INTRODUCTION
Wet and dry specimens of three rocks of approximately granitic composition were deformed at different experimental conditions. These experiments were carried out in order to study the extent to which water, either added or released from mineral reactions, may affect the mechanical properties and the microstructure of these rocks. The experiments were conducted in a Griggs' hot creep apparatus at temperatures and pressures up to 900 º C and 1000 MPa respectively; and most of them were performed in the constant strain rate mode (1 . 10 -8<?<1 . 10 -4 s-¹). All three rocks were low porosity rocks, and the wet specimens never contained more than 0.9 wt % of added water. The experimental results are, however, quite unequivocal: They show that wet rocks are weaker than dry ones. A significant difference is even found for rocks tested in uniaxial compression at room temperature. A few experiments carried out on specimens obtained from larger prestressed rock cores indicated that the central parts of these cores were more likely to contain more microcracks than the ends. Optical microscope examinations revealed that extensive microcracking takes place even at the highest temperatures, where other mechanisms such as slip and dislocation creep also become important. Very little partial melting could be observed except for the specimens deformed at 900º C.
We need to know more about these properties in order to understand the mechanics of large crustal movements. The influence of water on these properties is of particular interest, since water is by far the dominating pore fluid in the upper part of the crust. Water is known to affect the mechanical properties of rocks in many ways. A great number of reports (eg. Duba, Heard, and Santor, 1974) argue that cracking and faulting are enhanced in the brittle regime, provided that there is enough water present for the fluid pressure to reduce the effective confining pressure. This rather vague and thus unsatisfactory argument does not explain how the water actually affects the rock. The results, however, support another important claim: That the mechanical properties of rocks to a large extent depend on the amount of pores and microcracks, and on the directional distribution of these defects. A water pressure in the cracks and the pores will oppose the external forces acting to close them, and may, at least in places, prevent the cracks - and to a lesser extent even the pores - from collapsing. The friction between the surfaces of the cracks will thus be lost. In addition, water is also likely to affect the cohesion, ie. the short range interaction, between the individual mineral grains. Stress corrosion cracking suggested by Tullis and Yund (1980), and reported by Atkinson and Meredith (1981), could just be some sort of combination of the two effects mentioned above. Broch (1974), investigated the mechanical properties of a number of rocks under both wet and dry conditions, and found that dry rocks were stronger than water saturated rocks.
