9 Conclusions

In this paper we have discussed the structure generated by the instability of the line driving mechanism. We have considered self-excited structure in the sense that we introduce no explicit perturbations at the base of the wind. The emphasis has been on the evolution of structure as it moves out with the flow, described by the radial dependence of statistical quantities such as the clumping factor and the velocity dispersion.

The gas in the outer wind is compressed into a sequence of narrow dense clumps bounded on the starward side by a reverse shock and on the outside by a forward shock. This is the reason why the structure dissipates. The clumpiness of the wind is maintained by collisions between clumps moving at different speeds, not by establishing pressure equilibrium between the clump and its surroundings. This conclusion is robust within the assumptions of this study. A critical assumption in this regard is the floor temperature. Pressure equilibrium could be established if the dense clumps were allowed to cool to very low temperatures. This can only be verified through a detailed investigation of the energy balance in a structured wind, which is a complex problem and is deferred to a later study. The one-dimensionality of the models presented here is bound to be important for clump collisions, as clumps moving at different speeds have to collide, whereas in a multi-dimensional model they can pass each other without colliding.

Clump-clump collisions are also the reason why the strongest density contrasts are formed relatively far from the star (a few tens of stellar radii). Interestingly, this is the region where the millimetre continuum is formed. The results from this study can thus be verified by a comparison of radio and millimetre continuum fluxes.

We have shown that replacing the customary value of $10^{-3} \kappa_{0}$for the line-strength cut-off by a less artificial value drastically increases the amount of structure. This can be important for i.a. the production of X-rays and further simulations will benefit from imposing a cut-off as close as possible to the realistic value of $\kappa_0$

We have also shown that the outer-wind evolution of structure is a pure gasdynamical problem, in the sense that is not affected by external forces. This allows us to consider more economical models for future work.

Acknowledgements

We thank R. Blomme, D. Cohen, A. Feldmeier, K. Gayley, J. Puls and S. Van Loo for many stimulating discussions on line-driving and wind structure. This research was supported in part by NASA grant NAG5-3530 to the Bartol Research Institute at the University of Delaware. MCR gratefully acknowledges support from ESA-Prodex project No. 13346/98/NL/VJ(ic).