guidelines for conductor radius to avoid corona ?

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Mike Harrison by way of Terry Fritz <twftesla-at-qwest-dot-net>" <mike-at-whitewing.co.uk>

Anyone have any useful guidelines for the recommended radius of
curvature of (mostly cylindrical) conductors required to avoid corona
loss at various DC voltages....? 
I assume the relationship between radius and loss is somewhat
nonlinear - is there some kind of formula or rule-of-thumb to give
something like a "corona current per unit length" loss figure for a
given voltage and radius?