The Historical Pillorying of Thomas Jefferson The “Seismic Effect” of a DNA Study Gone Wrong M. Andrew Holowchak, Ph.D. Philosophy, Rutgers University, Camden
A
NDREW BURSTEIN PUBLISHED THE INNER JEFFERSON in 1996 and claimed it was highly unlikely that Jefferson was the father of any of Sally Hemings’s mulatto children. The deciding factor was Jefferson’s character. In 1997, Annette GordonReed published Thomas Jefferson and Sally Hemings: An American Controversy. The title, suggestive of an affair, was incautious and misleading. She never asserted that Jefferson fathered any of Hemings’s children or even that it was probable that he did. She merely satisfied herself with the weak and undemanding claim that Jefferson could have fathered Sally Hemings’s children. In 2000, Burstein published “Jefferson’s Rationalizations” and, in 2005, Jefferson’s Secrets. Firmly convinced of Jefferson’s paternity, he now maintains that Jefferson had a sex-only relationship that was undertaken for medical reasons. In 2008, Gordon-Reed published The Hemings of Monticello: An American Family. Here her guardedness is gone. Jefferson had a liaison with Hemings that was lengthy and mutually fulfilling. Why did Burstein make such a peripe’teian shift? What now makes Gordon-Reed so cocksure that there was a liaison and that it was mutually fulfilling? Both scholars have rather complex answers to my questions, but each is clear that the historical evidence coupled with the DNA findings implicates Jefferson. That I find puzzling, simply because the DNA evidence shows only that Jefferson could have fathered Sally Hemings’s last child, Eston, and the historical evidence that has some bearing on settling the issue is little and gauzy. Moreover, it has not changed much since each author’s first book was published. There is still scarcely any relevant evidence and certain-
ly nothing to decide the issue or even show unquestioned probability for one view. What has decided the issue for each scholar and numerous others can only be two things. First, there is the precipitant assessment of the scientists that conducted the DNA tests in 1998 that Thomas Jefferson’s paternity was the simplest and most probable explanation of the DNA results and its fallout. Second, there is the sophistic committee report of the Thomas Jefferson Memorial Foundation (hereafter, TJMF1) in 2000 that stated it was highly probable that Jefferson was not only the father of Eston Hemings, but also all the children of Sally Hemings. In short, paralogistic assessment of the evidence, not the evidence itself, has led to the sometimes vicious pillorying of a man that has devoted the salad years of his life to the service of others. The result has been seismic, or almost so. Following the lead of the TJMF, which stated that it was important for scholars to flesh out the probable nature of the relationship between Jefferson and Hemings, scholars have begun to do just that. They have built much on the sandy foundation of the putative liaison between Jefferson and Hemings by expatiating on the nature of the relationship mostly through pure speculation often cloaked as fact. GordonReed has been awarded numerous awards, among them a Pulitzer Prize, for her book. What is the rub? It is that there is no evidence, biological or historical, that shows Jefferson was the father of Eston Hemings or any other of Sally Hemings’s children. There is no evidence, biological or historical, that even makes it probable that Jefferson was the father of Eston or any other of Sally Hemings’s children. Nonetheless, the general public and most scholars are convinced of Jefferson’s intimate involvement with Hemings. Persons must be held accountable for this deplorable state of affairs. Culpability, I maintain firmly, must rest on the two groups of avowedly “disinterested” scholars—the scientists comprising the DNA committee and the members of the TJMF, minus the sole dissenter, at the time of the report—that have led astray the general public and other scholars. With the paucity of relevant evidence that has a bearing on Jefferson’s paternity, no conclusion other
than agnosticism is justifiable and agnosticism is insufficient to incriminate any person accused of any misdeed.
“The Biological Father of Eston Hemings Jefferson” In November of 1998, a DNA study was published in the prominent weekly publication Nature. What precisely did the DNA study show? DNA samples were taken from descendants of the following men: five male-line descendants of two sons of Field Jefferson (Jefferson’s uncle), one male-line descendant of Eston Hemings (last son of Sally Hemings), thee male-line descendants of three sons of John Carr (grandfather of Peter and Samuel Carr, nephews of Jefferson and brothers), and five male-line descendants of two sons of Thomas Woodson (Jefferson’s slave, who claimed to being the famous missing first son of Jefferson and Sally Hemings). The abstract, summing the findings, reads: There is a long-standing historical controversy over the question of U.S. President Thomas Jefferson’s paternity of the children of Sally Hemings, one of his slaves. To throw some scientific light on the dispute, we have compared Y-chromosomal DNA haplotypes from male-line descendants of Field Jefferson, a paternal uncle of Thomas Jefferson, with those of maleline descendants of Thomas Woodson, Sally Hemings’ putative first son, and of Eston Hemings Jefferson, her last son. The molecular findings fail to support the belief that Thomas Jefferson was Thomas Woodson’s father, but provide evidence that he was the biological father of Eston Hemings Jefferson. 2 What was the evidence that Jefferson was the biological father of Eston? The findings showed that Jefferson could not be ruled out as the father of Eston Hemings—viz., that Eston was descended from a male in Jefferson’s bloodline, which ruled out nephews Peter and Samuel Carr. The findings also refuted a loud oral tradition, which claimed that Jefferson was the father of Thomas Woodson, the alleged missing first child of Thomas Jefferson and Sally Hemings. The scientists—comprising three geneticists, a
pathologist, a statistician, and a biochemist—argued that the evidence, coupled with historical circumstances, made Jefferson the most likely candidate for paternity. They also agreed to the misleading title “Jefferson Fathered Slave’s Last Child,” which strongly suggests that it is Thomas Jefferson and not a Jefferson— someone in Jefferson’s bloodline, including Thomas Jefferson— that fathered Eston Hemings. In their own words: The simplest and most probable explanations for our molecular findings 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. The frequency of the Jefferson haploid is less than 0.1 per cent, a result that is at least 100 times more likely if the president was the father of Eston Hemings Jefferson than if someone unrelated was the father. We cannot completely rule out other explanations of our findings based on the illegitimacy in various lines of descent. For example, a male-line descendant of Field Jefferson could possibly [sic] have illegitimately fathered an ancestor of the presumed male-line descendant of Eston. But in the absence of historical evidence to support such possibilities, we consider them to be unlikely. 3 Their conclusion is a paralogism, for reasons to be given shortly. Next there came the Thomas Jefferson Memorial Foundation’s committee report, which came out not long after the article in Nature. The report of the committee, published in January of 2000, is titled “Report of the Research Committee on Thomas Jefferson and Sally Hemings.” Members met 10 times from December of 1998 to April of 1999 to discuss the scientific data and relevant historical evidence. They formed subcommittees and consulted with experts from other committees deemed relevant. After reviewing the available evidence—I concern myself nowise with whether or to what extent all the relevant available evidence was examined, but mention only that one prominent committee member wrote that the atmosphere was often vitriolic, more of a witch hunt than an open,
objective investigation 4—the committee settled on five conclusions, only the second of which needs to concern us here. The DNA study, combined with multiple strands of currently available documentary and statistical evidence, indicates a high probability that Thomas Jefferson fathered Eston Hemings, and that he most likely was the father of all six of Sally Hemings’s children appearing in Jefferson’s records. Those children are Harriet, who died in infancy; Beverly; an unnamed daughter who died in infancy; Harriet; Madison; and Eston. The committee’s conclusion, we shall see, is another paralogism. The study, with its misleading title and poorly supported conclusion, and the TJMF’s examination of it and the “currently available documentary and statistical evidence” in favor of Jefferson’s involvement with Hemings set astir the media and historians. Newspapers across the nation—including the Wall Street Journal, New York Times, Washington Post, Chicago Tribune, and Boston Globe—wrote of the scientific “proof” of Jefferson’s sexual relationship with Sally Hemings. Other media followed suit. 5 Historians too were quick to change their minds on Jefferson’s innocence. “Rarely in the writing of American history has the conventional wisdom about a debate reversed so completely,” writes R.B. Bernstein. 6 “Science shifted opinion from doubt to belief,” states Lucia Stanton, one of the members of the TJMF committee, “and most historians of the period now acknowledge that Thomas Jefferson was the father of Hemings’s known children.”7 Later she adds: “Now, because of the almost seismic effect of a scientific test, we will never again read the words of Jefferson and the members of his household in quite the same way”. She sums: “Science has thus provided a worthwhile lesson in the interpretation of evidence. A genetic test has roused us from a complacent, and sometimes selective, acceptance of our favorite letters, memoirs, or travel accounts.” 8 Stanton’s choice of words is, I think, Tartuffery, for as I have shown in Framing a Legend, the new revisionist “history” might be avant-garde, but it eschews rationality for innovation and
vilipends the notion that there is a right way and a wrong way to do historical research. For revisionists, anything goes.
“The Simplest and Most Probable Explanations� Let us now take a closer look at the first of the two paragraphs I have selected from the published DNA study. The simplest and most probable explanations for our molecular findings 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. The frequency of the Jefferson haploid is less than 0.1 per cent, a result that is at least 100 times more likely if the president was the father of Eston Hemings Jefferson than if someone unrelated was the father. Therein we find two arguments: an argument from simplicity and an argument from probability. The argument from simplicity might be cashed out as follows: 1. The frequency of the Jefferson haploid is less than 0.1 per cent. 2. So, it is at least 100 times more likely that Jefferson and not someone unrelated to him was the father of Eston Hemings (1). 3. So, the simplest explanation is that Jefferson is the father of Eston Hemings (2). The argument from probability uses the same evidence. It differs merely in its conclusion. 1. The frequency of the Jefferson haploid is less than 0.1 per cent. 2. So, it is at least 100 times more likely that Jefferson and not someone unrelated to him was the father of Eston Hemings (1). 3. So, the most probable explanation is that Jefferson is the father of Eston Hemings (2).
There is a difficulty for both arguments. Statement two, as a conclusion for premise one, is cloaked like an exhaustive and mutually exclusive disjunctive claim. It is not. What is misleadingly missing is that others with the Jefferson haploid are also at least 100 times more likely to be the father of Eston Hemings than someone unrelated. The argument should read: 1. The frequency of the Jefferson haploid is less than 0.1 per cent. 2. So, it is at least 100 times more likely that a Jefferson instead of a non-Jefferson was the father of Eston Hemings (1). 3. So, the simplest and most probable explanation is that a Jefferson is the father of Eston Hemings (2). When we flesh out the argument as it should be fleshed out—here I have conjoined simplicity and probability in the conclusion for the sake of convenience—simplicity and probability apropos of Thomas Jefferson vanish. Without such a fleshing out, it is an instance of the fallacy of false alternatives. Let us now turn to the second paragraph, which is a concession that simplicity and probability cannot be established unless there is more evidence. We cannot completely rule out other explanations of our findings based on the illegitimacy in various lines of descent. For example, a male-line descendant of Field Jefferson could possibly have illegitimately fathered an ancestor of the presumed male-line descendant of Eston. But in the absence of historical evidence to support such possibilities, we consider them to be unlikely. 9 This argument, fleshed out, is as follows (here again placing simplicity and probability into the conclusion): 1. Another male other than Jefferson with the Jefferson haploid could have fathered Eston Hemings.
2. There is currently no historical evidence to support claim 1. 3. So, the simplest and most probable explanation is that Jefferson is the father of Eston Hemings (1 & 2). This argument also is fallacious. Lack of evidence for one claim cannot be taken as evidence for another. From the premise There is no evidence for claim Δ, one cannot conclude Claim Δ is false. Furthermore, there currently is no historical evidence, other than hearsay—for the “testimony” of Madison Hemings, on which adherents of the Jefferson-did-it bandwagon heavily rely, is mere hearsay—to show that Jefferson was the father of Eston Hemings. 10 Moreover, premise 2 is false. There are historical data that offer evidence that Jefferson did not father Eston—e.g., Jefferson’s own statement of innocence and the testimonies of several others who were at Monticello at the time of the putative liaison. The open question concerns just how to weigh this evidence. Such criticisms of the arguments from simplicity and probability are damning and decisive, but it is not my principal objective to be critical. I wish to grasp why the addition of the DNA evidence to the gossamer extant historical evidence for Jefferson’s paternity has made such a difference in the minds of historians. Whatever the reasons we are given by scholars for their insistence on Jefferson’s involvement with Sally Hemings, the fact remains that the overwhelming majority of scholars only became convinced of Jefferson’s involvement after the DNA results were published. For the sake of heightened understanding, let us disregard the decisive objections I have marshaled against the framing of the arguments, given above. Here now I wish to frame charitably—and I admit, too charitably—the arguments from simplicity and probability to get a grasp of the usage of “simplest” and “most probable”. 1. The DNA tests show that Jefferson is one of several candidates apropos of paternity of Eston Hemings. 2. The weight of the historical evidence makes Jefferson the leading candidate apropos of paternity of Eston Hemings. 3. So, the simplest (or most probable) explanation is that Jefferson is the father of Eston Hemings.
Here, let us assume premise 2 is true, though I think it is likely false. As we shall see, not much rests on it, even it does happen to be true. In what sense, then, do we have the simplest explanation? In what sense, do we have the most probable explanation? Of competing etiological claims, Χ and Ψ, to say X is simpler than Ψ in an Ockhamistic is to say X explains all that Ψ explains and does so by invoking fewer assumptions or etiological principles. For instance, Natural Selection is simpler than Special Creation vis-a-vis accounting for the extinction of extant species and birth of new species. The latter assumes numerous divine interventions in biotic affairs. The former, working on the observation of the abundance of Nature, needs only a mechanism for selection. Do advocates of a liaison invoke simplicity in an Ockhamistic sense? If all other candidates for paternity were fully explored and there was found to be nothing to decide the issue one way or another, it might be conceded that Jefferson is the simplest explanation. He was around Monticello some nine months prior to the birth of Eston Hemings. Yet how do we know no other person with the Jefferson Y chromosome was around Monticello at the same time? There are numerous other difficulties. I cite one. Many scholars—e.g., Fawn Brodie, Gordon-Reed, and the members of the TJMF at the time of their report—work on assumption that Sally Hemings had one and only one father for all her children. On that assumption, they argue the simplest explanation is that Jefferson was the father for all Hemings’s children, for he was avowedly around Monticello nine months prior to the birth of each child, so the story goes, according to Winthrop Jordan. If we assume that all her children had the same father and if it is the case that Jefferson was around Monticello prior to the birth of each child, then simplicity, I suppose, does implicate Jefferson. Notice, however, the presuppositions that drive the argument. Notwithstanding the presuppositions, simplicity does not show Jefferson is most likely the father of Eston or all of Hemings’s children. Why? “Most probable” is ambiguous. An outcome can be “most probable” of several possible outcomes without being
“probable” simpliciter. I illustrate. In the roll of a fair pair of dice, the number “seven” is the most probable outcome. There is a 0.167 (6/36) probability of “seven” upturning in a roll of the dice, which is greater than any other outcome (e.g., “two” being 1/36; “three” being 2/36; and so on). Yet anyone who acknowledged that “seven” is the most probable outcome of any particular outcome and, therefore, bet that there is a greater probability of “seven” upturning than of anything else would be a buffoon. Likewise, one could grant that, given the current state of knowledge concerning Eston (or all of Sally’s children) and the whereabouts and activities of other Jefferson candidates, Jefferson is likely the most probable father of Eston (or any other or all other of Sally Hemings’s children)—i.e., there is a greater likelihood of Jefferson being the father than any other particular candidate. Yet it still does not follow that it is probable that Jefferson is the father of Eston (or any other or all of Hemings’s children). It might be, for instance, that there is a 0.30 probability that Jefferson, say, fathered Eston and no other person with the Jefferson Y chromosome has a probability that equals or exceeds that. Still, given that, there is a 0.70 probability that someone other than Thomas Jefferson fathered Eston. There is another cautionary point. Probabilities are especially sensitive. They can change remarkably with the addition of new evidence. Little to nothing is known about the likelihoods of any of the other physical matches. Thus any probabilities we assign at this juncture, given the relative paucity of information we have on other candidates and even Thomas Jefferson, are somewhat arbitrary.
“Serious Doubt … Can No Longer Be Sustained” For many scholars, the coup de grace was Frazer Neiman’s MonteCarlo data and Bayesian employment of it, so I need to say something about this third collop of “confirmatory” evidence before moving on. Neiman, director of archeology at Monticello at the time and member of the TJMF research committee, follows Winthrop Jordan in noting that Jefferson was at Monticello nine months prior to the birth of each of Sally Hemings’s children and that there is no
evidence of any other one person having been present nine months prior to the birth of each child. He writes, “[U]ntil now, the significance of this finding has rested on personal intuition.” Neiman ambitiously “outlines a quantitative means of combining that estimate with other evidence to produce an overall assessment of the probability that Jefferson fathered all of Hemings’ children.”11 If someone other than Jefferson was the father, Neiman asks, what is the probability of each of Hemings’s conception dates being within a period when Jefferson was at Monticello, not elsewhere? He asserts that Jefferson was home roughly 50 percent of the time since Sally’s first and last pregnancy, and so the scenario is comparable to flipping a fair coin, with the exception that coin flips are independent events. That is neither the case with Jefferson’s stays are Monticello, which were not random—being at Monticello on a given day makes the likelihood of being there the next day greater—nor with Hemings’s conception dates—getting pregnant one day makes it impossible to get pregnant the next. Neiman opts for the Monte-Carlo statistical method—a method of simulation-based inference, usually used for highly complex systems (e.g., physical systems) for which deterministic algorithms are unsuited, because there is considerable uncertainty about the inputs. It is a method that uses computer simulation to generate a relative-frequency probability assessment. Repeated random samplings take the place of a-priori calculation. Computers are employed to expedite the number of random tries. Consider a non-complex example—one who wants to figure out the probability of “six” occurring in one roll of a die. According to the a priori approach, on assumption of all outcomes being equiprobable, one merely places the expected outcome over all possible outcomes to arrive at the probability assessment of 1/6 or 0.167. Yet what if one has good reason to believe that the die is loaded, but has no means to ascertain just how? One then might approach the problem by rolling the die a large number of times— say, 1000—and simply guaging the probability by the frequency of successful outcomes over possible outcomes. That will give some indication of the bias of the die—the more rolls, the better the indi-
cation. The Monte-Carlo method is similar, only that the rolls of the die (or whatever outcome for which one desires a probability assessment) are done by a computer in an effort to generate quickly large amounts of data. Neiman does something similar with the probability of Jefferson being the father of any or all of Hemings’s children. He constructs four models, with slightly different parameters, and comes up with the following distribution-schemes, given the record of Jefferson’s stays at Monticello and the birthdays of Hemings’s children. The table he titles “Relative Frequency Distributions for the Number of Conceptions that Fall during or Three Days before a Jefferson Visit, for the Four Monte Carlo Models.” Model 1 Model 2 Model 3 Model 4
0.0 0.0 0.1 0.0 0
3.6 0.0 6.3 4.3 1
19.1 13.8 24.6 18.3 2
37.3 37.7 36.8 34.3 3
29.2 34.3 24.2 30.6 4
9.7 12.7 7.2 11.1 5
1.2 1.5 0.8 1.3 6
Given these distribution schemes, Neiman then uses Bayes’s theorem to derive a probability assessment for Jefferson being the father of all six children. p(J/e)
= p(J) x p(e/J) [p(J) x p(e/J)] + [p(~J) x p(e/~J)]
Given an extremely low a priori probability of 0.05 that Jefferson was the father of all six children [(J), assumed arbitrarily, just to get the ball rolling] and given a probability of 0.012 that Jefferson was not the father of all six children though he was present each time at conception (e/~J), Neiman arrives at an 84 percent a posteriori likelihood that Jefferson was the father of all six—a probability that increases commensurate with an increase in the a priori probability (e.g., an a priori probability of 0.10 generates an a posteriori probability of 92 percent and 0.50 a priori probability generates near certainty). On assumption of a very low prior probability of Jefferson being the father of all six children, we ultimately
arrive at an extremely high probability that he was, given the coincidence of him being present at Monticello nine months prior to the birth of each child. (Note how that coincidence drives the argument.) Moreover, the low a priori probability of Jefferson being the father of all six children that was given at the start does not take into consideration any other evidence that might also implicate Jefferson—e.g., DNA evidence. Factoring that evidence, the prior probability increases commensurately. At some point, it becomes ridiculous to consider Jefferson’s stays at Monticello and Hemings’s conception dates as independent. Coincidence is best explicated by causal relationship. Jefferson fathered all six children. The argument shows savvy, but it succumbs ultimately to what might be dubbed the fallacy of loading the deck—viz., it works only by assuming what needs to be assumed to gets the needed results. First, even if Jefferson was at Monticello nine months prior to the birth of each child, as others have pointed out, one gets the high degree of coincidence only if Sally Hemings was at Monticello every time she became pregnant. We know nothing of her whereabouts each time, and one cannot just assume she was at Monticello each time. Second, the birth dates are taken from Jefferson’s Farm Book, yet little is known about the precise time each birth was noted, because Jefferson was not present for all the births and, thus, likely obtained that information from Edmund Bacon or Sally Hemings. Much rides on that, since conception dates by Jordan were ascertained by working backward from the birth dates in the Farm Book. 12 Third, the argument says nothing about the 16 times Jefferson was at Monticello, between the conceptions of the first and last child, and Hemings did not get pregnant. Fourth, the argument merely takes for granted that other possible Jeffersonchromosome paternity candidates were not present at each conception. Finally, and this point attends on the last, Neiman’s argument works on assumption of there being one and only one father for all of Hemings’s children. That is gratuitous. There are other difficulties with the argument, which I address in chapter 5 of Framing a Legend: Exposing the Distorted History
of Thomas Jefferson and Sally Hemings, so I shall not iterate them here. Neiman ends his paper with unwarranted gasconade. “Serious doubt about the existence and duration of the relationship and about Jefferson’s paternity of Hemings’s six children can no longer be reasonably sustained.” 13 Neiman has stacked the deck. He has not vetted his data for accuracy and has not considering all the relevant data. Ken Wallenborn, the sole dissenting member of the TJMF committee, gives the reason why. On working through his argument, Neiman rushed into a room in which Wallenborn and another committee member sat, and excitedly ingeminated, “I’ve got him!” That was, as Wallenborn notes, “an inappropriately enthusiastic remark for someone who [was] working at Thomas Jefferson’s home.” 14 I add that it is a remark unpardonable for someone who is supposed to be engaged in objective research. Moreover, what sort of thrasonic thrill does one get by pillorying a man that has given so much of himself to his fellow human beings? 15 In sum, statistical arguments are only as good as the data that go into them. Let us go back to the loaded-die illustration of the Monte-Carlo method. One cannot run, with any assurance of legitimate results, a large number of random computerized rolls of that die on any computer to get information on just how that die is loaded unless first a sufficient amount of accurate data about just how the die is loaded is plugged into the computer. Contaminate the data and one contaminates the results. 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. Neiman does address one of the difficulties I have listed—the presumption that Hemings was at Monticello each time she got pregnant. To conclude that Sally Hemings must have been at Monticello when there is no evidence to show she was not is questionbegging. Neiman acknowledges that his conclusion applies to any other person, a Jefferson Doppelgänger, who might have been around Monticello each time Hemings was impregnated. “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.” He merely adds flippantly, “The chances that such a Jefferson doppelganger [sic] existed are, to say the least, remote.” 16 If we take “the probabilities that result apply to Jefferson or any other individual with identical arrival and departure dates” literally, we could then conclude that that there is a 99 percent probability that Jefferson fathered all six of Hemings’s children and a 99 percent probability that a Jefferson Doppelgänger, if one exists, fathered all six, which is impossible. Both sets of data would have to be accommodated and that would diminish the probabilities considerably. Defects notwithstanding, Neiman’s argument is uncritically cited by historians 17 as additional evidence of the scientific sort for a liaison. Jan Lewis for example writes, “Fraser D. Neiman’s ingenious statistical evaluation of the relationship between the pattern of Jefferson’s returns to Monticello and Hemings’ conceptions should quiet those who have resisted accepting Jefferson’s paternity.” 18 In reply to Neiman’s arrogant boast, “I’ve got him!” I merely reply, “No you haven’t.”
“As Well Established as … Many More Things in History” We have an imbroglio. As noted, numerous scholars, Burstein and Gordon-Reed are two of the most prominent, changed their mind after the DNA study. Yet scholars and scientists alike acknowledge that the DNA evidence is inconclusive. What is needed is something in addition to turn the tide in favor of an affair. That something is said to comprise “mounting circmstantial evidence” and a more measured approach to interpreting it. Yet there has not been new historical data added since the DNA study—at least, nothing ad rem. Why did so many scholars change their minds? The issue is perplexing. One can only conclude that the misleading manner in which the DNA study was written up along with its irresponsible title buffaloed numerous scholars, unfamiliar with biogenetic data. They were quick to jump uncritically on the Jefferson-did-it bandwagon and, once on the bandwagon, they felt compelled to stay on it, perhaps for fear of ridicule due to wishy-washiness. I suspect that is
the case with Burstein. Once an advocate of Jefferson’s innocence in the affair, he has become an advocate of Jefferson’s guilt. Were he to change his mind once more, he might never again be taken seriously as a historian. For Gordon-Reed, the DNA evidence gave her an opportunity to exploit Brodie’s thesis. With Jefferson being a candidate for paternity of Eston Hemings, there was at least something upon which she could hang a hat. Following the advice of the TJMF, she was at liberty to give free expression to her often wild speculations. It is important here also to say something about the DNA evidence in conjunction with the thin historical data, for most scholars plainly admit that neither by itself is compelling. If the scholars about whom Gordon-Reed has been critical, she asserts in Thomas Jefferson and Sally Hemings, would have merely said, “I don’t believe that Thomas Jefferson and Sally Hemings had a relationship … [b]ut there is no proof one way or another,” then “that would have been acceptable.” The problem is that scholars go the extra step and add, “Now let me tell you why the story cannot be true.” 19 This book was published in 1997. Consider what she says after the published DNA report not even two years later in an interview with Brian Lamb of C-SPAN Booknotes: [T]he results of a DNA test that was conducted after I wrote the book came in to suggest that one of the descendants of Sally Hemings was, in fact, a Jefferson, and so it lead a number of people—and myself included—to come to the conclusion that, coupled with the information that I write about and, you know, putting the DNA test together, that this story is true, as wellestablished as most—many more things in history have been established. So what has happened is that a sort of intellectual discussion about this topic has been supplemented by science in a way that, you know, is sort of unprecedented.20 Two things are worth noting. First, there is Gordon-Reed’s wishywashiness. She asserts the story is true, qualifies that by adding “as well-established as most…,” but again hedges, changes her mind, and says “…many more things in history [that] have been estab-
lished.” Such disingenuousness is frightening. Second and more to the point, she admits the DNA evidence alone is inconclusive: It merely shows that one cannot dismiss Jefferson in the paternity of Eston Hemings. Moreover, she admits in her earlier book, “But there is no proof one way or another.” That is explicit acknowledgement that the historical data are at best gauzy, immaterial, and whippy, at worst, nonexistent. In sum, nothing added to not much equals not much. There is a much more Brobdingnagian concern. Just how do the two sets of data fit together? Numerous scholars seem merely to assume that the two sets of data can merely be conjoined in some manner such that the probability of Jefferson’s involvement in an affair is thereby additively increased. That cannot be the case. If biological evidence should surface that implicates Jefferson with a great amount of probability, then historical evidence is unneeded. Jefferson, then, is very likely the father of Eston Hemings. Should overwhelming historical evidence surface that implicates Jefferson, then biological evidence is de trop. However the point is that the two sets of data do not work together. They are independent of each other. They form independent arguments. For illustration, let us consider the claim Fried calamari is the best appetizer for any meal. Upon what evidence is the claim founded? Let us assume three distinct arguments with three sets of data. First, several acclaimed food critics acknowledged a preference for fried calamari as an appetizer, among all other appetizers. Yet, since there is nothing near unanimity among noted food critics, we assign that claim a probability of, say, 0.17. Second, we find that more Europeans prefer fried calamari to any other appetizer. Since we know nothing about the preferences of nonEuropeans, we assign that claim a probability of 0.12. Finally, more cookbooks worldwide have recipes for fried calamari under their section “Appetizers” than they have for any other appetizer. Since this is the best signifier of global tastes, we assign this the highest probability: 0.33. Now, one cannot merely add these probabilities and assert There is a 0.62 probability that fried calamari is the best appetizer for any meal. That would be like knotting together, end to end, three weak pieces of rope and then trying to
pull a heavy truck with the rope. The rope would doubtless break. In short, all three arguments are independent. The best one can say is that we have two pathetically weak arguments and one weak argument. Nonetheless, scholars continually take the two bodies of evidence as dependent, not independent. Following the leads of the DNA committee and the TJMF, they assert without critical assessment that neither by itself amounts to much, but that the two together implicate Jefferson. “Spurred by the further confirming evidence of DNA testing in 1998,” writes Lucia Stanton, “historians will continue to contemplate the nature of the probable relationship between Sally Hemings and Thomas Jefferson.”21 Andrew Burstein writes of his failure to take seriously the “alternative picture” in The Inner Jefferson. The DNA evidence has narrowed down credible paths of historical exploration. With obvious hyperbole, he states, “Perhaps the DNA findings have not absolutely made Thomas Jefferson the father of his house servant’s children, but mounting circumstantial evidence makes him by far the most plausible father of these children, as most would now agree.” 22 He concludes thrasonically, “For most, Thomas Jefferson’s paternity of Sally Hemings’s children has now been convincingly established, yet there are those who insist on reminding us that it has not been absolutely established.” 23 R.B. Bernstein writes of the convergence of the DNA results, “close analysis of circumstantial evidence and oral tradition in Annette Gordon-Reed’s book,” and Fraser Neiman’s Bayesian argument in which he concludes “mathematically” in Bernstein’s words “the odds against anyone but Jefferson being the father were ten thousand to one.” He too mistakenly concludes that the independent arguments—here three and each weak—can be fashioned together to create a watertight case. 24 E.M. Halliday flippantly writes: “[I]n the autumn of 1998 careful DNA tests revealed a high scientific [sic] probability that Jefferson had indeed fathered the last one of Sally’s children, which along with circumstantial evidence strongly suggested that he fathered them all.” 25 Gordon-Reed says: “[T]he situation isn’t just about what the DNA test says. It has to be seen—the totality of the circumstances have to be considered as well.” 26 The prevailing
sentiment in all such instances is that the biological and historical evidence can added together such that together they yield high probability. Hogwash! The DNA study shows only that Jefferson is a possible candidate for paternity. The historical evidence in favor of a liaison is sinewy, slight, and circumstantial. Circumstantial evidence is inference-based, not incriminating. In this case, the inferences made by maligners and makebates are abundant. The weight of historical evidence favors Jefferson’s defenders, but it is not decisive. Agnosticism is the only rational alternative. Rationality, unfortunately, seems not to be a high priority for those insistent on pillorying Jefferson. Noting the independence of the biological and historical investigative strands is itself an opportunity for clarity on the issue. Imagine a future scenario both in which historical research turns up evidence that cogently supports Jefferson’s involvement with Sally Hemings and paternity of Eston Hemings and in which biological evidence independently and cogently implicates Jefferson in the paternity of Eston Hemings. Here the probability of two independent investigative strands reaching the same conclusion and offering cogent arguments is relatively low. Consequently, there would be good reason to accept the truth of the conclusion that Jefferson was the father of Eston. That, however, is nothing like what we have here. We are left with not much, and not much applies to both defenders and accusers of Jefferson. We are at an impasse. The biological arguments and historical arguments are strictly speaking independent lines of inquiry and offer independent means of evaluating the claim Thomas Jefferson was the father of Eston Hemings. Both lines concern probabilities; neither deals with certainties. The relevant evidence is exiguous, scrimpy. Nonetheless, for all intents and purposes, there is a sense in which the historical inquiry can lean on the biological data. Consider the hypothesis, Thomas Jefferson fathered Eston Hemings. The claim, The Y chromosome from the descendant of Eston Hemings will be identical to that of the descendants of Field Jefferson, follows as a deductive consequence and functions a prediction of the hypothesis to be confirmed or disconfirmed through ex-
periment. As it turns out, the samples are identical. Thus, the prediction is confirmed. Schematically, in hypothetico-deduction fashion: 1. If Thomas Jefferson fathered Eston Hemings (hypothesis), then the Y chromosome from the descendant of Eston Hemings will be identical to that of the descendants of Field Jefferson (prediction). 2. The then the Y chromosome from the descendant of Eston Hemings is identical to that of the descendants of Field Jefferson (experimental result). 3. Thus, Thomas Jefferson fathered Eston Hemings is confirmed. We must be guarded about what is being shown here. Confirmation of any scientific hypothesis does not amount to proof. It offers no certainty. It cannot even guarantee probability. It merely means roughly that we now have more reason to believe the hypothesis is true than we did prior to the experiment. Plug in Randolph Jefferson, Thomas’s brother, or anyone else with the Jefferson Y chromosome and the confirmation is nearly the same. The certainty from the DNA study lies not in knowing that Jefferson fathered Eston Hemings, but in knowing with a greater than 0.99 percent probability that the Y-chromosome match is caused by someone with the Jefferson haploid being the father of Eston Hemings. Given that bit of knowledge, it is wholly sensible for historians to look for evidence of paternity among those reasonable candidates with the Jefferson Y chromosome, not elsewhere. In that sense, and only in that sense, can history lean on biology. Thomas Jefferson, of course, would certainly be among the most likely candidates, if he would not be the most likely candidate, to investigate rigorously. Yet, as I iterate, being the most likely candidate among all others does not necessarily make it probable that he fathered Eston Hemings. It makes him merely the leading candidate among many plausible candidates.
Upshot
What seems to have happened is this. Prior to the DNA study, most scholars were convinced that there was little likelihood that Jefferson was the father of Eston Hemings, or any of Sally’s children. Out came the DNA study with its midleading title and paralogistic conclusion concerning Jefferson being the simplest and most probable explanation of Eston’s paternity. The media ran with the title and the scientists’ “consensus,” implicating Thomas Jefferson. The general public was swept away by overwhelming current of the media blitz. So too were the scholars. When the dust cleared and scholars took a long look at the DNA evidence, most were perhaps too swept away to return to the sobriety of agnosticism. They comforted themselves with the TJMF’s committee report, which made Jefferson the very likely father of all of Hemings’s children—cold comfort indeed given the biases of some committee members. Burstein in Jefferson’s Secrets apologized for his prior shortsightedness and argued that more intimate inspection of all available evidence showed Jefferson’s duplicity—his tendency to act on impulse and rationalize his actions. He not only had a lengthy affair with Hemings, but it was also only an in-chambers affair. Others, like Gordon-Reed, used the DNA study as a platform for a social-justice agenda—something prima facie laudable, but history ought to be truth-driven, not agendadriven. By showing that the statesman and slave had a loving, mutually fulfilling relationship for some 38 years, Sally Hemings becomes more than a slave and Jefferson becomes merely a man. Or as Gordon-Reed says in her evaluation of Barbara Chase-Riboud’s novel Sally Hemings: “In a small but important way, the humanity of Hemings [in the novel] is reemphasized. Jefferson’s humanity comes back into focus too. She is raised. He is cut down to size. Thus, two of the major requirement of black progress (restoration of black humanity and obliteration of the cult of the godlike white person) are fulfilled.” 27 The phrasing suggests strongly much more than equalitarian intention. Jefferson, being “cut down to size,” suggests comeuppance—comeuppance fueled by execration. This unfortunate scenario needs to be remedied for the sake of scholarly integrity. Cannons of rationality cannot be ignored or vil-
ipended, when found convenient. Much more investigative work needs to be done. That is not investigative work on “[t]he implications of the relationship between Sally Hemings and Thomas Jefferson … to enrich the understanding and interpretation of Jefferson and the entire Monticello community,” as the TJMF committee report says. 28 There is no unambiguous, cogent evidence of a relationship. Until such unambiguous and cogent evidence is found to decide the issue one way or another, it is as sensible to explore the implications of the putative relationship as it is to examine the nature of Bigfoot. Otherwise the cart drives the horse.
Notes 1
Now the Thomas Jefferson Foundation. Eugene A. Foster, M. A. Jobling, P. G. Taylor, P. Donnelly, P. de Knijff, Rene Mieremet, T. Zerjal, and C. Tyler-Smith, “Jefferson Fathered Slave’s Last Child,” Nature, 396, 5 Nov. 1998, 27. 3 Eugene A. Foster, et al., “Jefferson Fathered Slave’s Last Child,” 27-8. 4 White McKenzie Wallenborn, “A Committee Insider’s Viewpiont,” The Jefferson-Hemings Myth: An American Travesty, ed. Eyler Robert Coates, Sr. (Charlottesville: The Thomas Jefferson Heritage Society, 2001), 62-7. 5 See David Murray, “Anatomy of a Media Run-Away,” The Jefferson-Hemings Myth: An American Travesty, ed. Eyler Robert Coates, Sr. (Charlottesville: The Thomas Jefferson Heritage Society, 2001), 37-46. 6 R.B. Bernstein, Thomas Jefferson (New York: Oxford University Press, 2003), 196 7 Lucia Stanton, “Those Who Labor for My Happiness”: Slavery at Thomas Jefferson’s Monticello (Charlottesville: University of Virginia Press, 2012), 64-5. 8 Lucia Stanton, “Those Who Labor for My Happiness,” 94. 9 Eugene A. Foster, et al., “Jefferson Fathered Slave’s Last Child,” 27-8. 2
10
For difficulties with Madison’s testimony, see M. Andrew Holowchak, Framing a Legend (Amherst, NY: Prometheus Books, 2012), chap. 1 and chap. 2. 11 Fraser D. Neiman, “Coincidence or Causal Connection? The Relationship between Thomas Jefferson’s Visits to Monticello and Sally Hemings’ Conceptions,” William and Mary Quarterly, Vol. LVII, No. 1, 2000, 199-206. 12 William Hyland, Jr., et al., “Civil Action: Sally Hemings v. Thomas Jefferson,” American Journal of Trial Advocacy, Vol. 31, No. 1, 17, 17-21. 13 Fraser D. Neiman, “Coincidence or Causal Connection?” 210. 14 White McKenzie Wallenborn, “A Committee Insider’s Viewpiont,” 14. 15 See M. Andrew Holowchak, Dutiful Correspondent: Philosophical Essays on Thomas Jefferson (Lanham, MD: Rowman & Littlefield, 2012). 16 David Murray, “Present at the Conception,” 119-21. 17 E.g., R.B. Bernstein, Thomas Jefferson (Oxford: Oxford University Press, 2003), 196, and Jan Lewis, “Introduction,” The William and Mary Quarterly, Vol. 57, No. 1, 2000, 121-4. 18 Jan Lewis, “Introduction,” The William and Mary Quarterly, Vol. 57, No. 1, 2000, 122. 19 Annette Gordon-Reed, Thomas Jefferson and Sally Hemings: An American Controversy (Charlottesville: The University Press of Virginia, 1997), 226. 20 http://www.booknotes.org/FullPage.aspx?SID=119003-1, accessed 10 August 2012. 21 Lucia Stanton, “Those Who Labor for My Happiness,” 179. 22 Andrew Burstein, Jefferson’s Secrets: Death and Desire at Monticello (New York: Basic Books, 2005), 182 23 Andrew Burstein, Jefferson’s Secrets, 114-5. 24 R.B. Bernstein, Thomas Jefferson, 196. 25 E.M. Halliday, Understanding Thomas Jefferson (New York: HarperCollins Publishers, 2001), xi-xii.
26
http://www.booknotes.org/FullPage.aspx?SID=119003-1, accessed 10 August 2012. 27 Annette Gordon-Reed, “Engaging Jefferson: Blacks and the Founding Father,� The William and Mary Quarterly, Vol. 57, No. 1, 2000, 180. 28 http://www.monticello.org/site/plantation-and-slavery/icommittee-charge-and-overview, accessed 9 June 1012.