Geoscientists take it for granted that the Earth has a certain age (currently estimated at 4.54 Ga), but it is one divined from indirect evidence, lead isotopes in meteorites and ancient ores of lead derived from uranium. If ever geoscientists are to grasp the nature of the early planet the evidence would be geochemical, yet also second-hand because relics must lie somewhere in the mantle as the crust is constantly being changed. For decades it has been known that the mantle shows geochemical heterogeneity as a result of episodes of partial melting from which the oceanic and continental crust emerged. Even with such an ancient origin it seems intuitively likely that there should be some mantle that has not been interfered with. Now a group of geochemists from the US and Britain have presented evidence for just such ur-mantle (Jackson, M.G et al. 2010. Evidence for the survival of the oldest terrestrial mantle reservoir. Nature, v. 466, p. 853-856). Their data come from Cenozoic lavas collected on Baffin Island and in West Greenland, which gave an earlier clue for having melted from a truly antique source: they contain the highest ratio of helium produced in the Big Bang (³He) to that released by radioactive decay (⁴He). Repeated melting of the mantle gradually drives off, yet radioactive decay continually replenishes its complement of ⁴He, so the more reworked a mantle source for lavas is the lower its ³He/ ⁴He ratio. This notion is backed up by the lead and neodymium isotopes in the Baffin Island and West Greenland lavas and they suggest an age of formation of the mantle source between 4.45-4.55 Ga.

Convection over billions of years ensures a degree of mixing in the mantle, but such is the viscosity of the Earth that there is a good chance that some areas have remained unchanged, the more so if the bulk of magmatism involving deep mantle has been linked to narrow rising plumes. But what emerges from the rest of the geochemistry of these lavas? Provided they have not been contaminated by continental crust through which they have passed, it should be possible using models for the way different elements are contributed to or withheld from magma by mantle minerals to estimate the source mantle’s overall composition. The team did this, bearing in mind the uncertainties. Plotted relative to a ‘guestimate’ of the original bulk Earth based on the geochemistry of chondritic meteorites they sshoww a very good fit for those elements that are likely to be retained by mantle minerals during partial melting: the so-called ‘compatible’ elements. But the estimated source for the lavas seems to have been depleted in the ‘incompatible’ elements that are highly likely to enter magma as soon as partial melting starts. This pattern would be expected if the early mantle had undergone some kind of differentiation as a whole, and that would be a consequence of the entire mantle having been molten and then crystallising: some low-density minerals could preferentially have taken in incompatible elements and floated upwards to deplete those elements in the deep mantle. That is compatible with the idea of Moon formation as a result of a collision between the proto-Earth and a Mars-sized planet, which could have released sufficient energy in the form of heat tp completely melt the outermost Earth.

So the data reveal a great deal, especially that this ancient mantle may well have been the parent for all later mantle compositions as the Earth evolved by dominantly igneous processes. But they do not resolve the perennial debate as to whether the Earth accreted from a uniform mix of nebular material of which meteorites are relics, roughly the composition of chondrites, or heterogeneously from different materials that had condensed from incandescent vapour at different nebular temperatures at different times. Moon formation would have mixed up the latter efficiently in a mantle-wide magma ocean, so we may never know. However some of the oldest meteorites contain fragments of condensates that did form at different temperatures.