K (r)

Several authors recomend using 0.25 * length of the shortest side of 
the polygon (I mean rectangle). For other geometries I don't know but 
I supose that the algorithm in spatstat selects the apropriate range 
of radii.

The default is 0.25 * (length of shortest side of the bounding box of 
the window).   With a slight tweak.
Type:

     rmax.rule

in the R console, to display the code that does the calculation.

Adrian recommends, in the help for Kest(), that the user *not* set "r" 
(the vector of radii)  but rather
go with the default chosen by the software that he wrote.

And anyone who does not follow Adrian's recommendations in respect of 
point pattern analysis
is, IMHO, a damned fool! :-)

     cheers,

         Rolf Turner