strange to relate, this is actually correct behaviour
(on the part of rotating.sty, anyway).
i've tested it out with my latest release (what's in tex live 2008, iirc) and the results are as you report. however, with what svmono sets up, the results are correct. (the old version you had is one that doesn't pay attention to the class settings for float layouts; i deemed _that_ was a bug, and changed it a year or so back. and announced the change.
i've had lots of problems since, with the positioning stuff, which is why i took the trouble to test your thing out.
anyway, svmono sets things up with floats in float pages at the top of the page, and rotating copies that setup for rotated floats. (the latex "standard", such as it is, has them in the middle of the page, which is why the thing looks "right" for you with article.)
anyway, the documentation tells you that the values of the relevant skips are recorded in \rotFPtop and \rotFPbot; look at those after loading the package and you see
*\showthe\rotFPtop
> 0.0pt.
<*> \showthe\rotFPtop
?
*\showthe\rotFPbot
> 0.0pt plus 1.0fil.
<*> \showthe\rotFPbot
so all you need to do is:
Code: Select all
\setlength\rotFPtop{0pt plus 1fil}
and everything will be as you wanted. (whether the springer editors will like it is another matter...)