In 1913 W. Gaede published a theory of the molecular drag pump, including data from an experimental pump which bears his name. To the author’s knowledge, all reviews of this subject have reproduced Gaede’s theory without qualification. However Gaede himself noted that for molecular flow his theory is off by orders of magnitude. This discrepancy was forgotten by subsequent authors. At Varian SpA, it was found in both molecular and viscous flow, that Gaede's model may be corrected with the addition of an active "pumping" leak at the end of the channel, for which there is an exact and simple theory. Design of the channel to the leak-limited compression ratio is specified by a "golden rule". We also proved that a differential velocity between adjacent channel surfaces is not necessary for the pumping action. Gaede’s later designs suggest that he knew this. The modern Gaede model can now be applied to the pumping action of the Holweck pump in the direction of surface drag. What Gaede did not discuss is the possible influence of Bernoulli and inertial effects at high pressure. We will conclude with a discussion of the importance of inertial effects in the Gaede pump model.