Surface energy and surface stress have been thoroughly discussed in th
e literature. However, the widely accepted definition of surface stres
s as the strain derivative of the total free surface energy divided by
the surface area (Shuttleworth, Herring, etc) is inconsistent with th
e correct definition of a stress tensor (given in continuum elasticity
theory) as the strain derivative of the specific free energy. We anal
yse this anomaly in detail in order to prevent the erroneous use of th
e surface stress definition.