MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

There seems to be a few papers around with Erdős written as Erdös. For example:

MR0987571 (90h:11090) Alladi, K.; Erdös, P.; Vaaler, J. D. Multiplicative functions and small divisors. II. J. Number Theory 31 (1989), no. 2, 183--190. (Reviewer: Friedrich Roesler) 11N37

Would it be incorrect to cite such papers using Erdős instead?

share|cite|improve this question
I like the "erdos" tag. – Dan Petersen May 14 '10 at 7:05
Yes, definitely. In this case the printed name Erdös is an approximation of the correct typography and you should use the correct version Erdős if possible. – Gil Kalai May 14 '10 at 7:33
@Allen: Why? The generalization is completely natural, and answers to it easily specialize to answers to the original question. – Mark Meckes May 14 '10 at 15:28
Am I correct in my perception that J. H. S. and Gil Kalai agree completely, despite their opening sentences? – JBL May 14 '10 at 20:15
Alladi and Vaaler are among the very few people who have Erdős number 1 as well as Erdös number 1. – Gil Kalai May 19 '10 at 8:44
up vote 26 down vote accepted

We cite papers to show our respect to the authors and to help our readers find stuff. For the second purpose, I suspect most people would just type in names without diacritical marks, and most search facilities would find what you're looking for based on the letters alone, so it doesn't really matter. But for the first purpose, I think you should spell the name the way its owner would want it spelled, regardless of what some journal may have done.

share|cite|improve this answer
I disagree. We cite papers to help our readers find things. If for some reason the author's name was comprehensively mangled, we would need to cite the mangled version. So the principle is that we cite the name as printed in the paper. We are free to add explanations in the text. – Chris Godsil May 14 '10 at 8:49
This reminds me of a similar problem I ran into when writing my dissertation, with mathematicians who changed names (after marriage, say). For example, everybody in mathematical relativity knows that Mme. Choquet-Bruhat proved local existence and uniqueness for the Einstein equations. But the original 1952 paper was published under Y. Fourès-Bruhat. Whatever combination I chose to do (exacerbated by the fact I also cited other later papers of hers after her marriage to Gustav Choquet) it felt awkward. – Willie Wong May 14 '10 at 9:33
Chris: But I think Gerry's point is that, in this case, it doesn't matter, so we should stick to the correct form. But I agree that in a case like the one Willie describes, you probably should use the journal version of the name. – Matthew Daws May 14 '10 at 11:56
@Willie I did that. Admittedly, I only have one paper from before my name-change, so the confusion shouldn't be too massive. – Mikael Vejdemo-Johansson May 14 '10 at 14:00
I think they are both good points, and the way we should treat comprehensively mangled information is different from the way we should treat comprehensibly mangled information. – Andrew D. King Oct 10 '10 at 21:11

alt text

A proposed compromise

share|cite|improve this answer
Is there a badge for "first answer to make someone laugh out loud"? – Alon Amit May 14 '10 at 18:45
The definitive answer to "are there good math jokes?" – Peter Luthy May 15 '10 at 8:29
Is this the new smiley for "shouting while winking very hard"? – David Richerby 2 days ago

In response to Gil's compromise (because you can't put images in comments apparently), you can get the desired effect with \textdotacute from the tipa package from CTAN.

Example of \textdotacute


share|cite|improve this answer
Could you also make a hybrid of Paul and Pál? – S. Carnahan May 14 '10 at 18:19
@Scott, good question, but why stop there? Hungarian practice is family name before given name, so you would also need a hybrid between EP and PE. – Gerry Myerson May 14 '10 at 23:16
Perhaps life would be easier if the man had changed his name to an unpronounceable symbol. (Provided the symbol had a TeX command, of course.) – Nate Eldredge May 17 '10 at 13:54

My rule of thumb is to spell names as MR spells them. I just looked up Erdos in MR, and it turn out that there is a Paul Erdös as well as a Paul Erdős (different people). Since you mean the second, you should spell it correctly, or else put (sic) in your bibliography.

share|cite|improve this answer
Your first sentence could be interpreted either way. The listing of the paper (see the original question) copies the spelling from the paper, but if you click on the name, you get MR's canonical version of the name. – Ben Wieland May 14 '10 at 18:29

This is a matter of convention. One guideline is in paragraph 17.20 of the Chicago Manual of Style (15th ed):

Authors' names are normally given as they appear in the title pages of their books. Certain adjustments, however, may be made to assist correct identification (unless they conflict with the style of a particular journal or series. First names may be given in full place of initials. If an author uses his or her given name in one cited book in and initials in another (e.g., "Mary L. Jones" versus "M. L. Jones"), the same form, preferably the fuller one, should be used in all references to that author.

I would err on the side of consistency. Some bibliography styles in LaTeX/bibtex replace subsequent references to the same author with an em dash. If you use many different spellings of the same author's name, this behavior will break.

share|cite|improve this answer

Cite as it is in the journal. Are you absolutely certain that you know how the author wants his/her name spelled? Are you absolutely certain that the person you think is the author is the author, and not somebody else with the a similar name? Accents and spelling can change or be dropped when a person emigrates or to conceal an ethnic origin, and sometimes that is what is preferred by the person.

share|cite|improve this answer
There is no question that P. Erd\H{o}s spelled his name with the letter \H{o} and not with the letter \"{o}. (Reasonable argument might be possible about whether to one should spell his first name P\'{a}l or Paul.) – JBL May 14 '10 at 17:37
(And, I suppose, about whether using the phrase "first name" for his given name is appropriate.) – JBL May 14 '10 at 20:59
The question, as I interpreted it, wasn't just about Erd\H{o}s, but about the principle. I've had well meaning people (although not academics) remove the apostrophe from my name, lower-case-ize the "B", and often the "t" is removed because "O'Bryant" is an existing Irish name. But it's mine, and it's spelled how I spell it, not how somebody else thinks I meant to spell it. – Kevin O'Bryant May 14 '10 at 21:02
I agree very much with your last sentence, and in this case it means that the first sentence of your answer is exactly wrong -- the name as written in the journal is a misspelling of the name of the author in question. – JBL May 14 '10 at 21:25
I suppose to consider the full generalization, suppose the journal contained a flagrant typo, and then folded before it could issue a correction. Then should you be cited as O'Bryatn, despite that being nobody's intention, on grounds that we can't be absolutely certain how you want it spelled, no matter what you say about it? ;-) Citing with the typo and "(sic)" could aid some searches, but has nothing to do with how you want it spelled, just how some numpty type-setter spelled it. – Steve Jessop yesterday

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.