Huang et al. reply

Replying to F. M. Richter Nature472, 10.1038/nature09954 (2011) In our Letter 1 , we showed that the phenomenon of isotope fractionation in silicate melts by thermal diffusion, first reported in 1998 (ref. 2 ), can be characterised by a parameter Δ S T that is independent of temperature and composition. (Here Δ S T is the difference in the Soret coefficient, S T , between isotopes of a diffusing element.) Richter 3 questioned this finding by plotting Mg isotope ratio versus temperature data (figure 1 in ref. 3 ) for a subset of experiments from figure 3f of ref. 1 .