This system contains 1728 water molecules
(This is SPC water I think.)

Then I ran a short simulation for 170000 timesteps at 300Kelvin and 1 atm.
(that's when it crashed.  I'll worry about why later...)

Anyway, the average volume was 52149.8 (in Angstroms^3)
(for the last 80000 timesteps, after it had equilibrated)

Given that the mass of water is 18.0154 grams per mole, I'm getting
this value for the density:

density = (1728*18.0154/6.02214129e23) / (52149.8*1e-30*1e6)
        = 0.991 (in grams per mL)

I'm only looking for gross errors in the OPLSAA force-field.
So I'm satisfied with a 1% error.
But I realize this is not a particularly rigorous test.

Andrew 2014-5-21
