Wikipedia talk:Manual of Style/Archive 112

From Wikipedia, the free encyclopedia
Jump to: navigation, search
Archive 105 Archive 110 Archive 111 Archive 112 Archive 113 Archive 114 Archive 115
The following discussion is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.

Issues cataloged by Noetica

In the summaries of Noetica's recent series of excellent edits, he mentioned several issues:

  • The verbosity of the serial commas section
  • The rules for hyphens
  • The rules for en dashes
  • The section on possessives

Unfortunately my understanding is that all of these are dormant volcanoes. Is there any one of them which we'd like to erupt now, or should we just leave this list here to remind us to erupt them later? Ozob (talk) 02:55, 6 January 2010 (UTC)

You know, we just did dashes, hyphens and possessives. Let's do verbosity in serial commas. That way we'll be hashing out the best way to say something rather than the actual content of the rule. I'm pretty sure most of us are cool with the idea that giving our editors their freedom with the serial comma is a good thing. Darkfrog24 (talk) 03:10, 6 January 2010 (UTC)
Upon reexamining the section in question, I like the examples and think we should keep them, but the first paragraph could stand to be smoother. Darkfrog24 (talk) 03:12, 6 January 2010 (UTC)
Just took a crack at it! What do you think? Darkfrog24 (talk) 04:15, 6 January 2010 (UTC)
Ozob: Thanks for taking this up, but there is little point addressing these specifics that I raise in my edit summaries until the elephant on the page is taken properly into account: how is consensus to be defined, achieved, and recorded, and how and when does it lapse? I refer editors to the sections devoted to these urgent questions, below.
Meanwhile, just for the record, here are some remarks on the issues you single out:
  • The matter of serial commas is a hobby of several enthusiasts who are attracted to MOS work. But that doesn't mean the topic should be given much prominence or space in MOS. The issues are better addressed if they are summarised, with one or two salient suggestions added, and with a definite referral to Serial comma. That article is very sound and detailed; it is also the proper place for the committed hobbyists I mention. The same applies, in some degree, to other sections also.
  • The guidelines for hyphens are fine; but they ought to be looked at again along with any fresh consideration of en dashes.
  • En dashes will of course get more discussion – when we have proper ways of discussing sorted out, I hope. Of course guidelines can be polished or altered; but the recent discussion of en dashes shows that we are sometimes not equipped to achieve worthwhile and stable results.
  • The section on possessives is a wonderful case to highlight the failure of present ways here at WT:MOS. It cannot be fixed now. Bigger issues first, then the dots.
I would add the section on ellipses, which was controversial a couple of years ago but is now stable enough. This is another perennial awkwardness in the printed guides, and it is particularly sensitive when transplanted into the uncertain soil we till at Wikipedia. I have found a neat way to fix the ellipses guidelines; but I will not put it forward unless circumstances improve.
Darkfrog: Sorry, but I find nothing deep or worthwhile in your remarks above. You say we have "done" possessives, for example. "Done", indeed. Rarely on this page has so much good effort been so waylaid by sheer ignorance, ill will, and disrespect for process. Please try to gain from the other perspectives that we have on offer.
¡ɐɔıʇǝoNoetica!T– 00:18, 7 January 2010 (UTC)
Yes, we've "done" them in that we had a large discussion about them not long ago. That you don't see anything too deep in my comments does not surprise me because I was attempting to amuse at the time. Noetica, when you talk about how ignorant other people are and how hard you have to look for something valuable in what they're saying, I find myself distracted from what I imagine to be your larger purpose. I find it useful to have some other page upon which to do my ranting so that I can stick to business when I'm here. I hear LiveJournal's good. Darkfrog24 (talk) 04:57, 7 January 2010 (UTC)

Spaces in endash

Following User:Tony1's revert of my edit, I would like to make a plea that the rule for spacing the endash when one or both items of a discussion must be changed. "Seifert – van Kampen theorem" looks horrid when compared to "Seifert–van Kampen theorem", and the latter is what I have seen at least in every good mathematical publication. It might be helpful to provide on the talk page here a reference to what some other style guides have to say about the endash. I can't find much in those one my own shelf. Sławomir Biały (talk) 12:51, 11 December 2009 (UTC)

I concur with Sławomir Biały. My experience is primarily limited to mathematical publishing, but it's a very consistent rule that en dashes are not spaced; this is true in American Mathematical Society publications, Springer-Verlag publications, Cambridge University Press publications, etc., etc. One never sees a spaced en dash. Ozob (talk) 12:55, 11 December 2009 (UTC)
I don't have a problem with spacing the endashes in the examples given in the MoS between cities or days. But now looking in the ACS style guide, I see that they recommend that endashes should never be spaced when between authors' surnames (e.g. van't Hoff&endash;Le Bel). So this rule, which Tony1 has previously asserted is universal, does appear to have this notable exception. Sławomir Biały (talk) 13:07, 11 December 2009 (UTC)
According to the Oxford Style Guide, en dashes appear always to be unspaced. It gives the example of Einstein–de Sitter universe (unspaced). Sławomir Biały (talk) 13:13, 11 December 2009 (UTC)
En dashes are always unspaced according to the APA Publication Manual. They do not give any interesting examples, though. Sławomir Biały (talk) 13:22, 11 December 2009 (UTC)
The Oxford Style Guide is just one publication of many. Is that the American Psychological Association's manual (it's shyte, IMO). I've changed it again, hoping this will suit. diff. I don't think it's wise to introduce subjective judgement as to whether "the lack of space is [not] visually awkward in some cases", so I removed that notion—we should not risk promoting arguments about "awkwardness" on article talk pages. Tony (talk) 13:27, 11 December 2009 (UTC) PS I personally hate the lack of spacing in your example, which now reads "Steiffert–van". That is what is awkward. Tony (talk)
Actually, if you read all three of the above posts, the APA, ACS, and Oxford style manuals all agree on this point: en dashes are never spaced. And I doubt all three can simultaneously be disregarded as "shite". I think it is time to reassess this paragraph of the text altogether. Do any major publication manuals recommend spacing disjunctive en dashes? I know it has been previously asserted that this practice is widespread in publishing, but I have found zero evidence of this based on the manuals that I have so far checked. Sławomir Biały (talk) 13:35, 11 December 2009 (UTC)
I keep the Chicago Manual close at hand, and so I can confirm it also does not use spaced en dashes (although some of its recommendations about when to use en dashes are not what one would expect). But at least the changes so far cover the original problem, which was the article Seifert–van Kampen theorem. — Carl (CBM · talk) 13:55, 11 December 2009 (UTC)

The change regarding surnames looks OK to me. It's worthwhile to search for "Seifert–van Kampen" on Google Books and look at the examples that appear in print there. I found books by Springer, Cambridge Univ. Press, Oxford Univ. Press, the American Math. Society Press, and Wiley, all of which used an unspaced en dash in the name. — Carl (CBM · talk) 13:39, 11 December 2009 (UTC)

Unfortunately, CMOS gets confused about some things, and delivers pronouncements that look rather committee-driven (their "noun plus -ing" advice looks as though they can't decide one way or the other, which I guess is forgivable). They also don't always take their own advice. The 2003 edition (15th) is less oriented towards en dashes, which suggests a greater proportion of typewriter nerds, and people who write computer code, on the committee. Tony (talk) 14:37, 11 December 2009 (UTC)
Ignoring CMOS for a moment, do you give any weight to the usage of actual professional presses, such as the five listed above? Can you point to which style manual does ask for a spaced en dash in Seifert–van Kampen theorem? — Carl (CBM · talk) 15:00, 11 December 2009 (UTC)

So the ACS, APA, CMOS, and Oxford manuals all agree that there are no spaces around en dashes. Although Tony1 will likely find fault in all of these manuals, even he should acknowledge that these four are at least among the ten most popular and influential guides currently on the market. I propose that we do away with the recommendation of spaced en dashes altogether so that our own manual is in accord with what appears to be the norm in the publishing industry, unless there are objections backed by some concrete sources. Sławomir Biały (talk) 14:59, 11 December 2009 (UTC)

Many hard-copy style guides do in fact recommend spaced en dashes, and none recommends unspaced en dashes when they are interrupters. Just why a few should insist on jamming together the innermost elements of spaced items is a mystery, and one that should not influence how WP does things on the Internet. WP is not bound to follow any particular hard-copy style manual: its style guide is shaped in relation to its own set of conditions – for example, the computer monitor, the display font(s) and sizes we use, and the readership. The notion that we should write jam together innermost elements is damaging to the meaning and readability of the text, and frankly is just bad practice:
  • New York–Chicago route (that new transatlantic route from York to Chicago); or
  • 19 September 1901–13 October 1979 (is "1901–13" a year range stuck in there? Or some kind of code?)
This matter was settled years ago here, and there is absolutely no reason to allow your personal distaste for "Seifert – van Kampen theorum" to blow out into a general anti-spacing campaign. Next we'll be told that we are no longer allowed the option of using spaced en dashes as interrupters – but instead must use solely unspaced em dashes—to cleanse the text of spaced punctuation. Perhaps we should change the rules on mathematical symbols to squash them up and save space:
  • 4–2=2, rather than 4 – 2 = 2.
And no more 4 pm, but 4pm. What exactly is wrong with spacing an en dash to avoid ambiguity and reading difficulty? Tony (talk) 16:38, 11 December 2009 (UTC)
"Many hard-copy style guides do in fact recommend spaced en dashes, ..." – let me repeat my question: please name the one you are looking at when you say that. I would be interested in looking it up myself. — Carl (CBM · talk) 16:45, 11 December 2009 (UTC)
I did a brief query using Google Books, and in the resulting snippets found several instances of recommendations of spaced en dashes in place of unspaced em dashes (e.g., p. 80 of Bringhurst 2004, ISBN 0881792063), but found no instances of recommending spaced en dashes when the items being separated contain spaces. So I'd like to see a citation about the latter as well. Eubulides (talk) 17:15, 11 December 2009 (UTC)
Yes, I agree that em dashes are often replaced with spaced en dashes. The thing we are looking at here are "disjunctive" en dashes, such as those that replace the word "to" or the ones that link multiple authors of a theorem. — Carl (CBM · talk) 17:35, 11 December 2009 (UTC)
To be perfectly clear – and I see that Tony's post above somewhat blurs this – we are here only talking about "disjunctive" en dashes, those to which the Spacing subheading of the MoS applies. I'm perfectly fine with spaced en dashes in place of em dashes. Sławomir Biały (talk) 19:14, 11 December 2009 (UTC)

"Seifert – van Kampen theorem" might "avoid ambiguity and reading difficulty" (it doesn't for me; the spaced en dash looks too much like an "interruptive" one – like this one), but even if it did, that'd be not enough: it needs to be in actual use. Otherwise, it'd be like writing "seeked" because it's clearer than "sought". And I've actually read someone suggesting that, though not on Wikipedia. I concur with Carl in asking which source actually suggests "Seifert – van Kampen theorem". (I can't recall ever seeing anything like that, and as a physicist I read about things named after two people all the time.) --___A. di M. 20:57, 11 December 2009 (UTC)

(Also, you want "4 − 2 = 2" with a minus sign, not "4 – 2 = 2" with a dash. And that is in actual use. --___A. di M. 21:06, 11 December 2009 (UTC))

Try again! "4 − 2 = 2" with a minus sign ... Art LaPella (talk) 21:51, 11 December 2009 (UTC)
Damn C&P... fixed now.--___A. di M. 23:30, 11 December 2009 (UTC)
Fowler's Modern English Usage and Fowler and Fowler's The King's English do not use spaces with their dashes, but they don't really discuss the distinctions among the dashes and it's not clear to me whether they intend their usage to be normative or not. (And, I should correct my claim above: In mathematics, nobody uses disjunctive spaced en dashes.)
Style guide lists a few more possible guides we might look at: AP Stylebook, A Manual for Writers of Research Papers, Theses, and Dissertations, The Elements of Style, The Elements of Typographic Style, ISO 690, MHRA Style Guide, MLA Handbook, MLA Style Guide, and The New York Times Manual of Style and Usage. I'm curious to know if any of these approve of disjunctive spaced en dashes. At the moment, it looks to me like the typographic world is united against them. Ozob (talk) 00:32, 12 December 2009 (UTC)
"In mathematics, nobody uses disjunctive spaced en dashes." OK, this came up towards the top of my first google search. Tony (talk) 06:29, 12 December 2009 (UTC)
That citation is to a set of class notes, which incorrectly capitalizes the "van" in "van Kempen", so it's an example of how not to format "Seifert–van Kampen theorem". For what it's worth, my searches for the phrase found nearly universal preference for unspaced en dash. Anyway, rather than focus on this particular example, do you have a citation to a style guide recommending spaces in cases like these? That would be a much stronger argument. Eubulides (talk) 06:44, 12 December 2009 (UTC)
So I tried looking up the style guides I listed above in Google Books. AP Stylebook, Turabian, ISO 690, MLA Handbook, and MLA Style Guide have no previews. Strunk is available, but doesn't discuss the matter. Elements of Typographic Style has only a snippet view; searching for "en dash" reveals the statement, "5.2.1 Use spaced en dashes – rather than close-set em dashes or spaced hyphens – to set off phrases.", which is nice, but not relevant to the present discussion. MHRA Style Guide suggests using spaced em dashes! (5.2, p. 26) But on p. 6, 1.3.5, it commands that en dashes shall be unspaced. The New York Times Manual of Style and Usage says to always used spaced em dashes, and gives a reason: The space makes it easier to break lines in narrow newpaper columns? It tells us not to use an en dash for this, p. 96. Later on pgs. 102, 157, 166, and 328 it tells us that a minus sign is an en dash, which is just wrong.
Unless someone provides at least one style guide that recommends spaced disjunctive en dashes, then I think this discussion is settled. Ozob (talk) 16:02, 12 December 2009 (UTC)
I really don't know why this extended discussion is going on in the first place, when the The New York Times Manual of Style and Usage even recommends the use of the open set em-dash. Also you can find more info about this topic at Wiki: Dash. In particular, see the section,

According to most American sources (e.g., The Chicago Manual of Style) and to some British sources (e.g., The Oxford Guide to Style), an em dash should always be set closed (not surrounded by spaces). But the practice in many parts of the English-speaking world, also the style recommended by The New York Times Manual of Style and Usage, sets it open (separates it from its surrounding words by using spaces  or hair spaces (U+200A)) when it is being used parenthetically. Some writers, finding the em dash unappealingly long, prefer to use an open-set en dash. This "space, en dash, space" sequence is also the predominant style in German and French typography.

That is enough for me to allow the open-set en dashes for disjunctive purposes. Skol fir (talk) 05:26, 15 December 2009 (UTC)
The above quote is about a different matter: using spaced en dash as a stylistic alternative to unspaced em dash. That matter is already covered in the MoS, in a different part, and is not relevant to the discussion about spaced en dash for ranges and other disjunctions. Eubulides (talk) 06:20, 15 December 2009 (UTC)
  • No, no one is going to overturn a long-standing formatting requirement just like that, particularly in such a way that produces ambiguous and hard-to-read gobbledy such as:

19 December 2009–11 January 2010.

Just about every biographical article would be adversely affected by that construction alone. Harts prescribes spaced en dashes where there are spaces in the elements, but I am away from home at the moment and have no access to style guides. In any case, just what dead-tree style guides recommend is only part of the matter. Wikipedia is an online project, not a hard-copy one. We have our own particular readership and contexts. We are not dictated to by particular publications that set themselves up as house authorities or, indeed, more general guidelines. Tony (talk) 16:24, 12 December 2009 (UTC)

While I'm for the spaced en dash in general, it's just silly for last names. It's unanimously Schmidt–von Buren–Bellford theorem or Ansky–de Vries connundrum in every article, book, report, etc... that I've even seen.Headbomb {ταλκκοντριβς – WP Physics} 16:41, 12 December 2009 (UTC)
I would like to see what the latest Hart's says. I looked around this elsewhere on the Internet, and found this comment by a Noetica:
"Many authorities, British and American, allow that an en dash can be used to replace a hyphen in forming certain compounds or phrases, perhaps flanked by spaces where spaces or hyphens are already present in one or more of the joined elements: Albanian–Greek trade talks; New Zealand – Australian rivalry; South-East Asian – Japanese tensions. Details vary, but this sort of thing is common. So is the en dash in marking a range: a 10–7 majority; pp. 23–39. Chicago minimises such uses of the en dash. Those are the basics. More could be said."[1]
In light of this, I propose that we change the MOS wording as follows (see updated proposal below):
"Spacing: Disjunctive en dashes are unspaced, except that spaces are allowed when either item contains a space or hyphen (June 3, 1888 – August 18, 1940, but June–August 1940). Proper names typically do not require spaces around the en dash (Seifert–van Kampen theorem)."
This would not rule out our existing usage, but it wouldn't require it either. Eubulides (talk) 20:22, 12 December 2009 (UTC)

The latest Hart's has this example (p. 84): "the New York–New Jersey–Connecticut area". Note the lack of spacing. I think we should do away with the recommendation that disjunctive en dashes be spaced, but that (per Tony's advice) allow an exception when used to separate dates. Sławomir Biały (talk) 23:13, 12 December 2009 (UTC)

This seems to contradict Tony1's statement above. Does it give any direct advice about spacing or not spacing disjunctive en dashes? — Carl (CBM · talk) 03:02, 13 December 2009 (UTC)
No, we do not want people writing full date ranges with the innermost elements squashed. The current wording allows for last-name squashing, but as Noetica points out, the spacing is required to disambiguate and for ease of reading in many places. Tony (talk) 03:16, 13 December 2009 (UTC)
I can easily find usages by high-quality houses in which date ranges are written with unspaced endashes. For example, "January 1, 2001–December 31, 2002" in Fig. 2 (p. 1399) of Cao & Lin 2008 (doi:10.1016/j.enconman.2007.12.030), an Elsevier journal; and "29 June–6 July" in Table 1 (p. 159) of Tan et al. 2004 (PDF; doi:10.1007/s00484-003-0193-z), a Springer journal. In contrast, I can't find use of spaced endashes for date ranges in either academic publishing house. Absent some advice from a reliable style guide (and Noetica's comments are merely in a blog somewhere), I don't see any evidence that spaced en dashes are common practice for this situation, outside of Wikipedia, and I see lots of evidence that unspaced en dashes are common practice for it. Eubulides (talk) 05:39, 13 December 2009 (UTC)

←Back to "Seifert – van Kampen theorem". The treatment in the literature is chaotic, as usual with hyphens and dashes. Writers, particularly scientists, grope towards something that they think might look reasonable and logical. Some give up for want of proper guidelines and just bang in the old (unspaced) hyphen they grew up with on the typewriter as school children and undergraduates. Some book and article editors clearly don't know either, and have a vague stab or leave their author's raw text as it is, enforcing consistency if you're lucky. A quick survey revealed several basic patterns and underlying motivations.

  1. Most writers realise there's something inherently unsatisfactory about jamming a little hyphen blob in among this multi-word item, and opt for the en dash.
  2. There's a definite trend to acknowledge that even an en dash, squashed in, resonates poorly with the gap between "van" and "Kampen". I suspect most writers don't explicitly ascribe this to an undesirable linking of "Seifert" and "van" as though they alone were the double unit, to be reverse-disambiguated only on the subsequent rightward fixation; nevertheless, the urge to insert some kind of space is there, and a space to the right but not to the left of the en dash is not uncommon. Unfortunately, this looks like the punctuation you sometimes see in lists ("The Beatles– Yesterday"), which almost always looks untidy.
  3. While Eubulides dismissed with a wave of the hand my example of "Seifert – Van Kampen theorum" straight from the top of a google search, because they got the "V" wrong (should be "v"), the V is common, even if it is strictly speaking wrong. Here are examples of the mess out there, just in mathematics textbooks (no wonder the style guides can't agree or say nothing much about it):

Now, as we've discussed before on this page, when there's a mess out there, we need to ensure that our own articles do better. We set the standard internationally for encyclopedic text, and we should not join the race to the messy typographical bottom. WP chose the best and most logical out there and has been applying it consistently ever since. It has served us well with full date ranges, it avoids ambiguity in multi-word place names ("South Australia – Northern Territory border"), and it is used for theorums and other surname-derived scientific items. Tony (talk) 07:27, 13 December 2009 (UTC)

Oh, I quite agree that there is a lot of bad typography out there, but this doesn't address the point that "Seifert–van Kampen theorem" is used by many high-quality sources, as are date ranges like "January 1, 2001–December 31, 2002", and so far all the style guides we've found (including Harts) agree that unspaced en dash is the way to go for these sorts of examples, with zero style guides saying otherwise. (By the way, none of those Google Books URLs work for me; Google Books URLs are not reliable in general.) Given this apparent unanimity of style guides, why are we insisting that the Wikipedia style is right and that Harts etc. are wrong? Eubulides (talk) 07:56, 13 December 2009 (UTC)
Every one of the Google Books URL's works for me. In my professional estimation it is not Google Books that has a reliability problem. —Aladdin Sane (talk) 08:12, 13 December 2009 (UTC)
Google Books tends to behave differently according to which country your IP is in. They are operating at the very edge of what copyright laws allow, or even of what is merely tolerated, and obviously that differs a lot between legislations. No need for comments that sound like insults here. Hans Adler 08:35, 13 December 2009 (UTC)
Actually, google books screw up a lot of punctuation when it quotes the relevant search word underneath the entry title: spaces too. Now Eubulides, who says the publications that use the typography you seem to want are "high quality", and the ones that use spaced en dashes are not "high quality"? Tony (talk) 08:18, 13 December 2009 (UTC)
  • When I say the Google Books URLs don't work, what I mean is that when I follow them, I see a bunch of blank pages. I don't know whether that's because Google has blacklisted my country, or my IP address range, or whatever, but for the purpose of this discussion it doesn't really matter. Since I can't see the books, I don't know whether the books use high quality typesetting.
  • I am not proposing that spaced endashes be outlawed in this case, only that they not be required. For the usage in question we have seen high quality style guides suggest unspaced endashes (with no style guides suggesting otherwise), and we've seen high quality publishing houses (Elsevier, Springer) use unspaced endashes. So why are we prohibiting unspaced endashes?
Eubulides (talk) 08:44, 13 December 2009 (UTC)

So you want people to be able to write:

  • 29 December 1918–19 January 1920

do you? No thanks: it's a shocker. I could not stand by and allow WP's formatting to be degraded in that way, with a two-year range (1918–19) stuck in the middle. It has worked just fine, and will continue to do so; until this small band of editors started complaining about the formatting of the theorum, no one has ever said a thing about it. I am unconvinced about allowing non-spaced punctuation in such items even in that, but it would be a much less egregious change that removing the well-established convention for dates and other items. I think the wording, as recently inserted, goes quite far enough, allowing people who have a bee in their hat about surname-derived multiple-word scientific items to use unspaced en dashes. Anything more would be unconscionable. Tony (talk) 10:06, 13 December 2009 (UTC)

In general I would like to see more spaced dashes, not fewer, and I agree with Tony on the substantive case here. But Tony, you seriously need an attitude adjustment. And by the way, learn to spell theorem. --Trovatore (talk) 10:14, 13 December 2009 (UTC)
New Hart's Rules: The Handbook of Style for Writers and Editors, by R. M. Ritter, 2005, pp. 79–80 specifies unspaced en dashes almost exclusively. There's an exception for spaced en dashes as a "parenthetical dash", and there's an exception for en dashes used to replace letters. (They give as an example of the latter, "'F – – – off', he screamed." That's right, no extra space between the last dash and the letter "o".)
As a test, I'd like to ask everyone: How would you read the following sentences?
It was a cold January – December had been warm, but now Mother Nature had decided to punish us.
I hated Monday – Friday was my favorite day of the week – so I hit the snooze button and rolled over.
Now compare:
It was a cold January—December had been warm, but now Mother Nature had decided to punish us.
I hated Monday—Friday was my favorite day of the week—so I hit the snooze button and rolled over.
and:
It was a cold January–December had been warm, but now Mother Nature had decided to punish us.
I hated Monday–Friday was my favorite day of the week–so I hit the snooze button and rolled over.
Rather than make a pronouncement, I want to ask: What do the dashes look like they are saying in these three renderings? Which look good, and which do not? Ozob (talk) 17:08, 13 December 2009 (UTC)
Good examples. No doubt others can be cooked up of the New York–Chicago variety. Sławomir Biały (talk) 00:02, 14 December 2009 (UTC)
Rather than cook up examples, why not look at what high-quality publishing houses do? Let's take Tony's "shocker" example: it's standard practice in high-quality publishing houses to do it the "shocker" way. I used Google Scholar to search for "1 January 2001 31 December 2002" (in quotes) and found one example of it formatted with spaced endash, a PhD thesis at the U. of Canterbury, which is not a strong source. In contrast, all three instances I found at academic publishing houses used unspaced en dash (the sources are doi:10.1016/j.vaccine.2004.07.038, doi:10.1080/00365540500525161, doi:10.1016/j.resuscitation.2006.03.005; these are two published by Elsevier and one by Informa). So, again, I see zero support for the notion that high-quality publishers use spaced endash, whereas I do see some support for the idea that some less-formal sources use spaced endashes. But most of the informal sources use hyphen! so clearly we cannot be relying on less-formal sources for advice in this area. Eubulides (talk) 01:04, 14 December 2009 (UTC)
To me Ozob's point with these examples largely undercut's Tony's. As soon as we see something appearing in a wider context, the spaced en dash puts too much separation between its disjuncts, and can easily be confused for a spaced en dash qua em dash. The fact that publishers also seem to prefer the "shocker" way is also fairly compelling. Sławomir Biały (talk) 01:37, 14 December 2009 (UTC)

For comparison's sake, here's Tony's shocker done several ways:

29 December 1918-19 January 1920
29 December 1918 - 19 January 1920
29 December 1918–19 January 1920
29 December 1918 – 19 January 1920
29 December 1918—19 January 1920
29 December 1918 — 19 January 1920

and here's a fun cooked-up example:

I wanted pizza, but I was in New York-Chicago-style deep-dish was so much better!
I wanted pizza, but I was in New York - Chicago-style deep-dish was so much better!
I wanted pizza, but I was in New York–Chicago-style deep-dish was so much better!
I wanted pizza, but I was in New York – Chicago-style deep-dish was so much better!
I wanted pizza, but I was in New York—Chicago-style deep-dish was so much better!
I wanted pizza, but I was in New York — Chicago-style deep-dish was so much better!

(I'd call that first one the "Freshman Special".) I'm not convinced that any of the options make the shocker look good, but I do prefer it unspaced. And while I had to bake the pizza example to get something so bad, I think it illustrates that you can't space disjunctive en dashes and then use spaced en dashes anywhere else. Ozob (talk) 01:54, 14 December 2009 (UTC)

The problem is your "cooked up" example is patently false. Chicago deep-dish is not better that New York-style. ;-) oknazevad (talk) 19:15, 14 December 2009 (UTC) (Sorry, couldn't resist)
Sir, you are an infidel. :-) Ozob (talk) 00:59, 15 December 2009 (UTC)

I don't believe date ranges should be "squashed". They look better with a spaced en dash.  HWV258  03:47, 14 December 2009 (UTC)

  • I agree with HWV258. I’ve long been a Mac user and "into" fine typography and page layout. For Mac people, it’s trivially easy to use en and em-dashes. So, since about 1988, I’ve keep a keen eye peeled for the conventions professional publications used for en-dashes. Indeed, real-world practices are all over the map and publications that even have their own manuals of style fail to religiously adhere to them. Accordingly, it is impossible to point to the practices of notable, authoritative publications and periodicals to make a case for doing something this way or that. In a nutshell: MOS’ guidelines governing the use of en-dashes in ranges has long served us well. The newly added exception to the general rule is reasonable enough and I certainly see no need to expand the exceptions any further; to do so would result in a non-harmonious hodgepodge of punctuation style; that is not what we need in an all-volunteer, collaborative writing environment. The current guideline, IMO, produces text that is least ambiguous, most natural to the eye, and reads most smoothly. Greg L (talk)
I also agree that the newly added exception is adequate. Since Wikipedia is an electronic medium, it has one issue to contend with that is not required with printed media: wide variations in page width. Readers have different width screens, and have their browsers sized to various widths. This causes Wikipedia articles to break at various point in any text. If unspaced en dashes are required in more circumstances, this can cause line breaks at undesirable places, such as:
... 29 December 1918–19
January 1920
... 29 December
1918–19 January 1920
Of course these can be corrected by adding non-breaking spaces, but that is cumbersome. Squashing en dashes and adding non-breaking spaces creates larger non-breaking units of text, which is not as 'friendly' to various browser widths. Since Wikipedia is electronic, it should not be required to follow MOS for printed media. CuriousEric (talk) 05:55, 14 December 2009 (UTC)
The previous comment seems to assume that high-quality publishers like Elsevier and Springer don't publish via HTML. That is incorrect: all the sources cited above are available via HTML and use unspaced endashes. So these publishers don't seem to have a problem with using unspaced endashes in an electronic medium. And that's OK, because wide variations in page width mess up date ranges regardless of whether unspaced endashes are used. The following line breaks are all possible with spaced endashes, and they're all bad:
  • ........ 1 January 2001 – 31
    December 2002 ..............
  • ......................................... 1
    January 2001 – 31 December 2002
  • ....................... 1 January
    2001 – 31 December 2002 .......
  • ............. 1 January 2001
    – 31 December 2002 ............
  • 1 January 2001 – 31 December
    2002 .............................
In practice these line breaks are avoided by putting the date range near the start of a paragraph, or in an infobox, or inside {{nowrap}}, or whatever; and all of these fixes work equally well regardless of whether spaced or unspaced endashes are used. Eubulides (talk) 06:21, 14 December 2009 (UTC)
This is quite interesting: WP:NBSP says that non-breaking spaces should be used "on the left side of spaced en dashes, if necessary for comprehension", and later on the same page, WP:MOSDB says "En dashes are preceded by a non-breaking space per WP:DASH." At a bare minimum, we need to choose spacing rules that don't confuse the reader. My first impulse is that whenever a date has an internal space, we should require that that space be non-breaking, i.e., not 1 April 2010 but 1 April 2010. That avoids all of Eubulides's examples above. And I think we should do this regardless of the spacing of en dashes.
Returning to the spacing of en dashes: WP:EMDASH says that an em dash can be replaced by a spaced en dash. Do we really want to allow both that and disjunctive spaced en dashes? How would you read the following example? This time I'm not giving any hints:
They flew New York – Burbank – New York – Los Angeles had been the original plan, but bad weather forced them to reroute.
Ozob (talk) 15:01, 14 December 2009 (UTC)
  • They flew New York – Burbank – New York; Los Angeles had .... Or a period before LA. It's a little stumpy, so consider "They flew from New York to Burbank and back again; Los ...". On nbspaces, yes, they can be used, but you wouldn't bother for, say, full dates that open a bio article and clearly won't be at the end of the line. I think it's quite unnecessary to insert a nbsp between month and year in the example you give. The more clutter, the more we put off newbies and casual visitors from contributing. It's a matter of balance. I see some people putting nbspaces between every number and the word it numerates: that's boring (312 horses). Tony (talk) 15:15, 14 December 2009 (UTC)

Funny, I would have put the period after Burbank. I read Ozob's example as "They flew from New York to Burbank. Flying from New York to Los Angeles had been the original plan, but...", with no mention of the return trip. Maybe the use of dashes instead of appropriate prepositions should simply be discouraged when there's a strong possibility of confussion. That said, this is far afield from the thread starting conversation about names in scientific citations, which seems to use unspaced endashes in most of the references dug up. I know that jibes with what I've seen in my scientific career. oknazevad (talk) 19:15, 14 December 2009 (UTC)

Yes, when I originally thought of the sentence, I intended it as "... New York–Burbank. New York–Los Angeles ..." But that's why it's so striking: It can be completely misread because of the two different meanings of a spaced en dash. Whatever our rules on en dashes, contrived ambiguity like this should be forbidden!
Regarding non-breaking spaces, would you support requiring a non-breaking space between a month and a day? As in, September 1 or 29 February? Ozob (talk) 00:59, 15 December 2009 (UTC)

I dunno how comes everyone is lumping all types of "disjunctive en dashes" together. There's a significant difference between ranges such as 29 December 1918 – 19 January 1920, which are usually in places (such parentheses, tables or infoboxes) where they don't play any grammatical role in any sentence (otherwise from 29 December 1918 to 19 January 1920 is clearer and commoner), and adjectival name pairs as in Seifert–van Kampen theorem. In the former, the space improves readability (though I can't see how 29 December 1918–19 January 1920 is ambiguous) with no drawback, whereas in the latter having a " – " identical to an interruptive dash in the middle of a noun phrase will eventually lead to garden-path sentences or actual ambiguities such as in Ozob's examples. The EU style guide, for example, in paragraph 2.19 says, "En dashes are used to join coordinate or contrasting pairs (the Brussels–Paris route, a current–voltage graph, the height–depth ratio)", with no mention of spaces whatsoever, but in paragraph 3.15 it suggests "€ 20–30 million, 10–70 °C" when "the symbol or multiple ... do not change" and "100 kW – 40 MW" "[i]f the symbol or multiple changes". I should point a COI here. I tend to pick up sequences such as "van Kampen" as one unit, probably also because I myself have a space in my surname; but this might not be the case for people in countries where spaced surnames are so rare that descendants of immigrants eventually CamelCase their surnames. --___A. di M. 10:29, 15 December 2009 (UTC)

Endash proposal 2

The above discussion makes it quite clear that style guides and high-quality publishing houses use unspaced endash for date ranges (and other constructs where ranges contain spaces), and that there is not agreement among the editors here whether spaced or unspaced endashes should be used in this context. In the light of this discussion I propose that we change the MOS wording as follows:

"Spacing: Disjunctive en dashes are unspaced, except that spaces are allowed when either item contains a space. Proper names typically do not need spaces around the en dash (Seifert–van Kampen theorem). A spaced en dash can lessen confusion in some cases (29 December 1918 – 19 January 1920).

As with the previous proposal, this would not rule out our existing usage, but it wouldn't require it either, and it would allow unspaced endashes as other high-quality publishing houses do. Eubulides (talk) 20:30, 14 December 2009 (UTC)

Update in response to comments below: I trimmed " and can increase it in others (The June 15 – July 15 schedule – introduced the previous year – was unpopular.)" from the end of the proposed text. Eubulides (talk) 21:29, 14 December 2009 (UTC)
Not big on your second (counter) example. When read out loud, its intended meaning is entirely clear, so it doesn't seem to be a good example of potential confusion. Other than that, it seems like a pretty good take on it. oknazevad (talk) 21:22, 14 December 2009 (UTC)
The red-text example can be rewritten as:
  • The June 15 – July 15 schedule which was introduced the previous year was unpopular. or
  • The June 15 – July 15 schedule—introduced the previous year—was unpopular. or
  • The June 15 – July 15 schedule (introduced the previous year) was unpopular.
As there are preferred methods, I don't see it as a good example of spaced en dashes increasing confusion. I would suggest that the proposed wording (if we must have it) is simply ended after the date example in green.
 HWV258  21:21, 14 December 2009 (UTC)
Thanks for the comment; I shortened the proposed wording as you suggested. Eubulides (talk) 21:29, 14 December 2009 (UTC)

I'm happy with this version for the reasons adduced here and in the earlier thread. Sławomir Biały (talk) 00:01, 15 December 2009 (UTC)

  • No way. You three are just trying to force your way towards shunting the long-standing usage of spaced en dashes everywhere towards the opposite. The current wording is just fine, thanks very much. It is unconscionable to allow people to write dates or multi-word ranges/to-from/opposition constructions with the innermost units jammed together. There is no reason to change. You've got what you wanted, for surname-derived scientific terms – WP should not be running with the lowest, most illogical standards out there, but the highest. Tony (talk) 02:37, 15 December 2009 (UTC)
I'm also happy with this version. Ozob (talk) 00:47, 15 December 2009 (UTC)
I would support this version as well, as it gives editors freedom to employ formats endorsed by common style guides. Users shouldn't need be burdened with learning Wikipedia-specific styles in addition to the widely-employed styles with which they are already familiar. Christopher Parham (talk) 02:44, 15 December 2009 (UTC)
Eubulides's latest matra is that certain "academic" publishers are "high quality". Let me think about that. But ultimately, it doesn't matter. Most high quality" publishers use title case in subtitles, or to insert hyphens after "-ly" adverbs. Eubulides, are you next going to mount a campaign for title case, drawing on the practice of most "high quality" publishers? My reaction is, so ;what if other publishers use crap formatting? We take the best, not the worst, and this small band of people here is trying to change a long-standing rule to allow editors to use the worst. It's absurd. Tony (talk) 03:43, 15 December 2009 (UTC)
Let's not personalize this discussion, please. So far, no evidence has been given of any publisher or style guide consistently using spaced endashes when the operands include spaces. On the contrary: we've seen examples of high-quality style guides (Hart's) and high-quality publishers (Elsevier, Springer) that use unspaced endashes. This is not a "campaign" to do anything other than to give Wikipedia authors the freedom to use the same sort of formatting that our best sources do. There is nothing wrong with writing New York–New Jersey–Connecticut area, as Hart's does. Eubulides (talk) 04:04, 15 December 2009 (UTC)
Seconded. Let's take the best of what the publishing world has to offer. So far, "the best" would seem to be represented by OUP, CMOS, APA, ACS, and New Hart's, and they appear to unanimously rule that disjunctive en dashes are always unspaced. Once again I think we need to ask if there are any major publication guides that recommend spaced disjunctive en dashes. So far that point does seem to have been lost in the din. Sławomir Biały (talk) 18:24, 16 December 2009 (UTC)
"...as it gives editors freedom to employ formats endorsed by common style guides. Users shouldn't need be burdened with learning Wikipedia-specific styles in addition to..."—now I'm getting worried; and I didn't sign on for this aspect. Like it or not, this is Wikipedia, and we have the right to have our own standards. It's not about whether any one editor can "get it right", rather it's how an editor's contributions can eventually be standardised. We have worker bees who will gradually apply our standards, and very often the original editor need not be cognisant of the syntactical change to their work. I reject any wording that allows the free-ranging use of MOSs from all sorts of sources (and ultimate inconsistency between articles). That way madness lies. We can do better.  HWV258  04:16, 15 December 2009 (UTC)
Do you think the current wording allows for "the free-ranging use of MOSs from all sorts of sources"? It doesn't appear to do so. Christopher Parham (talk) 04:23, 15 December 2009 (UTC)
I was replying to the comment "...as it gives editors freedom to employ formats endorsed by common style guides" which suggests it does.  HWV258  04:27, 15 December 2009 (UTC)
I agree with HWV's concerns, and ultimately with Tony though to a less vehement degree. The long-established standard is quite sufficient as a standard for WP's purposes and variety of content – if any changes are merited (for proper names, etc), they should be introduced as the potential exceptions, where clarity or consistency dictates. Not the other way around. /ninly(talk) 04:36, 15 December 2009 (UTC)
Having made that comment I was referring the specific styles indicated in the proposal. Christopher Parham (talk) 04:40, 15 December 2009 (UTC)
Understood, however let's be careful about the precedence we enable here.  HWV258  04:50, 15 December 2009 (UTC)
The proposed change is not a device for letting editors use whatever style guide they want. It's merely a proposal to improve the quality of Wikipedia in one small area, by not requiring spaces where our best-quality sources say that spaces should be omitted. No wholesale precedent would be set: in no other situation that we know of is Wikipedia's style guide diametrically opposed to all high-quality style sources that we can find. Eubulides (talk) 05:08, 15 December 2009 (UTC)
  • I couldn't disagree more: it is just that: a free-for-all, "use whichever style guide you like" notion. The current wording has been significantly compromised by the addition of the last sentence: "Spacing: Disjunctive en dashes are unspaced, except when there is a space within either one or both of the items (the New York – Sydney flight; the New Zealand – South Africa grand final; June 3, 1888 – August 18, 1940, but June–August 1940). Exceptions are occasionally made where the item involves a spaced surname (Seifert–van Kampen theorem)." It has already strayed too far from the stable, unambiguous formatting that WP has happily used for years, but unhappily it looks as though we'll be forced to accept it for the sake of keeping the peace. However, allowing that concession to be an immediate slippery slope towards do what you like shows how unwise it is: I am determined that it not be treated as an open door to formatting chaos in the project. I do not want to see one article entitled "New York – Ontario trade" and other entitled "New York–Ontario border": that is the kind of thing we'll get. I certainly don't want to read squashed date ranges at the top of bio articles, as you proposal encourages, with an unintended year range stuck in the middle, and the associated reading difficulty and aesthetic blooper.

Are the arguments now being dragged out just for the sake of winning? Are those concerning the dangers of line-wrapping serious? For example:
"Until the publication of the Seifert–van
Kampen theorem" (unspaced en dash) is just as likely as
"Until the publication of the Seifert –
van Kampen theorem" (spaced en dash).

We are already asked to insert a non-breaking space before a spaced en dash. I note that no one has ever complained that the date-stamp after your signature wraps onto the next line if it's unlucky enough to be in the wrong place; yet they can do that, and you've passed over examples.

(2) The examples of potential ambiguity provided above—in which spaced en dashes of different role are wound together—seem to be contrived for the purpose: any editor worth their salt will try to avoid the use of en dashes as interrupters right next to en dashes in their other roles; I have very occasionally had to ask for this to be addressed in FACs (like, once or twice a year). We have semicolons, commas and parentheses at our disposal for this very reason. Perhaps advice concerning this point might be added to the en dash section.

(3) "Users shouldn't need be burdened with learning Wikipedia-specific styles in addition to"—that reminds me of a certain user who has been banned from this page for persistent trouble-making: every publishing house has its own rules, unless they can't be bothered thinking about it and say to follow CMOS or the dreadful APA or whatnot. Here, we are more professional; people who have an anti-MoS agenda, please put up their hands, because we need to know whether this will be extended into a full onslaught on this style guide, casting everything it advises as second to whatever external style guide one wishes to invoke against it. My question has not been answered as to whether there will soon be a move to allow title case in our article and section titles, just because most publications do it. Tony (talk) 06:01, 15 December 2009 (UTC)

To answer your last question, nobody in this thread (certainly not me) is proposing to allow title case in our article and section titles: Wikipedia's section-title practice is common in other high-quality publications, and that clearly distinguishes the title-case issue from the endash issue (as spaced disjunctive endash is quite uncommon in other high-quality publications). In practice, unspaced endashes are no more (or less) ambiguous than unspaced: one can contrive examples that make either look bad, but these are rare in practice and are easily worked around (for both styles). The proposed wording does not encourage unspaced endash in date ranges: on the contrary, it gives date ranges as an example where spaced en dashes are apropos. As for "Are the arguments now being dragged out just for the sake of winning?" and "that reminds me of a certain user who has been banned": again, let's please comment on the topic rather than make unfounded speculations about motivation. Eubulides (talk) 06:20, 15 December 2009 (UTC)

My two cents: I'd somehow distinguish the case of ranges from the case of name pairs functioning as an adjective (see my last post in the section above). But I can't think of any decent wording for that. Also, what would be wrong with Seifert–van Kampen theorem? When I remember to, I do the same for all surnames, regardless of whether there's a dash before them. --___A. di M. 10:42, 15 December 2009 (UTC)

Tony, what do you mean by "low standards" and "high standards"? Standards you dislike and standards you like? And are you proposing that we write "seeked" rather than "sought" because it's more logical, despite the latter being way, way more common? (No, I'm not serious, but it appears that you are when making arguments extremely similar to these ones.) --___A. di M. 10:53, 15 December 2009 (UTC)

Eubulides, "Wikipedia's section-title practice is common in other high-quality publications"—nope, almost all academic publications still use title case. It was with delight that I found that WP doesn't. ADM, (1) could you ask Eubulides what he means by "high quality" publications? He introduced the concept; I subsequently used it with a sense of irony (hence my quotes). (2) By "logical", I'm not referring to the lexicogrammar of English, which is only selectively logical, but to punctuation. Your question has suddenly made me realise that punctuation is set to a much higher standard of logic than the lexicogrammar. Perhaps this is true in all written languages; it's certainly the case in English. Tony (talk) 12:03, 15 December 2009 (UTC)
  • "almost all academic publications still use title case" That's certainly not my experience. Perhaps my branch of academia is different? Anyway, I just now visited my favorite Wikipedia article, Autism, and looked at its first 10 refereed academic sources that contained section headers, and found that about half of them used Wikipedia-style sentence case section titles, and about half used title case. Just for the record, the sentence-case sources were PMID 18258309, PMID 18414403, PMID 15858952, PMID 15749245, PMID 17168158, the title case users were PMID 17967920, PMID 19478850, PMID 17367287, PMID 17967921, and one source (PMID 19128068) used both styles: title case for major sections and sentence case for lower level sections. (A couple of the sources used small caps; I am counting small caps as lower case in the previous list.) So the argument that Wikipedia is almost alone in using sentence case does not hold water.
  • "could you ask Eubulides what he means by "high quality" publications". I've given Elsevier and Springer as examples of two publishers that generally use high quality typography (though of course the quality varies among the enormous number of publications that both houses generate). The PMIDs listed in the previous bullet are also from publications of good quality, some from Elsevier and Springer, and some from other high-quality academic publishers such as Annual Reviews and Nature Publishing Group. None of these guys are perfect, but they're all respectable examples.
Eubulides (talk) 18:07, 15 December 2009 (UTC)
(out of sequence) I'll observe further that of those two groups of PMID refs, the former group are all to journals with Title Case used for the journal name, while the latter group are to journals with ALL CAPS for the journal name. Compare Acta Paediatrica to PEDIATRICS. It appears that the chain of decreasing priority styles simply flows down from a different starting point, in a more refined equivalent of MSWord's Heading1, Heading2, etc. In some if not all cases it may boil down to the use of a registered trademark to protect the style in which the journal title is shown, as the simple, commonly used English words "Nature" and "Pediatrics" would not be eligible for such protection, only their expression in a specific typographic style, a difficulty not faced by journals with compound titles. But of course that's just my OR. LeadSongDog come howl 21:18, 15 December 2009 (UTC)
What concerns me at the moment is that Wikipedia should be beautiful and clear. We all acknowledge that, spaced or unspaced, improper use of endashes can be ambiguous or misleading. So we're left with trying to do as well as we can within the limitations of conventional punctuation. I think that the best option, in this case, is to follow the most common practice—which would be unspaced en dashes. Tony, I realize that you're about to object that they're ugly, that things get "jammed together". Sometimes that does happen, and this is why the proposal above permits a space. But most style guides agree that most of the time, spaces make the two sides of the dash too spread apart. Most of the people here think that too. You're also about to object that changing this style guideline would change long standing practice. Well, so did ending date autoformatting. Sometimes consensus moves. And also, I know that you feel like a few of us are pushing you quite hard. I intended not to push you specifically but to change the MoS. But you've been a strong and a good advocate for the existing practice, and the MoS will be better because of that. I would be happy to address other members of the opposition. When we finally do reach consensus, I believe Wikipedia will be more beautiful.
You have a good point that we shouldn't allow articles entitled both "New York – Ontario trade" and "New York–Ontario border", so maybe the proposal as it stands isn't sufficiently prescriptive. (Perhaps we could mandate unspaced en dashes, except when the items being connected contain an internal space and when the words adjacent to the dash could be interpreted as a range?) I'd like to see a specific counterproposal from you. What would you like the MoS to say? Ozob (talk) 12:29, 15 December 2009 (UTC)
Thanks for this comment; it led me to propose a further improvement; please see #Endash proposal 3 below. Eubulides (talk) 18:07, 15 December 2009 (UTC)
When you (Tony) said "we need to ensure that our own articles do better", what you exactly mean by "better"? I think a style is "better" than another if it is less confusing and more easily understandable for the readers (which, IMO, usually boils down to what the readers are more familiar with). YMMV; but it's better to make sure we're not talking at cross-purposes. --___A. di M. 12:54, 15 December 2009 (UTC)

Tony's first comment that loosening our recommendation on spacing will lead to a "free-for-all" is pure demagoguery. There is nothing outlandish about the idea that punctuation needn't be absolutely mandated. Indeed, as Tony points out, an editor is free to use commas, colons, parentheses, and so forth rather than dashes in many cases. Yet how can there be "freedom" in this aspect of writing, and not an apocalyptic "free-for-all" that Tony cautions us urgently against? Secondly, I also disagree with Tony's point that the MoS guide has "already strayed too far from the stable, unambiguous formatting that WP has happily used for years", and in fact I find quite the opposite to be true. Prior to the "compromise" solution, the MoS failed to capture what has been a de facto editing standard on Wikipedia, and what is certainly a de dicto standard in the rest of the world. The article Seifert–van Kampen theorem used the standard unspaced en dash for quite some time, and there are certainly many more examples of this same formatting that I will not list here (Riemann–von Mangoldt formula, for instance). It was only because of a change in the apparent status quo that I became aware of our somewhat idiosyncratic use of spaced en dashes and raised the matter here. I am a firm believer that the MoS should strive to be descriptive rather than prescriptive: that is, it should attempt only to codify the best and most common practices of experienced editors. Prior to the current "compromise", exactly the opposite was true of the guideline. Sławomir Biały (talk) 14:54, 15 December 2009 (UTC)

Actually, the prevalence of stylistic free-for-alls and ensuing editwars are the reason that MOS exists. Even with MOS being very explicit there is a lot of stylistic chaos (e.g., I cannot count the number of times, even in the last week alone, that I've had to fix things like "February 3 2009" and "3rd Feb. 2009", etc.) Specificity here is a Good Thing. — SMcCandlish [talk] [cont] ‹(-¿-)› 17:54, 16 December 2009 (UTC)
If this is truly just a matter of preventing edit wars, then surely the best thing to do is bring our own MoS into accord with that of the majority of off-wiki publication houses. As has been repeatedly attested above, every reliable style guide says that disjunctive en dashes are always used. Sławomir Biały (talk) 18:02, 16 December 2009 (UTC)

Endash proposal 3

"maybe the proposal as it stands isn't sufficiently prescriptive. (Perhaps we could mandate unspaced en dashes, except when the items being connected contain an internal space and when the words adjacent to the dash could be interpreted as a range?)" That's a good idea, and I would favor such a change as well. Here's some proposed wording to do that.

Spacing: Disjunctive en dashes are unspaced (the New York–Sydney flight), except that spaces should be used if either item contains a space and the words next to the dash could be misinterpreted as a range (29 December 1918 – 19 January 1920, because without spaces the "1918–19" could mislead the eye).

I am not withdrawing the #Endash proposal 2 wording; that wording would be OK too, though I think this new proposal is a bit better. Eubulides (talk) 18:07, 15 December 2009 (UTC)

Updated in respose to the comments below, by adding 'because without spaces the "1918–19" could mislead the eye' to the example. Eubulides (talk) 19:44, 15 December 2009 (UTC)
Uh, "New York" contains a space. — SMcCandlish [talk] [cont] ‹(-¿-)› 17:50, 16 December 2009 (UTC)
  • Support. This looks closer to standard practice here and elsewhere than the other proposals. —David Eppstein (talk) 18:26, 15 December 2009 (UTC)
  • Huh? "29 December 1918 – 19 January 1920" is a range. Is there a typo? --___A. di M. 19:32, 15 December 2009 (UTC)
    To help clarify this I added 'because without spaces the "1918–19" could mislead the eye' to the example. Eubulides (talk) 19:44, 15 December 2009 (UTC)
    I'd use an example where the lack of space would create a real possibility of misunderstanding, e.g. €10 – 100 million (without the spaces, the bottom of the range could be taken to be ten million euros, rather than ten euros). I think many readers wouldn't even notice that the string "1918–19" in the middle of "29 December 1918 – 19 January 1920" happens to equal the string used to refer to a 24-month period, until told so. --___A. di M. 20:17, 15 December 2009 (UTC)
    A better example would be welcome, but I'm afraid "€10 – 100 million" isn't a good one either, as even with the spaces I interpret it to mean 10 million euros to 100 million euros. I chose "29 December 1918 – 19 January 1920" because it was mentioned so often in the previous discussion. I also see no ambiguity with unspaced endash there, but the proposed wording doesn't claim there is an ambiguity, only that it can confuse the eye. Perhaps someone could propose a better example? Eubulides (talk) 20:30, 15 December 2009 (UTC)
    The problem with the date example is that even with the spacing it looks more like a range and less like a disjunction. Maybe it's the case that all disjunctions should be unspaced, and ranges should be spaced if and only if either side of the range itself contains a space? —David Eppstein (talk) 20:43, 15 December 2009 (UTC)
    Sorry, I don't follow this comment. A range is a disjunction. Perhaps you meant "all non-range disjunctions should be unspaced, and ranges should be spaced if and only if either side of the range itself contains a space"? If so, that rule would be OK too, except perhaps for the difficulty of stating it clearly and concisely. Eubulides (talk) 21:35, 15 December 2009 (UTC)
    The term disjunction is a bit misleading for a general readership since usually the elements being described as grammatically in disjunction are usually practically being conjoined (as in 1918–1919, a conjunction of two years). I think eliminating it would probably make the rule of thumb a lot clearer. Christopher Parham (talk) 21:47, 15 December 2009 (UTC)
    ... just noting in passing that the New York–Sydney flight is not a good example either, given that the next bit says spaces should be used if either element contains a space (as in New York). Sssoul (talk) 21:26, 15 December 2009 (UTC)
    No, the next bit of wording says that spaces should be used "if either item contains a space and the words next to the dash could be misinterpreted as a range". In the New York–Sydney flight this misinterpretation is implausible, so spaces should not be inserted around the endash. The wording that's currently in the MoS is incoherent, as the text says that spaces should not be used with proper names, but the examples use spaces with proper names such as "New York". Both #Endash proposal 2 and #Endash proposal 3 fix this obvious bug in the MoS. Eubulides (talk) 21:35, 15 December 2009 (UTC)
    It is not implausible: it is impossible. There's no airport in York, and even if there were one, if we meant that we'd write "the new York–Sydney flight" with a small en. --___A. di M. 21:59, 15 December 2009 (UTC)
    My pedantic side demands me to inquire just how it is that either form is superior to the flight from New York to Sydney. It also has some difficulty with the idea that 55 weeks equals 24 months. LeadSongDog come howl 22:20, 15 December 2009 (UTC)
    By "24-month period" I meant 1918–1919 (two years); unless I misunderstood, it has been claimed that if the spaces around the dash in 29 December 1918 – 19 January 1920 were removed, the result would be ambiguous as the "1918–19" resulting in the middle could be taken to refer to such a two-year period. --___A. di M. 22:41, 15 December 2009 (UTC)
    On the other hand, if one attempts to read the disjunction as 1918–19, then one is left with the unlikely garbage "29 December" and "January 1920" at the ends. Recall that above, Eubulides searched on Google Scholar for "1 January 2001 31 December 2002" and got some results which used unspaced en dashes. I plugged "December 1918 19 January" into Google Scholar and went looking. In fact, the first relevant hit I came to was "6 December 1918-22 October 1994", and it was on page 22 of the results. It's an obituary from Biographical Memoirs of Fellows of the Royal Society, Vol. 44, (Nov., 1998), pp. 239-252, [2], and the date range appears as a subtitle. I went out to page 36 of the results before giving up. I don't think that 29 December 1918–19 January 1920 looks good, and apparently, neither does anybody else! Maybe we should just prohibit ambiguous en dashing in prose (I presume that in tables, infoboxes, etc. it will be clear what is meant)? Ozob (talk) 23:50, 15 December 2009 (UTC)
    Surely there's a typo in the previous comment, as that source (The Royal Society) uses an unspaced endash (not a hyphen) in the date range: "6 December 1918–22 October 1994". So this is more evidence that an unspaced endash is common in date ranges of this form among high-quality publishers, even when there's an internal substring such that "1918–22" that would convey the wrong impression if looked at in isolation. By the way, searching Google Scholar that way is reaaally inefficient, as most of the "hits" won't be date ranges at all. You'll have better luck with queries like '"December 2004 5 January"' (that is, with double-quote marks around the query). This query gives me 8 results, of which two are date ranges: one of those is from Springer and uses unspaced endash, and the other is from RBF Consulting (a lower-quality source) and uses spaced hyphen. When I went looking for examples, that's the general pattern that I found: the amateurs used hyphens or slashes, often with spaces, and the pros used unspaced endashes. Eubulides (talk) 00:16, 16 December 2009 (UTC)
    Indeed there is a typo. I cut and pasted from the Google search results, but that gave me a hyphen instead of the original en dash. Ozob (talk) 00:47, 16 December 2009 (UTC)
  • The new flight from York to Sydney. Good one. There is no consensus for any of these proposals. Tony (talk) 02:45, 16 December 2009 (UTC)
  • (ec) New York–Sydney flights will begin next year looks (to my eye) much more like there will be additional flights from York to Sydney; while New York – Sydney flights will begin next year looks like flights from New York to Sydney will commence in the near future. If you allow unspaced en dashes where one of the items has internal spaces, you make room for ambiguities. I'm content with recommending spaced en dashes where at least one of the items has internal spaces as standard (allowing an occasional exception for multiple surnames, where there is much less chance for ambiguity). I don't want to have to read text two or three times to extract the intended meaning and our present wording suits my reading style fine. --RexxS (talk) 03:01, 16 December 2009 (UTC)
    The trouble with "recommending spaced en dashes where at least one of the items has internal spaces as standard" is that from all evidence it seems to be standard nowhere but here, and conflicts with established usage in many cases. —David Eppstein (talk) 03:15, 16 December 2009 (UTC)
    I read both of the examples above as "Flights from New York to Sydney will begin next year", never as "New flights from York to Sydney will begin next year." So for me and how I read, if the latter meaning is desired, then there is no solution with en dashes. I suggested before that maybe we should forbid ambiguous en dashing. I'm coming to like this idea, as all the examples put forward are convincing me that en dashes are a very tricky thing. (And I used to think they were so simple!) Ozob (talk) 03:53, 16 December 2009 (UTC)

←David, I thought you'd like to know that I received notification yesterday of "the 14th Biennial International Conference on Baroque Music at the Queen’s University, Belfast, 30 June – 4 July 2010". On the now-infamous "Steiffer – van Kampen theorem", I've already provided examples demonstrating that there is no established usage out there, including textbooks that use the spaced en dash; thus, the notion that spaced en dashes "[conflict] with established usage" is hard to justify. Tony (talk) 03:58, 16 December 2009 (UTC)

He was Seifert, not Steiffer. See Herbert Seifert. Ozob (talk) 12:07, 16 December 2009 (UTC)
I'm sure one can find many examples of brochures and the like that use spaced endashes for date ranges. But it's rare among higher-quality sources. I just now searched Google Scholar for that particular date, and found just one example doi:10.1093/em/can139 of it. This was published by the Oxford University Press, which used an endash "30 June–4 July 2010". In contrast, if you search the web, you can find many lower-quality sources that use spaced hyphen, or unspaced hyphen, or emdash, or even weirder combinations (such as an emdash with a space after but not before) to make the same announcement. So again, we see that a higher-quality source is using an unspaced endash for date ranges. I agree that in the wider world of self-published sources and lecture notes, one can easily find spaced endashes for that theorem, but the established publishing houses and style guides are uniformly in favor of unspaced endashes for disjunctions like that. Eubulides (talk) 07:08, 16 December 2009 (UTC)
I have no objection to spaced en-dashes in date ranges. I do object to them in the names of mathematics results. Between those two extremes, I don't have as strong opinions, but I would like to see either agreement that the spaces should only be used in ranges, or an example of a non-range disjunctive en-dash that a plurality of stylebooks or other sources would space. —David Eppstein (talk) 04:34, 16 December 2009 (UTC)
Eubulides, it's the queer notion of "high quality" and "low quality" sources that gets me. The "14th Biennial International Conference on Baroque Music at the Queen’s University, Belfast, 30 June – 4 July 2010" came in a letter from the Bach Network UK, along with its scholarly journal Understanding Bach. Shall I relay your concerns to them? On OUP, well ... let me tell you I've edited one book to be published by them. Hmmm ... like to have a look at their house rules? I can send them if you wish. Tony (talk) 07:58, 16 December 2009 (UTC)
It's safe to say that the Oxford University Press journal Early Music is edited to higher professional standards than promotional letters from the Bach Network UK. Even their scholarly journal Understanding Bach is lacking, as it uses unspaced hyphen for disjunction in the most-recent article published in it, with examples like "3-4 January 2009", "1734-35", and "the Buttstett-Mattheson controversy".[3] Eeeuwww; this is clearly bad style; but it is a new standalone journal and I suppose they'll learn eventually. Anyway, if those OUP house rules mention disjunctive endashes, then it'd be good to know what they say. Eubulides (talk) 09:41, 16 December 2009 (UTC)
The OUP managing editor replied to my emailed enquiry about en dashes that they were required (I had to ask, since the matter wasn't mentioned anywhere in their house rules). I had the distinct impression that they hadn't thought much about the details of en-dash usage, and while their use was definitely required, beyond that it was left up to writers. Hmmmph. So it was up to me to reign in chaotic usage among the 20 or so chapter contributors. Tony (talk) 11:00, 16 December 2009 (UTC)
Unfortunately, right now Wikipedia is as chaotic as the usage of those chapter contributors. Looking at Emmy Noether, one finds unspaced en dashes in the infobox, in section headings, and in prose. On 2003 Pacific hurricane season, however, every en dash is unspaced, excepting only the en dashes that appear in links to other articles. And both of these articles are FAs. Ozob (talk) 12:22, 16 December 2009 (UTC)
I fail to see how the use of unspaced disjunctive en-dashes in our FAs, consistent with the use of unspaced disjunctive en-dashes by high-quality publishers discussed earlier, constitutes "chaos" in any sense of the word. I was peripherally involved in the FA discussions for the Noether article and I remember consistency of dash usage being carefully examined in that case (the issue there being more em-dashes vs spaced en-dashes as punctuation between clauses of sentences). —David Eppstein (talk) 18:42, 16 December 2009 (UTC)
While "Seifert – van Kampen theorem" is attested, it is a very minoritarian usage, so saying "there is no established usage out there" sounds like a stretch to me. By that count, there's no established usage for the past tense of "seek", either, as both "seeked" and "sought" are attested.[4][5][6][7] Also, I can't see why most of you insist in treating ranges and adjectival compounds the same way. "Los Angeles–Chicago flight" is a construction exactly parallel to "London–Chicago flight", except that one name happens to contain a space and the other doesn't; therefore, spacing one but not the other would be illogical, and even slightly distracting if the two happened to be found in the same paragraph. On the other hand, "10–40 MW" and "100 kW – 40 MW" are not parallel constructions; can you see why? ― A. di M. — 2nd Great Wikipedia Dramaout 16:55, 16 December 2009 (UTC)
Honestly, I don't. I read "10–40 MW" as "from ten to forty megawatts" and "100 kW – 40 MW" as "from one hundred kilowatts to forty megawatts"; to me they are both ranges.
You have a very good point about the parallelism in "Los Angeles–Chicago flight" and "London–Chicago flight". The MoS would be wrong if it told us to write something like, "The Hurricane Katrina relief supplies were shipped New York – New Orleans and Chicago–Biloxi." Ozob (talk) 23:17, 16 December 2009 (UTC)
I referred that to the fact in the first range, one piece (the unit of measurement of the bottom) is implicit, but not in the second. "10 – 100 million" vs "10–100 million", where the bottom of the first range is taken to be 10 and the bottom of the latter is one million times as much, would be a clearer example of what I meant. OTOH even the former is likely to be misunderstood as from 10 million, except in contexts where it's clear one could expect a range spanning seven orders of magnitudes. ― A. di M. — 2nd Great Wikipedia Dramaout 23:46, 16 December 2009 (UTC)
  • Support either proposal 2 or 3. Sławomir Biały (talk) 17:13, 16 December 2009 (UTC)
  • Oppose: On what authority is it asserted that parallelism of the construction is the only or even most important consideration? I would say that prevention of ambiguity/confusability trumps it, considerably. And I'm a consistency fan. "Los Angeles–Chicago flight" is a misleading example, because most en.WP readers are familiar with these names and recognize them immediately, with no possibility of confusion. This might not be the case with a construction like "a slow Mallu-Chhitt–As Salatah al Jadidah route". Several proponents of rewording would oppose a spaced en-dash here (on the basis of the two terms being proper names, on the basis of the "London–Chicago flight" parallel, or both), which would be a very bad editing decision in my view. — SMcCandlish [talk] [cont] ‹(-¿-)› 18:14, 16 December 2009 (UTC)
    I had never heard of either location before, and yet I had no problems in understanding the phrase. And there's nothing to stop you from writing "a slow route from Mallu-Chhitt to As Salatah al Jadidah". ― A. di M. — 2nd Great Wikipedia Dramaout 18:51, 16 December 2009 (UTC)
    Agree with A. di M., the free text form is better. Not to put too fine a point on it, but adding wikilinks makes that even more legible as "a slow route from Mallu-Chhitt to As Salatah al Jadidah". By the second use, the reader has already seen the names and is so less likely to stumble. LeadSongDog come howl 19:47, 16 December 2009 (UTC)
  • I never suggested the free-text form wasn't better (I didn't address that issue at all). What I did address is the assertion that the unspaced version is always better than the spaced one, which you haven't counter-addressed. — SMcCandlish [talk] [cont] ‹(-¿-)› 06:43, 17 December 2009 (UTC)
  • Please don't vote: The discussion is still going, and we haven't reached consensus yet. Ozob (talk) 23:05, 16 December 2009 (UTC)

En dash Proposal 4

In rereading the previous subthread, it appears that the previous proposal caused confusion because it spent too much time on mechanics, and too little on motivation (the motivation being clarity). We also saw further evidence that major academic publishers such as Oxford University Press and The Royal Society use unspaced endash, with the only counterexample being a lower-quality source (a promotional letter from a musical society). Clearly there is not a universal consensus among editors here about spaced endash when a disjunct contains spaces: some would ban spaced endash entirely (since that's what major academic publishing houses do), some would always require spaced endash (since that was in the MoS for quite some time), and some would favor a compromise.

In the spirit of compromise I propose the following replacement wording:

Spacing: Disjunctive en dashes are unspaced, except that if either item contains a space, a spaced en dash should be used if needed to avoid confusion (29 December 1918 – 19 January 1920). Unspaced en dashes typically suffice between proper names (the Seifert–van Kampen theorem).

I'm not withdrawing the #Endash proposal 2 or #Endash proposal 3, either of which I could also live with. The point of this proposal is to have a simple commonsense rule that allows existing practice and is consistent with what seemingly every major academic publisher does. Eubulides (talk) 00:06, 17 December 2009 (UTC)

  • I can live with any of these proposals. Sławomir Biały (talk) 01:25, 17 December 2009 (UTC)
  • How about, more strictly: en-dashes in ranges of dates should be spaced when the dates in the range contain spaces (29 December 1918 – 19 January 1920). All other disjunctive en-dashes should be unspaced, even when the items being separated themselves contain spaces (1901–1978, the Seifert–van Kampen theorem). —David Eppstein (talk) 01:43, 17 December 2009 (UTC)
    This would lead to The European theater in World War II was active 1939–1945, concluding with the 16 December 1944 – 25 January 1945 Battle of the Bulge, which I think looks a little bit odd. Ozob (talk) 02:15, 17 December 2009 (UTC)
    I still don't know why we don't engage in writing English wherever possible: The European theater in World War II was active from 1939 until 1945, concluding with the 16 December 1944 to 25 January 1945 Battle of the Bulge? Wouldn't that approach be the best way of removing the need for this debate?  HWV258.  02:54, 17 December 2009 (UTC)
    I agree that words are often better than symbols. However, I think we should design our style manual to be robust against even the most contrived situations, and this is one of them. In my proposal below I simply forbid such nonsense, which I think is the most reliable way to handle it. Ozob (talk) 04:48, 17 December 2009 (UTC)
    So the spaced en dash as interrupter is "nonsense" too? You haven't complained yet about exactly the same logic that spaces it – the presence of spaces between the adjacent words. That is why we don't write them unspaced–like that. Em dashes in that function are regarded as being sufficient large to cope with no spaces. Tony (talk) 04:51, 17 December 2009 (UTC)
    I meant that contrived examples that are willfully ambiguous or confusing are nonsense. Sorry for the confusion. Ozob (talk) 04:58, 17 December 2009 (UTC)
    I would also be happy with David Eppstein's wording proposed above. I agree that prose containing mixed spaced and unspaced en dash for related items should be rewritten to avoid the jarring inconsistency, but that's true no matter what rule is allows spaced en dash (including the current MoS's rule). Eubulides (talk) 09:07, 17 December 2009 (UTC)

En dash Status Quo

My head is spinning, and other editors will suffer the same fate. Why introduce a quite new criterion ("if needed to avoid confusion"), which itself will cause confusion and result in lots of queries here on individual cases. You need to have thought through the issue to know the potential for confusion, and that is asking too much of the poor editors, who just want quick, simple advice; that is what the established rule provides, with a simple principle: spaced element(s), spaced dash; unspaced element(s), unspaced dash.

The current wording is a compromise, remember: that sentence Eubulides inserted at the end about the blessed Stifle it–van Camp thingeme as an exception. Let's stop torturing it and leave as is. Tony (talk) 01:43, 17 December 2009 (UTC)

  • Oppose. The status quo requires spaced en-dashes in disjunctions of spaced items in contradiction to all the evidence presented above that shows that unspaced en-dashes are used by most/all stylebooks and high-quality publishers. The status quo is "Tony gets his way against the consensus of most of the other editors in this thread". I don't see this as a compromise at all. —David Eppstein (talk) 01:47, 17 December 2009 (UTC)
You seem to be turning it into a personal matter. It's Tony against you. No, many other editors have expressed opposition to changing the status quo. Tony (talk) 04:21, 17 December 2009 (UTC)
  • Support: The issue is becoming a) far too polarized in the case of certain editors, and b) far too scattered and random in the case of everyone else. We cannot meaningfully come to consensus on 6 or more competing proposals, especially when new ones are arriving at the rate of 3 per day (i.e. doubling in under 24 hours). This (the whole meta-thread, not this sub-topic) should simply be closed as {{Resolved|No consensus; discussion restarted in new thread.}}, and a new discussion topic on the matter opened, devoid of any proposals anyone is being asked to !vote on and which are clearly causing some participants considerable irritation. Let's see what emerges from that before rushing to change anything. — SMcCandlish [talk] [cont] ‹(-¿-)› 06:40, 17 December 2009 (UTC)
  • "that sentence Eubulides inserted at the end about the blessed Stifle it–van Camp thingeme as an exception" First, I didn't insert that sentence. Second, "Seifert–van Kampen theorem" is not an exception: the universal practice among high-quality publishers and style guides we've found is to use unspaced en dashes in similar phrases, such as "New York–New Jersey–Connecticut area" (Hart's), "von Hippel–Lindau syndrome" (doi:10.1016/j.fertnstert.2008.11.013, Elsevier), etc. Wikipedia's current style rule not only fails to describe really happens on Wikipedia, it prohibits what high-quality publishers invariably do. Eubulides (talk) 09:07, 17 December 2009 (UTC)

En dash Proposal 6

I think the following proposal is the one that would be most consistent with itself and with outside style guides:

Spacing: Disjunctive en dashes in prose are unspaced. For example, 1914–18 or Navier–Stokes equations. En dashes remain unspaced when one of the items being joined have spaces. For example, New York–Chicago, not New York – Chicago, and 29 December 1918–19 January 1920, not 29 December 1918 – 19 January 1920. En dashes also remain unspaced when one of the items being joined has a hyphen. For example, Gell-Mann–Nishijima formula, not Gell-Mann – Nishijima formula. Disjunctive en dashes outside prose, such as in tables or infoboxes, may be spaced.
If a sentence would be unclear with an en dash, then it should be rewritten to remove the confusion. For example, New York–London flights will begin next year could be interpreted to mean New flights from York to London ... or Flights from New York to London .... If both interpretations are plausible in context, then the sentence should be rewritten so that it is unambiguous.

Ozob (talk) 03:45, 17 December 2009 (UTC) Update: Revise the last paragraph. Ozob (talk) 05:12, 17 December 2009 (UTC)

  • Support (mainly): Yes, that is agreeably consistent and verifiably agreeable. But really, you can imagine a context where New York–London flights is misinterpretable? Anyone capable of that majestic feat of misinterpretation is certainly clever enough to misinterpret New York–Chicago flights (direct!) as well.—DCGeist (talk) 04:05, 17 December 2009 (UTC)
  • Oppose. It's weird. Tony (talk) 04:19, 17 December 2009 (UTC)
    I'm weird. :-P Ozob (talk) 04:54, 17 December 2009 (UTC)
  • Support. This is fairly extreme in its repudiation of spaced disjunctive en-dashes, but I'm still willing to go along with this one. —David Eppstein (talk) 04:20, 17 December 2009 (UTC)
  • Oppose. I just don't believe that 29 December 1918–19 January 1920 looks better than 29 December 1918 – 19 January 1920. Aren't we getting a long way away from the original idea of allowing other styles (something I disagreed with anyway)? The current text looks like dictating a particular style.  HWV258.  04:39, 17 December 2009 (UTC)
    Yes, this proposal is very prescriptive, which could be a disadvantage. But it has the advantage of being consistent, simple, and in agreement with outside style guides.
    I agree that an unspaced endash in a standalone date may look bad. This is why there's an exception at the end for dates not in prose. In prose, I think it looks fine. Ozob (talk) 04:52, 17 December 2009 (UTC)
  • Oppose: Indeed, the preceding looks much worse and is a head-scratcher; it forces the reader to stop and ponder.—Finell 04:49, 17 December 2009 (UTC)
  • Just curious: Virtually every reputable style guide in the English language disagrees with you. What do you make of that?—DCGeist (talk) 04:52, 17 December 2009 (UTC)
  • You could also address the point raised: the 1918–19 in 29 December 1918–19 January 1920 is a head-scratcher. Why would we want to interrupt the flow of reading?  HWV258.  05:26, 17 December 2009 (UTC)
  • I'm happy to address that: I don't find it a head-scratcher, and nor do virtually all reputable style guides in the English language, which suggest a preference for precisely that style. What do you make of that?—DCGeist (talk) 05:32, 17 December 2009 (UTC)
  • There are those here that do find it a head-scratcher. This is WP, and we will let consensus here decide our style. That's all that matters.  HWV258.  05:41, 17 December 2009 (UTC)
  • Don't lecture me about consensus. Would you like me to lecture you about consensus? No, I thought not. Be respectful with those you disagree with, as I have.—DCGeist (talk) 05:57, 17 December 2009 (UTC)
  • To tackle your point: have you considered that with over three million articles (and with the most consensus-driven MOS creation process imaginable), perhaps they should consider adopting parts of our style guide? We have developed (quite happily) the standard of a spaced en dash between the birth and death dates in our biography articles. If this proposal gets up, every one of those will be changed. For what benefit? Because it matches the style in some other MOSs?  HWV258.  05:41, 17 December 2009 (UTC)
  • That's a fair argument. (See how nice it is to be respectful, friend?)—DCGeist (talk) 05:57, 17 December 2009 (UTC)
  • "every one of those will be changed" Well no, the actual Wikipedia consensus out there is about halfway between the alleged "status quo" and Category:Hyphen Luddites. Choosing the C's in Category:American farmers as a random sample, I found 9 spaced en dashes between birth and death dates, one unspaced en dash, one unspaced because there were years only, 8 with no birth–death range, and 6 hyphens! Art LaPella (talk) 06:34, 17 December 2009 (UTC)
  • Ah. Now, if I had your attitude, HWV258, I would take advantage of Art's research to start lecturing you about laziness, or deception, or something else inimical to "this is WP". But I'm not going to do that. You know why? Because I don't assume that, just because you disagree with me, you're full of crap.—DCGeist (talk) 07:17, 17 December 2009 (UTC)
  • Thanks Art LaPella for nicely demonstrating what happens when people take "standards" from wherever they choose. To DCGeist, I'm seriously hoping that in the cold, sober light of day you might back-track a little from your comments as my post didn't warrant them. Specifically, Art's research lends support to my statement. I referred to birth/death pairs, so the "8" are excluded. Of the 17 remaining cases, 15 will be altered by this proposal, so that's 88%. 88% (while not being "all") of the number of articles we have that contain birth/death pairs is still a lot of updates to WP (and my point-of-view is that the result of those edits will be to make the text less readable).  HWV258.  21:04, 17 December 2009 (UTC)
  • My point was that the hyphens won't be changed by a new proposal, for the same reason the hyphens haven't been changed by the existing guideline. Art LaPella (talk) 00:00, 18 December 2009 (UTC)
  • No one has got around to changing the hyphens yet (although I believe there is a good script that does it). Of course, when they do get changed, they will fall in line with the current (or by then, new) standard.  HWV258.  00:37, 18 December 2009 (UTC)
  • P.S. Nothing in my support for Ozob's proposal shall be construed as support for Chicago's oppressively bready pizza over, say, Sal and Carmine's on Broadway and 101st.—Native New Yorker DCGeist (talk) 05:19, 17 December 2009 (UTC)
    You and I are locked in mortal combat, like Ahab and Moby Dick. I shall vanquish your pizza, even at the cost of my life. Ozob (talk) 12:27, 17 December 2009 (UTC)
Then it's two against one. Think you can handle it, Midwestboy? On a lighter note, I support the proposal. It's far more consistant with the standard practice as shown in the various publishers' and reference style guides examined as part of this discussion. While we are not bound by such, it does fit the spirit of Wikipedia as a recorder of facts, not a maker of one. And, frankly, I fail to see how it could cause any confusion in anything but the most contrived examples, which are covered by the rewrite clause. oknazevad (talk) 19:29, 17 December 2009 (UTC)
  • I support this proposal too. I would also support it if the sentence "Disjunctive en dashes outside prose, such as in tables or infoboxes, may be spaced." were removed. Eubulides (talk) 09:07, 17 December 2009 (UTC)
  • Support the instruction, but the wording is bloated. I might come up with something equivalent but more concise later. ― A. di M. — 2nd Great Wikipedia Dramaout 10:34, 17 December 2009 (UTC)
  • Oppose – everything else aside (I agree with the other opposes and see no point in restating them), it appears contradictory. It states that the ambiguous ‘New York–Chicago’ should be used instead of the more logical ‘New York – Chicago’, but then it explains the problem with using the identically ambiguous ‘New York–London’. – MTC (talk) 12:00, 17 December 2009 (UTC)
    Neither of those are ambiguous in certain contexts. They flew New York–Chicago, for example. I suspect that all truly ambiguous situations are rare and contrived. Ozob (talk) 12:31, 17 December 2009 (UTC)

Alternate solutions for some of the wretchedness

Tony has given examples where an unspaced en dash is inherently ambiguous or requires a double-take. In other examples above, spaced en dashes used where only an en will do become ambiguous, or at least ugly, when used in conjunction with spaced en dashes used as an alternative to em dashes. At least this bit of wretchedness can be solved by deprecating the use of en dashes in place of em dashes. This change would have two additional advantages: (1) it would give Wikipedia another precious little bit of consistency; and (2) spaced en dashes used as em dashes are ugly. In professional publishing, this usage of en dashes is rare.

Also, just because an en dash can be used as a contraction doesn't mean that it always should be. We still have serviceable prepositions in to and through.—Finell 04:41, 17 December 2009 (UTC)

Oppose: first, since readers may interpret a spaced en dash as replacing an em dash regardless of what our manual of style says on the issue, no ambiguity is resolved. The way to resolve ambiguity is to just write it out in words, as in the previous proposal. Second, I believe the current guidelines provides a reasonable amount of flexibility to editors in formatting articles, rather than enshrining the aesthetic opinions of only one group of editors. Christopher Parham (talk) 14:07, 17 December 2009 (UTC)
Support deprecating use of spaced en dashes in place of em dashes. Actual em dashes are easier to read — no double-take is required, whereas I do often observe myself having to do a double-take on this alternative use of en dashes. (Actually, spaced em dashes are even easier to read than unspaced ones — less crabbed — but em dashes of any stripe are much better than the alternative spaced en dash.) --Pi zero (talk) 20:40, 17 December 2009 (UTC)
Actual spaced em-dashes — such as you use above and I am using here — are specifically disallowed by WP:MOS. But I like using them in talk pages. —David Eppstein (talk) 22:51, 17 December 2009 (UTC)
(edit conflict) Yeah, I'd replace the allowance for " – " and "—" with the allowance for " — " and "—". ― A. di M. — 2nd Great Wikipedia Dramaout 22:54, 17 December 2009 (UTC)

A broader perspective

So far, we're just discussing the "Spacing" sub-sub-point; but part of the problem is that the text above is nearly unreadable. I'd propose this:

About half the words of the current point 1. ― A. di M. — 2nd Great Wikipedia Dramaout 16:08, 17 December 2009 (UTC)

If you hide the proposal, I don't think people will read it. There is at least one typo in it: "100 m –10 km" is missing a space after the endash. While this proposal is clearly better than what we have, it insists too much on spaces around endashes: high-quality sources uniformly omit the spaces around the endash in "December 2009–January 2010" (e.g., doi:10.1080/00396330903461682, published by Routledge) and in "100 m–10 km" (e.g., [8], published by Pergamon) and we shouldn't prohibit this practice. I do like the brevity of the proposal, though. How about this rewrite instead?
  1. To indicate disjunction. There are two main roles:
    • To stand for to or through in ranges (1895–1910, 17–22 December). In prose sentences, complicated ranges are often better spelled out, as in The data were gathered from 16 December 2009 to 15 January 2010. Ranges should also be spelled out as needed to avoid confusion, for example if negative numbers or other dashes are involved (−10 to 10).
    • To stand for and, to, or versus between independent elements, as in Canada–New York border, a 3–1 score, blood–brain barrier, Seifert–van Kampen theorem, Michelson–Morley experiment (named after Albert Michelson and Edward Morley); contrast to the hyphenated forms Lennard-Jones potential (named after one individual, John Lennard-Jones) and Sino-Japanese trade (where the prefix Sino- lacks lexical independence).
    Spacing: Disjunctive en dashes are unspaced, except that if either end point of a range contains a space, a spaced en dash may be used if needed to avoid confusion, such as in ranges involving multiple units (100 kW – 40 MW) or dates (16 December 2009 – 15 January 2010).
Eubulides (talk) 20:26, 17 December 2009 (UTC)
Fixed the typo. I prefer "from ... to" rather than "from ... through". Dunno about "uniformly"; the EU style guide suggests "100 kW – 40 MW", but then, it suggests half spaces before percent signs. ― A. di M. — 2nd Great Wikipedia Dramaout 21:17, 17 December 2009 (UTC)
Thanks: the EU style guide is a counterexample for the case when units change. (The NIST style guide says that in this case one should always spell it out, which may be because they didn't want to deal with dueling style guides...) The EU style guide is not a counterexample for date ranges, though: it says unspaced endashes should always be used for those. I've changed my proposal accordingly. Eubulides (talk) 21:48, 17 December 2009 (UTC) PS. I changed "through" to "to" in the examples, and moved some examples into the spaced category to help make that clearer. Eubulides (talk) 00:45, 18 December 2009 (UTC)
"100 m–10 km". OMG, I wish I'd had that example before. It is the very reason the squashed en dash is not recommended. Don't you love the m–10? Really easy to read. It's the very reason no one writes en dashes as interrupters–like this–i.e., because of the spaces in the immediate vicinity. Tony (talk) 03:33, 18 December 2009 (UTC)
Surely that would be "100 m–10 km" ;-) In regard of the choice between "to" or "through" for the free text form, I would observe that using "to" avoids the BrE/AmE bifurcation on "thru". Of course the meanings are different in cases such as dates. "Through"/"thru" imply the end date is included in the range, while "to" is ambiguous on the inclusion.LeadSongDog come howl 15:59, 18 December 2009 (UTC)
Through is never used in UK English to refer to a span of time or dates. Kevin McE (talk) 16:32, 18 December 2009 (UTC)
"Never" is a bit strong. "through 1 January" gets 29 hits on google UK (vice 5 for "thru" and 825,000 for "to") LeadSongDog come howl 17:54, 18 December 2009 (UTC)
Really? I thought nobody does that because they have a different semantics than the unspaced en dashes—there are spaces in the immediate vicinity of this, too—and yet many people (especially in the US, I think) don't put spaces around the em dashes. (Indeed, to me both unspaced em dashes—like this and spaced em dashes — like this seem to be definitely more common than spaced en dashes – like this for "interruptive" dashes.) ― A. di M. — 2nd Great Wikipedia Dramaout 18:59, 18 December 2009 (UTC)

Too hasty

Why was this edit made (with a comment of This version has drawn the least objections, so it seems to be closest to consensus)? WP does not work by forcing changes based on "closest to consensus". As far as I'm concerned, we are still debating whether any change is needed.  HWV258.  04:21, 19 December 2009 (UTC)

  • Agree. There doesn't seem to be consensus one way or another on ANY of the proposals. Ohconfucius ¡digame! 04:25, 19 December 2009 (UTC)
    What the article had before certainly wasn't consensus—that's why we've been talking, after all. The discussion is a bit quieter now than it was a few days ago, and that version seemed to me to best reflect consensus, so I changed it. If you don't like it, WP:BOLD! That's how we got started, and I certainly won't be offended if that's what keeps us going. Ozob (talk) 04:31, 19 December 2009 (UTC)
    This is a style guide, and we should not be to-ing and fro-ing on a style guide. Editors could be reading and digesting this recent change right now. As subsequent changes have happened on the page, a simple revert is not that easy. Ozob: Could you please undo your change until we are all certain that consensus have been reached? Thank you.  HWV258.  04:39, 19 December 2009 (UTC)
    I undid the change for now. There's no rush. Eubulides (talk) 04:43, 19 December 2009 (UTC)
    If you don't like my edit, I would much prefer if you substituted your own preferred version. The present consensus, whatever it may be, is not for the old version. Furthermore, I don't think we can reach consensus if we don't make bold edits. Because I believe in WP:BOLD, I made the MoS reflect consensus as closely as I could. Where it fails it reflects my own failings. I am sure that with the input of others it can be made better, but I do not think that the right way to do that is to remain fixed on a rejected version. Ozob (talk) 04:58, 19 December 2009 (UTC)
    Thanks. I applaud your desire to be bold—but only in article space. In MOS space we need to wait for a real consensus before changing guidelines. We only shoot ourselves in the foot (in terms of being useful) by editing asymptotically.  HWV258.  05:14, 19 December 2009 (UTC)
    Oh, and just for everybody else's reference, here is precisely what I put on the MoS page. It's very slightly different from what Eubulides proposed above:
    1. To indicate disjunction. There are two main uses:
      • To stand for to or through in ranges (1895–1910, 17–22 December). Prepositions such as from and through are then omitted, for example, He served 1939–1941, not He served from 1939–1941. In prose sentences, complicated ranges are often better spelled out, as in The data were gathered from 16 December 2009 to 15 January 2010. Ranges should be spelled out when needed to avoid confusion, for example if negative numbers or other dashes are involved (−10 to 10).
      • To stand for and, to, or versus between independent elements, as in Canada–New York border, a 3–1 score, blood–brain barrier, Michelson–Morley experiment (named after Albert Michelson and Edward Morley); contrast to the hyphenated forms Lennard-Jones potential (named after one individual, John Lennard-Jones) and Sino-Japanese trade (where the prefix Sino- lacks lexical independence).
      Spacing: Disjunctive en dashes are unspaced, except that if either end point of a range contains a space, a spaced en dash may be used if needed to avoid confusion, such as in ranges involving multiple units (100 kW – 40 MW) or dates (16 December 2009 – 10 January 2011).
    Ozob (talk) 05:33, 19 December 2009 (UTC)
    I like this version too. I agree that the current version is badly broken (it's not even internally consistent). I reverted only because I wanted to give other editors more time to comment. Eubulides (talk) 06:00, 19 December 2009 (UTC)
  • WP:TLDR; there is way too much verbiage to read through here about a blooming dash. Why are any changes needed, and what is the nustshell version of the six proposals? SandyGeorgia (Talk) 04:46, 19 December 2009 (UTC)
    Yes, there is, isn't there? Here is my interpretation of events: Someone moved Seifert–van Kampen theorem to Seifert – van Kampen theorem to reflect MoS conventions on spaced en dashes. But spacing en dashes between authors' names is just not done in mathematics, where the Seifert–van Kampen theorem originates. So there were edits to this page, then reverts, and then the present discussion. The major points as I see them are beauty and clarity. We agree that there is a lot of bad en dash typography out there, but we don't agree what looks best. We agree that there are a lot of ways to confuse readers with dashes, but we don't agree what is least confusing. There were some compromise edits early on, but our progress has slowed.
    As I see it, there are three ways we can go:
    1. We can keep the old practice, which mandates spaces when one of the items being joined is spaced. Many articles (especially, but not always, FAs) already do this.
    2. We can reject the old practice and mandate no spaces in prose. This is what all print style guides require.
    3. We can space some things and not others, depending on what we think is the most clear and beautiful. This is what the MoS says presently.
    I have come to prefer unspaced en dashes in prose, but I would accept a certain amount of spacing. Others prefer spaced en dashes, but would accept a certain amount of unspacing. The discussion has gotten just a tiny bit quieter in the past few days, so I figured it was time to try editing the MoS again to see what people would think. Ozob (talk) 05:22, 19 December 2009 (UTC)
    • To try to answer SandyGeorgia's question more briefly: changes are needed because MOS:ENDASH currently requires spaces around endashes in phrases where high-quality print sources invariably omit the spaces, e.g., "the Chicago–New York route", "Seifert–van Kampen theorem", "von Hippel–Lindau disease". Eubulides (talk) 06:00, 19 December 2009 (UTC)
      • Part of this has been addressed. "...where high-quality print sources..."—we do not make changes to WP MOS pages simply because of what happens off-WP. We can take other points-of-view into account in our debate, but our consensus-driven process will (as always) decide the guidelines before changes are applied.  HWV258.  06:35, 19 December 2009 (UTC)

←There is clearly no consensus to change the status quo. Thank you, Eubulides, for reverting that change. Meanwhile, User:Noetica – who has been too busy to edit for months – emailed me to say he's been watching this thread with concern. He writes [edited by me for wikiformatting]:

These sources may be of some value:

A) Butcher's Copy-editing (4th edition 2006). This classic work is one of the most respected British guides. See the relevant page online (pp. 151–53). My commentary follows, drawing on salient points:

  1. Spaced en dashes (as opposed to spaced or unspaced em dashes) are now "most often used" for so-called parenthetical dashes.
  2. En dashes are also quite properly used to mean "and" or "to", in which case they are normally unspaced.
  3. On p. 152: "However, spaced en rules [en dashes] may be used between groups of numbers and words to avoid implying a closer relationship between the words or numbers next to the en rule than between each of these and the rest of its group." Three quite decisive examples follow, along with a caution that in no way detracts from the basic principle. A search for "en rule" in this work at Googlebooks confirms its robustness. See for example p. 131 and p. 246, where both the principle and the obvious caution are reiterated.

B) The Cambridge guide to English usage (Pam Peters, 2004). On p. 140: "A spaced en dash/rule is used when the words or numbers to be separated have internal spaces.

1 July 1991 – 2 June 1992"

This is essentially the same provisions as in source A (along with additional ones of interest), but more prescriptive. And there is NO restriction to dates; and there is NO provision for any alternative practice.

C) Texas State University's editorial style guide link. This is one of several academic sources online that prefer the general style given in sources above, though perhaps implicitly: "The event runs October 10–15. 6 a.m. – 9 a.m. (include a space before and after the hyphen or en dash in ranges of times)." This is one of several American sources in accord with the other sources cited.

D) The Cambridge guide to Australian English usage (Pam Peters, 2nd edition 2007). See pp. 155–56: Same wording as in source B.

E) Style manual: for authors, editors and printers (Wiley, 6th edition 2006). Probably the major Australian style guide; widely followed, especially by government publications: essentially the same ruling as above. For its prominence in Australia see Style_manual#Australia.

F) The Australian editing handbook (Elizabeth Flann and Beryl Hill, 2004). Same ruling as in source E and others.

There are others that I can't chase right now!

Finally, a nice example of practice from "established publishers". Spot the four ways of doing date ranges, in one table.

Tony (talk) 06:17, 19 December 2009 (UTC)

Thanks Tony1. In the light of this, I would like to ask Ozob where the justification for summary point 2 above ("We can reject the old practice and mandate no spaces in prose. This is what all print style guides require") originated? (underlining is my addition.)  HWV258.  06:42, 19 December 2009 (UTC)
Prior to this, nobody had found a style guide recommending a spaced en dash. Seemed just to me. Ozob (talk) 15:29, 19 December 2009 (UTC)
  • Thanks for those examples Tony. Obviously we have dueling style guides, since Hart's uses unspaced en dashes exclusively. Butcher's explicitly represents the compromise embodied in my draft (in the #A broader perspective subsection) and in Ozob's (earlier in this subsection). The other cited sources (most of which seem to be clones of Pam Peters) are also consistent with the compromise, in that they require spaces around endashes in ranges whose items contain spacies, and are silent about endashes in other uses (the examples given are all ranges, and it's not at all clear that Peters's rule was intended to apply to examples like "the Chicago–New York route"); the draft compromise allows this style. In practice, as we've seen, high-quality academic sources such as the OUP, Elsevier, and Springer seem to follow Hart's, which the draft compromise also allows. Eubulides (talk) 06:59, 19 December 2009 (UTC)
  • Interesting. "Groups of numbers and words". Now that I think about that, a somewhat significant difference between "Born–von Karman boundary condition boundary conditions" and "19 June – 24 September" is that in one the elements separated by the dash are all-alphabetical, and in the other they are alphanumerical. I hadn't noticed that before. ― A. di M. — 2nd Great Wikipedia Dramaout 12:53, 19 December 2009 (UTC)
  • And I don't understand what the "nice example" is supposed to show: they are all the same format, modulo typos. ― A. di M. — 2nd Great Wikipedia Dramaout 14:41, 19 December 2009 (UTC)

←I've received another email from User:Noetica, who is happy to provide information but feels that if he posts here directly he'll become too involved in the whole MoS thing again; he says he'll be in a better position to contribute directly towards the end of 2010, and for the moment it looks as though he's going to channel through me. BTW, I've heard a rumour that Noetica has just won a prestigious prize for his translations from mediaeval French and Latin; I'm trying to google the details.

In response to Eubulides's comment above – "Obviously we have dueling style guides, since Hart's uses unspaced en dashes exclusively.", Noetica says the following (which I quote, again with minor tweaks for wiki-formatting): They do not "duel". In fact they have great respect for each other. See the many references to New Hart's in Butcher's, for example this. Quite simply, New Hart's does not address certain subtleties that Butcher's addresses. Butcher's also pays respect to Peters' Cambridge Guide. And Peters, in turn, covers further issues with the en dash on which other sources (including Butcher's, New Hart's, and the Wikipedia MOS itself) are silent. MOS editors might give consideration to such complementarity: silence may mean nothing more than neglect, or perhaps a need for clarity and brevity; editors everywhere have to fill the inevitable gaps intelligently and flexibly. Edits to the MOS must adapt to the utterly unique Wikipedia environment, which is certainly distinct from the world of Elsevier and Springer.

Eubulides stated that "Butcher's explicitly represents the compromise embodied in my draft ...". Noetica: That is doubtful. Butcher's suggests solutions that "may be better" in certain cases, but those cases do not track yours closely. The general notice at the head of MOS already calls for common sense in applying it, and allows that there will be occasional exceptions. (Same for any style guide, of course!) Note the modifier "occasional".

"The other cited sources (most of which seem to be clones of Pam Peters) are also consistent with the compromise, in that they require spaces around endashes in ranges whose items contain spaces, and are silent about endashes in other uses (the examples given are all ranges, and it's not at all clear that Peters's rule was intended to apply to examples like 'the Chicago–New York route'); the draft compromise allows this style." Noetica: I do not believe this is correct. Peters herself says this: "A spaced en dash/rule is used when the words or numbers to be separated have internal spaces." That is unequivocal. It is irrelevant that her example happens to be a range.

"In practice, as we've seen, high-quality academic sources such as the OUP, Elsevier, and Springer seem to follow Hart's, which the draft compromise also allows." Noetica: I think we have seen no such thing: you have declared it, that's all. In fact those houses are not consistent, and do not always circumscribe their authors' choices. Would you, Eubulides, like to furnish us with Elsevier's definitive ruling on en dashes? I couldn't find it. Tony has edited for OUP; Springer's guidelines, with which I [Noetica] have worked, include nothing on these issues. Springer's published guide for authors itself uses dashes anomalously in citing a piece from this Springer book. The guide uses a spaced em dash (sic) between the title and subtitle: "Software engineering — from auxiliaries to key technologies". Those publishers are no paragons!

(Incidentally, for our sceptics on another issue: Springer uses spaced en dashes at the sentence level, not em dashes. In this they join Penguin, Routledge, Cambridge UP, and several other major houses.)

per Tony (talk) 15:20, 19 December 2009 (UTC)

I feel like we have made great progress now: Prior to this, all the style guides referenced on this page specified unspaced en dashes exclusively. Now we have contrary examples! We agree that we need to choose a dashing style that is consistent, beautiful, and clear. But as we can see now, there are several incompatible styles currently in use:
  1. No spaces: ACS, APA, CMOS, Hart–Oxford, MHRA
  2. Some spaces: Butcher, Flann and Hill, Peters, Style manual
  3. Unknown: AP Stylebook, MLA, ISO 690, Bringhurst
Opinion among those present is divided. I would prefer to sway the editors here my way or to be swayed the other way myself, but that seems unlikely at present. It may be that Noetica's correspondence and SandyGeorgia's presence moves some of us, but if not, I think we should go to the Village Pump and ask for other views.
I am in full agreement that we should continue to allow parenthetical spaced en dashes. And congratulations to Noetica! Ozob (talk) 16:23, 19 December 2009 (UTC)

Butcher's advice

Butcher's advice sounds like the most reasonable one to me. ― A. di M. — 2nd Great Wikipedia Dramaout 19:08, 19 December 2009 (UTC)

I agree that Butcher's advice is the closest approximation to what we do at Wikipedia. Following up on Noetica's remarks about style guides:
  • "They do not 'duel'." In that case, since Hart's explicitly gives "New York–New Jersey–Connecticut area" as a good example, and Peters explicitly says "A spaced en dash/rule is used when the words or numbers to be separated have internal spaces", we can therefore conclude that Peters is not talking about the same thing that Hart's is talking about, right?
[I really don't want to be here, but I also don't want to burden Tony with the task of channeling me. Later this week I'll have time to make a systematic submission about these dashes; but for now, just some quick responses on these particular points. On the road in the meantime: so I will not be able to make any rejoinders for a couple of days, alas.–Noetica] On Eubulides' point immediately above: Right? No, wrong. This could be read as a captious attempt to trade on accidental features of the expression in Hart's and Peters' guidelines. Hart's, yes: it wants no spaces. But you cannot presume from this anything about Peters' ruling. If she had expressed things meticulously so that the minutest detail was spelt out, her book would be unsellable. Her readership is ordinary authors and editors, not career quibble-mongers like us. Yes, I quibble with her myself, since elsewhere she commits glaring and substantial errors of fact. I have to email her about a couple of these. But her judgement is entirely rational here; and the general intent is quite plain. –¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
So you contend that Hart's and Peters' disagree on this point? That is what I meant by saying that we have "dueling style guides". Eubulides (talk) 19:13, 21 December 2009 (UTC)
  • "Butcher's suggests solutions that "may be better" in certain cases, but those cases do not track yours closely." I don't see why not. All of the examples where Butcher says it's OK to use spaces (p. 151) are ranges: "6.6 – 7.8", "18 September – 19 January", "c. 1215 – c. 1260"; this closely mimics the examples in the proposed wording, all of which are ranges: "100 kW – 40 MW", "16 December 2009 – 15 January 2010". And Butcher's suggestion to use these optional spaces "cautiously" closely matches the proposed text advice to use spaced rules only "if needed". I'd be perfectly happy to replace the proposed "if needed" with "cautiously" if the discrepancy between those two phrases is what's needed.
Good. Everyone acknowledges the need for caution; we can agree that this is a better way to put it. "When needed" suggests that editors can generally determine when a need arises. But not all can! All can be cautious, though; and all can consult colleagues, or re-fashion a sentence. Caution and flexibility are always required, not just here. Once more, that's covered in the header at the top of MOS. Meanwhile, we need simple, durable, and readable guidelines. –¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
Since the proposed wording is considerably shorter and simpler than what's in there now, it's a considerable improvement on the "simple" and "readable" front. I agree that style guides should not be changed on a whim, but this isn't a whim we're talking about here. Eubulides (talk) 19:13, 21 December 2009 (UTC)
  • "The general notice at the head of MOS already calls for common sense in applying it, and allows that there will be occasional exceptions." True, but the problem is that the current style guide says that endashes between proper names are "occasional exceptions". But this is incorrect. It's not an exception: it's common practice. High-quality publishers routinely omit spaces in such cases, and none of the style guides mentioned say anything about this particular point, except for Hart's which uses an example with unspaced en dashes.
The general and unequivocal nature of Peters' rule (see next point) shows that you miss the mark, in your last sentence here. The rest I don't quite follow. –¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
I'll follow up on the next point below. Let me try to rephrase the other: if the style guide says you must do X, and actual articles and high-quality publishers and style guides often do the opposite of X, then we can't rely on the "occasional exception" provision: the MoS needs to change. Eubulides (talk) 19:13, 21 December 2009 (UTC)
  • 'Peters herself says this: "A spaced en dash/rule is used when the words or numbers to be separated have internal spaces." That is unequivocal. It is irrelevant that her example happens to be a range.' No, it is quite relevant that her example is a range: it suggests that the rule is intended to apply to ranges, and is not intended to apply in other circumstances. The example in Hart's, which does not involve a range, is unspaced. The only way I see to resolve this without saying that there is a "duel" between Peters and Hart, is to say that Peters is talking about ranges and Hart is talking about other disjuncts.
Mere unsupported assertion. Peters is quite explicit in her rule, and her choice of example does not remove its generality. "It suggests" X only to those with a predisposition in favour of X; others are satisfied with the plain ruling as it stands: "A spaced en dash/rule is used when the words or numbers to be separated have internal spaces." –¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
We also need to see in what context she gives that rule, which most of us can't (the book is not on Google preview). For example, if it were in a subsection titled "Ranges", ... ― A. di M. — 2nd Great Wikipedia Dramaout 12:12, 21 December 2009 (UTC)
The context of her rule is not clear. It's not in a "Ranges" subsection. At any rate, "her choice of example does not remove its generality" is merely a repated argument for a broad interpretation of her rule, an interpretation that is undercut by her choice of a range for its example. Eubulides (talk) 19:13, 21 December 2009 (UTC)
  • '"In practice, as we've seen, high-quality academic sources such as the OUP, Elsevier, and Springer seem to follow Hart's, which the draft compromise also allows." Noetica: I think we have seen no such thing: you have declared it, that's all.'" No, I've cited several examples. I have systematically searched Google Scholar for these examples, and have found no counterexamples for the phrases I cited. I do not have access to Elsevier's style guide, and don't need it: all I need to do is to see what Elsevier journals publish. Similarly for OUP, Springer, and the rest. Even if there is nothing formal in their style guides, we can look at and rely upon what they actually do.
If you were right, we might accept also Springer's and Elsevier's inconsistencies as par for the course and nothing to worry about, right? We agree: these are not exemplars without blemish; and we should also doubt that they give the present question as fastidious attention as we do, here at MOS. But the situation is worse than you think. Follow Springer, you say? How about this, from a Springer publication:

... the London – New York route was especially critical ...

¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
Sorry, I cannot read that example; Google Books previews are unreliable and it doesn't display for me. From its title page, that source is a conference proceedings; these are typically edited to a lower standard than scholarly journals are, often simply by taking authors' manuscripts verbatim, and so the source is not the best example of high-quality scholarly publishing. I agree that a large publishing house such as Springer is not uniform, and I expect that one can find counterexamples to the general rule. But the general trend in academic publshing is quite clear. I just now searched for "London New York route", "London New York flight", "New York London route", and "New York London flight" in Google Scholar (with quotes), to find all examples publshed in scholarly journals that used endashes. Here is a list of everything I found:
Unspaced endash
Spaced endash
None.
You're free to repeat the queries. Eubulides (talk) 19:13, 21 December 2009 (UTC)
  • "Those publishers are no paragons!" You're quite correct. But they are high-quality academic publishers, among the best in the business. I see no reason to prohibit the style they use.
But which of the styles they use? In answering your last point I have demonstrated that Springer sometimes uses the style established at our MOS, not the one you assert they use. But even if Springer, OUP, and Elsevier were consistent with their style choices in the vexed matter of ranges and "disjunctive" dashes, it is by no means clear that we ought to follow them. Once more (since I have had to make the point far too often in these discussions), the crux is this: Wikipedia is unique. It confronts weighty problems of pan-anglophone, collaborative, dynamic online publishing that never intrude on the serene world of academic journals. The web is not paper, and very few Wikipedia contributors are professional editors; very many are not even experienced writers. No appeal to New Hart's, Chicago, Butcher's, or Elsevier practice is final. We have to fashion guidelines ourselves, for an entirely new situation. We must respect precedents, yes; but many precedents are vague, rashly conceived, or scarcely applicable in new contexts. We at MOS must above all respect the special needs of Wikipedia editors, if we are ultimately to serve the readership. That means no hasty or half-considered changes, which yield nothing but chaos and dismay.
Of course we can improve things; but take it slowly! Even Chicago makes changes to its guidelines – over years, not weeks or months. Think very carefully before assuming that the latest contrivance is necessarily the best. Stability is vital, if MOS to serve the community as it ought to. The present guidelines for en dashes have served well, and they have now been adjusted in one small detail: a change that everyone accepts. I urge that we keep things as they are, for now. I'll have more to say in a couple of days on some overarching issues for MOS and its associated pages. Until these issues are addressed, most of this fiddling is a sheer waste of time.–¡ɐɔıʇǝoNoetica!T– 12:02, 21 December 2009 (UTC)
  • "I have demonstrated that Springer sometimes uses the style" Only in a conference proceedings. You haven't demonstrated they use the style in their higher-quality publications. On the contrary, for that particular example scholarly journals seem to agree on unspaced endash, as I show above.
  • "Wikipedia is unique". That is not an argument for requiring a style that is used (as far as we can see) only by lower-quality sources.
  • "Think very carefully before assuming that the latest contrivance is necessarily the best." I don't see how this point bears on the discussion. For all we know, it's spaced endashes that are "the latest contrivance".
Eubulides (talk) 19:13, 21 December 2009 (UTC)
Eubulides (talk) 20:51, 19 December 2009 (UTC)

Re Butcher's remark on parenthetical dashes: The use of spaced en dashes as an alternative to em dashes requires more effort to read than using actual em dashes for em dashes. I'd expect that to be true even in localities where spaced en dashes are "now most often used" for parenthetical dashes: em dashes are instantly recognizable, exactly because they aren't as overloaded as en dashes. Effective communication should trump "now more often used" for our purposes here, even if one were to accept the "now more often used" claim — although, incidentally, I have doubts about the geographical (not to mention internet) distribution of that claim; it sounds like it may be anglocentric.

Spaced em dashes are, as I remarked earlier, even easier on the eye than unspaced, but even unspaced em dashes are a vast improvement over em-dash-wannabe spaced en dashes. --Pi zero (talk) 21:21, 19 December 2009 (UTC)

I have to really, really strenuously [imagine sweat and grunting] disagree with the notion that "foo—bar" is somehow easier to read/parse than "foo – bar". It requires pretty good vision (a concern many of you 20-year-olds don't think about) to tell the difference between "foo—bar" and "foo-bar" if you're a fast reader (and essentially impossible in some fonts). Un-spaced en-dashes are even worse, of course: "foo–bar" vs. "foo-bar". However, Pi zero, I have to say that "foo — bar" is overkill. — SMcCandlish [talk] [cont] ‹(-¿-)› 10:13, 20 December 2009 (UTC)
Don't run in the hall with those assumptions, you might do yourself an injury :-). 20 years ago I might have had less preference for spaced over unspaced, since my eyesight was better then.
So your personal preference is against spaced em dashes. But do you really feel that they damage the project, so as to justify the MOS proscribing them? 'Cause I think proscribing them damages the project. At the time that that proscription was put into the MOS, it was remarked that spaced em dashes were about twice as common on WP as unspaced em dashes, which suggests to me that more editors thought the spaced variety looked better. --Pi zero (talk) 15:04, 20 December 2009 (UTC)
A practice I've read about and seen somewhere is half-spaced em dasheslike this. ― A. di M. — 2nd Great Wikipedia Dramaout 11:29, 20 December 2009 (UTC)
Interesting idea, but your actual code is coming out on my display looking like full spaces.
foobar   ({{gaps|foo|—|bar}})
foo — bar   (foo — bar)
--Pi zero (talk) 15:04, 20 December 2009 (UTC)
On my display, they are narrower, but by just one pixel each. I guess they could be made even narrowerlike thisor maybe even like this. ― A. di M. — 2nd Great Wikipedia Dramaout 16:31, 20 December 2009 (UTC)
Instruction creep alert! Remember most editors don't know when or how to enter a dash, never mind a half space. Art LaPella (talk) 21:45, 20 December 2009 (UTC)
Nothing would forbid creating a template expanding to <span style="margin:0.15em">—</span> and suggesting it in the MOS, in the unlikely event that consensus emerged that half-spaced em dashes are acceptable. ― A. di M. — 2nd Great Wikipedia Dramaout 21:51, 20 December 2009 (UTC)

Something we can agree on?

So, we don't agree on spacing. That's fine; we'll work it out eventually. But I want to ask if we agree that the other text under point one is too long. Right now it's four bullet points, but I think we can get away with half that. A. di M., Eubulides, and myself have all made attempts at epitomizing it. If one of these is acceptable, then I'd like to go ahead and replace it (without changing the part about spacing). My first proposal is that we use what I proposed above. This, together with the spacing rule currently on the page, would give us:

  1. To indicate disjunction. There are two main uses:
    • To stand for to or through in ranges (1895–1910, 17–22 December). Prepositions such as from and through are then omitted, for example, He served 1939–1941, not He served from 1939–1941. In prose sentences, complicated ranges are often better spelled out, as in The data were gathered from 16 December 2009 to 15 January 2010. Ranges should be spelled out when needed to avoid confusion, for example if negative numbers or other dashes are involved (−10 to 10).
    • To stand for and, to, or versus between independent elements, as in Canada–New York border, a 3–1 score, blood–brain barrier, Michelson–Morley experiment (named after Albert Michelson and Edward Morley); contrast to the hyphenated forms Lennard-Jones potential (named after one individual, John Lennard-Jones) and Sino-Japanese trade (where the prefix Sino- lacks lexical independence).
      • Spacing: Disjunctive en dashes are unspaced, except when there is a space within either one or both of the items (the New York – Sydney flight; the New Zealand – South Africa grand final; June 3, 1888 – August 18, 1940, but June–August 1940). Exceptions are occasionally made where the item involves a spaced surname (Seifert–van Kampen theorem).

Would anyone object to this? I think the conciseness is a great improvement. Ozob (talk) 22:45, 20 December 2009 (UTC)

Let's keep with what you proposed earlier. Nobody has specifically objected to it. Eubulides (talk) 23:10, 20 December 2009 (UTC)
You simply do not have consensus to change the current text. That is plain. Tony (talk) 04:13, 21 December 2009 (UTC)
There is certainly not universal agreement on any version of the text. However, it is clear that the version currently installed is controversial and cannot be said to have consensus in any reasonable way. We should fix it in a way that has better consensus, and Ozob's proposal in #Too hasty is the best we have so far. Specific suggestions for improving it, to better reflect a broad consensus, are welcome. (If I had my druthers I'd change it to omit all unspaced endashes, but obviously that would lessen consensus so I'm not about to propose that.) Eubulides (talk) 04:23, 21 December 2009 (UTC)
The status quo gained consensus originally, which was some time ago. You need to establish consensus to change it. You have not done so: many editors have said they do not like the idea of allowing squashed en dashes willy-nilly. Tony (talk) 05:40, 21 December 2009 (UTC)
I'm afraid this is an inaccurate characterization of Ozob's compromise proposal in #Too hasty, and of other editors' comments. The compromise proposal does not allow squashed endashes "willy nilly", and other editors have not taken the extreme position that no change can be made to the existing text, or that the compromise is unacceptable. The consensus has clearly changed. Again, specific suggestions for improving the proposed wording to reflect consensus are welcome. Eubulides (talk) 06:04, 21 December 2009 (UTC)

Look, the change I'm suggesting right now has nothing to do with spacing. It has to do with tightening up the other text. I would normally not hesitate about making a change like this, but I posted here because the en dash section has become very sensitive. I'm glad I did, too.

Now, back to the text above: Do we have consensus to revise the four bullets under point one so that the entire point reads as above? Let me reiterate that this would not change the instructions on spacing. That is a separate issue. Ozob (talk) 08:43, 21 December 2009 (UTC)

I'm not sure the two issues are so easily disentangled, but I suppose we could give it a try. In looking over the proposal again, I see that it adds the following sentence, which was not in my earlier proposal (in #A broader perspective) and is not in the MoS now: "Prepositions such as from and through are then omitted, for example, He served 1939–1941, not He served from 1939–1941." I suggest removing this sentence, as "He served 1939–1941" looks a bit off and would be better written as "He served from 1939 to 1941." Eubulides (talk) 09:09, 21 December 2009 (UTC)
That sentence corresponds to the present instruction, "This is also the case when the nearby wording demands it, for example, he served from 1939 to 1941 and not he served from 1939–1941, in which from and to are complementary and should both be spelled out; similarly, between 1939 and 1941, not between 1939–1941." If you can think of a better example to put there we could substitute it. Ozob (talk) 09:25, 21 December 2009 (UTC)
I propose: To stand for to or through in ranges, especially in tables, parentheses, and the like (1895–1910, 17–22 December). (In prose sentences, ranges are often better spelled out with prepositions, e.g. from 1895 to 1910.) Do not mix prepositions and dashes (do not use from 1895–1910). Avoid using dashes for ranges when they could cause confusion, for example if negative numbers or other dashes are involved (use −10 to 10, not −10–10). ― A. di M. — 2nd Great Wikipedia Dramaout 09:59, 21 December 2009 (UTC)
This seems to be pretty OK (like the one similar to this). Although the "He served 1939–1941" example needs to be thrown out and something saner picked instead. It seems essentially to be status quo + allowing for "Seifert–von Kampen". Which is much better than the status quo. Headbomb {ταλκκοντριβς – WP Physics} 16:40, 21 December 2009 (UTC)
What about It was estimated that 600,000–800,000 people attended the festival, or something like that? ― A. di M. — 2nd Great Wikipedia Dramaout 17:11, 21 December 2009 (UTC)
Or maybe the winter of 2009–2010? Ozob (talk) 18:22, 21 December 2009 (UTC)
The above rewordings are OK, but they still make the text too long. There should be no need for parenthesized sentences here. I suggest the following rewrite of the first bullet instead:
  • To stand for to or through in ranges (1895–1910, 17–22 December, the 2009–2010 season). In prose sentences, ranges should not be mixed with prepositions (from 1895–1910) and are often better spelled out (from 1895 to 1910). Ranges should also be spelled out as needed to avoid confusion, for example when negative numbers or other dashes are involved (−10 to 10).
Eubulides (talk) 19:13, 21 December 2009 (UTC)
Usually, periods of less than 12 months spanning two different calendar years, such as financial years and the like, use a slash (2009/10); I'd do the same with "winter". BTW, Eubulides' version sounds fine to me (except that I'd explicitly add "avoid" before the red example, for greater clarity for colour-blind readers). ― A. di M. — 2nd Great Wikipedia Dramaout 21:06, 21 December 2009 (UTC)

OK, how about this for the bullets:

  • To stand for to or through in ranges (1895–1910, 17–22 December). Ranges should not occur with prepositions (There were 3,000–3,500 casualties, not There were from 3,000–3,500 casualties), and in sentences they are often better spelled out (There were between 3,000 and 3,500 casualties). Ranges should be spelled out if needed to avoid confusion, for example when negative numbers or other dashes are involved (−10 to 10).
  • To stand for and, to, or versus between independent elements, as in Canada–New York border, a 3–1 score, blood–brain barrier, Michelson–Morley experiment (named after Albert Michelson and Edward Morley). An en dash is not used for a hyphenated name such as Lennard-Jones potential, named after one individual, John Lennard-Jones, or an element that lacks lexical independence, such as the prefix Sino- in Sino-Japanese trade.

Ozob (talk) 04:44, 22 December 2009 (UTC)

Phrases like "There were 3,000–3,500 casualties" are acceptable but not the best, the style guide shouldn't be recommending them as good examples. Nor should the style guide recommend "There were between 3,000 and 3,500 casualties"; that's too wordy, and it should be "There were 3,000 to 3,500 casualties". And how did we get on the depressing (and lengthy) subject of casualties? Why not just stick with the shorter version? I don't see how the extra words have helped make things any clearer. Eubulides (talk) 18:10, 22 December 2009 (UTC)
Yeah, I suppose it's not the best example. The depressing subject of casualties came up because I was looking for a range that might appear naturally somewhere. As far as the comparison with your version, I was a little hesitant about the phrase "prose sentences" (after all, we don't usually write our articles in blank verse, but I suppose it's possible), I thought "mixed" was a little more vague than "with", and I thought "if needed" was a little more prescriptive (in a good way) than "as needed". What about:
  • To stand for to or through in ranges (1895–1910, 17–22 December). Ranges should not occur with prepositions (3,000–3,500 people, not from 3,000–3,500 people), and in sentences they are often better spelled out (3,000 to 3,500 people). Ranges should be spelled out if needed to avoid confusion, for example when negative numbers or other dashes are involved (−10 to 10).
  • To stand for and, to, or versus between independent elements, as in Canada–New York border, a 3–1 score, blood–brain barrier, Michelson–Morley experiment (named after Albert Michelson and Edward Morley). An en dash is not used for a hyphenated name such as Lennard-Jones potential, named after one individual, John Lennard-Jones, or an element that lacks lexical independence, such as the prefix Sino- in Sino-Japanese trade.
Also, note that the second point is not the same as my original proposal (I removed most of the parentheses). I don't know if you had an opinion on it or not.
Finally, I'm really hoping to get consensus on this, so I'd like to know: Tony and Noetica, what would you think of these changes? Ozob (talk) 18:24, 22 December 2009 (UTC) UPDATE: Change "between 3,000 and 3,500 people" to "3,000 to 3,500 people" as per Eubulides's suggestion below. Ozob (talk) 06:22, 23 December 2009 (UTC)
The 2nd point is OK; I have no preference between the older and newer versions. There's one example in the 1st point that still needs improvement, though: "between 3,000 and 3,500 people", though acceptable, is often suboptimal style, and "3,000 to 3,500 people" is typically better (it's certainly shorter). Eubulides (talk) 19:52, 22 December 2009 (UTC)
OK, I made that change above. Ozob (talk) 06:22, 23 December 2009 (UTC)
We don't write blank verse, but we do write tables, infoboxes, and other such kinds of non-prose text. ― A. di M. — 2nd Great Wikipedia Dramaout 12:40, 23 December 2009 (UTC)

Comments:

  • "To stand for ..." is better than the slightly different wording in the first three bullets.
  • Why omit "May–November"? It serves to illustrate that it's not only numerical items that require an en dash when ranged. Since the four examples at the top are all commonly misused, I suggest retaining them.
  • "He served 1939–1941" is strictly North American. In other varieties, the "from", and thus the spelled-out "to" are necessary. It would be better not to use this example. Therefore, the "In prose sentences, complicated ranges are often better spelled out, as in The data were gathered from 16 December 2009 to 15 January 2010" is probably unnecessary, especially as "often" is unclear. I'd say "often" it's less cluttered using an en dash when the items are long.
  • Why not retain this: "when a number range involves a negative value or might be misconstrued as a subtraction (−3 to 1, not −3–1)." It's so clear.
  • "Canada–New York border"—no, the en dash is spaced, as required by the "Spacing" section. In any case, can and, to or versus go there comfortably? I see it's in the existing text, too, but I'm unsure it works; same for "blood and brain barrier.
  • Do we have to link to those articles? Who would divert to consult them right here? The blue link colour fuzzes up the use of green and red. We could link to other items that are green if we're to take it to the extreme.
  • The existing second bullet is clearer explicitly mentioning "hyphen"; actually, it's better all round, except for your suggested opening.

BTW, I have to go change the succinct version of the MoS if anything changes here. In fact, when I produced that page, I found little I could save from the en dash text, even though overall the word count is about 40% of this bloated Manual. Tony (talk) 13:02, 23 December 2009 (UTC)

I hope that when we're done with this, the long version and the short version will be the same:
  • To stand for to or through in ranges (pp. 211–19, 64–75%, the 1939–45 war, 17 May – 22 November). Ranges should not occur with prepositions (3,000–3,500 people or 3,000 to 3,500 people, not from 3,000–3,500 people). Number ranges must be spelled out if they involve a negative value or might be misconstrued as a subtraction (−10 to 10, not −10–10)).
  • To stand for to or versus (male–female ratio, 4–3 win, Lincoln–Douglas debate, France–Germany border).
  • To stand for and between independent elements (diode–transistor logic, Antiqua–Fraktur dispute, Michelson–Morley experiment). An en dash is not used for a hyphenated name (such as John Lennard-Jones: Lennard-Jones potential, not Lennard–Jones potential) or an element that lacks lexical independence (such as the prefix Sino-: Sino-Japanese trade, not Sino–Japanese trade).
I noticed that the article blood-brain barrier has en dashes only in the page ranges in its citations; not once is "blood-brain barrier" en dashed. I'm not sure what's going on there, because "blood-brain" doesn't look like a compound adjective to me. Instead of leaving in something doubtful I substituted "Antiqua–Fraktur dispute".
Regarding borders, compare the following similar examples:
French-German border
France–Germany border
I'd interpret the first as a compound adjective, and therefore I'd hyphenate it. The second isn't, or at least I don't think it is, because neither "France" nor "Germany" is an adjective. The only way I can think to read it is "France to Germany border", in which case it ought to be where I put it under the second bullet point. Similarly, I think in "Canada – New York border" the dash stands for "to"; the corresponding compound adjective would be "Canadian-New York border", which is confusing because "New York" still looks like a noun. Ozob (talk) 18:08, 23 December 2009 (UTC)
I wonder why an en-dash joined item can't be a double adjective. It doesn't matter whether it's a noun or an adjective, does it? On "French–German border", I'm so used to "Franco-" that it jars a little, although it's not incorrect. Wouldn't "German–Polish" border be better, since neither particle has a common short form? Tony (talk) 07:34, 24 December 2009 (UTC)
  • "Blood-brain barrier" is a special case, as common medical practice is to use "blood" as an adjectival combiner ("blood-retinal barrier" is considerably more popular than "blood-retina barrier"), and so it's not clear whether a hyphen or an endash is more appropriate here. Some high-quality sources (e.g., doi:10.1038/nrd2368, Nature) use an endash, some (e.g., doi:10.1002/jps.21580, Wiley) use a hyphen; I don't see any overall preference among high-quality sources, and the MoS shouldn't express a preference either.
  • A typo: "))" should be ")".
  • Please remove "Antiqua–Fraktur dispute". It's in the wrong section (it's a versus, not an and) and we have plenty of examples already.
  • Please remove "17 May – 22 November" for the same reason you removed "Canada–New York border". The issue of spaced endashes should be discussed in the spacing paragraph, not here.
  • The last sentence is unnecessarily long. There's no need to give versions with both hyphens and endashes for the same thing. I suggest replacing:
    (such as John Lennard-Jones: Lennard-Jones potential, not Lennard–Jones potential)
    with:
    (Lennard-Jones potential, named after John Lennard-Jones)
    and replacing:
    (such as the prefix Sino-: Sino-Japanese trade, not Sino–Japanese trade)
    with
    (the prefix Sino- in Sino-Japanese trade)
Eubulides (talk) 23:58, 23 December 2009 (UTC)
Re. compound adjectives, I thought that was one of the main usages of hyphens, isn't it? ― A. di M. — 2nd Great Wikipedia Dramaout 16:23, 24 December 2009 (UTC)

I don't agree that the full date should be removed from the bullet. Let's remove all examples from the bullets, then, since the "Spacing" para provides the basis for the spacing of all examples, spaced or unspaced. In fact, the full date is very useful as a prominent example, because such constructions are almost ubiquitous on WP. The "Canadian–New York" example is a problem because it conflicts with the spacing rules.

Frankly, I see no reason to expend all of this time and energy in completely revamping the en dash bullets. Although I generally think the MoS provides too many examples to make its points, here, the examples seem to be a good idea, since editors need particular guidance WRT this usage. I do agree that the openings should be the same, as Ozob renders them. Tony (talk) 04:06, 24 December 2009 (UTC)

OK, it sounds like "blood-brain barrier" is inappropriate. I generally prefer to put in more examples rather than less, but I erred on Antiqua–Fraktur dispute, and I've removed it entirely.
I agree with Tony that we should have at least one example of a date that involves months or days, not just years. I realize that this intersects with our unfinished spacing discussion, but there's no reason why we can't change the spacing later. So I'd like to leave 17 May – 22 November. Now we have:
  • To stand for to or through in ranges (pp. 211–19, 64–75%, the 1939–45 war, 17 May – 22 November). Ranges should not occur with prepositions (3,000–3,500 people or 3,000 to 3,500 people, not from 3,000–3,500 people). Number ranges must be spelled out if they involve a negative value or might be misconstrued as a subtraction (−10 to 10, not −10–10).
  • To stand for to or versus (male–female ratio, 4–3 win, Lincoln–Douglas debate, France–Germany border).
  • To stand for and between independent elements (diode–transistor logic, Michelson–Morley experiment). An en dash is not used for a hyphenated name (Lennard-Jones potential, named after John Lennard-Jones) or an element that lacks lexical independence (the prefix Sino- in Sino-Japanese trade).
This is a lot of work for a few bullets, isn't it? But I think what we've got now is really good. If nobody raises any objections in the next few days, then I'll change the page. Ozob (talk) 05:27, 24 December 2009 (UTC)
Well, revamping them might not be 100% necessary, but it's not useless either. Right now, the first sub-point (starting "To convey the sense") isn't exactly an easy-to-read paragraph. ― A. di M. — 2nd Great Wikipedia Dramaout 16:23, 24 December 2009 (UTC)
Ozob, you write: "If nobody raises any objections in the next few days, then I'll change the page." Well, I object. For at least four reasons:
1. The detail is faulty. For example, you equivocate on the word range:

To stand for to or through in ranges (pp. 211–19, 64–75%, the 1939–45 war, 17 May – 22 November). Ranges should not occur with prepositions (3,000–3,500 people or 3,000 to 3,500 people, not from 3,000–3,500 people).

At first range denotes something that could be expressed either with a dash or with one or two propositions; but in your continuation it denotes something expressed with a dash. This is what you mean, I think:

To stand for to or through in ranges (pp. 211–19, 64–75%, the 1939–45 war, 17 May – 22 November). Range indications should not mix prepositions and dashes (3,000–3,500 people or 3,000 to 3,500 people, but not from 3,000–3,500 people).

(In fact, that through is almost exclusively US English, and should not be encouraged. Many others find it distracting and irritating. But that's a separate issue.)
2. I do not accept that these points should be discussed with consideration of spacing deferred. MOS needs coherent guidelines that work well together; so all the issues with en dashes ought to be thrashed out together, till there is consensus on their use broadly. I include the issues that have been raised with HTML entities versus Unicode. These are not "completely separate", since presenting editors with a comprehensive, workable practice is paramount. Tony has raised the matter of hard spaces preceding en dashes, and the absurdly long string that would be needed if the space and the dash were both coded in HTML. Such a thought is also highly relevant, in designing a usable guideline. But I have given up trying to persuade people of the need for a proper way to code hard spaces. Few here seem to appreciate that need. And while I'm at it: please set aside any suggestion to use templates for spaced en dashes (heaven help us all). Editors need to type intuitively: with few keystrokes, on real keyboards. They will quite rightly refuse tech-heavy "solutions" for basic punctuation.
3. The thread above is fragmented and convoluted. It is quite a task to keep up with what is proposed and why. I'm sure some interested editors have by now given up trying. Amid such chaos, there is no prospect of genuine, wide, durable consensus for change.
4. A general and very urgent problem: there is no means established here to record a consensus when it is achieved. The present guidelines for hyphens and dashes were ironed out in 2007, through much systematic discussion. Subject to minor alterations, they have stood (and served well) since then. But no one can easily find this out, and I see no reference to all that hard work in the present round of discussion. This is wasteful and inept. We desperately need better methods. I have raised this again and again, but I cannot recall anyone else registering the slightest interest. Ignoring history, we repeat it. Meanwhile, many of the editors whose needs we pretend to serve abandon MOS as useless. We come across as a bunch of jejune amateurs. (Well, some of us are!) Take a look at our Archive 108, where a number of perennial favourites are churned through for the zillionth fruitless time. Yes, you'll find en dashes there; but especially take a look through the great debate concerning possessives, culminating in the current travesty of a guideline. (That kindergarten-level exercise prompted my disappearance from MOS, soon to be resumed. Some of us have better uses for our time.) Read also of my attempt to establish a means to record the latest relevant discussion of any given guideline, so it could immediately be found by anyone interested. The response? Dull, uncomprehending, and carping.
In short, I do not agree to any further changes to the en dash guidelines. They are half-baked suggestions, no matter what the undisputed particular competences of the proponents.
I urge editors to focus intently on bigger matters first. When enough of you are ready to get serious, let me know! Till then, I'm busy elsewhere. (O, and Wavelength: DO consider emailing me, OK?)
¡ɐɔıʇǝoNoetica!T– 08:34, 25 December 2009 (UTC)
What's needed is a FAQ, available at the top of this and every other MOS talk page, that lists every such "perennial favorite", and a solid consensus here that any re-raising of an issue on that FAQ will be immediately tagged {{Resolved|Duplicate of MOS FAQ, item 123.}} and speedily archived on the spot. It should require a Village Pump resolution or RFC to change anything serious in MOS, or it will never stablize. Another, grander idea is to do like WP:WSS, which began as (and is still called, but really is not) a WikiProject: formalize and processify. Create Wikipedia:Manual of style/Proposals and revert any MOS change that does not go through the proposal process (successfully). An intermediate idea would be to make every thread here that proposes a change an actual proposal, like an XfD and close each one after X amount of time, with "closed as accepted", "closed as rejected", or "closed as no consensus" (closer should be an admin, and pages should be protected). Other ideas might come to mind, or some combination of these, but in 2010 I'd like us to look at seriously attempting to stablize MOS and keep it from being something every pundit and yahoo can change on a whim, and something everyone fights about a lot less. Few if any fights erupt over what does and does not constitute valid deletion criteria at WP:CFD, or various other detailed, nitpicky WP process things, yet MOS is nothing but constant fights and editwarring. – SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 10:49, 25 December 2009 (UTC)
I disagree about such a FAQ, SMC. That would be tantamount to telling new people that they don't count and could potentially give us more ownership over the MoS than we're supposed to have. The whole idea of Wikipedia consensus is that it can change over time. However, the idea that all significant changes to the MoS (other than say, wording) should go through discussion first is a good and reasonable one. Darkfrog24 (talk) 13:27, 25 December 2009 (UTC)
{{ndash}} turns out to be fewer keystrokes than &nbsp;&ndash; , though. (In an ideal world you'd use something like _-- which is even fewer keystrokes, but ...) ― A. di M. — 2nd Great Wikipedia Dramaout 13:57, 25 December 2009 (UTC)
  • "The thread above is fragmented and convoluted. It is quite a task to keep up with what is proposed and why." I agree that it would be helpful to have a single draft rewrite of the whole en dash section rather than these dribs and drabs.
  • "I do not agree to any further changes to the en dash guidelines" As things stand, there is no consensus for the current guidelines, and the guidelines are tagged. In effect this means that there are no en dash guidelines. This is not an accurate representation of the consensus here. The guidelines need to be modified to accurately record the consensus that we do have, which is what Ozob's draft and my draft (I don't care which) have tried to do. Constructive suggestions for improving these drafts would be welcome. It's clearly untenable to insist on no further changes.
  • "It should require a Village Pump resolution or RFC to change anything serious in MOS, or it will never stablize." First, spaces around endashes are not such a big deal. Second, the MoS should not be carved in stone, and there should not be a gigantic bureaucratic overhead to make minor changes like this; especially, as in the case here, where the change would be a simplification that reflects common practice better.
Eubulides (talk) 19:35, 25 December 2009 (UTC)
  • "The present guidelines for hyphens and dashes were ironed out in 2007, through much systematic discussion." I went back and read all the 2007 discussion on this talk page re the subject of dashes. (The biggest thread is at Wikipedia talk:Manual of Style/Archive 82 #Hyphens and dashes in the MoS; there are several other threads.) I couldn't find any comments about whether spaces should surround disjunctive en dashes. It appears that the 2007 discussion here provides no evidence for the claim that there was an earlier consensus on this issue. Eubulides (talk) 06:39, 26 December 2009 (UTC)
  • I'm not saying there shouldn't be discussion about change, but please understand that there is no justification for the comments I've seen floating around such as "there is no consensus for the current wording". No matter how you want to slice it, I'm afraid that if it is on the page for any length of time, then it is policy, and we don't look backwards. It is not correct to say that there is no consensus for the current wording. It is correct to say that there is now debate about the current wording.  HWV258.  07:22, 26 December 2009 (UTC)
  • There may have been consensus for the MoS wording at some point, but clearly there is not consensus now. So far we've seen no basis for the suggestion that spaced disjunctive endashes were discussed earlier and that an earlier consensus was established based on that discussion. If such a discussion existed, it would have been helpful for us to see it, to avoid repetition here. Eubulides (talk) 08:03, 26 December 2009 (UTC)
  • Your response edit comment included "Consensus can change". Since that misses the point I raised, I have to ask: why do you feel you always have to respond? It's okay to read something and think (to yourself): "fair point". You'll move ahead leaps and bounds on WP if you get out of the mindset of always having to respond. Cheers.  HWV258.  08:49, 26 December 2009 (UTC)
  • "That misses the point I raised" Actually, my previous bullet directly addressed your point "It is not correct to say that there is no consensus for the current wording". Stable material did establish a consensus, but this old consensus is no longer present. It would be helpful to move forward, by rewording the MoS to express the current consensus as best we can. "Why do you feel you always have to respond?" is a question I'd rather not respond to. Eubulides (talk) 09:23, 26 December 2009 (UTC)

Archive 82

[Outdent] Eubulides, you refer us to Archive 82, and you say this about your reading of it: "I couldn't find any comments about whether spaces should surround disjunctive en dashes. It appears that the 2007 discussion here provides no evidence for the claim that there was an earlier consensus on this issue."

You must have noticed, however, that the discussion was unusually orderly and thorough, and drew on material that Tony set up in userspace. A draft was prepared there, and many editors worked on it collegially (even more thoroughly). This "subcommittee" procedure kept WT:MOS free from its accustomed sprawl and clutter, but no one was left out of the loop in any way. The whole episode was a model of consensus-making, and we could all learn a lot from revisiting it.

The "subcommittee" discussion, along with the draft at its location linked at Archive 82, is apparently not preserved. But that draft was clearly discussed openly on this page, as the archive shows. After weeks of careful consultative development, the draft was duly incorporated in WP:MOS on 14 June 2007, with this edit. The draft text included the following provision, which attracted no dissent whatsoever:

All disjunctive en dashes (1. above) are unspaced, except when there is a space within either or both of the items (“the New York – Sydney flight”, “the New Zealand – South Africa grand final”, “3 July, 188818 August, 1940”).

The substance of this provision, having been scrutinised over the preceding weeks, has stood as a guideline for the community for almost a year two years and a half since it was incorporated in the Manual of Style.

The provision has enjoyed well-founded consensus since its inception. Now, some editors are challenging the provision. But that challenge does not by itself dispel consensus. The last time the matter was thoroughly tested, along with all matters concerning dashes, hyphens, and associated spacing, there was consensus. Nothing has yet happened to overturn that result.

In fact, we have not properly defined consensus for style matters here at WT:MOS, any more than we have thought to make provision for recording it. But by any standard, this must count as one of the classic examples of consensus.

Eubulides, you write above: "First, spaces around endashes are not such a big deal. Second, the MoS should not be carved in stone, and there should not be a gigantic bureaucratic overhead to make minor changes like this; especially, as in the case here, where the change would be a simplification that reflects common practice better." I respond to these two points:

1. Deals are big or small depending on the place and the context. Here, spacing around punctuation is indeed a big deal. We would be remiss if it were not. Printed style guides touch on the matter hardly at all; but we have to. Again and again we need reminding: this is all new. We are ahead of all current printed guides, because we are wrestling with a new medium, with new ways of writing, editing, and presenting text. (How do I know? I collect such guides, in four languages; I research them; I have taught concerning their provisions at tertiary level; I study and use and compare them almost every day.)

2. Of course MOS should not be carved in stone! No one says it should be. But neither should it be "writ on wind and water", subject to every passing fad or the latest indignant dissatisfaction. Especially with punctuation, where the available resources fall far short of the functions that punctuation must serve, no solution can ever be perfect. It is easy to find fault even with provisions that have stood for years. I do it myself with Chicago and New Hart's, all the time.

For these reasons, I maintain my opposition to any change concerning en dashes at this stage. I certainly will not shift from this stance while the discussion is conducted opportunistically and without all related matters being considered together. It would be easy for me to refute certain debating points made above in this thread. But that would be a waste of time, since the goal-posts are continually moved. What matters is this: there is obviously no consensus for further amendment. If anyone will now mistake a waning in the discussion here as a licence to amend these provisions in MOS, I and others will have a responsibility to revert them. We have spoken clearly and reasonably to support the present reasonable and clear consensus. Let people try again another time if they like, after they have thought and read more on these surprisingly subtle issues.

¡ɐɔıʇǝoNoetica!T– 11:26, 26 December 2009 (UTC)

The last proposal by Ozob, below, is definitely not a big deal, as it doesn't change the instruction at all; it just makes its wording more concise and clearer. (Or is there any situation where Ozob's proposal and the current wording in MoS give different advice, which I am missing?) ― A. di M.2nd Dramaout 12:10, 26 December 2009 (UTC)
It is a big deal in this way at least: it treats only a part of the material concerning en dashes, but the matter needs to be thought through as a whole, as it was in the reforms of 2007. For one thing, a fragmented approach risks alienating the editors who consult MOS. They don't want to have to track partial changes every couple of days. For another thing, I and others are not inclined to give such a fragment our full attention. I would still change some wording in the present proposal, in fact. But I don't care to discuss that now, for reasons amply laid out above. There are deals that are even bigger than this. They have priority.
¡ɐɔıʇǝoNoetica!T– 13:17, 26 December 2009 (UTC)
I agree with Noetica. At the moment there is no consensus for change.  HWV258.  19:27, 26 December 2009 (UTC)
While I'm sympathetic to Noetica's desire to get all the endash stuff done at once, and agree that it's more of a pain to do it step by step, Noetica has made no substantial objection to Ozob's latest proposal. Following up on some other comments by Noetica:
  • "that draft was clearly discussed openly on this page" Sure, but that's what I said: there was quite a bit of discussion here back in 2007, but the topic of spaced disjunctive endashes was not discussed here. The first serious discussion of the topic on Wikipedia appears to be this very thread, a thread that has demonstrated that there is no consensus to prohibit usage like "Chicago–New York flight".
  • "challenge does not by itself dispel consensus" This is not just some random driveby complaint. This is a series of postings by many editors objecting to the requirement, along with substantial evidence that the MoS requirement prohibits near-universal practice among high-quality scholarly publishers and widespread (though not universal) recommendation of style guides.
  • "spacing around punctuation is indeed a big deal ... this is all new. We are ahead of all current printed guides" Doesn't this contradict your earlier comment "Think very carefully before assuming that the latest contrivance is necessarily the best."? But at any rate, it's irrelevant whether one subscribes to the claim that the MoS is "ahead of all current printed guides". What matters is whether the MoS's relatively-minor spaced-endash provision reflects consensus. Clearly it does not.
Unfortunately Noetica's comments contained no constructive suggestions for improving the current MoS, in particular, for improving Ozob's draft improved text. I again urge constructive and collaborative engagement with the draft to help it reflect consensus better than the MoS does now. This is far more likely to improve the encyclopedia than threats to revert. Eubulides (talk) 07:41, 27 December 2009 (UTC)

I take strong exception to Noetica's claim that "The substance of this provision, having been scrutinised over the preceding weeks, has stood as a guideline for the community for almost a year two years and a half since it was incorporated in the Manual of Style" and "has enjoyed well-founded consensus since its inception". My take on the matter is: this provision has existed in the MoS for that long, but for most of that time it remained untested, in part because in past times most Wikipedia editors favored hyphens in place of en-dashes. As soon as it was actually tested on non-range disjunctions (by the recent moves and unmoves of several mathematics articles including Seifert–van Kampen theorem) the test brought to light strong objections that showed that it was not the consensus after all, and moreover that it was in complete conflict with standard usage (in mathematics and elsewhere) and with many published style guides. We can argue whether there was a consensus at one time for this provision, or whether it passed into the MoS without consensus because the editors at that time weren't paying attention to how broadly they worded those clauses, but I think that argument is moot: the point is that we have no consensus now for these clauses, and the claims of Tony et al that "no consensus to change = consensus to keep as is" have no foundation in logic. So, we need to build a new consensus, because the alternative is not the status quo but rather having no reliable guidance in the MoS regarding this issue at all. —David Eppstein (talk) 20:31, 27 December 2009 (UTC)

Responding point-by-point

← Responding point-by-point to Noetica:

  1. This is a good point. I prefer "Ranges should not mix prepositions and dashes" but if there is a good reason to prefer "range indications" then I'd be happy with that, too.
  2. I disagree. The specifics of spacing and entity references can be resolved separately from the present copyedit of the instructions on when to use en dashes.
  3. Yes.
  4. This is a good idea.

As the text stands now, it's:

To indicate disjunction. There are three uses.

  • To stand for to or through in ranges (pp. 211–19, 64–75%, the 1939–45 war, 17 May – 22 November). Ranges should not mix prepositions and dashes (3,000–3,500 people or 3,000 to 3,500 people, not from 3,000–3,500 people). Number ranges must be spelled out if they involve a negative value or might be misconstrued as a subtraction (−10 to 10, not −10–10).
  • To stand for to or versus (male–female ratio, 4–3 win, Lincoln–Douglas debate, France–Germany border).
  • To stand for and between independent elements (diode–transistor logic, Michelson–Morley experiment). An en dash is not used for a hyphenated name (Lennard-Jones potential, named after John Lennard-Jones) or an element that lacks lexical independence (the prefix Sino- in Sino-Japanese trade).

Again, if nobody raises any objections to this text in the next few days, then I'll change the MoS. Ozob (talk) 20:59, 25 December 2009 (UTC)

If you do that, Ozob, people will feel obliged to revert your edit. See my last extended contribution in the preceding subthread.
¡ɐɔıʇǝoNoetica!T– 11:29, 26 December 2009 (UTC)
I do not understand your position, so I have asked for mediation. See Wikipedia:Mediation Cabal/Cases/2009-12-24/Wikipedia:Manual of Style. Ozob (talk) 18:52, 26 December 2009 (UTC)
I fully support what Noetica has said. BTW, there's an unsolicited comment on my talk page about this thread. Tony (talk) 09:13, 27 December 2009 (UTC)
Tony, Noetica, HWV, and all those currently opposed to the text in the blue box above: What changes would it need to gain your support? Ozob (talk) 13:03, 28 December 2009 (UTC)
As I have already indicated, I will not countenance further changes to the MOS guidelines for en dashes until all the issues are put on the table here (along with any connected issues with hard spaces, hyphens, or em dashes), and we are all committed to finding a robust consensus, and an adequate means of recording this consensus. Why should I or anyone else want to waste still more time on small evanescent fluctuations in our guidelines? The current tag warning that these will occur is a disgrace, and ought to be removed. "The present language is likely to fluctuate"? Why? Fluctuate here on the talkpage, not in a text presented for the guidance of millions of editors. Less alarming tags are available, to alert editors that there is a discussion going on.
I take this stand not to be obstructionist, but to be serious and to respect the Project – by respecting the needs of our editors and the needs of the huge readership that we all serve.
Propose a draft that shows a unified and harmonious guideline, taking all factors into account, and we can get serious. Anything short of that is a disruption. (More from me later, on points made in other subthreads above.)
¡ɐɔıʇǝoNoetica!T– 00:21, 29 December 2009 (UTC)
Have to concur on the need to remove the warning. It presupposes a consensus that the wording there will change, yet I see no such consensus on that matter, only a very protracted debate on proposed changes. As for the latest proposed wording, I'm going to continue being pretty neutral on it. I think it has improved, but I honestly have not pored over every rationale pro and con for every part of it. – SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 00:44, 29 December 2009 (UTC)
The warning needs to stay because there is clearly not a consensus for the current wording. Unfortunately, despite repeated requests, those in favor of prohibiting "the Chicago–New York route" still has not proposed any specific wording that would capture the current consensus better: instead, all we have seen are comments that continue to insist on no changes whatsoever until all issues are resolved. That is a recipe for inaction. We need a better recipe. A specific proposal is on the table: please propose specific wording changes to it. Eubulides (talk) 06:56, 29 December 2009 (UTC)
I agree with Eubulides here. The discussion makes it obvious that we currently have no consensus. That includes no consensus for keeping the old wording. We can work towards coming up with a new consensus, or we can dig in our heels, but the latter course of action will not result in the old consensus prevailing: it will cause us to continue to have no consensus, and therefore no guidance from the MoS on this issue. —David Eppstein (talk) 07:16, 29 December 2009 (UTC)
Original wording is kept by default in guidelines and policies, since it takes consensus to change them. I agree that it's better to come to a clear consensus to either change the wording or keep it as is and not mess with it, but there isn't actually any "no guidance" doomsday scenario. — SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 09:55, 29 December 2009 (UTC)
I have replaced the tag with one informing editors that the section is under discussion and inviting them to join in, rather than warning them of "fluctuations". See my reasons above, and endorsement from SMcCandlish. I propose that we leave this tag in place till the matter is settled; and once more I suggest that we engage in broad good-faith dialogue, rather than dismissing clear evidence and copious argument with objections that counsels of caution are a "recipe for inaction". They are nothing of the sort. If you have not paused to define consensus, nor to provide for recording it, nor to consider how one part of a guideline might affect another, nor to survey the relevant archives with an open mind before forming your judgement, nor to listen dispassionately to those who (unlike you) are versed in the printed style guides, your precipitate action will lead to instability. But Wikipedia needs stability in its Manual of Style. I grow tired of repeating these plain truths, and of being misquoted or misread, and of obdurate refusal to look at the larger issues. Good editors leave MOS discussions because of such shortsightedness. More later perhaps, if I can find time to address further moves in this vexatious debate. However well-intentioned those moves might be, they are distinguished more by rhetorical surface than by substantive depth of argument.
¡ɐɔıʇǝoNoetica!T– 09:31, 29 December 2009 (UTC)
Eubulides has now substituted yet another tag. That's fine. Let's leave it there! –¡ɐɔıʇǝoNoetica!T– 09:38, 29 December 2009 (UTC)
Works for me. I'm actually a big fan of tagging guideline wording as disputed when there's a dispute about it, since it helps draw attention to the discussion and get it resolved faster. — SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 09:55, 29 December 2009 (UTC)

(outdent) Re: Eubulides's "A specific proposal is on the table: please propose specific wording changes to it." This has become so long-winded, heated, logically contorted and nitpicky on all sides that I for one would like to see a summary of what the alleged issues are – without any invective – with the extant wording, and a summary of what problems are seen with the proposed new wording. If we can do that without bashing each other, I suspect that compromise wording will be easy to arrive at, that resolves whatever faults the original wording had and doesn't introduce new problems. — SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 09:55, 29 December 2009 (UTC)

I did not see the discussion here before changing the tag back to Noetica's simple, unobtrusive informational (rather than emotive) tag. If editors who consult MOSDASH see the information tag, they will soon acquaint themselves with the debate here, and may even join it. Having a fierce red icon makes us look likely warmongers, and is more likely to turn off editors who may contribute to this discussion. Tony (talk) 10:38, 29 December 2009 (UTC)
A "summary of what the alleged issues are – without any invective – with the extant wording": okay. As for the first point, there's not wrong with what it says, but it is unnecessarily wordy; it could be trimmed by about 50% without making it any less clear, and even making it clearer and easier to read. The three following points' content is right, too, but they are so similar in scope they could be merged into one with about half the words. The only problem I have with the content is the "spacing" point: we have seen that the practice of spacing all dashes when the operands themselves contain spaces is not universal, and while sometimes it is clearer, sometimes it is less clear (namely, when it could be confused with an interruptive dash). Among the style guides which have been cited in these threads, I think Butcher's advice is the most reasonable one. But if everyone else disagrees, I have no strong problem with keeping the current advice (provided the exception for surnames is kept). ― A._di_M.2nd Dramaout (formerly Army1987) 13:02, 29 December 2009 (UTC)
Okay. I'd like to hear the other half before getting into details, but my initial reaction is that this would probably go smoother if we have a proposal for changed wording that changes nothing about what is actually advised, just trims the verbal bloat. Reach consensus on that and install it (or reach consensus that it isn't too wordy after all), then completely separately raise the issue of what to do or not do about the spacing question. They're really completely separate issues, and cleanup of bad prose shouldn't be held hostage by concerns about substantive changes to the advice, a weightier and more ponderous matter. — SMcCandlish   Talk⇒ ʕ(Õلō  Contribs. 14:11, 29 December 2009 (UTC)
That proposal already exists: it's outlined in color at the start of this subsection. Let's stick with the tag that Noetica, SMcCandlish, and I agreed on: it's more accurate to say that this thread is a dispute rather than merely a discussion. I don't see why there should be an exception just for surnames; high-quality scholarly sources prefer unspaced endashes even for non-surnames. In practice, among these high-quality sources, spaced endashes are the exceptional case, not the other way around. Eubulides (talk) 16:29, 29 December 2009 (UTC)

Let's return to the actual content discussion. Those of you who are opposed to the boxed text above (including but of course not limited to Noetica, Tony, and HWV), what would it take for you to support it? If you don't think you will ever support it, why? So far I have not seen any objections to the content, only to the process. Ozob (talk) 17:25, 29 December 2009 (UTC)

Relationship to endash–emdash disagreement

Isn't this disagreement over spaced vs. unspaced disjunctive endashes similar to the disagreement over spaced parenthetical endashes vs. emdashes? Some style guides recommend spaced endashes for parenthetical remarks and others recommend emdashes, with no general consensus, and with both styles allowed in Wikipedia articles. MOS:EMDASH says this:

'Spaced en dashes – such as here – can be used instead of unspaced em dashes in all of the ways discussed above. Spaced en dashes are used by several major publishers, to the complete exclusion of em dashes.'

For consistency shouldn't the MoS also say the following?

'Spaced en dashes can be used instead of unspaced en dashes for disjunctions, such as "10 January – 12 February", in which one or both items contain spaces. In this case spaced en dashes are used by several major style guides, to the complete exclusion of unspaced en dashes.'

After all, the situation is similar: we have dueling style guides in both cases, so in both cases the MoS should allow either style. What—if anything—is wrong with a consistent treatment of these two issues? Eubulides (talk) 06:54, 31 December 2009 (UTC)

No. It's completely different. Tony (talk) 07:57, 31 December 2009 (UTC)
Tony — I don't know whether this is your intent, but you are coming across as intransigent and uncommunicative, insisting on sticking to an old consensus that obviously doesn't exist. This sort of curt response to yet another attempt to make progress on this issue doesn't help. Could you please describe a process that you think can lead to a new consensus, since the old one doesn't exist and you keep shooting down all attempts to make progress on this issue? —David Eppstein (talk) 08:11, 31 December 2009 (UTC)
David Eppstein: intransigence breeds intransigence. This time, Tony has responded with a curtness that will get us nowhere – to a stonewalling refusal to accept the results of extraordinary collegial consultations that he initiated way back in mid-2007. That refusal itself will get us nowhere! Myself, rather than taking on your eminently defeasible assault on my articulated calls for caution and orderly procedure, I now have a more radical proposal to make. See the subsection I am initiating below this one.–¡ɐɔıʇǝoNoetica!T– 08:43, 31 December 2009 (UTC)
Don't accuse me of intransigence. I have already engaged in a significant compromise to the guideline on spacing. I'm not going to stand by idly while people wreck the formatting of date ranges and the rest. Tony (talk) 01:11, 1 January 2010 (UTC)

Proposal to defer discussion of dashes

It must be admitted, I think, that deliberations in this thread have stalled. SMcCandlish has intervened, attempting to break the impasse; but that appears unlikely to succeed. I propose that we move on, deferring any further review of the MOS guidelines for en dashes (and associated issues) till we have thoroughly dealt with some overarching issues:

  1. How are we to define consensus, for the crucial work that MOS performs within the Project?
  2. How is a MOS consensus to be achieved?
  3. How is a MOS consensus to be recorded, for all editors to see?
  4. When and how does a MOS consensus ever lapse?

Wavelength has opened that discussion, in a section below. I strongly recommend that we all give that discussion our sustained attention. It will take a lot of hard work to get things right: but what issue is more urgent? This present thread shows what happens when we ignore the larger questions. I say that Tony did groundbreaking work on consensus-building, for the whole matter of hyphens and dashes of all kinds, in mid-2007. I was there, so I know what an effort was made. If it is uncertain that there at least was a consensus in this case, it must surely be unclear in all other cases too.

We can return to dashes, spaces, and all of that later. Then we can work with the necessary procedures in place – procedures unique to MOS, perhaps. Nothing is lost by such a patient deferral. Later, we can establish a consensus that all editors will accept as one, even if some disagree with the details of the guideline. That disagreement is inevitable; but balanced, rational, and stable compromises are surely something we can achieve, and ought to work towards.

I now call on your support for this proposal, and I look forward to us all moving on: to work together in the consensus section.

¡ɐɔıʇǝoNoetica!T– 09:16, 31 December 2009 (UTC)

There already is WP:Consensus which is policy... ― A._di_M.2nd Dramaout (formerly Army1987) 22:52, 1 January 2010 (UTC)

Discussion of the proposal to move on: support and dissent

Certainly it will help to better document consensus, and we can keep the en dash section tagged while we temporarily turn our attention to the consensus-recording issue. Eubulides (talk) 10:25, 31 December 2009 (UTC)
Let's not diminish those weighty matters as calling for a short diversion of attention, while the en dash section remains flagged continuously as under discussion. Your cause, and everyone else's, would be better served by acknowledging the procedural failures of the current thread, then abandoning this thread, and starting anew with a clean slate when the time is right. Nothing is lost by doing that, and everyone gains. Please remove the tag, for now.
[I have reworked this as a new subordinate subsection, because the call I have made for collaboration on larger issues needs to remain clearly visible. It's in everyone's interest to work together on that.]
¡ɐɔıʇǝoNoetica!T– 11:17, 31 December 2009 (UTC)

I'm in no hurry as long as we seem to be making progress to a consensus rather than permanently deadlocked and as long as the old language in the MOS isn't used (again) to move the mathematics articles I care about to names nobody outside Wikipedia uses. Which is to say, I think your call for us to take a step back and look at how we establish and record our consensus here is a good thing. Let's keep in mind WP:CREEP as we do so, though. —David Eppstein (talk) 21:02, 31 December 2009 (UTC)

Nothing is lost by leaving the article in its current state, including the tag, while we discuss the procedural issue. Keeping the tag addresses concerns about mistaken use of the MoS to require a style where there is no consensus for the requirement. It shouldn't take that long to decide on procedural matters such as whether the talk page should have an FAQ. Eubulides (talk) 23:56, 31 December 2009 (UTC)
What is lost is that every visitor thinks the MoS is inherently unstable. I strongly object to the medium-term presence of this "dispute" tag; it should be reverted to the "information" tag. I will do so within 12 hours. Tony (talk) 01:07, 1 January 2010 (UTC)
I must comment on all of those points, Eubulides.
I have suggested a clean break, so that later we can have a new and better discussion of en dashes (or any other issues at the same level of specificity), when all procedural matters are sorted out. That means relinquishing the present discussion, responding in good faith to the good faith of those who seem to disagree with you. You persist in claiming that MOS is somehow in error as things stand, and that there is no consensus on the matters you and others have raised. But these are the very claims that motivate our proposed discussion of procedure. Let's not persist with such claims here and now: nor with any counter-claims. Let's start a fresh discussion later, on a firmer footing, to address your legitimate concerns.
If we all applied tags to sections of MOS that we consider faulty or still under review, or non-consensual, hardly any section would be free of them. I would certainly have the Possessives section tagged, and two more. Other editors would have theirs too, and the result would be chaotic and unseemly.
Finally, excuse me for mentioning it, but this may be relevant: I have had somewhat more experience of this MOS work than you have (see statistics for MOS and statistics for this talkpage). I can assure you that the issue of consensus, and associated procedural matters, will not be dealt with quickly or easily. That discussion has not even begun to consider the other pages of the Manual of Style (WP:MOSNUM and all the rest); yet how these all work together procedurally must be addressed, in full consultation with regular contributors to those pages who are rarely seen here at WT:MOS. There will also need to be wide community consultation, including perhaps an RFC. Please take seriously the much broader issues that have been very usefully raised in the narrower discussion of en dashes.
¡ɐɔıʇǝoNoetica!T– 01:34, 1 January 2010 (UTC)
  • "What is lost is that every visitor thinks the MoS is inherently unstable." No, the tag doesn't say that. All it says is that there's a dispute over the en dash section, which there is. Changing the tag to say that there's merely a discussion would make the tag less accurate. Please don't make the tag less accurate.
  • "I will do so within 12 hours." Setting a 12-hour deadline on New Year's Day is not really in keeping with the spirit of Wikipedia:There is no deadline. There really is no rush here, and there's certainly no need to set arbitrary deadlines.
  • "Let's not persist with such claims here and now: nor with any counter-claims" It's fine for us to temporarily focus on improving the dispute-resolution process: but that doesn't mean that we should in the meantime pretend that there is no disagreement.
  • "If we all applied tags to sections of MOS that we consider faulty" I am not in the habit of applying tags to the MoS, and it was not my idea to apply a tag in the first place. However, it cannot be denied that there is a serious and continuing disagreement over that section; also, there is a serious and reasonable concern that without the tag the MoS will be mistakenly interpreted to require a style for which there is no consensus. In cases such as these a tag is called for, and it's inappropriate to remove the tag.
  • "That discussion has not even begun to consider the other pages of the Manual of Style (WP:MOSNUM and all the rest)" The proposal was not to defer the matter for a massive project that would rationalize the entire MoS. Such a project would take months at the very least, and more likely would continue indefinitely. If that is in fact what is contemplated, then that's obviously too much delay, and we should instead apply standard conflict-resolution procedures to this dispute. Ozob has already suggested mediation as one step, and that would be fine with me as well.
Eubulides (talk) 08:17, 1 January 2010 (UTC)
Eubulides, point by point:
  • Remove the tag. If you take what I propose at all seriously, you will understand there will soon be no active process of dispute concerning en dashes. There will a fresh one, later.
  • Tony's declaration that he will do something within 12 hours if your side does not act in the meantime has nothing to do with Wikipedia's conventions regarding deadlines. And why do you say there is no rush to remove the tag, but act as if there were a rush to change the provisions for en dashes? (Rhetorical question only!) I was going to comment earlier, by the way, that Ozob's suggestion to act quickly is particularly unfortunate at this time of year, when regular editors may not be able to pay attention to MOS. Let's have patience all around.
  • No one is pretending there is no disagreement. You have been offered a way to make your case on firmer ground, and press what I have called your legitimate concerns, with the assistance of all hands. Others will make their case also, when it is not futile to attempt such a thing.
  • Nor am I in that habit, nor is Tony. Tags are to be used with caution, when editors need to be alerted to present moves. But here the suggestion is that we all leave this dispute behind, so that we can focus on ground rules. You want to mark what you take to be a lack of a consensus: but even what constitutes a MOS consensus is an issue that we are adjourning to consider! That has never been thought through, but this dash affair now highlights its urgency. Please understand: a clean break, a new beginning later.
  • How much later? It could well take a long time, yes. But it will be much quicker if we set smaller issues aside in the meantime, and work together. As for Ozob's suggested mediation, it is ridiculous. The terms of it are slanted, and Ozob misrepresents me from the outset. For example, I did not "threaten" to revert anything! Check above: I said that we would feel a responsibility to revert, and that we would feel obliged to revert, if there were further changes without consensus. I now suggest to Ozob that the mediation request be withdrawn.
¡ɐɔıʇǝoNoetica!T– 08:58, 1 January 2010 (UTC)
"And why do you say there is no rush to remove the tag, but act as if there were a rush to change the provisions for en dashes?" There must be a misunderstanding here, as I have consistently argued that there is no rush to change the provisions: for example, I reverted Ozob when he installed such a change (a change that I agreed with), purely because I thought the proposed change needed more than a day or two of discussion. Similarly, there is no rush to remove the tag, given that the dispute (obviously) is still present. A "clean break" in no way implies that the tag should be removed: whether we use a new way to resolve the dispute is independent of whether the dispute exists. Even though there's no rush, the process of course needs to take its course in a reasonable amount of time. If it takes months or years to establish a new way to record consensus, then there's something wrong with the process, and we should consider alternative means of dispute resolution. For starters, it would be a shame for mediation to be rejected out of hand. Eubulides (talk) 09:22, 1 January 2010 (UTC)
Very well. The mess of discussion weighing on us from above is so tangled that I can't easily immediately tease out all of the moves, and who made them. I wouldn't reject mediation out of hand if it were at all a reasonable proposition. This time, it patently was not. Look, I don't want to waste any more time on this. Do you? ... I thought not. Let's close the thread, and get on with something more important. (Sheesh, I said at the start that I really don't want to be here.)
¡ɐɔıʇǝoNoetica!T– 09:32, 1 January 2010 (UTC)
You are welcome to leave, then. I requested mediation in good faith and stated the situation as impartially as I could. I am extremely offended at your insinuation that I requested mediation only to further my own position. Ozob (talk) 16:55, 1 January 2010 (UTC)
You may have suggested mediation in good faith, Ozob. Read what you will into my words; but I do not say, nor do I insinuate, nor do I believe, that you made the suggestion only to further your own position. I do say that mediation was a ridiculous idea. I have articulated my position at length; if you miss something in it, you have only to ask for clarification, not act as if I had refused to clarify. As for my leaving, don't tempt me. I came here, most reluctantly, because no one was able to survey the printed style guides adequately. I am able to, so I cited some. Since then, it becomes apparent that no evidence concerning those guides will count in any case. Then my points about consensus were for long met with Jesuitical casuistry or with silence; but do you imagine that I am naive in such matters, or seek merely to prevaricate? Still no one takes notice when I argue that the printed guides and the learned journals in mathematics are not final arbiters of good practice for Wikipedia, since we are on uncharted seas. We must be the mapmakers, even though Wikipedia policy can be opportunistically cited against our making such necessary progress. I understand your frustration; please understand mine. We are both well motivated.–¡ɐɔıʇǝoNoetica!T– 20:53, 1 January 2010 (UTC)

If there is further discussion of en dashes or em dashes, please notify me on my talk page. Is there a way to automate this request? Anomalocaris (talk) 07:22, 6 January 2010 (UTC)

I don't think there's a way to automate requests like that. Anyway, as noted above, this en dash thread was suspended while we discussed possible ways to improve the consensus processes here, in #Defining consensus, #Achieving consensus, #Recording consensus, #Lapses in consensus, #Wikipedia:Consensus and Wikipedia:Manual of Style, #Wikipedia:Consensus and Wikipedia:Manual of Style and Wikipedia:Ignore all rules, and #Examples of consensus. An FAQ has resulted, visible at the top of the talk page, but otherwise the discussions have not resulted in any specific proposal for improving the consensus process, and seem unlikely to do so; so it may be time to resume the en dash discussion. If so, though, we should start a new thread rather than continue this one, as this thread is already about 240,000 bytes and its bulk makes it unnecessarily hard to edit the talk page. Eubulides (talk) 05:53, 14 January 2010 (UTC)
Eubulides: Good to see Anomalocaris showing interest in the issue, so let's advise him or her when a new section on en dashes is started.
It is wrong to say that, apart from the FAQ initiative, discussions "have not resulted in any specific proposal for improving the consensus process". See this from me, at #Achieving consensus, below:

We can also get more creative. Why not a dedicated supplementary page for MOS, replicating the structure of WP:MOS itself but with each section's content replaced by synoptic explanations of the relevant MOS content, and links to archived consensual discussions? Every section of WP:MOS, and some subsections, could have a discreet unobtrusive link to that supplementary page, for the use of MOS editors and other enquirers. Why not? This would provide means of stabilising MOS, identifying topics that need further treatment, and informing all future discussion on this talkpage. The same could be done for all associated pages forming part of the Manual of Style. (And indeed, the relations between all these pages still needs more examination and reform.)

Typically for this talkpage, no one appears to have read with care. Rather it has been misread as something more run-of-the-mill, something that we already have partially in place. I'll add to the section in a moment, to correct that. (We see what we are primed to attend to.)
Meanwhile, since development of a better understanding of consensus at MOS is not much progressed, any new discussion of en dashes will still be difficult. The same problems with consensus must arise again, though at least we all understand each other's positions better now, and we have a more complete set of rulings from printed guides on record. Another broad issue that is unresolved: the status of formal printed academic journals as potential models for practice on Wikipedia. I would argue that they are not the best models at all (even if they did show consistency). Our context differs from theirs in these ways at least:
  1. Academic journal content is academic and learned; ours is much less formal in much of its coverage (sports, film, popular music, cooking, and so on).
  2. Academic journal style is more permanent, adapted for a kind of rigour that suits use over subsequent decades; ours can be amended as the language changes.
  3. Academic journals are produced by highly trained academic writers, subject also to codified editing by professionals; Wikipedia writing and editing is a collaborative and amateur effort par excellence, unique and without precedent.
  4. Academic journal text exists on paper, in a form that will not be altered, so how it will appear is predictable; our text exists first on the web, in a form that must change its appearance, from user to user and even from viewing to viewing (because user settings and browser settings are not fixed). Sometimes our text also turns up quoted on the web, and also quoted in print: a further consideration for us at MOS.
  5. The formal editing standards and codes for academic journals are set out (in traditional printed guides) for such stable text on paper. The inchoate standards and codes for Wikipedia must respond not only to the amateur production methods and different contexts of viewing (mentioned above): they must also respond to the quite different behaviour of strings of characters in HTML, variously interpreted by different browsers (even beyond settings). We who work on WP:MOS – and other parts of Wikipedia's extended Manual of Style – we are at the forefront of developing such standards, and such a code. Denying this, we default to a benighted, pusillanimous, and merely reactionary understanding of our role in Wikipedia.
Most of what I say above applies to formal, mostly printed, works in general; and while we must respect precedents from them as codified in style guides, we must not do so unthinkingly and without adaptations.
I say all this now because soon I'll be saying nothing at all. WT:MOS will attract all sorts of editors, displaying a colourful succession of factional agendas, and superannuated notions of editing from days when Fowler first snared pedants with birdlime (see Papageno the Very Serious Fowler), as his successors still do. But MOS will retain the editors it deserves. Good luck!
¡ɐɔıʇǝoNoetica!T– 07:07, 14 January 2010 (UTC)
I stand corrected: you did specifically propose subpages for recording consensus. As for academic journals, I respectfully disagree with much of your comment. In particularly the claim "Academic journal text exists on paper" may have been largely true twenty or even ten years ago, but it's not true now: almost all journals of any note are online now, and most reading is done online. Much of this online stuff is HTML, and academic journals that use HTML (which, these days, means most of them) face pretty much the same problems that we do with browser incompatibilities. Eubulides (talk) 08:54, 14 January 2010 (UTC)
I strongly agree with Noetica's post above, except for one point: "Academic journals are produced by highly trained academic writers". Naaah. The standards of English in academic journals have plummeted over the past 15 years, due to two things: (1) the influx of a huge number of (otherwise good) manuscripts authored by non-native scientists and scholars who find they have to write in English to be heard; and (2) the erosion of discretionary time in academia, related to the extreme pressure placed on the profession to teach more, churn out more research, sit on more committees, which has lessened enthusiasm in the profession for pro bono journal editing and management (they receive little professional acknowledgement for this work—in particular, only a tiny advantage when competing for funding). Wikipedia has to do better than what we now find in most "academic" journals if it is to be respected on the Internet. Tony (talk) 09:15, 14 January 2010 (UTC)
Eubulides, see my more developed proposal now, below.
While it is true that many journals are accessible online, many of their offerings are in PDF format, not HTML. And even when they do use HTML, bringing their text into comparability with Wikipedia text, they seem reluctant to publish guidelines to match the new medium. Generally, they could learn a lot from our Wikipedia MOS! They probably will, if they have not already begun to do so. (A related point: I think I have spotted a correction that Britannica online has made to an article, since it was criticised at one of our articles. I prefer not to say which one. There must be lots of this going on.) In any case, have you earlier referred us to examples of HTML publishing by journals? I thought your points (and examples?) in the long discussion above concerned academic practice for the traditional medium of print, echoed perhaps in PDF.
Tony, you and I have close firsthand knowledge of slipping editorial standards in academic journals. We could exchange stories and have a good laugh. To make a point, I spoke above of an ideal, in which they do still work at getting things right. But my case is in fact stronger if they don't do that.
¡ɐɔıʇǝoNoetica!T– 10:53, 14 January 2010 (UTC)
Noetica has a point. Rather than say that no specific proposal has resulted, let's say that no specific proposal has resulted yet. However, I agree with Eubulides that the hiatus should end. While we may get some results from out attempts to define MoS consensus and streamline the process of achieving it, the basic business of the page should not wait for it.
Having worked closely with scientists—and edited their English—I have to concur with Tony. Many of the best minds in chemistry and physics and pharmacology did not grow up speaking English and it shows (though why the journals can't employ better copy editors is beyond me). Wikipedia should not set its standards of English presentation so low. However, this may not be true of academic journals outside the hard sciences. Darkfrog24 (talk) 14:47, 14 January 2010 (UTC)

The discussion above is closed. Please do not modify it. Subsequent comments should be made on the appropriate discussion page. No further edits should be made to this discussion.