What springs to mind when you think of a whale? Blubber, blowholes and flukes are among the hallmarks of the roughly 80 species of cetaceans (whales, dolphins and porpoises) alive today. But, because they are mammals, we know that they must have evolved from land-dwelling ancestors.

About 375 million years ago, the first tetrapods—vertebrates with arms and legs—pushed themselves out of the swamps and began to live on land. This major evolutionary transition set the stage for all subsequent groups of land-dwelling vertebrates, including a diverse lineage called synapsids, which originated about 306 million years ago. Though these creatures, such as Dimetrodon, looked like reptiles, they were actually the archaic precursors of mammals.

By the time the first mammals evolved 200 million years ago, however, dinosaurs were the dominant vertebrates. Mammals diversified in the shadow of the great archosaurs, and they remained fairly small and secretive until the non-avian dinosaurs were wiped out by a mass extinction 65 million years ago. This global catastrophe cleared the way for a major radiation of mammals. It was only about 10 million years after this extinction—and more than 250 million years since the earliest tetrapods crawled out onto land—that the first whales evolved. These earliest cetaceans were not like the whales we know today, and only recently have paleontologists been able to recognize them.

For more than a century, our knowledge of the whale fossil record was so sparse that no one could be certain what the ancestors of whales looked like. Now the tide has turned. In the space of just three decades, a flood of new fossils has filled in the gaps in our knowledge to turn the origin of whales into one of the best-documented examples of large-scale evolutionary change in the fossil record. These ancestral creatures were stranger than anyone ever expected. There was no straight-line march of terrestrial mammals leading up to fully aquatic whales, but an evolutionary riot of amphibious cetaceans that walked and swam along rivers, estuaries and the coasts of prehistoric Asia. As strange as modern whales are, their fossil predecessors were even stranger.

Pioneers who cleared land in Alabama and Arkansas frequently found enormous round bones. Some settlers used them as fireplace hearths; others propped up fences with the bones or used them as cornerstones; slaves used the bones as pillows. The bones were so numerous that in some fields they were destroyed because they interfered with cultivating the land.

In 1832, a hill collapsed on the Arkansas property of Judge H. Bry and exposed a long sequence of 28 of the circular bones. He thought they might be of scientific interest and sent a package to the American Philosophical Society in Philadelphia. No one quite knew what to make of them. Some of the sediment attached to the bone contained small shells that showed that the large creature had once lived in an ancient sea, but little more could be said with any certainty.

Bry’s donation was soon matched, and even exceeded, by that of Judge John Creagh from Alabama. He had found vertebrae and other fragments while blasting on his property and also sent off a few samples to the Philadelphia society. Richard Harlan reviewed the fossils, which were unlike any he had seen before. He asked for more bones, and Creagh soon sent parts of the skull, jaws, limbs, ribs, and backbone of the enigmatic creature. Given that both Creagh and Bry said they had seen intact vertebral columns in excess of 100 feet in length, the living creature must have been one of the largest vertebrates to have ever lived. But what kind of animal was it?

Harlan thought the bones were most similar to those of extinct marine reptiles such as the long-necked plesiosaurs and streamlined ichthyosaurs. He tentatively assigned it the name Basilosaurus. He wasn’t certain, though. The jaw contained teeth that differed in size and shape, a characteristic of mammals but not most reptiles. Why did the largest fossil reptile that ever lived have mammal-like teeth?

Harlan traveled to London in 1839 to present Basilosaurus to some of the leading paleontologists and anatomists of the day. Richard Owen, a rising star in the academic community, carefully scrutinized every bone, and he even received permission to slice into the teeth to study their microscopic structure. His attention to such tiny details ultimately settled the identification of the sea monster. Basilosaurus did share some traits with marine reptiles, but this was only a superficial case of convergence—of animals in the same habitat evolving similar traits—because both types of creature had lived in the sea. The overall constellation of traits, including double-rooted teeth, unquestionably identified Basilosaurus as a mammal.

A few years later, a scientist handling a different specimen with his colleagues pulled out a bone from the skull, dropped it, and it shattered on the floor. When the unnerved scientists gathered the fragments, they noticed that the bone now revealed the inner ear. There was only one other kind of creature with an inner ear that matched: a whale.

Not long after the true identity of Basilosaurus was resolved, Charles Darwin’s theory of evolution by means of natural selection raised questions about how whales evolved. The fossil record was so sparse that no definite determination could be made, but in a thought experiment included in On the Origin of Species, Darwin speculated about how natural selection might create a whale-like creature over time:

In North America the black bear was seen by [the explorer Samuel] Hearne swimming for hours with widely open mouth, thus catching, like a whale, insects in the water. Even in so extreme a case as this, if the supply of insects were constant, and if better adapted competitors did not already exist in the country, I can see no difficulty in a race of bears being rendered, by natural selection, more and more aquatic in their structure and habits, with larger and larger mouths, till a creature was produced as monstrous as a whale.

Darwin was widely ridiculed for this passage. Critics took it to mean he was proposing that bears were direct ancestors of whales. Darwin had done no such thing, but the jeering caused him to modify the passage in subsequent editions of the book. But while preparing the sixth edition, he decided to include a small note about Basilosaurus. Writing to his staunch advocate T.H. Huxley in 1871, Darwin asked whether the ancient whale might represent a transitional form. Huxley replied that there could be little doubt that Basilosaurus provided clues as to the ancestry of whales.

Huxley thought that Basilosaurus at least represented the type of animal that linked whales to their terrestrial ancestors. If this was true, then it seemed probable that whales had evolved from some sort of terrestrial carnivorous mammal. Another extinct whale called Squalodon, a fossil dolphin with a wicked smile full of triangular teeth, similarly hinted that whales had evolved from meat-eating ancestors. Like Basilosaurus, though, Squalodon was fully aquatic and provided few clues as to the specific stock from which whales arose. Together these fossil whales hung in a kind of scientific limbo, waiting for some future discovery to connect them with their land-dwelling ancestors.

In the meantime, scientists speculated about what the ancestors of whales might have been like. The anatomist William Henry Flower pointed out that seals and sea lions use their limbs to propel themselves through the water while whales lost their hind limbs and swam by oscillations of their tail. He could not imagine that early cetaceans used their limbs to swim and then switched to tail-only propulsion at some later point. The semi-aquatic otters and beavers, he claimed, were better alternative models for the earliest terrestrial ancestors of whales. If the early ancestors of whales had large, broad tails, that could explain why they evolved such a unique mode of swimming.

Contrary to Huxley’s carnivore hypothesis, Flower thought that ungulates, or hoofed mammals, shared some intriguing skeletal similarities with whales. The skull of Basilosaurus had more in common with ancient “pig-like Ungulates” than seals, thus giving the common name for the porpoise, “sea-hog,” a ring of truth. If ancient omnivorous ungulates could eventually be found, Flower reasoned, it would be likely that at least some would be good candidates for early whale ancestors. He envisioned a hypothetical cetacean ancestor easing itself into the shallows:

We may conclude by picturing to ourselves some primitive generalized, marsh-haunting animals with scanty covering of hair like the modern hippopotamus, but with broad, swimming tails and short limbs, omnivorous in their mode of feeding, probably combining water plants with mussels, worms, and freshwater crustaceans, gradually becoming more and more adapted to fill the void place ready for them on the aquatic side of the borderland on which they dwelt, and so by degree being modified into dolphin-like creatures inhabiting lakes and rivers, and ultimately finding their way into the ocean.

The fossil remains of such a creature remained elusive. By the turn of the 20th century the oldest fossil whales were still represented by Basilosaurus and similar forms like Dorudon and Protocetus, all of which were fully aquatic—there were no fossils to bridge the gap from land to sea. As E.D. Cope admitted in an 1890 review of whales: “The order Cetacea is one of those of whose origin we have no definite knowledge.” This state of affairs continued for decades.

While analyzing the relationships of ancient meat-eating mammals in 1966, however, the evolutionary biologist Leigh Van Valen was struck by the similarities between an extinct group of land-dwelling carnivores called mesonychids and the earliest known whales. Often called “wolves with hooves,” mesonychids were medium- to large-sized predators with long, toothy snouts and toes tipped with hooves rather than sharp claws. They were major predators in the Northern Hemisphere from shortly after the demise of the dinosaurs until about 30 million years ago, and the shape of their teeth resembled those of whales like Protocetus.

Van Valen hypothesized that some mesonychids may have been marsh dwellers, “mollusk eaters that caught an occasional fish, the broadened phalanges [finger and toe bones] aiding them on damp surfaces.” A population of mesonychids in a marshy habitat might have been enticed into the water by seafood. Once they had begun swimming for their supper, succeeding generations would become more and more aquatically adapted until something “as monstrous as a whale” evolved.

A startling discovery made in the arid sands of Pakistan announced by University of Michigan paleontologists Philip Gingerich and Donald Russell in 1981 finally delivered the transitional form scientists had been hoping for. In freshwater sediments dating to about 53 million years ago, the researchers recovered the fossils of an animal they called Pakicetus inachus. Little more than the back of the animal’s skull had been recovered, but it possessed a feature that unmistakably connected it to cetaceans.

Cetaceans, like many other mammals, have ear bones enclosed in a dome of bone on the underside of their skulls called the auditory bulla. Where whales differ is that the margin of the dome closest to the midline of the skull, called the involucrum, is extremely thick, dense, and highly mineralized. This condition is called pachyosteosclerosis, and whales are the only mammals known to have such a heavily thickened involucrum. The skull of Pakicetus exhibited just this condition.

Even better, two jaw fragments showed that the teeth of Pakicetus were very similar to those of mesonychids. It appeared that Van Valen had been right, and Pakicetus was just the sort of marsh-dwelling creature he had envisioned. The fact that it was found in freshwater deposits and did not have specializations of the inner ear for underwater hearing showed that it was still very early in the aquatic transition, and Gingerich and Russell thought of Pakicetus as “an amphibious intermediate stage in the transition of whales from land to sea,” though they added the caveat that “Postcranial remains [bones other than the skull] will provide the best test of this hypothesis.” The scientists had every reason to be cautious, but the fact that a transitional whale had been found was so stupendous that full-body reconstructions of Pakicetus appeared in books, magazines and on television. It was presented as a stumpy-legged, seal-like creature, an animal caught between worlds.

Throughout the 1990s, the skeletons of more or less aquatically adapted ancient whales, or archaeocetes, were discovered at a dizzying pace. With this new context, however, the stubby, seal-like form for Pakicetus depicted in so many places began to make less and less sense. Then, in 2001, J.G.M. Thewissen and colleagues described the long-sought skeleton (as opposed to just the skull) of Pakicetus attocki. It was a wolf-like animal, not the slick, seal-like animal that had originally been envisioned. Together with other recently discovered genera like Himalayacetus, Ambulocetus, Remingtonocetus, Kutchicetus, Rodhocetus and Maiacetus, it fits snugly within a collection of archaeocetes that exquisitely document an evolutionary radiation of early whales. Though not a series of direct ancestors and descendants, each genus represents a particular stage of whale evolution. Together they illustrate how the entire transition took place.

The earliest known archaeocetes were creatures like the 53-million-year-old Pakicetus and the slightly older Himalayacetus. They looked as if they would have been more at home on land than in the water, and they probably got around lakes and rivers by doing the doggie paddle. A million years later lived Ambulocetus, an early whale with a crocodile-like skull and large webbed feet. The long-snouted and otter-like remingtonocetids appeared next, including small forms like the 46-million-year-old Kutchicetus. These early whales lived throughout near-shore environments, from saltwater marshes to the shallow sea.

Living at about the same time as the remingtonocetids was another group of even more aquatically adapted whales, the protocetids. These forms, like Rodhocetus, were nearly entirely aquatic, and some later protocetids, like Protocetus and Georgiacetus, were almost certainly living their entire lives in the sea. This shift allowed the fully aquatic whales to expand their ranges to the shores of other continents and diversify, and the sleeker basilosaurids like Dorudon, Basilosaurus and Zygorhiza populated the warm seas of the late Eocene. These forms eventually died out, but not before giving rise to the early representatives of the two groups of whales alive today, the toothed whales and the baleen whales. The early representatives of these groups appeared about 33 million years ago and ultimately gave rise to forms as diverse as the Yangtze River dolphin and the gigantic blue whale.

Studies coming out of the field of molecular biology conflicted with the conclusion of the paleontologists that whales had evolved from mesonychids, however. When the genes and amino acid sequences of living whales were compared with those of other mammals, the results often showed that whales were most closely related to artiodactyls—even-toed ungulates like antelope, pigs, and deer. Even more surprising was that comparisons of these proteins used to determine evolutionary relationships often placed whales within the Artiodactyla as the closest living relatives to hippos.

This conflict between the paleontological and molecular hypotheses seemed intractable. Mesonychids could not be studied by molecular biologists because they were extinct, and no skeletal features had been found to conclusively link the archaeocetes to ancient artiodactyls. Which were more reliable, teeth or genes? But the conflict was not without hope of resolution. Many of the skeletons of the earliest archaeocetes were extremely fragmentary, and they were often missing the bones of the ankle and foot. One particular ankle bone, the astragalus, had the potential to settle the debate. In artiodactyls this bone has an immediately recognizable “double pulley” shape, a characteristic mesonychids did not share. If the astragalus of an early archaeocete could be found it would provide an important test for both hypotheses.

In 2001, archaeocetes possessing this bone were finally described, and the results were unmistakable. Archaeocetes had a “double-pulley” astragalus, confirming that cetaceans had evolved from artiodactyls. Mesonychids were not the ancestors of whales, and hippos are now known to be the closest living relatives to whales.

Recently scientists determined which group of prehistoric artiodactyls gave rise to whales. In 2007, Thewissen and other collaborators announced that Indohyus, a small deer-like mammal belonging to a group of extinct artiodactyls called raoellids, was the closest known relative to whales. While preparing the underside of the skull of Indohyus, a student in Thewissen’s lab broke off the section covering the inner ear. It was thick and highly mineralized, just like the bone in whale ears. Study of the rest of the skeleton also revealed that Indohyus had bones marked by a similar kind of thickening, an adaptation shared by mammals that spend a lot of time in the water. When the fossil data was combined with genetic data by Jonathan Geisler and Jennifer Theodor in 2009, a new whale family tree came to light. Raoellids like Indohyus were the closest relatives to whales, with hippos being the next closest relatives to both groups combined. At last, whales could be firmly rooted in the mammal evolutionary tree.

Adapted from Written in Stone: Evolution, the Fossil Record, and Our Place in Nature, by Brian Switek. Copyright 2010. With the permission of the publisher, Bellevue Literary Press.