Although the ice that makes up the upper parts of the Greenland and Antarctic ice sheets is annually layered, for times before about 70 ka the layering disappears because of plastic deformation. Earlier ages have to estimated from models of the deformation, and a second check is to match the data records from ice cores against those from sea floor sediments. Different processes contribute to those records: for instance, the marine record of oxygen isotopes in benthonic forams tracks the changing volume of ice locked on land, while the same record from ice cores depends on the air temperature above the ice cap. The correlation does seem to work, however. But not, it seems, for the very deepest ice recovered from beneath Antarctica (see Yet further back in the Antarctic ice in the December 2005 issue of EPN) which extends to around 800 ka.

French scientists involved in the EPICA Dome C ice-core project have cunningly discovered a means of checking on the otherwise undateable deep Antarctic ice (Raisbeck, G.M. et al. 2006. ¹⁰Be evidence for the Matuyama-Brunhes geomagnetic reversal in the EPICA Dome C ice core. Nature, v. 444, p. 82-84). The core penetrated to an estimated time that should include the most recent magnetic reversal, dated very precisely to 778±2 ka. Although the exact details of how the magnetic field behaved during this reversal, it is known that when its polarity flips the intensity of the field becomes very small. While the field is stable it is sufficiently strong to deflect charged particles, both in the Solar wind and in cosmic rays, so that less pass through the atmosphere. Cosmic rays are so energetic that they can perform isotopic transformations, one product being ¹⁰Be. So if the magnetic field decreased so the proportion of ¹⁰Be in the atmosphere would go up. Raisbeck and colleagues have examined the ¹⁰Be record in the EPICA core in great detail. In a 10 m thick section from a depth of almost 3.2 km the isotope rises to a peak, which they interpret as the signature of the reversal. If correct, this gives a ‘golden spike’ against which the depth to age conversion can be refined.

Balmy shores of the Precambrian

Before the appearance of fossil organisms that could give clues to past climates the only sources of information take the form of proxies. One of the best examples might seem to be the oxygen isotope composition of carbonate rocks that relate to sea-surface temperature. In fact it isn’t useful for the Precambrian because estimates of SST depend on being able to identify the shells of planktonic animals and use their d¹⁸O as a proxy. That is a pity, because limestones are common throughout the geological record and various aspects of their geochemistry have been used extensively as proxies for other crucial information, such as the relationship between their strontium isotope composition and the pace of continental weathering. Another palaeothermometer relies on the same temperature dependent fractionation of oxygen isotopes between seawater and the precipitation of dissolved silica to form cherts, whose d¹⁸O decreases with temperature. The trouble is that silica is notoriously prone to being remobilised and reprecipitated as pH changes in the fluids within sedimentary rocks. Some results from Precambrian cherts gave such low d¹⁸O that seawater temperature would have been tens of degrees higher than they were during the Phanerozoic, but they have been wisely suspected of having been affected by much later alteration by warmer fluids passing through cherty sequences. Now the approach has been given a boost by geochemists at the French National History Museum (Robert, F. & Chaussidon, M. 2006. A paleaotemperature curve for the Precambrian ocean based on silicon isotopes in cherts. Nature, v. 443, p. 969-972).

François Robert and Marc Chaussidon analysed the silicon isotopes in cherts for which oxygen isotope data are available. Since the two isotopic systems would both change, yet would behave differently during hydrothermal or metamorphic alteration, if the results correlate well both should be undisturbed. Except in samples that show the lowest d¹⁸O values (i.e. highest temperatures) there is a good correlation. That finding validates many of the O-isotope seawater temperatures, but Si isotopes fractionate during precipitation too, again in relation to temperature. So Robert and Chaussidon take Precambrian ocean temperature data to a new level with estimates based on two methods. Their results are fascinating: as well as confirming a decline from around 70°C 3.4 Ga ago to between 10 to 40°C in the Phanerozoic, the d³⁰Si data show sharp downward ‘spikes’ at about 2.5 Ga and 1.8 Ga. Between about 1.5 Ga to 600 Ma ocean temperature was steady at around 20°C, so there is no sign of continually cold oceans through the period of ‘Snowball Earth’ events – the number of samples cannot yet resolve the individual events, but the ‘Cryogenian’ is an obvious target for more work. The data are also important as they hint at all kinds of possible biological outcomes for such global warmth, and explanations are definitely needed.  Does the record suggest greater geothermal heating, or was it an outcome of the greenhouse effect? Will more details show periods of changing burial of organic carbon? Whatever, the Precambrian has become a stranger world to contemplate.

See also: de la Rocha, C.L. 2006. In hot water. Nature, v. 443, p. 920-921.