QuickLesson 13: Classes of Evidence—Direct, Indirect & Negative

Thomas Jefferson's Farm Book

Evidence can be messy. Because it is a mental construct, it rarely gives us the clear and simple answers that we seek. Sources, by contrast, are physical; we can touch them, see them, smell them, hear them. Information is also physical, visible, audible. Evidence, however, is intangible. It’s only what we think certain information means. That’s all it can be―until we make something concrete from it by processing it and molding it into a meaningful and convincing form.

How do we do that?

Step One: Understand the Various Types of Evidence

Creating substance from the evidence we perceive, but others may not, requires us to understand the nature of evidence. At its core, evidence comes in three basic types: direct, indirect, and negative—the latter two being sometimes lumped together under the catchall label circumstantial. All three basic types can be used alone or in combination to provide “proof” of an identity, a kinship, a circumstance, or any other point critical to our research.

Direct Evidence

As the name implies, direct evidence is information that directly addresses the issue at hand. It’s not merely something relevant. It plainly offers an answer to a specific research question. It may not provide as complete an answer as we would like. It may not even provide an accurate answer. But it specifically states something about our problem. It is then up to us to decide its veracity and its weight.

Indirect Evidence

Much of the information we find does not provide an explicit answer to any problem we have defined, yet it can seem potentially relevant―especially if we are thoughtful researchers who watch for patterns and parallels within all resources that we use. This indirect evidence might support direct evidence we have found. It might weaken the direct evidence we want to believe. It might seem tangential although we can’t yet say how. Whatever role it plays, it carries no weight until and unless we combine it with other evidence to arrive at an answer or construct an argument for our conclusion.

Negative Evidence

Sherlock Holmes famously spoke of “the sound of the dog not barking.”  Nothing could more graphically define negative evidence. It’s the absence of what should happen under a given set of circumstances. A watch dog is expected to bark. If it doesn’t, then its very silence attests the likelihood of certain things.

Step Two: Define the Research Question

Whether any piece of information is evidence depends upon the research question we seek to answer. The same piece of information might be immensely valuable evidence for one research problem and totally irrelevant to another, even though it deals with the same person or event. As researchers, we gather information on a chosen topic; but not until we define a specific question can we determine whether and how a piece of information becomes relevant evidence. For example:

  • If we are biographers, we seek information that relates to a certain name within a certain sphere; then we attempt to determine whether that information relates to our person-of-interest or some other individual of the same name. Our research question might be: Is this John Smith, who was killed at the Battle of Wilson’s Creek, the same John Smith whose farm was bushwhacked by Quantrill’s Raiders? After we form this question, each new finding would be weighed against it to determine whether the new information provides evidence to answer our question.
  • If our research focuses upon an event—that Battle of Wilson’s Creek, let’s say—then our research question might be this: Why did the battle cause the Union to effectively lose control of Southwestern Missouri, considering that neither the Confederate nor Union forces could claim victory and both sides suffered equivalent losses? With this research question, we would likely search the relevant records for information that deals with psychological, political, and tactical aspects, as opposed to personal associations or identifying markers.
Step Three: Proceed with Research

Once our question is defined, each new piece of information is then evaluated against the question to determine its relevancy. If it is relevant, it constitutes evidence. If it does not shed light on our question, it is not evidence for that issue but might prove useful for another question.

Step Four: Correlate New and Old Findings

With each new piece of evidence, we ask: Does this support or contradict what we have already found? What does our body of evidence represent at this point? Do we have multiple pieces of direct evidence, all supporting one conclusion, or does some of it point in a different direction? Does all the compatible evidence trace back to the same origin or does each piece have independent origin? Do we have multiple pieces of indirect evidence that, together, suggest a possibility no record explicitly states? Is there negative evidence that niggles at us, whispering doubts we need to resolve?

How strong is the totality of the evidence we have accumulated? Does it support a conclusion? Does it point to new records? Does it suggest a new interpretation of our working hypotheses? In this stage of research, we weigh all pieces of evidence against each other, analyzing the strengths and weaknesses of each, individually and comparatively.

Step Five: Resolve Conflicts in the Evidence

No question is reliably answered until all conflicting evidence is logically and decisively resolved. Typically, this requires additional research because the correlation and analysis process will have spotlighted holes in the work we have done up to this point. Even if we think all resources have been exhausted, we often have to revisit materials already used—this time playing Devil’s Advocate, deliberately trying to disprove whatever hypotheses we’re forming.

Step Six: Organize our Evidence into a Logical, Written Argument

EE’s QuickLessons typically offer a case at point. However, case studies of QuickLesson length can never do justice to the sheer messiness of evidence or adequately demonstrate why we need a solid understanding of all its forms. So, in lieu of a simple example, QuickLesson 13 draws from a published case study that is an exemplar in the analysis of questionable sources, the correlation of conflicting information, and the evaluation of hotly contested interpretations.

In this QuickLesson, we will deal with just one research question the author posed. Elsewhere, QuickLesson Study Groups will likely want to evaluate the whole article, identifying each research question, examining each piece of evidence, analyzing it for type and strength, and studying how the author used that evidence to build a solid and convincing case.

Case at Point: Sally Hemings’s Children

Early in the presidency of Thomas Jefferson, a disgruntled political activist used the media to attack Jefferson at a personal level. Specifically, he broadcast local rumors that linked the president to Sally Hemings, the enslaved “quadroon” half-sister of Jefferson’s deceased wife. To quote a few lines from one of the sensational accounts, a rant that appeared in the 1 September 1802 edition of the Richmond, Virginia, Recorder:


It is well known that the man … for many years past has kept, as his concubine, one of his own slaves. Her name is SALLY. The name of her eldest son is TOM. His features are said to bear a striking although sable resemblance to those of the president himself. The boy is ten tor twelve years of age. The delicacy of this arrangement must strike every person of common sensibility. … By this wench Sally, our president has had several children. There is not an individual in the neighbourhood of Charlottesville who does not believe the story, and not a few who know it....

For two centuries, political writers, historians and, now, geneticists have debated two related questions:

  • Who, exactly, were Sally Heming’s children?
  • Who, exactly, fathered those children?

Both are essentially genealogical questions. Ironically, neither had been studied by the standards of genealogical scholarship.1 In 2000, in the wake of genetic reports that one proved child of Sally carried the Jefferson Y-chromosome while one alleged child did not, the peer-reviewed journal of the National Genealogical Society undertook the challenge of weighing the evidence by genealogical standards. It published its findings in a special September 2001 issue titled Jefferson-Hemings.

The lead article, “Sally Hemings’s Children,” examined all the available evidence, drawing from relevant original sources and a plethora of derivatives. It considered primary and secondary accounts. It weighed the evidence from the perspectives of all who created or contributed evidence. The conclusions drawn by the lead investigator Helen F. M. Leary, a past-president of the Board for Certification of Genealogists and the field’s leading specialist in families of the Upper South, is unequivocal on three critical issues:2

  1. Sally Hemings was the mother of seven children born 1790–1808:
    • “Child of unknown gender,” born early-to-mid-1790; lived ‘but a short time’ according to a younger brother Madison.3
    • Harriet Hemings, born 1795; died 1797.
    • Beverly Hemings, born 1798; left Monticello at the age of 23–24, married and lived as white, according to Madison.
    • Harriet Hemings, born 1801, left Monticello at 20–21, married and lived as white, according to her brother Madison.
    • Female child, born 1799; died as infant.
    • Madison Hemings, born 1805, died 1877 in Ohio.
    • Eston Hemings, born 1808; died 1856, Madison, Wisconsin.
  2. Historical and genetic evidence consistently point to Thomas Jefferson as the father of all seven children.
  3. Sally was not the mother of Thomas Corbin “Tom” Woodson, a man whose modern descendants assert him to be the son of Thomas Jefferson and Sally Hemings,4even though his male-line offspring carry Woodson, rather than Jefferson, Y-DNA.5

The third issue forms the question for this QuickLesson: Was Tom Woodson the son of Sally Hemings? Jefferson apologists assert that he was. In their perception of the evidence, a conclusion that Sally bore Woodson then becomes evidence that she promiscuously bore children by multiple men, which is then offered as evidence that she was not Jefferson’s “faithful companion,” as alleged.6

Leary’s examination of the evidence relating to Tom Woodson covers all three basic types: direct, indirect, and negative.7 Part of that evidence will be presented here, to illustrate the differences between the three evidentiary classes. The evidence selected here will focus upon Tom Woodson’s period of birth, Sally Heming’s age at the time of his birth, and the well-identified pool of enslaved people in Thomas Jefferson’s household.

Census Information

(Direct evidence)

Leary found Thomas “Tom” Woodson on four federal censuses: 1820, 1850, 1860, and 1870. The age data these records provide place his birth into the following age brackets:

1820 (26–45):           8 August 1775–7 August 1794

1850 (age 66):          2 June 1783–1 June 1784

1860 (age 75):          2 June 1784–1 June 1785

1870 (age 80):          2 June 1789–1 June 1790

Analyzing this information, we might conclude the following:

  • In all four cases, the evidence is direct, because each census offers an explicit statement of age. The fact that the 1820 census offers an age bracket spanning 20 years, rather than the specific age we would prefer, does not change the fact that the census makes a direct assertion. The fact that none of the four censuses provide an exact birthdate (the “full answer” researchers usually prefer) does not change the fact that each makes a direct statement regarding Woodson’s age.
  • The first three censuses offer compatible direct evidence, from which one can conclude that the birth likely occurred about the middle of 1784. The fact that the 1850 and 1860 censuses appear to differ by one year does not alter their compatibility, given not only the frequency with which census ages err but also the fact that ages often were given as of the date a household was visited rather than the “official census date” decreed by the Census Bureau (i.e., 1 June in 1850, 1860, and 1870).
  • The last census presents conflicting direct evidence, a conflict the author had to resolve. In doing so, she notes (among other points) that the 1870 enumerator “rounded off” the ages of household members: Thomas is said to be “80,” his son James (actually 32) was said to be “40,” etc. Rounding off to a 5- or 10-year guesstimate was a common practice among enumerators when the provider of the data did not know exact ages.8
Deed Information

(Negative and indirect evidence)

About two years after Tom Woodson’s marriage, he was explicitly named in a deed executed by a neighbor. Robert Renick, of Sinking Creek, Greenbrier County (now West Virginia), identified adjoiners to the land he and his wife were selling. One of those adjoiners was “Tom Woodsen.” While the deed information provides direct evidence of Woodson’s residence, on the surface it might seem irrelevant to Woodson’s age. However, it acquires relevance to that question when correlated with other pieces of evidence both negative and indirect. Key to the interpretation is the experiential knowledge that individuals in this time and place were named as adjoiners on deeds under one of two circumstances: they owned the land or they occupied the land by a legal lease.

  • Negative evidence: Land ownership. Under Virginia’s common-law practices, land ownership does not speak to age; even minors could own land. However, Leary’s search through all extant deeds, probate files, and tax rolls for Greenbrier yielded no evidence that Tom Woodson bought land, inherited land, or paid tax on land. The absence of such a record—i.e., the absence of what should be found under this set of circumstances—provides evidence that Tom Woodson did not own this piece of land. By elimination, he would have held the land by lease.
  • Indirect evidence: Age restrictions upon legal leases. Also by Virginia’s common law, an individual had to be 21 or above to execute a legal lease. That law does not speak directly about Tom Woodson’s age. Thus, it is not direct evidence. It is indirect evidence because it is relevant to the question and can be combined with other evidence (information from the deed) to deduce that Tom Woodson was over age 21 on the date the deed was executed, 23 June 1807. This indirect evidence thereby places his birth before 23 June 1786.9 It also compatibly supports the age evidence derived from the censuses.
Family Information

(Indirect evidence)

According to family records put forth by Woodson descendants, the eldest child of Tom Woodson and his wife Jemima was born 28 January 1806 in Greenbrier. That date places the conception of this child, Lewis Woodson, about April 1805. If Tom Woodson was—as the family also alleges—the oldest child of Sally and Thomas Jefferson, then he would have been barely 15 at the time he fathered this child by the 22- or 23-year-old Jemima.

Comparing the situation against male marital and paternity norms in Woodson’s society, Leary notes that fifteen was “a highly uncommon age for fatherhood in that society [and] a truly remarkable age for a boy to establish a stable and lifelong marriage.” Both the birth information for the son and the patterns Leary applied represent indirect evidence for a conclusion about Tom Woodson’s age. From these, Leary concludes that the “census evidence [of] Woodson’s mid-1784 birth date” coupled with the indirect evidence provided by the son’s information, produces for Tom “a more credible male marital age of twenty to twenty-one.”10

Thomas Jefferson’s Farm Book

(Direct, indirect, and negative evidence)

Jefferson's Farm Book—1794 slave list

Among the thousands of pages of farm and financial records left by Jefferson, a goodly number are relevant to this problem. The document depicted above (click to enlarge) provides evidence of two particular matters: (1) Sally’s age; and (2) the identity of Jefferson’s enslaved people―specifically, it answers the question whether in 1794 he owned a slave child named Tom who could be the child born to Sally in 1790.


(Direct and indirect evidence)

The image is a page extracted from Jefferson’s farm book. Its header titles it, “Roll of the negroes Nov. 1794, and where to be settled for the year 1795.” On this roll, after the name of each enslaved person, there usually appears a notation of his or her year of birth and sometimes the month. Occasionally, there is a stated occupation. As shown in column 1 of this document, the individuals assigned to Monticello for 1795 included Sally 73 [i.e., born 1773], listed immediately after Sally’s sister Critta and Critta’s child James.11

This cryptic piece of information also provides both direct and indirect evidence, depending upon the question posed:

  • Direct evidence: When was Sally Hemings born?
  • Indirect evidence: Was Sally old enough to be the mother of Tom Woodson, born about the middle of 1784?

Given that all records bearing upon Sally’s age place her birth in 1773 (exact date is never stated), Leary concluded: “Arguing that Thomas Woodson was Sally’s son requires us to accept the premise that Sally bore him at an extremely unlikely age (ten to twelve) or that Woodson himself made all the passages from boyhood to manhood at ages that range from virtually impossible to highly precocious.”12


(Negative evidence)

The 1794 roll provides no listing of a slave child named Tom. The only enslaved person of that name, one Tom Shacklef[or]d in column 1, has no cited age. He was, however, clearly an adult, given that he and a male bracketed with him are said to be the “carters” at Monticello—i.e., the trusted men who hauled goods in and out of the plantation.

Two relevant considerations—the absence of a slave child Tom on this roll and the fact that, unlike her sister Critta and other mothers on this roll, no child is credited to Sally—both provide negative evidence that the Tom Woodson who surfaced in distant Greenbrier County could not be the first child born to the young Sally Hemings in 1790. This negative evidence supports the direct evidence provided by Sally’s son Madison in his biographical account wherein he stated that his mother’s first child lived "but a short time.”

James T. Callender’s Assertion

(Conflicting direct evidence)

The author of the scurrilous newspaper articles, James Thompson Callender, alleged in 1802 that Jefferson and Hemings had a son “ten or twelve years old” whose name was Tom. That allegation stands as direct evidence. However, when compared to the records above it presents conflicting direct evidence that has to be resolved. Among other evidence used by Leary, we find firsthand (primary) information offered by one Thomas Turner, who was a personal acquaintance of Jefferson.

Thomas Turner’s Assertion

(Direct evidence used to resolve conflict)

Turner, whom Leary identifies as another Jefferson foe as well as a personal family acquaintance, published a piece on 31 May 1805 in the Boston Repertory. There, he stated that Jefferson and “the black, (or rather mulatto) Sally … have cohabited for many years, and the fruit of the connexion abundantly exists in proof of the fact. … The eldest son (called Beverly,) is well known to many.”

Turner’s identification of Sally’s oldest son as Beverly, rather than Tom, is an identification all firsthand evidence supports. Turner’s assertion also is compatible with the family account given by Sally’s son Madison. It is supported, as well, by all the entries in Jefferson’s farm book and other papers that identify Sally’s children.

Using this range of evidence to reconcile the conflicting evidence found in Callender’s rants, Leary concludes:

“The slave whom Callender derisively calls ‘president Tom’ did not exist, at least not as a 1790-born son of Sally Hemings and Thomas Jefferson. He may have existed as the child born in 1790 to Sally, who died shortly afterward. Or he may have existed as a redheaded almost-five-year-old named Beverly, who was at Monticello. SALLY and TOM typeset in all-capitals in Callender’s explosive articles was an eye-catching and far more suggestive combination than SALLY and BEVERLY would have been—and Callender’s style was gross exaggeration rather than outright invention.”13

The Bottom Line

Evidence is not a source. It is not the actual words that appear on a piece of paper or those uttered by an informant. Words have varied meanings that change across time. In any situation, words have nuances. Words taken out of context can seem to say something quite different from what the originator intended. The words that our sources provide do convey information, but that information may be interpreted in disparate ways—as America has seen from the conflicting views about the relationship between Sally Hemings and Thomas Jefferson.

Too often, history researchers focus on the search for sources, hoping to find explicit answers to their questions. They may indeed find explicit answers, but those answers may be wrong. They may also find a plethora of sources that make no direct assertions and yet prove invaluable for a reconstruction of historical events or past lives. Achieving reliable “proof” of any point requires us to understand evidence, its varied classes, the distinctions between them, and how each can be used to build an argument in support of a reliable conclusion.

1. The best gateway to the many studies that have been published on the relationship between Thomas Jefferson and Sally Hemings is offered by the Thomas Jefferson Foundation at its website,  Monticello. Particularly for the Foundation’s own analysis of the historical evidence, see “Report of the Research Committee on Thomas Jefferson and Sally Hemings” (www.monticello.org/plantation/hemings_report.html : published January 2000).

2. Helen F. M. Leary, “Sally Hemings’s Children: A Genealogical Analysis of the Evidence,” Jefferson-Hemings: A Special Issue of the National Genealogical Society Quarterly, NGSQ 89 (September 2001): 165–207. This QuickLesson will not cite the abundance of sources that Leary discusses; they are well-cited in her article. Here, the focus is on how the author analyzes evidence and draws conclusions.

3. Madison Hemings, as a longtime resident of Ohio's Ross County and its offshoot Pike County, was known locally as Thomas Jefferson’s son. The 1870 census taker, after recording Madison’s data, added a note: “This man is the son of Thomas Jefferson!” Three years later, the local newspaper editor published an interview with Madison, chronicling his family and his life. For the census notation, see 1870 U.S. Census, Ross County, Ohio, population schedule, Huntington Township, p. 599, dwelling 49, family 49; National Archives microfilm publication M593, roll 1263. For the interview, see “Life among the Lowly: Number 1,” Pike County Republican, Waverly, Ohio, 13 March 1873.

4. For example, see Byron W. Woodson Sr., A President in the Family: Thomas Jefferson, Sally Hemings, and Thomas Woodson (Westport, Conn.: Praeger, 2001).

5. The Y-chromosome evidence for Jefferson, Woodson, and Carr—the latter being the male line of Thomas Jefferson’s nephews, who some alleged to be the father of Sally’s children—is examined from a genealogical perspective by geneticist Thomas H. Roderick in “The Y Chromosome in Genealogical Research: From Their Ys a Father Knows His Own Son,” NGS Quarterly 88 (June 2000):134–36.

6. Eyler Robert Coates, “Jefferson’s DNA and Sally Hemings,” The Jeffersonian Perspective (http://eyler.freeservers.com/JeffPers/jefpnotg.htm : last accessed 31 October 2012), para. 11, asserts: “All of this suggests that Sally undoubtedly had children by at least two different men, and possibly more, but she was not truthful in identifying who the fathers were …especially since she would have strong motives to name the greatest man around rather than someone of lesser stature.”

7. Leary, “Sally Hemings’s Children,” 173–80.

8. Ibid., 174–75.

9. Ibid., 175–76.

10. Ibid., 175.

11. “Roll of negroes Nov. 1794, and where to be settled for the year 1795,” Farm Book 1774–1824, p. 30; Thomas Jefferson Manuscripts, Coolidge Collection; Massachusetts Historical Society, Boston. Copy provided 2001 by Helen F. M. Leary to Elizabeth Shown Mills, Editor, National Genealogical Society Quarterly.

12. Leary, “Sally Hemings’s Children,” 178.

13. Ibid., 179–80.

How to Cite This Lesson

Elizabeth Shown Mills, “QuickLesson 13: Classes of Evidence―Direct, Indirect & Negative,” Evidence Explained: Historical Analysis, Citation & Source Usage (https://www.evidenceexplained.com/content/quicklesson-13-classes-evidence%E2%80%94direct-indirect-negative : [access date]).




Evidence can be messy. Because it is a mental construct, it rarely gives us the clear and simple answers that we seek. Sources, by contrast, are physical; we can touch them, see them, smell them, hear them. Information is also physical, visible, audible. Evidence, however, is intangible. It’s only what we think certain information means. That’s all it can be―until we make something concrete from it by processing it and molding it into a meaningful and convincing form.