The problem of strain hardening has been approached by preparing series of similar cylindrical chalk samples front outcrop material. The samples in each series could then be given a certain degree of deformation beyond the elastic limit. Afterwards the samples could be tested in standard triaxial tests at different confining pressures in order to determine the resulting yield envelope. Such experiments were performed both under compressional and extensional conditions.
Under compressional conditions the preceding loading resulted in a corresponding stretching of the elastic region as expected. But this apparent strengthening was not strengthening in a general sense. The material cohesion was greatly reduced (like the tensile strength). Under extensional conditions there was little or no effect of previous straining of the samples. It is therefore proposed to replace the words strain hardening by compressional hardening, to indicate that hardening is only hardening towards further compression.
The results from the extension tests defined one single failure line, irrespectively of previous straining, saturation fluid and chalk type. Assuming that the Mohr Coulomb criterion gives a valid description of failure in chalk, it is shown that this failure line corresponds to a critical state line for extension.
Failure criteria like Drucker Prager that are symmetric around the hydrostatic axis in stress space, cannot give a complete description of chalk behaviour.
A general criterion that includes intermediate stress and reflects both compressional and extensional failure, must have more like a convex triangular cross section with the hydrostatic axis in stress space, like the Lade's criterion.