profile. A small (to minimise soil disturbance) earthquake frequency of 0.6 Hz was fired up front in each centrifuge test to compare the structural behaviour observed for the two soil profiles in the absence of external damping. An earthquake of 1.0 Hz frequency was fired at the same stage in each centrifuge test to obtain the secant shear moduli. For the most part the earthquake frequencies, magnitudes and durations were kept the same to aid comparison between different configurations. The soil layout in Fig. 2 is specific to the dense–loose–dense sand case, though the instrumentation layout applies to both soil