The loss mechanisms responsible for the observed attenuation in soils are often unclear and controversial. This is particularly the case with the small-strain damping Dmin in air-dry sands. Ultimately, physical explanations must accommodate the observed effects of confinement, strain level, frequency, and load repetition. Three hypotheses are explored herein: measurement bias, thermoelastic relaxation, and adsorbed layers. Micro and macroscale experimentation using photoelasticity, thermal infrared imaging, atomic force microscopy and resonant column testing are complemented with conceptual analyses. Results show that Mindlin-contact friction cannot explain the observed response of the small-strain damping ratio Dmin and thermoelastic loss is suggested. While thermoelastic relaxation is inherently frequency dependent, the superposition of multiple internal scales in soils can justify the observed low dependency on frequency. Moisture condensation leads to adsorbed water layers on grain surfaces, which has a small but observable effect on shear modulus and a significant influence on damping ratio. Participating lossmechanisms at small-strains may involve distortion and motion of adsorbed layers and hydration force hysteresis. Hysteretic capillary breakage at contacting asperities gains relevance when the strain exceeds the elastic threshold strain; this strain coincides with the strain range when frictional losses begin to dominate. Finally, the damping ratio in air-dry sands is very small, and causality-based attenuation–dispersion relations predict modulus dispersion about 1% per log cycle, therefore the medium can be considered non-dispersive for practical purposes.