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?

Larry,  I think the issue here might be related to wording.  The laser does not actually emit the attosecond pulse.  I believe that vswitchs alluded to this in his response to me.  I believe optics are used to generate the actual attosecond pulse.  The laser emits light at the characteristic wavelength (not a fraction of a wavelength).  Once the laser pulse is shaped it can become something else.  If the pulse starts out pretty wide from the laser then its spectrum is highly monochromatic.  But, after being shaped its spectrum changes.  For purposes of this problem, I think the fact that the pulse originates in a laser which is monochromatic is not highly pertinent.

Given the above, one thing to note is that 10^-18 s is not the only characteristic time.  If we assume the pulses are emitted at approximately 10^-14 s then the "square" wave is really a non-50% duty cycle wave with an "ON" time much much less than its off-time.  The "ON" time is 10^-18s and the Tperiod= 10^-14s.

The area of darkness between the pulses actually contain an infinite sum of fourier components which destructively add up so that the light intensity is zero there.  So, the waves never "disappear" just because the pulse does.  Thus, you never really have fractions of a wave propagating around.