Wikipedia talk:Manual of Style/Dates and numbers/Archive 28

Units
The rules state for conversion and the example given
 * For example, "a 100 millimetre (4 in) pipe ..."

is completely against the international standard SI. After a number the abbreviation form should always be used!!!! AnyFile 17:49, 23 August 2005 (UTC)


 * We should aim to write language that as many people as possible will understand. The average person on the street is not an expert in SI units, and we should not expect them to be. It seems perfectly reasonable to spell out some units - indeed, it is desirable to do so, jguk 18:05, 23 August 2005 (UTC)


 * The SI units are only that dramatically important in contexts like science which actually use it.
 * Also, remember that that text is in prose. There was no consensus by my memory on the spelling out of numbers (Oxford and Chicago both recommend spelling out "a hundred", but there is no such suggestion here), and therefore we can assume here that, because it is in prose, "100" is a number that is essentially a word (poor way to explain, I know), abbreviated to digits for simplicity.  Therefore it makes perfect sense to spell out "millimetre". Neonumbers 07:35, 25 August 2005 (UTC)


 * If we spell out "millimetre" and "kilometres", certainly we must spell out "inch" and "miles" as well. Remember that for the vast majority of the planet's inhabitants, "mm" and "km" are readily understood (not only by "experts in SI units") but "in" might cause some confusion (especially with the word "in") and lots of people will never ever have encountered "mi" (despite being familiar with the 1,609-metre mile). Of course there are a few countries where "in" and "mi" might be more easily understood than "mm" and "km", and these countries are very important bases for the English-language Wikipedia, so I'm not proposing dropping them altogether or anything, just being a bit consistent. (I would probably write "400,000 km", not "400,000 kilometres", but I wouldn't revert someone changing it. Non-standard or not, if it's an extra service to the reader, then who am I to go against it?) -- Jao 23:49, 27 September 2005 (UTC)

I am not certain this has been addressed, but after reviewing other pages for the answer to this question, I believe there is no rule with regard to spelling out numbers under a certain amount (e.g., "John had 1 album" vs. "John had one album"). The Uniform System of Citation (Bluebook) requires spelling out numbers from one to ninety-nine in the main text while one to nine are to be spelled out in footnotes. This does not include volume or page numbers (e.g., "John mentioned the lost album on page 2 of his autobigraphy." Anyone know the status of the rule, if any? Jtmichcock 20:36, 29 August 2005 (UTC)


 * I don't believe consensus was reached on the issue, so it's up to the author. In my opinion, numbers twenty-one through ninety-nine should be spelled out in nonscientific articles when numbers are used infrequently. When I edit I spell out numbers one through ten in nonscientific articles and leave the rest as they are (that's just my personal preference/style). &mdash;Wayward 00:04, 30 August 2005 (UTC)


 * Well, I thought numbers from one to ten were always spelt out, so "one album" not "1 album", and I don't recall anyone who objected to that, so if that's not on the guideline, and I thought it was, it should be there.
 * My personal opinion matches that of Wayward, with the addition of simple multiples of nice round numbers (so that the entire thing is two words max, e.g. four million, four trillion but not thirty-two thousand). Personal opinion only though.  There was no consensus reached, and when I say that I could live without a guideline on something like I am now, then anyone who knows me well enough will know that the need for it is absolutely minimal (there's nothing that bothers me more than non-guidelines.)  Neonumbers 08:17, 1 September 2005 (UTC)

I'm going to start spelling out everything for all those cases where numbers are used to start a sentence&mdash;and there are a zillion of them on Wikipedia. That's a much bigger problem than this.

A distinction should also be made between counting numbers (integers) and measured quantities. When 5 means anything within about 0.5 of 5.0, then 5 should usually be written as a digit rather than spelled out. Too many style guides don't even mention that when discussing the spelling out of numbers. Gene Nygaard 11:56, 1 September 2005 (UTC)
 * I'm not convinced that spelled-out numbers is a big issue in general. I agree that its awkward to see numerals at the beginning of a sentence, and it would be awkward to see "four million eight hundred and fifty-seven thousand", for example, although "four million" would be fine. Otherwise, I don't think it really matters whether we write 99 or ninety-nine. Unlike some other changes I advocate for, I don't find either hard to read, nor do I find it jarring to see "99" in one article and "ninety-nine" in another. Chuck 15:53, 6 September 2005 (UTC)

If you want an example of a spelled-out-numbers-policy 'The Chicago Manual of Style' would be a good book to look at. --kop 20:19, 19 September 2005 (UTC)

Geographical coordinates
The current templates for geographical coordinates display with ugly additional blanks. These occur after the minute and second symbols and before the NESW indicators, like in 33°56' 24"  N 118°24'  00"  W. (to show the effect I purposely replaced the symbols by quotes). Is this intentional? Is there a remedy? I usually see coordinates in other sources formatted as 33°56'24"N 118°24'00"W. &minus;Woodstone 10:27:04, 2005-08-25 (UTC)

Units - superscript
Now that it's been a nice while since the units debate died down, I want to bring a few points up - minor ones, not fundamental ones like the one about conversions. The first is the removal of the line suggesting ¹²³ rather than &lt;sup&gt; tags to make 1 2 3. The reason for this is that the superscript, as you can see, makes the lines uneven. I think ¹ is a relatively new character, so maybe just ² and ³. Come to think of it, there are no situations where you'd need ¹.

So, a suggestion to add a line suggesting the use of ² and ³ where appropriate (e.g. m² rather than m2). This isn't asking much, but this is a style guide and it's worth putting that in just for clarification. Any thoughts? Neonumbers 11:24, 25 August 2005 (UTC)
 * Although my user stylesheet takes care of the extra leading, I think it’s a sound (re-)addition. You won’t need ¹ indeed (⁴ rather), but it’s not a new glyph, it’s for instance in ISO 8859-1. Christoph Päper 16:57, 25 August 2005 (UTC)
 * P.S.: If you want to improve your user stylesheet or a sitewide one, add a line like this to it:

sup, sub {line-height: 0.1em ; font-size: 1ex ;}


 * The default style sheet of my browser (Firefox 1.0.4) doesn't add excess leading when superscripts are used, so this must be a browser-specific problem. It should be pointed out that the Unicode Consortium and the W3C recommend using superscript markup rather than superscript characters because having two ways to represent the same info causes problems for search engines and their users.  There is not a superscript version of every character, so you have to use markup sometimes, e.g. 10n or s−1. Therefore, we might as well use markup all the time and be consistent. Indefatigable 19:58, 25 August 2005 (UTC)

One question: where do you find your user stylesheet? (you can answer on my talk page if you want.)

I wanted to refer only to units here, so mathematical equations et cetera are a different story. Just units.

Now, about the -1, this ties in with something I wanted to discuss but gave way in order to speed things up: representing division in units. I wanted the guide to recommend using a forward slash (what was its proper name? solidus, was it?), so instead of m s-1, use m/s. If we do that, there would be no problem (or at least, I can't think of any units that use something to the fourth power).

On that division thing, the complete line I wanted to add was to recommend using a solidus, and, where more than one unit is below the division line in e.g. newtons per amperes per metres (which is actually teslas if I remember right) would be N/(A m) except with thin spaces. Anyway, that's another issue, but the solidus part ties in with this I guess. Neonumbers 08:11, 1 September 2005 (UTC)
 * The farad (1 F = 1 C/V) is probably the most common unit with a power of four. It can be expressed in SI base units like this:


 * The middle dot is preferable over a (thin) space. The first or third variant of the first or third row should be recommended IMHO. — Christoph Päper 15:22, 1 September 2005 (UTC)


 * I'd say anything with two or more slashes and no parentheses is unacceptable. Throw the first row out.
 * The first set of parentheses in the third row is superfluous. No need to encourage that.
 * A horizontal bar, no matter how you make it, doesn't work well in running text and should be limited to offset formulas. The table method is clumsy and far too complicated to try to explain and encourage, and would appear to be difficult or impossible to set up so that the number appears separately before the units in this text as opposed to math markup version.
 * Math markup is often ugly. But a bigger problem when units of measure are involved is the difficulty in getting editors to add either the \mathrm you have used or \mbox to get the symbols upright rather than italic.
 * Mixing characters and html superscripts in the same unit is ugly. Mixing them in the article as a whole is less of a problem.
 * I assume that the preference Neonumbers has for the character superscripts and the solidus (obviating the use of negative superscripts) has to do in part with that notion being more familiar to most people.
 * However, one major advantage of the negative superscripts method is that it is unambiguous. The order in which the units are placed doesn't matter.
 * We already have far too many "J/mol K" and similar undisambiguated units in Wikipedia (in normal mathematics rules, multiplication and division are equal in priority and proceed left to right absent grouping, but that is not what is intended here). Can Neonumbers come up with a way to ferret them out and fix them?
 * I don't see any big need to prefer either km&sup2; or km2 as a general rule in the simple cases such as the area of a country or its political subdivisions. Gene Nygaard 17:01, 1 September 2005 (UTC)
 * I agree in general, although I kind of like multiple slashes. I've thought about yet another possibility: A²·s⁴&frasl;kg·m² That's not a very good one, though. There are Latex packages to ease the handling of units, e.g. SIunits or fancyunits, but they're not installed on the Wikipedia servers. &mdash; Christoph Päper 12:16, 3 September 2005 (UTC)

Well, now I know that there is a unit that uses the fourth power.

With regard to ambiguity: The SI brochure specifically details how to resolve this. It says, for example, the multiple solidi are never to be used (i.e. never A²·s⁴/kg/m²), because this, as well as any multiplication after the solidus (e.g. A²·s⁴/kg·m²), is ambiguous. In any case where there is more than one unit with a negative exponent, i.e. after the solidus, brackets are to be used, so it should be A²·s⁴/(kg·m²) &mdash; or, of course, using negative exponents. On the chemical infobox, I think last time I checked parentheses were used (as an example, and I'm too lazy to check again now), so I think it was J/(mol·K).

I quote from The International System of Units (SI), the manual:
 * The solidus is not followed by a multiplication sign or by a division sign on the same line unless ambiguity is avoided by parentheses. In complicated cases, negative exponents or parentheses are used to avoid ambiguity.

You're correct, my preference for the solidus is partly because people are more familiar with it. (Personally, I like negative exponents, though I don't think should they be used here.) My preference, however, also comes back to the superscript numeral characters: the superscript html tags (at least in my IE7 browser) make the lines uneven, the use of solidi would eliminate their need. (On that note, does anyone know of any units that use a fourth power, like the farad does, but doesn't have its own name, like "mole per cubic metre"?)

My focus with the superscript numerals is actually with smaller cases like area and volume, rather than complicated cases. So for m², that doesn't cause uneven lines (= extra leading?), whereas m2 does. They also look significantly different. This is significant because (here comes my rant about consistency) this is one project by one team and in something like this should have one way. I'll refrain from reciting the extended version of my consistency rant. Neonumbers 13:20, 3 September 2005 (UTC)


 * The chemical infoboxes have been sporadic corrections over time, and finally probably some templates with correct J/(mol·K).  Most of the elements are probably now correct; many chemical compounds still have "J/mol·K" (see dinitrogen tetroxide, an often used and often linked to article because it is a rocket fuel, benzothiazole, many more).
 * Some moments such as area moments of inertia (SI units m4) involve units to the fourth power.   Note that the dimensions here are simply length to the fourth power.  Thermal radiation depends on temperature to the fourth power (see Stefan-Boltzmann law).  Gene Nygaard 14:48, 3 September 2005 (UTC)


 * As far as the different looks go, my tired old eyes prefer being able to distinguish squared from cubed. So I like the superscripts better.  Gene Nygaard 15:04, 3 September 2005 (UTC)


 * Maybe I'm too stupid to see it, but how is anything like A²·s⁴/kg/m² ambiguous? J/mol·K isn't ambiguous either, it's just plain out wrong when J/mol/K (= J/(mol·K)) is meant. Christoph Päper 08:44, 6 September 2005 (UTC)
 * Because it could mean either (J/mol)·K or J/(mol·K)? — Omegatron 14:14, September 6, 2005 (UTC)
 * No, J/mol·K (= J·K/mol) cannot mean J/(mol·K)! Every 5th-grader gets this right, sadly some people with university degrees don't. Units shouldn't be context-sensitive. Christoph Päper 20:03, 6 September 2005 (UTC)
 * Right. But even better is to avoid using "/" at all. No way J·K·mol-1 will be confused with J·K-1·mol-1. (No matter how we code its writing) Nabla 21:42:55, 2005-09-11 (UTC)


 * It's ambiguous because, as Omegatron said, a reader wouldn't necessarily know for sure that the author doesn't have a misconception such that the author thinks he's writing A2&middot;m2&middot;s4/kg, instead of A2&middot;s4/(kg&middot;m2). As Neonumbers pointed out, in SI, only one solidus is allowed, unless you enclose the denominator in parentheses. Adding the parentheses is easy. Regarding superscripts, I prefer the html &lt;sup&gt; superscripts instead of the special character superscripts. As Indefatigable mentioned, using &lt;sup&gt; tags is more consistent and portable, and more logical. And, I usually prefer a solidus rather than a negative exponent, like Neonumbers. --Simian, 2005-09-28, 03:39 Z

Eyesight saving says use &lt;sup&gt; tag. Simplicity of editing says use &lt;sup&gt; tag (that is, I have no ² key on my keyboard). Anti-rule-creep says why have any rule. Consistency says pick either one, but only one. Therefore, if anything, I support using &lt;sup&gt;, otherwise, lets just leave it out. Chuck 16:05, 6 September 2005 (UTC)
 * The (US) English keyboard layout, which I assume you are using, is very supoptimal indeed. Most other Lating ones have ², often at AltGr+2. Christoph Päper 20:03, 6 September 2005 (UTC)


 * I haven't been here for about a month because I've been caught up in other things and haven't had time to participate in Wikipedia. I probably won't be able to check this discussion again for another while &mdash; but I want to say this now that I've read the discussion that's followed my last comment.
 * It seems as if everyone apart from me and Christoph are for the tags, and that no-one seems to consider the extra leading involved important. I would very much like to point out the leading that 2 causes &mdash; note the gap between the line with the superscript and the one above it.  I'd also like to point out that the ² character is very conveniently placed as a link below the edit box, alongside ¹ and ³ and also all the other §±×źÂ.  So it's not really that much harder to enter, and you're not doing it that often.  But I can't argue against the sup tags being easier to read; they do, after all, make bigger text.  My eyes have no problem with it, but then again that's me.
 * I'm quite reluctant personally to accept sup tags, but I'll have to if everyone apart from me wants them standardised. If I don't manage to find time, and you all decide to standardise sup tags and put it like so on the manual, I won't complain &mdash; much.  Neonumbers 08:31, 3 October 2005 (UTC)

Dates linking convention currently ludicrous
We link to dates like this Month Day Year. The reason we do this is because there is a bit of software magic that makes dates so linked automatically display for logged-in users in their preferred format.

Unforunately this convention makes the links themselves almost complete useless. Have you tried whatlinkshere for your birthday? You get thousands and thousands of links across hundreds of years. For relatively recent dates (say the last hundred years?) it would be much much better to link to Month Day Year. Then whatlinkshere suddenly becomes a massively useful resource for finding out what when on when in history.

Ok the display preference software will be broken temporarily. This is no big deal:
 * 1) If people really care, then a developer can be cajoled into making a relatively simple change to the data-matching regex.
 * 2) Display preferences of logged-in users are secondary at the moment to creating a quality resource.

I am open to suggestions on what content should go at the target of each day. Recent dates can redirect to the relevant part of the archived current events page. Pages refering to pre-2002 will in time grow as people examine the what links here for that page. Initially they should perhaps be a small page pointing to the year and day. E.g. September 3 1939 will point readers at September 3 1939.

I'll await some comments before changing the policy page, but I see this as a big step forward in how we present our information and hope you'll support it. Pcb21| Pete 13:46, 25 September 2005 (UTC)


 * As correctly stated above, the current linking allows logged-in users to change the display format of dates. This is a most useful feature, that should not be given up. Linking as is allows to look up events in a year (having some historical coherence) and events on a month+day (meaningful for commemorations). Looking up all articles that contain a certain precise date, does not look particularly useful to me. &minus;Woodstone 14:24, 25 September 2005 (UTC)


 * Looking up a particular anniversary or year would only ever be one more click away than it is now. I think you are being short-sighted in thinking that linking to a particular day is not useful. We already have about 200 words on every single day since mid-2002 as part of the current events archiving process. As time goes on, that archive is becoming more and more useful - but we should make it easier to access.


 * As for your point about display preferences - I already said it would be an easy software change to extend to work with full dates. We should not let the tail of a few logged-in users wag the dog of making content accessible. Pcb21| Pete 14:53, 25 September 2005 (UTC)


 * If you wish, you can link to full dates in YYYY-MM-DD format in a single link and have the date preferences work, look at 2005-09-25. If you want to go changing split links to this unified format, you could do so, although i doubt it is worth the trouble. DES (talk) 15:00, 25 September 2005 (UTC)


 * A few comments
 * I'd preferences option to be independent from the linking process.
 * We already have things like 1997 screwing up the commas in the date formats.
 * If all full dates are linked as such for preferences purposes, we need an option to suppress display of the year (something like piping in links, but not screwing up the preferences display)
 * Sensible options for a range of dates would also be nice; neither August 17-23 nor 17-23 August work in the commonly used preferences, forcing us to clutter up the display without using the piping: August 17-August 23 or 17 August-23 August
 * We still need preferences links for partial years, even if the year can be suppressed in specific dates. For example,
 * Christmas comes on 25 December every year.
 * As is, we also have too many links to unnotable years and months standing alone, in situations where preferences have no effect on the display. Gene Nygaard 15:19, 25 September 2005 (UTC)


 * Another thing on my wish list would be the ability to display only the first three letters of the month. This would be especially useful in tables.
 * The use of ordinals such as 31st August 1965 is another problem that should be dealt with. I can get by fine without them, but articles which use them often have inconsistent date formats, no matter what your preferences setting.  Gene Nygaard 15:23, 25 September 2005 (UTC)


 * To DESiegel - that format seems to do the opposite of work! You are forced into the poor linking structure (i.e. linking to anniversary and year separately) even when you explicitly don't want to be. I agree with much of what Gene has said, particularly about having pointless links to unnotable years simply to enforce preferences. Yes there a lot of preferencing issues we could deal with... but more fundamentally we need to separate out display preference and linking completely. Then linking comes back to what in should be about - creating a network of articles. Preferences working on regexes independently of links could then get as funky as the community and the developers desire - I am aware however that the software will not change overnight, so we should push on linking to full dates where appropriate and the software will catch up as it has done in so many ways before. Pcb21| Pete 15:27, 25 September 2005 (UTC)
 * I would myself prefer that there was a way to tag a date for preference formattign without linking it -- the two are really separate functions. But that must wait on a feature beign added to the software. In the meantime, I do avoid and remove links to years when these aren't needed for preference formatting (since a year without a day and month won't be preference formatted anyway) and have no particualr relevance to the articvel at hand (which is the normal case). But i can't see intentionally breaking the current preference system as a way to encourage the develiopers to work on a new system. DES (talk) 05:36, 26 September 2005 (UTC)


 * It's a method that's worked for the last three years ;). Pcb21| Pete 11:12, 26 September 2005 (UTC)
 * Yes it ahs worked, and it does work. But does it work well? To be more exact, it works but has certian undesireable side effects, including the ones detailed here, and including the encouragement to link years when there is no good reason to (not even to suppor the preference mechanism. Other forms of wiki markup are not done with links, why should this one be? why not soemthing like <<26 September 2005>> where anything in the <<>> would be recognized as a date and formatted according to user preferences? I don't say this is vital, i do say it would be an improvement. DES (talk) 14:34, 26 September 2005 (UTC)


 * Linking years does in fact support the preference mechanism.
 * A corollary, as I pointed out above, is that links such as 1999 and 1987 screw up that support of the preferences mechanism:
 * One of these, either April 13, 1999 or 7 May 1987, will be wrong for most people using preferences (except those actually like the way the editors of the American edition of the Guinness Book of Records do it, using that silly "April 13 1999" format). But even for them, they will then see the normal links without "year in ____" as screwed up.
 * Compare the same one without the "year in ____" link:
 * April 13, 1999 or 7 May 1987
 * But without any year link, it's just like with the "year in ____" link
 * April 13, 1999 or 7 May 1987. Gene Nygaard 15:55, 26 September 2005 (UTC)


 * Note also that it is even worse for those using the "2001-01-15 16:12:34" preference; both dates are screwed up, unless there is a link to the unadorned year without piping. Gene Nygaard 16:04, 26 September 2005 (UTC)

I agree that there should be a change to the date preference mechanism. As I have said before, create date-object should be done with a method different to that for real linking. As examples of how ludicrous the linking has become, dates:
 * rank highly on the list of articles with multiple links to the same article. The top 125 are all dates. There are pages with 1000 links to 2000.
 * rank highly on the list of most referenced articles
 * rank highly on the list of most incoming links

There is a widespread false belief that all calendar elements must be linked each time they occur. So people link solitary elements such as Tuesday or January or 2001 even if they are repeated many times in the same article. These have almost zero added value. I sometimes remove such links but the response is often an immediate revert. Overlinking of solitary date elements is one of the silliest things in Wikipedia articles. I wish the guidance in the Manual of Style could be more explicit.

Date preference formatting deals with two issues with distinctly different priorities:
 * Ambiguity. Digit only dates can be misinterpreted. For example '6/2' could be 6 February or June 2. This cause miscommunication and so for any solution is a 'requirement'.
 * Style. Word dates cannot be misinterpreted but some people prefer certain formats. For example some people like '6 Feb 1802' and others like 'Feb 6, 1802'. I could live with a Wikipedia that did not modify word based dates. I am sure that others could too. But for some people it seems important. It is a 'nice to have'.

Please see a discussion that I raised a while back in Wikipedia:Village pump (proposals)/Archive (archived) involving users such as Tim Starling, Thryduulf and Cyrius. Bobblewik 18:10, 26 September 2005 (UTC)


 * ISO 8601 (yyyy-mm-dd) is unambiguous and is an international standard. It's been used for almost two decades on usenet and no one has any problem with misinterpreting or understanding it. New users quickly learn the simple, logical, unambiguous format. There's no ambiguity with 2005-06-02 because it's an international standard. One encounters unnecessary additional complexities by introducing ambiguities, instead of just using the simple, international standard already in place worldwide. Using ISO 8601, instead of a plethora of arbitrary formats and personal nuances, will eliminate a significant portion of problems in automation. It's also the only choice that's indiscriminate to any particular group, because it's already an international standard. --Simian, 2005-09-28, 01:07 Z


 * It is a standard. But it is not a standard that applies to usage such as Wikipedia display of dates.  Gene Nygaard 02:50, 28 September 2005 (UTC)

Numbers in words
This is quite a complicated topic that would need extensive discussion here. Setting guidelines on this matter would require much discussion on elaboration on many exceptions - See http://webster.commnet.edu/grammar/numbers.htm for an example. I think any guidelines should be as simple as possible with as few exceptions as possible. Zero to ten suits me fine, and there are still plenty of exceptions within that. --JimWae 20:00, 25 September 2005 (UTC)


 * User:Babajobu started a discussion of this at Village pump (policy), now basically dormant, which should be moved here, or at a minimum summarized here. Gene Nygaard 20:09, 25 September 2005 (UTC)


 * Summary is that there was a lot of resistance to a guideline suggesting that numbers under one hundred be spelled out in non-scientific articles. I had thought that book style, with the general rule of spelling out these numbers, would be more appropriate for Wikipedia than newspaper style, in which only numbers under ten spelled out. But it seems that there are more Wikipedians for newspaper style than for book style. One thing that's clear from the Village pump discussion is that any guideline must emphasize the distinction between counting numbers and numbers in measurements. Babajobu 20:40, 25 September 2005 (UTC)


 * Wikipedia is a co-operative (more or less) project among people with different styles. A single person, or a group with a clear line of command, can decide which elements from Chicago they want to use & be consistent without needing to endlessly enforce it (those projects come to an end - whereas wikipedia apparently never does). There will never be a permanent consensus to accept any guidelines like this that are not already nearly-universally recognized. Keep all guidelines simple & widespread & easy to follow. --JimWae 20:47, 25 September 2005 (UTC)


 * I feel like this discussion is basically a solution looking for a problem. Does anyone really care if I write ninety-eight or 98? (Obviously they do, I was just being rhetorical.) Chuck 17:47, 26 September 2005 (UTC)


 * As you suggest, plenty of people do care. Some people criticise digit format because it is not 'proper' style according to whichever styleguide they prefer. However, nobody has yet suggested that digit format reduces comprehension. In the case of references to centuries (e.g. 7th century), digit format is the default.


 * Digit format for centuries is the default on Wikipedia, not elsewhere. Babajobu 14:38, 27 September 2005 (UTC)


 * Google "20th century" 70,400,000 hits
 * "twentieth century" 38,100,000 hits
 * Gene Nygaard 15:12, 27 September 2005 (UTC)


 * Good style is not an internet popularity contest. Babajobu 16:19, 27 September 2005 (UTC)


 * It is also a matter of language independence. Digit format is more language independent and that is important to Wikipedia which, unlike many publications, is an online international resource. Native readers of any individual language in Wikipedia should be prepared to accept a little style unfamiliarity in exchange for greater access to non-native readers of that language. Bobblewik 09:50, 27 September 2005 (UTC)

I didn't know there was an on-going discussion about this. I changed the 16th to 21st centuries to number in words format. Phil objected on my talk page. Uncle Ed 16:33, 28 September 2005 (UTC)


 * Well, this really depends on the context doesn't it? For example, as someone else pointed out, 10th century is the more common form for history, but Tenth Amendment to the United States Constitution is the prefered form for amendments. You really have to bring a sense of the cultures and/or subcultures being represented by any page to Wikipedia, and since numbers are so primative to most cultures, I'm not sure that you can resonably make a blanket rule. I would propose the following guideline:
 * In reference to a specific historical event, date or fact, use the convention that is most commonly applied.
 * In reference to mathematics or measurement use digits (e.g. "1+1=2" not "one + one = two" and "20 meters", not "twenty meters").
 * When a link is required, and the link has digits, don't "pipe" the link to re-name it (e.g. not ).
 * If a number has a fractional or decimal component express it as digits (e.g. "3.141" not "three point one four one") except where traditional usage is spelled out (e.g. "half & half cream" not "1/2 & 1/2 cream").
 * For 100 or greater, always use digits.
 * For 10 or greater, use digits when specifically refering to the number itself (e.g. "after 22 comes 23") or when a seqence of three or more numbers is required (e.g. "we skipped 50, 51, 52, and 53").
 * When cultural context demands digits or words, use what is appropriate.
 * In all other cases, use words.
 * Seem reasonable? I think this ends up meaning roughly the same thing, but with obvious exceptions that are common usage accounted for. -Harmil 13:46, 6 October 2005 (UTC)