A New Story for Stonehenge

In the summer of 2019, I walked up a hillside in west Wales with Richard Bevins, a researcher who specializes in petrology, the study of the origins, compositions, and structures of rocks. We climbed through a landscape tufted with reeds. A drystone wall ran on our right. The territory opened into grassland, reaching up to a ridgeline etched against the sky. Our destination, Carn Goedog, jutted out above us. In Welsh, “carn” refers to a mound of rocks; “goedog” means full of trees. At close quarters, we could see that the rocks of Carn Goedog were dappled with lichen. I looked in vain for the feature that gave them their geologic name—spotted dolerite. “To see the spots, you have to get inside,” Bevins said.

Bevins and his collaborators believe that Carn Goedog is the source for a slew of the “bluestones” used to build the prehistoric monument Stonehenge, in southern England. Picture Stonehenge and you may first envision its huge standing stones, or sarsens. The bluestones are smaller; the biggest is a little less than ten feet tall, and weighs more than three tons. Around three dozen bluestones stand among the sarsens—there are different ways of counting—with more buried underground.

The rocks used for monuments like Stonehenge typically come from no more than a dozen miles away. Most of the sarsens have been traced to a part of the Marlborough Downs, fifteen miles from Stonehenge. But the distance between Stonehenge and Carn Goedog is a hundred and forty miles—roughly three and a half hours by car. If Bevins and his colleagues are right, then the bluestones had been transported much farther than any other material used for stone circles in Europe.

The mystery of the bluestones dates back to at least the seventeen-twenties. The antiquarian William Stukeley, an early authority on Stonehenge, wrote that, compared with the sarsens, the bluestones were “of a different sort,” and harder to place. The “provenancing” of the bluestones—the identification of their geographic source—began in earnest in 1923, when the geologist Herbert Henry Thomas posited that they had come originally from west Wales. Thomas had examined the light shown through thin slices of bluestone under a microscope and analyzed them alongside samples of rocks gathered in the field; he identified a number of outcroppings in the Preseli Hills, an area of rolling uplands close to the coast, as likely sources. In particular, Carn Alw, which is just under three-quarters of a mile from Carn Goedog, struck him as a particularly promising location. Until relatively recently, Thomas’s theory was regarded as canonical.

The revision began with a kind of social coincidence. In 2008, Bevins received an e-mail from a retired geologist named Rob Ixer, with whom he’d once provenanced a collection of axe-heads. Ixer was convinced that he’d stumbled upon a more accurate way to provenance the bluestones. Earlier that year, two archeologists, Tim Darvill and Geoff Wainwright, had conducted the first excavations inside Stonehenge in more than forty years, and asked Ixer to examine some of the thousands of stone fragments brought up to the surface. Around that time, Ixer happened to be speaking with another archeologist, Mike Parker Pearson, who’d mentioned that he had a shoebox full of similar bluestone fragments, which had been gathered in 1947. “He told me he had the shoebox under his desk and at his feet,” Ixer recalled. Parker Pearson sent the box—which had been lent to him by the Salisbury Museum, and essentially forgotten following a few studies a half century earlier—to Ixer. Inside were twelve fragments, the largest about four or five inches across. One was thought to be from a stone tool, made of sarsen rock; the others were bluestones that could not be traced to individual parts of the monument but which had clearly been brought to the Stonehenge site from somewhere else.

The work of provenancing has changed since Thomas’s time. Many petrologists have come to focus on the chemical analysis of rock fragments, comparing the mix of elements found within them. In their axe-heads paper, from 2004, Ixer and Bevins had argued for a more multifaceted approach. They maintained that, in addition to performing chemical analyses, petrologists should examine a rock’s appearance under a microscope, using both light shining through the rock—Thomas’s technique—and light reflecting off the opaque minerals on its surface. Ixer, therefore, had the shoebox fragments cut into sections thirty microns wide—thin enough for light to pass through. He reviewed them under a microscope, using both penetrating and reflective light.

Ixer was unable to make a connection between the shoebox fragments and any specific rocks in Wales. So he decided to e-mail Bevins. When Bevins examined the thin rock sections under a microscope, something in the way the light looked reminded him of samples that he had collected decades earlier, while doing his Ph.D. field work, in the nineteen-seventies. Back then, he’d collected a sample from Craig Rhos-y-felin, an outcropping at the foot of the Preseli Hills, a little north of where Thomas had thought the Stonehenge bluestones had come from. Bevins unearthed his old samples and had them sectioned, too. He found that those from Craig Rhos-y-felin matched five of the shoebox fragments.

It was “one of those eureka moments,” Bevins told me. The findings suggested that Thomas had identified the right general area but the wrong specific outcropping as the source of the bluestones. The fragments in the box, at least, had come not from Carn Alw but from Craig Rhos-y-felin. Thanks to new techniques, the question of the origins of the bluestones had been reopened.

Our understanding of Stonehenge has evolved with time. In the seventeenth century, the antiquarian John Aubrey linked the monument to pre-Roman Druids. In the eighteenth century, Stukeley recognized a relationship between the alignment of Stonehenge and the solstice, and attempted to demonstrate that the monument could not have been built by the Romans by taking a series of measurements which proved not to correspond to Roman units. More sophisticated investigations began in 1901, after William Gowland, a metallurgist and archeologist, was appointed to oversee excavations at the site. These uncovered no metal artifacts in the relevant layers of soil, confirming that Stonehenge was a Neolithic monument, built by early farmers in the last period of the Stone Age and the transition to the use of metal tools beginning in the Copper Age.

In 1918, the site was transferred from private ownership into the care of the British government. William Hawley, a former British Army officer turned archeologist, dug there for seven seasons; he showed that Stonehenge had been built not all at once but in successive stages. Archeologists worked to stabilize the stones in the nineteen-fifties and sixties, setting them in concrete to stop them from falling over. Then, in the nineties, they started using high-precision radiocarbon dating on excavated animal bones and antler picks, and so found evidence that some parts of Stonehenge dated back to as early as 3000 B.C., with two phases of construction following over the next thousand years.

Since 2001, there have been at least ten major archeological projects at or around Stonehenge, along with many smaller ones; many have involved techniques unavailable to previous researchers, such as high-precision radiocarbon dating, ground-penetrating radar, and isotope analysis. In 2008, the Strumble-Preseli Ancient Communities and Environment Study, or SPACES project, showed that ancient people had broken up the bluestones and used the fragments to make axes and other small artifacts. Around the same time, the Stonehenge Riverside Project produced evidence that the bluestones had been erected first, forming a different, wider circle around the edge of the monument before being moved closer in. Archeologists also reëxhumed the cremated remains of prehistoric people buried at Stonehenge, and recovered evidence of another bluestone circle within a henge at West Amesbury, on the bank of the River Avon, little more than a mile away.

All this archeological work has produced a wholly new understanding of Stonehenge. Parker Pearson told me that, in the nineteen-fifties, it was widely thought that the monument had been “designed by Bronze Age Mycenaeans”—people who lived in Greece between 1600 and 1100 B.C., and who also built large stone structures there. The theory was that a Mycenaean architect might have directed local builders in Britain. But radiocarbon dating, which showed Stonehenge to predate the Mycenaeans, had consigned that theory to the scrap heap. “Researchers now know that its earlier stages were constructed more than a thousand years earlier, in the Neolithic,” Parker Pearson said. On Salisbury Plain, Neolithic burial monuments, made of earth, generally date to around 3600 B.C.; Stonehenge’s first stage, consisting of a circular ditch enclosing the wide circle of holes that are believed to have once held the bluestones, was built shortly after 3000 B.C., with the larger sarsen stones arriving around five centuries later.

Today, Parker Pearson said, the total picture was that, instead of three phases of construction, Stonehenge appeared to be a dynamic monument built in perhaps half a dozen stages. It all began “with a circle of Welsh bluestones encircling what was Britain’s largest burial ground,” he told me; later, the sarsens were added, “aligning with the eternal movements of the sun and moon.” Stonehenge, he concluded, was “not a Druids’ temple or astronomical observatory” but “a monument of remembrance.”

Having established a connection between the shoebox samples and Craig Rhos-y-felin, Bevins and Ixer contacted Nick Pearce, a professor at Aberystwyth University, and asked him to do a chemical analysis of their fragments. Bluestone, as a category of rock, contains two major subcategories, rhyolite and dolerite, neither of which occurs naturally near Stonehenge. Hundreds of rhyolite samples have been gathered there, and four of the bluestones are made of it. But none of those stones were a petrological match for Craig Rhos-y-felin. It seemed that the shoebox fragments belonged to a missing bluestone.

The shoebox samples contained crystals of a mineral called zircon, which, because it is inert and largely unaffected by temperature and pressure, can be used to fingerprint a rock. Pearce used a laser to vaporize the zircon, then analyzed its chemical composition. He found that it matched zircon contained in many loose fragments found at both Stonehenge and Craig Rhos-y-felin. Bevins and Ixer now believe that the rhyolite fragments are part of what’s called Stone 32D—a “stump” of a bluestone that lies underground at Stonehenge. Stone 32D was exposed in excavations in the nineteen-fifties; photographs from those excavations suggest that its composition is similar to that of rocks at Craig Rhos-y-felin. Bevins and Ixer suspect that 32D was taken from Craig Rhos-y-felin to Stonehenge, where, at some later stage, for some reason, it was broken up. (Bevins has applied for permission to excavate the stone and test it.)

Pearce also analyzed some dolerite fragments, not from the shoebox, that had been drilled out of some bluestones in the nineteen-eighties. Using a mix of petrography and whole-rock chemistry—a process that involves crushing a fist-sized rock sample and examining its components—he, Bevins, and Ixer matched fifty-five per cent of Stonehenge’s dolerite fragments to Carn Goedog, where I’d later venture with Bevins. This was a further revision to Thomas’s theory. Back in the nineteen-twenties, Thomas had concentrated on three sites in the Preseli Hills: Carn Alw for the rhyolite bluestones, and Carn Meini and Cerrig Marchogion, two nearby outcrops, for the dolerites. The new work suggested that he had been close but slightly off. (Carn Goedog lies roughly in the middle of his three candidates.)

On the basis of these findings, Parker Pearson and his team started excavating at Craig Rhos-y-felin in 2011. They uncovered evidence of what Parker Pearson described as “a little hearth and activity area,” and also a few hazelnut shells. The shells were dated to the centuries before 3000 B.C.—roughly the time when Stonehenge’s first stage of construction began. Parker Pearson became convinced that Craig Rhos-y-felin had been a small bluestone quarry site. “They took one or possibly two pillars and no more,” he said. Bevins took me there during our trip; we drove past a cottage, parked, and walked to a grove flanked by slanting gray rock. “This phenomenal cliff face—as a geologist, I would say it’s not natural,” Bevins told me. It looked to him like a quarry.