As an example one can use the results of Bowen (1988) to estimate the shock strength at around 2 stellar radii and compare the minimum required shock speed with the actual speed of the shock. In Bowen’s model without dust (his fig. 5) the shock strength in terms of a pressure ratio is around 15. The propagation speed of the shock is around 15 km/s which is close to the minimum speed for isothermal shocks above which sound waves are dragged out by the flow. This makes it very hard to conclude whether or not the simulations yield Vsw > 0. In Bowen’s figure 5 the shocks are clearly not isothermal but the radiative cooling law was necessarily primitive. A more complete treatment of the cooling would probably make the shocks more radiative and therefore more closely isothermal. This would also reduce the pressure ratio at the shock. Both ef2 On the propagation of sound waves in a stellar wind traversed by periodic strong shocks be valid. For an isothermal saw-tooth wave this means that the pressure ratio at each discontinuity should be less than 3=2 + 1=2 p 5 2:6. For adiabatic saw-tooth waves in a gas with a polytropic equation of state the pressure jump should be less than 1 + 1=4( + 1) + 1=4 p ( + 1)2 + 16. This is approximately 3