The idea of dwarf and giants stars, but not the nomenclature, was first established by Eijnar Hertzsprung in 1905; his first diagrams in support appeared in 1911. In 1913 Henry Norris Russell could demonstrate the effect far more strikingly because he measured the parallaxes of many stars at Cambridge, and could plot absolute magnitude against spectral type for many points. The general concept of dwarf and giant stars was essential in the galactic structure work of Harlow Shapley, Russell's first graduate student. In order to calibrate the period-luminosity relation of Cepheid variables, he was obliged to fall back on statistical parallax using only 11 Cepheids, a very sparse sample. Here the insight provided by the Russell diagram became critical. The presence of yellow K giant stars in globular clusters credentialed his calibration of the period-luminosity relation by showing that the calibrated luminosity of the Cepheids was comparable to the luminosity of the K giants. It is well known that in 1920 Shapley did not believe in the cosmological distances of Heber Curtis' spiral nebulae. It is not so well known that in 1920 Curtis' plot of the period-luminosity relation suggests that he didn't believe it was a physical relation and also he failed to appreciate the significance of the Russell diagram for understanding the large size of the Milky Way.