We infer the fine structure of a sub-Moho boundary layer (SMBL) at the top of the lithospheric mantle from high-resolution seismic observations of Peaceful Nuclear Explosions (PNE) on superlong-range profiles in Russia. Densely recorded seismograms permit recognition of previously unknown features of teleseismic propagation of the well known Pn and Sn phases, such as a band of incoherent, scattered, high-frequency seismic energy, developing consistently from station to station, apparent velocities of sub-Moho material, and high-frequency energy to distances of more than 3000 km with a coda band, incoherent at 10 km spacing and yet consistently observed to the end of the profiles. Estimates of the other key elements of the SMBL were obtained by finite difference calculations of wave propagation in elastic 2D models from a systematic grid search through parameter space. The SMBL consists of randomly distributed, mild velocity fluctuations of 2% or schlieren of high aspect ratios (???40) with long horizontal extent (???20 km) and therefore as thin as 0.5 km only; SMBL thickness is 60-100 km. It is suggested that the SMBL is of global significance as the physical base of the platewide observed high-frequency phases Pn and Sn. It is shown that wave propagation in the SMBL waveguide is insensitive to the background velocity distribution on which its schlieren are superimposed. This explains why the Pn and Sn phases traverse geological provinces of various age, heat flow, crustal thickness, and tectonic regimes. Their propagation appears to be independent of age. temperature, pressure, and stress. Dynamic stretching of mantle material during subduction or flow, possibly combined with chemical differentiation have to be considered as scale-forming processes in the upper mantle. However, it is difficult to distinguish with the present sets of Pn/Sn array data whether (and also where) the boundary layer is a frozen-in feature of paleo-processes or whether it is a response to an on-going processes; nevertheless, the derived quantitative estimates of the SMBL properties provide important constraints for any hypothesis on scale-forming processes. Models to be tested by future numerical and field experiments are, for example, repeated subduction-convection stretching of oceanic lithosphere (marble-cake model) and schlieren formation at mid-ocean ridges. It is also proposed that the modeling of the observed blocking of Sn and Pn propagation at active plate margins offers a new tool to study the depth range of tectonics below the crust-mantle boundary. Finally, the deduced schlieren structure of the SMBL closes an important scale gap of three to four orders of magnitude between structural dimensions studied in petrological analysis of mantle samples (xenoliths or outcrop of oceanic lithosphere) and those imaged in classical seismological studies of the lithosphere.