65. Thomas Jefferson to Robert Smith, July 1, 1805, quoted in Dumas Malone, Jefferson the President: First Term, 1801–1805 (Boston: Little, Brown and Co., 1970), p. 222.
66. Monticello report, Appendix F, “Review of the Documentary Evidence,” p. 2.
67. Gordon-Reed, Thomas Jefferson and Sally Hemings: An American Controversy, p. 146.
68. Text of the June 19, 1805 Washington Federalist article as reproduced in Rebecca L. McMurry and James F. McMurry, Jefferson, Callender, and the Sally Story: The Scandalmonger and the Newspaper War of 1802 (Toms Brook, Va.: Old Virginia Books, 2000), pp. 107–110.
69. Malone, Jefferson the President: First Term, 1801–1805, pp. 218–223.
70. Monticello report, Appendix F, “Review of the Documentary Evidence,” p. 2.
71. Randall to Parton, in Flower, James Parton, p. 239.
72. “There are such things, after all, as moral impossibilities.” Ellen Randolph Coolidge to Joseph Coolidge, October 24, 1858, Coolidge Collection, University of Virginia Library.
73. Gordon-Reed, Thomas Jefferson and Sally Hemings: An American Controversy, pp. 107ff.
74. Ellen Coolidge to Joseph Coolidge, October 24, 1858, U.Va. Library.
75. On this generally, see Jean M. Yarbrough, American Virtues: Thomas Jefferson on the Character of a Free People (Lawrence: University Press of Kansas, 1998), and also Andrew Burstein, The Inner Jefferson: Portrait of a Grieving Optimist (Charlottesville: University Press of Virginia, 1995).
76. Jefferson to William Short, October 31, 1819, in Jefferson: Writings (ed. Merrill D. Peterson, New York: Library of America, 1984), pp. 1432–33.
77. See generally Elizabeth Abbott, A History of Celibacy (New York: Scribner, 2000).
Individual Views of
Professor Forrest McDonald
16
Individual Views of Professor Forrest McDonald
* * *
I have read the drafts of the Jefferson-Hemings Scholars Commission report and want to go on record as endorsing it without reservation.
I should like, however, to make an addendum to be included with the final report if possible. Lest anyone think I have acted out of preconceived ideas favoring Jefferson, let it be known that I am an unreconstructed Hamiltonian Federalist, and out of my admiration for Alexander Hamilton I have long been disposed to believe the worst about Thomas Jefferson. Indeed, for nearly four decades I assumed, without thinking much about it, that the allegations regarding a Thomas Jefferson-Sally Hemings relationship were founded in fact. Since the kindling of the current controversy over the DNA evidence and Eston Hemings, however, I have studied the subject as thoroughly as I could, have read the evidence produced by Herbert Barger, and have of course followed the work of the Scholars Commission—the result being that I have entirely abandoned my earlier assumption. Thomas Jefferson was simply not guilty of the charge.
Forrest McDonald
Distinguished University Research Professor
University of Alabama
Individual Views of
Professor Thomas Traut
17
Does the DNA Analysis Establish Thomas Jefferson’s Paternity of Sally Hemings’ Children?*
* * *
Because of his remarkable contributions to both the discussions about, and the actual writing of the original documents encoding the American concepts and principles for liberty and justice in a new democratic society, Thomas Jefferson remains an icon of liberal thinking and of democratic ideals. This lofty stature, however, continues to make him a tempting target for those hoping to expose some flaw in his character.
A long-standing hypothesis, with little historical foundation, claims that Jefferson is the father of one or more children born by his slave, Sally Hemings. Because DNA analysis has become a powerful tool for identifying any individual, and because parents pass on their DNA to their descendants, it would appear that such a scientific test might help to accurately test the paternity hypothesis, if living descendants of the original characters in this historic drama could provide DNA samples.
A team of eight scientists, led by Dr. Eugene A. Foster, a retired pathologist in Charlottesville, Virginia, undertook to perform such a DNA analysis.1 They had obtained blood samples from the descendants of two putative sons of Sally Hemings (Thomas Woodson and Eston Hemings), though there is debate as to whether Thomas Woodson was actually a child of Sally Hemings. They also had blood samples from five descendants of Field Jefferson, the uncle of Thomas Jefferson, and samples from three descendants of John Carr, the father-in-law of Thomas Jefferson’s sister.
Background on DNA Analysis
Most of the DNA that encodes the human genome is located in the nucleus of cells, and is distributed among twenty-three different chromosomes. All such chromosomes exist in pairs, with one member of each pair coming from each parent. For twenty-two of these chromosomes it is very difficult to distinguish which member of the pair is the paternal chromosome. Only with the XY chromosome pair is it obvious and therefore easy to make this distinction, because the Y chromosome determines maleness and is passed from father to son. An additional benefit is that the Y chromosome is the smallest of all the chromosomes. It has only a small amount of genetic information; it does not undergo as much variation with time in different generations, so that the Y chromosome in a modern descendant should still greatly resemble the ancestral Y chromosome of 200 years ago.2,3,4 Thus, the clear benefit of using only the Y chromosome is that it specifically and reliably follows the paternal line of descent.
The DNA in human chromosomes consists of a duplex of two DNA strands. Each strand is a linear molecule, composed of only four types of nucleotides, designated by the letters A, C, G, and T.5 DNA is organized as informational units, genes, and each gene has the information to define the amino acid sequence of a specific protein, which are the functional agents in cellular biochemistry. Because it requires three consecutive nucleotides, defined as a codon, to designate a specific amino acid, there are actually sixty-four codons to define the twenty different amino acids found in our proteins. That is, for each amino acid there are usually three possible codons, and sometimes five or six codons. An important benefit of this is that frequently a codon in a gene can undergo a change at one nucleotide, and this altered codon will have the same identity for the amino acid that was originally specified. Therefore many alterations in our DNA have absolutely no effect, because they cause no change in the protein that is coded by that gene.
Like our DNA, proteins are also linear molecules, containing the twenty different types of amino acids in a string, or sequence, that is again unique for each protein. Alterations in our DNA are also tolerated because of the twenty types of amino acids that make up proteins; often several different amino acids can be placed at some specific position in the protein sequence without changing the shape of the protein, and therefore without changing the function of the protein. Again, this feature makes it permissible for certain DNA alterations (mutations) to occur, because by substitution of a single amino acid they cause a change in the protein that has no harmful consequences.
Types of Changes in DNA
When cells divide, the entire DNA content in the nucleus must first be duplicated to yield two complete sets of DNA that are in theory identical. The actual duplication of the parental DNA is performed by enzymes which are remarkably efficient, but which are not perfect. Occasionally they will make an error, so that the newly copied DNA may now include some small alteration. Such alterations are of two types:
1. nucleotide substitution
2. insertion or deletion of nucleotide(s)
If a typist were to simply type all the text in some book, occasionally the typist might strike the wrong letter key in a particular word. This would be comparable to a nucleotide substitution. If the typist becomes distracted and types the same sentence three times in succession, this would be an insertion because there is now extra text. Or, the typist could miss several sentences in
the original text, and produce a slightly abbreviated copy. This is a deletion.
The above changes are not always harmful, and therefore are retained in the copied DNA. In the same way that we can often still understand a word even though it contains a typographical error, substitutions in DNA are often tolerated. And while insertions of duplicated text are unattractive, they again may not change the meaning. Deletions are more likely to be serious.
Through time many such changes, or mutations, may occur in our DNA when they cause no harm to the person(s) carrying these slightly altered genes. Even harmful mutations may continue in the twenty-two autosomes, because these come in pairs and the matching chromosome may still have a normal gene. Therefore, in our population all people have inherited comparable, but often unique changes in their DNA. These changes are very specific to a family line, though some of them have naturally spread more widely into our population, depending on how far back in time a particular change first occurred. The result of this continuing alteration of our DNA is that among living humans, for any two individuals more than 99% of their DNA will be identical when compared position by position along the DNA sequence. At the same time, enough variation has arisen in the human lineage that each individual probably has at least one truly unique feature.
Strategy of the DNA Analysis
Because variations or changes in DNA are caused by one or more specific alterations in a very large sequence, how do we detect such changes? Three major experimental approaches exist, which differ in their complexity and expense, and also in their resolution or accuracy:
Method 1. Sequencing a complete segment of DNA.
Method 2. Restriction fragment length polymorphism (RFLP).
Method 3. Identifying unique sites by amplifying the DNA with the polymerase chain reaction (PCR).
Method 1 is the most accurate, but also the most difficult. Clearly, one could completely sequence the entire Y chromosome from two or more men. Then, by carefully comparing these linear sequences, we could detect any position at which the nucleotide in the separate sequences was not identical—thereby implying that one or more had undergone a change. While this procedure would be very accurate, the sheer size of a single chromosome, in terms of the number of total nucleotides to be sequenced, makes this an enormously complex and therefore exceedingly expensive effort. Even the human Y chromosome, the smallest of all twenty-four human chromosomes, contains an estimated 35,000,000 bases in its sequence.6
Scientists have therefore devised two simpler procedures that are likely to give us a meaningful test. Method 2 depends on the ability of a select group of DNA-cutting enzymes to accurately recognize a short string of nucleotides, like a short string of letters in a sentence, and chemically cut the DNA at this position, thereby producing two smaller DNA molecules. These DNA-cutting enzymes are called restriction enzymes. Because it is very simple to measure the physical size of a DNA molecule, then by using several such restriction enzymes one can easily test how many cuts are made (by the number of smaller DNA restriction fragments produced), and estimate where the cuts are made (by the variable size of the DNA fragments).
As more changes have occurred between two DNA molecules that are being compared, the likelihood increases that at least one such change will alter a recognition site for a restriction enzyme. This will be directly visible in the number of DNA fragments produced, and by changes in the size of such DNA fragments. Thus, when cut by a single restriction enzyme, the DNA fragments will vary as to the number and size when DNA from very different human groups are compared. That is, one DNA molecule may contain four sites at which it is cut by a single enzyme while a different DNA molecule may contain six sites for cutting. If the two DNA molecules are initially the same size, then the number and sizes of fragments will be different when the sample molecules are cut four times or six times. Such variations are referred to as “restriction fragment length polymorphism,” or simply by the initials RFLP. Note that large segments of DNA need not be accurately sequenced by this procedure. Instead it helps to determine if two or more different DNA molecules have the same few, small identical sites. The observed variation in such RFLPs was an early method for a DNA identification test.
Method 3 uses a different approach, based on identifying by sequence small sites in the DNA at which change occurs more frequently. Clearly this is only possible if the sites being changed are not essential for our normal health and function, and therefore are maintained over generations. Imagine that one wanted to find where, in older automobiles driving down a street, damage was likely to exist. Clearly damage would not be likely in the engine or carburetor if the cars are actually running. But one might readily observe dents in door panels, sagging bumpers, or rust on the fenders. This is the type of damage that an automobile may accumulate with age, while otherwise running normally. Now, if different car models were constructed with different quality sheet metal or different construction standards, then it might happen that Fords had problems with door panels, while Hondas were more likely to have rusty fenders, and so forth. This analogy is intended to suggest why specific changes may occur and be maintained within a family line, and that they are likely to be observed only at specific sites in the DNA. Such positions where various changes in the DNA are observed are then the sites of polymorphism.
Helpful to such comparative studies is the fact that some regions in a DNA molecule are more susceptible to change. Currently studies of the human Y chromosome have led to a series of defined sites for restriction enzymes or PCR probes that are both readily detectable and also shown to have varying frequencies of occurrence among different groups of people.7,8,9,10,11,12,13 Such established DNA sites are therefore useful in efforts to compare separate DNA samples without an inordinate amount of work. For the work by Foster et al. on the Y chromosomal DNA, nineteen separate sites had been identified as useful because variations frequently were observed there. As shown in Figures 12–16, these include seven bi-allelic markers, eleven microsatellite short tandem repeats (STRs), and the minisatellite MSY1. The latter is considered one site, though it contains various arrangements of four different DNA patterns, whose type or identity is identified in parentheses.14
Once the exact sequence of such individual sites is known, it is possible to detect their existence by using a small DNA probe (synthesized chemically) designed for that specific site, and then using the polymerase chain reaction (PCR) technique to make very many copies of this DNA segment, if in fact the Y chromosome being tested contains this segment. This highly amplified set of DNA molecules will all be small and identical, and therefore very easy to measure. This method clearly depends on knowing in advance the sequence of sites that are likely to show variation in a population (polymorphism). Only in the last ten years have enough such polymorphic sites become identified for the Y chromosome to make Method 3 more applicable.15
All our chromosomes come in pairs. The set of chromosomes inherited from one parent is half the total, and is referred to as a haploid set, while the combination of these two haploid sets from our two parents equals a diploid set. When only one member of a chromosome pair (the Y chromosome from the XY pair) is analyzed for its DNA, the DNA pattern is then designated a haplotype.
For the work by Foster et al. both Methods 2 and 3 were used in an analysis that included nineteen distinct sites in the DNA haplotype at which changes have been observed in the general population.
It is important to appreciate that many changes may have occurred that would be undetected by this analytic procedure, because only nineteen small sites within a very large amount of DNA are being examined. However, any observed change would be a true and positive result. As an illustration of this, consider that one wished to verify if a typist had correctly typed the entire text of an encyclopedia, and could examine only nineteen pages of the newly typed text. Clearly many errors could have occurred on the unexamined pages, and these would not be detected. But, any error that was observed on the nineteen pages being exam
ined would be direct proof that a change had occurred.
Results of the DNA Analysis by Foster et al.
While Dr. Eugene Foster, the lead author of the Nature paper, organized this DNA project, it appears that most of the specific scientific work was done by the seven co-authors, who were scientists working in England or in Holland. Because these scientists had recently published papers on examining Y chromosome polymorphisms in humans,16 they appear to be well qualified for the scientific work involved.
Normally Y chromosomes are carried by males who are designated as XY, for this chromosome pair, in contrast to females who are designated as XX. Because Thomas Jefferson had no sons that lived to adulthood, it was not possible to get a Y chromosome sample from a direct descendant of Thomas Jefferson. Foster and associates therefore relied on samples derived from descendants of Field Jefferson. As the paternal uncle of Thomas Jefferson, this individual and his direct male descendants would be expected to share the same Y chromosome inherited by both Thomas Jefferson and Field Jefferson from their common ancestor, Thomas Jefferson II—the grandfather of Thomas Jefferson (see Figure 12).
Figure 12. The Jefferson Family Line. Five living descendants of Field Jefferson, designated by code (J##) have the DNA haplotypes shown, as determined by Foster et al. The haplotypes are also by a numerical code, where the number represents how often that microsatellite or minisatellite sequence segment was repeated. One needs only compare the numbers, at a given position from left to right, to see if they are the same or different. It can then be seen that the first four men have identical haplotypes, while J50 has an extra STR segment at one position, an insertion, indicated in bold. All five men would be considered to have the Jefferson haplotype. This is the family tree presented in the DNA study by Foster et al. that omitted Randolph Jefferson and his sons, who have been added in Figure 16.
The Jefferson-Hemings Controversy Page 55