I've recently been researching how to measure strain using strain gauges. I came upon the subject of shunt calibration and hit a mental roadblock. From what I understand, shunt calibration simulates a strain across a strain gauge using a comparatively large resistor placed in parallel with said strain gauge. This virtual strain can be worked out rather simply and the readout obtained from the shunt load (i.e. the value output from the measuring system) can be correlated to the strain. My problem is that, from what I can see, shunt calibration is done with one resistor. How does one account for the deviance of the strain gauge through a range of possible loads if one only has one calibration point?