Résumé
A series of two-stage torsion experiments on Carrara marble were conducted to constrain the influence of known prior deformation on rheological response and microstructural evolution. Comparison with previous experimental data on Carrara marble flow and fabric evolution during single-stage torsion deformation experiments provided direct insights into the significance of initial deformation at various conditions. Our experiments were conducted at 727 degrees C temperature and 300 MPa confining pressure, while maintaining a constant strain rate of 3 x 10(-4) s(-1) on the periphery of the cylindrical samples. Under these conditions, the marble is known to deform in power-law (n = 6-10) ductile flow. All torsion experiments were performed with a Paterson type gas-medium testing machine equipped with a torsion actuator module.;Prior (D1) and subsequent (D2) deformation are accomplished by two torsion experiments in sequence on same sample segments. The effect of D1 strain history is investigated during D2 by applying counter-clockwise torsion to a sandwich sample consisting of three segments with different D1 rotation sense. D2 samples experienced continued, first and reversed shearing deformation in top, centre and bottom segments, respectively. D2 bulk strain was chosen equal to D1 strain in top and bottom segments.;D1 experiments followed the typical single-stage deformation behaviour of Carrara marble under the applied experimental conditions. Yielding was followed by strain hardening until a peak stress was reached at a shear strain around 1, after which work softening occurred. Weakening gradually evolved into a constant stress regime. During hardening, a shear microstructure and crystallographic preferred orientation (CPO) developed. Afterwards the volume fraction of smaller dynamic recrystallised grains increased continuously, resulting in a recrystallisation microstructure and CPO at shear strains of 5 and higher. The new D2 experiments displayed a strain variation between homogeneously deformed sample segments that increased with increasing D1 strain. The stress-strain behaviour of the D2 bulk sandwich samples showed less pronounced work-hardening and -softening when compared with single-stage deformation experiments. Furthermore, constant flow stress was attained at increasingly lower strain with increasing D1 strain. In most D2 segments, fabric development is equivalent to single-stage experiments at corresponding absolute strain. The fabrics differed markedly from those in single-stage experiments in the case of reversed two-stage deformation at moderate strain (D1 shear strain 1 and 2.6). Experiments show that grain shape by shearing of relict grains is defined by finite strain and thus affected by strain reversal. Recrystallisation is controlled by absolute strain and not influenced by strain reversal.