What was measured in Millikan's study of the photoelectric effect?

Millikan's photoemission paper of 1916 is of great historical importance because it provided the first unambiguous experimental support for Einstein's theory of the photoelectric effect. His graphical presentation of the results, for the correlation of stopping potential with frequency of the incident light on a sodium surface, is widely reproduced in textbooks, both at the high school and undergraduate levels. By far the most significant result that emerged from the experiments is the slope of this graph, which provided an accurate value of the Planck constant h. However, some modern textbooks also emphasize the intercept on Millikan's graph, even though Millikan made little use of this. As previously emphasised, the intercept does not give a measure of the work function of the illuminated sample, but measures that of the collector surface. Millikan made this point very clear, and his experimental plots have been corrected for this, but it is still not generally realized that this is so. Even with Millikan's correction, his intercept does not give the true work function for sodium. Neither of his surfaces was metallic, though this was assumed by Millikan and is assumed in textbook accounts. It is shown here that the sodium surface was almost certainly heavily contaminated with sodium hydroxide. The consequences for the interpretation of the experiment, and for the use of these results in teaching, are explored.