I have found that the 6L7G tube works as well as or better than the 6A8G in this circuit, as it is driven by an external oscillator (6K6 wired as a triode, which could probably just as well be a 6K5G or 6J5G). With the 6L7G, the 2nd control grid is injected with the oscillator signal and the 1st control grid receives the RF. To me it seems that this configuration leads to better reception (I haven't checked out gain differences yet). The 6A8G uses the 2nd control grid for RF and the 1st grid for oscillation. As the 6L7G isn't used as often, its prices don't seem to be as high, so this would be a nice inexpensive alternative to the 6A8G in this circuit configuration.
T.