no magic.
just seems to work better then smaller coils with HF frequencies and there
usually isn't enough spare cord to make up bigger coils.
I suppose a more "magical" number would be 6 inches, and the correct number
of turns would depend on the frequency that worst affects the component in
which you are trying to reduce the RFI.
Think of a coaxial balun/choke that is recommended on dipoles and J-polls
and so on in place of a proper balun.
The choke will keep RF off the cable and prevent it from acting like an
antenna.  This may not get rid of the RFI altogether but it does seem to
work very well for speaker wire and long runs of cable.
I was getting into my wife's computer on all bands, and especially bad on 20
meters even at 5 watts.  I finally narrowed the problem down to a 20 foot
piece of two conducter XLR cable used for a microphone into the soundcard of
the computer.  It was the only one i had on short notice and so I just
coiled it up into about a 6 inch diameter coil with a few feet on each end
to reach the mic and the PC, and the RFI vertually disappeared.  If i unplug
the cable, the RFI goes away completely.
So, coiling cables can certainly assist with aleviating interpherence
issues.
73
Colin, V A6BKX