Category talk:Articles with invalid ISBNs

Template:ISBN-13 has been proposed for deletion

Mediawiki software already recognizes ISBN-13s and links them to Special:Book sources all by itself. This template, {{ISBN-13}}, looks like it must have been created before Mediawiki support was available. It seems possible that widespread use of the template might make it harder for SmackBot to do its checking, and I joined in the call for its deletion. If you have an opinion, you are welcome to add it to the discussion. EdJohnston 03:21, 23 April 2007 (UTC)

10 or 13

Should we use 10 or 13 number combos now? Bearian (talk) 01:51, 15 February 2008 (UTC)

Opinions, of course differ. My take is that we should show only the ISBN-13s for any books published after 1 January, 2007, and preferably show only the ISBN-10 for older books. The reason for the apparent nostalgia is that some catalogs and book-ordering systems will find an entry for an ISBN-10 and not for the ISBN-13 of the same book. At least, this was true when the invalid-ISBN project was last active. There is a simple algorithm to convert one style of ISBN into the other (until the moment that the 13s start to be issued out of the unconvertible range), so two ISBNs are not more informative than one. I believe that Rich F. is somewhat more favorable to the ISBN-13s than me, so you might ask him as well. EdJohnston (talk) 04:36, 15 February 2008 (UTC)

Wow, thanks. Bearian (talk) 16:14, 19 February 2008 (UTC)

Could wikipedia's ISBN page be adapted to automatically convert ISBN-13s to ISBN-10s for sites that only accept those, and to show both forms? --Random832 (contribs) 13:34, 9 May 2008 (UTC)

New lists generated by WikiProject Check Wikipedia

There is now a series of lists with invalid ISBNs at WikiProject Check Wikipedia. -- User:Docu 16:05, 13 April 2009 (UTC)

Placement of this template?

What is the proper placement of this template in a reference? For example:

• <ref>{{cite book |title=My title here |isbn=1234567890 {{Please check ISBN}} |year=2010}}</ref>
• <ref>{{cite book |title=My title here |isbn=1234567890 |year=2010}} {{Please check ISBN}}</ref>

Could someone please add this to the documentation? Thanks! GoingBatty (talk) 22:20, 3 June 2012 (UTC)

It's not intended for manual placement, but is (or was) added by a bot. The bot always placed the {{Please check ISBN}} immediately after the ISBN concerned, and was designed not to re-add a {{Please check ISBN}} if one was already present in that position. So, the first form is correct, because the second could result in a double-tagging. --Redrose64 (talk) 15:50, 4 June 2012 (UTC)

Thanks for the information, Redrose64! So if {{Please check ISBN}} is not intended for manual placement, what is the appropriate way for a human to mark that an ISBN needs to be fixed? Thanks! GoingBatty (talk) 17:15, 4 June 2012 (UTC)

Personally, I've always made an effort to determine the correct number. If I possess the same edition of the book it's easy, but otherwise I try to work it out, and fix the article. --Redrose64 (talk) 10:17, 5 June 2012 (UTC)

Sure, figuring it out if the best option. However, if a user doesn't have the book or the ability/desire/time to work it out, what is the appropriate way for a human to mark it so someone more knowledgable can fix it? GoingBatty (talk) 02:20, 6 June 2012 (UTC)

There are two fixes, one is to simply change the text to "id=ISBN 1234567890 {{Please check ISBN}}" which renders "ISBN 1234567890". This is what I should probably have done, but I was under a little time pressure. The better fix is to change the way the template renders, so that it is consistent with the rest of Wikipedia, which fixes this as a side effect, and is the reason there was no issue last time I did this exercise. I proposed this at the appropriate talk page, however someone thinks that folk need a link to ISBN (on every number no less) (if it were so they should change the default behaviour, of course, not perpetuate a fudge) and I was not minded to fight overmuch, even though it seemed to me that there was consensus for the change. Rich Farmbrough, 15:49, 6 June 2012 (UTC).

The template should have been placed outside the citation template, or there should have been a |check= parameter added to the templates, which could be used to say "check the doi", "check the ISBN" etc... However, some people were too hasty and we have the current mess. Headbomb {talk / contribs / physics / books} 23:59, 15 August 2012 (UTC)

{{Please check ISBN}} should not be placed in the |isbn= parameter of the Citation Style 1 templates. Doing so will corrupt the citation COinS metadata. With the Lua upgrade to the CS1 templates, malformed ISBNs are caught, categorized and reported.

—Trappist the monk (talk) 20:55, 13 April 2013 (UTC)

Yeah, well AFAIK only one user was doing that, and that user is currently blocked until 23:06, 25 March 2014. Even if unblocked they will still be prevented (by ArbCom enforcement) from making those kind of automated edits ever again. --Redrose64 (talk) 22:22, 13 April 2013 (UTC)

Thanks, I didn't know that. Still the template exists and people will use it. Just trying to make sure they don't use it incorrectly.

—Trappist the monk (talk) 23:32, 13 April 2013 (UTC)

Working on these

I'm working on these atm, trying to thin it down... I'm using the tool here, to get hyphenations. The Library of Congress has a tool here that will give the 'valid' check digit....lots of times that seems to be the problem, and with the right check digit it pops right up on WorldCat. Otherwise, I'm actually searching for the book itself in WorldCat and Google books, and if I still can't find it, it's usually a non-English book. If I can, I'll check the publisher's website, but by then I'm usually finding dead websites. At this point I'm commenting them out. Some of these errors are ISSNs or WorldCat #'s just marked wrong. Revent (talk) 04:13, 13 May 2013 (UTC)

Appearance in reflist

Terry Pratchett (previous version, ref#3) -- is that the intended display of "Category:Articles with invalid ISBNs" while this category's template is in place (since deleted)? The first two articles now in the category display no such thing (Brian King #3) and a more helpful glaring red note (Early flying machines #31).

I write here because there is no Talk at Special pages:BookSources and I don't know about Special pages space (documentation, help, talk). --P64 (talk) 00:37, 12 October 2013 (UTC)

The template {{please check ISBN}} should not be included in a CS1 citation. When there are errors in the ISBN (length, improper check digit), CS1 detects, flags, and categorizes them automatically. I've fixed Early flying machines.

—Trappist the monk (talk) 01:22, 12 October 2013 (UTC)

P64, can you be more specific? I went ahead and fixed the two bad ISBNs in Brian King though I guess that was not the point of your question. I don't know which template you're referring to when you say that a template was deleted. It's disappointing to see that Category:Articles with invalid ISBNs is up to 3000 entries. I used to work on emptying that category. EdJohnston (talk) 01:42, 12 October 2013 (UTC)

Trappist has pointed to crucial incompatibility of the citation and maintenance templates that previous editors had used.

FYI those still reading, The appearance of ref #3 in our Terry Pratchett biography surprised me. See the previous version, linked above, because there is no ISBN problem or template {{Please check ISBN}} currently.

Regarding the backlog, I infer that A–E volume is all the current workers are able to handle. Visibility at the top of the list works in local elections, too, I understand. --P64 (talk) 02:00, 12 October 2013 (UTC)

P64, I see a variety of problems in the reference list of the Pratchett article. I'm not sure which of those you are directing to our attention. The 'bad ISBN' message for ref #3 is certainly ugly, and it seems to have been introduced by this edit by Yobot in mid-2011. Can you say what you mean by "this category's template is in place (since deleted)"? Thanks, EdJohnston (talk) 03:03, 12 October 2013 (UTC)

Yobot should not be blamed. The Yobot edit made no alteration to ref in question (which at the time was , not ). Yobot's edit altered |birthplace= to |birth_place= but at the time, both were valid - {{Infobox writer}} was being transitioned in connection with a possible merge into {{infobox person}}, and |birthplace= ceased to be recognised the following day.

The error message did not exist at the time, since it was only generated after {{cite book}} was converted to Lua in early 2013; but had it been in existence then, the error message would already have shown on the page, because the ref itself was not altered by Yobot's edit.

The problem referred to by P64 refers to the visible wiki markup [[Special:BookSources/05525131075|05525131075 [[Category:Articles with invalid ISBNs]]]] which is what happens when {{Please check ISBN}} is placed inside the |isbn= parameter of a {{cite book}} etc. AFAIK it was only Rich Farmbrough and his bots that were doing that; and indeed, the {{Please check ISBN}} was added on 06:42, 13 May 2012. It shows as raw wiki markup because the |isbn= parameter expects a valid ISBN with which to create a link to [[Special:BookSources/isbn]], and links cannot be nested. The red error message - had it been visible at the time - would have been cleared by the edit of 15:38, 3 December 2012 which removed the superfluous digit; but that edit did not remove the {{Please check ISBN}} which is why P64 was seeing it. --Redrose64 (talk) 10:16, 12 October 2013 (UTC)

Now that we have the Lua version of {{cite book}}, it seems unnecessary for bots such as Rich F's to do any more runs to find invalid ISBNs. But I wonder if the error message continues to place the article in the invalid ISBN category. EdJohnston (talk) 14:39, 12 October 2013 (UTC)

It does.

—Trappist the monk (talk) 14:42, 12 October 2013 (UTC)

It's not necessary inside cite templates, but many ISBNs are not in cite templates. there is a Checkwiki report though. All the best, Rich Farmbrough, 16:15, 12 April 2014 (UTC).

Backlog

Thanks for the preceding responses.

Ed Johnston noted the count of 3000+ articles sadly. Category intersection may help set priorities, at least personal priorities for some editors. For example, CategoryIntersect.php now shows [1] that there are 94 of these Articles with invalid ISBNs among the 18000+ Good articles. --P64 (talk) 20:07, 12 October 2013 (UTC)

I started on these and the even more compelling FAs, and made good progress until I had to spend today dealing with Arbcom. All the best, Rich Farmbrough, 04:33, 12 April 2014 (UTC).
04:33, 12 April 2014 (UTC)

FA's done, GA's going. All the best, Rich Farmbrough, 06:12, 12 April 2014 (UTC).

Then, of course, it's on to Category:Pages with ISBN errors, where there are currently 35 FAs and 173 GAs. I've been working on that category; it's down from 8,000 to 5,270, and I have a simple and quick AutoEd script that fixes ISBNs in about half of the citations. Many of the errors are two ISBNs in a single citation parameter, which is easy to fix. – Jonesey95 (talk) 06:20, 12 April 2014 (UTC)

There are 40 FAs now... All the best, Rich Farmbrough, 16:20, 12 April 2014 (UTC).

Correcting invalid ISBNs when they are as-printed and they resolve in some databases

I came across a case in article Willy Ronis that I haven't encountered (or at least haven't noticed) before. The ISBN given for "Willy Ronis: Photographs, 1926-1995" is ISBN 0-905836-89-X, which is invalid. It does not resolve on google books or amazon but it does resolve to the correct work via worldcat. I'm guessing that this is correctly reproducing an error in the original work (incorrect check digit). The ISBN 0-905836-89-8 and the 13 digit equivalent ISBN 978-0-905836-89-8 are both valid and resolve to the correct work via google books and amazon but not via worldcat. It seems plausible to me that where a work has an invalid ISBN printed in it, libraries are more likely to list the work under the as-printed ISBN, whereas booksellers are more likely to list the work under the publisher's intended (valid) ISBN. Normally I've been assuming that invalid ISBNs do not resolve and replacing them with the valid equivalent, but now I've noticed that this one does resolve, should it be changed or not?TuxLibNit (talk) 22:03, 14 July 2014 (UTC)

Hi. Thanks for noticing. I have this book, and I added it to the Ronis article, and can confirm that 0-905836-89-X appears both in the colophon and the barcode. -Lopifalko (talk)

I noticed this strangeness when I was fixing invalid ISSNs. Sometimes an invalid ISSN would link to the referenced magazine in worldcat, but not anywhere else on the web (except copies of Wikipedia and maybe the magazine's web site). When I got to worldcat, the invalid ISSN was usually not listed in the catalog information for that journal. I think worldcat's maintainers somehow scoop up ISSNs that are misprinted in print sources and index them in order to ensure that people can find the source. That doesn't make the ISSN valid, it just means that people make mistakes and worldcat does a nice job of allowing for human error. – Jonesey95 (talk) 01:34, 15 July 2014 (UTC)

I've added this to the Ronis book listing: "ISBN 978-0-905836-89-8 (the ISBN is misprinted in the book as ISBN 0-905836-89-X)". If you think it could be written better than this then please say. -Lopifalko (talk)

I didn't respond earlier because I was happy with this resolution but having recently learned more about worldcat I thought it might be relevant to explain how I now think worldcat gets these invalid numbers. I think that worldcat works by electronically aggregating the records of many libraries so that the invalid ISBNs are found because they've been entered by a librarian in at least one library database somewhere. For example, starting from here if you are brave enough to ignore the scary red warning and search in worldcat anyway you'll find several results. Selecting the first and entering "United Kingdom" as your location under "Find a copy in the library" gives a list of libraries. Following the "Book" link for eg "National Library of Scotland" eventually finds a record in that library and selecting "Staff/MARC21 view" takes you here, where MARC-21 field 020 shows the isbn, tagged with a "z" to indicate it is invalid. I think that normal library practice (policy?) is that one should simply use the value(s) as printed on the work rather than use other means to try to find a valid number instead. IMO this is a much more straightforward rule than wikipedia's "prefer a valid value if available".TuxLibNit (talk) 18:04, 6 August 2014 (UTC)

Nine Princes in Amber and some questions

I like this; I like trying to correct the ISBNs and want to work on the backlog. I have tried to do some work on a few pages and wish to ask on some things I have noted.

There are some, more or less, obvious cases, like was the case with Hat matrix, where ISBN had last digit 6, while 8 produces a correct one. Or with Ōnin War, where two digits were reversed. With these, I found the book on some databases and saw the correct ISBN and made deduction based on "typo-hypothesis".

Then there was the case with A Big She-Bear. Published in 2013, in Georgian, it clearly should have an ISBN. It was marked as having been hyphenated wrong. I did some research and it seems the only extensive record with that ISBN connected with that book was at Goodreads, if I remember correctly. Also, I somehow convinced myself that Google found references to Georgian libraries (no general library sources seem find it, tho, the book or the number). So I just removed the hyphenation. But it does nag at the back of my mind if this was correct...

And then there is the Nine Princes in Amber. A well known, popular book and I know it exists as I have read it. But it was published in 1970. I am not entirely sure they had ISBN back then. The Library of Congress has this book, but mentions no ISBN, nor do any other libraries, as far as I gather. But there is an ISBN listen in the article. The only places where this particular number produces results are user edited sites (Goodreads) or places like Amazon, where people sell the first edition. Where on earth did that number come from, then? It is not the ISBN for any later edition either, it seems. My question is, what I should think about this kind of stuff? Am I making some kind of mistake? Am I the crazy one?

Also, some kind of comments on the cases I have mentioned above would be much appreciated :)

Sincerely, your faithful servant, Voltteri (talk) 05:03, 2 February 2015 (UTC)

@Voltteri: This page has only 20 watchers, so you might not get all of the answers here. Anyway:

• if only the last digit of an ISBN is wrong, this sometimes happens when somebody has taken the ISBN-10 and prefixed it with "978-" without realising that in ten cases out of eleven, the last digit needs to be recalculated - or that somebody has taken the ISBN-13 and removed the "978-" without realising that in nine cases out of ten, the last digit needs to be recalculated. In the case of Hat matrix. that explanation doesn't fit, because ISBN 0-471-17082-8 may be written ISBN 978-0-471-17082-2 and neither ends with a 6.
• why A Big She-Bear should be marked with {{Please check ISBN|reason=does not match hyphenation rules and is probably invalid}} - that was in the article when it was created, so it's possible that Magularia (talk · contribs), who created the page, simply copied it from another article without appreciating its purpose. I often see new articles with {{pp-move}} at the top, which cannot be correct because the page has never been protected - I suspect that many articles are created by copypaste from something related. There is nothing intrinsically "wrong" with the hyphenation of ISBN 978-9941-451-14-0 and it is a valid ISBN as it stands. Hyphenation of ISBNs is often misunderstood - the hyphens or spaces are not necessary, and are included merely to break up the long sequence of digits into smaller groups, that are more easily read. I could put ISBN 978-994-145-114-0 and it would be equally "valid" as far as ISBN databases are concerned.
• The date on which ISBNs began isn't a one-size-fits-all rule. They began during the 1960s in the UK, with the SBN system, which had nine digits. Not all publishers began at the same time: Ian Allan issued their first, SBN 7110-0000-X, in February 1968. When it went international, and became ISBN, books that already had a SBN kept that number but prefixed with a zero, i.e. ISBN 0-7110-0000-X. Books that had already been published before the introduction of the system were often issued ISBNs retrospectively, normally when reprinted, but not always. Puffin Books for example, already used their own coding system which began with PS1 - they simply replaced the "PS" with "SBN 14-03" plus sufficient zeros to make eight digits, to which the appropriate check digit was appended; so PS1 (published 1941) became SBN 14-030001-5 and then ISBN 0-14-030001-5; similarly, PS132 (pub. 1959) became SBN 14-030132-1 and then ISBN 0-14-030132-1.

If a book has an ISBN, use it - but if it doesn't, don't use one from Amazon etc. unless you can be sure that it's the same edition, or at least that the content and pagination are the same. ISBN 1-85227-582-0 and ISBN 0-86369-921-9 are purportedly the same book, hardback and paperback respectively (with differing covers), but page 61 (out of 124) is noticeably different. --Redrose64 (talk) 10:49, 2 February 2015 (UTC)

@Redrose64: Thank you very much for your reply. I had realized this page has not been very active, but at least you saw my questions, so all's well that ends well :) If I may, I would like to ask some clarifications and add some comments:

• It makes sense that recalculations are sometimes omitted in such a process. I have not seen such cases yet, but I have only done a couple, I'll keep an eye out for those.
• I hadn't thought that adding the check request could have been done as a mistake by the editor, I thought it only happens as a bot edit of some type. Very revealing thought, thank you for bringing it up. So, are there cases where the hyphenation matters? Can merely the hyphenation itself be a problem or does it always come with digit errors too? Are "wrongly" added hyphenations something the Special:BookSources picks up?
• I understand now that the date is not clear cut. However, how can one be certain a book does not have an ISBN, as I think is the case with Nine Princes in Amber? If no library or similar "good" source lists one, can the absense of evidence be seen as evidence of absense? Do you have any theories where this particular number in that article came from?
• And finally, if the, say, first edition has no ISBN, but a later edition has, can it be substituted or not? If we are talking about an article about a book and the ISBN is listen in the info box, do we interpret that the article only deals with the particular edition of the book or just any edition or some combination? Which policy applies here?

Yours, Voltteri (talk) 11:25, 2 February 2015 (UTC)

Hyphenation shouldn't matter, since Special:BookSources strips out all hyphens and spaces anyway. But what does matter is the use of the wrong character for the hyphen: it only works properly if you use the one that has its own button on the keyboard, sometimes called hyphen-minus. If you use a true hyphen (‐) U+2010, true minus sign (−) U+2212, en-dash (–) U+2013 or any other type of short horizontal line, it probably won't be recognised.

Different editions of a book always get different ISBNs; but different printings use the same ISBN as the first printing of that edition. It is important to use the correct one, since the content and pagination may differ. For example, there is a book called "The Story of London's Underground" by John R. Day that first appeared over 50 years ago and is now in its 11th edition. The first edition of 1963 (which had no ISBN) measured 6.75" by 4", had 15 chapters and 154 pages; the fourth edition (1971, ISBN 0-85329-016-4) was also 6.75" by 4", but now has 16 chapters and 190 pages; the sixth (1979, ISBN 0-85329-094-6) measures 8.5" by 6", also has 16 chapters but 168 pages; the tenth edition (2008, ISBN 978-1-85414-316-7) measures 11" by 8.5", 23 chapters, 224 pages. The extra 36 pages of the fourth edition are not all in the additional chapter - earlier parts of the text were expanded too, so what had been on page 152 of the first edition was on page 158 of the fourth. --Redrose64 (talk) 12:03, 2 February 2015 (UTC)

Hyphenation rules and She-Bears

Hyphens are used to separate different parts of the ISBN. The first part but only in 13 digit ISBNs) is 978 or 979 which effectively says "this is an ISBN" to distinguish from other EAN codes. The second part is a language-area code, 0 and 1 for English, 2 for French, 3 for German and so forth - 9941 is the code for Georgia. 451 is the publisher code and is valid under the hyphenation rules for Georgia:

0-0
10-39
400-899
9000-9999

These rules allow one "big" publisher "0" with 10,000 ISBNs, 30 with 1,000 ISBNs, 500 with 100 ISBNs and 1000 with 10 ISBNs.

Our publisher is 451 "Saunje" and 14, the serial number, probably the 15th book published by them with an ISBN (they may have already used up one or more previous ranges).

So given that the ISBN is valid, why is it marked as "probably wrong"? There is clue on the Category page:

Note that the ISBN hyphenation rule-base is a living document. While bots try to keep up, as of 11 May 2012 there is a known issue with certain ranges which have recently been assigned to countries, notably numbers beginning (798) 610 and (798) 611.

In 2011 the bot that tagged the article this was copied from had no knowledge of 9941 (and a number of other 99... ranges). (The Library of Congress converter still doesn't know about this range.) Subsequently it learned about these and removed any tags that were imposed as a result. The text was not pasted back into WP until after that bot had stopped running.

All the best: Rich Farmbrough, 16:02, 17 June 2015 (UTC).