The topography of laboratory induced shear fracture surfaces of Wester
ly granite was studied. Three types of fracture surfaces were examined
: (1) a fresh fracture from the shear failure of an intact sample unde
r polyaxial loading (sigma2 = 40 MPa > sigma3 = 15 MPa); (2) a shear f
racture subjected to frictional sliding of 100 mum under polyaxial loa
ding; (3) a shear fracture subjected to frictional sliding of 800 mum
under conventional triaxial loading (sigma1 > sigma2 = sigma3 = 40 MPa
). Both sliding distances are within the range of the grain size of We
sterly granite. The results are represented by a power spectral method
. Similar to the power spectra from natural rock surfaces, the power s
pectra of the induced shear fracture surfaces fall off about 2 orders
of magnitude per decade increase in spatial frequency. No corner frequ
ency exists in the power spectra over a spatial frequency range from t
hat corresponding to the profile length to the Nyquist frequency. A sl
ope break in the power spectrum was identified, however. It separates
a steeper low frequency segment from a less steep high frequency segme
nt. The spatial frequency at the slope break corresponds to a waveleng
th of several hundred microns which is on the scale of the microcracki
ng and contact breaking on the fractures. Upon re-examining power spec
tra of natural fault traces and fault surfaces obtained in previous st
udies, we noted similar slope breaks. We suggest that this slope break
may have significant implications in the scaling problem. Both the in
duced fracture surfaces and natural faults exhibit topographic charact
eristics different from those of sawcut surfaces, which have been wide
ly used in laboratory rock friction experiments. In the present study,
we observed that even a small amount of sliding (less than a grain si
ze) already results in significant mismatches between the paired slidi
ng surfaces in the direction normal to sliding.