[R] 'symbols' not plotting correct circle radii

Jim Lemon jim at bitwrit.com.au
Wed Jul 18 14:17:02 CEST 2012

On 07/18/2012 10:11 PM, Jim Lemon wrote:
> On 07/18/2012 09:42 PM, Sarah Goslee wrote:
>> On Wed, Jul 18, 2012 at 6:42 AM, Jim Lemon<jim at bitwrit.com.au> wrote:
>>> On 07/18/2012 08:02 PM, Stuart Leask wrote:
>>>> Hi there.
>>>> I have been plotting some circles using 'symbols', with radii
>>>> representing
>>>> my data, but the radii looked incorrect.
>>>> It seems to happen with a single circle too:
>>>> Symbols ( 0, 0, circles = 40, xlim = c(-40, 40), ylim= c(-40, 40))
>>>> If I put a ruler up to my monitor (technology!) to compare the radius
>>>> with the axes, the circle isn't radius 40; it is closer to 15...
>>>> I suspect there is a simple answer.
>>> Hi Stuart,
>>> The answer does not appear to be simple. While the documentation for
>>> symbols
>>> says that the "inches" argument controls the size of the circles, it
>>> appears
>>> to do so with reference to the user units of the plot, not to that
>>> quaint
>>> unit descending from the twelfth part of someone's foot. So, whatever
>>> numbers you pass as "circles" seem to be scaled to a maximum
>>> dimension of
>>> whatever "inches" is in user units, at least on the x11 graphics device.
>>> From ?symbols:
>> If inches is FALSE, the units are taken to be those of the appropriate
>> axes.
>> Is that not what's wanted here?
>> Sarah
> My interpretation was that Stuart wanted circles with radii that were
> the stated value of the "circles" argument in user units. As the
> "inches" argument is TRUE (1) by default, it only seems to serve as a
> scaling factor for the values of the "circles" argument. So, the
> combination of "circles=40" and "inches=1" yields a circle with a radius
> of 1 user unit. When I reran the example with "inches=0.5", I got
> circles with a maximum radius of 0.5 user units. I haven't tested this
> on any device other than x11.
> Jim
Oops, I just realized that I left out saying that you were correct.

More information about the R-help mailing list