The backscatter "signatures" were found to vary systematically with annealing temperature but were relatively independent of the subsequent cooling history, reflecting the dependence of grain size and hardness on heat treatment. Both grain size and hardness changes dramatically over a small range of annealing temperature around 1050 degrees centigrade and this sudden change in physical properties was accurately identified by corresponding pronounced changes in backscatter amplitude. This steel exists almost entirely in a single phase (austenite) irrespective of the method of cooling to room temperature. Thus, it is argued that it is reasonable to expect that grain size variations can be monitored to a fair precision in this steel, by ultrasonic methods.

A theoretical relationship between backscatter power and the shear wave attenuation coefficient due to scattering is presented. In excellent agreement with this theory, the measured backscatter power was found to be directly proportional to third, first and inverse power of the mean grain diameter D for the ranges 10 to 60-75 µm, 60-75 to 150 µm and > 150 µm respectively. Correspondingly the attenuation coefficients obtained from these power law fits are

where is the wavelength of shear waves. Alternatively attenuation coefficients measured directly from the backscatter amplitudes and experimental D values and spanning a range of almost two orders of magnitude had a precision, as good as 10-14% for most samples. Grain sizes spanning a range of about 1.5 orders of magnitude can in principle be measured to a precision, at best 8%, but typically 18%.

In the theoretical appendix, simple derivations (not found in previous literature) of the scattering laws for the three regimes are presented. It is also shown that scattering in the steel due to causes other than grain boundaries, e. g. dislocations, inclusions and grain boundary precipitation, is relatively insignificant at the operating frequency employed.