Writing in the mid 1990s, Rabinowicz comments on adhesive wear rates between materials with differing hardness. He is surprised by the lack of work conducted on this topic, but says that the studies completed suggest that the ratio of the wear rates is proportional to the ratio of the hardness squared.
(W_hard/W_soft) = alpha * (H_soft/H_hard)^2
The figure shown in the book shows a lot of scatter: it is almost as bad as fatigue data. Consequently, I am not sure how useful the hardness relationship is in predicting the wear rate. Although the exact relationship may be unclear, the fact that there is a correlation cannot be disputed. It is also important to note that however large you make the hardness ratio, you cannot entirely eliminate adhesive wear from the harder surface.
Source: Friction and Wear of Materials by Rabinowicz
