Wikipedia talk:Manual of Style/Dates and numbers

1/3rd, 1/4th, 1/5th, 1/6th, ...

I'm often seeing 1/3rd, 1/4th, 1/5th, 1/6th, ... I wonder whether this is worth a mention. JIMp _talk·cont 18:12, 29 July 2009 (UTC)

I think that use is incorrect. Bubba73 (talk), 23:00, 7 August 2009 (UTC)

There are times when something like that, although not in customary style, might avoid ambiguity or obscurity because a diagonal slash can mean many different things, e.g., the 1/3 Blankshire Fusilliers, 4/4 time, 135/80 systolic over diastolic blood pressure or shillings and pence as in 2/6. However, "one-third", "three-quarters", etc., would probably serve just as well in many of those cases. —— Shakescene (talk) 00:08, 8 August 2009 (UTC)

Then in such cases let it be "one-third", "three-quarters", etc. These are incorrect. JIMp _talk·cont 00:23, 8 August 2009 (UTC)

Question: changing prose to measurements

Is converting prose descriptions of approximate distances to numeric measurements in line with this page's guidance? A specific example is this edit. I understand the impulse here, and the new text is certainly not incorrect, but it seems to change the tone of the prose. Powers ^T 13:31, 30 July 2009 (UTC)

Normally whole numbers from zero through 9 are expressed in words, not numerals. The conversion would be in numerals. --Jc3s5h (talk) 15:34, 30 July 2009 (UTC)

I believe this page indicates that measurements are usually expressed in numerals, and the change I linked does follow that guidance. My question, though, is whether changing "just over a mile" to "just over 1 mile (1.6 km)" is desired practice. Powers ^T 17:44, 30 July 2009 (UTC)

Upon further examination, I see that the MOSNUM has contradictory guidelines:

• Do not use spelled-out numbers before symbols for units of measurement: write five minutes, 5 minutes, or 5 min, but not five min.
• Measurements, stock prices, and other quasi-continuous quantities are normally stated in figures, even when the value is a small positive integer: 9 mm, The option price fell to 5 within three hours after the announcement.

Perhaps whoever wrote those bullet points mistakenly thought that minutes were not continuous, and that only whole minutes could be counted. --Jc3s5h (talk) 18:12, 30 July 2009 (UTC)

I wrote the first of them, and no, I don't believe that only whole minutes could be counted. But I guess that "measurements" in the latter does not actually mean any dimensionful quantity, but just stuff which is actually measured, not ballpark estimates such as "just over a mile" or "I'll be there in five minutes", or metaphorical statements such as "One inch of love is one inch of shadow" or "The snake is long seven miles". (The bullet was added by Pmanderson which is now banned from talk pages of style guidelines, but I'm going to ask him on his talk page.) --A. di M. 20:05, 31 July 2009 (UTC)

I think that it's best to indicate when something's an estimate or approximation (as by using words or the indefinite article), so I wouldn't change "a mile" to "1 mile" unless some 1-mile boundary, limit or radius was involved (e.g., the Four-minute mile). On the other hand, in Wikipedia, some consideration has to be given to those in metric countries and we should not require them to do conversions in their heads. Only the original author could really say whether it would be better to say "(about a kilometer and a half)" or "(about 1.6 km)". —— Shakescene (talk) 22:03, 31 July 2009 (UTC)

I wouldn't say that only the original author could really say what the best conversion is. In fact any who has access to the sources is in just as good a position. Generally speaking "a mile" is probably best converted to "a mile (1+1⁄2 km)". JIMp _talk·cont 04:04, 1 August 2009 (UTC)

PMAnderson told me that the reason why he doesn't consider minutes to be continuous is that you'd normally say "9.30 minutes" but rather "nine minutes and eighteen seconds". But I don't think that this is the only point: I'd not use numerals in "I walked about two kilometres this morning" if that's a rough guesstimate (as opposed to a measurement). As for the article, I'd go with "just over a mile (about 1.6 km)". --A. di M. 17:56, 4 August 2009 (UTC)

Minutes are not continuous; time is. JIMp _talk·cont 23:29, 4 August 2009 (UTC)

Thanks for the input, everyone. I've changed some of the {{convert}}s back to prose but added metric conversions as appropriate. Powers ^T 12:50, 5 August 2009 (UTC)

Bot to unlink dates?

Why not have a bot to unlink dates, since that is the policy? Is it because a few should be linked and the bot can't decide which to keep? Bubba73 (talk), 19:46, 31 July 2009 (UTC)

Something like that. Powers ^T 19:53, 31 July 2009 (UTC)

Actually, there is a bot in the works (approved by the community); see WP:DATEBOT. In a nutshell, only dates that are used for autoformatting will be delinked; date fragments (such as March 3 or 1999) will be left linked. There is an exclusion list, to which editors can add articles that they don't want to be delinked if they can explain why both the year and the month-day fragments are useful. Dabomb87 (talk) 19:55, 31 July 2009 (UTC)

What we actually need is another code to format dates (i.e. not link them). Then the bot could make the replacement. Bubba73 (talk), 22:58, 7 August 2009 (UTC)

Can we have a bot which works out which is the currently predominant date format in a particular article, and then converts any other dates in that article to that format? Alarics (talk) 16:07, 10 August 2009 (UTC)

There was discussion about that, but it was rejected as it was outside the scope of the bot task that was approved by the community. Dabomb87 (talk) 18:17, 10 August 2009 (UTC)

So how do we get something done about it? Alarics (talk) 18:41, 10 August 2009 (UTC)

Start another discussion about it, I assume. The bot can do it eventually, but the community has to approve that task first. Dabomb87 (talk) 18:48, 10 August 2009 (UTC)

Numero sign / Number sign

In reference to Archive 106, Question on #1 vs number one, should there be a section in MOSNUM outlining the proper usage of the numero sign versus number sign?

Another undiscussed usage is in a table header (where space is limited): Is it proper for one to write "№ of shows" or should one write "Number of shows"? See the example in the header of this table. —Preceding unsigned comment added by Gmoose1 (talk • contribs) 14:59, 7 August 2009 (UTC)

This is worth developing. The hash sign (#) or as it is often called in North America the pound sign is still somewhat of an annoyance to Europeans (and you can imagine my annoyance with calling it pound, being an english person). In England it is common to write N^o, and has been since at least my copy of Johnson's Dictionary which is er 1753? Somewhere like that. Considering the differences, for a worldwide, but English-reading, audience would it just be easiest and simplest to write "The number (or numbers)". But take e.g. house numbers should we write oh house 22 or number 22? Both I think would be acceptable in British English. (A children's programme in the late 70s hosted by Sandi Toksvig was called No. 73). I can fill in all the sources details etc later but just gave me pause for thought. SimonTrew (talk) 15:51, 7 August 2009 (UTC)

Number Ten, Downing Street, for example ("No. 10"). Keep in mind that on US keyboards, the # sign is usually present, but № (which I just pulled from the Symbols menu below the edit box) is usually not. Nor are superscript a and o (or for that matter superscript th). But as an Englishman who came to the States at age 11, I still think # is a bit unnnatural in some contexts. (It's also slightly ambiguous because in old-fashioned commercial shorthand in inventories, invoices, etc. @ can stand for "each" and # for "pounds", among other things.) —— Shakescene (talk) 21:40, 7 August 2009 (UTC)

Yeah exactly, I think it is never going to be pleasant to everyone to say #10 or N^o ten or whatever. Why don't we just reccommend saying e.g. Number 10 (for downing street?), of course if the context makes it clear it means the residence of the UK PM (or metaphorically his office or powers, as in say the White House, by analogy). I just think it would be the most simple and also more worldwide to say that. SimonTrew (talk) 12:00, 8 August 2009 (UTC)

I've never been a fan of the pound/hash symbol, except perhaps in tables and other environments where space is particularly tight (even then, I'd avoid if possible). I don't much like the squishy little "№" symbol from the edit-tools, as Shakescene has presented it above. Tony (talk) 12:59, 8 August 2009 (UTC)

So to form a rule for the manual of style and summarize the above, we could say "Number of shows" if there is enough space, but if space is tight (eg: in a table), then one should write "№ of shows"? Not sure if we're ready to form a rule for house numbering? As for the symbols that should be used: I think both N^o and No. should be avoided as № is readily available in the Symbols box and a wiki article is available to counter ambiguation. Furthermore, # should be strictly avoided in this context as it causes confusion for British English audiences. American English audiences recognize # and No. have the same meaning. Any disagreements? Lars (talk) 16:31, 10 August 2009 (UTC)

Two concerns with that. First, № appears archaic to most American eyes, like something you'd find on a 1910s-era cash register. Second, in fixed-width fonts such as that used in edit windows, № is so squished as to be nearly unreadable. Powers ^T 17:50, 10 August 2009 (UTC)

I have never heard of "№". What we usually put is "No.". Alarics (talk) 18:40, 10 August 2009 (UTC)

"№" is indeed difficult to read in monospace, and may be enough to argue that we should go with "No." or "N^o". It is also not commonly used (like ‽), which may support the argument that it appears archaic. So, how about "No." versus "N^o"? Any thoughts? The latter is more specialized, and may improve readability as it does not appear to be the word "no" at first glance. Lars (talk) 19:33, 10 August 2009 (UTC)

Anomalies in the template for converting measurements.

I have discovered an anomaly with using the 'convert' template. Look at this:

4,700 square miles (12,000 km²)

4,699 square miles (12,170 km²)

When converted into square kilometres, 4699 sq miles comes out 170 km² more than 4700 sq miles. Lose one square mile, gain 65.6 square miles!

I think the problem may occur because of rounding when a number ends in two zeros.

Therefore it is risky to rely blindly on the conversion template. It may give anomalous results.

Please let me know if this issue should be raised elsewhere. Michael Glass (talk) 06:45, 8 August 2009 (UTC)

{{convert|4700|mi2|sigfig=4}} gives 4,700 square miles (12,170 km²). You just have to use the significant figures setting. For many articles that rounding is appropriate; if you write 4,700 nobody is going to expect it to be exactly that. If you write 4,699 they will expect it to be exactly that. SimonTrew (talk) 12:07, 8 August 2009 (UTC)

Well, in my humble view, this is another reason not to use a conversion template, but to allow fellow editors, including newbies, maximum control over all aspects of the construction. I'm all for keeping it simple, and if that means using a calculator, so be it. Tony (talk) 12:57, 8 August 2009 (UTC)

I disagree. If a newbie just throws the numbers in, a more experienced editor can use {{convert}} later. The thing about using the template, if e.g. something changes in MOS then EVERY article that uses it will get that change. SimonTrew (talk) 13:45, 8 August 2009 (UTC)

Well I do agree to the extent that it can be fiddly to get right and there are some things it simply can't do. But on the whole, if you can do it with the template, you should. SimonTrew (talk) 13:53, 8 August 2009 (UTC)

I think the trouble is that as the rounding cuts in automatically unless it is overriden. The default position can be a trap for the unwary. An over-precise conversion can be overridden; the opposite may not be noticed by the uninformed.Michael Glass (talk) 14:08, 8 August 2009 (UTC)

The template reads the precision of the input number and matches the precision of the output accordingly. This is in accord with the MoS and normal mathematical practice. There is no anomaly. An over-precise conversion can be overridden, sure, but are the uninformed any more likely to fix an over-precise than an under-precise one? In the majority of cases the template will not give the wrong precision. 4,700 sq mi ≈ 12,000 km² is correct as is 4,699 sq mi ≈ 12,170 km². It is the conversion templates without this type of default rounding which are more prone to produce output with incorrect precision. To those unwary I can only say "Get wary." JIMp _talk·cont 15:11, 8 August 2009 (UTC)

It's a bit of a circular argument that it conforms with MoS since this is a discussion on a subpage of the MoS.

I do agree however that yeah if you put in 4,700 it is unlikely it is going to be exactly 4,700 (of whatever) and so misleading to a ridiculously over-precise value. If you pur in 4,699 presumably you mean that (and not 4,698 or 4.701) and should be converted more accurately. It comes down to common sense. My little book here gives logs and other things such as trig functions to 8 sig figs. Most floating point maths used in computers is at double-precision and has about 12 sig figs (decimal) in the mantissa. If it is good enough for rocket science and subatomic modelling (and it is cos I have programmed it) then I think it ridiculous to be more precise than that. There are all kinds of more precise representations but in real science and engineering that is plenty: in fact single precision (5 or 6 sig figs decimal) is usually good enough, but with modern processors using a double is as fast, if not faster, than a single. And the same applies to humans as computers. Of course propagation errors may occasionally require higher representations in intermediate results, but on Wikipedia that hardly applies, unless the underlying templates are so off that they preduce an obviously bizarre result such as 1 cm = 1.00001 cm or whatever.

I see no problem here. These templates are a bit fiddly and I am sure if we were starting from scratch we would change a lot of things, but we aren't and we can't. It is one of those things an editor just has to do. The {{cite}} templates are also trick but everyone expects those to be used. SimonTrew (talk) 04:55, 9 August 2009 (UTC)

If we put the template in question up against a statistically representative sample of typical editors with calculators, I'd put my money behind the template to give the most appropriate levels of precision for the typical measurements you find on Wikipedia. JIMp _talk·cont 08:06, 9 August 2009 (UTC)

The behaviour contradicts the principle of least surprise, but as a mathematician I think this is one of the situations where that is actually justified. Besides, I don't see how we can avoid it. Consider:

0.47000 square miles (1.2173 km2)	4.7000 square miles (12.173 km2)	47.000 square miles (121.73 km2)	470.00 square miles (1,217.3 km2)	4,700.0 square miles (12,173 km2)
0.4700 square miles (1.217 km2)	4.700 square miles (12.17 km2)	47.00 square miles (121.7 km2)	470.0 square miles (1,217 km2)	4,700 square miles (12,000 km2)
0.470 square miles (1.22 km2)	4.70 square miles (12.2 km2)	47.0 square miles (122 km2)	470 square miles (1,200 km2)	4,700 square miles (12,000 km2)
0.47 square miles (1.2 km2)	4.7 square miles (12 km2)	47 square miles (120 km2)	470 square miles (1,200 km2)	4,700 square miles (12,000 km2)
0.5 square miles (1.3 km2)	5 square miles (13 km2)	50 square miles (130 km2)	500 square miles (1,300 km2)	5,000 square miles (13,000 km2)

Notice that in the last column we always write 4,700, whether we mean 2, 3 or 4 significant digits. Similarly, in the penultimate column it's not clear whether 470 has 2 or 3 significant digits. The template needs to guess. We can't make it guess correctly in all situations, but if it makes sure to under-, rather than overestimate the number of significant digits in doubtful cases, then it's more likely to be wrong right than if it does the opposite. And even if the template gets the intended number of significant digits wrong, then except in situations where a human reader can infer it from the context it is usually correct and encyclopedic to round the numbers.

But there is an unrelated anomaly in the top right cell of the table. I am taking this to Template talk:Convert, because it's clearly a bug that needs fixing. Hans Adler 09:40, 9 August 2009 (UTC)

Do we really need this whole conversion business at all? As stated above, the conversion is not 100% reliable and it's not "official info" anyway. Maybe we should just use English units in American/British articles and metric units everywhere else. Offliner (talk) 10:29, 9 August 2009 (UTC)

But this is the English-LANGUAGE Wikipedia, not tied to any one country.

In the UK and Ireland metric is commonly mixed with Imperial units in everyday life; in the UK some things such as road distances and heights, and pints of beer served on draight, MUST be measured in Imperial; in Canada metric is most commonly used, with occasional use of US Customary simply because of the close ties with the US; in Australia and New Zealand exclusively metric and so on.

Last night I was talking to someone from mexico and said that Cambridge was about 80 km from London, whereas I would never say that to an English person (I would say about 50 miles). Yes, the conversions need to exist. Again, if using {{convert}} and the MoS then says "don't use conversions except in this or that circumstance" we can probably change the templates and 90% of articles will immediately conform (the remainder being for things like historical use of units, or quoted sources, etc). So, even if the conversion is not particularly useful in itself, simply as a marker that "this is a measurement" is. I know {{val}} also stands for that purpose but the same applies, mutatis mutandis. It is also extremely helpful for people translating across different Wikipedia (what am I supposed to do, for example, if I translate a French article about an American car?) SimonTrew (talk) 10:51, 9 August 2009 (UTC)

To Hans: I would not write 4,700 if only the first two digits are significant: instead, I'd write 4.7 thousand; likewise, I'd not write "50" if only the first digit is significant (third cell of bottom row): I'd write "fifty" (or even "about fifty", if appropriate). And I'd like the template to assume that the zero in "50" is significant, for this reason. I once read about an editor converting numbers such as 187, 190, 191, and 194 and being surprised that the second one was converted with less precision. --A. di M. 13:12, 9 August 2009 (UTC)

Just noticed a bug: {{convert|470.00|mi2|km2}} gives 470.00 square miles (1,217.3 km²), with one 0 after the point instead of two. --A. di M. 13:25, 9 August 2009 (UTC)

I might say 4.7 thousand square metres too, though many don't, and the template has the capacity for a number of such constructions. However, if you see "4,700", the safest assumption is that it's precise to the nearest 100. As for the bug mentioned above, it should be fixed as soon as an admin puts in the new code. JIMp _talk·cont 17:48, 9 August 2009 (UTC)

Hum... maybe for 4,700 I agree, but in cases such as 5,000 I think that assuming that none of the zeroes is significant is excessive. When giving numbers with two sigfigs, there is a 1-in-10 chance that the second one is 0. So I'd assume that numbers such as 500,000 have two significant digits. What do you others think? --A. di M. 22:15, 9 August 2009 (UTC)

Which should generally boil down to giving an output of at least two significant figures which is just what the template does. So, yeah, I think that's just about right. JIMp _talk·cont 00:32, 10 August 2009 (UTC)

I think a template that automatically treats a zero as not a significant figure is a bit of a worry. Take the areas of the states of the United States. it would be bizarre to think that the areas of the states of Washington and North Dakota were less accurately surveyed than the other 48 states simply because the areas in square miles happened to end in two zeros [1]. Michael Glass (talk) 13:02, 10 August 2009 (UTC)

So, your first step is to compare these with the areas of other states and base your rounding on that. Without this comparison how else could you justify assuming the zeros were significant? Zeros may or may not be significant whether you use the template, a calculator or do the conversion in your head, you have to deal with this. With nothing else to go on standard practice is to regard the zeros as not significant (and generally they aren't). JIMp _talk·cont 13:42, 10 August 2009 (UTC)

Truncating the year (not in a date range)?

Another user has questioned my copy-editing of an article in which I reformatted some two-digit dates into 4-digit format (for example, in "from 1930 until -54" I changed "-54" to "1954", and in the passage "...in 1954 XYZ bought the blah-blah-blah, then XYZ merged ZYX into the QRS, which in -55..." I changed "-55" to "1955"). He points out (correctly, as far as I can tell) that nothing in WP:MOSNUM expressly states that, except for the closing date in a date range such as "2005–06", all four digits should be included when expressing a four-digit year. Apparently this user has been expressing years in a two-digit format whenever a four-digit year had previously appeared in the same paragraph.

I have a hunch that this issue isn't addressed in MOSNUM because the question never arose earlier -- contributors have assumed that years should be expressed using all digits, with the notable exception of date ranges (which have been discussed several times). Now that the question has arisen, I think that MOSNUM needs to be revised to clarify that every stand-alone mention of a year should use all of the digits in that year. Accordingly, I would like to append the following to the second bullet in WP:YEAR: "When listing a year outside the context of a year range, use all digits (1954, not '54 or -54)."

Please comment on this proposal. --Orlady (talk) 03:29, 10 August 2009 (UTC)

I can't see really how this comes under WP:YEAR at all, but in my opinion "from 1930 until -54" can then only mean to the year -54, which there never was one of, so balderdash. Or it is supposed to mean a dash, then it is simply essentially repeanting the word "until" (tautologically and ridiculously) and so it is a range, and does not conform to WP:YEAR. I think his argument boils to nonsense and you are correct. He is simply using a bizarre form of abbreviation for dates, which I don't think conforms. That's my opinion, and I would take all of those tacks to say his -54 etc is simply not conforming. SimonTrew (talk) 13:18, 10 August 2009 (UTC)

I agree. Legislating against this eccentric notation is about as necessary as legislating against a specific misspelling of a word. The relevant section is already too long, and we shouldn't make this worse because of a single editor. Hans Adler 14:03, 10 August 2009 (UTC)