I have at last sacrificed a pair of ordinary supermarket alkaline batteries. In other words, they are unlikely to have unusual capacity. The continuous recording from a Superlux S502 stereo mic, phantom power on, ran for 4 hours and 6 minutes. It created two files of almost 1 hour 34 mins each, and one file of 59 mins. But the same batteries had also done my previous tests, which perhaps accounted to 20 minutes of intermittent use. So to record a typical live event, that's all you would need, I'd say. Of course with more fancy batteries you'd get longer time.
Niels, can you suggest something I could record, ideally with that Superlux mic, that might indicate whether it would be any good for field recording? I have no immediate prospect of being able to go somewhere rural but suburban background noise after midnight would be possible, which I might get the chance to upload to YouTube in about a week. But as you would not be familiar with the context, the environment, I am not sure it would be meaningful. Also I only have foam covers for the mics which really don't help with actual wind noise. Hmmm...
I do have a plan to record some music from a source connected to a Mackie mixer which has mic level outputs. Feed in some loud music then mute the input channels of the Mackie. What would then be recorded would be the output noise of the mixer plus the input noise of the recorder. Then I'd normalise it, and when played back through headphones such that the music part was as loud as you'd care to listen, would the worst-case-scenario muted-source noise be audible, or rather significantly audible? If it could be heard, the point would be moot as it would not be possible to say whether the recorder or the Mackie was creating the noise. But if it was, in practice, silent, then that would show that self noise would not be a problem with the recorder alone, at least in the context of music recording eg classical music under concert conditions in a church or hall.