A cylindrical sample, 0.5 m in diameter and length, was obtained by excavation from 3.7--4.2 m depth below ground surface in a surficial deposit of fractured glaciolacustrine clay. The sample was enclosed in a triaxial cell in a cold room to impose conditions close to field temperature and stress. Hydraulic testing using a hydraulic gradient of 2 provided a saturated hydraulic conductivity of 7¿10-10 m/s, which is only slightly larger than the matrix hydraulic conductivity. The cubic law provided a mean hydraulic aperture of 5--6 &mgr;m for the four continuous vertical fractures in the sample. A column of immiscible-phase trichloroethene (TCE) imposed incrementally at the top of the sample provided an entry-pressure-derived aperture equal to 17 &mgr;m for a parallel-plate fracture. TCE dense nonaqueous phase liquid (DNAPL) that flowed through the sample produced dissolved-phase diffusion haloes in the matrix that indicated the preferential fracture pathways. These haloes indicated that DNAPL flowed through only 5--15% of the visible, oxidation-stained fractures; hence natural fractures had variable apertures along their length, and the larger aperture regions provided channels for fluid flow. Using the hydraulic test results and applying the cubic law only to the aperture segments of confirmed DNAPL flow, an equivalent hydraulic aperture of 8--11 &mgr;m was obtained, which is greater than the conventional mean hydraulic aperture and smaller than the local aperture determined from the DNAPL entry pressure. These differences are large in the context of fluid flux, which is proportional to the aperture cubed. ¿ 2000 American Geophysical Union