various people wrote various things...

I don't think theissue is overheating as there doesn't seem to be any
correlation between speed, time to cutout, time paused (as long as it
is longer than about a second) and time to next cutout following the
pause. After spinning teh front wheel by hand until the light cut out
the resistance to turning *seemed* to reduce suggesting an open
circuit (I say seemed, becuase the difference in resistance is most
easily observed as the wheel stops but this is also about the time
that the light comes back to life. I will investigate the contacts in
the front lamp and try swapping it out (although my other lamp has a
light sensitive switch in the lamp head so that might comlicate
things). MG, do you have experience of teh switch unit being a common
failure point with shimano units? I might also see if I can bypass
the switch altogether

thanks everyone
james