This paper is very interesting;
http://xxx.lanl.gov/PS_cache/arxiv/pdf/ ... 3454v1.pdf
it is written by two of the authors who supplied the rule of thumb which supplied the popularly-quoted radius, Valencia and Sasselov.
But this one is much more comprehensive, it seems to me; it gives a wide range of densities and radiuses, all dependent on a tripartite division of components between metallic core, rocky mantle, and watery hydrosphere.
A feasible waterworld could be as much as 80% water according to this paper.
There seem to be two different models for the formation of worlds like Gl 581 c; if it formed in the dry inner part of the solar nebula, it would be mostly metallic core and rocky mantle; the maximum feasible density (according to the paper) gives a planet with 65% metallic core, 35% mantle, minimal water, a diameter of 1.5 Earth's and a gravity of 2.2 gees.
The other model concerns a Neptune-like ice core world which formed beyond the snow line, and migrated inwards; the minimum density (80% water) gives a radius twice Earth's and a gravity of 1.2 gees.
Two very different worlds; I don't know if there is any way to distinguish them at this stage.
