Abstract. The detachment of rock fragments from fractured bedrock on hillslopes creates sediment with an initial size distribution that sets the upper limits
on particle size for all subsequent stages in the evolution of sediment in landscapes. We hypothesize that the initial size distribution should
depend on the size distribution of latent sediment (i.e., fracture-bound blocks in unweathered bedrock) and weathering of blocks both before and
during detachment (e.g., disintegration along crystal grain boundaries). However, the initial size distribution is difficult to measure because the
interface across which sediment is produced is often shielded from view by overlying soil. Here we overcome this limitation by comparing fracture
spacings measured from exposed bedrock on cliff faces with particle size distributions in adjacent talus deposits at 15 talus–cliff pairs spanning a
wide range of climates and lithologies in California. Median fracture spacing and particle size vary by more than 10-fold and correlate strongly
with lithology. Fracture spacing and talus size distributions are also closely correlated in central tendency, spread, and shape, with b-axis
diameters showing the closest correspondence with fracture spacing at most sites. This suggests that weathering has not modified latent sediment
either before or during detachment from the cliff face. In addition, talus at our sites has not undergone much weathering after deposition and is
slightly coarser than the latent sizes because it contains unexploited fractures inherited from bedrock. We introduce a new conceptual framework
for understanding the relative importance of latent size and weathering in setting initial sediment size distributions in mountain landscapes. In
this framework, hillslopes exist on a spectrum defined by the ratio of two characteristic timescales: the residence time in saprolite and weathered
bedrock and the time required to detach a particle of a characteristic size. At one end of the spectrum, where weathering residence times are
negligible, the latent size distribution can be used to predict the initial size distribution. At the other end of the spectrum, where weathering
residence times are long, the latent size distribution can be erased by weathering in the critical zone.