High Integrity in Sensor Networks studies the effect of compromised data on sensor network data analysis and data fusion algorithms. We are at the start of a study to better understand the compromised data itself so that we can model it in such a way to characterize this affect on analysis. The accelerated life experiment we describe here was instrumented to see how the accuracy of a temperature and humidity sensor changes across mote and sensor systems and across time as the battery dies. We found that the error in the measurements was stable over time, but the bias in the measurements varied greatly across the motes. The experiment gave us more information on the important variables to include in calibration function, so that we can estimate the true measured value from the reported value of the sensor.