1 INTRODUCTION
The problem of transfering the experimental data from laboratory to field conditions is a fundamental question of geomechanics. A number of trials have already been undertaken to solve the problem, both analytically (Holliday & Tharkay 1964), (Hashin 1964), and experimentally (Bieniawski 1968), (Gustkiewicz 1975), (Cyrul 1984), however, a satisfactory solution of the problem seems to be very distant. Nevertheless, recent investigations have provided ideas and data on the most imporatant factors that should be taken into consideration and thoroughly analyzed. These investigations allow for some of the general questions addressed prior to the start of any costly investigation program. Let us outline some of the principal questions and attempt to give a reasonable answer.
1.1 Should our approach to the above mentioned question be deterministic?
The anwser to this question is not simple but neither it is impossible. As a matter of fact the most popular approach, in most of areas of geomechanics, has been that of the deterministic type. A number of new physical relations and phenomena in rocks under load have been discovered using the deterministic approach. However, most of the achievements do not exceed the laboratory scale of consideration. Analytical solutions have usually been successfully verified experimentally using selected type of rocks and sophisticated equipment. Transformation of those relations into a large in situ scale has usually been unsuccessfull due to numerous assumptions and simplifications originating from so called scale (size) effect. Accordingly, the deterministic approach to mechanical behaviour of rocks has a limit which is set much below practical applications of the approach to mining engineering.
1.2 Should our approach be probabilistic?
A very complex structure of rocks that includes substantial random components seems to support such an aporoach. Random heterogeneity of rocks associated with anisotropy of almost any sedimentary rocks, complicated by random sets of different type discontinuities, gives a real picture of the rock structure. Utilizing deterministic tools to study such a complex structure seems to be unrealistic. However, the probabilistic approach has also several weak points. Whereas the deterministic hypothesis can be verified by one experiment, any probabilistic hypothesis usually needs great deal of data to be verified. The collection of the necessary data is very time consuming and costly, and therefore can be as unrealistic as that of the deterministic type. Another weak point of the probabilistic approach originates from the fact that the structure of rocks is not purely random. This of course depends on the scale of considerations. It is, however, obvious that many types of rocks, besides an evident random component in their structures, also possess a very visible deterministic one. Therefore, we face the the question of the employment of a compromise to our considerations which, however, should be scientific in nature. It seems that contemporary science has at its disposal such a tool which can be applied to some problems of rock mechanics and is called geostatistics. Later, geostatistics will be used to evaluate the data from the indentation tests and to make an inference of mechanical properties of rocks. It should be emphasized that this is the first application of geostatistics in the evaluation of mechanical properties of rocks.
