Theoretical considerations and observational data support the idea that mergers were more frequent in the past. At redshifts z = 2 to 5, violent interactions and mergers may be implicated by observations of Lyman-break galaxies, sub-mm starbursts, and active galactic nuclei. Most stars in cluster ellipticals probably formed at such redshifts, as did most of the halo and globular clusters of the Milky Way; these events may all be connected with mergers. But what kind of galaxies merged at high redshifts, and are present-epoch mergers useful guides to these early collisions? I will approach these questions by describing ideas for the formation of the Milky Way, elliptical galaxies, and systems of globular clusters.

10 pages; LaTeX source. Invited review at "After the Dark Ages: When Galaxies Were Young", 9th October Astrophysics Conference, University of Maryland, October 12-14, 1998. Also available at http://www.ifa.hawaii.edu/faculty/barnes/mga.html