We present 2 days' measurements of the hemispherical-directional reflectance factor (HDRF) of snow made at fine spectral and angular resolution with the Automated Spectro-Goniometer (ASG) for the range of solar zenith angles ($theta$0 = 40¿--50¿) and snow textures (surface grain size = 80--280 ¿m). Measurements of the stratigraphy of snow texture and density accompanied each day's suite of measurements. These measurements represent the most detailed available in terms of angular and spectral resolution. The HDRF for fine grain, faceted snow exhibited a local backscattering peak at the view zenith near the solar zenith angle, whereas those for medium grain, clustered snow did not have a local backscattering peak. The HDRF decreased at all wavelengths for an increase in grain radius from 80 ¿m to 280 ¿m. However, the decrease in HDRF in the visible wavelengths was largest at $theta$r = 80¿ in the forward direction and largest for λ > 1.8 ¿m near $theta$r = 30¿ in the backward direction. As solar zenith angle decreased from 47¿ to 41¿, the HDRF increased near nadir for λ ≤ 1.03 ¿m but decreased with coherent angular structure for λ > 1.03 ¿m. We compared forward radiative transfer modeling results with the HDRF measurements. The forward model used single-scattering parameters for ice spheres with radii that matched the surface-area-to-volume ratio derived from stereological analysis of snow samples and a stratigraphic distribution of optical depths from measured density and modeled extinction efficiency. All HDRF models underestimated reflectance for λ > 1.30 ¿m and had large absolute errors in the perpendicular plane. Mean absolute RMS errors in reflectance for the fine grain, faceted snow case were 0.09 at λ = 1.3 ¿m and 0.14 at λ = 1.85 ¿m. Mean absolute RMS errors for the medium grain, clustered snow were 0.04--0.06 at λ = 1.3 ¿m and 0.04--0.06 at λ = 1.85 ¿m. The models for the more spherical medium grain snow had better overall spectral and angular fits than those for the nonspherical fine grain snow. The spherical radii inferred from the surface-area-to-volume ratio from stereological analysis of snow with nonspherical particles have a greater effective path length than the actual snow particles, resulting in underestimates of hemispherical-directional reflectance.