What is the evidence against the snowball earths?
A number of objections have been raised against the snowball earth hypothesis, many of which have been more or less successfully rebutted.
  1. The global deposits are probably diachronous (not the same age everywhere) and are no more extensive than Phanerozoic (Cambrian to Recent, or 542-0 Ma) glaciations collectively.

    Rebuttal: There is mounting paleomagnetic evidence that ice-sheets reached the ocean even in the tropics during each of the three proposed snowball earths, and mounting chemostratigraphic evidence that the glaciations were broadly synchronous.

  2. The Gaskiers glaciation (circa 580 Ma) in eastern Newfoundland (Canada) has been shown from precise U-Pb zircon dating of multiple volcanic ash layers to have lasted less than 1.0 million years (Bowring et al., 2003). Short-lived glaciation contradicts the "hard" snowball earth hypothesis, which predicts they last for millions of years. This result, combined with the limited distribution of broadly contemporaneous glacial deposits, the lack of reliable paleomagnetic evidence for low-latitude glaciation, the absence of banded iron formation, and the poorly-developed or absent "cap" carbonate, makes it doubtful that the Gaskiers Formation represents a snowball earth.

    Proviso: The predicted duration of a snowball earth with thin tropical sea ice (Pollard & Kasting, 2005) is only 8% the length of the original "hard" snowball formulation (Caldeira & Kasting, 1992).

  3. The sedimentary character and great thickness (1000's of m) of Sturtian and Marinoan glacial deposits locally point to the existence of fast-flowing wet-base glaciers. These should be absent if the ocean was totally ice covered and the equatorial climate was as cold and dry as present Antarctica.

    Rebuttal: Up to 90% of the total ice drainage from Antarctica is routed through narrow corridors of fast-flowing wet-base ice known as ice streams, and thick piles of sedimentary debris (till) at present and former ice-stream mouths attest to their potency as agents of erosion and transport. Evidence for a 635-Ma (Marinoan) ice stream has recently been found in northern Namibia.

  4. Indicators of open water—for example wave ripples, far-travelled ice-rafted debris, and biomarkers of phototrophism—are found within Sturtian and Marinoan glacial strata.

    Rebuttal: A potential weakness in this argument is that most glacial deposits left after an ice age formed while the ice was in its final retreat. There would of course be open water at that time irrespective of the maximum ice extent.

  5. Polygonal sand wedges 2-3 m deep are found at a Marinoan-age permafrost holrizon near the paloeo-equator in South Australia. Deep sand wedges form in present permafrost by thermal expansion-contraction under strong seasonal forcing, which should not exist on the equator unless the Earth's rotation axis was highly oblique. This would make the polar climate warmer than the equator, and low-latitude glaciation would not require a snowball earth.

    Rebuttal: Paleomagnetic evidence for low-latitude carbonate deposition conflicts with the high-obliquity hypothesis because carbonates form preferentially in the warmer parts of the ocean because their solubility decreases with temperature (and pressure). Mechanical modeling suggests that deep sand wedges could form under diurnal (day-night) forcing if the ground was permanently frozen and very cold, as would be the case in the ablative zone (ice-free area) of a snowball earth.

  6. If continental weathering was reduced for millions of years while hydrothermal exchange at mid-ocean ridges continued unabated, the strontium isotopic composition of seawater should change (i.e., become less radiogenic, lower 87Sr/86Sr ratio). In fact, no significant change is observed between carbonates deposited before and after the Sturtian or Marinoan glaciations.

    Rebuttal: The lowering of seawater pH that would accompany the buildup of CO2 would cause dissolution of any sea-floor carbonate not covered by ice. Geochemical modeling suggests that the flux of isotopically-conservative strontium from carbonate dissolution over the course of a snowball glaciation could reduce the magnitude of the isotopic shift to within the resolution of the data. Osmium isotopes may be better suited than strontium for this important test of the snowball hypothesis.

  7. If CO2 built up to high levels as predicted in the snowball earth hypothesis, seawater pH would drop and CaCO3 would become severely undersaturated. Even if saturation was maintained through carbonate dissolution under the ice, the abiotic precipitation of post-glacial cap carbonates required 5-20 times oversaturation (dependent on nucleation kinetics and crystallization inhibitors) implying that "cap" carbonates should not form during the post-glacial sea-level rise (transgression) accompanying ice meltdown.

    Rebuttal: These are serious criticisms that may relate to why Sturtian cap carbonates, unlike Marinoan ones, formed only after the post-glacial transgression.

  8. Given that mid-ocean ridge hydrothermal exchange acts as a source of Ca2+ and a sink for Mg2+ with respect to seawater, the Ca/Mg ratio of the latter should be very high at the end of a snowball earth. This favors cap carbonates composed of calcite (CaCO3) as opposed to dolomite (Ca0.5Mg0.5CO3). However, Marinoan cap carbonates almost invariably begin with dolomite ("cap dolostone" sensu stricto).

    Rebuttal: Evidence suggests that cap dolostones formed from surface waters dominated by meltwater, not from deepwaters affected by hydrothermal exchange. However, Mg-rich silicate rocks (e.g., mafic and ultramafic volcanics) cannot be weathered fast enough to supply Mg on a time-scale of a few thousand years as estimated for glacial meltdown. Dolomite weathering is faster but can only supply cations with a Ca/Mg ratio of 1.0. The present ocean does not precipitate dolomite despite a Ca/Mg ratio of 0.2, and dolomite forms in the laboratory from waters with Ca/Mg ratio of 1.0 only at temperatures >75ÜC (137ÜF). The dolomite in cap dolostones could be secondary but it would be difficult to explain the stratigraphically precise change from dolomite to calcite at the top of a single 5-10 cm thick layer of sea-floor barite mapped for over 150 km along strike in the Mackenzie Mountains of NW Canada.

  9. No mass extinctions of palynoflora are observed coincident with the Sturtian and Marinoan glaciations in central Australia, contrary to expectation that snowball earths and their greenhouse aftermaths would constitute significant evolutionary bottlenecks.

    Rebuttal: The Cryogenian fossil record is extraordinarily impoverished, consisting of a few leiospheres (simple thin-walled organic microspheres) and local testate amoebae.
  10. Multiple geomagnetic polarity reversals are found in Marinoan cap dolostones in different areas and this observation conflicts with the time-scale for post-glacial transgression (i.e., meltdown of grounded ice sheets) of thousands of years. The reversals imply that sedimentation rates for cap dolostones were <4 cm kyr-1 (assuming 100 kyr reversal frequency). This is >50 times slower than the sedimentation rate implied if cap dolostones were laid down in ~2000 years, the estimated time-scale for glacial meltdown. Reproduceability of the reversals at three sites in South Australia suggests that they are not a product of secondary remagnetization.

    Rebuttal: The slow sedimentation rate implied by reversals of 100 kyr or greater frequency contradicts the low terrigenous content (typically <2% insolubles) of cap dolostones, an array of primary structures implying high sedimentation rates, and smoothly varying isotopic profiles with little scatter compared with other stratigraphic units in host successions.