[R] lwd less than 1

[Prof Brian Ripley]

On Thu, 19 May 2005, Jacques VESLOT wrote:

> Thank you for helping,
>
> Actually, lwd=.5 in plot() or par() brings neither error nor warning message, 
> but lwd below 1 seems not to be taken into account, as plot(..., lwd=1) and 
> plot(..., lwd=.5) look quite the same.

On what device: see below?

> Besides, it is mentioned in the lwd section of the par() help page :
> " (...) some devices do not implement line widths less than one".

(The people who write the documentation do tend to know what it says:
you omitted `The interpretation is device-specific' and it seems did not 
look in the device-specific documentation.)

This _is_ documented under both ?postscript and ?pdf as

      Line widths as controlled by 'par(lwd=)' are in multiples of
      1/96inch.  Multiples less than 1 are allowed.  'pch="."' with 'cex
      = 1' corresponds to a square of side 1/72 inch.

You haven't mentioned the device (or version of R) you are using, nor the 
commands used nor how you are viewing the output.

I checked the sources: the quartz() device is restricted to lwd >= 1, so 
this might have resulted from plotting on that and then copying the plot.

> What I am looking for is how to get readable plots (with not too wide lines) 
> after size reduction...

Using postscript() or pdf() directly works for me, and the code is the 
same on all R platforms.

> Paul Murrell a écrit :
>> 
>> Jacques VESLOT wrote:
>> 
>>> Someone, who uses R under Mac, wants to insert a couple of small plots 
>>> (each with several lines) in an article, but he has to reduce plots' size 
>>> significantly. He did it (in pdf or enc. ps) but, unfortunately, 
>>> everything is reduced but lines' width. Besides, 'lwd' argument in par() 
>>> or plot() can't be less than one.
>> 
>> IANAMU (I am not a Mac user) either, but you should definitely be able to 
>> set lwd to less than 1.   What happens when you try par(lwd=.5) or 
>> plot(1:10, type="l", lwd=.5)?

>>> I am not a Mac user and I don't know whether it is a Mac-related problem 
>>> or not.
>>> 
>>> Could somebody please give me a way to solve this problem, either to 
>>> reduce line width under R, or to get reduced lines' width when reducing 
>>> plot size ?

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
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