Analysis of pulse rise time data produced from repetitive piezoelectric and magnetostrictive seismic sources shows that the change in rise time as a function of distance is source dependent. Thus a linear rise time law of the form &tgr;=&tgr;0+CT/Q, with C source independent, cannot be used to assess the attenuation of rock in situ. In fact application of such a law to the data results in a significant overestimate in the value of Q. A numerical example is also presented to show that in terms of filter theory a linear rise time law is ill founded. An alternative, source independent method of assessing the attenuation is employed for both constant-Q and non-constant-Q models, and it is found that either model may be made to give a reasonable fit to the data. However, it is shown that if a non-constant-Q mechanism is operating, then there is probably less than a 5% change in Q over the frequency range 0.80--70.0 kHz.