Effects of rounding errors on postmortem temperature measurements caused by thermometer resolution.

Beginning 7 h after death, a datalogger was used to measure the temperature in the external auditory canal of an adult male body placed in a refrigerated room. The sequence of measured values approximated a single exponential function with a correlation coefficient of 0.998475. This suggests that the starting time of body cooling in the refrigerated room under constant temperature can be calculated with less error using any two data points recorded by the datalogger. However, the results of such calculations varied widely and longer postmortem intervals demonstrated greater calculation errors. Periodic errors also appeared. Mathematical simulations showed that this variation was caused by rounding errors, which represent the difference between the thermometer readings and the true temperature. The resolution of the thermometer was 0.1 degrees C, a normal specification; however, even this led to noticeable rounding errors. Therefore, significant errors may influence postmortem interval estimations using other body temperatures. When body temperatures are used to determine the time of death, a method that minimizes rounding errors should be considered.