Mass flow calorimetry represents a useful technique for studying calorimetric behavior of a number of systems. However, a systematic error can be introduced at the level of the calibration constant variation. The error is developed directly from linear regression analysis equations, and should be generalizable to any application of the statistical methodology, possibly even for the non-linear variants of regression analysis. This error is developed and illustrated with some recent data purporting to 'prove' cold fusion.