including vertical compression, much higher than could be accounted for by the weight of overlying strata. A similar result was obtained in Gorge Power House tunnel in granitic gneiss beneath a ridge.

The method used to obtain the stresses in the rock face, consisted of setting several strain gauges in the form of a rosette on a rock surface, cutting a slot around them into the rock, thereby freeing the rock enclosed by the slot from the stresses in the rock face ; measuring the resulting strains ; and computing the stresses on the basis of the elastic properties of the rock measured on samples in the laboratory. However, the elastic properties of a mass of rock in place may be considerably different from those measured on samples in the laboratory and on this account there is some doubt about the reliability of this method for determining rock stresses. The results obtained in these tests have been discussed by Terzaghi and Richart⁽¹⁾.

Numerous investigations, using a considerable variety of methods, have been made in France during the past 10 years. The methods and results have been reviewed Talobre⁽²⁾.

The methods used generally involved measuring the stresses directly, in place, on the surfaces of underground openings thereby avoiding the uncertainties inherent in strain relief methods where measurements of elastic properties on laboratory samples are required.

It was not usual to find that a direction of principal stress was vertical, and there was no general tendency for those approaching verticality to be the greatest. In a general way also it was found that stresses increased with thickness of cover but there was no strict relationship between them.

Most of these determinations appear to have been made in connection with hydro-electric projects which are usually in rugged terrain, and it seems possible that the non-verticality stress is due to the effect of the surface topography. During a short but valuable discussion on the existence of high horizontal stresses Berthier⁽⁷⁾ stated that measurements have been made (in France) where the horizontal stresses were considerably greater than the vertical stress.

Several determinations of rock stresses have been made in the Snowy Mountains of south-eastern Australia in granite and gneiss by Alexander and Worotnicki. Reference has been made to these tests previously by the author(8) and since they are now the subject of a full paper to this Conference by Alexander⁽⁹⁾, only a brief mention will be made of them here. Several tests were made in sites beneath exceedingly rugged terrain, and undoubtedly the stress field at these sites is influenced by the surface topography. After making allowance for these topographic effects, it appear that the horizontal stresses are at least of the same order and in some cases probably greater than the vertical stress. In the one specially selected to be free from complications arising from topography and shape of