Experimental data on stability of a three-dimensional supersonic boundary layer on a swept wing are presented. The experiments are performed on a swept wing model with a lenticular profile with a 40 • sweep angle of the leading edge at a zero angle of attack. The supersonic boundary layer on the swept wing was laminarized with the use of distributed roughness. A pioneering study of interaction of traveling and stationary disturbances is performed. Some specific features of this interaction are identified. The main reason for turbulence emergence in a supersonic boundary layer on a swept wing is demonstrated to be secondary crossflow instability.Introduction. The problem of turbulence emergence and development of methods of transition control in three-dimensional boundary layers has been studied by many researchers. Creating a small commercial supersonic aircraft of the new generation is supposed to involve new technologies, in particular, passive control of the laminarturbulent transition (flow laminarization) in the boundary layer with the use of microscale roughness distributed over the surface of a swept wing near the leading edge.It was shown [1, 2] that instability of the boundary layer on a swept wing can be controlled with the use of distributed roughness at subsonic flow velocities. The method of passive control has the following features. Microscale roughness elements are applied parallel to the leading edge of the wing, at a distance equal to 1-5% of the wing chord. Based on calculation results for the most unstable stationary mode, the spanwise step between the roughness elements is chosen, which should be approximately (0.50-0.55)λ st (λ st is the wavelength of the most unstable stationary mode in the direction parallel to the leading edge of the wing). The roughness elements used in [1, 2] were cylinders 6 μm high, which were located near the leading edge of the swept wing. A change in the spanwise distance between the roughness elements was found to affect the position of the laminar-turbulent transition region. For instance, the use of distributed roughness with a 12-mm step along the wing span (or with a step multiple to this value) made the transition region approach the leading edge approximately by 35%, while the use of an 8-mm step increased the laminar flow region by 11%.Distributed roughness was first used for passive control of the transition in a supersonic boundary layer on a swept wing in the experiments [3,4] with the use of the method developed for subsonic velocities [1,2]. Saric and Reed [3,4] reported that they used surface microroughness to delay the transition to turbulence in a three-dimensional boundary layer on a wing model with a subsonic leading edge. For a supersonic leading edge, the boundary layer was observed to remain laminar on the entire model. Even the use of roughness elements with a step λ st did not lead to boundary layer tripping. It should be noted that the values of the transition Reynolds numbers given in [3,4] are overrated by an order of magnitude. Zuccher et...