Lone Survivors

by Chris Stringer

  A good example of the enhanced power of radiocarbon dating came when four colleagues and I investigated one of the enduring mysteries of the Paleolithic record of Britain. Representations of Ice Age art are extremely rare in Britain, and two of the only examples known (or claimed) are from Robin Hood Cave in Derbyshire, found in the 1870s, and from the town of Sherborne in Dorset. Both showed a rather similar profile of a horse engraved on a flat fragment of bone. While the Derbyshire example was discovered by prehistorians in a cave alongside Paleolithic artifacts of appropriate age (about 14,000 years old), the “Sherborne bone” was discovered in 1912 by schoolboys from the local public school, in the vicinity of a quarry from which no comparable material had ever been reported. Serious doubts were soon raised about the authenticity of the Sherborne discovery, but direct radiocarbon dating could not have been contemplated when application of the method would probably have destroyed most or all of the object. The advent of AMS dating at Oxford University allowed us, in 1995, to drill a tiny sample from it and date the bone to about six hundred years old, while microscopic studies of the engraving showed that it was probably carried out quite recently with a metal implement, rather than a flint tool. This result was in line with suggestions from one of the staff at Sherborne fourteen years after the “discovery” that a boy had probably copied the engraving from an illustration of the Robin Hood specimen in their school library, in order to play a joke on their science teacher!

  But even AMS dating is not perfect, since it finds and produces a date from whatever radiocarbon is in the sample; even a small amount of contaminant radiocarbon can greatly affect an age estimate, especially when the sample is 30,000 or 40,000 years old, and only a tiny fraction of its original radiocarbon is still there. Fortunately, new preparation procedures such as acid-base-wet oxidation (ABOX) dating for charcoal samples and ultrafiltration for bone are largely overcoming the problems of contamination in dating Paleolithic materials and are giving increasingly trustworthy determinations. The advantages provided by ultrafiltration were very well demonstrated through the redating of bone samples from Gough’s Cave in Cheddar Gorge, Somerset. This is one of Britain’s most spectacular tourist caves but also one of our most important Upper Paleolithic sites. Excavations spread over more than one hundred years have revealed quantities of stone artifacts together with human and animal bones representing its late Ice Age inhabitants. Revised radiocarbon dating has now shed further light on the nature of the human presence here, and on the timing of the return of people to Britain after a period of Ice Age abandonment lasting about 10,000 years. Prior to this new research, it was uncertain when occupation took place and how different parts of the archaeological story fitted together, but it now seems that Gough’s Cave was one of the first sites to be used by hunters of wild horses and red deer when people returned to Britain after the peak of the last glaciation.

  This transformation was achieved by the dating specialist Tom Higham and the archaeologist Roger Jacobi, using ultrafiltration pretreatment on animal bones butchered or worked by the Stone Age humans, and on the remains of the humans themselves. Previously, the radiocarbon dates obtained had only made it possible to tie occupation down to a span of about 1,500 years. Now, much greater confidence can be ascribed to dates that show almost all the Upper Paleolithic material in the cave accumulated over as little as two to three human generations, centering around 14,700 years ago. Interestingly, this date corresponds precisely to a dramatic warming of climate recorded in the composition of annual layers of ice in Greenland. These archives suggest that the previously ice-covered Atlantic Ocean defrosted in about five years. Among the material dated at Gough’s were bones of several of the humans, some of which show patterns of cut marks interpreted as evidence of cannibalism. Before, it had been thought that these might have belonged to a more recent phase of activity than the one associated with the horse and deer hunting, but we now know they were precisely the same age. Thus the animals, and the people who preyed on them, represented some of the first colonizers of Britain after the peak of the last Ice Age. As the climate rapidly warmed, herds of horses and deer must have migrated across Doggerland, now submerged under the North Sea, and the hunters followed.

  A much older British fossil that I have been involved in studying was found in 1927 at Kent’s Cavern, in southwest England. After its discovery, the anatomist Arthur Keith described this fragment of upper jaw as a modern human, but it had to wait another sixty years to achieve further fame, when it was one of the first fossil humans to be dated by the radiocarbon accelerator at Oxford. The estimated age of about 35,000 years made it one of the oldest modern humans in Europe; subsequent detective work on the Kent’s Cavern archives by Roger Jacobi suggested that it could date from even earlier. So, in 2004, we decided to borrow the specimen from Torquay Museum and restudy it, using every scientific approach we could muster. The team I assembled involved researchers such as Erik Trinkaus and Tim Compton, specialists in CT and ancient DNA (techniques that I will discuss in chapters 3 and 7), curators and conservators, and Higham and Jacobi. Careful examination and CT modeling confirmed Erik’s hunch that one of the teeth had been glued back into the wrong socket; a new reconstruction was made, allowing the sampling of the tooth roots for ancient DNA and ultrafiltered accelerator dating. Sadly, both of those attempts ultimately failed, but accelerator dating of animal bones found around the fossil indicates that its real age is some 40,000 years, and it may record an early spread of modern humans to western Europe.

  Other physical dating methods that can be applied to fossil and archaeological materials beyond the limits of radiocarbon dating have also been developed or enhanced in the last twenty years. These include uranium-series (U-S) dating, which is based on the radioactive decay of different forms of uranium. Accumulation and measurement of the so-called daughter products are possible in substances like stalagmites and corals. The former has been very useful in cave sites, while the latter has been used to examine past changes in sea levels around tropical and subtropical coasts and, as mentioned already, to check the accuracy of radiocarbon measurements. One of the holy grails of dating has been to get uranium decay methods to work on fossil bones. However, this has proved notoriously difficult because, in contrast to stalagmites and corals, which are essentially sealed after deposition, bone continues to be open to the accumulation or loss of uranium (for example, as groundwater percolates through it). This means that its physical clock can run very erratically. Nevertheless considerable progress has been made recently, and I will discuss some of the results as applied to the Broken Hill fossils of Homo heidelbergensis in chapter 9.

  A number of other methods depend on the fact that crystalline substances such as sand grains, flint, or the enamel of a tooth store up changes in electrons within their crystal structure from the radiation they receive from their surroundings, once they are buried. The amount of change (corresponding to radiation damage) can be measured from the accumulated energy released in the sand or flint when treated with a laser beam (optically stimulated luminescence, or OSL) or by heating (thermoluminescence, or TL), while in tooth enamel, the accumulated changes in the electrons can be detected using microwave radiation (electron spin resonance, or ESR). For any of these methods to work, the radiation signal must first be set at zero—for example, when a tooth begins to grow (ESR)—or set back to zero when the previous signal is wiped out as sand grains are bleached by exposure to the sun, or when flint or clay is strongly heated in a fire (luminescence). Provided the rate of subsequent accumulation of radiation damage in the material can be estimated from the environment in which it was buried, the length of time it has been in the ground (for example, in a Cro-Magnon fireplace or a Neanderthal butchery site) can be estimated.

  As with radiocarbon dating, procedures have continuously been refined, so now even single grains of sand can be dated by luminescence. Equally, in the case of ESR, where previously a large chunk of a tooth had to be sacrificed, we have moved to a
situation in which, using the microscopic technique of laser ablation, it is now possible to directly date a tiny area of fossil human tooth enamel. Another potential complication of ESR dating is the fact that fossils take up uranium when they are buried; hence they contribute to their own accumulated radiation dose. Estimating the rate of uranium uptake is critical (did most of it get in soon after burial or did it come in gradually?), but this unknown can now be addressed by combining, or “coupling,” an ESR determination with a U-S date on the same piece of enamel, and looking for the age estimate that is most compatible when comparing the two.

  An excellent example of the tremendous impacts that luminescence and ESR dating have made on human evolution came out of the Middle East, from the famous Israeli caves of Tabun and Skhul (Mount Carmel) discussed in chapter 1. I was fortunate enough to be involved in some of the pioneering work on dating these sites in the late 1980s and early 1990s, since the Natural History Museum has a share of the human fossils, artifacts, and sediments from them.

  Ranges of the main dating methods for recent human evolution.

  These fossils had a key role in developing ideas of Neanderthal–modern human relationships: essentially, did they represent one single rather variable population, perhaps 40,000 years old, or did the more modern-looking Skhul people succeed Neanderthals like Tabun, and perhaps evolve from them? Even when more Neanderthal fossils were added to the mix from Israeli sites like Amud and Kebara (both with quite complete skeletons from apparent burials), and more modern-looking skeletons were added from the site of Qafzeh (near Nazareth), the picture did not get any clearer. Relative dating using similarities in stone tools suggested that they were all rather close in age, while a radiocarbon date on some charcoal from Tabun suggested the Neanderthal from there was not much more than 40,000 years old. In turn, based on the known succession of Neanderthals and modern humans in Europe about 35,000 years ago, it was assumed that a similar sequence would be found in the Middle East, though perhaps it would be a little older. Thus in the early 1980s it seemed reasonable to assume that the Qafzeh and Skhul “moderns” were about 40,000 years old, and two different evolutionary scenarios were proposed in the region. Erik Trinkaus favored the view that the Neanderthals had evolved fairly rapidly into moderns there, while I took the view that there had been a replacement of the Tabun people by the Skhul and Qafzeh moderns—but we were both wrong! And there was already a clue as to why we were wrong in some relative dating work on animal remains from the sites.

  Qafzeh Cave, like many of the sites, contained fossilized rodent remains as well as the human burials, and these can provide useful information not only about the local environment but also about the dating of sites. Pioneering studies of these small mammal remains by Israeli researchers suggested that Qafzeh could in fact be older than the Neanderthal sites rather than younger. This led the archaeologist Ofer Bar-Yosef to propose that the Qafzeh early moderns could date to as much as 70,000 years. Yet such an age was clearly beyond the reach of radiocarbon, so how could this view be tested? At last, with refinements in ESR and luminescence dating that came during the 1980s, this became possible.

  The first significant application of these emerging techniques (thermoluminescence applied to flints that had been heated in a hearth) came from French–Israeli collaborations and initially seemed to reinforce the expected pattern in the Middle East, dating the recently discovered Neanderthal burial at Kebara to about 60,000 years. However, shortly afterward in 1988, the first application was made to the site of the Qafzeh early modern material, giving an astonishing age estimate of about 90,000 years, more than twice the generally expected figure, supporting or even exceeding the relative dating suggested by the rodents! Next up were the Skhul and Tabun sites, and for these I started working with dating specialists like Rainer Grün and Henry Schwarcz. Henry, a Canadian, is the doyen of dating in this time range, and Rainer, a German now working in Canberra, had studied and worked with him. Analyzing samples of animal teeth from both the sites for ESR dating gave equally revealing results. Within three years we had shown that the Skhul early moderns were at least as old as the Qafzeh ones, while the deep Tabun Cave sequence covered hundreds of thousands rather than tens of thousands of years. We also suggested that the Neanderthal burial from Tabun was much older than the 40,000-year radiocarbon date: it was perhaps as old as the moderns from Skhul and Qafzeh.

  There was obviously a much more complex sequence than any of us had envisaged, and in some ways the expected chain of events was turned on its head: the modern-looking people from Skhul and Qafzeh were older than the Kebara Neanderthal. Further work showed that they were also older than the Amud Neanderthal. Thus they could not have evolved from these late Neanderthals, and, puzzlingly, those late Neanderthals were in the Middle East after the early modern humans, and not before them. Continuing dating work using all the available techniques now suggests that the Skhul and Qafzeh people actually range from about 90,000 to 120,000 years old, while the Tabun Neanderthal is most likely about 120,000 years old. So the emerging scenario is one where populations apparently ebbed and flowed in the region, which makes sense given its geographic position between the evolving worlds of the Neanderthals to the north and early moderns to the south.

  Bar-Yosef suggested that the moderns came up into the region during a particularly warm and wet period, about 120,000 years ago, but as the succeeding Ice Age cooled and dried the north, the Neanderthals were pushed down there and took over the region—an intriguing reversal of the usual Replacement model! In fact, I think such changes could have been long-standing and even more complex, going back deep into the evolutionary history of the two species. At times when conditions were favorable, one or the other group, or perhaps both, would have moved into the region, while at times of severe aridity it might even have been completely abandoned. Whether populations were generally pushed there by unfavorable conditions in their home territories, or were pulled there by climatic ameliorations leading to population expansions, we do not yet know, but new climatic data are emerging.

  The potential of ESR to match the ability of AMS radiocarbon in directly dating human fossils is at last being realized. In 1996 the first of an increasing number of applications of this technique to significant human fossils was made when Rainer Grün and I collaborated with colleagues including James Brink from South Africa to date the Florisbad human skull. This fossil, which had been found in 1932, is actually rather incomplete but seems to combine a large and fairly modern-looking face with a strong brow ridge and somewhat receding forehead. For many years it was assumed to date from about 40,000 years ago, based on a radiocarbon date from peat deposits at the site, and on that basis it seemed to be a relic hanging on in the margins of southern Africa, while moderns were evolving and spreading in western Asia and Europe. As such, it supposedly demonstrated the backward role of Africa in modern human evolution—the primitive Florisbad humans merely marking time until moderns arrived from farther north and replaced them. However, the fossil preserved one upper molar tooth, and a tiny fragment of its enamel was taken to Rainer’s lab in Australia for ESR dating—with sensational results. The fossil was not 40,000 but about 260,000 years old! Thus its potential role in human evolution was revolutionized at a stroke: rather than representing a southern African equivalent of the Neanderthals, on the brink of extinction, it could instead have been an ancestor to us all.

  There are some situations where even the best physical dating techniques need a helping hand, and combinations of physical and relative methods are required. The Neanderthals apparently disappeared about 30,000 years ago, but the factors leading to this and the time scale for their demise are still fiercely debated. While accelerator radiocarbon dating gives excellent precision in the measurement of an age, it does have problems of accuracy compared with calendar years during this critical period of time, both because the rate of formation of radiocarbon in the atmosphere was unusually variable then, and because even a tiny amount of con
tamination from young or old carbon will make a significant difference to the age obtained. As I explained earlier, this latter problem is being addressed through techniques that very effectively remove contaminants before dating is attempted. But to address the former issue, there were fortunately other significant events in Europe during that period to provide new and potentially very accurate ways of relative dating. As I discuss further in chapter 4, a massive volcanic eruption took place in the Campania region of central Italy about 39,300 years ago (which we know from argon dating). As well as enormous quantities of local deposits such as lava, pumice, and ash, the eruption also produced much finer volcanic dust, known as crypto- or microtephra because it cannot be seen with the naked eye. This microtephra may be ejected into the upper atmosphere and travel for many thousands of miles, and the Campanian Ignimbrite—from the Latin words igni (fire) and imbri (rain)—settled eastward as far as Russia and North Africa.

  The CI, as it is known, has now been found in dozens of archaeological sites, including the famous Russian localities at Kostenki, in levels which we already knew from radiocarbon dating were at least 35,000 years old. Each volcanic eruption took place as a result of unique combinations of factors like chemical composition, temperature, and pressure, and thus can be “fingerprinted” and recognized. So wherever the special CI chemical signature is found in an archaeological site, we can be pretty confident that the level concerned, with its associated fossils and artifacts, was laid down just over 39,000 years ago. In turn, all such sites can be correlated to this age by a lattice of synchronous volcanic deposits.