Palaeomagnetic and geological data from Neoproterozoic glacial deposits indicate that glaciations were long-lived (lasting for millions of years)2,3 and locally associated with iron formations. The glacial deposits are covered by extraordinary sequences of carbonate sediments called 'cap' carbonates, which have unusual textures and low δ13C values. The snowball Earth hypothesis can explain all these observations3, whereas a semifrozen (slushball) Earth does not.

In the snowball Earth hypothesis, a runaway ice–albedo feedback leads to a planet frozen to the Equator4. Extremely large increases in carbon dioxide are required to terminate the glaciation and overcome the climate stability imposed by the high planetary albedo5. In Hyde et al.'s model1, with all the continents covered in ice, volcanic emissions without chemical weathering would cause atmospheric CO2 levels to rise. But with ice-free tropical oceans, even a modest rise in CO2 would cause the tropical glaciation to be short-lived. The exact duration would depend on the extent of chemical equilibration between sea water and calcium carbonate in the deep ocean, but it would be far less than the roughly 10 million years estimated from analysis of basin subsidence3. It is also unclear how iron could accumulate in sea water to produce banded iron formations if tropical oceans were ice-free. Moreover, the model of Hyde et al. predicts a progressive deglaciation from the Equator to the poles occurring at much lower CO2 concentrations, no higher than during the Cretaceous period.

Support for extreme increases in CO2 in the aftermath of the glaciations comes from cap carbonates, which are nearly ubiquitous features of Neoproterozoic glacial deposits. These carbonate rocks have distinctive features, including “knife-sharp” contacts with glacial deposits and unusual textures indicative of rapid precipitation on the sea floor6. The cap carbonates, which have been singled out as a climate paradox of Neoproterozoic geology7, are predicted by the snowball Earth hypothesis to be a consequence of intense carbonate and silicate weathering in the aftermath of the deglaciation. It is hard to reconcile the global occurrence of such a pulse of intense carbonate precipitation immediately after the termination of the ice ages with the progressive retreat of ice at moderate CO2 levels predicted by the Hyde et al. model.

Excitement over whether a semifrozen Earth might explain the geological observations stems from concern for the survival of eukaryotic life in such extreme and extended glaciations8. The critical feature is the survival of groups of photosynthetic algae that evolved before the glaciations. The survival of metazoans, as discussed by Hyde et al., is less problematic because such organisms (if they existed) could live wherever primary producers (photosynthetic or chemosynthetic) were still active. Photosynthetic algae could survive a series of glaciations in refugia near volcanic islands, such as Iceland and Hawaii, or beneath thin equatorial ice cover9. Evolution might well be stimulated by this prolonged genetic isolation, and by perturbations of biogeochemical cycles during the postglacial, ultra-greenhouse climate. This is consistent not merely with the survival of eukaryotic life, but also with the coincident radiation of metazoa and other groups8.