Recently physicists have managed to build "attosecond lasers", lasers which emit pulses 10^-18 seconds long, interrupted by much longer periods of darkness (at least 10^-14 seconds). Before them, lasers emitting femtosecond (10^-15 seconds) pulses have been around. Assuming they produce visible light, what colour is it?

Thanks for the hint.  However, I haven't been frustrated.  I just hadn't spent the time to Fourier transform the non-50% duty cycle pulse.  It should contain integer multiples of the fundamental frequency f=10^14 cycles/second.  This is near the visible.  The intensity goes down for each higher level integer multiple.  For example, there may be components like 1,3,5,7 X the fundamental frequency.  The peak is at 1X the frequency.  But, the higher order harmonics have a smaller intensity.  So, something like 5X would be visible.

The actual peak intensity is not in the visible since 10^14 cycles/second is infrared.  But, since we don't see infrared, what we see of the pulse is only a fraction of its total intensity.

As I mentioned in my previous email, there should be a series of spectral components which constructively add up at the narrow peak of 10^-18s width and destructively add up everywhere else (i.e. the dark areas). 

I'll post something more thorough later.