'Superwinds' - galactic-scale mass outflows driven by the collective effect of supernovae and winds from massive stars - have a rich legacy in theoretical astrophysics. They figure prominently in models for the formation and evolution of galaxies (e.g., Dekel & Silk 1986; White & Frenk 1991; Mathews 1989; Berman & Suchkov 1991) and of the intergalactic medium (e.g., DeYoung 1978; Ostriker & Cowie 1981), for the dynamics and structure of the interstellar medium and Galactic Halo (e.g., MacLow & McCray 1988, hereafter MM; Heiles 1990; Dettmar 1993), for the origin of the cosmic X-ray background (e.g., Bookbinder et al. 1980), and for some types of QSO absorption-lines (e.g., York et al. 1986).
Despite their potential importance, superwinds have remained largely in the realm of theoretical deus ex machina until quite recently. This situation has changed dramatically in the last several years, and our primary goal in this paper is to review how X-ray, optical, IR, and radio observations of starburst galaxies have been used to discover superwinds and to quantify many of their basic physical properties. This review is designed to complement two other recent reviews by Dettmar (1993) and Begelman (this volume). While Dettmar's emphasis is on the 'disk-halo connection' in typical spiral galaxies, the emphasis in the present review is on starburst galaxies (e.g., galaxies in which the gas depletion time is much less than a Hubble time and in which the region of star-formation is largely concentrated to the central kpc or so of the galaxy). Likewise, we leave to Begelman the discussion of outflows driven by powerful AGNs. It is indeed possible that some of the outflows we will discuss in this review are driven by AGNs rather than by starbursts. As we have argued elsewhere (Heckman, Lehnert, & Armus 1993), this is not likely to be true in the majority of cases however.
In order to place the importance of these multi-waveband observations in context, we will begin our review in Section 2 with a brief summary of the basic theoretical and astrophysical underpinnings of the superwind concept. After extensively discussing the most relevant observations in Section 3, we will describe in Section 4 how our recent 'census' of superwinds can be used to estimate the rate at which they eject mass and energy in the local universe. We will then conclude the review in Section 5 with a summary of the potential implications superwinds have for many of the topics considered during this Summer School. We direct motivated readers to the paper by Heckman, Armus, & Miley (1990 - hereafter HAM) for a more detailed discussion of some of the material to follow.