[lxgr]

But it's still an approximation, not an estimate, right? (The fact that you can list the mean radius with sub-decimal precision suggests as much.)

To me, an estimate suggests that there's error bars; an approximation suggests that there's variance that we can quantify (or at least we're very confident about our error bars).

[feoren]

There are error bars on absolutely everything that we ever measure.

[lxgr]

Sorry, missed a "large". It would sound pretty strange to me to claim that the distance between New York and London is "an estimated x kilometers" (with single-kilometer precision), even though there is tectonic movement etc.