DETERMINING THE AIR/WATER PARTITION COEFFICIENT TO EMPLOY WHEN CALIBRATING FORENSIC BREATH ALCOHOL TEST INSTRUMENTS

Calibrating and testing forensic breath test instruments using wet-bath simulators heated to 34 degrees C is fundamentally dependent on the assumed air/water partition coefficient for ethanol employed. Models developed from the early experimental work of Harger et al. (1950), defining the partition coefficient as a function of temperature, have formed the basis for many calibration and testing protocols. However, it appears these models may have been developed from the inappropriate fitting of log-transformed data by linear least squares. Fitting the data appropriately by nonlinear regression yields a model that, if employed, would calibrate an instrument lower by 2.5% relative to the naive model. Moreover, the more appropriate model is in better agreement with the more recently published models of Dubowski (1979) and Jones (1983). Forensically, the effect on instrument calibration is insignificant with a maximum difference of only 2.5%.