The remarkable success of the English as colonists, compared to other
European nations, has been ascribed to their “daring and persistent
energy”; a result which is well illustrated by comparing the progress of
the Canadians of English and French extraction; but who can say how
the English gained their energy? There is apparently much truth in the
belief that the wonderful progress of the United States, as well as the
character of the people, are the results of natural selection; for the
more energetic, restless, and courageous men from all parts of Europe
have emigrated during the last ten or twelve generations to that great
country, and have there succeeded best. . . . Obscure as is the problem
of the advance of civilisation, we can at least see that a nation which
produced during a lengthened period the greatest number of highly
intellectual, energetic, brave, patriotic, and benevolent men, would
generally prevail over less favoured nations. Natural selection follows
from the struggle for existence; and this from a rapid rate of increase.
It is impossible not to regret bitterly, but whether wisely is another
question, the rate at which man tends to increase; for this leads in
barbarous tribes to infanticide and many other evils, and in civilised
nations to abject poverty, celibacy, and to the late marriages of the
prudent. But as man suffers from the same physical evils as the lower
animals, he has no right to expect an immunity from the evils conse
quent on the struggle for existence. Had he not been subjected during
primeval times to natural selection, assuredly he would never have
attained to his present rank.

WITH SETTLEMENT and the invention of agriculture, human societies embarked on a trajectory quite different from the foraging life that had hitherto been their only choice. The new behaviors that had now been developed allowed people to construct complex societies and urban civilizations.

They learned to treat strangers as kin, at least in the context of reciprocal exchanges and trade. They coordinated their activities through religious rites. They defended their territory against neighboring tribes, or attacked them when the moment seemed propitious. With settlement came specialization of roles, administrators to take control of surpluses, priests to organize religious ceremonies, headmen and kings to manage trade and defense.

The first cities started springing up in southern Mesopotamia some 6,000 years ago. Uruk, in what is now Iraq, sprawled over some 200 hectares (500 acres) with large public buildings. The city required armies of laborers and an administration to recruit and feed them. As societies became more intricate, their operation demanded more sophisticated skills and perhaps more specialized cognitive abilities, ones at least that no forager had had occasion to exercise. The invention of writing around 3400 BC opened the way to the beginning of recorded history. The first great urban civilizations emerged in Egypt, Mesopotamia, India and China. The next phase of the human experiment had begun.

Genetics, which illumines many aspects of prehistory, yields even greater returns when applied to the historical past because it can be related to known people or events. DNA can be used to analyze populations, saying who came from where, which helps understand mixtures of people like those of the British Isles. DNA faithfully records who slept with whom throughout the ages, a matter of historical interest in cases like the secret family of Thomas Jefferson. And with populations that have married within themselves for centuries, like those of Jews, DNA can reach back to the time of the patriarchs.

Geneticists may in future be able to trace back human lineages or pedigrees to all times and places, providing a genetic framework for exploring almost every historical period. Meanwhile a promising start has been made, as is evident from the following cases.

The Secret Strategy of Genghis Khan

In the year 1227 the Mongol conqueror Genghis Khan died, perhaps in a fall from his favorite horse. His empire stretched from the Caspian Sea to the Pacific Ocean and included much of Russia, China, and Central Asia. His followers brought his body home to a hill in northeast Mongolia. To keep his burial place secret, all those who interred him are said to have been killed, and their assassins were dispatched in turn.

Whether or not that story is true, Genghis’s tomb remains secret and has defied two recent attempts, one by a Japanese expedition, one by Americans, to locate it. But while archaeologists were frustrated in their search for Genghis’s hoard, geneticists engaged on a quite different task stumbled across a more vital part of Genghis Khan’s legacy.

A team led by Chris Tyler-Smith of Oxford University had analyzed the Y chromosomes of some 2,000 men from populations across the Eurasian land mass. They noticed that many of the chromosomes fell into a single cluster. Some chromosomes in the cluster were identical at each of 15 sites tested and others were just one mutational step removed from this master sequence. The striking feature of the cluster was that the owners of its Y chromosomes did not all come from a single population, as would have been expected, but from regions all over Eurasia.

A clue to their origin was that the Y chromosome with the master sequence was particularly common in Inner Mongolia. A quarter of the men tested from this region carried the master sequence chromosome or its close derivatives. Another clue was that only 16 of the 50 or so Asian populations studied included men with the master sequence, yet all but one of these 16 live within what were the borders of the Mongol empire at the time of Genghis’s death. The one exception was the Hazara of Afghanistan and Pakistan, who are thought to be descendants of Mongol soldiers sent to garrison the region.

Tyler-Smith and his colleagues believe the master sequence chromosome must be that of the Mongol royal house. It would have been carried by Genghis Khan and by the male relatives he sent to administer the regions of his far flung empire. Dating methods suggest the cluster started to form around 1,000 years ago, the time that Genghis’s dynasty began its ascent to power.²⁹¹

Mongol soldiers doubtless raped many women during their extraordinarily cruel and murderous campaigns. But there may be a more significant reason for the existence of so many men carrying the specific chromosome of the Mongol royal house: Genghis accumulated a large harem in which he seems to have labored with surprising industry. The fourteenth century Persian historian Rashid ad-Din, who served as chief minister of the Mongol government of Persia, wrote that Genghis Khan had nearly 500 wives and concubines, and that it was his practice to take women into his harem as booty whenever he conquered a new tribe.

Another Persian historian of the Mongol empire, ’Ata-Malik Juvaini, includes without further explanation the following observation in his History of the World Conqueror, completed in AD 1260: “Of the issue of the race and lineage of Chingiz-Khan there are now living in the comfort of wealth and affluence more than 20,000. More than this I will not say but shall rather avoid [the subject] lest the readers of this history should accuse the writer of these lines of exaggeration and hyperbole and ask how from the loins of one man there could spring in so short a time so great a progeny.”²⁹²

Genghis’s interest in procreation was shared by his sons, one of whom is credited with 40 sons. It seems to have been a deliberate policy of Genghis and his heirs to father as many children as possible. “It’s pretty clear what they were doing when they were not fighting,” comments a historian of the Mongol period, David Morgan of the University of Wisconsin.²⁹³

From the proportion of Mongol royal house Y chromosomes in their sample, Tyler-Smith and his colleagues have been able to calculate just how well Genghis succeeded in his procreative program. An astonishing 8% of males throughout the former lands of the Mongol empire carry the Y chromosome of Genghis Khan. This amounts to a total of 16 million men, or about 0.5% of the world’s total.

The second most common Y chromosome in East Asia, after that of Genghis Khan, is one that probably belongs to Giocangga, the patriarch of the Manchu rulers who governed China as the Qing dynasty from 1644 to 1912. The Qing imperial nobility consisted of male descendants of Giocangga and his grandson Nurhaci, who founded the dynasty. The nobility was highly privileged and its members were able to keep many concubines. In addition the Qing nobility used marriages to cement political alliances with other peoples of northern China such as the Mongols.

Tyler-Smith has detected the Manchu chromosome in 7 northern populations though not in the Han, the major Chinese ethnic group. He believes the chromosome belongs to the Manchu royal house because of its frequency, its geographical distribution and the fact that its founder, according to genetic evidence from the chromosome itself, lived some 500 years ago— Giocangga died in 1582. Tyler-Smith estimates that the Manchu Y chromosome is carried by 1.6 million men living today.²⁹⁴

A third patriarch, one with an estimated 2 to 3 million living descendants, has come to light through a study of Irish Y chromosomes. He may well be Niall of the Nine Hostages, an Irish high king of the fifth century A.D. whom some historians had regarded as a probably legendary figure.³⁶⁴

The genome offers a unique new window into history, one that is especially illuminating when DNA evidence can be combined with historical evidence. The cases of Genghis, Giocangga, and Niall of the Nine Hostages raise the question of whether large-scale procreation isn’t just a perk of political power but may be a salient, even if unconscious, motivation for it.

A History of Britain, from the Genome’s Viewpoint

The genome often holds surprising answers for historical questions that involve lineages. Consider the matter of English surnames. Commoners acquired surnames between AD 1250 and 1350, apparently for the convenience of feudal record keepers who needed to differentiate between tenant farmers with the same first names. The surnames were not highly original. They tended to be a person’s profession (Smith, Butcher), or a patronymic (Johnson, Peterson), or derived from some landscape feature (Hill, Bush). Historians assumed that the same name had been invented many times over, so there would be no reason to assume that people with the same surname had a common ancestor in the thirteenth century. George Redmonds, however, a historian of British surnames and place names, came to feel that many English surnames had single progenitors. “But it was never possible to prove it genealogically because we don’t have enough evidence,” he says.²⁹⁵

That began to change when Redmonds’s advice was sought by Bryan Sykes, a geneticist at Oxford University. Sykes had been invited to give a talk to scientists at Glaxo Wellcome, a large British drug company, which in the mid-1990’s was beginning to take an interest in the human genome. The organizers of the conference at which Sykes was to speak asked him several times if he was related to the company’s then chairman, Sir Richard Sykes. He kept saying no, not that he knew of. Even the company chauffeur who arrived to drive him to the conference asked the same question.

Sykes was about to repeat his usual denial but suddenly a thought crossed his mind. “Maybe Sir Richard and I were related after all, but without realizing it,” he writes. “And, more to the point, maybe I could prove it by a genetic test.” Sykes asked the chauffeur to wait and rushed back to his lab for a genetic sampling kit (essentially a swab to brush cells off the inside of the cheek). At the conference, which his namesake was attending, he asked him for a sample.²⁹⁶

The two men had grown up in quite different parts of the country. Bryan Sykes’s family lived in Hampshire, in southern England, Richard Sykes had grown up in Yorkshire, in the north. Apart from both having been trained as scientists, they seemingly had nothing else in common. But it turned out there was something else: they possessed the same Y chromosome.

Y chromosomes, of course, are bequeathed from father to son just as are surnames. After the test with his namesake, Bryan Sykes wondered if other Sykeses too might be related to one another. Research showed that there were many Sykeses living in Yorkshire and that the surname itself was derived from a Yorkshire word, sike, meaning a moorland stream. Sykes picked some 250 of his namesakes at random from the Yorkshire area and sent each a letter asking for a sample of his DNA. About a quarter obliged, and from analysis of their cells a distinct pattern emerged. About a half carried the identical Y chromosome, or one that was just a single mutational step away from it. The rest had a miscellany of unrelated Y chromosomes.

Several interesting conclusions followed. First, there was just one real Sykes Y chromosome. All the men who carried it were presumably descended from the first bearer of the surname. That meant the surname had been assigned only once or, if more than once, all other lines had ended without male heirs and no longer existed.

As for the 50% of Sykeses who did not carry the true Sykes Y chromosome, their cases must have been largely the result of what geneticists delicately refer to as a “nonpaternity event” at some point in their family tree, meaning the biological father was not the same as the father of record. Adoption is one possible explanation for nonpaternity, though it probably wouldn’t account for many cases.

If half of Sykes men alive today have a nonpaternity event somewhere in their genealogy, doesn’t that raise considerable doubt about the virtue of Sykes wives through the ages? Bryan Sykes argues this is not the case. Assuming there have been 23 generations of Sykes since the first Mr. Sykes in the thirteenth century, an infidelity rate of merely 1.3% per generation would account for the fact that only half of contemporary Sykes men carry the correct Y chromosome. This compares very favorably with the nonpaternity rates of contemporary populations, Sykes comments, which run from 1.4% to 30%, though most fall in the 2 to 5% range.²⁹⁷

“I’m proud to say I have the aboriginal chromosome,” Sykes replied when asked whether he was a true Sykes or one of the out-of-wedlock kind. His early ancestors seem to have been a rough lot; they appear regularly in court records of the fourteenth century as having incurred fines for cutting down trees or stealing sheep. “Nonetheless, their wives were faithful through all this,” Sykes says.²⁹⁸

Redmonds, the local historian, has traced the earliest Sykeses to the villages of Flockton, Slaithwaite and Saddleworth in West Yorkshire. The first mention of the name is a court record of AD 1286 referring to a Henri del Sike of Flockton. There are still Sykeses living in Flockton. Redmonds was able to locate the plot of land of which Henri del Sike was tenant, a farm that straddled a stream between two parishes. He took his geneticist friend to visit. “There was no sign of the farmhouse which my ancestor, the very first Sykes, had occupied, but even so, it felt quite extraordinary to be here,” Sykes wrote. “Looking round at the old mill, the track and the stream, it seemed that nothing in the landscape had greatly changed. Nor had it. The field and croft boundaries were as they had been in the late thirteenth century when Henri del Sike was living here. As I stood, I could almost hear the voices of the children—my ancestors—laughing as they threw pebbles into the stream.”²⁹⁹

Sykes has analyzed the Y chromosomes of three other English surnames and found that, as with his own, each can be traced to a single bearer. Because of his research it now seems that many English surnames once had a single bearer, and even the commonest ones like Clark and Smith may be descended from only a few originals.

Genetic analysis is at the least a new tool for historians and may one day support a new kind of history, possibly somewhat at variance with the conventional kind. English schoolchildren are taught that their history really begins with the Roman invasion of 55 BC and Caesar’s defeat of the Celtic tribes who opposed him. The true bearers of the English heritage, the textbooks imply, are the Anglo-Saxons, later invaders whose Germanic language was the ancestor of English. The defeated Celtic inhabitants of Britain are assumed to have been pushed back into the hinterlands of Wales and Scotland and largely disappear from most history books.

But a survey of British Y chromosomes shows that the Y chromosomes characteristic of Celtic speakers, far from having disappeared, are carried by a large proportion of the male population of Britain. Nowhere does the indigenous population seem to have been wiped out, either by the Anglo-Saxons who invaded from Denmark and northern Germany in the sixth and seventh centuries AD, or by the Danish and Norwegian Vikings who arrived in the ninth and tenth centuries.³⁰⁰ (Two other groups of invaders, the Romans and the Normans, probably arrived in numbers too small to have left a demographic mark.)

The Y chromosomes common among Celts have a particular set of DNA markers known to geneticists as the Atlantic modal haplotype, or AMH. AMH Y chromosomes are also found, it so happens, in the Basque region of Spain, whose inhabitants are thought to represent the original inhabitants of Europe. AMH-type Y chromosomes are particularly common in places like Castlerea in central Ireland, which no invaders ever reached. This suggests that the chromosomes are the signature of the first hunter-gatherers who arrived in Britain and Ireland toward the end of the Pleistocene ice age 10,000 years ago.

Given the similarity between Basque and Irish Y chromosomes, some geneticists suspect that people who had used Spain as a southern refuge during the Last Glacial Maximum started to move northward as the glaciers melted. Many may have traveled by boat up the west coast of Europe, entered the waterway between Ireland and England and settled on each side of it.³⁰¹

The carriers of the AMH Y chromosomes presumably spoke a language like Basque or some other tongue belonging to the first Paleolithic inhabitants of Europe. So it is a puzzle that the chromosome is now associated with Celtic, an Indo-European language that spread to Britain only in the first century BC, along with ironworking technology and agriculture. The solution is presumably that the Celtic way of life became widespread in Britain mostly by cultural transmission, not by a large invasion of Celts. The cultural shift evidently included the adoption of Celtic language by the original inhabitants of the British Isles.

Another layer in this puzzle is that British mitochondrial DNA—the genetic element inherited solely through the female line—shows a different pattern from the Y chromosomes. The mitochondrial DNA generally resembles that of northern Europe. This suggests that the Celtic speakers in Britain obtained many of their wives from northern Europe, perhaps in exchange with European Celts, perhaps by pillage and rapine.³⁰²

The historian Norman Davies opens his recent history of the British Isles by noting that the mitochondrial DNA recovered from bones buried some 8,000 years ago in a cave in the Cheddar Gorge matched that of a local schoolmaster, proving the continuity of the human population of the region. ³⁰³ The genome is already being welcomed by historians as a rich new source of unexpected information.

The Origin of Icelanders

England was invaded by Vikings from both Denmark and Norway. The influence of the Danish Vikings can be seen most strongly in Y chromosomes from York and Norfolk in the eastern regions that bore the brunt of the Danish invasions. The Norwegian Vikings operated to the north of the Danes. In the ninth century AD they captured the Orkney Islands to the northeast of Scotland and made them a base of operations. Norn, a form of old Norse, was spoken on the islands until the eighteenth century. Norwegian Vikings have left a strong genetic signature among Orcadians, as Orkney Islanders are known, but their traces can also be seen farther afield, particularly in Iceland.

From their base in Orkney the Norwegian Vikings sailed around the northern coast of Scotland and down the waterway between Britain and Ireland, making settlements on both the British and Irish sides. In AD 870, the Vikings discovered Iceland, several days’ sail to the northwest of Scotland. Apart from some Irish hermits, who quickly left, the island was uninhabited. News of this virgin territory, with no hostile natives, soon got around, and for 60 years there was a steady stream of settlers. Immigration ceased in AD 930, perhaps because many of the trees had been chopped down, prompting an ecological crisis, and there was no more unclaimed farmland left. The island was then essentially closed to new immigration until modern times.

Iceland’s genetic history has received much attention, both for its intrinsic interest as an isolated human population and because its population has become a leading source for discovering the genetic roots of common diseases ranging from cancer and heart disease to asthma and schizophrenia. These diseases are thought to result from several errant genes acting in combination. The errant genes are very hard to detect because each makes only a small contribution to the overall disease. For various reasons, including an excellent system of medical records, Iceland offers many advantages in searching for such genes. In 1996 Kari Stefansson, an American-trained Icelandic neurologist, put together a high powered genetic analysis company, DeCode Genetics, which has enjoyed considerable success in identifying disease genes in Icelanders and other populations. The company and its large pharmaceutical partners hope to develop diagnostic tests and drugs on the basis of the Icelandic findings. It is therefore of considerable interest to know if Icelanders are genetically similar enough to other populations, particularly those of the United States and Europe, for discoveries about their patterns of genetic disease to be relevant elsewhere in the world.

Icelandic records from the twelfth and thirteenth centuries, notably documents known as the Book of Settlements and the Book of Icelanders, indicate that although Norse Vikings directed the immigration to Iceland, the inflow included people from the Norse settlements in the Orkneys and the coastal regions of Scotland, northern England and Ireland. Most of these Norse invaders, after the initial conquest, had intermarried with the local population. Assuming these Vikings brought their families, many if not most of the women in the founding Icelandic population would have been British or Irish, and in either case of Celtic origin. The Book of Settlements mentions only a small proportion of the founding settlers by name but of those whose ancestry is recorded, only 5% of the men came from the British Isles but 17% of the women. In addition, the Vikings captured slaves in raids in both countries, many of whom were probably women.

Icelandic historians have developed the case that their country was probably founded by men who were mostly Norse and women who were mostly from the British Isles, especially Ireland. This claim of descent from two important peoples, the Vikings and the Celts, helped to differentiate the Icelanders of the nineteenth and twentieth centuries from their much-resented rulers, the Danes. “The result of this conflation is the dominant modern concept of Icelandic origins, one that fuses the nobility and heroism of the Norse with the literary and other cultural traditions of the Irish and other peoples of the ‘Celtic fringe,’” write a group of Icelandic and other experts.³⁰⁴

Given the power of modern genetics to deconstruct complex populations like that of England, it should be a simple problem to analyze the genetics of Iceland and check the validity of the historians’ position. But it’s not so easy. Comparison of Icelandic Y chromosomes with those in Scandinavia and the British Isles confirms that most of the male founders were indeed Norse, though not overwhelmingly so: some 20 to 25% of Iceland’s founding fathers appear to have had Gaelic, meaning Celtic, ancestry, with the rest being of Norse origin.³⁰⁵

The founding mothers are much harder to trace. The patterns of mitochondrial DNA found among Icelanders today look generically European but without greatly resembling those of any particular country.³⁰⁶ They look, well, Icelandic. The reason is probably genetic drift, the random gain or loss of genetic signatures between generations, accentuated by the violent fluctuations in Iceland’s population since the settlement. The Black Death killed 45% of the population in 1402-1404. A smallpox epidemic in 1708 reduced the population by 35%. Famine, the aftermath of a volcanic eruption, caused a 20% decline in 1784-1785. After each of these population declines and expansions, the characteristic mix of mitochondrial DNA signatures would have changed, pushing the population farther down a separate path from that of its source populations.

With the genetic answers being so Delphic, a team of researchers has resorted to the old technique of craniometry. They measured Icelandic skulls from the settlement period and compared them with medieval skull collections from Ireland and Norway. Unfortunately the Icelandic skulls were not in good enough condition to tell their sex. Overall, they seemed very similar to the Norse skulls and less like those of Ireland. The researchers say that “although our results do not preclude a significant Irish or other contingent among the settlers of Iceland, we conclude that the founding population was of predominantly (60-90%) Norwegian origin.”³⁰⁷

To help in its quest for disease genes, DeCode Genetics has assembled a genealogical database of the Icelandic population that extends back 1,100 years into the past. It is based on calfskin documents that hold the first 300 years of records, on church archives, and on the data from three complete censuses that were held starting in 1703. DeCode’s genealogist, Thordur Kristjansson, reckons the database includes the names of about half the Icelanders who have ever lived, including 85% of those born in the nineteenth century and almost everyone who lived in the twentieth century.³⁰⁸ The database has enabled DeCode researchers to explore the historical dynamics of a human population in fascinating detail.

One finding is that generation times are shorter in mother-to-daughter lines of descent than in father-to-son lineages. The average interval between generations was 29 years in female lineages going back to 1698, 32 years for male lineages. The difference presumably reflects the simple fact that women tend to be younger than their husbands.

A greater surprise is how many people in one generation leave few or no descendants in the next. DeCode has traced the ancestry of all 131,060 Icelanders born from 1972 to 2002 back to two cohorts of ancestors. Of all contemporary Icelandic women born since 1972, 92% are descended from only 22% of the women born in the 1848-1892 cohort, and 86% of contemporary men are the progeny of just 26% of this group.

The progeny pyramid narrowed even more steeply going back to an earlier generation of ancestors, those born between 1698 and 1742. Because of the incompleteness of the genealogy for earlier centuries, the pedigree of many contemporary Icelanders could not be traced that far back. Nevertheless, DeCode researchers found that just 7% of the women born in the early eighteenth century period are the ancestresses of 62% of contemporary women, while 10% of the men of this period fathered 71% of contemporary males.³⁰⁹ Most people, in other words, have lines of descent that eventually go extinct, at least in a population the size of Iceland’s, while just a few ancestors give rise to the majority of subsequent population cohorts.

This difference in reproductive success seems to be due largely to genetic drift, the force of which depends on the size of the population. Iceland’s population fell to a post-settlement low of 33,000 after the 1708 smallpox epidemic but steadily increased from the beginning of the nineteenth century to its present level of 290,000. Even during this expansion, the influence of genetic drift was still at work. Before the end of the Pleistocene, there may have been many human populations no bigger than this, offering much grist for drift to work on.

Jewish Origins

The population history of Jews has been studied more than that of most other groups and has yielded one surprise after another. The population’s first remarkable feature, from which all the others follow, is that Jews have to a significant extent married among themselves over the centuries. Jewish communities, in other words, have been largely endogamous, at least until recent times, which means the population’s gene pool has had time to develop its own private history, and this genetic history has shed light on many historical events.

An important consequence of endogamy is that the gene pool is not diluted through intermarriage and so the selective pressures that may act on a population are able, over time, to favor specific genetic variations. A striking possibility, plausible though not yet confirmed, is that one particular Jewish community, the Ashkenazim of northern and central Europe, lived for a long time under a harsh selective pressure that raised certain variant genes to high frequency. These variant genes are well known to physicians because of their serious side effects—when inherited from both parents they cause a variety of serious diseases. But the variant genes can hardly have become so common through their role in promoting disease. They must confer some special benefit, and that, the hypothesis goes, is increased intelligence.

The selective pressure, according to this idea, was the restriction of Ashkenazim by their European host populations to a small number of occupations that happened to require higher than usual mental agility. The pressure lasted from about AD 800 to 1700. If true, the hypothesis, described further below, has several interesting implications, including that it would represent a very recent and dynamic example of human evolutionary change.

Judaism is a religion, open to others to convert to, and it has long seemed that religion and culture, not necessarily genetics, were the common elements of and between the world’s various Jewish communities. But in 2000 a team of geneticists led by Michael Hammer of the University of Arizona reported that men from many far flung Jewish communities have the same set of variations on their Y chromosomes. The variations are not exclusive to Jews but are common throughout the Middle East.³¹⁰ The finding meant that the founding fathers of Jewish communities around the world were drawn from the same ancestral Middle Eastern population of 4,000 years ago from which other peoples, such as Arabs, Turks and Armenians, are also descended. These generic Middle Eastern Y chromosomes, part of the J branch of the worldwide Y chromosome family tree, are both a common link between men of different Jewish communities and proof that their communities must have remained genetically separate from their non-Middle Eastern host populations.

But genetics points to a very different story with Jewish women. A team under David Goldstein of University College, London, surveyed Jewish communities of Germany and eastern Europe, known as Ashkenazi Jews, as well as those of Morocco, Iraq, Iran, Georgia, Bukhara, Yemen, Ethiopia and India. Unlike the case with the Y chromosome, they found that each Jewish community has its own pattern of mitochondrial DNA variations, evidence that Jewish women, unlike Jewish men, do not all come from the same ancestral population.

Mostly, the mitochondrial DNA in each Jewish community doesn’t closely resemble that of any other population, meaning that the geographic origin of the founding mothers of Jewish communities cannot be identified for certain. However, in several cases it looks as if it could come from the host community. For example, among the Bene Israel, the Jewish community of Bombay in India, the commonest pattern of mitochondrial DNA is just one mutation away from a pattern common among non-Jewish Indians.

The explanation proposed by Goldstein and his colleagues is that the founding fathers of Jewish communities came from the Middle East, the founding mothers from the host population in each country.³¹¹ The Jewish men, arriving perhaps as traders and presumably unmarried, took wives from the local population in each country, and it seems then converted their wives to Judaism. Once the community was established and reached sufficient size, it became closed; no more wives were taken from the host population, and community members married among themselves. With no fresh infusions from the local population, the mitochondrial DNA in each Jewish community fell under the influence of genetic drift, making it look less and less like that of the local version from which it originated.

If this explanation is correct, the members of a Jewish community are generally a genetic admixture between Middle Easterners (the founding fathers) and the host population of each country (the founding mothers). This could explain why Jews often resemble the people of their host country, yet also in some respects resemble one another.

The genetic findings are generally compatible with Jewish historical accounts, though not in every detail. The ancestral Jewish population is ancient but came from a mix of Middle Eastern men, DNA analysis indicates, not a single patriarch. Many Jewish communities have accounts or traditions of how they were founded, often to escape persecution or at the invitation of a friendly potentate. The Iraqi Jewish community (whose members now live mostly in Israel) is said to have been founded after the destruction of the first temple in 586 BC. The Bene Israel of Bombay say their ancestors fled to India to escape the persecution of Antiochus Epiphanus, who ruled from 175 to 163 BC. The DNA analysis in general confirms that Jewish communities are ancient, though it cannot place an exact date on their founding. But the circumstance it suggests for their origin, that of single Jewish men taking local wives, indicates that at least some Jewish communities probably began as trading outposts, not by the mass emigration of families.

The modern Jewish population falls into three main groups, based on ancestral place of origin. Ashkenazi Jews lived mostly in Germany and eastern Europe and, from at least the sixth century AD, spoke a common language, Yiddish; Sephardic Jews are those expelled from Spain and Portugal in AD 1492 during the Spanish Inquisition; and Oriental Jews are those who have always lived in the Near East. Of the 5.7 million Jews living in the United States, some 90% are of Ashkenazic origin; of the 4.7 million Jews in Israel, 47% are Ashkenazic, 30% Sephardic and 23% Oriental.³¹²

Jewish status, except for converts, is now defined by maternal descent. This practice, however, goes back only to Talmudic times, the period from around 200 BC to AD 500. In ancient Israel, tribal affiliation was determined by patrilineal descent, as were the two castes of hereditary priests, the cohens and the levites. After the destruction of the temple, the cohens were left with little to do and power passed into the hands of the rabbinate. The rabbis established matrilineal descent as the basis of Jewish identity. It is sometimes suggested they did so in wise appreciation of the fact that maternal descent is a fact and paternal descent only a probability; but a modern scholar, Shaye Cohen of Harvard University, believes rabbinic tradition and the influence of Roman law are likelier reasons.³¹³

The patrilineal priestly tradition still exists, and has afforded geneticists another deep insight into Jewish history. Cohens and levites continue to carry out ceremonial roles in certain congregations. Cohens are called first to the reading of the Torah in synagogue, and are asked on special occasions to bless the congregation. (The cohen’s blessing, signaled by holding up the hand with a split between the middle and the ring fingers, is familiar to many non-Jews; it was adapted by Leonard Nimoy, who remembered seeing it as a boy in synagogue, as the Vulcan greeting for his role as Spock in Star Trek.)³¹⁴

Oral tradition holds that all cohens, or cohanim, are descended from Aaron, the brother of Moses and the first high priest. The Jewish priesthood is thought to have been established some 3,300 years ago and to have passed from father to son ever since. This fact was on the mind of Karl Skorecki, a medical researcher at the Technion-Israel Institute of Technology in Haifa, one morning when he was sitting in synagogue and the Torah was being read. The cohen doing the first reading was a Sephardic Jew. Skorecki, whose family is Ashkenazic, himself comes from a line of cohanim. The thought occurred to him that though he and the Sephardi differed strongly in physical appearance, they must both have inherited the same Y chromosome from Aaron, if oral tradition was correct.³¹⁵

Skorecki called Michael Hammer, the University of Arizona geneticist, who agreed with his inference and set about analyzing the Y chromosomes of cohanim from both the Ashkenazic and Sephardic communities. Despite the millennium or so for which the two communities have been separate, and the geographical distance between them, Hammer and his colleagues found that the cohanim of both groups did indeed possess a distinctive genetic signature.

The signature is a set of DNA sequences at two specific sites on the Y chromosome. It is known as the cohen modal haplotype, a geneticist’s phrase meaning the set of DNA variations typical of cohens. The Hammer team detected the cohen modal haplotype in 45% of Ashkenazic cohanim and in 70% of Sephardic cohanim.³¹⁶ The finding substantially confirmed the oral tradition that cohanim are descended from a single individual. This person was presumably a founding high priest and could perhaps have been Aaron himself if indeed there was an Aaron; some modern scholars believe the great patriarchs of Israel may have been more a part of legend than of history.³¹⁷

To learn more about when the ancestor of all the cohanim might have lived, another team of geneticists including Skorecki and David Goldstein has looked at the variations that have developed on the cohen modal haplotype. The Goldstein team estimates that about 106 generations must have occurred to account for the observed amount of variation that has built up on the cohen modal haplotype. Assuming 30 years per generation, this means the ancestor of the cohanim lived some 3,180 years ago (or 2,650 years ago, if a generation time of 25 years is preferred).³¹⁸ A general date of about 3,000 years ago is of particular interest since it would place the first cohen at the beginning of First Temple Period of Jewish history.

The fact that only 50% or so of cohens, depending on the population, carry the cohen Y chromosome means that the rest must result from a discrepancy, at some point in their lineage, between the biological father and the father of record. Adoption cannot be invoked since the priesthood cannot be transferred to adopted sons, which leaves infidelity as the explanation. But as with the case of the English Sykeses, it takes only a small rate of nonpaternity in each generation to produce a large proportion of males with discrepant paternity many generations later.

Since the cohen lineage stretches back three times as far as that of the Sykeses, the fidelity of cohen wives must have been even higher. James Boster, an anthropologist at the University of Connecticut, calculates on the basis of the Skorecki team’s figures that the rate of nonpaternity was 1.2% per generation among Ashkenazic cohanim and 0.4% among Sephardic cohanim. (This estimate would of course not pick up any cases where a cohen’s wife had taken another cohen as her lover.)

Such infidelity rates are extremely low compared with the nonpaternity rates of 5% and more that are assumed typical of contemporary Western societies. Boster and his colleagues ask how cohanim through the ages secured such exemplary fidelity from their wives without resorting to the coercive measures used by men in other societies, such as purdah or chastity belts. They point to Jewish law and custom, under which intercourse is regarded as ritually impure from the beginning of a woman’s menstruation until seven days after its end, whereupon it is the husband’s duty to make love to her. Indeed he must do so immediately on her return from the ritual cleansing bath. This sage religious obligation has a strong consequence on the biological plane: it ensures that first intercourse, after several days abstinence, coincides with the three day period of peak fertility prior to ovulation. “This practice, coupled with extreme sanctions against adultery, . . . could account for these very high degrees of paternity certainty,” the researchers observe.³¹⁹

Levites, according to their genetics, have a more complicated story. Levites are a junior priesthood to the cohanim, with fewer duties and obligations. By tradition, levites consist of all male descendants of Levi who are not also cohanim. The exclusion arises because Levi, the third son of the patriarch Jacob, was also an ancestor of Aaron. About 4% of Jewish men are levites, the same proportion as are cohanim.

The Y chromosomes of Ashkenazic and Sephardic levites show no particular similarity. So, unlike the case with the cohanim, there is no identifiable male levite lineage that precedes the Ashkenazi-Sephardi split. There is, however, a strong genetic signature common to 52% of Ashkenazic levites. It is a set of genetic variations belonging to a branch of the world Y chromosome tree known as R1a1. To judge by the amount of variation on these levite R1a1 chromosomes, the original ancestor seems to have entered the Jewish community about 1,000 years ago, roughly the time when Jewish settlement in northwest Europe began, in other words at the founding of the Ashkenazic community.³²⁰

The geneticists who discovered the R1a1 signature among the levites, a team that included Skorecki, Hammer and Goldstein, note that outside the Jewish community the R1a1 chromosome is relatively common in the region north of Georgia, in the Caucasus, that was once occupied by the Khazar kingdom. The Khazars were a Turkic tribe whose king converted to Judaism in the eighth or ninth century AD.

The geneticists propose that one or more of the Khazar converts may have become levites, accounting for the R1a1 signature among today’s Ashkenazic levites. But Shaye Cohen, an expert on Jewish religious history, believes it unlikely that converts would become levites, let alone founding members of the levite community in Europe. The Khazar connection is “all hypothesis,” in his view.

The genetic findings about cohen and levite ancestry are just genetics; they have no bearing on who is or is not considered to be a cohen or a levite. “Genetics is not a reality under rabbinic law,” Cohen observes.³²¹

Ethnic origins and hereditary priesthoods have opened two windows on Jewish history; a third has been created by the study of genetic diseases. Every population has its own particular set of genetic diseases, but those of Jewish communities in the United States and Israel have come under particular medical scrutiny, which is one reason why so many have been documented. The diseases are known as Mendelian, because they are caused by a single mutation and inherited in an obvious pattern; this stands in contrast with the so-called complex diseases, like cancer or diabetes, which can be caused by many contributing genes and are not inherited in any clear pedigree.

So far at least 40 different Mendelian diseases have been detected in Jewish populations.³²² Some of these diseases occur in non-Jewish populations as well, some are common to several Jewish communities, and some are restricted just to the Jews of a single community. The diseases are of course studied so as to help the patients but incidentally they yield many interesting clues to population history.

A disease called familial Mediterranean fever is caused by an errant gene that occurs among Ashkenazi, Iraqi and Moroccan Jews. It is also found in Armenians, the Muslim Druze sect and Turks. All present versions of the gene seem to be descended from a single ancestor who must have lived about 4,000 years ago in the ancient Middle Eastern population from which Jews and other ethnic groups are descended.

Later, the Jewish religion was founded and its adherents developed their own genetic history as they started to marry among themselves. The Jewish population may have grown to about a million people before suffering a terrible decline in AD 70, the year of the destruction of the temple in Jerusalem by a Roman army. That event began the diaspora, the dispersal of Jewish populations around the Mediterranean world. The largest Jewish community, the Ashkenazim of central and eastern Europe, may have reached 150,000 or so people by AD 1095, the year of the first crusade and the beginning of the persecution of Jews by Christians.

The Ashkenazi Jewish population is of particular interest because it has produced many individuals of high intellectual achievement, both in Europe and among the Ashkenazim who fled to the United States and elsewhere in the wake of Nazi persecution. Another attribute is a distinctive set of Mendelian diseases. The mutations that cause these diseases can hit at random anywhere in the genome, so would not be expected to favor any particular category of gene. But no fewer than four of the Ashkenazic Mendelian diseases affect the cell’s management of chemicals known as sphingolipids, so called because their discoverer could not resolve the sphinxlike riddle of what they did. The four sphingolipid diseases are Tay-Sachs, Gaucher, Niemann-Pick and mucolipidosis type IV. Another cluster of four diseases affects the cell’s system for repairing DNA. These are the BRCA1 and BRCA2 mutations which can cause breast and ovarian cancer, Fanconi’s anemia Type C and Bloom syndrome.

The sphingolipid diseases in particular are reminiscent of the group of mutations that cause blood disorders like sickle cell anemia, and which are now recognized as defenses against malaria. When malaria suddenly became a threat some 5,000 years ago, natural selection favored any mutation that offered protection, even if it carried serious disadvantages. Diseases like sickle cell anemia are the result of that quick fix. The sickle cell mutation, though devastating for individuals unlucky enough to inherit a copy from each parent, offers substantial protection against malaria for the much larger number in the population who inherit just a single copy.

Evolution has probably engineered many quick fixes like this in the human genome. Later, as the generations pass and better mutations turn up, evolution is generally able to improve on the quick fix or favor variant genes that diminish the side effects of the first mutation. This is why a batch of harmful mutations affecting a common pathway is the fingerprint of a recent evolutionary response to some sudden selective pressure.

Turning back to the four sphingolipid diseases, they look awfully like an evolutionary quick fix, a set of mutations selected because of some advantage gained by disrupting sphingolipid metabolism. So if that advantage was protection against disease, what disease could it have been? The puzzle is that carriers of the sphingolipid mutations don’t seem to enjoy unusual immunity to any specific disease.

“A second hypothesis,” writes Jared Diamond, after discussing the idea that the variant genes conferred greater resistance to tuberculosis, “is selection in Jews for the intelligence putatively required to survive recurrent persecution, and also to make a living by commerce, because Jews were barred from the agricultural jobs available to the non-Jewish population.”³²³

The suggestion that one group of people may be genetically more intelligent than another is a sensitive subject, not least because it opens the door to the argument that if some groups are smarter, others may be less so. The idea Diamond floated was not followed up, and indeed the geneticists who next looked at the sphingolipid diseases suggested they had grown common not through natural selection but because of a quite different mechanism known as a founder effect.

If a population gets squeezed down to small numbers by some calamity, and then expands, its gene pool will be an amplified version of that of the few individuals who survived the disaster. If one of the survivors carried a generally rare mutation, the mutation will be much commoner in the new expanded population than it is in the general human population. The relatively high incidence of the usually rare mutation in the expanded population is called a founder effect, after the founder who carried the mutation.

Recently Neil Risch, now of the University of California, San Francisco, concluded that the four sphingolipid diseases must have become common among the Ashkenazi Jewish population because of founder effects. He noted that the four diseases had similar properties to the other Ashkenazic Mendelian diseases, such as having arisen very recently, in the last 1,100 years. Because all the Mendelian diseases seemed therefore to have arisen through the same cause, he argued, that cause must be founder effects, since natural selection wouldn’t favor such a miscellany of different mutations. ³²⁴

A similar conclusion was reached for different reasons by Montgomery Slatkin of the University of California, Berkeley.³²⁵ He calculated that if there had been two bottlenecks in the Jewish population, at AD 70 following the destruction of the temple, and at sometime after AD 1100, the founder effects caused by these two population reductions could explain how the Ashkenazic disease genes had gotten to be so common. Slatkin’s calculation did not rule out natural selection, but since a founder effect was possible, that provided the most economical explanation, in his view.

But a new and substantially buttressed case for natural selection, with the need for extra intelligence posited as the driving force, has now been advanced by Gregory Cochran, a physicist turned population geneticist, and Henry Harpending, an anthropologist at the University of Utah.³²⁶ They agree with Risch that all the diseases arose from the same cause and at about the same time. But the cause must have been natural selection, not founder effects, because in testing other, non-disease causing Ashkenazic genes, Cochran and Harpending could see no evidence for any of the reductions in population size required to cause Risch’s founder effects. Nor is there any clear historical evidence, they say, that the Ashkenazi Jewish population ever dwindled to the low numbers needed to generate a founder effect.

Having argued that natural selection must therefore have been the reason that the Ashkenazic mutations became so common, Cochran and Harpending next ruled out disease as the agent of selection. The Ashkenazic population, they note, lived in the same cities as their European hosts and suffered from the same diseases, yet Europeans show no similar pattern of mutations.

But there was a significant difference between Ashkenazim and Europeans, Cochran and Harpending argue, and it lay in the special range of occupations to which Ashkenazi Jews were restricted by their Christian hosts.

The origin of the Ashkenazi Jews is obscure but they were established in northern France by shortly after AD 900. Most had become moneylenders by AD 1100 because Christians forbade usury, and this continued for several centuries. Moneylending was an intellectually demanding profession, not least because the Indian numerals in use today, and specifically the concept of zero, did not become widespread in Europe until around 1500. Figuring out xvii percent of cccl, without the use of zero, is not a straightforward computation.

Jewish communities became subject to particular persecution after the First Crusade, launched in AD 1095. They were expelled from England in 1290, from France in 1394, and from various regions of Germany in the fifteenth century. Many migrated to Poland, where they lived first as moneylenders and then served as the managerial class for the Polish authorities, particularly in such roles as tax farming. (The tax farmer would pay a nobleman the tax due, then try to recoup the sum, with profit, from the peasantry.) Being frequently uprooted and forced to start over again, there was continual pressure on families to survive and find ways of being useful to their unpredictable hosts. “From roughly 800 AD to 1650 or 1700 AD, the great majority of the Ashkenazi Jews had managerial and financial jobs, jobs of high complexity, and were neither farmers nor craftsmen. In this they differed from all other settled peoples of which we have knowledge,” Cochran and colleagues write.

Restrictions on Ashkenazi employment were lifted around 1700, bringing to an end a period of some 900 years during which most of the population would have had to earn a living in occupations requiring more mental ability than most. Given what is known about the heritability of intelligence, the Cochran team calculates that even in as little as 500 years there would have been time for the intelligence of the Ashkenazi population to have been raised appreciably.

The authors cite evidence suggesting that sphingolipid mutations serve to foster the growth and interconnectedness of neurons, sometimes by lifting natural restraints. They believe that all the Ashkenazic disease mutations, in ways that remain to be discovered, serve to promote the extra cognitive skills that the Ashkenazic population needed in order to survive.

The outcome, they say, is that Ashkenazim have an average IQ of 115, one standard deviation above that of northern Europeans, although some measurements put it at only half a standard deviation higher. This is the highest average IQ of any ethnic group for which reliable data exist. Such an advantage may not make much difference at the average, where most people are situated, but it translates into a significant difference at the extremes. The proportion of northern Europeans with IQs greater than 140 is 4 per thousand but the figure for Ashkenazim is 23 per thousand, a sixfold difference.

This may have something to do with the fact that Ashkenazim make up only 3% of the U.S. population but have won 27% of U.S. Nobel prizes. Ashkenazim account for more than half of world chess champions. “Jews and half Jews, who make up about 0.2 percent of the world’s population, have won a total of 155 Nobel prizes in all fields, 117 in physics, chemistry and medicine,” writes the anthropologist Melvin Konner.³²⁷

Jewish folklore holds that intelligence was fostered not by occupation but by channeling the cleverest children to become rabbis. The rabbis were able to have more children, the folklore explanation holds, because they were sought as husbands for the daughters of wealthy families. “Talmudic academies served as systems of selection,” writes Konner. “Whatever we think of what was studied, the process culled the best minds in every generation of Jews for more than a thousand years. Rising stars among these bright young men would board with successful merchants, and matches would be made between them and the merchants’ daughters.”

But the Cochran team gives short shrift to this explanation, saying there were not enough rabbis—only 1% of the population—to make a genetically significant difference. As further proof of their thesis, they cite the fact that the two other main branches of the Jewish community, Oriental Jews and Sephardim, lived mostly under Muslim rulers who often forced them into menial jobs, not the intellect-demanding ones imposed on Ashkenazim. Oriental Jews and Sephardim score similarly to northern Europeans with no elevation in IQ, as would be predicted under the Cochran team’s thesis.

Among Ashkenazim, some 15% carry one of the sphingolipid or DNA repair mutations, and up to 60% carry one or other of all the disease mutations. (Most of these diseases are only harmful if a mutated gene is inherited from both parents, and some others are not fully “penetrant,” a geneticist’s term meaning a person can carry the mutation but doesn’t necessarily have the disease.)

In summary, the Cochran group has taken two well accepted phenomena—the odd pattern of Ashkenazic Mendelian diseases and the notable intellectual achievement of Ashkenazim—and has attempted to establish a link between them. The argument is necessarily extended, but is carefully developed at each stage. “It’s certainly a thorough and well argued paper, not one that can easily be dismissed outright,” said Steven Pinker, a cognitive scientist at Harvard. Though several aspects of the argument cross disputed academic territory—it assumes that intelligence is heritable and that IQ scores are a reliable measure of it—it has the virtue of making a clear and testable prediction: that people carrying one of the Ashkenazic mutations should do better than average on IQ tests. As of this writing, the test has not been conducted.

Despite the existence of genetic diseases that can be called Jewish, in the broader context Jews are doubtless highly similar to other populations in the west Eurasian or Caucasian branch of the human family. Their genesis as a distinctive group resembles that of Icelanders. Just as Jews appear to be a mixture of Middle Easterners with various European or other Middle Eastern populations, Icelanders also are probably a mix of two Caucasian populations, Norwegians and Celts. Both Jews and Icelanders have practiced endogamy, the necessary step for keeping one’s gene pool to oneself, Jews for religious reasons, Icelanders for geographic ones. Icelanders have been genetically separate for 1,000 years, but are still so similar to other Europeans that they can serve as a test bed for discovering European disease genes; Jews have been separate for just 2,000 years longer.

DNA and the Secret Family of Thomas Jefferson

Leading American historians for years denied a startling circumstance that was clearly attested to in the historical record: Thomas Jefferson, the third president of the United States, fathered an unacknowledged family with his slave mistress Sally Hemings.

Here is some of the evidence that historians of Jefferson found reason to disbelieve:

“It is well known,” the journalist James T. Callender wrote in the Rich mond Recorder on September 1, 1802, the second year of Jefferson’s first presidency, “that the man, whom it delighteth the people to honor, keeps, and for many years past has kept, as his concubine, one of his own slaves. Her name is SALLY. The name of her eldest son is TOM. His features are said to bear a striking although sable resemblance to those of the president himself.”

In 1873 a son of Sally Hemings, Madison Hemings, gave a long biographical statement to an Ohio newspaper, the Pike County Republican. He told how his mother, then aged around 13, had been sent to Paris, where Jefferson, then a widower, was American ambassador. Sally’s role was to be a servant to Maria, one of Jefferson’s two daughters.

“Their stay (my mother and Maria’s) was about eighteen months,” Madison Hemings related. “But during that time my mother became Mr. Jefferson’s concubine, and when he was called back home she was enceinte by him. He desired to bring my mother back to Virginia with him but she demurred. She was just beginning to understand the French language well, and in France she was free, while if she returned to Virginia she would be re-enslaved. So she refused to return with him. To induce her to do so he promised her extraordinary privileges, and made a solemn pledge that her children should be freed at the age of twenty-one years. In consequence of this promise, on which she implicitly relied, she returned with him to Virginia.

“Soon after their arrival, she gave birth to a child, of whom Thomas Jefferson was the father. It lived but a short time. She gave birth to four others, and Jefferson was the father of them all. Their names were Beverly, Harriet, Madison (myself) and Eston—three sons and one daughter. We all became free, agreeably to the treaty entered into by our parents before we were born.”

It is difficult to believe that a 68-year-old Ohio carpenter, as Madison Hemings then was, would be moved on chance encounter with a journalist to invent an account of such specificity and poignancy. It contained many details that could be independently checked. Jefferson freed very few slaves, but he let all of Sally’s children go free. Winthrop Jordan, a historian at the University of Mississippi, documented in 1968 that Jefferson, despite his many absences from Monticello, was present at the time of conception of all Hemings’s known children.

But apart from Jordan, who stated that a liaison between Jefferson and Hemings was a possibility, a long line of Jefferson historians dismissed Madison Hemings’s account. Merrill Peterson, the first historian to give it scholarly study, conceded that Madison’s recollection “checks remarkably well with the data accumulated by scholars on Jefferson’s domestic life and the Monticello slaves.” But he chose to reject its central claim, that Madison was Jefferson’s son, with the defective argument that since Jefferson’s enemies wanted the story to be true, it must be false. The Jefferson-Hemings liaison was a legend, he wrote, sustained by the hatred of the Federalists, the propaganda of the British, “the Negroes’ pathetic wish for a little pride,” and the cunning of slave auctioneers thinking they would “get a better price for a Jefferson than for a Jones.” The “overwhelming evidence of Jefferson’s domestic life refuted the legend,” Peterson assured his readers in 1960.³²⁸

A third of a century later, the historian Joseph J. Ellis took the same position. He derided the story as a “piece of scandalous gossip” that had affixed itself to Jefferson’s reputation “like a tin can that then rattled through the pages of history.” Jefferson historians had no desire to know what might be in the tin can; they just wanted to boot it as far away as possible. “Within the community of Jefferson specialists, there seems to be a clear consensus that the story is almost certainly not true,” Ellis wrote in 1996. “After five years mulling over the huge cache of evidence that does exist on the thought and character of the historical Jefferson, I have concluded that the likelihood of a liaison with Sally Hemings is remote.”³²⁹

The community of Jefferson specialists found much more to their taste a self-serving story concocted by the Jefferson family to protect his reputation. Two of Jefferson’s grandchildren put it about that Jefferson’s nephews Peter and Samuel Carr were the fathers of the light-skinned slaves at Monticello. The Carrs were the sons of Jefferson’s sister, which could explain why the young slaves so resembled Jefferson.

There the matter rested, so far as scholars were concerned, and might well have solidified into accepted fact, but for the trespass of two outsiders onto the historians’ carefully groomed turf. An African American lawyer, Annette Gordon-Reed, weighed the same evidence available to the historians but came to the opposite conclusion. A Jefferson-Hemings liaison was very likely, she argued, though it could not be proved. That finding led her to a harsher, but not so unreasonable, judgment that “those who are considered Jefferson scholars have never made a serious and objective attempt to get the truth of this matter.”³³⁰ Gordon-Reed had no genetic evidence available to her; she simply interpreted the available historical evidence more skillfully than a generation of professional historians had done.

The second outsider to the issue was Eugene Foster, a pathologist who had recently retired from Tufts University to Charlottesville, Virginia. Foster had no particular interest in Jefferson, but Charlottesville is Jefferson country, and a friend asked him one day if DNA fingerprinting might shed any light on the Hemings issue. Foster decided it wouldn’t—forensic DNA analysis can identify individuals and resolve paternity but can’t reach back up a genealogical tree because of the shuffling of DNA between generations. Then he learned of work on the Y chromosome, which is passed unchanged from father to son except at its very tips, and realized it could hold the answer.

First Foster needed a sample of Jefferson’s Y chromosome. Unfortunately Jefferson had no male descendants, but his paternal uncle Field Jefferson would have carried the same Y chromosome, assuming no illegitimacy in the Jefferson male line. With the help of Herbert Barger, a Jefferson family historian, Foster located 5 male descendants of Field Jefferson and wrote explaining his project and asking for a sample of their blood.

Next, he needed a Y chromosome of the Carr brothers, the leading suspects in the view of historians and the Jefferson family members. He obtained blood from three male-line descendants of the three sons of John Carr, the grandfather of Peter and Samuel.

FIGURE 11.1. THOMAS JEFFERSON’S FAMILY WITH SALLY HEMINGS.

A Y chromosome analysis of Thomas Jefferson’s family performed by Eugene Foster and Chris Tyler-Smith showed that Eston Hemings, a son of the slave Sally Hemings, carried the same Y chromosome as that of Thomas Jefferson’s male relatives and was therefore highly likely to have been Jefferson’s son.

At specific sites on the chromosomes, short DNA sequences are repeated a number of times, as in ATATAT. The number of repeats changes quite often between generations, so can be used to identify different lineages. In this case the repeats at 11 sites have been used to fingerprint the different Y’s. A natural shift in repeat numbers at the 9th site has occurred in the rightmost of Field Jefferson’s descendants

Eston Hemings had the same 11 repeats as the Field Jefferson descendants, so would have acquired his Y chromosome from a Jefferson family member, who from the historical evidence was almost certainly Thomas Jefferson, the third president of the United States. The finding strongly supports contemporary rumors that Jefferson had fathered a secret family with Sally Hemings, his slave mistress.

Of Sally Hemings’s descendants, Foster collected blood from a male-line descendant of Eston, her youngest son, who was born in 1808 and is said to have borne a striking resemblance to Thomas Jefferson.

In addition he took samples from 5 male line descendants of Thomas Woodson. There is a strong oral tradition among long-separated branches of the Woodson family that Thomas was a son of Thomas Jefferson who was sent away from Monticello as a boy. There is one other reference, besides James Callender’s, to a slave son of Jefferson named Tom. But there is no documentary evidence showing Thomas Woodson’s presence at Monticello, nor is he named by Madison Hemings in his list of Sally’s children.

Foster had his blood samples analyzed in the laboratory of Chris Tyler-Smith, the Y chromosome expert at Oxford University, with the following results:

• All 5 male-line descendants of Field Jefferson turned out to carry the same distinctive set of markings on their Y chromosome, making it highly probable that the same Y chromosome was carried by the third president.

• All 5 male descendants of Thomas Woodson carried non-Jeffersonian Y chromosomes, ruling out the idea that Jefferson was Thomas Woodson’s father.

• All three male-line Carr descendants carried the same Y chromosome, proving this was the true Carr family Y chromosome.

• The Y chromosome of Eston Hemings’s male-line descendant was a perfect match to the Jefferson family Y chromosome, and differed from that of the Carrs.

The “simplest and most probable explanations for our molecular findings,” Foster and his colleagues wrote, “are that Thomas Jefferson, rather than one of the Carr brothers, was the father of Eston Hemings Jefferson, and that Thomas Woodson was not Thomas Jefferson’s son.” They could not rule out the possibility that some male Jefferson other than Thomas had fathered Eston, they wrote. “But in the absence of historical evidence to support such possibilities, we consider them to be unlikely.”³³¹

The DNA evidence by itself does not prove conclusively that Thomas Jefferson had an unacknowledged family with Sally Hemings. Nor does the historical evidence by itself. But for two entirely independent kinds of evidence to point so strongly to the same conclusion makes a robust case. The historian Joseph Ellis certainly felt so, and to his credit admitted error. “The new evidence persuaded me that I had been wrong, and I felt a kind of moral and professional obligation to say that,” he said.³³²

Barger, the historian who helped Foster, was not pleased by this outcome. He has assailed Foster’s findings and proposed other male Jeffersons as the father of Eston, just as the Jefferson grandchildren did on an earlier occasion. But none of these ad hoc candidates can be shown to have been present at Monticello at all of Sally Hemings’s conceptions as Thomas Jefferson was.

Jefferson had a strange and special tie with Sally Hemings, one only possible in the divided world of slave and free. No portraits of Sally survive, but she may well have reminded Jefferson of his beloved wife Martha, being as she was Martha’s half sister. Both were daughters of John Wayles, Martha by his wife Martha Eppes, Sally by the slave Elizabeth Hemings, who became Wayles’s mistress after his wife’s death.

Jefferson’s feelings for Sally, a subject of much speculation, are simply unknown. In all his correspondence he mentions her just once. At Monticello his white family and his unacknowledged black family lived side by side, but even in private he seems to have paid no special attention to Sally’s children. Madison learned to read, he says, “by inducing the white children to teach me the letters and something more.” As for Jefferson, “He was not in the habit of showing partiality or fatherly affection to us children. We were the only children of his by a slave woman.”

Some mysteries lie beyond the power even of DNA to resolve.