The quasi-static gravitational collapse of clouds should lead to fragmentation into small Jeans-unstable cloudlets with a column density similar to that of the clouds as a whole. This criterion defines the mass and size of newly-formed fragments as the collapse proceeds. One could in principle expect that all collapsing matter should end up in a massive singularity at the centre of the collapsing configurations. However, the onset of star formation is shown here to stop the collapse. This should happen if the newly-formed stars produce winds while ramming through the left-over cloud and in this way cause the stirring required to stabilize the collapse. An inmediate consequence of this is the velocity dispersion generated in the star-forming region, which is supersonic in the case of massive clusters and detectable upon the appearance of massive stars in giant HII regions. The stirring cause by the supersonically moving wind-driven sources is also shown to cause a distinct cloud structure, or filling factor, in excellent agreement with recent observations of regions of violent star formation. The two effects, i.e. the acquired velocity dispersion and the filling factor caused in the parent cloud, allow us to differentiate between high- and low-mass clusters. In both cases, however, the disruptive energy from massive stars ends up erasing the clues stored in the gas during cluster formation.

Introduction

There are several trivial but definitive conclusions regarding the history of spheroidal stellar systems that one can infer from the present agglomeration of stars in globular clusters and galactic bulges.