Normal galaxies span wide ranges of luminosity and many other parameters such as density, intensity, metallicity, extinction, or light-to-mass ratios. This great diversity in properties, and the complex mix of physical conditions within each system, have frustrated attempts at deriving simple models for the behavior of these ``unremarkable'' galaxies. Surveys of galaxies in the local Universe have thus become particularly valuable for codifying the empirical evidence. Survey data are typically analyzed to generate statistical information or to address specific questions, usually using global or integrated properties. For instance, luminosity functions are derived, or correlations between parameters are established. In such statistical studies, close attention must be paid to the sample used, for its completeness, selection biases, parameter-space coverage and other aspects could affect the results substantially. Statistical results are usually interpreted in phenomenological terms, explaining the observed trends for instance as reflecting the mixing of two components with distinct properties.
A complementary approach is to study in great detail a few nearby, spatially well resolved systems, and to derive insight from the similarities or contrasts among these cases. Such detailed case studies tend to search for physical insight by relating observational trends to the variation of a physical parameter of the system. While similar insights can also be pursued with statistical studies, case studies can make more convincing arguments based on detailed physical models of the ISM on small scales where homogeneity can be assumed.
Ultimately, one would like all these diverse interpretations to be collected into a self-consistent and complete picture. The main obstacle to such unification is that each observable Pobs is a complicated integral over the physical system:
where the integrals are taken over the solid angle of the galaxy and
through the line-of-sight respectively, and the contribution at each
point is a function of radiation intensity U, density n, temperature T,
and weighting functions XP such as geometry or optical depth
effects; the form and dependencies of fP
vary greatly among observables. While the availability of spatially
resolved data
simplifies the problem by narrowing down the solid angle and
therefore the variability of fP, it does not eliminate it entirely.
The empirical pursuits concerning normal galaxies typically aim at
deriving a better description of the infrared properties, especially in terms
of identifying the ``fundamental parameters'' that stand behind the many
correlated parameters, and pinning down the precise significance of observables
in terms of physical parameters. Another main goal is to pin down
similarly the fundamental sequence defined by the evident progression
of infrared properties, and to understand its key significance. Is the true
sequence simply defined by the infrared spectral energy distribution?
Is it driven by intensity variations? Or by the UV content of the
heating spectrum,
and therefore the galaxy's content of young stars? Are there secondary
parameters defining truly significant multi-dimensional families of properties?
Can low-luminosity analogs be defined for objects with extreme properties,
making more cases available for study in the Local Universe? An improved
empirical understanding of the data in these terms will help improve our
understanding of extreme systems, help us better plan high redshift surveys,
and understand their biases, and will feed directly into deciphering the
data on
the infrared cosmic background and on source counts at faint levels.
The physical understanding of star formation on the scale of galaxies must
start by addressing the episodic behavior of star formation, and the apparently
chaotic behavior on the kpc scale within disks. Can
a cycle be identified with well defined phases? Can those phases
be distinguished by observations, and can physical drivers and inhibitors
for episodes be identified? Can a galaxy then be understood in
terms of a superposition of phases of many overlapping episodes, and can
the factors be identified which regulate the ``steady-state'' of the whole disk
and the correlations between global parameters? Ultimately, physical models
of star formation in galaxies would be used to simulate the history of
star fromation and chemical evolution in the Universe, and construct more
reliable models of primordial galaxies to guide the search for those
first-generation objects.