I believe this was intended to be a followup to PR#8713 but has opened a new report, so I am replying to re-file it in the correct place.
You can read the source code to find out what happens (as anyone else looking into this would need to). I currently have Argument \code{inches} controls the sizes of the symbols. If \code{TRUE} (the default), the the symbols are scaled so that the largest dimension of any symbol is one inch. If a positive number is given the symbols are scaled to make largest symbol this height in inches (so \code{TRUE} and \code{1} are equivalent). If \code{inches} is \code{FALSE}, the units are taken to be those of the appropriate axes. (For circles, squares and stars the units of the x axis are used.) Note that it is not just width and height: the units for the whiskers are also relevant. Indeed, the scaling for boxplots is chosen so that the largest width, height or whisker length is one inch. On Thu, 30 Mar 2006, [EMAIL PROTECTED] wrote: > concerning the symbols manpage: > > > it probably should actually say > > "If 'inches' is 'FALSE', the units for the width and height of the > symbols are taken to be those of the x axis and y axis, respectively..." > at least for 'boxplots' symbols this is empirically the actual > behaviour. if other symbols (e.g. circles) actually _are_ using x axis > units, the difference between 'isotropic' and 'anisotropic' symobls > should be made clear. > > the current text gave me the false impression that I needed to set asp=1 > in order to get correct heights for the boxplots symbols. > > joerg > > ______________________________________________ > R-devel@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel > > -- Brian D. Ripley, [EMAIL PROTECTED] Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595 ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel