It is shown that, in regions near peak strains, improved accuracy of measurement with a strain gauge or extensometer, can be obtained by taking readings over more than one distance including each point. A better estimate of the true strain can then be derived than will be given by a single reading. Peak errors in the region of 8 to 10 per cent can thus be reduced to 1 or 2 percent.

Two methods are described: one in which measurements are made at equal intervals along a line, keeping the gauge length constant; the other using different gauge lengths centred on the point concerned, the latter being suitable where a grid can be marked on the surface.