Dr Penman wrote kindly about my paper, but his letter does not alter the fact that:
(i) Terzaghi failed to confirm Mohr's hypothesis; there is not a 'true' curved envelope for peak strength data; any approximation by a straight line with cohesion and friction parameters in Mohr's equation is unsound;
(ii) there is a sound definition of 'friction' in the limited class of material that is the true subject of 'soil' mechanics.
Coulomb's soil mechanics differentiated between granular material whose disturbed parts do not adhere, and solid rock with tensile adhesion. Soil mechanics is about a particular class of granular materials which have an 'angle of repose' which Coulomb called their 'angle of friction'.
This angle is seen in conical heaps of such soil, submerged below water or standing dry in air.
Shear strain of a dense soil continuum causes dilation, called 'interlocking' by Taylor, which produces peak strength data that Terzaghi misinterpreted. The 'friction' which resists slip of solid surfaces in contact is a different subject of study; Coulomb's soil is a continuum. Coulomb's 'angle of repose' implies that his soil has critical states.
If I were writing the book on Critical state soil mechanics today with Peter Wroth, I would describe critical states as occuring below a slope of Coulomb's soil at the angle of repose.
Roscoe once had to consider the properties of a tip of broken china plates; we did not expect such material to fit a critical state model. In contrast, I had experience with roads and embankments made of well compacted selected fill which led me to consider the c=cu or the =d analyses to be better starting points for geotechnical design of stable soil bodies than analyses using large values of Terzaghi's 'true' c and small values of 'true' .
Ideas set out in Critical state soil mechanics evolved in my undergraduate teaching and later lectures. In my 1980 Rankine lecture I explained other ground behaviour mechanisms seen in centrifuge model tests. Soil compacted below 50% equivalent liquidity is brittle; high hydraulic gradients across fissuring ground cause pipes, channels, or liquefaction.
For much construction, soil tests could find the critical state properties of the selected fill materials, the designs could aim for ductility, and construction bugets could include the cost of tests to failure of centrifuge models.
Where 'observational methods' are used they could be based on model tests to failure. Benjamin Baker wrote that if an 'Engineer' had no experience of failure of retaining walls of his own construction it merely proved that his practice was not extensive. He regarded papers on failures as more valuable than papers on unproven theory or on design and construction of works which have not failed.
Engineers who put thought into a review of the Mohr Coulomb equation, and who comission centrifuge tests to failure of models, will find that both these activities give them new insights into Ground Engineering.
Professor Andrew N Schofield
University of Cambridge