“Junk Science”: The problematic Jefferson-Hemings “conception window” statistical analysis

“Junk Science”: The problematic Jefferson-Hemings “conception window” statistical analysis

“Junk Science”: The problematic Jefferson-Hemings “conception window” statistical analysis

“Junk Science”: The problematic Jefferson-Hemings “conception window” statistical analysis

Without objective scholarship comes the risk of fiction. Hence, history becomes an illusion.

– Cynthia Burton

Only in fiction do we find that the loose ends are neatly tied. Real life is not all that tidy.

– Walter Cronkite

After the infamous Jefferson-Hemings DNA study in November of 1998, Monticello set out to conduct its own study:

On December 21, 1998, Dr. Daniel P. Jordan, president of the Thomas Jefferson Foundation, appointed a research committee of Monticello staff members, including four Ph.D.’s and one medical doctor, and charged the committee with evaluating the DNA study of Dr. Eugene Foster and associates, assessing it within the context of all other relevant historical and scientific evidence, and recommending the impact it should have on historical interpretation at Monticello.

Former surgeon and volunteer Monticello guide Dr. Ken Wallenborn was a Monticello Research Committee member who noted what he saw to be a biased exercise in searching for any form of damning evidence to prove Jefferson fathered Sally Hemings’ children:

When the committee was assembling for one of its meetings in February 1999, the head of the Archaeology Department at Monticello [Fraser Neiman] dropped a packet of papers on the table next to me and said (and this is exactly how another member of the committee and I recollect it): “I’ve got him!” He repeated this statement again and then explained his ‘Monte Carlo Simulation.’ This just seemed to be an inappropriately enthusiastic remark for someone who is working at Thomas Jefferson’s home.

Fraser Neiman’s statistical study, or “Monte Carlo Simulation”, analyzing Jefferson’s residence at Monticello when Hemings’ children were conceived, was dubbed by the USA Today as the “most provocative finding” of the Monticello Research Committee’s report. However, the study was published in a non-scientific magazine with no official peer review. When scientists and statisticians analyzed the study they criticized it for its core mistakes and bias. Professor of Law and History David N. Mayer pointed out the fundamental flaws in the way the conception window analysis was applied:

The Monte Carlo approach estimates the probability of a given outcome by comparing it to a very large number of random outcomes generated by a simulation model. Neiman’s study rested on two unsupported postulates: that there could only be a single father for all of Sally Hemings’ children, and that rival candidates to Thomas Jefferson would have to had to arrive and depart on the exact same days he did. Here, the assumption of random behavior makes little sense, because the visits to Monticello of the other candidates for paternity—Jefferson’s friends and relatives (including his brother Randolph, Randolph Jefferson’s [five] sons, and the Carr brothers)—were not random occurrences; they certainly would have been far more likely to occur after Jefferson’s return to Monticello from extended absences in Washington or elsewhere. The final impression one gets of the Niemen study is of a simulation whose parameters were deliberately set to “get” Thomas Jefferson as the father of Sally Hemings’ children. (Scholars Commission p.302)

Dr. M. Andrew Holowchak described Neiman’s process as a fait accompli:

Neiman has gotten the results he had desired to get, simply because he has plugged in exactly the sort of data that would assure him of those results. Change the data and you change the results.

Thirteen eminent scholars reviewed the DNA study and the Monticello Research Committee Report over the course of a year in 2000-2001.The goal was as Thomas Jefferson described, “to follow the truth wherever it may lead”, and in the process demonstrate proper historical and scientific scholarship. The “Scholar’s Commission” examined Neiman’s statistical study and came to a more likely lower probability of Thomas Jefferson paternity:

Even without considering Thomas Jefferson’s advanced age (sixty-four) and health, if the question is changed from trying to place a single suspect at Monticello nine months prior the birth of all of Sally’s children to simply trying to identify the Jefferson men who were likely to have been in the Monticello area when Eston Hemings was conceived, the statistical case for Thomas Jefferson’s paternity of Eston, based on DNA evidence alone, falls below fifteen percent. (Scholars Commission p.10)

Dr. Wallenborn also described how Neiman was informed of basic editorial mistakes that ultimately went uncorrected:

When his article was listed in Appendix I of the TJMF Research Committee Report and simultaneously published in the William and Mary Quarterly in January 2000, it contained a serious and glaring error that had been pointed out to him. This error was his statement that the “molecular geneticists found the Jefferson Y-haplotype in recognized male-line descendants of Thomas Jefferson”! He should have said descendants of Field Jefferson, Thomas Jefferson’s uncle. Why the TJMF allowed this significant error to be published in their report and in the W&M Quarterly remains unanswered. Future historical researchers will possibly quote this erroneous statement and think that the DNA sample came directly from Thomas Jefferson’s direct descendants and that this cinches the case in the Sally Hemings paternity story.

Neiman’s statistical analysis makes a number of reaching assumptions in order to build a statistical case, but without those assumptions, the entire statistical model falls apart. For example, Neiman refuses to acknowledge a number of key possibilities: that another Jefferson male could have impregnated Sally, that Sally might not have been monogamous, and that either Sally or Jefferson might not be present at Monticello during her conceptions. An honest statistical model cannot be computed without factoring in such alternative hypotheses. For example:

  • Sally may have conceived elsewhere. There are no comparable records establishing Sally Hemings’ whereabouts during the same period. Monticello’s own tours describe how slaves were sometimes given a “slave pass” which allowed them to travel off the mountain. Slaves such as Sally’s sister Critta were also sent out to work at other farms, and evidence indicates that Sally may have been absent at Monticello at times. There is simply no record of Sally’s comings and goings. (Jefferson Vindicated, p. 106)

  • Sally may not have been monogamous. Unmarried women who have multiple children may have them by multiple fathers. Sally’s own mother and two of her sisters each had multiple children by multiple fathers. For example, Sally’s older half-sister Bett was the mother of eight children by multiple men.[1] Madison Hemings said Sally’s mother Betty Hemings “had seven children by white men and seven by colored men — fourteen in all.”[2] As former slave Henry Bibb wrote in 1849: “It is almost impossible for slaves to give a correct account of their male parentage….”[3] This was an unfortunate reality at the time, and it is entirely possible if not likely that Sally could have fit the pattern. Historian Eyler Coates: “But the Monticello report states, in effect, that because Sally’s patterns of conception match Jefferson’s presence at Monticello it should be taken as evidence that all her children were by the same father, i.e., Thomas Jefferson. In this way, the committee uses one assumption to support another.”[4]
  • It is impossible to assume all the dates of births were accurate. The dates utilized for the study were from Jefferson’s own hand-written farm book, which he loosely maintained when he was away as President and often made entries months later and sometimes years afterward. For example, when he returned to Monticello after a long absence, he asked others to inform him of their recollection of the dates of any slave births that occurred while he was gone.
  • Another Jefferson male relative could have been the father, particularly of Eston Hemings. Dr. David Murray, Director of the Statistical Assessment Service describes, “[T]he act of appearing at Monticello should not be viewed as itself a causal procreative act. The Report’s mathematical model is likewise incapable of ruling out the prospect that Thomas Jefferson’s visits to Monticello co-occurred with some other event, such as a visit by nearby brother Randolph (or any other of the crowd who traveled with the president), who comes to see his brother just when his brother is there. This is not idle speculation, since a record was found at Monticello showing an invitation to Randolph to visit Thomas Jefferson exactly coincident with [the conception window for Eston Hemings]. But because no one found a record of Randolph’s actual arrival, the Report declines to pursue the Randolph connection…. It is equally possible to declare that [other Jefferson males] were present but unrecorded as it is to claim that no other Jefferson male was ever-present during those time periods.”[5]

In fact, the biggest oversight of the theory and statistical study is the refusal to acknowledge any other Jefferson male as a paternal candidate. Sadly, many historians — and even “Jefferson scholars” — have not taken the time to study the lives of Jefferson’s male relatives, let alone as candidates for paternity.[6]

Dr. Wallenborn said that the committee as a whole seemed to be actively trying to avoid an honest consideration of other candidates as if a pre-determined result was being sought.[7] This mindset drove the statistical study, which historian Eyler Coates described:

It begins to appear obvious that the model was set up to produce the desired results, and was not realistically designed to take into consideration other possible explanations. It was designed to indict Jefferson, and it should come as no surprise that the conclusion confirms such a high probability as was “discovered” by such a ruse.

Dr. Murray also noticed a pattern of bias that managed even to make its way into the statistical equation:

It is nevertheless instructive to see how the [Monticello] Report handles the absence of evidence in other circumstances. We don’t know where Sally Hemings was at the time of conception. Of this, the Report can only say, “There is no documentary evidence suggesting that Sally Hemings was away from Monticello when Thomas Jefferson was present.” That is, we see a double standard. When there is no documentary evidence that brother Randolph was there during a conception date, the Report concludes that therefore he was not…. By the Report’s insistence, Sally Hemings is there unless proven otherwise, while Randolph is not there unless proven that he is.

Similarly, Dr. Murray noted how the Monticello Report avoided recognizing Tom Woodson as the possible first son of Sally Hemings (Woodson descendants had no DNA match with Jefferson DNA), since it would disprove the Report’s insistence that Sally was monogamous:

The same maneuver is applied to Tom Woodson, who is dismissed as Sally’s child because there is no documentation of his birth to her, even though oral history links him to Sally quite strongly.

Tucked away in one of the footnotes to the statistical analysis is Neiman’s unsupported reasoning as to why there can be no alternative male candidates:

Because the model outcomes are tabulated against Jefferson’s arrival and departure dates, the probabilities that result apply to Jefferson or any other individual with identical arrival and departure dates. The chances that such a Jefferson doppelganger existed are, to say the least, remote.[8]

Yet Neiman never describes how he determines the “remoteness” of such an individual, and never in the study is it considered for computation. Strangely, Neiman’s “remoteness” seems to be a function of the assumption that either Jefferson sired all six children or the doppelganger is some one who likewise fathered all six children. Why should it be assumed that there is one and only one person who must tread in Jefferson’s exact chronology? Dr. Murray addresses the assertion and describes how the study should have been constructed:

That demand seems to follow from nothing more than the previously formed assumption that Sally was not promiscuous. From a probability point of view (and especially given the awkwardness of Tom Woodson’s [non-Jefferson] paternity), there could have been doppelganger sub1, who matches Jefferson’s itinerary during only one stay, and who is responsible for only one child, and then doppelganger sub2, who matches Jefferson’s itinerary on another occasion, and fathers another Hemings child, and so forth. Maybe six doppelgangers, maybe only two or three; who knows? Moreover, maybe it’s someone already at Monticello throughout the period continuously, who therefore doesn’t have to “match” Jefferson’s comings and goings, but is just opportunistically “there” and takes advantage during circumstances that present themselves when Jefferson is coincidentally home. And so forth. I am not arguing that we have evidence of these scenarios. But it would seem that those likelihoods must at least be specific and allowed to function in the model as alternative hypotheses, and not just dismissed. As it stands, the argument sticks with only the most unlikely character, the one perfect Jefferson doppelganger – who presumably also wrote a parallel Declaration of Independence – who is offered as the [only] logical alternative.

Dr. Murray goes on to characterize Neiman’s simplistic yet faulty logic:

Neiman’s argument appears to be something like this: “If Thomas Jefferson fathered children, then Thomas Jefferson must have been present when those children were fathered. Children were fathered at Monticello. Thomas Jefferson was present at Monticello when children were fathered. Therefore, Thomas Jefferson fathered those children.”

Regardless, Neiman asserts with full certainty that the statistical study proves beyond a doubt that Jefferson is the father of all of Sally Hemings’ children. Neiman declares:

…the chance is just 1% that [Jefferson’s] presence was a coincidence…. How likely is it that this could have occurred by chance if Jefferson was not the father?

But Dr. Murray proposes a simple analogy to better illustrate the scenario:

Imagine vases were broken on six different occasions at Monticello. No vases were known to have been broken when Thomas Jefferson was away. Thomas Jefferson was home when all six vases were broken. There is evidence that some Jefferson [male] likely broke one of the vases. There is no evidence that Thomas Jefferson or indeed any Jefferson broke any other vases. Are we willing, therefore, to subscribe to the conclusion that there is a 99% probability that Thomas Jefferson broke all six vases? Mercifully, courts of law are not likely to do so, there being no necessary connection between being present and the act of surreptitiously shattering glass.

When speaking to the press, Neiman asserted not only confidence about his statistical analysis, but also its own historical significance:

“The DNA evidence applies to only one child. This shows he, in all likelihood, fathered all six. Serious doubts about [Thomas Jefferson’s] paternity of all six children cannot reasonably be sustained. This statistical analysis is more powerful… than the genetic finding.”[9]

Rather than publish his statistical analysis in a scientific journal and submit it to peer review, Neiman solicited a cursory review by two demographers he personally knew and published the study in the William and Mary Quarterly, a humanities journal of early American history and culture.

Similar to the way Dr. Foster published his DNA “historical analysis” in a science journal with no peer review, Neiman published his “statistical analysis” in a history journal with no peer review.

Steven Corneliussen, a science writer at the Thomas Jefferson National Laboratory, felt the need to understand the findings since the national lab he worked for carried Jefferson’s name. After he examined both Foster’s DNA study and Neiman’s statistical analysis, he was shocked and labeled them both “science abuse.”

According to Corneliussen, scientific participants in the controversy – who were not scholars qualified to put the results into historical perspective – abused science’s special authority, leading to misreporting that hobbled public understanding. The flawed statistical analysis only made things worse, and Corneliussen described them both as “intellectual disrespect: abuse of the special authority of science.”[10]

Whether or not Hemings and Jefferson had children together, misreported DNA and misused statistics have skewed the paternity debate, discrediting science itself.

Neiman made fundamental errors that statisticians recognized immediately. Neiman had merely established an exact calendar date of 267 days prior to birth for the conception date of each of Sally Hemings’ children. However, Neiman had overlooked the simple reality any parent knows: births rarely ever occur precisely 267 days after conception, the presumed average gestation length.

In reality, conception windows are a bell curve that spreads over the period of as much as five weeks, which should include at least three weeks before and two weeks after the presumed conception date.[11] The US National Institute of Environmental Health Sciences states that “only 4% of women deliver when predicted and only 70% within 10 days of their estimated due date.”[12] That means that one in 25 women give birth on their due date. Neiman failed to account probabilistically for variability within the four-to-five weeks conception window, which throws off the computations of the entire statistical analysis, particularly when Jefferson was only present at Monticello briefly around the time of conception date. The Scholars Commission noted how Jefferson’s absence from Monticello reduces the statistical likelihood significantly:

It is also inaccurate to say that Thomas Jefferson was at Monticello nine months before each of Sally’s children was born… Using Dr. Nieman’s figures, for example, we find an estimate that Beverly Hemings was conceived on July 8, 1797 (a nine-month gestation would have begun on July 1), and that Thomas Jefferson had not been at Monticello since May 5 and did not return until July 11. While it is certainly possible that Sally did not become pregnant until after the estimated conception date, it seems somewhat more likely statistically that she conceived prior to Jefferson’s return. Roughly ninety percent of mothers give birth within two weeks of their estimated due date, permitting us to identify a four-week conception window during which Beverly was likely conceived. For more than sixty percent of this conception window, Thomas Jefferson was not present at Monticello…. if, based solely on his visitation patterns, the odds are that Jefferson was not the father of Beverly Hemings, it follows ipso facto that there is less than a fifty-fifty chance that he was the father of all of Sally Hemings’ children.[13]

Professor Forrest McDonald, a staunch Hamiltonian and critic of Jefferson, was a member of the Scholars Commission and took issue with Nieman’s calculations. He “observed both that the calculations ignored the fact that 1808 was a leap year; and, far more importantly, ignored the fact that Thomas Jefferson was away from Monticello for as much as nine days overlapping Sally’s probable conception window.”[14]

Corneliussen described how the editors of the humanities journal William and Mary Quarterly overstepped their role by publishing a scientific study outside both their mandate and capacity as historians:

In my view, when the William and Mary Quarterly made itself a venue for science, it became obligated to require science’s common practices [such as technical review]… Among the study’s deficiencies is one problem so fundamental that, just by itself, it cancels any possibility of the study contributing usefully. By failing to apply an obviously necessary biostatistical technique, Neiman failed to account for the distinct statistical chance—in one case, greater than fifty percent chance—that at the time of conception in four of the six cases Neiman designated for study, Jefferson could actually have been absent from Monticello.

Expert genealogist Dr. Edwin M. Knights criticized the Jefferson paternity claims, describing how near certainty is needed to ascribe paternity:

Paternity studies, which include both genetic and non-genetic evidence, calculate a statistical probability of paternity. The non-genetic evidence, known as prior probability, is combined with the results from studying genetic loci from one or more alleged fathers. In most cases, a probability of paternity requires a minimum standard value of 99 percent.[15]

Yet Neiman’s statistical analysis — lacking any thorough technical review for errors — became one of the three main pillars to assert Jefferson’s paternity of all of Sally Hemings’ children, joining Foster’s 1998 DNA study and the overarching historical revisionism.

Dr. Robert F. Turner, chair of the Scholars Commission, summed up the expert assessment of Neiman’s conception window analysis:

Candidly, to borrow a term used by several of my colleagues during our Dulles sessions of the Scholars Commission, Neiman’s “Monte Carlo” study struck me as being “junk science” long before I had become involved in a technical discussion with scientists who confirmed its fatal deficiencies. Another term that was used during the Dulles meetings was “GIGO” —computerese for “Garbage In, Garbage Out,” or “if your input is not reliable, your output from the computer will be no better.” As our report reflects, none of us was impressed by it.[16]



[1]  Cynthia H. Burton, forward by James A. Bear, Jr., Jefferson Vindicated, p. 107, 109

[2] Monticello Report, Appendix E at 21. See also, JEFFERSON AT MONTICELLO 26 n.1.

[3] Lucia Stanton, SLAVERY AT MONTICELLO 21 (1996).

[4] Eyler Robert Coates, Sr. The Jefferson-Hemings Myth: An American Travesty, p.97.

[5] David Murray, Ph.D.,  The Jefferson-Hemings Myth: An American Travesty, p.120-1.

[6] Annette Gordon-Reed’s book Thomas Jefferson and Sally Hemings: An American Controversy does not once mention Randolph Jefferson or his sons, as if they didn’t exist, since their candidacy for paternity would put into question her entire thesis. Joseph Ellis, author of American Sphinx: The Character of Thomas Jefferson and who also wrote the accompanying essay to Foster’s DNA article in Nature, in a phone call with Herb Barger revealed that he was unaware that Jefferson even had a brother.

[7] Ken Wallenborn, The Jefferson-Heming Myth – An American Travesty, p.55-68.

[8] Fraser D Neiman, Coincidence or causal connection? The relationship between Thomas Jefferson’s visits to Monticello and Sally Hemings’s conceptions. William and Mary Quarterly, January 2000

[9] USA Today, Jan. 27, 2000

[10] Corneliussen, Steven. Sally Hemings, Thomas Jefferson, and the Authority of Science.

[11]  Sarah Kliff,  Doctors get due dates wrong 96.6 percent of the time, Vox, Jul 19, 2014

[12] Pregnancy length ‘varies naturally by up to five weeks’, BBC, 7 August 2013

[13] Robert F. Turner, The Jefferson-Hemings Controversy: The Report of the Scholars Commission, p.126-7.

[14] Robert F. Turner, The Jefferson-Hemings Controversy: The Report of the Scholars Commission, p.130.

[15] Edwin M. Knights, M.D., Genealogy and Genetics: Marital Bliss or Shotgun Wedding? Family Chronicle, March/April 2003

[16] Robert F. Turner, The Jefferson-Hemings Controversy: The Report of the Scholars Commission, p.126.

How Sally Hemings arrived in Paris

How Sally Hemings arrived in Paris

How Sally Hemings arrived in Paris

— SALLY WAS NOT THE SLAVE Jefferson requested


In 1782, Jefferson’s wife Martha died, leaving behind three surviving daughters. His wife’s death devastated Jefferson, who swore on her deathbed that he would never remarry. Around this same time, Jefferson was elected by Congress to travel to serve as a Minister to France, but he declined.

Two years after later, Jefferson was again elected by Congress in 1784 to serve as Minister to France “for the purpose of negotiating treaties of commerce.” This time he obeyed the request and in 1785 moved to Paris, accompanied by his oldest daughter Patsy (Martha), along with James Hemings (Sally’s older brother), who would serve as chef and learn the art of French cooking. Contrary to popular belief, Jefferson did not bring Sally Hemings along with him to Paris, nor did he later request Sally to join them in Paris. Rather, he requested as a last resort an older and more mature slave named Isabel to accompany his youngest daughter Polly (aka Maria/Mary) to Paris.

Not long after arriving in France, Jefferson was informed of the death of his youngest daughter Lucy back in Virginia. This was devastating to Jefferson, since Lucy was the fourth child Jefferson had lost and only three years after the immobilizing loss of his wife. Stricken with grief, he sent a request for his daughter Polly to travel to France to join him and his other daughter Martha, who were then his last two surviving children.

Jefferson wrote to Francis Eppes, his sister-in-law’s husband, requesting that he locate “some good lady” to accompany Polly, or if not a “careful gentleman” along with a slave woman, specifically suggesting an older nurse named “Isabel”.  Nowhere does he mention Sally Hemings in the letter:

With respect to the person to whose care she should be trusted, I must leave it to yourself and Mrs. Eppes altogether. Some good lady passing from America to France, or even England, would be most eligible; but a careful gentleman who would be so kind as to superintend her would do. In this case some woman who has had the small-pox [inoculation] must attend her. A careful negro woman, as Isabel, for instance, if she has had the smallpox, would suffice under the patronage of a gentleman. —Thomas Jefferson[1]

However, Isabel was either recovering from childbirth or ill (presumed from the smallpox inoculation) and was not well enough to make the journey overseas.[3] Instead, Sally Hemings, Polly’s personal maid and likely her childhood friend, was chosen to accompany her, contrary to Jefferson’s written request. At the time, Polly was 8-years-old, and Sally was approximately 14-years old.

Two of the only existing eye-witness accounts of Sally Hemings’ demeanor came during her voyage to France.

One account was from Captain Andrew Ramsay, the ship captain who was entrusted to bring the girls across the Atlantic Ocean to London on their way to Paris. His impression of Sally was relayed by Abigail Adams, wife of future President John Adams (who was then serving as Minister to England). Captain Ramsay had personally delivered the girls to Abigail Adams for their brief stay in London. Upon their arrival, Mrs. Adams wrote to Thomas Jefferson saying that Captain Ramsay had observed Sally and Polly during the voyage and felt that Sally was immature and not a capable caregiver to Polly:

The old Nurse [Isabel] whom you expected to have attended her, was sick and unable to come. She has a Girl [Sally]… with her, the Sister of the Servant [James Hemings] you have with you… The Girl who is with her is quite a child, and Captain Ramsey is of opinion will be of so little Service that he had better carry her back with him. But of this, you will be a judge. She seems fond of the child and appears good naturd [sic]. —Abigail Adams[4]

Ten days later, after having had the chance to observe the two children for herself for an extended period, Abigail Adams wrote again to Jefferson about her concern of Sally’s incapacity as a caretaker for Polly:

The Girl she has with her, wants more care than the child, and is wholly incapable of looking properly after her, without some superior to direct her.  —Abigail Adams[5]

These historical, first-hand characterizations of Sally Hemings as a naïve and immature child differ markedly from the portrayal of her as an attractive, developed woman who traveled with Jefferson to Paris and swept him off his feet, as depicted in Hollywood remakes of the scenario. In these fictionalized accounts, Jefferson allegedly fell in love with Sally Hemings while in Paris and conceived a child with her there, later to be born in America supposedly as a slave named Tom Woodson. However, in 1998 that claim was proven false when DNA results for five different Woodson descendants showed no connection at all with Jefferson-male DNA. Thus some other man was the father of that child.

To make the notion of a relationship between Jefferson and Hemings more believable, in the CBS miniseries Thomas Jefferson and Sally Hemings: An American Scandal, the actress who played the part of Sally Hemings was 25-years old, an entire decade older than Sally’s age at the time. The feature film Jefferson in Paris employed the same tactic, misrepresenting the true age and appearance of Sally Hemings as she was portrayed by an actress that was 23-years old.

In the New York Times, Historian Phyllis Lee Levin, author of the historical biography Abigail Adams, criticized the modern and historically inaccurate versions of Sally Hemings in Paris:

If you compare [the] version of “Dashing Sally” Hemings, “admired from Monticello to Paris as a great beauty” to the on-the-spot descriptions of her forlorn appearance when she first turned up on Abigail Adams’s London doorstep, you have no choice but to congratulate Jefferson’s so-called slave mistress for effecting one of the most remarkable make-overs of all time…. One wonders, given her startling metamorphosis from bedraggled to “dashing,” if Sally Hemings, for the sake of a movie, hasn’t been turned into a Colonial Cinderella. —Phyllis Lee Levin[6]

The eye witness characterizations by both Captain Ramsay and Abigail Adams are also in sharp contrast to the description made over 80 years later by Sally Hemings son Madison Hemings, who, in his old age, claimed that four years after her arrival in Paris the then seventeen-year-old Sally Hemings would negotiate a treaty with Jefferson — one of the smartest minds in history — to have her children freed when they reached 21 years of age, although for some reason did not negotiate her own freedom. Madison himself was not 21 but 22 when he was freed, and his older brother Beverly didn’t leave Monticello until he was 24.[7] Alternatively, the only record of a “treaty” or an agreement was reached not with Sally Hemings but between Jefferson and Sally’s brother James Hemings, who negotiated his own freedom. Other than Madison Hemings’ claim, there is no record of any “treaty” between Jefferson and Sally Hemings.[8] Jefferson did free James Hemings after their return to Virginia, but James found it difficult living on his own as a free black man, and unfortunately he later committed suicide.[9]

The claim of a treaty with Sally Hemings is based on the theory that she had leverage over Jefferson since she was supposedly pregnant with his baby (presumed to be Tom Woodson) at the time. The Woodson family oral history had always claimed that they were direct descendants of Thomas Jefferson. However, in 1998 that claim was proven false when DNA results for five different Woodson descendants showed no connection at all to any Jefferson male DNA. Thus some other man was the father of that child.

There has never been an account by any of the hundreds if not thousands of people who came in contact with Jefferson that noted any behavior or hint of a relationship between him and Sally Hemings. Many historians and novelists have theorized and fictionalized these notions but there are no first-hand accounts of such behavior. There are many descriptions of Jefferson’s disciplined composure and the way he carried himself around people, but there is practically nothing on Sally Hemings. In fact, the only first-hand observations of Sally Hemings demeanor in the entire historical record come from Captain Ramsay and Abigail Adams upon her arrival in London. Beyond these two descriptions, everything else is speculation.


[1] Letter from Thomas Jefferson to Francis Eppes [30 August 1785]
[3] The Jefferson-Hemings Controversy: The Report of the Scholars Commission, p 148.  Cynthia H. Burton, Jefferson Vindicated, p.105, letter from Abigail Adams to Thomas Jefferson.
[4] Abigail Adams letters to Thomas Jefferson, June 26 & 27, 1787
[5] Abigail Adams letter to Thomas Jefferson, July 6, 1787
[6] Letter to the New York Times, April 18, 1995. http://www.nytimes.com/1995/04/25/opinion/l-jefferson-s-love-life-doesn-t-equal-history-a-colonial-cinderella-054795.html
[7] The Jefferson-Hemings Controversy: The Report of the Scholars Commission, p 116-117
[8] Thomas Jefferson, Agreement with James Hemings, 15 September 1793
[9] Maria Godoy, Behind The Founding Foodie, A French-Trained Chef Bound By Slavery, NPR, October 17, 2015.

Image: Notre Dame bridge in Paris 1751. Nicolas Raguenet, La joute des mariniers, entre le pont Notre-Dame et le pont au Change, 1751. Oil on canvas, 47 x 83.5 cm. Carnavalet Museum, Paris. Photo: Claudio Gallego.

Scientists Denounced the 1998 Jefferson-Hemings DNA Study, But Historians and Media Paid No Attention

Scientists Denounced the 1998 Jefferson-Hemings DNA Study, But Historians and Media Paid No Attention

Scientists Denounced the 1998 Jefferson-Hemings DNA Study, But Historians and Media Paid No Attention

“The genetic findings my collaborators and I reported in the scientific journal Nature do not prove that Thomas Jefferson was the father of one of Sally Hemings’ children. We never made that claim… It is true that men of Randolph Jefferson’s family could have fathered Sally Hemings’ later children.”                      Dr. Eugene Foster, lead author of the 1998 DNA study


The infamous Thomas Jefferson — Sally Hemings DNA study titled Jefferson fathered slave’s last child was originally published November 5, 1998 by the British scientific journal Nature. Since 1998 it has been widely accepted by media and academia that the study established a definitive DNA match, and that belief has seeped into our public consciousness. But was a DNA match actually established?

Before the Nature article, most Jefferson scholars had dismissed the notion of a Jefferson-Hemings affair, citing unsubstantiated hearsay, erroneous claims and an overall lack of any real evidence. Utilizing Y-chromosome analysis that had only became available two years before, the 1998 DNA study sought to provide scientific evidence to settle the matter. However, only after the study’s publication did it become apparent that lead author Dr. Eugene Foster withheld from his co-authors and the editors at Nature the possibility of other Jefferson male relatives as paternity suspects. As a result, Foster’s conclusions were not subject to proper transparent peer review before publication.

Scientists and medical experts called the study into question after publication, particularly the problematic/inaccurate headline Jefferson fathered slave’s last child which Foster later had to retract, admitting that it was “misleading”. This retraction came after scientists around the world criticized the author’s overall interpretation of the DNA results, with experts even calling it “an abuse of the scientific press.”  So what was it that scientists found issue with?

Origins of the DNA study

Dr. Eugene Foster, a retired pathologist, became involved at the request of his friend, amateur historian Winifred Bennett. In 1997, Bennett suggested to Foster that he collaborate with her to conduct DNA testing of descendants of Jefferson, Hemings and other families to determine scientifically whether any matches occur that would indicate paternity. Bennett described her wish to travel with Foster and interview the various descendants when blood samples were taken and write a book on the topic. Once the results were analyzed, Bennett would then publish her book in partnership with Foster when he was to simultaneously release the DNA findings to the press.

Foster initially passed on the idea, but after researching the topic he found it to be possible to isolate the male Y-chromosome from the male lines of various family descendants. Once isolated, the y-DNA signatures of the family lines could be compared for potential matches. He then set out with Bennett to visit and test the y-DNA of family descendants to establish whether Thomas Jefferson fathered any of Sally Hemings’ children.

In order to locate Jefferson descendants, Foster reached out to the Thomas Jefferson Memorial Foundation (now called the Thomas Jefferson Foundation, aka Monticello) which recommended Herb Barger, a Jefferson family genealogist. Barger was instrumental in locating descendants from various family lines in order to gather their DNA, which appeared in the article’s chart that presented the y-DNA findings.


Initial Study Results

The DNA chart published within the Nature article contains two main columns of information:  on the left is ancestry with lists of descendant bloodlines, and on the right is the actual y-DNA information that was obtained from analyzing the blood samples of the descendants. (Since this study in 1998, DNA science has advanced in its ability to analyze and establish both paternal and maternal ancestry, however such an updated study of Jefferson and Hemings descendants has yet to be conducted.)

Starting at the top of the chart, President Thomas Jefferson appears first, although notably there is no corresponding DNA data next to his name and left blank. Next is Sally Hemings’ last son Eston Hemings. Next is Carr family descendants. Dabney Carr was Thomas Jefferson’s best friend who married Jefferson’s sister. It is actually Dabney Carr’s sons Peter and Samuel (Jefferson’s nephews) who are the focus because they were long thought by historians to be primary suspects for paternity of Hemings’ children. After Dabney Carr’s death at the young age of 30, his sons Peter and Samuel were taken in and reared by Jefferson at Monticello. The two brothers were later known to fraternize with the slave women. In fact, the two men were said to have even confessed to relationships with Sally Hemings and her sister Betsy Hemings.

Last on the chart are the descendants of Thomas Woodson, the rumored love-child of Thomas Jefferson and Sally Hemings, conceived when in Paris together.

Tom Woodson is a bit of a mystery since no records have been found of his birth. He was speculated to have been born in 1790 shortly after Sally Hemings returned with the Jefferson family from Paris where Jefferson had been US Minister to France. The first accusation was circulated in a tabloid of the time in 1802 by the scandalous anti-Jefferson writer James Callender. He claimed there was a tall, “light-skinned boy” about 10 to 12-years old bearing a “striking resemblance” to President Jefferson, even though Callender had never even been to Monticello or spoken with anyone there. Given the assumed dates of conception and birth, the tabloid originally claimed Jefferson fathered Woodson from an affair with Sally Hemings while together in Paris. Such a match would have solely implicated Thomas Jefferson and ruled out any other paternity candidates back in America. The Woodson family’s oral history has long maintained that they were direct descendants of Thomas Jefferson and Sally Hemings.

However, the DNA results of the numerous male Woodson descendants showed no connection at all to Jefferson’s family DNA.

This can be seen when the DNA markers (number sequence) for the Jefferson family male descendants on the top right of the chart are compared with the Woodson family male descendants’ markers on the bottom right. There is not even a remote match between the Woodson DNA bloodline and Jefferson family DNA. The results negated the extensive oral history and firm belief that the Woodson family had held that they were all descendants of Thomas Jefferson. More significantly, it proved false the original 1802 tabloid attack on Jefferson that first claimed he fathered a look-alike son named “Tom” with his “Congo harem” slave “black Sally,” (even though Sally Hemings was “mighty near white” with “long straight hair.”)

The Carr brothers’ DNA was examined, but again the DNA markers did not line up with Eston Hemings. Even though there was no match for their paternity of Eston, it is important to note that it doesn’t rule out a potential Carr DNA match for any other children of Sally Hemings (Harriet, Beverly, and Madison) since at the time, no y-DNA samples were available for any of them for this study — only for descendants of Eston Hemings.

Last we come to compare the y-DNA of Eston Hemings descendants to the extended Jefferson male descendants. DNA markers show an apparent match between Eston Hemings male line DNA and Jefferson’s uncle’s male line DNA. However, what is significantly missing from this chart and study is any y-DNA data for direct descendants from Thomas Jefferson himself. Jefferson had two daughters who carried his family line, but his only son who could carry his y-DNA died in infancy. Jefferson’s little-known brother Randolph was his nearest blood relative with five sons of his own, however Foster’s study did not make an effort to locate DNA from Randolph’s male line, nor even mention him in his initial study.

Instead, Foster had to go further up the Jefferson family tree for Jefferson male DNA:  past Thomas Jefferson’s father Peter, up to his grandfather and ultimately over to Thomas Jefferson’s paternal uncle Field Jefferson, who was the brother of Thomas Jefferson’s father Peter Jefferson. Field Jefferson had 4 sons who then collectively had 15 sons of their own. It was from these descendants that the Jefferson family male y-DNA was collected, tracing all the way back through six generations of descendants.

Thus, in the absence of Thomas Jefferson’s male-line DNA, it was from uncle Field Jefferson and his male line descendants’ DNA that the Jefferson male y-DNA signature was established as a match with Sally Hemings’ last son Eston Hemings.  (At the time, no y-DNA was retrieved from any of Sally Hemings other children; only Eston Hemings.)

Even though Thomas Jefferson’s DNA was not available, Foster assumed it would be exactly the same as his uncle Field Jefferson’s male-line descendants. Although there were at least 22 Jefferson males alive at the time with the same y-DNA — five of which lived nearby and were prime suspects for paternity — Foster put his thumb on the scales with his conclusion: “The simplest and most probable explanation for our molecular findings are that Thomas Jefferson, rather than one of the Carr brothers, was the father of Eston Hemings.” The accompanying headline of the article took it even further to definitively state: “Jefferson fathered slave’s last child.”

The Scientific Backlash

Scientists around the world refuted Foster’s analysis. Dr. Jane Rees, Nature Scientific Correspondence Editor, later stated in an email, “We received various replies to the original Scientific Correspondence by Foster, and simply did not have enough space to publish them all.” DNA expert Dr. Thomas Traut analyzed the DNA data and noted the error in Foster’s interpretation of the data and the headline: “This title is not a statement of fact.… It, therefore, represents an unfortunate lapse, because Foster et al. omitted any presentation of the number of adult male Jeffersons living in 1807 in that region of Virginia.”

Dr. David Abbey pointed out the error in one of only two letters to the editor that Nature chose to publish:

“…the authors did not consider all the data at hand in interpreting their results. No mention was made of Thomas Jefferson’s brother Randolph (1757-1815), or of his five sons. Sons of Sally Hemings conceived by Randolph (or by one of his sons) would produce a Y-chromosome analysis identical to that described by Foster…Further collaborative data (for example, the whereabouts of any of those who might have been involved at conception) are needed to confirm that Jefferson did indeed father his slave’s last child, as claimed in the title. We know Thomas Jefferson was there, but how about Randolph Jefferson and his sons?”

And it seems that this omission was not by accident. Foster later admitted in an interview with the US journal Science that his assistant, the genealogist Herb Barger, had previously informed him that Randolph and his son Isham were likely candidates for paternity. Reed Irvine, Founder of Accuracy in Media, pressed Foster to admit his deception:

“Dr. Foster acknowledged in an interview with me that he had not informed the editors of Nature that Thomas Jefferson had a younger brother, five nephews and two cousins who could have passed the distinctive Jefferson Y chromosome on to the male children of Sally Hemings. He admitted that he had known of these other Jeffersons, but he said rather testily, ‘I don’t believe I have to account for all the history.'”


Randolph Jefferson: An Overlooked Primary Suspect

Re-examining the history and Thomas Jefferson’s correspondence at the time, it turns out that the book Thomas Jefferson and his unknown brother Randolph includes a relevant letter from Thomas to his brother. Randolph, who was then a widower, was invited by his brother to come to visit — which would place him at Monticello a little over 9 months prior to the birth of Eston Hemings. Randolph also coincidentally appeared at Monticello immediately after Eston’s birth.

Randolph Jefferson lived in the vicinity at Snowden Farm, which was half a day’s ride from Monticello. When Thomas Jefferson was President and living in Washington DC, Monticello often was kept locked up. Jefferson would send advance word to his family that he was going south to visit Monticello, which prompted friends and relatives to flock there to see him upon his return. At times guests numbering more than 50 would stay for weeks. Therefore whenever he would return to Monticello, all manner of friends and relatives would come to join him, particularly for the funeral of Thomas Jefferson’s daughter Mary in April of 1804, which also happened to be exactly 9 months before the birth of Madison Hemings, one of Sally Hemings’ sons and the older brother of Eston Hemings. With so many people visiting, there are almost no records of the comings and goings of Randolph, the Carrs’, or of any other close male relatives to Monticello. It is simply impossible to claim that Thomas Jefferson was alone at Monticello at any given time, and that none of the other 22 Jefferson family males were present.

Of the Jefferson males, Randolph Jefferson in particular is a bit of an enigma and rarely appears in history books. In fact, what little is known of him comes from only a few sources, one of which is the Memoirs of a Monticello Slave by Isaac Granger Jefferson. He described Randolph as being much different than his brother Thomas: “Old Master’s brother, Mass Randall, was a mighty simple man: used to come out among black people, play the fiddle and dance half the night; hadn’t much more sense than Isaac.” In fact, a common description of Randolph was that he was impressionable, unaccomplished and not very intellectually curious, which is perhaps why he was overlooked by the history books.

In addition, Eston Hemings’ family’s oral history stated that they were descended from a Jefferson uncle. Julia Westerinen told the Washington Post “It was never really talked about in our family… We were told we were related somehow to an uncle of Jefferson’s.”. At the time, Randolph Jefferson was known at Monticello as “Uncle” and “Uncle Jefferson.”

Within days after the birth of Eston Hemings, Randolph Jefferson strangely appeared at Monticello (there is no known letter inviting him). During that visit Randolph drafted his new will with the assistance of his brother Thomas, bequeathing his estate specifically to his five legitimate sons and daughter. Did Thomas or even Randolph’s children come to know of a relationship and encourage him to get his estate in order? Did Thomas, as the older brother who helped rear Randolph and advise him throughout his life, encourage Randolph to straighten out and change his ways?

Shortly after Eston’s birth, Sally Hemings was relocated from her private cabin to the south terrace colonnades under main house to live with her sister Critta. Could the relocation have been an effort to protect Sally Hemings from male visitors such as Randolph? Perhaps, because after moving out of her solo cabin and into the shared dwelling — closer to Thomas Jefferson — she stopped having children. Coincidentally, Randolph Jefferson also remarried and within a year he had a child with his new wife.

Herb Barger, the Jefferson family genealogist who located the descendants, alerted Foster about Randolph Jefferson being a prime suspect for paternity who would have the same Jefferson male y-DNA. Barger also warned Foster about Randolph’s sons (closer in age to Sally Hemings) being suspects who were known to fraternize with the slave women, particularly Randolph’s son Isham and his distant cousin John Garland Jefferson. Isham spent time being reared at Monticello, and John Garland stayed at Monticello to study law. In an email, Barger warned Foster early in the study:

…a Jefferson/Hemings match WOULD NOT NECESSARILY mean the father to be President Jefferson, but possibly Isham (or another Jefferson around that area at that time)…. I believe CAUTION would be in order if a Jefferson/Hemings match is returned because President Jefferson, Isham and the Field Jefferson blood donors would have the same DNA.

Foster replied to Barger, thanking him for the information about Isham Randolph Jefferson, saying: “This is exactly the kind of information that will have to be considered if it turns out that there is a Y-chromosomal match.” He even went on to state: “the DNA evidence in itself can’t be conclusive for a variety of reasons.”

Barger then noticed that the frequency of Foster’s replies to his emails and calls began to significantly diminish. After a long period with no response, Barger wrote an email to Foster warning of the need to be thorough and careful with the interpretation of the DNA findings, and the danger of the media jumping to sensational conclusions, which, if erroneous, could be impossible to retract:

“I have cautioned all concerned that raw results may need interpretation and discussion by those of us with information that may have a bearing on the matter. I would like to be IN on the results BEFORE they hit the media because you know, the first headlines are lasting impressions and any corrections and clarifications would be way back on later pages at later dates. This topic is too HOT and significant to treat it as another “sensational” headline in today’s fashion. The final outcome will have a very profound historical effect (whatever the results may indicate) and should be studied and announced with great care and dignity…”

Despite Barger’s integral contribution to the study, Foster and the Thomas Jefferson Foundation did not invite him to the press conference at Monticello where the study was first shared with the press on November 1, 1998. Barger watched the coverage from home as the nightmare scenario that he predicted played out in the media. Despite Foster’s assurances, neither Randolph Jefferson —nor any other Jefferson males alive at the time — were ever mentioned as possible suspects for the paternity of Sally Hemings’ son Eston. In fact, Foster’s article gave the impression that Thomas Jefferson was the only candidate.

So why did Foster depart from the scientific method and never mention Randolph Jefferson, his sons, or any other Jefferson male in the article? Did he withhold crucial evidence of other potential suspects from his co-authors and his editors at Nature, who had they known would likely have insisted the article mention of such paternity candidates? Why was his “simplest explanation” of paternity by Thomas Jefferson devoid of impartial, unprejudiced consideration of all scientific possibilities?

As Professor Traut and others called out the erroneous Nature headline Jefferson fathered slave’s last child as “not a statement of fact,” Steven T. Corneliussen, coincidentally of the Thomas Jefferson National Accelerator Facility, clarified: “In fact, that evidence suggests no specific Jefferson.”

Others in the scientific community, such as Dr. Gary J. Davis, wrote to Nature and pointed out another scenario that Foster had failed to mention:

“…any male ancestor in Thomas Jefferson’s line, white or black, could have fathered Eston Hemings.… Plantations were inbred communities, and the mixing of racial types was probably common. As slave families were passed as property to the owner’s offspring along with land and other property, it is possible that Thomas Jefferson’s father, grandfather, or paternal uncles fathered a male slave whose line later impregnated another slave, in this case, Sally Hemings.”

Two months after the publishing of the Nature article, in a rare rebuke of another scientific journal, the preeminent US Journal Science publicly called foul:

“Contrary to headlines that splashed across the country in November, there is no conclusive proof that former U.S. President Thomas Jefferson fathered an illegitimate child by his slave Sally Hemings. At least five of his family members are candidates for paternity of Sally’s child, researchers admit in a letter in tomorrow’s issue of Nature.”

The next day, Nature published Foster’s mea culpa in a reply to the critiques:

“It is true that men of Randolph Jefferson’s family could have fathered Sally Hemings’ later children. Space constraints prevented us from expanding on alternative interpretations of our DNA analysis, including the interesting one proposed by Davis. The title assigned to our study was misleading in that it represented only the simplest explanation of our molecular findings: namely, that Thomas Jefferson, rather than one of the Carr brothers, was likely to have been the father of Eston Hemings Jefferson…. We know from the historical and the DNA data that Thomas Jefferson can neither be definitely excluded nor solely implicated in the paternity of illegitimate children with his slave Sally Hemings.”

Foster used the excuse of “space constraints” that kept him from mentioning all the alternative interpretations, however, there appeared to be enough space to simply add the letter “A” to the headline, which would have made it more accurate: A Jefferson fathered slave’s last child.

And regarding overall space constraints, Foster’s article was accompanied by a sister article on the topic:  an essay on the political significance of the DNA study, which was questionable for a scientific journal to publish. In fact, it took up more space than the Foster article, and repeated inaccurate claims of a definitive DNA confirmation: “Now DNA analysis confirms that Jefferson was indeed the father of at least one of Hemings’ children.”

The journal Natural Science wrote perhaps the most scathing summary critique:

“Rather than accept the authority of the editor of Nature or some other journal in the determination of scientific truth, both the media and the public at large should be skeptical about all scientific claims until they have been evaluated, not only by peer-reviewed journals but also in the open forum of scientific and public discussion. In particular, the public should be skeptical about scientific claims that support political interests.  When such claims lack intrinsic scientific significance (as in the case of those made in the Foster paper), their publication in a scientific journal should be recognized for what it is: an abuse of the scientific press.” [March 19, 1999]

Backlash from the Legal Community

In addition to criticism from scientists, Foster found his DNA interpretation also being criticized by legal experts. Attorney Thomas R. Moore pointed out in the New York Times that based on the DNA evidence “no court of law would hold that Thomas Jefferson had a child by Sally Hemings.”

Historian Willard Sterne Randall reached out to the FBI for clarity:

“I don’t deal in inferences. As a former journalist, I got on the phone and called the head of the DNA lab at the FBI, Jennifer Smith, who told me this case wouldn’t hold up in court because the DNA has to be direct from the father. That is an impossibility because [Jefferson’s] only son died, and as long as there are other possibilities of people who had access to Sally Hemings the case would be thrown out of court on those grounds.”

Put more bluntly by geneticist Carl Ladd, Supervisor of the DNA Unit of the Connecticut State Police: the Jefferson DNA study “doesn’t mean bingo as to whether Jefferson is or is not the father… it is not conclusive.”

Foster backpedaled in his published reply to the lawyer in the New York Times stating, “the genetic findings my collaborators and I reported in the scientific journal Nature do not prove that Thomas Jefferson was the father of one of Sally Hemings’s children. We never made that claim.” However his article’s headline “Jefferson fathered slave’s last child” explicitly made this claim, and an editor at Nature confirmed Foster’s duplicity to Reed Irvine:

“[Foster’s assistant Herb] Barger told Dr. Foster about Randolph and the seven other Jefferson men who could have fathered children by Hemings. A Nature editor claimed Dr. Foster did not share this information with them and that he approved the headline that declared without any qualification that President Jefferson was Eston’s father.”

This was verified by Dr. Jane Rees, Nature Scientific Correspondence Editor, when she replied to Herb Barger by email on Jan 11, 1999: “Foster did see a precursor to the title used before publication, and so we do not feel that the title used was imposed on him in any way.”

However, in a disingenuous effort to save face, Foster would pretend that the headline was a mistake on the part of Nature, and he told the Washington Post he was frustrated and working hard to fix their error:

“Dr. Eugene A. Foster, one of the primary researchers in the DNA study, has tried to rein in these stories, but to little avail. In January he repudiated as “misleading” the headline — “Jefferson fathered slave’s last child” — that accompanied a November article he coauthored in the journal Nature, an article that touched off the Jefferson-Hemings uproar.”

Overlooked Eye-Witness Testimony

Historically, one of the most compelling eye-witness accounts comes from Captain Edmund Bacon. In Jefferson’s Garden Book, he describes Bacon as his Monticello overseer “since 1806 and before that working for him in various jobs.” Bacon’s father went to school with Jefferson and was his first Monticello overseer, and Bacon’s oldest brother was the overseer while Jefferson was in Paris as US Minister to France. Bacon was thoroughly familiar with all the people and activities on the Monticello mountaintop, and in his retirement he dictated his memoirs in interviews with the Reverend Hamilton Pierson. Bacon described how he saw not Thomas Jefferson but another man coming out of Sally Hemings’ cabin on many mornings:

“[Jefferson] freed one girl some years before he died, and there was a great deal of talk about it. She was nearly as white as anybody and very beautiful. People said he freed her because she was his own daughter. She was not his daughter; she was _____’s daughter. I know that. I have seen him come out of her mother’s room many a morning when I went up to Monticello very early.”

Pierson he said he omitted certain names and facts “in regard to the intemperance, and other vices… of some who were connected with Mr. Jefferson’s family.” Could the name of the man have been omitted by Pierson in order to protect a Jefferson relative? It is currently unknown who that regular visitor to Sally Hemings’ cabin was, but one thing is clear:  Edmund Bacon said definitively it was not Thomas Jefferson he saw with Sally Hemings, but another man.


And could it be possible that in the evenings, while Thomas Jefferson was entertaining relatives and guests with music and intellectual discussions in the main house at Monticello, his brother Randolph preferred  “to come out among black people, play the fiddle and dance half the night?” Which could have led to a sexual relationship with Sally Hemings? The power dynamic of the brother of Monticello’s master making advances upon a slave such as Sally Hemings likely would not have afforded her with much ability to refuse him. And as many historians have emphasized, a relationship between Thomas Jefferson and a slave just doesn’t fit his character. However, it could very likely fit his brother Randolph, who rode and served with four militiamen who had black slave mistresses, including one man whose mistress was Mary Hemings, one of Sally Hemings’ sisters.

If this were a legal case, Foster would be in contempt of court for withholding evidence and hiding key suspects from the judge and jury. Scientific experts would testify to his faulty analysis and even condemn him for his willful suppression of evidence, reminding the court why scientific studies should always be subject to thorough peer review to avoid bias. And finally, legal experts would point out how the overall evidence does not come close to meeting the threshold needed for a conviction of paternity.

But this scientific and legal critique all happened outside the attention of the mainstream media — which had already run with the first sensational and unvetted story. Foster’s initial article rushed to a biased judgment and provided an a priori conviction based on theory rather than empirical evidence. Where perceptions matter most, the court of public opinion passed judgment without the benefit of the full evidence or facts.


The Media Feeding Frenzy

The news of Foster’s original DNA article reverberated and became one of the most heavily covered stories of 1998. Much of the coverage demonstrated a remarkable flight from careful and skeptical reporting. All too often, the news stories, commentary, and analysis transformed an intriguing but admittedly indeterminate scientific finding into a dead certainty, with headlines such as “DNA Link; Paternity Proved,” and “Adulterer on Mt. Rushmore,” with the article even charging Jefferson of “statutory rape”. National Public Radio’s account was typical, the correspondent announcing:

The proof is finally in. The president not only did have an illicit sexual affair, he fathered at least one child with his lover… DNA testing has ended (that) debate.

U.S. News and World Report was one of the first publications to break the story and featured an inaccurate genealogical chart that showed a direct DNA line between Thomas Jefferson and the modern descendant of Eston Hemings, even though the y-DNA testing had to come instead from descendants of Jefferson’s uncle Field Jefferson. This incorrect visual depiction was replicated by other news outlets.

Many journalists went on to turn the DNA results into a sort of referendum on the state of race relations and presidential politics happening at the time. Dr. David Murray, Director of Research at the Statistical Assessment Service, described a hostile environment where anyone questioning the results risked being accused of racism:

Are there skeptics and resisters? Indeed, but they are not only inconsequential, they are suspect as to motive. Since the effect of the findings is a moral vindication for slave narratives (moreover, they affirm the sexual oppression of women) and hence, by extension, for the role of the entire African American community in American life, those who today resist the facts (or even express uncertainty) should be seen as not only benighted, but indeed, must be suspected of low motives.

In an editorial headline on the topic, the New York Times characterized those who questioned the DNA claims: “Hemings-Jefferson Deniers, Desperate.” A letter to the Washington Post even went so far as to state, “Like the Holocaust deniers and others… Jefferson’s purported defenders have a problem with the truth.” The dynamic created a toxic environment that cut off any critical discourse.

Overnight, Jefferson’s stature was diminished, as he was consumed by accusations of hypocrisy and white supremacy. He was accused of being a sexual predator, and several reports even cast him as a child molester. His entire legacy and principles came into question, as Jefferson and all that he stood for was now being seen through the darkest of lenses. The Chicago Tribune all but convicted and condemned Jefferson:

“Now along comes DNA testing evidence, the same leveler that exposed the sexual peccadilloes of President Clinton, to provide clear and convincing genetic evidence that Jefferson fathered at least one child by his slave mistress, Sally Hemings. What a relief. Now Jefferson can be brought down off the god-like pedestal on which some have tried to elevate him, ignoring oral history provided by Hemings’ family…. How are we to view Jefferson now? How about “deadbeat dad”? That’s what you call fathers who run away from their responsibilities to their children.”

However, there were a rare few in the media who thoroughly researched the topic and steered clear of the sensational bandwagon. Only a handful of media published corrections — often regulated to the back pages — and even fewer noted the failure of the press to correct the record. The Wall Street Journal stated “the backtracking comes a little late to change the hundreds of other headlines fingering Jefferson.” The article pointed out that there were no fewer than eight Jefferson relatives who were potential candidates for some or all of Hemings children. “The research could never finger Jefferson individually with complete accuracy,” said Dr. Stephen Quake, Stanford University Professor of Bioengineering and Applied  Physics. “All that the DNA indicated in the end was that descendants of Jefferson’s paternal uncle have the same Y chromosome as a male-line descendant of Eston Hemings, the youngest son of Sally Hemings.”

The Washington Post ombudsman issued a sharp critique of no less than seven of its own published articles for their overreaching claims, some of which were front page features.

“In reporting on the Jefferson-Hemings story these six months, The Post often has failed to make clear what is fact (DNA testing shows that a Jefferson fathered Eston Hemmings but not which Jefferson), what is speculation and what is convenient.”


An “abuse of the special authority of science”

The US journal Science called out Nature for its editorial failures:

“Y chromosome data cannot be used to identify individual paternity within the Jefferson clan. That’s a job for historians,” Foster says. But that’s not how it sounded in the headlines on the initial Nature report and on an accompanying comment by geneticist Eric Lander of the Massachusetts Institute of Technology and historian Joseph Ellis of Mount Holyoke College in South Hadley, Massachusetts. The Foster article was titled: “Jefferson fathered slave’s last child,” and the comment included a heading that said: “Now, DNA analysis confirms that Jefferson was indeed the father of at least one of Hemings’ children”…. Nature staffer Rosalind Cotter agrees that “the whole thing really was rushed through.”

According to science writer Steven Corneliussen, scientific participants in the controversy – who were not scholars qualified to put the results into historical perspective – abused science’s special authority, leading to misreporting that hobbled public understanding. A later flawed statistical analysis of the pattern of Jefferson’s presumed presence during conception windows only made things worse, and Corneliussen described both as “intellectual disrespect: abuse of the special authority of science.”

“Whether or not Hemings and Jefferson had children together, misreported DNA and misused statistics have skewed the paternity debate, discrediting science itself.”

The legal standards for paternity in the State of Virginia require near certainty: “Parentage between child and man must be established by (i) Scientifically reliable genetic tests which affirm a 98% probability of paternity.”

With at least eight other Jefferson male candidates for paternity alive and nearby Monticello at the time, Foster’s results proved at best a 12.5% probability, which is nowhere near the legal standard. Foster failed to meet even the basic standard of proof beyond a reasonable doubt in the State of Virginia, which states: “It is not sufficient to create suspicion or probability of guilt, but the evidence must establish guilt beyond a reasonable doubt. It must exclude every reasonable hypothesis except that of guilt.

It has been said that “a defining mark of good scientists is that they go to great pains to distinguish between what they know and what they don’t know.” When Foster’s initial DNA article in the journal Nature claimed the “simplest and most probable explanation” is that Jefferson fathered slave’s last child, Foster ignored exculpatory evidence and put his thumb on the scales, egregiously departing from the scientific method. No amount of backpedaling could put the genie back into the bottle.



A month after the scientific community had rebuked his findings, in a moment of frankness Foster revealed: “This whole affair has been conducted by amateurs. I include myself….”

While Foster lost friends and lost credibility in the scientific world, he was warmly embraced by the historical and social sciences academic world, which invited him to speak at numerous universities, including Yale. Despite the fact that his only expertise was as a pathologist (who specializes in the causes and effects of diseases), he was invited to England to speak about slavery. In his speaking tour he was at times even referred to as “Historian Eugene Foster.”

And what became of Winifred Bennett and her partnership with Foster to have her book and his article come out together? When Foster began speaking with Nature about publishing his article, Foster went quiet and stopped talking to Bennett. Months later she was caught by surprise and devastated when she saw Foster’s press conference at Monticello on the nightly news. Her agreement with Foster to release his study together with her book vanished as she watched Foster breaking the news without her, but instead with Dan Jordan, the President of the Thomas Jefferson Foundation. For Bennett, it marked the end of her book project and the end of her friendship with Foster. She later described to Herb Barger how months later she caught up with Foster and confronted him:

I said “Gene, what is it you want? Do you want money?” He said, “No. I want fame.” Well, he just was willing to sacrifice me for his fame.