Abstract
Public and institutional discourse increasingly treats three distinct propositions as a single settled fact: that Thomas Jefferson fathered the children of Sally Hemings, that the 1998 Y-chromosome study demonstrated this, and that Jefferson and Hemings concluded a negotiated agreement in Paris in 1789 securing freedom for her future children. This paper disaggregates the three claims and assigns each its proper evidentiary weight. It argues that a Jefferson male fathered Eston Hemings is well established; that Thomas Jefferson specifically fathered all six recorded children is probable on convergent circumstantial grounds but not demonstrated; and that the Paris agreement rests on a single late, secondhand, editorially mediated source and cannot bear the confidence now routinely placed upon it. The paper further contends that the corrective against historical dismissal of testimony from enslaved persons is properly served by applying ordinary source criticism evenly, not by exempting any testimony from it — an exercise that damages the Randolph family’s counter-narrative far more than it damages Madison Hemings’ account.
1. The Problem of Compression
The Jefferson–Hemings question has passed through three phases in a generation. Before 1997, the dominant scholarly position, articulated most influentially by Dumas Malone, held the allegation to be a partisan slander unworthy of serious treatment (Malone, 1970). Between 1997 and 1998, Annette Gordon-Reed’s forensic reexamination of the documentary record shifted the burden by demonstrating that the case against paternity rested on evidence at least as weak as the case for it (Gordon-Reed, 1997). After November 1998, a genetic finding reframed the debate entirely.
What has happened since is compression. Three propositions of unequal strength have fused into one, and the confidence properly attaching to the strongest has been silently transferred to the weakest. The propositions are:
P1. A male of the Jefferson line fathered Eston Hemings (b. 1808).
P2. Thomas Jefferson himself fathered Eston, and by extension the five other children Sally Hemings bore between 1795 and 1808.
P3. Jefferson and Hemings concluded an agreement in Paris in 1789 under which she consented to return to Virginia in exchange for privileges and a pledge that her children would be freed at age twenty-one.
P1 is near-certain. P2 is probable. P3 is a family tradition recorded once, sixty-four years after Jefferson’s death, by a man born sixteen years after the events described. These are not the same order of claim, and the historiographical cost of treating them as one is that the discipline’s own standards become unavailable for use on the question — a cost borne disproportionately by whichever side of a future dispute happens to hold the weaker documentary hand.
2. What the Genetic Evidence Established
2.1 The finding
Eugene Foster and collaborators compared Y-chromosome haplotypes across male-line descendants of Field Jefferson (Thomas Jefferson’s paternal uncle), of John Carr (grandfather of Peter and Samuel Carr), of Thomas Woodson, and of Eston Hemings. The Jefferson and Eston Hemings lines matched on a rare haplotype. The Carr lines did not match. The Woodson lines did not match (Foster et al., 1998).
Three results follow. First, a Jefferson male fathered Eston. Second, the Carr brothers — named by Jefferson’s grandchildren as the guilty parties — are excluded. Third, the Woodson family’s claim of descent from the “Tom” of the original 1802 allegation is unsupported.
2.2 What the finding did not establish
Thomas Jefferson was not tested and cannot be. He left no surviving legitimate male-line descendants; the test proceeded through his uncle’s line precisely because his own was unavailable. The Y chromosome is shared by every male in the patriline and cannot discriminate among them. At the relevant date, roughly twenty-five adult Jefferson males were living in Virginia, of whom perhaps eight were within plausible reach of Monticello (Turner, 2011).
The Nature headline — asserting flatly that Jefferson fathered his slave’s last child — was not the authors’ formulation and misstated the result. Foster published a clarification within two months acknowledging that the data were consistent with paternity by any Jefferson male and that the title overstated the certainty of the inference (Foster et al., 1999; see also Abbey, 1999). The accompanying commentary by Lander and Ellis (1998), which linked the finding to contemporaneous political controversy, further encouraged the conflation.
2.3 The alternative-candidate problem
The most frequently advanced alternative is Randolph Jefferson, Thomas’s younger brother, resident about twenty miles distant at Snowden. Isaac Granger Jefferson, formerly enslaved at Monticello, recalled that Randolph would come among the enslaved community in the evenings to play the fiddle and dance (Bear, 1967). Thomas wrote to Randolph in August 1807 inviting a visit, within a plausible conception window for Eston; whether the visit occurred is unrecorded. Randolph’s several sons were also of age by that date.
This objection has genuine force, and it is not adequately met by observing that no positive evidence places Randolph at Monticello at the moment in question — the same absence of positive documentary evidence characterizes the affirmative case. What limits the objection is its scope: it addresses Eston, and by extension perhaps Madison (b. 1805), but Randolph’s sons were children when Beverly (1798) and Harriet (1801) were conceived, and no evidence of any kind places Randolph at Monticello across the full 1795–1808 span. An account that resolves one birth while leaving five unexplained is weaker than one that resolves all six, and the paternity of all six by a single father, while not proven, is the natural reading of a stable household arrangement.
3. The Circumstantial Case for Thomas Jefferson
The case rests on convergence. No single element is decisive; the question is whether the elements are more economically explained together than separately.
3.1 Presence correlation
Jefferson recorded Hemings’ children in his Farm Book (Betts, 1953). He was resident at Monticello during every plausible conception window for all six births, and no conception occurred during his prolonged absences — absences that were frequent and often lengthy during his vice-presidency and presidency. Neiman (2000) subjected this pattern to Monte Carlo simulation and estimated the probability of its arising by chance at approximately one percent.
The critique is legitimate and should be stated. Six observations is a thin base for a probabilistic claim. Household activity of many kinds clustered around the master’s periods of residence, since the house population and its visitors expanded then; the correlation may reflect opportunity for a class of men rather than for one man. Nonetheless, no other Jefferson male is known to satisfy the presence condition across the whole period, and Thomas Jefferson satisfies it perfectly.
3.2 The manumission anomaly
Jefferson freed almost no one. Over a lifetime of holding more than six hundred people in bondage, he manumitted or permitted the departure of roughly ten, and every one of them was a Hemings (Stanton, 2012). Beverly and Harriet left Monticello in 1822 without pursuit — Harriet, by his overseer’s recollection, given money and put on a northbound stage (Pierson, 1862). Madison and Eston were freed by the will of 1826.
This pattern demands explanation. It does not, however, uniquely require the paternity explanation. The Hemings family were in all likelihood the children of John Wayles, Jefferson’s father-in-law, and therefore half-siblings of his late wife Martha. A general policy of preference toward his wife’s kin accounts for much of the pattern without recourse to a further hypothesis (Stanton, 2000). The paternity thesis explains the selection within the Hemings family — why these four and not their cousins — somewhat better than the alternative does, but the margin is narrower than it is often represented to be.
3.3 The collapse of the family counter-testimony
The Randolph descendants’ account, long treated as authoritative, does not survive examination. Ellen Randolph Coolidge, in an 1858 letter, attributed the Hemings children to Samuel Carr; Thomas Jefferson Randolph, in conversation with the biographer Henry Randall, attributed them to Peter Carr. They contradict one another, and the 1998 study excludes both (Gordon-Reed, 1997; Foster et al., 1998). Martha Randolph is reported to have asserted that Jefferson was absent from Monticello for a period fatal to the paternity claim; the Farm Book does not sustain the assertion. Edmund Bacon’s memoir states that he personally observed another man leaving Hemings’ quarters in the early mornings — but Bacon did not begin service at Monticello until 1806, after four of the six children were born, and the memoir was taken down more than fifty years later (Pierson, 1862).
The collapse of the counter-testimony does not establish the affirmative claim. It does remove the principal historiographical reason for which the affirmative claim was rejected for a century and a half.
3.4 The unreliability of the originating allegation
The public charge originated with James Callender in the Richmond Recorder in September 1802. Callender was a paid political writer nursing a specific grievance — Jefferson’s refusal to appoint him postmaster at Richmond — who never visited Monticello and named no source. The single testable element of his account fails: he described a son “Tom,” roughly twelve years old, and no such person appears in Monticello’s meticulous records. The Woodson descendants’ haplotypes do not match the Jefferson line (Foster et al., 1998; Rothman, 2003).
That a hostile and unreliable source approximated the general shape of a situation is weak corroboration at best. Callender’s charge belongs to the history of the controversy, not to the evidence for it.
3.5 Assessment
The Thomas Jefferson Memorial Foundation’s research committee concluded in 2000 that the weight of evidence indicated a high probability of Jefferson’s paternity of Eston and most likely of all six children (Thomas Jefferson Memorial Foundation, 2000). The Scholars Commission convened by Robert Turner concluded the opposite, twelve of thirteen members judging the allegation unproven, with a single dissent from Wallenborn (Turner, 2011). Both bodies carried institutional and reputational commitments that a careful reader should weigh.
The defensible formulation is that Thomas Jefferson’s paternity is the best available explanation of the assembled evidence, held with the confidence appropriate to a strong inference to the best explanation — not the confidence appropriate to a document. Historians should be able to say this without embarrassment. Much of what is known about the eighteenth century is known on exactly this footing.
4. The Paris Agreement
Here the evidentiary floor drops away, and the disparity with P1 and P2 is severe.
4.1 The single source
The claim originates entirely in the memoir of Madison Hemings, published 13 March 1873 in the Pike County Republican of Waverly, Ohio, as the first installment of a series titled “Life Among the Lowly” edited by S. F. Wetmore (Hemings, 1873). Madison stated that his mother became pregnant by Jefferson while in France, that she understood herself to be free under French law and refused to return, and that Jefferson induced her return by promising extraordinary privileges and pledging that her children would be freed at twenty-one.
The source is subject to four limitations, none of which is a moral judgment on the witness:
Chronological. Madison Hemings was born in 1805. The negotiation he describes occurred in 1789. He is transmitting maternal testimony at a remove of eighty-four years from the event and roughly thirty-eight years from his mother’s death.
Mediated. Wetmore was a Republican partisan assembling a series with a discernible purpose. The extent of his hand in the prose cannot be measured. Some phrasing is idiomatic to nineteenth-century journalism rather than to speech.
Uncorroborated. No contemporary document of any kind supports the agreement. Jefferson’s Paris accounts are detailed and survive; they contain no reference to it. No letter, memorandum, or third-party observation attests it.
Unattested in its foundational premise. The Paris pregnancy — the leverage on which the entire bargain is said to rest — is recorded nowhere but in this memoir. Madison stated the child died soon after birth. No record of the birth exists.
Israel Gillette Jefferson’s account, published in the same series later that year, offers general support but is compromised by shared provenance and by claims of knowledge he was not positioned to possess (Jefferson, 1873).
4.2 Consistency with subsequent events
The pattern of later manumissions partially fits. Harriet departed at about twenty-one; Madison and Eston were freed at or near twenty-one. But Beverly appears to have departed at roughly twenty-four, and Sally Hemings herself was never legally freed — Martha Randolph gave her “her time” informally around 1827, an arrangement short of manumission (Stanton, 2000). A negotiator who secured a binding pledge for her children and nothing whatever for herself is not impossible, but the fit is looser than summaries suggest.
4.3 The James Hemings analogy
The strongest argument for the plausibility of the agreement is that Jefferson demonstrably made an arrangement of exactly this character with Sally’s brother. James Hemings, likewise in Paris and likewise able to petition for freedom under French law, entered into a written agreement in September 1793: he would return to Virginia and train a successor in French cookery, after which he would be freed. Jefferson honored it in 1796 (Gordon-Reed, 2008; Stanton, 2000).
This raises the prior probability that a similar arrangement existed with Sally. It supplies no evidence that one did. And a pointed asymmetry deserves notice: Jefferson wrote the James Hemings agreement down. He recorded very nearly everything. The absence of any written trace of a parallel arrangement is not conclusive — an arrangement of this kind would have been the one thing he had reason not to record — but the argument that its absence is explained by its embarrassment is an argument that cannot be tested, and unfalsifiable arguments should not be counted as support.
4.4 The simpler alternative
There is a plainer account of Sally Hemings’ return that requires no bargain. She was sixteen, in a country whose language she was still acquiring, and her entire family — mother, brothers, sisters, nieces, nephews — was in Virginia. Freedom in France meant permanent separation from every person she knew and an isolation without resources. Historians who assume a negotiated agreement should be asked to explain what the assumption accomplishes that this does not.
5. Source Criticism Applied Evenly
The historical profession’s long refusal to credit testimony from formerly enslaved persons was an intellectual and moral failure, and its correction is among the more durable gains of recent historiography. Gordon-Reed’s contribution was precisely to insist that Madison Hemings be read by the same rules as Ellen Coolidge (Gordon-Reed, 1997).
The point deserves emphasis because it is frequently misapplied. Even-handed criticism does not exempt Madison Hemings from scrutiny; it subjects him to the same scrutiny as everyone else. Under that scrutiny, he fares comparatively well. His account is late and secondhand — but it is internally coherent, it is not contradicted by any document, and its testable elements survive testing. The Randolph family’s account is also late, is internally contradictory as between two named perpetrators, and is refuted in its specifics by the 1998 study. Where the two conflict, the historian has better reason to prefer Madison Hemings.
Faring better than a discredited alternative is not the same as being documentation. A source may be the best available on a question and still be insufficient to establish it. Stanton and Swann-Wright’s work on the transmission of oral tradition within the Monticello families demonstrates both the durability of such memory and the ways it consolidates, compresses, and acquires narrative shape across generations (Stanton & Swann-Wright, 1999). Both findings apply here. The tradition is worth taking seriously, and taking it seriously includes recognizing what it is.
6. Conclusions
On P1. A Jefferson male fathered Eston Hemings. This is established to the standard of genetic evidence and should be stated without hedging.
On P2. Thomas Jefferson is by a wide margin the most probable father of all six of Sally Hemings’ recorded children. The conclusion rests on convergent circumstantial evidence — presence correlation, the manumission anomaly, and the failure of every proposed alternative to account for the full sequence — reinforced by the collapse of the family counter-testimony. It is a strong inference and should be labeled as one.
On P3. The Paris agreement is a family tradition preserved in a single source, published eighty-four years after the events, by a witness born sixteen years too late to observe them, through an editor with an agenda, with no corroboration and with its foundational premise unattested. It is plausible, partly consistent with what followed, and unverifiable. It should be presented as tradition, attributed to its source, and not asserted as fact.
The general lesson is about the transfer of confidence. The genetic result was rigorous and narrow. In public transmission its rigor detached from its narrowness and migrated to propositions the study never addressed — most conspicuously to a Parisian negotiation about which the Y chromosome is entirely silent. Institutions that describe P3 in the declarative are borrowing certainty from a finding that does not extend to it, and doing so in a domain where the discipline’s credibility on contested questions is not a resource to spend lightly.
References
Abbey, D. M. (1999). The Thomas Jefferson paternity case. Nature, 397(6714), 32. https://doi.org/10.1038/16268
Bear, J. A., Jr. (Ed.). (1967). Jefferson at Monticello: Memoirs of a Monticello slave and Jefferson at Monticello. University Press of Virginia.
Betts, E. M. (Ed.). (1953). Thomas Jefferson’s farm book, with commentary and relevant extracts from other writings. American Philosophical Society.
Brodie, F. M. (1974). Thomas Jefferson: An intimate history. W. W. Norton.
Foster, E. A., Jobling, M. A., Taylor, P. G., Donnelly, P., de Knijff, P., Mieremet, R., Zerjal, T., & Tyler-Smith, C. (1998). Jefferson fathered slave’s last child. Nature, 396(6706), 27–28. https://doi.org/10.1038/23835
Foster, E. A., Jobling, M. A., Taylor, P. G., Donnelly, P., de Knijff, P., Mieremet, R., Zerjal, T., & Tyler-Smith, C. (1999). The Thomas Jefferson paternity case. Nature, 397(6714), 32. https://doi.org/10.1038/16266
Gordon-Reed, A. (1997). Thomas Jefferson and Sally Hemings: An American controversy. University Press of Virginia.
Gordon-Reed, A. (2008). The Hemingses of Monticello: An American family. W. W. Norton.
Hemings, M. (1873, March 13). Life among the lowly, No. 1. Pike County Republican.
Jefferson, I. G. (1873, December 25). Life among the lowly, No. 3. Pike County Republican.
Lander, E. S., & Ellis, J. J. (1998). Founding father. Nature, 396(6706), 13–14. https://doi.org/10.1038/23802
Malone, D. (1970). Jefferson the President: First term, 1801–1805. Little, Brown.
Neiman, F. D. (2000). Coincidence or causal connection? The relationship between Thomas Jefferson’s visits to Monticello and Sally Hemings’s conceptions. The William and Mary Quarterly, 57(1), 198–210. https://doi.org/10.2307/2674364
Onuf, P. S., & Lewis, J. E. (Eds.). (1999). Sally Hemings and Thomas Jefferson: History, memory, and civic culture. University Press of Virginia.
Pierson, H. W. (1862). Jefferson at Monticello: The private life of Thomas Jefferson, from entirely new materials [Recollections of Edmund Bacon]. Charles Scribner.
Rothman, J. D. (2003). Notorious in the neighborhood: Sex and families across the color line in Virginia, 1787–1861. University of North Carolina Press.
Stanton, L. (2000). Free some day: The African American families of Monticello. Thomas Jefferson Foundation.
Stanton, L. (2012). “Those who labor for my happiness”: Slavery at Thomas Jefferson’s Monticello. University of Virginia Press.
Stanton, L., & Swann-Wright, D. (1999). Bonds of memory: Identity and the Hemings family. In P. S. Onuf & J. E. Lewis (Eds.), Sally Hemings and Thomas Jefferson: History, memory, and civic culture (pp. 161–183). University Press of Virginia.
Thomas Jefferson Memorial Foundation. (2000). Report of the research committee on Thomas Jefferson and Sally Hemings. Monticello.
Turner, R. F. (Ed.). (2011). The Jefferson–Hemings controversy: Report of the Scholars Commission. Carolina Academic Press. (Original report issued 2001)
