Models of hydrolytic weakening based on the influence of chemical environment on the concentrations of point defects stimulated a series of experiments by Ord and Hobbs <1986> in which single crystals of natural quartz were deformed in a solid medium apparatus under a range of buffered oxygen, water, and hydrogen fugacities. Microstructures in these specimens have been examined using visible light and transmission electron microscopy in an attempt to identify processes responsible for the observed relationships between chemical environment and strength. Preheating treatments, conducted for ~20 hours at the pressure, temperature, and chemical conditions of each ensuing deformation experiment, modified each specimen microstructurally such that the quartz was not in the form of a low-dislocation-density single crystal when deformation testing began. Specimens fractured axially during pretreatment, particularly those in chemical environments for which water fugacity remained high. Fractures later healed giving rise to arrays of fluid inclusions in regions with moderate densities of dislocations. Some deformation and recrystallization, concentrated at healed fractures and specimen end zones, also accompanied pretreatment. ''Mn-buffered'' specimens, for which fO2 and fH2O, were high, were densely fractured; in less strained regions, subsequent deformation was restricted to the vicinity of healed fractures, while regions of higher plastic strain and recrystallization show evidence from inclusions for abundant early fractures, now obliterated. ''Ta-buffered'' specimens, of low fO2 and fH2O, are also deformed close to axial fractures, but here fractures are comparatively rare.

The marked variation in the degree of fracturing seems to be an example of stress-corrosion cracking in response to chemical environment. Dislocations are developed in material away from healed fractures in one deformed Ta-buffered specimen. If point defects introduced by diffusion are responsible for water weakening of quartz crystals, the complex microstructures seen in the deformed specimens of Ord and Hobbs <1986> require that point defect distributions were neither uniform nor controlled by bulk diffusion from the sample edges. The clear association between plastic strain and healed fracture does point to important local effects either of ''water'' upon dislocation activity or of healed-fracture structure having been involved in the nucleation of dislocations. Regardless of the details, we infer from the dependence of the degree of fracturing upon chemical environment, together with the high correlation between locations of plastic strain and regions of earlier fracture, that the relationship between chemical environment and macroscopic strength of natural quartz crystal is not simply a matter of water weakening as a direct result of high point defect concentrations. Instead, fracture, inferred to be important in natural fluid-rock interactions, also seems to be an experimentally critical process which might nucleate the dislocations necessary to initiate plastic deformation in the laboratory. ¿1991 American Geophysical Union