Specimens cut from a single synthetic quartz crystal and oriented normal to r( 1 101) were deformed in a gas apparatus at 3 kbar confining pressure and temperatures from 475° to 900° C. There was a large variation in stress—strain behaviour between specimens, which could be correlated with inhomogeneity in OH concentration as revealed by infrared absorption. This inhomogeneity was also revealed by the optical microscope as a banding in the distribution of strain parallel to the (0001) growth surfaces. Other optical features in the specimens were deformation bands parallel to the {101̄0} prism planes, undulatory extinction and, on a finer scale, optical lamellae of several orientations, especially prismatic and rhombohedral. Detailed transmission electron microscope observations were made at each temperature after various amounts of strain. The dislocation structures revealed can be broadly grouped into those produced up to 550°C which are suggestive of relatively low temperature behaviour, and those above 550°C which indicate that climb and other high-temperature recovery processes are active. A notable feature in the low temperature range is the generation of clusters of dislocation loops around submicroscopic inclusions already present in the initial material: this effect provides an alternative to the multiplication of grown-in dislocations and gives the equivalent of a much higher initial dislocation density than would the grown-in dislocations alone. Other low-temperature features include intense banding of tangles of dislocations, some of which can be associated with optical lamellae. At higher temperatures, the dislocations become more curved and intertwined and form many small, isolated loops and dipoles and occasional networks: also, bubbles appear above 800° C. At the lower temperatures, dislocation densities of the order of 10 9 cm −2 are soon reached in the yield region, rising to a maximum of over 10 10 cm −2 at larger strains ; the maximum densities reached at high temperatures are one to two orders of magnitude less. The most active slip planes appear to be m 10 10 and m 10 11 , with c [0001] and a 〈2 11 0〉slip directions, respectively. In discussion, constraints that the electron microscope observations can place on observations suggesting a substantial role of multiplication of the Frank-Read type, especially at inclusions, as opposed to the kinetic type of multiplication source thought to be the basis of the law used in the Haasen model. From this discussion, it can be concluded that further attempts to apply microdynamic theory to quartz would be much helped by more detailed experimental work, paying special attention to the following aspects: 1. (1) The use of homogeneous specimen material, well characterized in composition and initial defect content. 2. (2) Independent measurements on dislocation velocities and multiplication rates. 3. (3) Further study of work-hardening and recovery processes.