The fractionation of 13C between calcite and graphite, ∆(Cc-Gr). is consistently small (2.6-4.8 permil) in 34 assemblages from upper amphibolite- and granulite-facies marbles of the Grenville Province. In 25 samples from the Adirondack Mountains, New York, it decreases regularly with increasing metamorphic temperature. The fractionations are independent of absolute δ 13C values of calcite (-2.9 to +5.0). For T = 600-800° C, the Adirondack data are described by ∆( Cc- Gr) = -0.00748 T (° C) + 8.68. This good correlation between ∆ and T suggests that carbon isotope equilibrium was attained in these high-grade marbles and that the theoretical calculations of this fractionation by Bottinga are approximately 2 permil too large in this temperature range. Because of the relatively high temperature sensitivity suggested by these results and by Bottinga's calculations, and the pressure independence of isotope fractionation, ∆(Cc-Gr) may provide a very good thermometer for high-grade marbles. Comparison of this field calibration for ∆(Cc-Gr) vs temperature with results from other terranes supports the utility of ∆(Cc-Gr) for geothermometry and suggests that graphite is much more sluggish to exchange than calcite, that exchange between calcite and graphite occurs at temperatures as low as 300°C, and that equilibrium may normally be attained only when peak metamorphic temperatures are greater than 500-600°C. Because 13C exchange is an unavoidable metamorphic process at temperatures above 300°C, high values of δ 13C(Gr) in moderate- to high-grade carbonate-bearing rocks do not provide a sufficient criterion to infer an abiogenic origin for the graphite.