> I can only track down this to the step where the `refer' ms-macros > are inserted (i.e. the `refer' output for my original document). > the relevant modified excerpt from the latter is: [snip]
I think this has to do with the way ref*add-xxx and ref*field construct the reference entry as a series of "words" separated by spaces. I guess there is no provision for having two fields "stick together" without a space in between, so this has been *faked* (gasp!) with the following construct: .de ref*add-N .ref*field N \z( "" ")" .. Now, "\z( xyz" might appear to look like "(xyz" under some circumstances, but it's a poor substitute and certainly won't work at the end of a line. My suggestion: the above sucks badly anyway (italic opening parenthesis and roman closing parenthesis (!?)), so I would simply do away with all that fancy stuff and substitute .de ref*add-N .ref*field N .. which also corresponds more to the style used by journals such as Nature and Physical Review. If you absolutely want to prevent a linebreak between two fields I guess you need to redesign the ref*add and ref*field macros.
