"Scalability" was a major theme of the conference. It arises from two main concerns: To what extent can we scale local starburst systems to cosmically distant ones? And to what extent can we scale the physics of modest to extreme star formation amplitudes? The evidence, fortunately, is that scalability is good, implying modest rather than fundamental adjustments with changes in environment and scale.
The premier example of scalability is the Schmidt-Kennicutt "law," under which SFR GAS1.4. The quantities refer to global averages over the surfaces of individual galaxies. The relation applies over a remarkable 6 decades. As noted above, a similar degree of scalability applies to the radio/far-IR correlation for star forming systems.
Other encouraging, if less firmly established, examples of scalability include:
Congruences in EM spectral shape for starbursts across a wide range of environments and amplitudes.
The continuity of starburst properties across a large range of amplitudes. It is possible to define a scaling sequence between the nearby (3.5 Mpc) archetypal starburst M82 (L ~ 2 × 1010 L), more distant ULIRGs (1012 L), and high redshift SCUBA sub-mm sources (1013 L).
The smooth increase of starburst activity with lookback time exhibits no evidence of a transition point where starbursts suddenly become more important.
Continuity of Lyman break galaxies (LBG's) at z 3 with more local systems. Careful studies, lately including GALEX data, show that properties (sizes, surface brightnesses, masses, kinematics) of LBG's are continuous with those of lower redshift luminous blue compact galaxies, some of which may be the progenitors of local dE galaxies (i.e. dynamically hot systems).
The mass-metallicity-extinction relation, which changes only slowly with redshift and has no transition points. The abundance scale seems to decrease smoothly with redshift.
The duration of starburst episodes is t ~ 100 Myr and seems similar at all redshifts. Individual galaxies may experience a number of such episodes.
The IMF for star formation on the scales of star clusters or galaxies now appears to be universal except in a small number of SSC's where there may be changes in MLOW. The massive star IMF appears universal, which is very important for analyzing feedback processes. (Progress here has been excellent despite many complications, e.g. limited spatial resolution, large differential extinction effects, and mass segregation.)
Star formation histories of nearby galaxies may all be similar for a given gas density, despite the presence of "noise" which gives rise to minibursts. It is important to understand the disk self-regulation mechanism.