The Hertzsprung phenomenon, in which secondary bumps in the light or velocity curves of cepheid variables occur at systematically earlier phases as the period lengthens, has been known for about 50 years. As part of his programme on non-linear computations of cepheid models, Christie has interpreted the effect as a progressive wave which originates near the surface, travels to the core and is reflected back. He found that the phase varied with period in the correct sense, but to make his models fit both the Hertzsprung and period-luminosity relations he had to assign to them masses only about a half of those expected from evolutionary theory. The discrepancy is one of the outstanding problems in the field.