Actually, turnout numbers don't necessarily have anything to do with radius. The turnout # has to do with the amount of divergence from the through track. A #8 diverges 1 unit for every eight units of through track, a #6 diverges 1 unit for every 6, and a #4 diverges 1 for every 4. As far as Atlas turnouts go (all I know about), the #4 is actually about 4.5.
With these measurements the Atlas #4 diverges 12.5 degrees from the main, a #6 - 9.5 degrees, and a #8 - 7.1 degrees.
Hope that helps.