# ~/cite : Is the Tie Actully Recommended? [duplicate]

When making citations, I've seen both styles: with the tie/non-breaking space (~\cite{refnm}) and without it (\cite{refnm}). There are a few questions & answers on this topic in the more general sense, recommending reasonable things like a tie before a \ref. However, \cite never seems to be mentioned in these lists...

I happen to be mostly concerned about numeric styles like IEEE, which I think is unlikely to produce bad line breaks if a tie is used, but what about other styles, and is it really recommended?

Interestingly, the IEEE Editorial Style Manual only says the following about references, which I think a few other guides have turned into "should be one the same line" (I'm pretty sure they mean as opposed to super/sub-script).

When they are, they appear on the line, in square brackets, inside the punctuation. Grammatically, they may be treated as if they were footnote numbers [...]

I did look at Why should I put a ~ before \ref or \cite?, and I thought that \cite deserves a separate question - I think it's pretty different case from \ref. I guess I'm looking for a more stylistic opinion.

## marked as duplicate by Stefan Pinnow, TeXnician, Tiuri, Troy, marmotJan 5 at 19:12

• The main question is, do you want to allow or forbid "Jones and Smith(linebreak)[1]"? – Teepeemm Jan 5 at 17:38

With a numeric reference style the tie is certainly recommended, because you don't want to have

blah blah blah blah blah blah blah blah as seen in the paper by Uthor
[2] blah blah blah blah blah blah blah blah blah.

Better to abound with ties than to spare them. In cases where you get underfull or overfull lines due to ties, you can exercise your own judgment about where to allow line breaks, perhaps doing some rewrite or carefull removing a tie. Even more so if your document is submitted to somebody else and you have no control on the final line width. In this case the ties can and will be managed by the copy editors and you should not trust the output you get in the preprint version.

If the citations use author-year style (like in \citep or \citet of natbib or the more advanced commands of biblatex, a tie will usually be of a hindrance:

blah blah blah blah blah blah blah blah as proved in
Uthor~(2002) blah blah blah blah blah blah blah blah

blah blah blah blah blah blah blah method is successful
(Uthor,~2002) blah blah blah blah blah blah blah blah

are not a problem (the ties will be supplied by \citet and \citep).

Cross references, that is \ref or \eqref are different beasts and should always have a tie. But cases such as

theorems \ref{thm:main}~and~\ref{thm:other}

are good like this: it's better if the numbers end up in the same line than getting

theorems 11.35
and 12.3

So judgment is always to be exercised.