For many years there has been debate regarding why shock wave reflection off a solid surface has allowed regular reflection to persist beyond the incidence angles where it becomes theoretically impossible. Theory predicts that at some limiting angle the reflection point will move away from the wall and Mach reflection will occur. Previous studies have suggested that the paradox could be due to the presence of the growing viscous boundary layer immediately behind the point of reflection, and some numerical studies support this view. This paper takes the approach of establishing an experimental facility in which the theoretical assumptions regarding the surface of reflection are met, i.e. that the reflecting surface is perfectly smooth, perfectly rigid, and adiabatic. This is done by constructing a bifurcated shock tube facility in which a shock wave is split into two plane waves that are then allowed to reflect off each other at the trailing edge of wedge. The plane of symmetry between the waves then acts as the perfect reflection surface.

Through a careful set of visualization experiments, and the use of multivariate analysis to take account of the uncertainty in shock Mach number, triple-point trajectory angle, and slightly different shock wave arrival times at the trailing edge, the current work shows that the transition from one type of reflection to the other does indeed occur at the theoretical value. Conventional tests of reflection off a solid wall show significantly different transition results. Furthermore, it is also shown that the thermal boundary layer plays an important role in this regard. It is thus confirmed that viscous and thermal effects are the reason for the paradox. Reasons are also suggested for the counter-intuitive behaviour of the reflected shock wave angle.