par(mkh) {was [R] Plot Character Sizes in rw-0.90.0}

Peter Dalgaard BSA p.dalgaard at
Thu Dec 2 17:08:06 CET 1999

Martin Maechler <maechler at> writes:

> 1) Implementation unchanged + using  mkh gives a warning saying 
>    that mkh is practically ignored.
> 2) setting par(mkh) or  <highlevelplot>(* , mkh = ..)
>     gives a warning (``mkh : setting cex'')
>     and then behaves as if "cex" was used.
> 3a) Even more S compatibility:
>    Setting par(mkh) or  <highlevelplot>(* , mkh = ..)
>    changes "cex" accordingly *IF* pch=<numeric> .
> 3b) Similar but more logical; need some "notation" : 
>     The symbol size used in the core function plot.xy()  is
>            cexbase * Pcex   (--> plot.c, l.1089)
>     where "cexbase" is par("cex") and  Pcex is the "cex" argument of 
>     plot(),points(),...
>     Now we could change this to
>            cexbase * mkh * Pcex
>     for the case of numeric pch (internally pch <= 31 or something) where
>     mkh is *either* par("mkh") or the mkh arg. of plot.xy().
> 4) complete S compatibility (cex *not* working for pch=<numeric>)
>    This is completely out of question for me!

Um, as I read my S docs, cex works for numeric pch, but only if mkh=0
which is the default, and mkh works for pch=1 only. That mkh doesn't
work for non-numeric pch is a bit odd, and we could fix that if we
could decide on a reasonable definition of the height of such symbols
(probably, it should depend on the font, not the character). 

But is the real issue not that one will sometimes want to set the size
in "screen-inches", rather than by scaling the defaults? Looks wrong
to me to have it scaling with anything except the pixel resolution.

   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at             FAX: (+45) 35327907
r-help mailing list -- Read
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at

More information about the R-help mailing list