David J Taylor wrote: > > Would it be any better for the routine to lie, and just give to the > nearest MHz, rounded down? That would result in an excessive error for those not using time synchronisation software, and would produce extreme instability if the frequency was very close to the exact MHz.