The Composition–Age–Velocity Connection
We stressed that the observed composition of stars is uniform. All stars have, to a very good approximation, the same composition. The justification for this fundamental result was the success of the Harvard spectral type classification, based on spectral line intensity ratios. The observations demonstrated time and again that, to a very good approximation, the overwhelming majority of stars (at least those nearby) exhibit similar line ratios for the same effective surface temperatures, thus indicating similar compositions. This general tenet survived until the beginning of the 1950s, when it became clear that this was only approximately correct. The problem was then completely turned round, to ask what was the source of the small difference in the abundances of the heavy elements.