Perhaps the most helpful criticism I received while working to finalize this volume came from Henry Wiencek, whose books on slavery and the Founding Fathers include the 2003 prize-winning volume, An Imperfect God: George Washington, His Slaves, and the Creation of America. He contacted me in late 2007 in connection with some research he was doing on Jefferson and slavery, and in the process argued (in my view quite persuasively) that Jefferson’s notation that Harriet II and her brother Beverly had “run” in 1822 must be interpreted in the light of other existing evidence to mean he allowed them to leave Monticello in search of a better life. He noted (as I had as well) that Monticello overseer Edmund Bacon later spoke of having put Harriet on a stage to Philadelphia with fifty dollars from Mr. Jefferson, but also reminded me that granddaughter Ellen Randolph Coolidge later stated: “It was his principle (I know that of my own knowledge) to allow such of his slaves as were sufficiently white to pass for white men, to withdraw quietly from the plantation; it was called running away. … ”85 Although I quoted that sentence in Chapter Six of my original Individual Views, in retrospect I clearly failed to give it the weight I now believe it warrants. I have thus made appropriate changes to my Individual Views, and am most grateful to Mr. Wiencek for pointing this out.
To be sure, Ellen referred specifically to slaves who were white enough to pass “for white men,” but that would cover Beverly—and the perhaps more elaborate measures of having Bacon put Harriet on a stage with a significant amount of cash could reflect a greater concern for her physical safety because she was a woman (and, in any event, that incident is already persuasively documented by Bacon himself). So, thanks in large part to Mr. Wiencek, I am now persuaded that Jefferson’s notations about Beverly and Harriet having “run” in 1822 are not reliable evidence that either of them left Monticello without Thomas Jefferson’s consent. This, in turn, suggests that Jefferson may well have informally “freed” Harriet within a few months of her twenty-first birthday, which occurred in May 1822.
But by the time Harriet turned twenty-one, Beverly would have been twenty-four. So allowing a twenty-three- or twenty-four-year-old Beverly to “withdraw quietly” from Monticello, and having Bacon put a twenty- or twenty-one-year-old Harriet on a stagecoach, is still hardly serious evidence that this was the result of some sort of “treaty” in which Thomas Jefferson promised an adolescent Sally Hemings he would free all of their children when they turned twenty-one if she would return to Monticello with him from Paris. Ellen’s use of the plural “men” suggests that Beverly was not the only light-skinned slave to be thus permitted to withdraw quietly, and her language suggests that this was a consistent policy. We do not have sufficient evidence to determine with authority how often this occurred, but from Ellen’s statement the motivating factor was that slaves had to be “sufficiently white to pass for white men. … ” Assuming this was true, and that Beverly was as light-skinned as Harriet or Eston, then this would explain his departure without any need for a “treaty” commitment made in Paris to persuade a reluctant Sally to return to Virginia.
Science Writer Steven T. Corneliussen
Subsequent to the release of our April 2001 report I have benefited greatly from the assistance of Steven Corneliussen,86 a science writer at the Thomas Jefferson National Accelerator Facility (Jefferson Laboratories) in Newport News, Virginia. Among his other activities, Mr. Corneliussen is a media advisor for the American Institute of Physics and a leader of the effort to preserve Virginia’s Fort Monroe, where the Civil War’s first self-emancipators helped push slavery toward collapse. Corneliussen is an agnostic on the paternity question but a student and critic of what he calls “Hemings-Jefferson science abuse,” by which he means misuse of the special authority of science in the paternity controversy. He charges that this abuse began when Nature’s editors confused the public worldwide concerning the DNA evidence by conflating molecular findings and historical interpretation of molecular findings. He charges that the abuse worsened when Dr. Fraser D. Neiman’s statistical study confused both the public and “credulous historians” concerning the correlation between Hemings’ conceptions and Jefferson’s visits to Monticello. Corneliussen believes that, irrespective of whether or not Jefferson and Hemings were parents together, this abuse discredits science itself and requires scrutiny because of science’s special authority on matters in public discourse.
After the 2000 Neiman study confidently claimed that statistical science proved that Jefferson fathered Hemings’ children, Corneliussen—skeptical that statistical science can be usefully applied in a two-century-old paternity mystery—began inquiring whether experts would be vetting the study independently. Only later did I come to know him. After we had exchanged a number of e-mails, I asked him to review Chapter Five of my Individual Views, concerning Dr. Neiman’s study. Most of the members of the Scholars Commission were not professional scientists, and I wanted to get another opinion on my very critical assessment of Dr. Neiman’s scholarship to make sure I was not being unfair. Corneliussen is not a scientist either, but on his own initiative—having long since realized that no one in science was going to review a scientific study sequestered in a humanities journal—he informally sought the scientific judgments of two of his colleagues concerning the statistical study. He called on Dr. William C. Blackwelder, a biostatistician and Fellow of the American Statistical Association, and Dr. David R. Douglas, a Fellow of the American Physical Society whose particle-accelerator work involves computer simulations akin to those on which Dr. Neiman built his statistical study.
Later Corneliussen published an excellent op-ed article in the Richmond Times-Dispatch summarizing his view that the paternity controversy has involved abuse of science’s special authority. In part, he elaborated on Dr. Blackwelder’s and Dr. Douglas’s judgment that, even setting aside the statistical study’s questionable assumptions and other problems, “two failings in particular fatally undermine” it. First, correlation or association cannot prove causation. Corneliussen reported that Dr. Blackwelder “emphasized that at most statistical science can only investigate whether Jefferson was present more often” at times of conceptions “than would be expected by chance alone.” Second, the study failed to account for the varying probabilities with which conception can take place on one of the days during a window of time approximately nine months before birth. Dr. Blackwelder had criticized this failing for the case of Sally Hemings’ son Beverly by noting, as Corneliussen put it, that “Jefferson’s presence for less than 50 percent of the window in [Beverly’s] case means an overall probability of less than 50 percent for Jefferson’s presence in all six. So just by itself, this one absence upends the entire study.” (As I have explained elsewhere,87 Sally Hemings likely had five children, not six. Corneliussen engaged Dr. Neiman’s claims as he found them.)
Corneliussen notes that the statistical study merely “awarded itself a fudge factor” of sorts concerning Beverly’s case, and that the failure to account for a distribution of conception-window probabilities mattered notably in three more cases as well. Using historical data recorded in Dr. Neiman’s article and in Cynthia H. Burton’s Jefferson Vindicated, and building on observations by Burton and in my Individual Views, Corneliussen wrote, “Depending on biological assumptions, Jefferson missed a week of [Hemings’] unnamed third child’s conception window. He missed a day or more of Madison Hemings’. A Lynchburg trip caused the same for Eston’s. That means Jefferson could have been absent when the only child linked to [a Jefferson father] by DNA was conceived.” Corneliussen charges that the “statistical study called itself a ‘probabilistic evaluation,’ but didn’t even try to compute” the four absences’ statistical implications. He asserts that this point is crucial in that it shows that whatever may be the qualitative significance of the conceptions-coincidences evidence, the Neiman study is not merely weak as a scientific, quantitative claim, but is completely useless.
Professor Gordon-Reed, in a letter to the editor of the Times-Dispatch
accusing Corneliussen of “smoke and mirrors,”88 apparently completely missed this qualitative-vs.-quantitative distinction. According to Dr. Douglas, in a letter of his own,89 her letter showed no awareness that Corneliussen was condemning not the qualitative argument about the conceptions’ coincidences, but what purported to be a quantitative proof backed by the authority of statistical science. Dr. Douglas re-asserted that distinction and challenged as well Professor Gordon-Reed’s charge that in this matter and also in the DNA matter Corneliussen was attacking historians, when in fact he was criticizing scientists.
Corneliussen points out something else that’s crucial as well. I mentioned at the start of Chapter Five that the statistical study has been hailed by believers in the paternity allegations as likely to persuade those who disbelieve. He notes that R. B. Bernstein’s Thomas Jefferson, which the eminent scholar of the Jefferson era Professor Gordon S. Wood has called “the best short biography of Jefferson ever written,”90 declares the Neiman study one of “three pillars”91 of parenthood proof, along with historical evidence and the DNA evidence. Corneliussen and I have both noted that historian Jan Lewis has argued similarly.
Corneliussen remains agnostic about what the historical evidence and the DNA evidence together may show, but his commentary piece concludes: “[T]hanks to DNA confusion and the bogus statistical study, many today believe that science itself has proven the paternity. It hasn’t.”92 Anyone interested in a full airing of Corneliussen’s criticisms of Hemings-Jefferson science abuse will want to see his essay at TJscience.org, “Sally Hemings, Thomas Jefferson, and the Authority of Science.” The essay’s thumbnail summary says, “Whether or not Hemings and Jefferson had children together, misreported DNA and misused statistics have skewed the paternity debate, discrediting science itself.”93
The National Genealogical Society Quarterly Special Issue
In September 2001 the National Genealogical Society issued a “Special Issue” of its quarterly journal entitled, Jefferson-Hemings: A Special Issue of the National Genealogical Society Quarterly. A sense of what is to come appears in the one-page “Editor’s Corner” in which we are told “Monticello—ruled by the Jeffersons and populated by the Hemingses—is the symbol, if not the seat, of the world known as American slavery.”94 It is unclear whether the editor believes Jefferson created American slavery, or just that slave life at Monticello was typical of the evil institution throughout the south. Either way, her assessment is far from accurate, as slave life at Monticello was far better than at most plantations in eighteenth- and early nineteenth-century America.
The lead article in this special issue is entitled “Sally Hemings’s Children: A Genealogical Analysis of the Evidence,” and was written by Helen F. M. Leary. This article has already been dissected brilliantly by a genealogist who has spent decades studying the Jefferson family and other families in the Monticello area, and little purpose would be served by my attempting to duplicate that exhaustive, nearly two-hundred page effort.95 But a few comments may be in order. For example, Ms. Leary asserts that “the conception of each Hemings child coincided precisely and exclusively with Jefferson’s visits to Monticello.”96 Again, one wonders what she is trying to say. Is she telling us that Jefferson arrived “precisely” on the day Sally conceived each of her children and left the following morning, that Sally always became pregnant when Jefferson visited Monticello, and that Jefferson was the “exclusive” potential father present on that day? Whatever she is trying to say, we have no evidence that any of these possible interpretations is true.
In reality, there is very little—“precise” and “exclusive” or otherwise—that we really know about Sally Hemings. We do not even know with any certainty where Sally was when she became pregnant with her various children. The silliness of Leary’s assertion that “precise” evidence exists in this matter is apparent when we look at what we believe we know about the conception of Sally’s son Beverly. As discussed in Chapter Five of my Individual Views, Monticello scholar Dr. Fraser Neiman calculates that Beverly was conceived on July 8, 1797.97 From Jefferson’s records, it appears that he had been away from Monticello for more than two months prior to the estimated conception date and did not return until July 11—three days after the estimated conception. Now it is certainly possible that Dr. Neiman erred in his calculations or that Sally gave birth later than the “norm”; but, to mention just one example, we have no idea “precisely” where Randolph Jefferson was on July 8 or any other day in July 1797. Without knowing Randolph’s location—or that of his sons—how can anyone pretend to “know” that Thomas Jefferson’s presence was “exclusive”?
In a footnote on page 174 of her article, Ms. Leary seems to suggest that I attempted to mislead readers. She quotes me as saying in Chapter Ten of my Individual Views that “There was nothing in the DNA tests to cast doubt on Woodson’s status as the son of Sally Hemings,” and asserts this is “a statement that is correct but misleading to those unfamiliar with the nature of DNA evidence. Because Y-chromosomes exist only in males, a Y-chromosome test cannot provide any evidence on maternity.”98 I will submit to the reader the judgment of whether I attempted to “mislead” anyone with this statement. In reality, I had already carefully made her point in Chapter One, and the point I was making in the clause she quotes99 was that the shift in position by some advocates of the Hemingses’ story in suddenly assuming that Thomas Woodson was not Sally’s child was not warranted by anything in the DNA tests. I remain agnostic about the maternity of Thomas Woodson, but I find it noteworthy that many who once seemed certain he was conceived in Paris by Thomas Jefferson and Sally Hemings, and others who had doubts about Thomas Jefferson’s paternity but accepted Woodson’s claim to be the son of Sally Hemings, suddenly found it convenient to “assume” he was not Sally’s child either once the DNA excluded Thomas Jefferson as a possible father.
Leary also asserts without the slightest explanation that all of Sally’s children “were likely offspring of the same man.”100 Indeed, unwarranted assumption seems to be Ms. Leary’s favorite analytical tool. She notes Jefferson’s long and close friendship with George Wythe—noting Jefferson called Wythe “my earliest & best friend”—and then notes rumors that Wythe may have fathered a son by a “free mulatto who kept his home.” From this she concludes: “Obviously convinced by Jefferson’s stoicism in the face of Federalist pressures concerning the Hemings children, Wythe named Jefferson as executor and entrusted to him the responsibility for overseeing Brown’s property and education.”101 Is this really all that “obvious,” or might the choice of Jefferson as executor have been influenced by the fact that the two men had been best of friends since Jefferson’s days as Wythe’s student at William & Mary?
Ms. Leary repeats the common canard that the DNA test of Eston Hemings disproved “the family claim that Sally’s children greatly resembled Jefferson because they were fathered by a Carr nephew. … ”102 On the contrary, the test showing Eston Hemings was likely fathered by a Jefferson male said nothing about the paternity of any of Sally’s other children. This is so obvious that it is difficult to understand why apparently serious people keep repeating the claim.
It is clear that, like many genealogists, Ms. Leary places little stock in “oral traditions” passed down through generations about their ancestry. But it is a bit bizarre for her to say, without the slightest bit of hard evidence, that “[l]ike many ‘family traditions,’ the ‘Jefferson’s uncle’ story” passed down by Eston Hemings’ descendants “was a whitewashing of the truth, made necessary by the circumstances.”
Ms. Leary seeks to rebut our observation that Sally appears to have stopped having children about the time Randolph Jefferson remarried and Thomas Jefferson retired from the presidency and returned full-time to Monticello by arguing:
A more logical explanation for the end of Sally’s pregnancies is the fact that Jefferson’s daughter Martha and her family came back to live at Monticello in that year 1809—a long-cherished goal for
Jefferson. The Randolphs made the move in March, bringing children aged one to seventeen, and, presumably, some of their own servants. With the house now filled to capacity and Martha installed as chatelaine, continuing the liaison would have been problematic and dangerous, lest its discovery cause Martha to return to her own plantation, Edgehill.103
Unlike other parts of her article, this reasoning is, indeed, “logical.” But it raises yet another problem Ms. Leary fails to address. Reasonable people can agree that it would be “problematic and dangerous” for Thomas Jefferson to carry on a sexual liaison with Sally Hemings with his beloved daughter and a house full of grandchildren at Monticello. But at Jefferson’s insistence, Martha and her family returned to Monticello virtually every time Thomas Jefferson did during the entire period of his presidency. So the same “logical explanation” that persuades reasonable people that he would not have carried on the relationship after 1809 would have applied with equal force during the years when Sally was having children.
Leary argues that the reason Madison Hemings did not claim to be Thomas Jefferson’s child until more than four decades after Jefferson’s death was because, prior to ratification of the Thirteenth Amendment in “1870” (sic—the Thirteenth Amendment was ratified on December 6, 1865), Beverly and Harriet and all of their children would have been “at risk of capture and re-enslavement had their identity been discovered. That possibility, however remote, was fearsome.”104 This speculation is neither scholarship nor genealogy, it is supposition and fantasy. According to Wetmore, Madison had not been in touch with Harriet in a decade and did not know whether she was alive or dead. All he revealed about Beverly was that he went to Washington and then “married a white woman in Maryland” and they had a daughter. For “prudential reasons” he did not give the name of Harriet’s husband. Had Madison actually been seriously concerned for their welfare, all he needed to do was omit their names and even the geographic areas in which they lived from his alleged account of his family background.
The Jefferson-Hemings Controversy Page 64