The bold part could really skew your power distribution well outside of any reasonable generalisation unfortunately - if you are only putting vocals through, you probably have less LF content to start with, and if the lead vox does any screamo type stuff you could easily exceed a "best guess" estimate of power delivered to the top cabs.
I hadn't thought of that. It's a good point.
You amp is however only rated about 4dB more powerfully than the top speakers, so assuming it's kept out of clip (you did suggest this was the case) and any screamo type events are of short duration, you'd probably "get away with it", albeit I'd have been experiencing some significant cheek clenching were I in your shoes at the time.
That's about it! I did keep the vocals at the low end of intelligible rather than make them too prominent - more for fear of being shut down by the venue's owner for being too loud than worrying about speaker damage!
Thanks to everyone for your input. I am trying to think a bit more academically than practically here though. I gave the details of last weekends event more as a reason for my thinking than as a case study.
For an average music input or pink noise, I would like to know if there is a way of working out the LF:HF power distribution given the upper and lower limits and the crossover frequency. There must be a way of working it out, it's just that I don't know what it is!
Steve.