INTRODUCTION?
It is now widely accepted that the brittle fracture of rock is a process from microcrack proliferation to the formation of the main fracture. According to the Griffith theory of brittle fracture of rock, the main fracture surface is formed by the coalescence of a great number of favorably oriented microcracks. However, the mechanism by which microcracks coalesce into the main fracture has not been clearly known yet, though some hypothesis, e.g., [1], has been proposed. Many geologists often use the steps or antisteps on fault planes as an indicator of the direction of fault displacement, but the genesis of these steps is still a problem under discussion [2]. With these problems as the subject of this paper, a series of triaxial compression tests of a sandstone were made at moderate temperatures and confining pressures. The specimens were loaded to three different stress levels, viz., 100%, 80% and 50% of the corresponding compressive strength and then the tested specimens were examined both macroscopically and microscopically. From the observation and analysis of microcracks in thin sections prepared from tested specimens, some interesting phenomena were found. On the basis of test results, a mechanism was proposed for the transition from microcrack proliferation to the formation of the main fracture and also for the genesis of antisteps on fault planes.
TESTS AND RESULTS
The rock tested is a sandstone consisting of detritus, quartz (about 35%), feldspar and calcite incompletely cemented by calcareous cement. The initial porosity is around 3% and the average grain size is about 0.3mm. Cylindrical specimens 20 mm in diameter and 50 mm in length were cut from a same block of the sandstone and triaxial compression tests of air-dried specimens were made under the temperature and pressure condtions of 20°C, I MPa; 45°C, 39 MPa; 75°C, 65 MPa; 105°C, 91 MPa and 135°C, 117 MPa. The axial strain rate was 5x 10-5s -'. Under these conditions, the elastic modulus and compressive strength of the sandstone varied within ranges of 400-700 MPa and 4-10 GPa, respectively. The test was conducted by the 8000 kN servo-controlled high p-T creep test apparatus in the Geodynamical High Temperature and High Pressure Laboratory, Institute of Geophysics, Academia Sinica. Silicone oil was used as the pressure medium and the specimen was sealed in a rubber sleeve to prevent it from oil penetration. For every test condition, a specimen was first loaded up to failure to deterimine the compressive strength under that condition. For each of the three test conditions: 7512 ,.65 MPa; 10512,91 MPa and 13512,117 MPa, two more specimens which were loaded to 80% and 50% of the corresponding compressive strength were added. The specimens after failure or unloading were carefully examined macroscopically to know their overall appearance and were then sliced to prepare thin sections for microscopic observation. The thin section was perpendicular to the fracture surface and parallel to the specimen axis. Microscopic observations were carried out with an Olympus-Vanox polarization microscope. Cracks in thin sections were divided into four types according to their lengths, viz., cracks of types I (0.1-0.Sram), 2 (0.5-1.0ram), 3 (1.0-2.Sram) and 4 (> 2.Smm). With the aid of the movable objective table of the microscope, cracks of the four types were counted separately and their angles of inclination to the direction of ó were measured. In order to understand the variation of crack density with the distance from the main fracture, microcrack counting was also performed in three consecutive fields of view which had their centers aligned perpendicular to the trend of the main fracture.
