Those tolerance fits are per. ANSI B4.1 which I think are based on standard room temps 20C or 68F.
Those tolerances are on a diameter fit so I support for a fit of a feature based on radius one would half the number to achieve the same fit - as which a profile of surface tolerance. Though for a fit to exist there must be an opposite surface of reference which circles you back to using the tolerance variations as a feature of size or diameter.