashes are used for specific textual functions, then they should have a width that looks appropriate for those functions. One case that's been mentioned is the case of a condensed typeface.

On the other hand, it is nice if the names of things aren't misleading, so the width of an en-space ought to be half an em for that reason.

However, because em-dashes and en-dashes as sorts are what we had to work with, phototypesetting keyboards ended up with keys marked "em dash" and "en dash", and now we have Unicode codepoints with those names. Instead of dashes named after their functions - and that is the only reason the two principles cited above are in conflict.

So far, all of this has already been said, and it's also stuff that pretty well everyone can agree with.

The controversy is, therefore: given this mess, what do we do about it? Do we make fonts that will, if used naively, produce text that looks funny, or do we avoid that by consciously making fonts that include "en-dashes" and "em-dashes" that aren't an en or an em in width?

My personal opinion is that the first alternative is clearly unacceptable, and thus, in the interim, we have to go with the second alternative to have more happy faces among our customers... which does not mean we shouldn't also press to have the problem fixed properly by creating new Unicode codepoints with appropriate names, and having them become the preferred ones used by word processing software and so on.

Christian Thalmann

Am I the only one who often finds 500 unit en-dashes and 1000 unit em-dashes optically too short? Especially the en-dash, used with full spaces as Gedankenstrich in German, looks flimsy at 500 units. 

George Thomas

@Christian Thalmann -- Yes, about the en dash. That's why many commercial type shops during the 70s devised ways of making a 3/4-em dash to replace the en dash.

ClintGoss

George Thomas said:

... devised ways of making a 3/4-em dash ...

... which became U+2015 HORIZONTAL BAR (I'm guessing that's what you're referring to?).

The Wikipedia "Dash" page, Note A, says "In Cambria and many other fonts, the length of the horizontal bar is equal to ¾ em dash or 1½ en dash."  

ClintGoss

To further the confusicopia, some fonts change dash widths between styles:

Times New Roman v7.00 sets the advance width (ADW) of U+2014 Em Dash at 2048 for Rg, Bd, and BdIt, but Em Dash Italic has an ADW of 1821.

Bug? Interesting that 1821 is 16:9 of 1024 ... possibly intentional? These daftly dashes be dashed ...