The history of star formation determines the evolution of galaxies and the generation rate of production of for heavy elements. It has been traced by deep Hubble Space Telescope (HST) imaging followed up with large ground-based telescopes. Even at modest redshifts, however, these techniques only probe the rest-frame ultraviolet. Here, SAFIR will make a critical contribution. Interstellar dust can absorb nearly all the UV in star-forming galaxies. In the best-studied starburst galaxies such as M82, a debate raged for more than a decade on how to correct even the near-infrared emission for interstellar extinction. Such corrections are poorly determined for galaxies at high redshift, so there are large uncertainties in the star-forming rate for z>1 when most of the heavy elements were created. These uncertainties can be removed only by measuring the far-infrared emission from dust heated by young stars in these galaxies. The importance of this approach is underlined by the large cosmic far-infrared and submillimeter energy density discovered by COBE. This background has been partially resolved by ISO in the very far-infrared and is thought to arise from starburst galaxies at redshifts of up to z~3. SAFIR will resolve most of this highredshift background into individual galaxies. This will allow us to image the dominant phases of dust embedded star formation and nuclear activity throughout the Universe. Since ultradeep optical images (e.g., Hubble Deep Field) reveal many galaxies too faint to contribute significantly to the submillimeter diffuse background, an entirely new population of optically faint, young galaxies must be responsible for it. A full understanding of star formation in the early Universe requires that we extend far-infrared and submillimeter measurements to these small systems. In this respect, SAFIR is crucial for understanding the processes by which primordial structure in the Universe leads to the first galaxies in it. In this luminosity range and over 1