I'm assuming there is no variable speed drive involved (vsd supply would of course generate harmonics).

Saturation of the motor iron can result in harmonic excitation currents being drawn from the power system.  This (saturation) may result (or be increased) from any of the following:
- operating the motor at a higher voltage than nameplate (or at unbalanced voltages)
- operating the motor at a lower frequency than nameplate
- bad  motor design which puts the core in saturation under normal operation
- change in motor core characteristics (doesn't seem too likely).
- stator turn to turn short causes increased volts per turn (I'm skeptical of whether a motor can continue to operate after turn to turn short but some people think it can... like CSI who makes flux-sensing devices to detect a trn to turn short).
- applied voltage has a dc component (not very likely)
Anyone else?

It's also worth pointing out that "overvoltage" could be the result of miswiring... in particular hooking up a 480v wye winding as 480v delta.

The harmonics produced from all of the above scenario's (except dc offset) would likely be odd harmonics.  

Out of curiosity... are you investigating power system harmonics that you believe have originated from a motor? Can you provide more details?

electricpete,

Do induction motors sink harmonics?  If so which and why.

I doubt if induction motors sink harmonics.  Motors are primarily inductive and their impedance is going to increase with frequency.  Capacitor sink harmonics because their impedance decreases with frequency and so they present a lower impedance path for harmonics.

Redtrumpet's answer on the sinking of harmonics sounds good to me.  

One other thought on the original post... the measurement system may introduce harmonics into the measurement if a ct or clamp-on probe is saturating.