Afterthought: certainly don't turn the wick right up in the hope of 'Class A' operation. No, really, don't do it.
Well, since you asked...in the interest of research... and with a little adjustment (ahem) my Nait was temporarily fudged to just squeeze the necessary 90mV of bias to run the test entirely in Class A. Result: 0.0135% THD (whereas just 7.0mV of bias resulted on 0.0165% THD) And at this level the bias was galloping away while I watched. The bias setting represents nearly 400mA of standing current, and the meter drifted well past the 500mA mark in under 15secs of the very abbreviated test run. Left unchecked the amp would have blown in short order. Now look at the 'gain': a difference less than 0.003% (< -90dB difference) but with the certainty of killing the amp. So don't believe anyone that says more bias is better.... at least without checking all the possible consequences in detail.