The widths of the Iranian and Tibetan collision segments of the Alpine belt are found to be proportional to the local convergence rates. Both segments are plateaus, but they stand at different elevations. The simplest reason for the width-rate correlation is that both increase with distance from the poles of relative rotation of the colliding plates. The plateau morphology and the height difference cannot be explained by a viscous or an elastic-plastic lithosphere. Instead, the difference in plateau height can be explained by proportionality to the total mass flow into the collision belt. The width-slip-rate relation provides a tool for estimating ancient plate motions. For example, the inferred Uralide collision rate was 5 to 9 cm/yr, with a pole of rotation at lat 81°N, long 75°E.