Supernova (SN) explosions of massive stars are a dominant source of kinetic
energy in the interstellar medium. Strong, supersonic stellar winds
are also an important source in the case of the extreme most massive
stars ( 40
*M*_{}).
Mechanical feedback structures the ISM,
with immediate consequences corresponding to supernova remnants
(SNRs), stellar wind-driven bubbles, and superbubbles resulting from
combined SNe and winds from multiple stars.

While the SNRs evolve, in the simplest description, according to the
Sedov (1959)
model, the wind-driven bubbles and superbubbles are
thought to evolve according to a similar,
Sedov-like adiabatic model for constant energy input
(Pikel'ner 1968;
Castor et al. 1975):
the central supersonic wind drives a shock
into the ambient ISM, piling up a radiatively cooled, dense shell;
and a reverse shock near the source thermalizes the wind's kinetic
energy, thereby generating a hot (10^{6} - 10^{7} K),
low-density
(*n* ~ 10^{-2} - 10^{-3} cm^{-3}) medium
that dominates the bubble
volume (Figure 3). This heating process is
believed to be the origin of the
diffuse hot, ionized medium (HIM) in the interstellar medium.
Assuming that the hot bubble interior remains adiabatic, the
self-similar shell evolution follows the simple analytic relations,

(3.1) |

where *R* and *v* are the shell radius and expansion velocity,
*L* is
the input mechanical power, and *t* is the age. Once SNe begin to
explode, they quickly dominate *L*, and the standard treatment is to
consider the discrete SNe as a constant energy input (e.g.,
Mac Low & McCray 1988).
Hence, we may write *L* in terms of the SN parameters:

(3.2) |

where *N*_{*} is the number of SNe,
*E*_{51} is the SN energy, and *t*_{e}
is the total time during which the SNe occur.

There are several approaches to testing the standard, adiabatic shell evolution, and by extension, our understanding of mechanical feedback. In the first instance, we can examine the properties and kinematics of individual shell systems and carry out rigorous comparisons with the model predictions. Secondly, we can also examine statistical properties of entire shell populations in galaxies, and compare with model predictions. And thirdly, we can carry out spatial correlations of shells with regions of recent star formation, to confirm the existence of putative stellar progenitors. All three of these methods require high spatial resolution, and thus the Local Group offers by far the best, and often the only feasible, laboratory.