When you 'meg' a motor, you are no longer testing for a perfect open circuit, but instead testing the slight leakage current that is always expected from a real insulation system.
It is critical that you not meg at too high a voltage, because doing so can _damage_ the insulation system. Without a proper specification for the test, giving the allowable voltages and leakages, the test is worthless and potentially damaging to the motor.
It seems totally reasonable to me that you be able to test the motor in the water, but it also seems totally reasonable that the test voltages and allowable leakage currents (insulation resistance) would be different for the submerged and the dry motor.
Tom: do you guys have different test specifications when you test motors in the water?
-Jon