# What Died and When: Reading Earth’s History in Eras and Fossils

URL: https://sciencestruck.com/geology/earth-history-fossils
Category: Earth Science
Published: 2026-08-31T22:25:28
Updated: 2026-08-31T22:25:28
Image: https://sciencestruck.com/_astro/earth-history-fossils.B18FiHOU_ZdceHo.webp
Find the right roadcut in eastern Montana and you can put one hand across the end of the Mesozoic. Below your palm, sandstone with duckbill bone in it and pollen from a warm coastal forest. Across the middle, a band of grey clay about as thick as a fingernail, carrying iridium, an element that is scarce in Earth’s crust and abundant in asteroids. Above the clay, fern spores by the million, and from there upward, in that outcrop and in every outcrop on the planet, no non-avian dinosaur ever again.

That clay band is the boundary between the Cretaceous and the Paleogene, and it explains how the whole geologic time scale was assembled. Almost every line on the chart, at least across the 538 million years since animals began leaving good fossils, sits where a geologist found a spot in the rock column where the fossils change abruptly and marked it. The names came from the places where those changes were first traced out. The numerical ages came a century later, bolted on once radioactivity gave anyone a clock. Each line on the chart marks a place where the fossils changed, which makes the whole column one long strip chart of what lived, what stopped living, and roughly when.

Below is the whole 4.5 billion years at cruising altitude: the hierarchy, the two methods that put numbers on it, the long quiet Precambrian, the fossils that mark the lines, the three eras of complex life, the extinctions that reset the deck, and what the finished scale actually gets used for now.

## Why does the geologic time scale look like a layer cake?

Because in undisturbed sedimentary rock, that is genuinely how it stacks: mud, sand, and shell settle on top of what was already there, so each layer is younger than the one beneath it. Anyone who has ever seen the Grand Canyon in cross-section, or a highway cut through shale, has seen the raw data the scale was built from. The chart is a filing system for those layers.

The filing has four working levels, nested inside one another like drawers. **Eons** are the largest: Hadean, Archean, Proterozoic, and Phanerozoic, the last of which means “visible life” and covers everything since the Cambrian. Eons divide into **eras** (the Phanerozoic gets three: Paleozoic, Mesozoic, Cenozoic). Eras divide into **periods**, which are the names most people half-remember from school: Cambrian, Jurassic, Cretaceous. Periods divide into **epochs**, and epochs into ages, which is where working stratigraphers actually live. We are in the Phanerozoic Eon, [Cenozoic Era](https://sciencestruck.com/biology/cenozoic-era), Quaternary Period, Holocene Epoch, and the units get shorter as you go down because the record gets better as you get closer.

The names are a geography lesson in disguise. Cambrian comes from Cambria, the Roman name for Wales. Ordovician and Silurian are named for the Ordovices and Silures, two tribes Rome fought in that same corner of Britain, and the two men who named those periods spent years arguing about where one ended and the other began. Devonian is Devon. Permian is Perm, in Russia. Jurassic is the Jura mountains, and Cretaceous comes from _creta_, Latin for chalk. Every one of those names was assigned before anybody on Earth could tell you the age of a rock in years.

Somebody has to keep the chart consistent, and that somebody is the International Commission on Stratigraphy, the body within the International Union of Geological Sciences that ratifies boundaries and publishes the International Chronostratigraphic Chart. When the ICS defines a boundary, it does it physically: a specific bed, at a specific outcrop, with a bronze marker driven into the rock. The formal name is a Global Boundary Stratotype Section and Point. Everyone calls it a golden spike. The Devonian begins at a marked bed on a hillside in the Czech Republic, and every Devonian date anywhere on Earth is measured back to it.

The layer cake analogy takes you about that far and then quits, and it is worth knowing where. Real strata get tilted, folded, faulted, and in some mountain belts turned completely upside down. Erosion removes whole slabs of the sequence and leaves a gap called an unconformity, which is a surface representing time that used to be there. No single place on Earth has the full stack. The chart is a composite, stitched together from thousands of partial sections that overlap the way tree-ring records overlap.

## How do geologists actually read time in the rocks?

With two independent toolkits that answer two different questions. One gives you order, the other gives you years, and neither substitutes for the other.

The ordering half is older and, honestly, more elegant. In 1669 the Danish anatomist Nicolas Steno, who had been dissecting shark heads and noticed that the “tongue stones” people found in Malta were shark teeth, wrote down the rules that still govern field geology: layers were laid down flat, they extended sideways until something stopped them, and the ones on the bottom went down first. That last rule is superposition, and in the field it never fails: the layer underneath went down first, unless something later turned the sequence over. Add two corollaries and you can sequence almost any outcrop. A fault or an intrusion must be younger than everything it cuts through. A chunk of rock included inside another rock must be older than its host. Then William Smith, surveying canals in England in the 1790s, added the fossil rule: the same succession of fossil species turns up in the same order everywhere, so a fossil assemblage identifies a layer even hundreds of miles from where you last saw it.

None of that yields a single year. It yields sequence, which is enough to build the entire chart, and in fact the chart was finished in outline before anyone knew what an atom was.

The years came from radioactive decay. Certain isotopes fall apart at a rate that does not care about heat, pressure, or how the rock was treated, and each has a half-life, the time for half of a sample to convert. Uranium-238 decays through a long chain to lead-206 with a half-life near 4.47 billion years, conveniently the same scale as the planet, while uranium-235 runs its own chain to lead-207 at about 704 million years, which gives a single grain two clocks that have to agree. Potassium-40 to argon-40 covers the millions. Carbon-14, with [a half-life of about 5,730 years](https://gml.noaa.gov/education/isotopes/decay.html), runs out of usable signal somewhere around 50,000 years, which is why nobody carbon-dates a dinosaur. That misconception comes up in every classroom, and the answer is arithmetic: after 66 million years there would not be a single atom of the original carbon-14 left to count.

There is a practical wrinkle worth knowing, because it explains why volcanoes matter so much to paleontologists. Radiometric dating works on minerals that crystallized from a melt, which means igneous rock, not the sandstone your fossil is sitting in. So geologists date the volcanic ash beds sandwiched above and below the fossil layer and bracket it. Zircon is the favourite mineral for this: when it crystallizes it accepts uranium into its structure and rejects lead almost completely, so any lead inside a zircon grain arrived by decay. The rock doesn’t lie, but it does leave things out, and a good date always comes with an uncertainty attached because the measurement has one. The published age for the Cretaceous-Paleogene boundary is 66.043 million years, with a stated error of a few tens of thousands of years; the round 66 in a textbook is that same figure with its error bars filed off. The full comparison of the two approaches lives in [relative vs absolute dating](https://sciencestruck.com/geology/relative-vs-absolute-dating), which walks through both methods properly.

## What was happening in the Precambrian, Earth’s longest chapter?

Almost everything, and very slowly. The Precambrian is not a formal unit but an informal supereon covering the Hadean, Archean, and Proterozoic together, running from Earth’s formation to the base of the Cambrian at about 539 million years ago. That is roughly seven-eighths of the planet’s history in one lump, and it gets one-eighth of the space on most classroom charts, which tells you something about how much of the record survived.

The earliest stretch left almost no rock at all. The oldest intact rocks known are gneisses in northern Canada dated to around four billion years, and the oldest surviving pieces of Earth are older still: individual zircon grains from the Jack Hills of Western Australia, recycled into a much younger sandstone, that date to about 4.4 billion years and carry a chemical signature suggesting liquid water was already around. Everything before that is reconstructed from meteorites, which formed at the same time as the solar system and were never resurfaced.

Life shows up early and stays simple for an extraordinarily long time. The most visible evidence is stromatolites, domed and layered mounds built by mats of microbes that trapped sediment and grew upward toward the light, one paper-thin lamina at a time. The oldest widely accepted examples come from roughly 3.5-billion-year-old rocks in the Pilbara region of Western Australia, and living ones are still growing in Shark Bay a few hundred miles away, which is the closest thing science has to a time machine you can wade into. A closer look at what are stromatolites and how do [they form](https://sciencestruck.com/science-facts/what-are-stromatolites-how-do-they-form) covers the building process layer by layer.

The single largest event of the Precambrian is written in rust. Cyanobacteria in those mats made oxygen as a waste product, and for hundreds of millions of years the oceans absorbed it by oxidizing dissolved iron, which precipitated out as the banded iron formations that supply most of the world’s iron ore today. Once the iron ran out, oxygen accumulated in the atmosphere, at around 2.4 billion years ago in what is called the Great Oxidation Event. For the anaerobic life that dominated the planet, it was a poisoning. It also made complex cells, and eventually animals, possible.

The last stretch of the Proterozoic gets strange: global glaciations severe enough that ice may have reached the tropics, followed by the Ediacaran, whose fossils are soft, quilted, fronded things that mostly cannot be assigned to any living group. The Ediacaran is also the only Precambrian period with a golden spike; every older boundary is defined by an agreed round number of years instead, because the fossils are too sparse to draw lines with. Four billion years will not fit in one section, and [the precambrian era](https://sciencestruck.com/biology/the-precambrian-era) deserves the long version: the ice, the oxygen crisis, and those fronded Ediacaran things at proper length.

## How do fossils mark where one era ends and the next begins?

Through index fossils, which are the working tools of the whole system. An index fossil is a species that ticks four boxes: it was geographically widespread, it existed for a short slice of time, it was abundant, and it is easy to identify without a specialist. Hit all four and its presence in a layer pins that layer to a narrow window, anywhere on Earth.

Short-lived is the box people find counterintuitive. A species that survived 200 million years is useless for dating because it tells you almost nothing; a species that appeared and vanished inside a million years is a precision instrument. Free-floating and free-swimming organisms make the best indexes because ocean currents scattered them across every basin regardless of local habitat. Trilobites index the Paleozoic. Graptolites, colonial drifters that look like pencil marks on shale, are so good for the Ordovician and Silurian that the entire Ordovician is subdivided by graptolite zones. Ammonites do the same job for the Mesozoic. Conodonts, the tiny tooth-like elements of an eel-shaped animal nobody identified until a body fossil turned up in the 1980s, index a huge stretch of the column. Foraminifera, single-celled and shelled, do the Cenozoic and pay for themselves in the oil industry. There are working examples for every period, and a guide to what index fossils are, with examples, runs through the whole roster.

The golden spikes are defined by these species, usually by a first appearance rather than a last one, because an origination is a sharper and more reliable signal than a disappearance. The base of the Cambrian, and with it the whole Phanerozoic Eon, is defined at Fortune Head in Newfoundland by the first appearance of _Treptichnus pedum_, which is not a body at all but a burrow: a branching trace left by an animal that was probing sideways through sediment in a way nothing had done before. The line that opens the age of visible life is drawn on a hole in the mud.

Which raises the obvious question of how any of this survives. Bodies rot; the fossil record exists because a small percentage of them got buried fast enough to beat the decay. Minerals seep into bone and shell and fill the pore space, a process called permineralization. Shells dissolve and leave a mold that later fills to make a cast. Leaves and fish compress to a carbon film. Insects get caught in resin that hardens to amber. Footprints, burrows, and coprolites record behaviour rather than anatomy. The full range is worth a browse through [types of fossils](https://sciencestruck.com/geology/types-of-fossils), and the important consequence is bias: the record over-represents hard-shelled marine animals in shallow water with rapid burial, and badly under-represents soft-bodied life, small animals, mountain habitats, and anything that lived where sediment was eroding instead of accumulating. When a chart shows a group appearing at a particular date, the honest reading is that the group is first _preserved_ then, which is a claim about mud as much as about biology.

## What are the three great eras of complex life?

The Phanerozoic splits into three, and the boundaries between them are the two worst days in the history of animals. Paleozoic means ancient life, Mesozoic middle life, Cenozoic recent life, and the naming was done by people looking at how alien the fossils in each block looked compared with what lives now.

### Paleozoic, roughly 539 to 252 million years ago

It opens with the Cambrian Explosion, a stretch of perhaps twenty million years in which most major animal body plans alive today first show up in the rock, along with a fair number that did not last. The Burgess Shale in British Columbia preserves the soft parts of that fauna in fine detail, which is why we know Cambrian oceans held five-eyed swimmers and spiked worms and not just shells. From there the Paleozoic builds reefs, puts jaws on fish, gets plants onto land in the Silurian, grows the first forests in the Devonian, and then buries so much undecayed plant matter in Carboniferous swamps that we are still burning it. If you want one period at close range, an overview of the flora and fauna of the Ordovician gets down to the individual reef builders.

### Mesozoic, roughly 252 to 66 million years ago

The Mesozoic begins in a wrecked world and spends most of the Triassic rebuilding. Dinosaurs and mammals both appear in the Triassic, which surprises people who assume mammals arrived to fill the gap afterward; they coexisted with dinosaurs for about 150 million years, mostly at small body size. The Jurassic and Cretaceous are the ammonite’s world as much as the dinosaur’s, and the Cretaceous adds two things that changed the surface of the planet: flowering plants, and coccolithophores, microscopic algae whose calcite plates rained down in such volume that they built the White Cliffs of Dover.

### Cenozoic, 66 million years ago to now

Mammals and birds inherit the vacancies and move into them quickly. By about 50 million years ago the first whales are in the water, descended from hoofed land animals that walked back into the shallows, and their earliest relatives still have functioning legs. Grasses spread widely enough in the Neogene to create a whole new kind of landscape, and the grazers answer with high-crowned teeth built to survive a lifetime of chewing plants full of silica. The poles freeze, with ice building on Antarctica from around 34 million years ago, and the Quaternary then cycles in and out of ice ages on an orbital rhythm you can read in the oxygen isotopes of deep-sea foraminifera. We are 11,700 years into the current warm interval, and the last mammoths, on Wrangel Island in the Arctic Ocean, died about 4,000 years ago, which is well after the Egyptian pyramids were finished. There is a longer account of the cenozoic era for anyone who wants the whole 66 million years epoch by epoch.

Here is the whole Phanerozoic in order, with ages rounded to the nearest million years and one marker fossil apiece. Copy it onto a wall if it helps; a class of mine kept a hand-lettered version taped above the pencil sharpener for eleven years.

-   **Cambrian** (539 to 487): trilobites and small shelly fossils, the first hard parts in quantity.
-   **Ordovician** (487 to 443): graptolites, straight-shelled nautiloids, the first reef-building bryozoans and corals.
-   **Silurian** (443 to 419): sea scorpions, jawed fish, the first vascular land plants.
-   **Devonian** (419 to 359): armoured placoderms, ammonoids, the first forests and the first four-limbed vertebrates.
-   **Carboniferous** (359 to 299): lycopod coal swamps, giant arthropods, the first amniote eggs.
-   **Permian** (299 to 252): sail-backed synapsids, fusulinid foraminifera, the last trilobites.
-   **Triassic** (252 to 201): ceratitid ammonoids, the first dinosaurs, the first mammals.
-   **Jurassic** (201 to 145): ammonites in enormous variety, sauropods, _Archaeopteryx_.
-   **Cretaceous** (145 to 66): flowering plants, rudist bivalves, coccolith chalk.
-   **Paleogene** (66 to 23): nummulite foraminifera, early whales, mammals spreading into every vacated role.
-   **Neogene** (23 to 2.6): grasslands, three-toed horses, the first hominins.
-   **Quaternary** (2.6 to now): mammoths, dire wolves, us.

## What wiped the slate clean along the way?

Five events stand out badly enough in the record to get their own name, and four of them sit precisely on a boundary between named units, which is the point. The chart’s biggest divisions are where they are because the fossils on either side barely resemble each other.

The end-Ordovician event, around 445 million years ago, came with a heavy glaciation of the southern supercontinent and a sea-level drop that drained the shallow shelves where nearly everything lived. The Late Devonian arrived as a series of pulses spread across several million years, and they fell hardest on reef builders; reefs took tens of millions of years to rebuild to anything like their earlier scale.

The end-Permian, about 252 million years ago, is the worst thing that has happened to life on this planet. Published estimates of marine species loss generally run above eighty percent, and the land record is comparably grim; trilobites, which had made it through everything for nearly 300 million years, ended here. The favoured explanation is the Siberian Traps, a volcanic province that erupted a volume of basalt hard to picture and, worse, cooked its way through coal and evaporite deposits on the way up, loading the atmosphere with carbon dioxide and sulfur. The oceans warmed, acidified, and lost oxygen. Recovery took millions of years, and the rock immediately above the boundary in many places is a monotonous fossil-poor mudstone, which is what an empty ocean looks like from the outside.

The end-Triassic, near 201 million years ago, is associated with another giant volcanic province, this one along the rift that was opening the Atlantic, and it cleared the field of competitors for the dinosaurs.

Then the Cretaceous-Paleogene, 66 million years ago. In 1980 a team including the physicist Luis Alvarez and his geologist son Walter reported an unexpected concentration of iridium in the boundary clay, first from a section near Gubbio in Italy, and proposed an asteroid impact. The crater turned up later, buried under the Yucatán Peninsula and offshore sediment at Chicxulub, roughly 180 kilometres (110 miles) across. The boundary clay itself has since been found on six continents, often with shocked quartz and glass droplets in it, and above it in North America comes the fern spike, the pollen signature of a landscape recolonized by the plants that show up first after a burn.

Those five are the famous ones, and they are not the only ones; the record holds a longer list of smaller resets, and a rundown of the major and lesser [extinction events](https://sciencestruck.com/biology/major-lesser-extinction-events-in-earths-history) in Earth’s history covers the ones that missed the famous five. The pattern worth carrying away is that extinction concentrates the information. A slow steady turnover would give geologists no lines to draw. Catastrophes give the whole planet the same page break at the same instant, and that is what makes global correlation possible at all.

## What do the rock layers themselves tell us between the big events?

Quite a lot, and at a resolution fossils cannot touch. Where a fossil zone pins a layer to within a million years or so, a single sedimentary bed can record the few hours it took to settle out.

The clearest example is [graded bedding](https://sciencestruck.com/experiments/graded-bedding): a bed that is coarse at the bottom, gravel or coarse sand, and grades smoothly upward to silt and mud at the top. That gradation happens because a sediment-laden current slowed down and dropped its load in order of grain size, the heaviest first. Most graded beds in the deep-sea record are turbidites, laid down by an underwater avalanche that ran down a continental slope in a few hours and then settled out over the following days. A stack of a thousand of them is a stack of a thousand separate afternoons, separated by centuries of nothing much. The mechanics get their own treatment under graded bedding.

These structures do a second job that saves careers. In a mountain belt where the strata have been folded past vertical, a graded bed tells you which way was up when it formed, because the coarse end was always the bottom. So do mudcracks, which taper downward, and ripple marks, and the way certain shells settle. In overturned country, using superposition without checking a way-up indicator first will hand you the entire sequence backwards.

Other layers tick like clocks. Varves, the paired light and dark laminae in some glacial lake beds, are a summer and a winter each, and can be counted like tree rings. Storm beds, evaporite layers, ash falls, and rhythmic couplings driven by orbital cycles all give the same gift: a piece of the column with a known duration. Every boundary age on the chart is a stack: a radiometric date on an ash bed, tied to a fossil zone, tuned against orbital cycles, and cross-checked against the magnetic reversals recorded in the same rock. When four independent methods agree on a boundary, that is why the number is trusted, and where they disagree the ICS says so and the chart gets revised.

## Why bother learning a timeline this long?

Because people are paid to use it on ordinary working days. Every barrel of oil ever produced was found partly by biostratigraphy: a micropaleontologist looking at foraminifera in the mud coming up from a drill bit, identifying the zone, and telling the rig how deep it is in the section. Mineral exploration, groundwater mapping, and seismic hazard work all depend on being able to say that this formation here is the same age as that formation two hundred kilometres away.

Climate science leans on it harder every year. The only records of a planet with much higher atmospheric carbon dioxide are in the rock, and the closest thing to a natural experiment in fast carbon release is the Paleocene-Eocene Thermal Maximum, about 56 million years ago, when a large slug of carbon entered the ocean and atmosphere and the deep sea warmed by several degrees. Sorting out how fast that carbon went in, and how long the planet took to draw it back down, requires the time scale to be right to within tens of thousands of years. The same goes for extinction: the only way to say whether current rates are unusual is to have a background rate from the fossil record to compare them against.

The chart is also still an argument, which is the part I like best. The relevant ICS subcommission voted the proposal for a formal Anthropocene Epoch down in 2024. The stratigraphers were not satisfied that any single horizon could carry the boundary worldwide, and a unit the chart cannot pin to one bed in one outcrop is a unit the chart cannot use. That is the same fight Sedgwick and Murchison had about the Cambrian and Silurian in the 1830s, conducted with better instruments and worse manners.

You can start reading this at street level. Limestone facing on an old bank building is often packed with crinoid stems and shell fragments, and a hand lens turns a lunch break into a Paleozoic seafloor. Any roadcut with visible layering will show you superposition, and if you are lucky, a graded bed or a ripple mark that fixes which way was up. Stay off the loose stuff at the base of a fresh cut, since roadcuts shed rock without warning; look from the shoulder, wear eye protection if you are hammering anything, and get permission before working in a quarry. Then find the surface where two beds meet and ask what happened on that day, or in the ten million years that surface quietly deleted. That question, asked at one outcrop after another for two centuries, is the entire chart.