Although bone is piezoelectric, the signal from bent bone cannot be described by the piezoelectric effect if the samples are assumed to be homogeneous. To account for the symmetry properties and z-dependence of the signal from dry bone in cantilever bending, a theory postulating a polarization proportional to the gradient of the stress was proposed. The present work is concerned with a further examination of the applicability of the stress gradient theory to both dry and wet bone.

Our measurements on dry bone samples cut from a bovine tibia indicate that the signal is not inversely proportional to the thickness squared, contrary to the prediction of the stress gradient theory for a homogeneous sample. Our measurements show that the moduli responsible for the signal in bent bone vary with position, a finding consistent with some earlier reports that the piezoelectric moduli determined in uniform stress measurements vary with position. These results imply that bone cannot be considered homogeneous.

We then show that variations in the piezoelectric moduli can also account for the z-dependence and symmetry properties of the signal observed in bent bone. The magnitude of the variation needed to produce the signal we typically measure is dd_yzz/dy ≈ 3×10⁻¹⁹ c.g.s.e.s.u. Therefore, since bone is not homogeneous, it is not necessary to introduce the phenomenological stress gradient theory to account for the otherwise anomalous electromechanical effect in bent bone, and we conclude that it is the variations in the piezoelectric moduli that produce the electromechanical effect in bent bone. This result unifies the electromechanical phenomena for dry bone under one effect, the piezoelectric effect, and underscores the fact that the inhomogneities in bone can have a large effect on macroscopic measurements.

We present a mathematical model for the piezoelectric behavior of an osteon in which an osteon resembles a radially-poled cylinder. There is some evidence that this model is applicable to an osteon, but further experimental work is needed to test the model.

We have considered the possibility that trapped charge may contribute to the electromechanical effect in bone through the electrostriction effect and conclude that although this theory may be applicable to some polymers, it is not applicable to bone. We also suggest that the electromechanical effect observed in poled polyvinylidene fluoride in cantilever bending may be due to spatial variation in the piezoelectric moduli.

Our measurements on wet bone show that the signal from bent wet bone exhibits the same z-dependence and symmetry properties that are characteristic of dry bone and that the magnitude of the effect in wet bone is of the order we would expect if the signals had the same origin. However, we find that the sign of the signal observed in wet bone Is always positive, even for samples that have a negative signal when dry. Therefore, it is possible that more than one effect contributes to the signal observed in wet bone.