I'm not sure that I agree with Gengis here. I always thought that an aircraft INU was "schuler tuned", so that the tendancy to drift or oscillate about its position was cancelled out. If this is the case then surely a serviceable INU would not exhibit this trait and schuler effect would not be measureable.
I certainly remember that a crew report of oscillation with a period of about 80-90 minutes was cause for an INU removal due to failure of the internal schuler circuits.