Snowball Earth


Artist's conception of the planet during the Cryogenian, clad in Snowball Earth glaciers and sea ice.
Our planet is thought to have been completely frozen over during the Neoproterozoic. From space, Earth would have looked like a big snowball.

The “Snowball Earth” glaciations were a series of ice ages during the Neoproterozoic era of geologic time, mainly confined to the Cryogenian period, but perhaps also into the Ediacaran period, too. These ice ages were thought to have been so profound that perhaps the entire surface of the planet froze over, all the way from the poles to the equator. In a 1992 paper, Caltech geophysicist Joe Kirschvink quipped that from a vantage point in outer space, the planet would have looked like a giant snowball. The evocative name stuck, and there has been an avalanche of scientific studies investigating the Snowball Earth glaciations ever since.

Depending on how you count, there were two episodes of Snowball Earth glaciation during the Neoproterozoic, or perhaps three.

A close-up look at the Neoproterozoic portion of the geologic time scale. The final era of the Proterozoic eon is the Neoproterozoic, and there are three periods in the Neoproterozoic: the Tonian (1000 Ma to ~720 Ma), the Cryogenian (~720 Ma to ~635 Ma), and the Ediacaran (~635 Ma to 541.0 +/- 1.0 Ma). There were two Snowballs during the Cryogenian, and one in the Ediacaran.
A close-up of the Neoproterozoic Era as depicted in the ICS timescale, showing the Cryogenian and Ediacaran periods.

They are called the Sturtian (about 720 to 660 Ma) and the Marinoan (about 645 to 640 Ma), with the Gaskiers (or Ediacaran, at about 580 Ma) being the possible third event in the sequence. There’s pretty good evidence that the Sturtian and Marinoan ice ages were true “Snowballs,” with glaciers flowing at sea level at the equator, but the Gaskiers event seems to have not been as extensive.

Did I get it?

Evidence of glaciation

Why do geologists think there were ice ages during the Neoproterozoic? These cold times in Earth history are times when glaciers grew, flowed, and moved sediment around. Though the ice itself melts (and thus cannot be preserved though geologic time), its impacts on the sedimentary record are distinctive and durable. Ancient glaciers left behind three kinds of signatures: (1) striations, (2) tillites, and (3) dropstones. These signatures are found in Neoproterozoic strata in many places: for the Sturtian, there are 39 localities documented, on six continents. For the Marinoan glaciation, there are 48 places where glaciogenic signatures have been found.


Striations are scratches or grooves that result when a glacier takes a particle of rock (like a cobble or boulder) and drags it against the bedrock, gouging out a scratch mark that shows the direction the glacier was flowing. These are subtle features in cross-section, and they show almost no volume, so the best way to see them is on the ancient bedrock surface over which the glacier flowed. But this is rarely exposed, so this is the least common kind of evidence for Snowball Earth.

Photograph of a striated glacial pavement in Norway, with overlying tillite.
Striated glacial pavement (unconformity surface) below Neoproterozoic Smalfjord tillite, western Norway. Photo by Galen Halverson; reproduced with permission.

Tillites & diamictites

Photograph showing a 1 m by 0.5 m outcrop of purple colored diamictite with pink clasts. A pencil provides a sense of scale.
Diamicitite bearing faceted cobbles and boulders of granite, near Konnarock, Virginia.

Glaciers produce deposits of exceptionally poorly sorted sediment: giant particles like boulders are dumped in the same spot as very fine particles like silt and clay. On the continents, this distinctive “boulder clay” is officially called “till.” Similar extremely poorly sorted deposits occur in the marine realm, where glaciers flow into the sea. To get a sense of this style of sorting, and a general sense of the diversity of rock types making up large clasts, as well as the shape of these clasts, we have compiled some representative imagery here.

Some of these examples are modern, and some are ancient. Regardless, they all share key characteristics worth noting.

Check out this gigapixel panorama of Pleistocene-aged till from Alberta, Canada:

You’ll notice there that (1) it’s really poorly sorted, (2) there’s a lot of small stuff —the “matrix” — which “supports” the overall deposit and that means that (3) the biggest bits — the “outsized clasts” — are pretty much isolated from one another, making them look like they are “floating” in the soup of the surrounding clay and sand.

Here’s a 3D model, showing a similar looking deposit from Ireland:

Photograph showing a large round (cobble-sized) clast within a poorly sorted diamictite, with a pencil for scale.
Gaskiers tillite, Newfoundland.

Again, this is Pleistocene rather than Neoproterozoic, but the darned-near-present is the key to the past. Note the large faceted clasts “floating” (i.e., not touching each other) in all orientations within the finer-grained matrix. When till is lithified (turned to rock through compression and cementation), we call it tillite. The Snowball Earth glaciations are recorded by tillite deposits.

Another rock name frequently used in describing poorly sorted sedimentary deposits is diamictite. Diamictites are poorly sorted, but the name is purely descriptive, and doesn’t necessarily imply a glacial origin for the deposit. For instance, a landslide could produce a very poorly sorted mix of sedimentary grains, a diamicton, which might then be lithified to make a diamictite. Both rock types contain large “out-sized” clasts (cobbles, boulders) is isolated from one another, surrounded by finer-grained, poorly sorted sediment. All tillites are diamictites, but not all diamictites are tillites.

  • Photograph showing a large (cobble-sized) clast within a poorly sorted diamictite, with a pencil for scale.
    A large (cobble-sized) clast pf porphyritic rhyolite within diamictite of the Gaskiers Formation, western Avalon Peninsula, Newfoundland.
A close-up photograph showing a scratched and faceted pebble, held in a person's fingers. The pebble has a series of flat faces, one of which shows a clear patter of intersecting gouges and scratches of varying depth.
A faceted shape bearing intersecting scratches is strong evidence of a clast having been ground down through glacial action.

One way to distinguish whether a diamictite had a glacial origin (that is, whether it is a tillite) is on the basis of the character of the larger clasts. Some of the boulders and cobbles in a true tillite will show faceted shapes, where they’ve have flat faces ground down by being dragged along over resistant surfaces. These facets will frequently show randomly-oriented scratches, too. These features together provide compelling evidence of the past grinding action of glacial ice.

Because scratched-up clasts are relatively uncommon, many geologists will be cautious and conservative when describing poorly sorted sedimentary deposits, referring to them by the broader and less specific “diamictite” even if they suspect the deposit is glacial in origin (glaciogenic; that is, a tillite or its marine equivalent).

Here are two hand samples of diamictite to examine. Which one is clearly glaciogenic? Which one is equivocal? Challenge yourself to explain the evidence you’re keying into to make the distinction.


Photograph showing two outsized granite clasts in metadiamictite. A lens cap provides a sense of scale.
Outsized clasts in foliated metadiamictite of the Fauquier Formation, Virginia Blue Ridge. Foliation is vertical in this photo.

Of course, like any rock, diamictites are subject to metamorphism if mountain-building occurs in their region after they form. Metamorphism transforms diamictites into metadiamictites, and tillites into metatillites. Metadiamictites show the original sedimentary poor sorting, but then also bear a foliated metamorphic overprint, giving them a palimpsest texture that indicates this more complicated history.

These are potentially-glacial deposits that were deposited in regions that later experienced mountain-building and recrystallization. If the regional metamorphism is intense, original primary sedimentary structures can be destroyed. If metamorphism is moderate, it can generate confusion, as metamorphic foliation wraps around outsized clasts in the same way that sedimentary laminations might in an unmetamorphosed dropstone-bearing deposit. Geologists working in metamorphic terranes have to be extra careful in making sedimentary interpretations of such rocks.

Compare the two formation mechanisms here:

Animated GIF showing the formation of a dropstone in layered deposit of sediments, producing truncated pre-dropstone strata on the bottom, and draped post-dropstone strata above.
Formation of a dropstone in layered sedimentary strata. Compare with the development of foliation that wraps around a metasedimentary porphyroclast.
Animated GIF showing the evolution of foliation around a porphyroclast; foliations starts off weak, but becomes more pronounced through compression, creating a pattern that wraps around the central chunk of resistant rock.
Formation of foliation wrapping around a porphyroclast in metasedimentary strata. The pattern resembles the relationship between a dropstone and laminated strata.

Here is a gallery of metadiamictites to peruse:

  • Photograph showing a large faceted pink granite clast in metadiamictite. A lens cap provides a sense of scale.
    Port Askaig tillite, Islay, Scotland.

Did I get it?


A cartoon conceptual diagram showing the characteristic relationships between a dropstone and the strata it drops into. Older layers are truncated (stop abruptly against the side of the dropstone), indicating they were pierced by the dropstone, while subsequent layers are draped on top of it.
Cartoon cross-section of a dropstone, in cross-section, showing key relationships to the surrounding sedimentary layers.

A third and final signature of glaciation in the Neoproterozoic era is the presence of dropstones in offshore marine sedimentary deposits. These are fine-grained, thin-bedded or laminated deepwater deposits of clay and silt, into which drop much larger particles, such as pebbles, cobbles, or boulders. The dropstones pierce into the soft pre-existing sedimentary layers, disrupting their continuity. We say that the pre-dropstone layers are truncated. Post-dropstone layers of fine sediment are draped atop them.

There’s more than one way to make dropstones today. After all, you could paddle a canoe out to deep water with a few cobbles onboard, then toss them in with a splash. Or a tree could grow through many years, its roots wrapping around cobbles in the soil. If the tree were to be blown over in a storm, it might be washed out to sea, root-wrapped cobbles and all. When, at sea, the wood finally rots, the cobble would be freed to drop into deep water sediments.

Photograph showing a ~3cm wide dropstone in 1- to 8-mm thick sedimentary layers of the Kingston Peak diamictite.
Dropstone in Kingston Peak diamictite, Sperry Wash, Death Valley region, California.

But there were neither trees nor canoes during the Neoproterozoic. The only plausible candidate for transporting diverse clasts offshore where they could be dropped into thin sedimentary layers was ice. The idea is that when outlet glaciers from the Snowball Earth ice sheets reached the sea, where they calved off icebergs full of gravel, sand, cobbles, and silt. These icebergs floated out to deep water, melted, and let their load of sediment fall through the water below.

Cartoon of larger-scale process here?  Simplified version of this:

In the Neoproterozoic, ice was the only way to get dropstones, and so the presence of Neoproterozoic dropstones = Neoproterozoic glacial ice.

It’s worth pointing out that there’s no size implication with the term “dropstone.” Dropstones can be huge, but they can also be small. Here’s a very small example: a mere 40 microns across, imaged with a scanning electron microscope:

Here is a gallery of dropstones to peruse:

  • A photograph showing a dropstone.
    An internally-laminated dropstone in Kingston Peak Formation, exposed in Sperry Wash, near southern Death Valley, California.

Did I get it?

Evidence of low-latitude of glacial deposits

We have now established that there is plenty of evidence of glacial activity preserved in the rock record from the Neoproterozoic. But it is also true that there is plenty of glacial activity on the modern Earth, and no one would mistake our modern planet for a giant snowball. So the question becomes: why do we think that the Snowball Earth glaciations were global freeze-overs, not just local glaciations (as in the modern world)? The answer comes from examining paleomagnetic signatures within the sedimentary rocks.


The study of ancient magnetic signatures within rocks is paleomagnetism. Geologists who pursue this important work are paleomagneticians, but they are a cheeky lot, and like to call themselves “paleomagicians.” They collect very-precisely-oriented samples of either volcanic or sedimentary rocks, and then place those samples within a machine that can measure the orientation of their faint remnant magnetic field very precisely. This yields several kinds of magnetic information, but for our purposes we will be focused on determinations of paleo-inclination, which in turn can speak to the paleo-latitude at which the sample formed.

How does this work?

An illustration of the shape of Earth's magnetic field as expressed at Earth's surface. The torus-like shape produces vertical lines of magnetic force at the poles, and horizontal lines of magnetic force at the equator.
The shape of Earth’s magnetic field is a torus. Lines of magnetic force are vertical at the poles, and horizontal at the equator, with intermediate values in between.

The Earth’s magnetic field is generated due to flow in its molten outer core, which is made of magnetically-conducting iron/nickel alloy. The magnetic force penetrates through the overlying mantle, crust, ocean, and atmosphere, and even extends out into space around the planet. The shape of that magnetic field is the relevant issue here. It is shaped like a very fat doughnut, with a dimple-like “hole” over the north magnetic pole and the south magnetic pole. Technically, this shape is called a “torus.”

As expressed at Earth’s surface, lines of magnetic force are vertical at the poles, and horizontal at the equator. In the diagram at right, consider the angular relationship between the “shaft” of the arrows and the surface of the Earth. At the equator, they are parallel, and at the poles you will note a perpendicular relationship. In between, the angles vary systematically. This angle (ranging between 0° at the magnetic equator and 90°at the magnetic poles) is called the angle of inclination. Because it varies by north-to-south position on the planet’s surface, it is used as a record of ancient latitude (paleo-latitude).

Lava flows and sedimentary deposits form at Earth’s surface as more or less horizontal sheets. Prior to consolidation (cooling and crystallization in lava flows, settling and lithification within clastic sedimentary strata), magnetically-susceptible grains such as magnetite are free to move, and they align themselves with the surrounding, permeating planetary magnetic field. Once lithified or crystallized, the grains have no more freedom of movement, and are locked in place as a durable record of the orientation of the Earth’s magnetic field at the moment that layer formed.

Equatorial sea-level glaciers imply high albedo

Photograph of a finely-bedded sandstone with tidal layers gently folded in the middle of the ~25 cm wide outcrop.
Gently folded tidal rhythmites of the Elatina Formation of central Australia, strata that show both evidence of glaciation and paleo-latitudes as low as 4 degrees from the equator. Photo by Galen Halverson; reproduced with permission.

A series of papers in the mid-1980s to late 1990s established measurements of paleomagnetic inclination (and thus paleo-latitude) for Marinoan-aged sea-level marine strata in several locations, particularly central Australia. These strata formed about 635 Ma and include signatures of glacial influence (such as dropstones), but they also show tidal rhythms preserved as varves. The idea behind studying these special strata is that (1) we know they are glacial, but (2) they also show fine layers that we can rely on as representing “horizontal” at the time they formed. Collecting carefully-oriented samples from such rocks allows for the precise measurement of their subtle magnetic signatures. Luckily, these samples have also been undisturbed since the time that they formed: their magnetism hasn’t been reset by a metamorphic event, for instance. The quality of the paleomagnetic measurements is robust, meaning that they are trustworthy and have passed several tests validating the results. And those results are: the sediments formed within 10° of the equator.

A map showing the position of Neoproterozoic Australia relative to the paleo-equator at about 635 Ma. 15 sampled sites range in paleo-latitude from about 5 degrees north of the paleo-equator up to about 17 degrees north.
Marinoan-aged glaciogenic sedimentary rocks in central Australia show near-equatorial paleolatitudes.

Now, this is a big deal, because the equator is usually pretty warm, and not a place where we would expect continental ice sheets to be flowing into ocean waters, shedding dropstone-laden icebergs.

Similar measurements of Sturtian (~710 Ma) sedimentary and volcanic rocks in ancestral North America show it too was positioned astride the equator during Neoproterozoic time.

This is important, because generally speaking the equator is a very warm part of the planet’s surface: it’s where the incoming rays of sunlight are most direct. As you move toward the poles from the equator, temperatures generally decrease. Therefore, if indeed there were glaciers flowing into the sea at such equatorial latitudes, it implies an exceptionally frigid climate for the planet as a whole.

Furthermore, because snow and ice are highly-reflective, such low-latitude ice would imply an overall very reflective planetary surface. In other words, an icy equator implies an icy planet, which would reflect away a lot of otherwise-warming incoming sunlight: it would have a high albedo. Among naturally occurring substances, snow and ice have the highest albedo (they are most reflective) and open ocean water has the lowest (it is most absorptive of incoming solar radiation).

Did I get it?

Cap carbonates

One surprising feature of Snowball Earth glaciogenic deposits is that many of these distinctive sedimentary rocks with their characteristic signatures of glaciation are comformably overlain by deposits of limestone and/or dolostone. Because the carbonate strata always appear on top of the glaciogenic strata, they are dubbed “cap carbonates.”

Photograph with annotations showing the Noonday dolomite cap carbonate comformably overlying the glaciogenic Kingston Peak diamictite.
Noonday dolomite cap carbonate comformably overlies glaciogenic Kingston Peak diamictite in Sperry Wash, near southern Death Valley.

The most astounding thing that cap carbonates tell us is that temperature conditions abruptly switched from the chill of glacial conditions to balmy tropical warmth. Shallow water limestones form best today in places like the Bahamas, or offshore of Florida. The cap carbonates are conformable with respect to the glaciogenic sediments on which they are deposited. They are also much more extensive, covering vastly more area than the glaciogenic sediments alone, suggesting they were deposited during a major transgressive event. Several lines of evidence point to the rate of cap carbonate deposition being very rapid. The comformable relationship between pre-Snowball carbonates (warm) to during-Snowball glaciogenic sediments (cold) to post-Snowball cap carbonates (warm again) implies extreme temperature oscillations during the Neoproterozoic. These strata stretch our conception of the limits of climate change on Earth.

Overall, the cap carbonates show a transgression (as glacial ice melted and was added as water to the ocean) and a massive chemical precipitation of carbon-containing rock (which has major implications for climate controls). These extensive deposits indicate the ocean was in a state of carbonate oversaturation despite what must have been extraordinary levels of CO2-induced ocean acidification in the wake of millions of years of volcanic outgassing with no carbon-removal processes operating on Earth’s surface (see discussion below for details).

The cap carbonate transgressive sequences for Sturtian and Marinoan glaciations are distinctive and consistent in deposits of that age around the world.

Unusual characteristics

Photograph showing skinny columnar stromatolites in cap carbonate. A pen provides a sense of scale: each stromatolite is about 2 cm in diameter, but half a meter tall (at least!).
Columnar “tubestone” stromatolites from Maieberg Formation, Namibia. Photo by Jay Kaufman, reproduced with permission.

Many cap carbonates show some primary sedimentary structures that suggest the post-Snowball hothouse Earth was a crazy place.

Not only was the temperature quite high, but apparently the cap carbonates were deposited very rapidly. This makes sense if we consider that the world’s surface consisted of a lot of pulverized rock interacting in shallow water with an atmosphere supercharged with carbon dioxide. As the CO2 dissolved into the seawater to combine with calcium ions (and some magnesium ions too), huge quantities of carbonate were precipitated and blanketed the seafloor. As the glaciers melted, a lot of fresh water was added to the oceans, and sea level rose globally, flooding the recently-deglaciated land surface.

Overall, the cap carbonates include two major subunits: an initial dolostone, representing shallow conditions, and a later limestone, representing a deeper depositional setting. Within this transgressive sequence are several distinctive primary sedimentary structures:

    1. tubestone stromatolites
    2. giant wave ripples
    3. aragonite crystal fans
    4. barite seafloor cements
Photograph showing skinny columnar stromatolites in cap carbonate. A rock hammer provides a sense of scale: each stromatolite is about 2-3 cm in diameter, but half a meter tall (at least!).
Columnar “tubestone” stromatolites from Sete Lagoas, Brazil. Photo by Jay Kaufman, reproduced with permission.

A distinctive primary sedimentary structure in the cap carbonates is the tubestone stromatolite. Stromatolites are fossilized microbial mats as we see in other carbonate rocks throughout geologic time. However, the cap carbonate stromatolites are often very thin, and grow in tight proximity to other another.

One interpretation is that these stromatolites have this distinctive form because they were growing in relatively shallow water (within the depths to which sunlight penetrates) and that the water was precipitating fine-grained carbonate rapidly, so that a form of natural selection takes over, and only those stromatolites that grow upward are those that don’t get buried survive to photosynthesize another day. The result is a dense collection of finger-like projections, with space in between, sometimes with short-lived microbial lamination bridging the gap over the fresh deposits of chemically-precipitated fine-grained carbonate. These finger-like stromatolites are sometimes organized into broad dome-like structures or mounds.

Above the tubestone stromatolites are accumulations of structures that resemble ripple marks, a signature bedform occurs-the giant wave ripple. These odd, large (1-6 m wavelength), symmetrical ripples grew upward without much sideways movement of the wave crest. They were initially thought to represent violent hurricane waves in the post-Snowball greenhouse climate, but this doesn’t explain why they grew straight upward. A preferred explanation is that they represent the record of a deeper wave base combined with the rapid cementation of the seafloor during cap carbonate deposition.

Photograph of cross-section through many layers of cap carbonate, showing small upward-branching pink crystal fans. Coin for scale.
Multiple layers of aragonite crystal fans in cap carbonate at Sete Lagoas, Brazil. Photo by Jay Kaufman, reproduced with permission.

Another interesting feature that suggests very rapid deposition of the carbonate is layers full of aragonite crystal fans. (Aragonite is a polymorph of calcite.) These large blade-like crystals of CaCO3 occur in layers within the overall cap carbonate sedimentary sequence. The long axis of the crystals is perpendicular to the ancient seafloor. Astonishingly, these crystals seem to imply such rapid crystallization of carbonate from the seawater that huge crystals several centimeters long grew upward into the supersaturated water column, and were buried as fast as they could form! It’s very difficult to imagine conditions stable enough for 10 cm tall crystals poking up all  over the seafloor, like a ‘bed of nails,’ and persisting for any length of time in violently oscillating ocean waves.

In Marinoan cap carbonates (but not Sturtian), barite seafloor cements appear exactly at the transition between the older, shallow-water dolostone deposits and the younger, deeper-water limestone deposits. In structure, these are similar to the aragonite crystal fans described above, but the mineral is different: it is a barium sulfate called barite, BaSO4. The inference is that this distinctive layer formed when sulfate from surface meltwaters and barium (derived from feldspar weathering?) in deeper ocean waters mixed for the first time. If this is right, then the barite seafloor cements mark passage of the chemocline between these two reservoirs as transgression proceeded.

Did I get it?

Return of the BIF

Photo of an outcrop of rock showing a dropstone of quartzite in banded iron formation.
A dropstone (of quartzite) in banded iron formation strata of the Rapitan Group, Northwestern Canada. Photo by Galen Halverson; reproduced with permission.

There is an additional chemical sedimentary rock that also appears in some Snowball Earth sedimentary sequences: banded iron formation (BIF). This is cited as evidence of radically altered oceanic chemistry during the Snowball glaciation, and therefore an indirect evidence of extensive sea ice cover.

As we saw in the Archean, the chemical precipitation of banded iron formation is a consequence of anoxic ocean waters. It is a broad generalization that BIFs are an “extinct rock” that “died out” at the Paleoproterozoic / Mesoproterozoic boundary with the build-up of copious free oxygen in the planet’s atmosphere and oceans. But that turns out to be not quite true.

During the Neoproterozoic Snowball Earth glaciations, BIFs make a reappearance in the geologic record. They were “resurrected” because the conditions that allow their formation occurred again, this time due to the Snowball.

A graph showing the relative abundance of banded iron formation through Earth history. There is a small bump in the early Archean (3.9 Ga), then a *big* bump that lasts through the Archean and Paleoproterozoic (3.5 Ga to 1.8 Ga), and then no more BIF until a third bump, also small, that coincides precisely with the timing of the Snowball Earth glaciations during the Neoproterozoic (0.7 to 0.55 Ga). Formation names and localities are labeled for reference.
A plot showing the relative abundance of banded iron formation (BIF) through geologic time. Note the small “BIF blip” that coincides with the timing of the Neoproterozoic Snowball Earth glaciations.

Because BIFs are the geologic signature of an anoxic ocean, their reappearance after a long absence is cited as strong evidence that sea ice over of Earth’s oceans was indeed global. That is, they were entirely or almost-entirely sealed off from interacting with the oxygenated atmosphere. Any dissolved oxygen in the oceans was consumed through chemical reactions, but was not replaced by diffusion from the atmosphere. This allowed deep sea black smokers to again permeate ocean water with dissolved iron.

Did I get it?

When were the Snowballs?

We have already mentioned that when most geoscientists talk about the “Snowball Earth,” they mean the glaciations that occured during the Cryogenian period of the Neoproterozoic era, specifically the Sturtian and the Marioan glaciations. The Sturtian occurred approximately 720 to 710 Ma, and the Marinoan around 660 to 635 Ma. Both the Sturtian and Marinoan glaciations show a global distribution of their sedimentary deposits. Their ages are not determined by fossils, but by relationships with rock units that are datable using isotopic methods (ash deposits, cross-cutting dikes, inclusions of datable rocks etc.).

Consider this hypothetical demonstration of the relevant information:

A cartoon cross-section of a glacial deposit (tillite/diamictite) and surrounding geologic units. The tillite includes a clast of granite, dated at 1.2 Ga. The tillite is above strata that include a volcanic ash layer dated at 800 Ma. The tillite is cut across by a granite dike, which has been dated at 500 Ma.
The age of this glacial deposit must be: (a) younger than the age of the clasts it contains, including the granite clast dated at 1.2 Ga, (b) younger than the volcanic ash layer stratigraphically below it (800 Ma), and (c) older than the granitic dike that cuts across it (which is 500 Ma). The inclusion date is redundant to the information provided by the ash in this case. We can constrain the age of the glacial deposit to be between 800 Ma and 500 Ma, based on the available information.

These techniques are nothing special for Snowball Earth only, but merely the application of general relative dating principles to the question of the timing of Neoproterozoic glaciation.

Some scientists also consider the slightly younger Gaskiers glacation (635 to 542 Ma, during the Ediacaran period of the Neoproterozoic) as a “Snowball Earth” glaciation. However, the paleomagnetic signatures on Gaskiers deposits are relatively high-latitude, not tropical or equatorial. There are also a lot fewer sites that host Gaskiers-aged strata, though it is recorded from at least eight small paleocontinental landmasses. (This is in contrast to dozens apiece for the Sturtian and Marinoan.) Considering the Gaskiers is younger than the Marinoan and Sturtian, and therefore less likely to have been destroyed by subsequent rock cycle processes, this implies it was truly not as extensive in distribution. Furthermore the Gaskiers glaciation has no known associated banded iron formations, suggesting that ocean water continued to exchange oxygen with the atmosphere during glaciation, and therefore no global sheath of sea ice existed then. Finally, dating of volcanic ash deposits to constrain the age of the Gaskiers glaciation suggest that the entire ice age lasted for only about 340,000 years, which is inconsistent with the multi-million year expectation based on the “Snowball Cycle.”

Photograph showing an outcrop of poorly sorted diamictite with a bunch of different rock types as large clasts, with a pencil for scale.
Gaskiers diamictite, Newfoundland.
Photgraph showing an outcrop of Gowganda Tillite, with a dark matrix and isolated clasts of various sizes of pink granite. A coin provides a sense of scale.
Gowganda tillite, Ontario. Photo by Ron Schott, reproduced with permission of his estate.

If we go back in time, to the Paleoproterozoic era, we find that there was another episode of glaciation recorded, though whether it was also a true “Snowball” (in the sense of ice reaching all the way to the equator) is a difficult question to answer. It is sometimes called the Huronian glaciation or Makganyene glaciation. It dates to between 2.4 and 2.1 Ga, and is best preserved in North America in Ontario’s Gowganda Formation, a striking jet-black tillite that formed about 2.3 Ga.

(Being positioned in southern Ontario meant that the Gowganda’s glacial deposits were ideally positioned for plucking by Pleistocene ice age glaciers. This means that in some very young glacial deposits, we can find pieces of very old glacial deposits!)

Did I get it?

Starting and stopping a Snowball

The Snowball Earth glaciations have important implications for our understanding of the role of greenhouse gases in regulating Earth’s climate. As you have learned, the planet’s surface temperature depends on several key variables, including the amount of energy the Sun emits, the planet’s position relative to the Sun, the ability of the planet to redistribute heat on its surface (mainly via ocean currents), and the ability of the planet’s atmosphere to retain heat that would otherwise be radiated out into space.

It is in this last category that greenhouse gases matter. They allow visible light from the Sun to pass through them unimpeded, but block infrared radiation – the kind that the planet’s rocks and oceans re-emit after basking in the sunlight for a while.

The most accepted model for the initiation of the Snowball Earth glaciations depends exquisitely on the level of the greenhouse gas carbon dioxide, CO2, in Earth’s atmosphere. The explanation goes like this: in the late Proterozoic, a supercontinent called Rodinia [LINK TO GRENVILLE OROGENY DISCUSSION] formed in Earth’s equatorial region. Because of its direct rays of incoming sunlight, the tropics are today home to (1) warm temperatures and (2) a humid climate with lots of rain. Lots of water vapor can be stored in warm air. Assuming that these same warm, wet conditions applied to the tropical belt during the Neoproterozoic, there would have been enhanced chemical weathering of the rocks in that belt. The reactions of chemical weathering consume CO2, such as the process by which feldspar (the most common mineral in Earth’s crust) is converted to clay (a widespread constituent of sedimentary rock). This means that as chemical weathering proceeded, CO2 levels in the atmosphere would have fallen. Typically, weathering-induced CO2 drawdown is balanced by volcanic CO2 emissions. But if the rate of tropical-weathering-induced CO2 drawdown was greater than the volcanoes could compensate for, atmospheric CO2 levels would have fallen. And because that means there would be less of a greenhouse effect, the temperature would have gotten colder.

The next stage of the story takes our attention away from the equator and focuses it instead on the polar regions, where incoming sunlight is much less direct. As a general rule, these are the coldest parts of our planet. As global CO2 was drawn down, ice would have begun to grow at the poles, forming floating sea ice that grew and grew through time, reaching into progressively lower and lower latitudes.

Sea ice has a high albedo: about 92% of incoming sunlight is reflected away. Very little energy is absorbed from the incoming sunlight. In contrast, ocean water has an extremely low albedo: only about 8% of incoming sunlight is reflected away, and the rest is absorbed as heat energy.

As sea ice built up in the polar regions, its high albedo reflected away more and more of the energy the planet was receiving from the Sun. This constantly-increasing albedo was an amplifying feedback system that took the CO2-induced cooling and made it even more pronounced. Eventually, when the ice reached about 33° of latitude, the albedo-induced cooling effect was unstoppable, and the remainder of the planet’s surface, including the equator, froze over in as little time as a decade.

The Snowball Earth had  begun. The average global temperature at this time is estimated to have been -50° C.

3-frame animation of a conceptual cross section of the Proterozoic carbon cycle. In the first frame, there is a balance between volcanic emission of CO2 and deposition of carbonate sediment, with tropical weathering (rain "washing CO2 out of the air") providing the intermediate step. Meanwhile, subduction continues to feed CO2 to the volcanoes. In frame 2, Lots of tropical weathering has resulted in a drawdown of CO2, and thus a diminished greenhouse effect. In frame 3, the planet has frozen over, and the carbon cycle is shut down, with the exception that plate tectonics (driven by Earth's internal heat) continues to drive subduction, so the volcanoes continue to emit CO2 into the air of the frozen world.
How the Neoproterozoic carbon cycle is thought to have changed to initiate a Snowball Earth glaciation (and sow the seeds for its eventual melt-off).

Sea ice sealed off gas exchange between the liquid ocean below and the atmosphere above, resulting in anoxic waters developing in much of the sea. It was at this time that banded iron formation was resurrected. Up above, during the unbelievable cold of the Snowball, much of the the carbon cycle would no longer be functional. Chemical weathering rates would be effectively nil. The planet was locked in an icy death grip, so reflective it could never warm up again… right?

A conceptual diagram showing changes in the Earth system that initiate and end a Snowball Earth glaciation: (1) carbon dioxide drawndown through tropical weathering allows ice to begin forming at the poles, (2) Sea ice reaches ~33 degrees of latitude, the tipping point for (3) the runaway ice albedo feedback, freezing the planet over, (4) five million years or so of "Snowball Earth" glaciation, during which CO2 from volcanoes again builds up in the atmosphere, (5) CO2 induced warming causes a rapid melt-off, leading to (6) a post-Snowball "greenhouse" Earth, during which time cap carbonate is deposited, and (7) as CO2 is again drawn down, temperatures again begin to fall and ice builds up anew at the poles.
The theoretical cycle of events that turned on and off the two “hard” Snowball Earth glaciations.

Wrong! (Obviously, since we’re here to talk about it.) You will recall that plate tectonics is not driven by energy from the Sun. Instead, it operates from Earth’s internal heat, driven in part by radioactive decay. So under the sea ice and under the glaciers, the plates continued their slow lateral movement. More to the point, subduction continued and seafloor spreading continued, and both of these processes resulted in volcanism. During the Snowball Earth, volcanoes continued to erupt, producing lava and ash but more importantly, emitting gases. These gases included water vapor (about 60% by volume) and carbon dioxide (about 15% by volume). The water vapor would have frozen, but the CO2 would not.

As the volcanoes continued to erupt, CO2 levels in the atmosphere began to build up again. With no weathering to extract the CO2 from the air, the amount of this greenhouse gas began to increase unchecked. Eventually, after about 5 million years or so, enough CO2 would have accrued that the resulting greenhouse heat retention would have overcome the albedo’s cooling effects. A few patches of sea ice would have melted –probably at the equator — and the exposed ocean water would have efficiently absorbed incoming solar energy. This melted more ice, which exposed more ocean water, and once again an amplifying feedback loop took over, rapidly melting off the sea ice (and glaciers) in short order.

The Snowball Earth was over; the post-glacial hothouse had begun. With an atmosphere chock full of CO2 and a low albedo allowing maximum absorption of incoming solar radiation, the average global temperature at this time is estimated to have been +50° C.

The CO2-charged atmosphere and pulverized glacial sediment (full of calcium ions) then equilibrated, reacting to produce a tremendous amount of carbonate sediment that rapidly precipitated in the oceans. As swollen, warm seas transgressed onto the continents, the cap carbonates were laid down, rapidly and perhaps violently. Gradually, CO2 was drawn down again.

Overall, this story is thought to have repeated at least twice (Sturtian and Marinoan), and perhaps with lesser intensity a third time (Gaskiers). Because of its repetition, some geoscientists have suggested it as a kind of cycle: the Snowball cycle.

Did I get it?

Problems with the traditional Snowball Earth model

The Snowball Earth is a controversial topic among some geologists. The evidence for widespread glaciation is incontrovertible, but whether the entire planet froze over entirely (a “hard” Snowball) is less agreed-upon. Perhaps it was cold, but not quite -50° C. After all, somehow life survived through the Snowball, and it is hard to imagine how that would work. Perhaps there were “refugia,” protected places or warmer-than average places: adjacent to deep-sea hydrothermal vents, perhaps, or polynyas that are maintained by currents flowing from beneath. So there are ways of getting around that objection.

A harder problem to surmount is the way that glaciers flow. In the modern world, where we observe glacial ice, it flows downhill, from its zone of accumulation toward its zone of ablation. We have ample evidence that there was glacial flow, glacial erosion, and glacial deposition. But in order for that all to happen, a functional hydrological cycle is probably a requirement. Glaciers only flow in the modern world because they get fresh ice forming at their upstream end, due to precipitation of snow. But that precipitation wouldn’t be possible without evaporation of water from somewhere else. If the planet were truly frozen over during the Snowball Earth glaciations, there should have been dramatically reduced potential for evaporation, not only because of the low temperatures (meaning low amounts of energy to drive evaporation), but more importantly because of the impact of the sea ice acting as a physical barrier: sealing off potential exchange between the ocean and the atmosphere. If the Earth experienced a true Snowball, then how did the glaciers keep flowing?

For this reason, a substantial number of geoscientists prefer a milder “Slushball Earth” model as their interpretation for Neoproterozoic glaciation.

Did I get it?

Impact on animal evolution

One of the most intriguing coincidences in the geological record is that “immediately” after the Snowball Earth glaciations, some sedimentary layers hold the very first evidence of multicellular animals preserved as fossil remains. These are the Ediacara, namesakes of the Ediacaran period.

A photograph showing a diversity of fossil Ediacara from Mistaken Point, Newfoundland, with a pencil for scale. The fossils are 10-30 cm long. Some are feather-like, some have round discs at one end, with stalks emerging from those discs. One huge one is a big branching feature. Another looks like a collapsed balloon or small pizza.
Diverse fossil Ediacara on a bedding plane at Mistaken Point, Newfoundland. Pencil for scale.

Exactly what the Ediacara were isn’t entirely clear. Animals, probably. What we know is this:

    • They were big enough to see with the naked eye, ranging between a couple of centimeters and a couple of meters in length.
    • They had morphologically distinct body parts, suggesting that they had differentiated tissues, if not organs.
    • They lacked hard parts such as bones, shells, horns or teeth. (The fossils are preserved as impressions of their soft bodies in shale and sandstone.)
    • Many appeared to be sessile (rooted in one spot), though others were probably mobile, on the basis of track-like trace fossils associated with some of the body fossils. Some resemble jellyfish, and so may have floated in a similar fashion.
    • Most showed bilateral symmetry, though there are others with apparent radial symmetry, and at least one has apparent trilateral (3-part!) symmetry.
A photograph of a pair of fossil Fractofusus, both looking like feathers or flattened spindles. They are 4-12 cm in length.
A pair of fossil Fractofusus, distinctive Ediacaran organisms, on a bedding plane at Mistaken Point, Newfoundland. Pencil for scale.

In Newfoundland, the end of the Gaskiers glaciation occurred at 579, and the oldest Ediacara fossils are about 5 million years younger. There are sparse examples for several million years, and then at Mistaken Point, a legendary Lagerstätte on the southeastern Avalon Peninsula, a 565 Ma ash bed smothered a huge number of Ediacara on a single bedding plane. A few meters above that, the same thing happened again. This unusual circumstance provides for exceptionally precise time constraints on the age of the organisms killed by the ash fall. Mistaken Point is the oldest known fossil assemblage of diverse, large, biologically complex fossilized animals.  All told, Ediacara lived for 33 million years prior to the Cambrian “explosion.” The fact that Ediacara evolved so soon (geologically speaking) after the end of the Snowball Earth suggests the great global catastrophe may somehow have nurtured the conditions for the evolution of these first animals.

But how? One obvious answer is by wiping out whatever came before, and “leveling the playing field” for up-and-coming animals with their new evolutionary innovations. Another straightforward variable was doubtless the increase in shallow sea habitat as glacial melting caused sea level rise, flooding the continents with shallow seas, perhaps an ideal habitat.

More esoteric explanations hinge on an increase in oxygen levels after the Snowball, and how those oxygen levels then provided sufficient fuel for oxygen-consuming animals to exists for the first time. Or perhaps because collagen is an important binding compound that holds animal bodies together, and collagen requires plenty of oxygen to form, that may be the most important aspect of the rise in oxygen. Exactly why oxygen rose isn’t entirely clear, though.

Obviously, being an animal comes with some big advantages, as large bodies with sensing tissues distributed throughout are able to sense gradients well (hot/cold, salty/fresh, toward a greater concentration of food molecules/away), and their large size (relative to microbes and single-celled protists) also potentially helps buffer against short-term hostile variations in environmental conditions. These are both innovations that natural selection could emphasize through time.

Did I get it?


The evidence for widespread glaciation during the Neoproterozoic’s Cryogenian and Ediacaran periods is indisputable. Several packages of glacial sediments (tillites, marine units bearing dropstones, etc.) have overlying “cap” carbonates, interpreted as warm-climate limestones laid down during post-glacial transgressions. One model to explain these glacial deposits is the “Snowball Earth,” which suggests the glaciation was triggered by (1) CO2 drawdown due to intense tropical weathering of the crust that then (2) led to a runaway ice-albedo feedback loop. The planet escaped the Snowball due to (3) volcanic outgassing building CO2 levels up anew, and then in the aftermath of the glaciers melting, CO2 dissolved into seawater, combined with Ca and Mg ions, and (4) laid down thick packages of carbonate rock. The Sturtian and Marinoan glaciations both show banded iron formations associated with their glacial deposits, suggesting widespread (if not global) sea ice cover triggering ocean anoxia. A third glaciation, the Gaskiers, is younger, shorter, and less widespread than the other two, but shortly after it ended, the first fossil animals (called Ediacara) show up in the geologic record.

Further reading

Bentley, Callan. (2010) “The Konnarock Formation,” Mountain Beltway,

Butler, R.F. (2004) Paleomagnetism: Magnetic Domains to Geologic Terranes. Originally published by Blackwell in 1984, 248 pp. Updated online 2004.

Kirschvink, Joseph L. (1992) “Late Proterozoic Low-Latitude Global Glaciation: the Snowball Earth,” in: The Proterozoic biosphere : a multidisciplinary study. Cambridge University Press, New York, pp. 51-52.

Klein, Cornelius. (2005) “Some Precambrian banded iron-formations (BIFs) from around the world: Their age, geologic setting, mineralogy, metamorphism, geochemistry, and origins,” American Mineralogist 90 (10): 1473–1499.

Harvard Smithsonian Center for Astrophysics in association with the Harvard University Center for the Environment. (2007) Many Planets, One Earth; video. ISBN: 1-57680-883-1

Hoffman, Paul, Alan Jay Kaufman, Galen Halverson, & Daniel Schrag. (1998) “A Neoproterozoic Snowball Earth,” Science 281. 1342-1346.

Pu, Judy P., Samuel A. Bowring, Jahandar Ramezani, Paul Myrow, Timothy D. Raub, Ed Landing, Andrea Mills, Eben Hodgin, & Francis A. Macdonald. (2016) “Dodging snowballs: Geochronology of the Gaskiers glaciation and the first appearance of the Ediacaran biota,” Geology 44 (11): p. 955–958.

Sohl, Linda, Nicholas Christie-Blick, & Dennis Kent. (1999) “Paleomagnetic polarity reversals in Marinoan (ca. 600 Ma) glacial deposits of Australia: Implications for the duration of low-latitude glaciation in Neoproterozoic time,” GSA Bulletin 111 (8); August 1999; p. 1120–1139.

Walker, Gabrielle. (2004) Snowball Earth. Bloomsbury; 288 pages.

A bibliography of Snowball Earth papers compiled on the website