Intramolecular isotope distributions can constrain source attribution, mechanisms of formation and destruction, and temperature-time histories of molecules. In this study, we explore the D/H fractionation between central (-CH 2 -) and terminal (-CH 3 ) positions of propane (C 3 H 8 ) -a percent level component of natural gases. The temperature dependence of position-specific D/H fractionation of propane could potentially work as a geo-thermometer for natural gas systems, and a forensic identifier of specific thermogenic sources of atmospheric or aquatic emissions. Moreover, kinetically controlled departures from temperature dependent equilibrium might constrain mechanisms of thermogenic production, or provide indicators of biological or photochemical destruction. We developed a method to measure position-specific D/H differences of propane with high-resolution gas source mass spectrometry. We performed laboratory exchange experiments to study the exchange rates for both terminal and central positions, and used catalysts to drive the hydrogen isotope distribution of propane to thermodynamic equilibrium. Experimental results demonstrate that D/H exchange between propane and water happens easily in the presence of either Pd catalyst or Ni catalyst. Exchange rates are similar between the two positions catalyzed by Pd. However, the central position exchanges 2.2 times faster than the terminal position in the presence of Ni catalyst. At 200°C in the presence of Pd catalyst, the e-folding time of propane-water exchange is 20 days and of homogeneous exchange (i.e., equilibrium between central and terminal positions) is 28 minutes. An equilibrated (bracketed and time-invariant) intramolecular hydrogen isotope distribution was attained for propane at three temperatures, 30°C, 100°C and 200°C; these data serve as an initial experimental calibration of a new position-specific thermometer with a temperature sensitivity of 0.25‰ per ˚C at 100 ˚C. We use this calibration to test the validity of prior published theoretical predictions. Comparison of data with models suggest the most sophisticated of these discrepant models (Webb and Miller, 2014) is most accurate; this conclusion implies that there is a combined experimental and theoretical foundation for an 'absolute reference frame' for positionspecific H isotope analysis of propane, following principles previously used for clumped isotope analysis of CO 2 , CH 4 and O 2 (Eiler and Schauble, 2004;Yeung et al., 2014;Stolper et al., 2014).