2002ApJ...577..206E

The origin of the empirical laws of galactic scale star formation is considered in view of the self-similar nature of interstellar gas and the observation that most local clusters are triggered by specific high-pressure events. The empirical laws suggest that galactic scale gravity is involved in the first stages of star formation, but they do not identify the actual triggering mechanisms for clusters in the final stages. Many triggering processes satisfy the empirical laws, including turbulence compression and expanding shell collapse. The self-similar nature of the gas and associated young stars suggests that turbulence is more directly involved, but the energy source for this turbulence is not clear, and the small-scale morphology of gas around most embedded clusters does not look like a random turbulent flow. Most clusters appear to be triggered by other nearby stars. Such a prominent local influence makes it difficult to understand the universality of the Kennicutt and Schmidt laws on galactic scales. A unified view of multiscale star formation avoids most of these problems. The Toomre and Kennicutt surface density thresholds, along with the large-scale gas and star formation morphology, imply that ambient self-gravity produces spiral arms and giant cloud complexes and at the same time drives much of the turbulence that leads to self-similar structures. Localized energy input from existing clusters and field supernovae drives turbulence and cloud formation too, while triggering clusters directly in preexisting clouds. The hierarchical structure in the gas made by turbulence ensures that the triggering time scales with size, thereby giving the Schmidt law over a wide range of scales and the size-duration correlation for young star fields. Reanalysis of the Schmidt law from a local point of view suggests that the efficiency of star formation is determined by the fraction of the gas above a critical density of around 10⁵ m(H₂) cm^–3. Such high densities probably result from turbulence compression in a self-gravitating gas, in which case their mass fraction can be estimated from the density distribution function that results from turbulence. For Wada & Norman's lognormal function that arises in whole-galaxy simulations, the theoretically predicted mass fraction of star-forming material is the same as that observed directly from the galactic Schmidt law and is ∼10^–4. The unified view explains how independent star formation processes can combine into the empirical laws while preserving the fractal nature of interstellar gas and the pressurized, wind-swept appearance of most small-scale clouds. Likely variations in the relative roles of these processes from region to region should not affect the large-scale average star formation rate. Self-regulation by spiral instabilities and star formation ensures that most regions are in a marginally stable state in which turbulence limits the mass available for star formation and the overall rate is independent of the nature of the energy sources. In this sense, star formation is saturated to its largest possible value given the fractal nature of the interstellar medium.