This is true below 1/0 as well. A change of 3 in AWG is a doubling in cross-sectional area, and a change of 6 is a quadrupling. The ampacity of AWG 2 is much less than 4 times the ampacity of AWG 8, for any insulation temperature rating.
Cheers, Wayne
That's true throughout the sizes. If you plot ampacity as a function of kcmil, you will not get a straight line. You will get the same mathematical trend throughout the sizes, so there isn't a sudden change in the physics at any size in particular.
By contrast, if you plot DC resistance as a function of kcmil, you will get a straight line. That is why it is in your advantage to parallel for local feeders, whereas it isn't necessarily as much of an advantage to parallel for long ones when voltage drop plays a controlling role.
Like I said, it is probably the relative uncertainty of manufactured wire tolerances, that prohibit parallel sets below 1/0. Suppose the manufacturing tolerance on all wires is +/- 1 kcmil. There is a big difference in ampacity between a #8 wire that could be anywhere from 15 kcmil to 17 kcmil. Whereas if a #1/0 could be anywhere from 105 kcmil to 107 kcmil, the relative difference in ampacity is a lot less.