Wikipedia talk:Manual of Style/Captions

Centered captions

Look ma, I'm centered!

Every so often I run into images whose captions are centered (in older cases with the deprecated <center>...</center>, or with {{center}} or something related). This has always struck me as a Bad Thing, and I generally remove it whenever I see it. I thought it might be worth adding a short note to the subpage here about that. As just a first idea, maybe something along the lines of: "Image captions should not be centered without a situation-specific reason." Thoughts? –Deacon Vorbis (carbon • videos) 14:14, 29 April 2020 (UTC)

Absolutely! Any thoughts on what an acceptable reason might be? Johnbod (talk) 14:18, 29 April 2020 (UTC)

Caption length and citations in captions

For the past few days I have had a discussion with User:पाटलिपुत्र regarding the image captions in the article on Ashurbanipal, which I worked on and got to GA level a few months ago. The main issue present here is that I believe पाटलिपुत्र's work on the images and captions in this article was excessive and unnecessary and since neither me nor पाटलिपुत्र is budging on the issue I feel like it is appropriate to attempt to get broader consensus for what the captions should look like.

This is what the article, as edited by पाटलिपुत्र, looked like prior to me changing things around again. The images are of wildly different sizes, many of the captions are long (several being six lines or longer) and (I would argue) contain excessive and unnecessary detail. For instance, the caption of the image in the section on Ashurbanipal's family, there to illustrate him and his wife depicted dining, goes into detail on every other thing in the mural. I argued that Wikipedia:Manual of Style/Captions states that captions should be "succinct" but पाटलिपुत्र argued back that they believed their captions were succinct (as the word does not necessarily mean "brief"). Granted, their latest versions of the captions (reverted by me) are considerably better in this respect, I still believe some detail in them was excessive and not relevant to the context the images were used.

The other issue is that पाटलिपुत्र wants to use citations in the image captions, arguing that Wikipedia:Citing sources says "Image captions should be referenced as appropriate just like any other part of the article". I argued that "as appropriate" could be considered a matter of opinion here, but that regardless of what the rule/guideline says, image captions are very rarely cited in the articles themselves. I pointed to the TFA:s over the last four days and how none of them had any citations in their image captions, even when the captions were long and contained information not immediately apparent from the image themselves. I also argued that citations and detailed info might be more appropriate for the commons page of the image rather than the article itself and that using citations in the captions to the degree that पाटलिपुत्र wants is actually detrimental to the article as it makes it stand out from every single other article in this respect. The image depicting Ashurbanipal dining had four citations alone for its captions in पाटलिपुत्र's original version (and two in their latest version).

This page states that "More than three lines of text in a caption may be distracting; instead, further information can be provided in the article body. And remember that readers wanting full detail can click through to the image description page" so perhaps others can weigh in on how much detail is appropriate in image captions and when they need to be cited (and if it is, as पाटलिपुत्र appears to interpret it, in every case, why is this never enforced during the GA or FA process?). Ichthyovenator (talk) 12:46, 27 July 2020 (UTC)

Rather to my surprise, I actually prefer the "bad" version you link above. Imo, both versions to some extent have the wrong information, or too much in the captions, but this is not in the descriptions, but the other stuff. Wikipedia:Manual of Style/Visual arts cautions against saying "Exhibited at the British Museum", "Exhibited at the Pergamon Museum", "Now exhibited at the British Museum" (all your version) because museum displays change - many of these photos were taken at the BM's recent exhibition. Are you even sure they are still on display? They have a lot of Assyrian reliefs that aren't. In the case of specific objects with a decent individual museum page, I think a citation link to that is certainly a good idea, and removes the need to give the museum accession number in the caption. In such cases it is wrong to say captions are "very rarely" cited, and you've quoted the policy support yourself. Most of the time, I don't even give the museum in the caption (some have such long names, and it is all in the image file). Here I might make an exception - can one just go to "BM"?. Btw, I think it is you who has introduced several WP:ERA-breaching "BCE"s into the captions - the article has always used "BC". There's a "center" I suspect is the wrong engvar too. I think the explanation of the details in the "garden party" relief is necessary, either in the text or in a caption. The relief is is not normally on display, but was in the exhibition, & even though I had read about it I did not spot the hanging head garden ornament until the display caption prompted me. The mixed sizes and layout in the "bad" version didn't bother me, except for a sandwich in the first section, on my settings. Several of your paragraphs are too long & should be split, which will help with this. I wonder if it is necessary to even say "BC" after the first few dates. Perhaps it is. Johnbod (talk) 15:27, 27 July 2020 (UTC)

I don't think I explicitly called पाटलिपुत्र's version "bad"; I just wrote out the things I did not like with it/was unsure of. पाटलिपुत्र introduced several changes which I found good immediately as well, for instance, they replaced several images with better alternatives and added Ashurbanipal's name in cuneiform to the lede. "Exhibited at ..." could be removed, yes. "In such cases it is wrong to say captions are "very rarely" cited, and you've quoted the policy support yourself": yes I quoted the policy but the issue is that I have yet to see another article where the captions are cited to the same extent as पाटलिपुत्र's version. I questioned whether enforcing this policy to such an extent in a single article was worthwile since no (few?) other articles do the same. I can see where you are coming from but in my experience this kind of citing is most commonly done on Wikimedia Commons, not in the articles directly (as पाटलिपुत्र pointed out this is however just my personal experience and not something I could explicitly find in the rules/guidelines). The most similar example I can find is Cleopatra where there are quite a few long and cited captions, but also many uncited ones. For another historical figure as an example, Julius Caesar has a lot of images but only 2 cited captions. If others think the citations are worthwile to include I'll of course be on board, though.

To my knowledge I was not the one who introduced "BCE" into the captions, no. I have consistently used BC throughout my work on the Assyrian kings. Note the version of the article that was GA reviewed (it has a single "BCE" in a caption; probably because I copied this image and its caption from some other article). On the detail on the garden scene; the head is an interesting detail in the relief but my argument was whether it was necessary to bring up in a section otherwise only about Ashurbanipal's family or if the caption should just focus on the part of the image that is relevant for that section (the king and the queen). I don't know which paragraphs you consider to be too long, and that has little to do with the image captions, but you are welcome to split any long paragraphs.

What do you mean by "both versions to some extent have the wrong information"? Ichthyovenator (talk) 16:40, 27 July 2020 (UTC)

I meant "the wrong information to include in captions" as in the examples I then gave. Johnbod (talk) 22:05, 27 July 2020 (UTC)

Oh, right. I was worried that you had found some factual errors in the text. Ichthyovenator (talk) 08:26, 28 July 2020 (UTC)

Reviving earlier discussion on citations in captions

@Johnbod: @Ichthyovenator: I note there is a previous discussion from 2018 (Do captions need a reference?), to which participated User:EEng, User:Ke4roh, User:Gronk Oz User:Domdeparis and User:Rp2006, and in which the consensus seems to have been that: "If the caption recites something which is in the article text, and sourced there, then no cite is needed in the caption. But anything stated in the caption only should carry a source". I agree with this statement, but it seems the discussion faltered at some point, and the draft proposal was finally not translated into policy. Could now be the right time to do so? Wikipedia:Citing sources already has: "Image captions should be referenced as appropriate just like any other part of the article", but a confirmation and possibly a bit more precision on Wikipedia talk:Manual of Style/Captions would be welcome. पाटलिपुत्र Pat (talk) 20:56, 28 July 2020 (UTC)

The "Paleologus and Beal" plantation on a 1685 map of Barbados, marked with a pineapple (to the left, below "Topp")

@पाटलिपुत्र: I actually agree with you that this is would be a good amendment to the policy. I also believe that simple captions (I have included an example from one of my articles) do not need to be cited at all, even if the information (in this case for instance that this map is from 1685) is not immediately obvious from the image itself. This goes along with the thinking that Wikipedia:Citing sources says that the caption "Belshazzar's Feast (1635)" would not have to be cited for the image File:Rembrandt-Belsazar.jpg, despite the year not appearing in the image itself.

I'm going to do a 180 and say that the level of detail in some of your captions in the Ashurbanipal article probably warranted citations (though I do not think all the captions needed to be that long and detailed). I do not think all the captions needed citations but longer captions for the garden scene and perhaps the palace and garden image for instance, with citations, is probably only a good thing. I also agree that more precision Wikipedia:Manual of Style/Captions in regards to what to include/what not to include in a caption and when to cite and when not to cite is a very good idea. Had there been more precision this entire argument would probably have been avoided. Ichthyovenator (talk) 21:26, 28 July 2020 (UTC)

Summarizing the discussion linked in the OP (and reading between the lines a little), there were three rough versions of proposed text (which probably needs to end of in WP:V, not MOS):

• Version 0: To the extent a caption recites something stated and sourced in article text, no cite is needed in the caption; but anything stated only in the caption should carry a source.
• Version 1 (modified): No cite is needed for a caption's uncontentious, straightforward statement of who or what an image shows, drawn from the images's file description page. Further, to the extent a caption recites something stated and sourced in article text, no cite is needed in the caption; but anything stated only in the caption should carry a source.
• Version 2: A caption's uncontentious, straightforward statement of who or what an image shows does not need to be sourced; further, to the extent a caption recites something stated and sourced in article text, no cite is needed in the caption. But anything stated only in the caption should carry a source. If citations are warranted on the image description page to establish facts about the image, those citations should be repeated in the caption.

Version 2's stuff about citations are warranted on the image description page seems to me a nonstarter, because most images are at Commons, and Commons doesn't use cites. I think V0's provisions should be easy for everyone to get behind. V1 goes further, and expresses what I think is considered good practice, but I'm not sure everyone will be comfortable enshrining it explicitly. It does seem like an issue that ought to be resolved. EEng 22:54, 28 July 2020 (UTC)

I would support documenting this in the MoS, and I agree with EEng#s that version 1 is probably the best. The reference to "uncontentious, straightforward statement" allows for some discretion by the editor, which is important. This would be a guideline, not actually a policy, so when situations arise where it is not appropriate it does not have to be enforced blindly. --Gronk Oz (talk) 02:54, 29 July 2020 (UTC)

Thank you for the feedback. For the sake of clarity and legitimacy, we could also simply expand on the statement from Wikipedia:Citing sources, which is already in policy, with the proposal from Version 1, with a few tweaks:

• Version 3: Image captions should be referenced as appropriate just like any other part of the article (Wikipedia:Citing sources). No cite is needed for a caption's uncontentious, straightforward statement of who or what an image shows, or when the caption simply recites something stated and sourced in article text. But anything which is stated only in the caption should be referenced. A description or attribution of the image resulting from specialized knowledge should also be sourced.

In this proposal, I removed the mention of the content of the file description page from Commons, which is rarely referenced and can be quite unreliable. I added the need to reference the description or attribution when it is the result of scholarly work, as when describing an archaeological artifact or an object from a museum. पाटलिपुत्र Pat (talk) 05:13, 29 July 2020 (UTC)

You're right about Commons descriptions being unsourced, but nonetheless identifications in captions are often drawn from the Commons description, without a citation being given in the caption. For example, the article Richard Nixon carries this image with a caption asserting it's Nixon in high school. But there's no cite in the caption to support that assertion. EEng 05:30, 29 July 2020 (UTC)

For the record, with "citations on commons" I was referring to something like the detailed image description page of this image, with clear source info and a reference that supports what the image depicts. I was specifically asked to add such info to the page of this image as part of a GA review (but not sure if there exists a policy for this). I like the "version 3" proposal by पाटलिपुत्र above but I don't think the reference to the commons page needs to be removed, it can easily be edited and sources added (though I'm not sure making that a requirement would be worthwile). Ichthyovenator (talk) 07:52, 29 July 2020 (UTC)

• I don't especially like these proposals. The most common information to be only in captions and not in text will be basic details on paintings, including portraits in biographies - subject, artist, date and museum/owner. These will normally be taken from the Commons image file, which is is not generally an RS, except in the cases of museum releases with metadata etc. At the same time it is usually correct. I don't think demanding other references for such information is a good idea - it will be generally ignored, & is likely to result in the mass removal of pictures if GA/FAC reviewers insist on it. User:Ichthyovenator, you started out wanting the removal of caption citations, and have moved to a proposal insisting on them! better just to leave imo. Johnbod (talk) 15:22, 29 July 2020 (UTC)

I am not supporting some insistance on caption citations; I believe that long and detailed captions would need to be cited, and that some of the longer captions in the article which spurred this discussion might have been warranted, but I still fundamentally believe that image captions should rarely be long/detailed enough to warrant citations. If captions are short and to the point, as most of them should be, there is no need to police citations in them. Ichthyovenator (talk) 15:33, 29 July 2020 (UTC)

I don't see that any of the proposed wordings actually say that. Johnbod (talk) 15:36, 29 July 2020 (UTC)

Then perhaps that could be clarified as well. Captions being brief and to the point would follow the "criteria" for them being succinct (depending on how one interprets that word) so I don't see how anything conflicts with that. I think "No cite is needed for a caption's uncontentious, straightforward statement of who or what an image shows, or when the caption simply recites something stated and sourced in article text" from पाटलिपुत्र's proposal says what I'm thinking. Ichthyovenator (talk) 15:44, 29 July 2020 (UTC)

Eeng's Nixon example illustrates the point - "uncontentious" is a treacherous word. After a few years on WP, one realizes that virtually nothing is "uncontentious". Portraits from a few centuries ago have all sorts of issues over their subject, authorship and date, which, yes, should be noted in refs. But apart from the subject, if they are illustrating a biography, the date and painter may not really matter all that much, and insisting that is referenced is likely to reduce usage. Johnbod (talk) 16:10, 29 July 2020 (UTC)

Video timestamps

When I see videos in mainspace, they sometimes include a note of how long they are in the caption. I'd like to see what the community thinks of this practice, and how best to format the note if it is included. You may !vote from the list below with a combination of a number and a letter, and feel free to add additional options.

• Option 1: Do not include a length note.
• Option 2: Include a length note whenever desired, on a case-by-case basis.
• Option 3: Generally include a length note in normal circumstances.

• Option A large: Example caption. (4:07)
• Option A small: Example caption. (4:07)
• Option B small: Example caption. (4:07 min)
• Option C small: Example caption. (4:07 min)
• Option D small: Example caption. (4 min 7 s)
• Option E small: Example caption. (4 min 7 s)
• Option F small: Example caption. (4 minutes 7 seconds)
• Option G small: Example caption. (4 min 7 sec)
• Option H small: Example caption. (4 min 7 sec)
• Option I small: Example caption. (4 m 7 s)
• Option J small: Example caption. (4 m 7 s)

I didn't list out all the large options to keep the list short, but if you want to !vote e.g. Option 2C large, feel free to do so. {{u|Sdkb}} ^talk 21:31, 4 October 2020 (UTC)

Examples and syntax

Option A large: (4:07) Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4:07) Option A small: <small>(4:07)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4:07) Option B small: <small>(4:07 min)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4:07 min) Option C small: <small>(4:07 {{abbr|min|minutes}})</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4:07 min) Option D small: <small>(4 min 7 s)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 min 7 s) Option E small: <small>(4 {{abbr|min|minutes}} 7 {{abbr|s|seconds}})</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 min 7 s) Option F small: <small>(4 minutes 7 seconds)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 minutes 7 seconds) Option G small: <small>(4 min 7 sec)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 min 7 sec) Option H small: <small>(4 {{abbr|min|minutes}} 7 {{abbr|sec|seconds}})</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 min 7 sec) Option I small: <small>(4 m 7 s)</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 m 7 s) Option J small: <small>(4 {{abbr|m|minutes}} 7 {{abbr|s|seconds}})</small> Contra dancers at the 2019 Youth Dance Weekend in Weston, Vermont (4 m 7 s)

Survey

• Option 2 Let's keep it flexible and subject to consensus. (Summoned by bot) --I dream of horses (Contribs) Please notify me after replying off my talk page. Thank you. 22:34, 4 October 2020 (UTC)
• Option 3E small If it's all done manually I wouldn't expect usage to be so consistent at first, but with templates consistency will improve and maybe some sort of automatic media metadata functionality miracles and wonders... anyways 3E small seems like the best format to converge on. --▸₷truthious Ⓑandersnatch◂ 23:05, 4 October 2020 (UTC)
• D - Would prefer 1 but I don't think everyone gets #1 so IMHO D is the next best option here. –Davey2010^Talk 20:25, 5 October 2020 (UTC)
• 3E. It is helpful to know how long the video will be before playing it. The other options are ambiguous. Kaihsu (talk) 10:19, 7 October 2020 (UTC)

  @Sdkb, I dream of horses, Struthious Bandersnatch, Davey2010, and Kaihsu: I have added several more options, because some of those initially presented are using inconsistent abbreviation styles ("min" but "s").  — SMcCandlish ☏ ¢ 😼  11:58, 26 October 2020 (UTC)

See https://physics.nist.gov/cuu/Units/outside.html – Kaihsu (talk) 17:55, 26 October 2020 (UTC)

@SMcCandlish: It's good to have the other options but I'm sticking with 3E small because as Kaihsu points out, the disparity is due to 'm' as a unit already signifying "meters". --‿Ꞅtruthious 𝔹andersnatch ͡ |℡| 03:49, 27 October 2020 (UTC)

• 3 and H (though see below about potential use for J). We need to use a consistent abbreviation style, not mix two of them on the same line. An issue is that m and s have other meanings (e.g. m is the standard symbol for metre/meter). If we went with any version that uses {{abbr}}, that linking should only be done on first occurrence (just as, for example, we only use {{circa}} for the first appearance in the page of "c.". That is to say, all the "different" options that differ only by use of {{abbr}} or not are really the same option just used at different places in the page. And a side point is that MOS:ABBR would want these with a period/point: min., sec., m., or s., unless per MOS:NUM we have proof that they are standardized as unit symbols without them. Even in British/Commonwealth English style guides, points are dropped only from abbreviations that are contractions like Dr, not from truncations like Prof., thus: min., sec. regardless of MOS:ENGVAR. I would add that m. is distinct from m, so a version of J with periods would be viable. Given that last point, both H and J could then be permissible, with H being normally preferred, but J usable in infoboxes, tight tables, and other situations of constrained horizontal space. Cf. WP:CREEP and WP:MOSBLOAT: don't rule out options without a good reason. PS: The heading is a bit misleading in that this could also apply to audio.  — SMcCandlish ☏ ¢ 😼  11:58, 26 October 2020 (UTC)

  SMcCandlish, thanks for all those thoughts! Looking at MOS:POINTS, I think you're right that all of these should have periods.

  Regarding replication of {{abbr}}, that would be quite tricky technically to implement. I think we might be saved from that, however, in that MOS:REPEATLINK allows for repetition of links in captions, and applying that same principle would allow for repetition of {{Abbr}}. It makes sense, since people often skim through a page and encounter videos, which is different from someone reading a paragraph where we assume they read the paragraph above.

  And regarding audio, that's another good point. This RfC is already complex enough as it is, so I don't want to change it now, but I assume that whatever the result is will probably be applied to audio files as well unless anyone wants to argue that they should do something different. {{u|Sdkb}} ^talk 19:18, 26 October 2020 (UTC)

  Well, if we do something with a template specific for that, whether to abbr-ize it can simply be a template parameter. I don't think there's any issue with using abbr twice on widely separate media in the same page, but doing over and over and over again in a gallery or something would be really annoying. On points: I notified WT:MOSNUM about this thread, and if it is in fact conventional to use, say, "min" and "sec" (across multiple fields, or as part of some kind of intl. standard), someone there will know about that, I would think. But otherwise the default would be "min." and "sec."  — SMcCandlish ☏ ¢ 😼  19:31, 26 October 2020 (UTC)

• 3 A large. (4:07) is already fairly common and can only become more so as the world becomes more tech-savvy. The uber-popular YouTube uses the m:ss format all over the site, although they have less need for the parens. As for "large", I need a better reason than saving a couple of millimeters of screen real estate to shrink text from the default size. <small>...</small> tags would violate MOS:SMALL (accessibility) if used within already-reduced text. ―Mandruss ☎ 12:43, 26 October 2020 (UTC)

  Mandruss, the main appeal of making it smaller, at least to me, is that it helps indicate that the timestamp is set apart from the rest of the caption, not part of its content. E.g. for the hypothetical caption Senator Wales gave one of the longest filibusters on record (40 min 7 s), it's not very clear that the parenthetical refers to the length of the video, rather than the length of the filibuster, whereas it would be clearer if the timestamp were reduced size.

  If we determine that we can't use size as a differentiator due to accessibility concerns, perhaps we could consider using color or something else. {{u|Sdkb}} ^talk 19:03, 26 October 2020 (UTC)

  Such an ambiguity would be very rare and might warrant an IAR exception subject to local consensus. Not significant enough to justify small for all cases (or a different color for that matter). ―Mandruss ☎ 19:27, 26 October 2020 (UTC)

• Small almost-always bad per WP:ACCESSIBILITY. Will return I suspect... --Izno (talk) 16:31, 26 October 2020 (UTC)
  • "Small" here doesn't mean "smaller than the minimum required by that guideline". That said, I really don't care if it's small. I'd be in favor of 3H even if the small tag were removed from it.  — SMcCandlish ☏ ¢ 😼  19:33, 26 October 2020 (UTC)
    • @SMcCandlish: Aren't captions already smaller than normal text? <small>...</small> makes text 85% of its original size, and so using it on a caption pushes that caption under the 85% guideline of MOS:SMALL. — Goszei (talk) 18:40, 30 October 2020 (UTC)
      • Right you are, that's something I overlooked. That rules out all "small" options within captions. Depending on the percent reduction of captions, which I don't know how to find, something between default size and "small" might work (see Template:Font size templates), but I'm not motivated to pursue that since I favor the default size. ―Mandruss ☎ 19:48, 30 October 2020 (UTC)
        • Ah, I didn't notice the captions were smaller already. So, yeah, we wouldn't want small tags or templates causing <85% size.  — SMcCandlish ☏ ¢ 😼  20:05, 30 October 2020 (UTC)
      • @Sdkb: Please be aware that the "small" options should be off the table, per this. ―Mandruss ☎ 14:34, 31 October 2020 (UTC)
• Option 2, generally favouring the shorter styles. Johnbod (talk) 17:07, 26 October 2020 (UTC)
• I'd actually make a completely different suggestion. Seems to me the purpose of providing the length is to give the reader an idea of the time investment he'd be signing up for if he clicks through, and precision isn't needed for that. So my suggestion is to have the guideline (which guideline?) say,

  If the linked video is more than a two or three minutes long, consider including its duration rounded to minutes: Still from video of the incident (4 min).

EEng 16:58, 31 October 2020 (UTC)

• I'd be more amendable if much of the rest of the world took that approach. See YouTube, Netflix, Amazon Prime Video, Hulu, VLC media player, Windows Media Player, etc. I could go on with further investigation, and I'm not aware of any widespread precedent for your method. (Netflix uses e.g. "2h 41m" on their pages describing DVDs, but we're not talking about DVDs here, and DVDs are on their way out anyway.) I doubt you need education on the user benefits of standardization, or on the corollary user costs of everybody thinking they have an idea that's a little better than everybody else's ideas and to hell with standardization. The m:ss format is well established, and AFAICT closer to a de facto standard than any alternative. ―Mandruss ☎ 19:26, 31 October 2020 (UTC)