integer function correct_digits(rerr) result(d)
      double precision rerr
      d = 0
      do
         if (rerr < 5*10.0d0**(-(d+1))) then
            d = d+1
         else
            return
         end if
      end do
   end function correct_digits