New UVLAMP measurements of experimental noble gas crystal/melt partitioning values (including He) suggest reasonably incompatible behaviour for both olivine and cpx and no significant fractionation of noble gases relative to one another. This is consistent with models of noble gas incorporation at crystal lattice sites in both crystals (1). However the determined D values of approximately 8 x10^-4 for cpx and 5 x10^-3 for olivine suggest a small but significant amount of noble gas might be retained in the mantle after melting. It is also apparent that He is three orders of magnitude less incompatible than U and Th in olivine. As opx is predicted to show similar characteristic to olivine, melting to produce a highly depleted harzbugitic (low-cpx) mantle would involve the preferential removal of U+Th relative to He. This in turn would allow a relatively undisturbed primordial/radiogenic ³He/⁴He ratio to be retained in association with low He abundance. Thus, recycling of previously depleted mantle into the source region of “hot spots” provides one possible explanation for the paradox of high 3/4 He ratios previously thought to indicate an undegassed, primordial lower mantle reservoir, with low He abundance indicating a degassed source (2). Preliminary UVLAMP depth profiles for noble gas diffusion in mantle minerals confirm that although sub-solidus diffusive removal of He relative to other noble gases from a gas-rich mantle plum is theoretically possible, the short distances involved are unlikely to produce an effect that can be sustained though a hot spot melting event. The slow diffusion rates and lack of fractionation of noble gases in our partitioning experiments suggests that low He/Ar (and Ne/Ar) ratios observed at hot spots are most likely to be features inherited from the source, or subsequently imposed by some shallow level process. In our partitioning experiments, it proved surprisingly difficult to grow olivine crystals that are free of bubbles, even from volatile undersaturated melts. These bubbles nucleate on the crystal surface and can become included as the crystals grow. Although inclusions can be avoided using our micro-analytical technique, their bulk effect is to produce high crystal+bubble/melt D values and fractionation of light from heavy noble gases due to decreasing solubility in the melt for the latter. If such bubble capture occurs in nature as suggested by (3), the cumulative crystal rock would show enrichment in light noble gases and the residual melt will be depleted. This effect is similar to degassing but may be independent of saturation and depth of emplacement and decoupled from other volatile behaviour. The removed volatiles will also be retained in the cumulate rock rather than degassed. Differences in crystal growth processes and bubble trapping during MORB and IOB emplacement could contribute towards different He/Ar and Ne/Ar ratios. 1. R.A. Brooker et al., Nature 423, 738-741. 2003 2. D.L. Anderson, Proc. Natl. Acad. Sci. USA 95, 4822-4827. 1998 3. J.H. Natland, J. Pet. 44, 421-456. 2003.