We present a procedure for comparing strain events at different sites, and apply it to observations for the Homestead Valley earthquake sequence of 15 March 1979. Coseismic strain steps occurred on two laser strainmeters at Piñon Flat Observatory (PFO) and on a creepmeter spanning the San Andreas fault at Wiebe Ranch (XWR); the creepmeter records were studied by Leary and Malin (1984). The PFO coseismic strains are consistent with a source model derived from near-field geodetic measurements, while the XWR coseismic extension is 1200 times too large and of the wrong sign. The XWR instrument showed an anomalous extension about 20 hours before the earthquake (Leary and Malin, 1984), but no such signals were detected on the PFO strainmeters at this time. To compare these preseismic observations we must make some assumption about the preseismic source, and we choose to assume that any precursory slip occurred on the fault plane that eventually ruptured. Under this assumption, the preseismic XWR and PFO records cannot be reconciled unless the preseismic dislocation is left-lateral or mostly thrust, and the fault zone at XWR magnifies strain by a factor of 2900 or more. This strain magnification implies that the fault-zone shear modulus at XWR is 107 Pa, three orders of magnitude smaller than the shear modulus of typical crustal rocks. However, earth-tide observations at XWR constrain the strain magnification at this site to be less than about 55. Thus the preseismic extension at XWR is probably not a precursor to the Homestead Valley earthquake.