There are systematic variations between relative noble gas abundances and lithophile tracers such as 87Sr/86Sr, εNd and La/Sm in a suite of basaltic glasses from the South East Indian Ridge (SEIR). 4He/40Ar⁎ (where 40Ar⁎ is 40Ar corrected for atmospheric contamination) correlates positively with 87Sr/86Sr and La/Sm but anticorrelates with εNd. The large range in 4He/40Ar⁎ observed in the glasses is due to fractionation during magmatic degassing caused by the very different solubilities of He and Ar in silicate liquids, whereas 87Sr/86Sr, εNd, La/Sm, etc. are insensitive to magmatic processes but rather reflect mantle heterogeneity. Thus, there is a curious situation in this suite of basalts where tracers of mantle heterogeneity (87Sr/86Sr, εNd, La/Sm, etc.) correlate with a tracer of magmatic volatile processes (4He/40Ar⁎). Here, we propose that “enriched” mantle (with high La/Sm and 87Sr/86Sr, low εNd) also has a higher C concentration than “depleted” mantle. Magmas derived from enriched mantle will therefore have higher initial C concentrations, leading to a greater fraction of CO2 degassed and thus a higher 4He/40Ar⁎ ratio on eruption. Simple solubility-determined fractional degassing models show that the range in 4He/40Ar⁎ observed in SEIR basaltic glasses can be generated if the mantle C concentration varies by a factor of 2 over the length of the ridge, consistent with independent estimates of C concentration heterogeneity in the MORB mantle. The correlations between lithophile tracers and 4He/40Ar⁎ can be reproduced by mixing between a depleted endmember with 87Sr/86Sr = 0.70275, εNd = 8.2 and [C] = 12 ppm and an enriched endmember with 87Sr/86Sr = 0.70360, εNd = 5 and [C] = 24 ppm, followed by degassing. The proposed degassing model allows us to estimate the initial C concentration (i.e. prior to degassing) of each SEIR basalt (for which Sr or Nd isotopes are available); using independent Nb concentration data (Mahoney et al., 2002), we show that C/Nb ratios prior to degassing along the SEIR are relatively constant, probably with a C/Nb ratio of 200±100. However, although the constancy of C/Nb in these samples is a robust conclusion, the estimated C/Nb ratio itself is model dependent. We also use these data to evaluate volatility of Re during degassing of MORBs; Re is known to be moderately volatile during subaerial and shallow marine volcanism, although it is not known if this element is also volatile at conditions appropriate to MORB emplacement. Although there is a (poor) correlation between Re/Yb (Yb being a non-volatile element of similar apparent bulk compatibility to Re) and 4He/40Ar⁎ in these samples, it is more likely that this correlation results from Re/Yb variation in the mantle source and is not due to loss of Re during magmatic degassing.