1

So there is a debate about how to edit ellipses in posts on the site. This has come to a head in this latest post.

Here is what needs to be decided as an official policy so that we don't end up with edit wars/community wikis from too many edits.

What do we want for a site policy?
1. use three periods ...
2. use three periods with a space in between . . .
3. use a narrow-space plus dots (separate each dot with U+202F NARROW NO-BREAK SPACE)
4. use a proper UTF8 ellipsis …
5. let the poster decide which method to use and otherwise leave it alone

  • 9
    No5 is the most sensible, I think... – Mari-Lou A Sep 30 '13 at 17:56
  • I'm a typography nut myself, and while I appreciate that #3 is the best answer typographically, even I feel that it's too impractical to enter from a keyboard. – Bradd Szonye Oct 1 '13 at 0:56
  • By the way, a slightly better option than #2, with good support on most browsers and some keyboards, is to use a non-breaking space between each dot. – Bradd Szonye Oct 1 '13 at 0:57
  • That "latest post" example is somewhat unfortunate. The issue of ellipsis typographic representation does crop up in comments against an answer, but the first comment to the question itself really is just "Grammar Nazi" peeving over one user's preference/requirement for a definite article in "on the one hand". – FumbleFingers Oct 1 '13 at 20:48
  • 1
    I noticed that the last edit also removed the spaces from around the ellipsis. Please don't do that – sources differ as to how much space they need inside, but they almost always need extra space outside. – Bradd Szonye Oct 2 '13 at 0:40
  • 4
    This is an issue? There are people standing in line in the cold waiting for a pumpkin spice latte being told that there is no more pumpkin spice latte. – Mitch Oct 2 '13 at 2:16
  • It may be worth noting that – in this site's main font anyway – my eye can't see much difference between #1 and #4. Worst... peeve… ever... – J.R. Oct 3 '13 at 22:09
  • Your so-called “proper UTF-8 ellipsis” virtually never looks right. See Bringhurst. – tchrist Oct 4 '13 at 3:35
  • @tchrist, how exactly do you mean “virtually never looks right”? According to Wikipedia, Bringhurst recommends using either flush dots, thin-spaced dots, or the Unicode ellipsis character, and I cannot think of any examples (except in dealing with shoddy typefaces) where the precomposed ellipsis does not “look right”. – Janus Bahs Jacquet Oct 6 '13 at 17:26
  • @JanusBahsJacquet A perfectly reasonable question. The problem with the font ellipsis is that it looks compressed, causing the spacing fore and aft (which should be balanced) to look wrong. If you look at my image below, where I demonstrate the four styles, the font ellipsis looks at position 2 looks by far the worst in this font, or at least just as bad as position 1. I agree with Bringhurst that the thick spaces of position 4 are over the top, which is why I routinely apply his recommendation at position 3. Special circumstances can vary. – tchrist Oct 6 '13 at 17:42
  • Oh, oops—I had apparently not read quite far enough when I asked that. I have to admit, though, that I disagree with both you and Bringhurst here: to me, spaced ellipses (even thin-spaced ones) virtually never look right. In your image, I find #1 and #2 perfectly fine (if a bit on the narrow side—an issue I have long had with Georgia), but both #3 and #4 much too wide. Personally, I consistently use the precomposed glyph, both because it is semantically better and because I think it looks better in most faces. I do agree that some faces could use slightly wider precomposed ellipses, though. – Janus Bahs Jacquet Oct 6 '13 at 18:10
11

The only possible correct answer for a site policy is option 5: let the poster decide, and leave it the [expletive deleted] alone.

Edit wars like in the linked example are laughable. It's an ellipsis in an internet post, folks, not a government ID number on a tax return.

If you encounter a post where someone used regular spaces between the periods and it's breaking across two lines, then you can edit to fix it; but please also fix all typos and misspellings in the post while you're at it. If they merely used non-breaking spaces and you like your ellipses space-free (or vice versa), then please, for the love of all that's holy, go find something better to do with your time.

  • There's nothing better to do with their time. This then implies that there's no love for anything holy. sobs silently – Mitch Oct 3 '13 at 19:10
  • Preach it, sistah! +1 – Robusto Oct 19 '13 at 15:21
6

Personally I could live with any of 1,2 or (preferably) 5. Expecting people to use symbols that are not easily available on all keyboards and setups seems unnecessarily complex and, which is more, will simply be ignored by the vast majority of users.

If I had to choose, I would go for 5. While I appreciate correct punctuation as much as anyone and will be quick to edit out things like "I am wondering.........." which are so common in internetspeak, I feel that requesting strict adherence to typographical conventions is not really reasonable on a public Q&A site. In any case, the conventions themselves are not so clear here with different style guides suggesting different variants.

Then there are the various problems associated with some people not having set their browser to auto detect character encoding or who have set it to non UTF8 defaults. They won't be able to see options 3 and 4 correctly which makes these, in my opinion, more effort than they're worth.

Given that each of these options has its adherents and all of them will be understood I think the best option is 5.

  • I agree, although only if #2 is entered using non-breaking spaces. A line break in the middle of an ellipsis is terrible. – Bradd Szonye Oct 1 '13 at 2:05
  • Anyone who sets their browser to force any text encoding on pages as a default is a dolt and must be prepared to suffer the consequences themselves. They cannot be taken into consideration in site policy, or we would end up unable to use en and em dashes, curly quotes, diacritics of any kind, etc. – Janus Bahs Jacquet Oct 6 '13 at 17:17
  • @JanusBahsJacquet woah! Most people don't even know what text encoding is. Also, remember that some of us habitually use non-latin alphabets and that many webpages do not explicitly set the encoding being used. In the latter case, encoding has to be set explicitly. – terdon Oct 6 '13 at 17:24
  • @terdon, I agree entirely with your first sentence. Most people don’t even know what text encodings are—and therefore, they don’t actively set their browsers to force a text encoding on web pages as a default. These days, I would say the number of sites that do not define an HTML charset is rather small (though ironically, SE appears to be one of them)—but even when you run into one, you are overriding the browser default for that particular site. This is different from setting the browser default, which would be necessary to mess up SE’s display. – Janus Bahs Jacquet Oct 6 '13 at 18:01
  • @JanusBahsJacquet believe me I have had this issue with my father (which is kinda why I took offense at dolt ;). He would have set his encoding because it was necessary to see a badly written Greek page (depressingly common) and then, other pages would still have the default set. All I am saying is that there are various situations where a user could have set the default encoding for valid reasons and without being a dolt. – terdon Oct 6 '13 at 18:07
  • The dolt comment was perhaps a bit harsh—but anyone who sets their browser’s default encoding to anything but automatic detection without knowing what they’re doing or even what text encodings are, is acting very unwisely. Moreover, such incorrectly set-up sites are becoming increasingly rare in my experience. I regularly use sites written in at least five different scripts, and I haven’t had to force an encoding for a long time. Definitely long enough that I do so only on a per-page basis, never in the browser’s default settings. – Janus Bahs Jacquet Oct 8 '13 at 14:42
  • No, I meant changing the encoding through the view manu, not the settings. It just sometimes persists to the next page is all. – terdon Oct 10 '13 at 3:08
4

Unfortunately, typography and ease of use are at odds here, plus different style guides and typographers disagree about the details.

The one thing that nearly everyone agrees on is that there should be some extra space between the dots, and it should never break across lines. The exact amount of space varies according to context and taste, but NARROW NO-BREAK SPACE (U+202F) and NO-BREAK SPACE (U+00A0) are both reasonable choices. I prefer the latter because it's easy to type on a Mac keyboard (Option+Space).

I personally avoid the Unicode HORIZONTAL ELLIPSIS (U+2026) because very few typefaces actually have a well-rendered ellipsis glyph, so it's just as hard to type as other solutions and typographically inferior. For example, the ellipsis character in the question renders on my Mac indistinguishably from three unspaced periods.

Because there's no ideal solution here, I think it's usually best to leave ellipses alone when editing unless they are grossly incorrect. It's easy to actually make them worse if you're not careful (e.g., accidentally replacing non-breaking spaces with ordinary ones).

When writing new posts, I recommend encoding ellipses roughly in this order:

  1. Separate periods with NO-BREAK SPACE (U+00A0) if you can type it easily (Option+Space on Mac keyboards,   in Markdown): “. . .”

  2. Use unspaced periods: “...”

  3. Use HORIZONTAL ELLIPSIS (U+2026) – not recommended because it renders the same as unspaced periods in the ELU font, and it's harder to type (Option+Semicolon on Mac keyboards, … in Markdown): “…”

  4. Separate periods with NARROW NO-BREAK SPACE (U+202F) – not recommended because it has rendering problems similar to HORIZONTAL ELLIPSIS (U+2026), and it's even harder to type (  in Markdown): “. . .”

  5. Do not separate periods with ordinary spaces.

  • 1
    Your #1 looks just like periods separated with ordinary spaces to me. – Kit Z. Fox Oct 1 '13 at 1:52
  • I don't actually know what the narrow space looks like, as I'm too lazy to figure out how to enter U+202F. It's possible that it doesn't even render correctly without proper browser and font support. I think I might drop it further down the list. – Bradd Szonye Oct 1 '13 at 1:57
  • 1
    @KitFox Based on your feedback, I decided to not recommend the narrow space. – Bradd Szonye Oct 1 '13 at 2:01
  • @KitFox I finally figured out how to encode U+202F here. There actually is a difference between that (#4) and full-width spaces (#1), but it's subtle and probably not worthwhile. – Bradd Szonye Oct 1 '13 at 2:17
  • 1
    Thin space looks better, but it breaks, which is unacceptable. – tchrist Oct 4 '13 at 3:37
  • @tchrist The narrow no-break space recommended in the post does not seem to break (tested by adjusting paragraph width pixel by pixel) in the sense of allowing neighboring characters to display on different lines, at least on my machine. What did you mean? – Tyler James Young Sep 9 '14 at 16:27
  • @TylerJamesYoung U+202F NARROW NO-BREAK SPACE will of course not break, but U+2009 THIN SPACE will. For convenience, I have a file with things like this in them for easy mouse grab-n-dump: x . . . x (NARROW NON-BREAK SPACE, BOTH OUTSIDES) x . . . x (NARROW NON-BREAK SPACE, INSIDE ONLY). x. . . .x (NARROW NON-BREAK SPACE) I use the part within the two x characters for grabbing, depending on what I need. – tchrist Sep 9 '14 at 17:14
  • @tchrist Interesting. I use AHK hotstrings to replace what I type with special characters according to rules I define in a text file. My newest one (:?*:...::. . .) replaces “...” with “. . .” (the ? and * just override the default hotstring settings requiring a preceding space and an “ending character” such as another space). I just wasn’t sure why you were bringing up U+2009 in the first place since it wasn’t mentioned in Bradd’s post and a character he does mention (U+2026) already meets the aesthetic and functional criteria you present. – Tyler James Young Sep 9 '14 at 18:10
  • @TylerJamesYoung He brought up U+2009 because he prefers its width to U+202F for spacing ellipses, but unfortunately it is not usable because it’s a breaking space. – Bradd Szonye Sep 9 '14 at 19:10
  • @BraddSzonye Ah! Underlying my confusion is the fact that those two display at the exact same width for me. I gather there are configurations where that’s not the case. – Tyler James Young Sep 9 '14 at 19:32
2

If I can restrain myself from editing out the first word in all those ELU questions starting with a redundant So (like these three), I see no reason why others can't just accept the OP's typography.

Having said that, I don't entirely subscribe to Kitfox's option #5 here. Whilst I don't think mods (or anyone else, for that matter) should put any effort into "standardising" this particular aspect of typography (in Answers, too? - how about Comments?), I don't see any reason to rail against anyone who wants to do it. Unless the OP in question objects, in which case he has the final say.

We've been given clear guidelines on some typographical conventions adopted by one of our more "professional" users. Which is all very nice, but I doubt even he seriously thinks it would be worthwhile attempting to enforce consistency across all ELU posts.

2

I really wish people would read this posting — which I hereunder quote in full.


This is in Bringhurst. To quote from the other posting’s citation, since nobody else seems to think this is a duplicate but me, I point out that in his highly recommended Elements of Typographical Style, Robert Bringhurst writes on page 82 of version 3.2 of that book:

Most digital fonts now include, among other things, a prefabricated ellipsis (a row of three baseline dots). Many typographers nevertheless prefer to make their own. Some prefer to set the three dots flush … with a normal word space before and after. Others prefer . . . to add thin spaces between the dots. Thick spaces (ᴍ/3) are prescribed by the Chicago Manual of Style, but these are another Victorian eccentricity. In most contexts, the Chicago ellipsis is much too wide.

Flush-set ellipses work well with some fonts and faces but not with all. At small text sizes – in 8 pt footnotes, for example – it is generally better to add space (as much as ᴍ/5) between the dots. Extra space may also look best in the midst of light, open letterforms, such as Baskerville, and less space in the company of a dark font, just as Trajanus, or when setting in bold face.

[. . .]

In English (but usually not in French), when the ellipsis occurs at the end of a sentence, a fourth dot, the period, is added and the space at the beginning of the ellipsis disappears. . . . When the ellipsis combines with a comma, exclamation mark, or question mark, the same typographical principle applies. Otherwise, a word space is required fore and aft. The ellipsis is a graphic word.

That means that it should be four dots at the end and without a space before the first, but three dots in the middle and with a space, and they are usually spaced out between each.

  • I will have to go to the store . . . and stay there.

  • I will have to go to the store. . . .

The first should be written with a dash instead:

  • I will have to go to the store — and stay there.

The ellipsis has become stigmatized in recent years, as txkspkng kiddos have taken to using a slightly mangled version of the ellipsis in lieu of any and all other possible punctuation, whether commas or periods, dashes or parentheses, question marks or quotation marks.


And also this one, which gets into the code points used:


Yes, you do put a space in front of three of them, but not in front of four of them. The open questions are whether to use three or four, and whether to put spaces not just fore or aft, but between them. The short answers to those two questions are respectively

  • that you use four without a leading no-break space if it is the end of a sentence,

  • and that you almost always want to set them with thin no-break spaces between them, but this varies a bit depending on your face and point size.

Here follows a longer and more professional treatment. . . .

In his The Elements of Typographical Style, Robert Bringhurst writes on page 82 of version 3.2 of that book:

Most digital fonts now include, among other things, a prefabricated ellipsis (a row of three baseline dots). Many typographers nevertheless prefer to make their own. Some prefer to set the three dots flush … with a normal word space before and after. Others prefer . . . to add thin spaces between the dots. Thick spaces (ᴍ/3) are prescribed by the Chicago Manual of Style, but these are another Victorian eccentricity. In most contexts, the Chicago ellipsis is much too wide.

Flush-set ellipses work well with some fonts and faces but not with all. At small text sizes – in 8 pt footnotes, for example – it is generally better to add space (as much as ᴍ/5) between the dots. Extra space may also look best in the midst of light, open letterforms, such as Baskerville, and less space in the company of a dark font, just as Trajanus, or when setting in bold face.

[. . .]

In English (but usually not in French), when the ellipsis occurs at the end of a sentence, a fourth dot, the period, is added and the space at the beginning of the ellipsis disappears. . . . When the ellipsis combines with a comma, exclamation mark, or question mark, the same typographical principle applies. Otherwise, a word space is required fore and aft. The ellipsis is a graphic word.

I should add that if you do use thin spaces to space out your dots, you want to use U+202F NARROW NO-BREAK SPACE, not U+2009 THIN SPACE, because it is a single symbol, and must not be line-broken. You probably also want to control the line breaking before the three-dot form of the ellipsis by using U+00A0 NO-BREAK SPACE there. Notice how different these four scenarios work out:

  • No spaces: stuff ... here
  • Font ellipsis: stuff … here
  • Thin spaces: stuff . . . here
  • Thick spaces: stuff . . . here

Which for me looks like this:

ellipses demo

To my mind, the first two are both too skinny, and the last one looks too fat, leaving the third version to occupy the so-called Goldilocks position of being “just right”. It is indeed option number three, the one with thin spaces, which I have used in this posting – except when demonstrating alternatives.


Personally, I never, ever “type” these characters. I employ murine snarf-n-barf after invoking @Reg’s “sudo make mez a unicode” spell.

$ unichars '\p{space}' 
U+0009  -- CHARACTER TABULATION
U+000A  -- LINE FEED (LF)
U+000B  -- LINE TABULATION
U+000C  -- FORM FEED (FF)
U+000D  -- CARRIAGE RETURN (CR)
U+0020  -- SPACE
U+0085  -- NEXT LINE (NEL)
U+00A0  -- NO-BREAK SPACE
U+1680  -- OGHAM SPACE MARK
U+180E  -- MONGOLIAN VOWEL SEPARATOR
U+2000  -- EN QUAD
U+2001  -- EM QUAD
U+2002  -- EN SPACE
U+2003  -- EM SPACE
U+2004  -- THREE-PER-EM SPACE
U+2005  -- FOUR-PER-EM SPACE
U+2006  -- SIX-PER-EM SPACE
U+2007  -- FIGURE SPACE
U+2008  -- PUNCTUATION SPACE
U+2009  -- THIN SPACE
U+200A  -- HAIR SPACE
U+2028  -- LINE SEPARATOR
U+2029  -- PARAGRAPH SEPARATOR
U+202F  -- NARROW NO-BREAK SPACE
U+205F  -- MEDIUM MATHEMATICAL SPACE
U+3000  -- IDEOGRAPHIC SPACE

I have other commands that tell me for more than everything you ever wanted to know about a particular code point. For example:

$ uniprops -a 'NARROW NO-BREAK SPACE'
U+202F ‹ › \N{NARROW NO-BREAK SPACE}
    \s \h \pZ \p{Zs}
    All Any Assigned Blank InGeneralPunctuation Changes_When_NFKC_Casefolded CWKCF Common Zyyy Z Zs General_Punctuation Gr_Base Grapheme_Base GrBase HorizSpace Print Separator Space Space_Separator SpacePerl XPerlSpace White_Space WSpace X_POSIX_Blank X_POSIX_Print X_POSIX_Space
    Age=3.0 Bidi_Class=Common_Separator BC=CS Bidi_Class=CS Block=General_Punctuation Canonical_Combining_Class=0 Canonical_Combining_Class=Not_Reordered CCC=NR Canonical_Combining_Class=NR Script=Common Decomposition_Type=Nb Decomposition_Type=Nobreak DT=Nb
       Decomposition_Type=Non_Canon Decomposition_Type=Non_Canonical DT=NonCanon East_Asian_Width=Neutral Grapheme_Cluster_Break=Other GCB=XX Grapheme_Cluster_Break=XX Hangul_Syllable_Type=NA Hangul_Syllable_Type=Not_Applicable HST=NA Joining_Group=No_Joining_Group JG=NoJoiningGroup
       Joining_Type=Non_Joining JT=U Joining_Type=U Line_Break=GL Line_Break=Glue LB=GL Numeric_Type=None NT=None Numeric_Value=NaN NV=NaN Present_In=3.0 IN=3.0 Present_In=3.1 IN=3.1 Present_In=3.2 IN=3.2 Present_In=4.0 IN=4.0 Present_In=4.1 IN=4.1 Present_In=5.0 IN=5.0 Present_In=5.1
       IN=5.1 Present_In=5.2 IN=5.2 Present_In=6.0 IN=6.0 SC=Zyyy Script=Zyyy Sentence_Break=Sp SB=Sp Word_Break=Other WB=XX Word_Break=XX
  • 2
    This is good advice typographically, but I think it's too burdensome for the average user to make it a policy. I care about this kind of stuff and yet even I can't be bothered to use it unless it's something I can easily type. – Bradd Szonye Oct 4 '13 at 6:46
  • 1
    This is great guidance . . . for a dissertation. My hat's off to you if you want to type your ellipses that way. And at least this will always be here if I ever need it, so I'll favorite the question. – J.R. Oct 4 '13 at 13:06
  • @BraddSzonye Except for the Mac’s convenient and intuitively named pre-composed (read: NFC) accented characters like é, ñ, ç or curly quotes like “ ” ‘ ’, all reachable with a trivial Alt+whatever sequence (like Alt-n and n for ñ, I never type the non-ASCII code points. I pull them in with my mouse (hence murine snarf-n-barf) from somewhere that has them, whether from the IPA keyboard, or from my own cross-platform custom Unicode tools, or from any number of other resources. – tchrist Oct 6 '13 at 17:55
  • @tchrist That's cool, but it's too big a hassle for me to set up and use, and I'm pretty well motivated and tech savvy. Not worth the payoff. – Bradd Szonye Oct 6 '13 at 18:04
  • 1
    @BraddSzonye If you ever want to borrow a bit of my OCD, I have more than enough to go around. ) – tchrist Oct 6 '13 at 19:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .