To the Editor.
—The article by Mackowiak et al1 raises several interesting points regarding clinical thermometry, but their results should be interpreted with some skepticism. The authors are correct in stating that many of Wunderlich's observations are not relevant to modern clinical thermometry. Before accepting their data as the true range of normal, however, we should be sure that the instrument is indeed measuring what we wish to measure. First, the authors report that the steady-state error of the Diatek 500 thermometer (Diatek Inc, San Diego, Calif) is 0.05°C to 0.07°C. However, it appears that the authors are not using the thermometer in a steady-state mode (called "monitor" mode by the manufacturer) but instead are using the "predictive" mode, in which the thermometer predicts the temperature based on the rate of rise of the temperature of the sensor rather than the actual equilibrium temperature. We have found in vitro