Making Absence Findable Without Making Fabrication Admissible: Institutional Design for the Recovery of Unnamed Maternity

Abstract

The four preceding papers in this series established an evidentiary typology for founding-era paternity claims, demonstrated that the standard applied to the Jefferson–Hemings case was anomalously strict relative to like claims, proposed a genetic-genealogical program for identifying the mother of William Franklin, and documented a durable pattern of historiographical incuriosity toward the mothers of acknowledged illegitimate children. This paper takes up the institutional question the first four defer: what should be built, and by whom.

Its central finding is that the incuriosity is not merely a habit of biographers but a property of retrieval infrastructure. Archival description standards, documentary edition indexing, and genealogical proof criteria each require a name as the entry condition for a record, which renders nameless persons unfindable and — more consequentially — uncountable, since a phenomenon that cannot be enumerated cannot be studied as a phenomenon. The paper proposes placeholder authority records as the primary remedy, converting an unindexed silence into a searchable and aggregatable object.

It also states the tradeoff plainly. Every mechanism that lowers the barrier to finding these women also lowers the barrier to unfounded claims of descent from prominent men. Restrictive lineage and proof standards are not merely exclusionary; they are load-bearing against fabrication, and they were built in response to actual fabrication. The design problem is therefore to separate discoverability, which should be expanded aggressively, from admissibility, which should not be relaxed at all. The paper argues these are severable, specifies four controls that sever them, and maps each control onto the evidence tiers and tractability tiers established earlier in the series.

1. The Problem Restated as a Design Problem

1.1 What the fourth paper found

The fourth paper documented that where a prominent American man acknowledged a child born outside marriage, the acknowledgment functioned as a terminal answer, and the mother was not thereafter pursued. It located the mechanism in the direction of the record: name, property, and standing transmitted patrilineally, and archives accreted around those transmissions.

That is a finding about historians and about archives. It is not yet a design finding, because it does not say what any institution should do differently.

1.2 The design finding

The design finding is narrower and more actionable: the name is the entry condition for the record, and everything downstream inherits that condition.

Archival description under Describing Archives: A Content Standard and its international counterparts organizes access around authority records for persons, families, and corporate bodies (Society of American Archivists, 2022; International Council on Archives, 2004). An authority record requires an identity. A person known only as “an unnamed servant in the Franklin household, Philadelphia, c. 1730” has no authority record, appears in no name index, and is retrievable only by someone who already knows to read the specific document in which the phrase occurs.

Documentary editions inherit the same condition. The Franklin Papers, the Jefferson Papers, and their peers index persons by name with great thoroughness (Labaree et al., 1959–; Boyd et al., 1950–). A researcher can search these editions exhaustively and never encounter the maternity question as a question, because there is no index term under which it lives.

Genealogical standards inherit it a third time. The Genealogical Proof Standard requires reasonably exhaustive research, complete source citation, analysis and correlation, resolution of conflicts, and a soundly reasoned written conclusion (Board for Certification of Genealogists, 2019). Applied honestly, it favors lines that generated documents — which is to say, definitionally, lines that had property, litigation, or church membership to record. It is not biased against poor women by intent. It is biased against them by the structure of what survives.

1.3 Why this is an institutional-maintenance pattern

None of these systems was designed to erase anyone. Each was designed to solve a real retrieval problem, and each solves it well. The erasure is a byproduct that no single actor produces and no single actor is positioned to correct — the archivist follows the standard, the editor follows the index convention, the certifying body follows the proof standard, and each is behaving correctly.

This is the familiar shape of institutional insulation: an outcome that persists because it is nobody’s output. The remedy accordingly has to be specified at the level of the standard, not the level of the practitioner, and this is what makes the problem tractable at all. Practitioner exhortation would fail. Standard revision might not.

2. The Primary Remedy: Placeholder Authority Records

2.1 The proposal

Where a source documents an acknowledgment of paternity outside marriage, or otherwise establishes that a specific person existed whose name the record does not preserve, the describing institution creates an authority record for that person constructed from known attributes rather than from a name.

A record might read: Unnamed woman, mother of William Franklin. Female. Philadelphia or vicinity, c. 1728–1731. Status undetermined. Known from: acknowledgment of paternity by Benjamin Franklin. Related entities: Franklin, Benjamin; Franklin, William. Evidence tier: n/a (existence certain, identity unknown).

The record is not a claim about who she was. It is a claim that she existed, which is not in dispute, plus an index point at which everything subsequently learned can attach.

2.2 What this accomplishes

Findability. A researcher working on Philadelphia servant women in the 1730s can now encounter the question. At present she cannot, unless she arrives already knowing.

Countability. This is the more important gain and the less obvious one. Individual placeholder records aggregate. Once several hundred exist across participating repositories, the pattern documented in the fourth paper becomes measurable rather than argued — one can state how many acknowledged illegitimate children in a given period and region have identified mothers, how that proportion varies by the mother’s apparent status, and whether it changed over time. The fourth paper asserted a pattern from a case series. Placeholder records would permit testing it.

This is the same conversion the standardization audit produced when it redefined its object from an unnamed phenomenon into a measurement problem. The redefinition was not a retreat; it was what made the object susceptible to evidence.

Accumulation. Partial findings currently have nowhere to go. A researcher who eliminates four of thirty candidates has produced real knowledge that will die with her article. A placeholder record gives negative results a home, which is the precondition for any cumulative program.

2.3 Precedent

The proposal is less novel than it may appear. Archival practice has moved substantially toward describing persons who left no papers of their own, particularly enslaved persons named in the records of those who held them. Projects including Enslaved: Peoples of the Historical Slave Trade, the Slave Voyages database, and repository-level reparative description initiatives have established both the technical means and the professional legitimacy of building person-records from attributes scattered across documents created by others (Tang et al., 2020; Eltis & Richardson, 2010; Antracoli et al., 2019). The Records in Contexts conceptual model accommodates agents described by relation and attribute rather than by name (International Council on Archives Expert Group on Archival Description, 2021).

What is proposed here extends that work by one step: from describing named persons who were previously unindexed to describing unnamed persons whose existence is nonetheless documented.

2.4 Cost and funding

Retrospective description is fundable. The National Historical Publications and Records Commission’s access grants and NEH’s Humanities Collections and Reference Resources program both support exactly this category of work, and both have funded reparative and retrospective description in recent cycles. The unit cost is low relative to digitization; the labor is professional archival labor, and the output is machine-readable.

The realistic entry point is a pilot at a single documentary edition — the Franklin Papers being the obvious candidate given the case in the third paper — producing a schema that others can adopt.

3. The Tradeoff, Stated Without Hedging

3.1 Discoverability and admissibility travel together by default

Every mechanism that makes an unnamed woman findable also makes her nameable. An index point that invites legitimate research invites illegitimate attachment with equal ease. A record reading unnamed mother of William Franklin is an empty slot beside a famous name, and empty slots beside famous names attract claimants.

This is not hypothetical. The genealogical record contains a long history of fabricated descents from prominent figures, and the professional apparatus of proof standards, peer-reviewed genealogical journals, and lineage-society verification exists substantially because of it (Jacobus, 1930/1968; Mills, 2017). Gustave Anjou’s fabricated pedigrees remain in circulation a century after their manufacture, and the labor of extracting them from the record vastly exceeded the labor of producing them.

3.2 The specific hazard in this domain

Three features make paternity-adjacent claims unusually vulnerable.

Asymmetric verification cost. Making a claim of descent costs a paragraph. Refuting one requires reconstructing multiple generations of documented ancestry, often for lines that generated few documents. The asymmetry is severe enough that unrefuted claims accumulate simply because refutation is unaffordable.

Genetic evidence is easily overstated. The 1998 Y-chromosome study constrained paternity to the Jefferson patriline and was reported as identifying Thomas Jefferson personally; the lead author published a correction (Foster et al., 1998, 1999). If a rigorous study in Nature was misreported this way, informal claims will be misreported worse. Autosomal segment matches at the depths relevant to this series fall in ranges where false positives are common, and a match presented without triangulation or threshold disclosure is nearly uninterpretable (Durand et al., 2014; Ball et al., 2016).

Moral pressure runs one direction. The fourth paper’s finding — that these women were neglected because they were poor, enslaved, or without kin — creates a legitimate presumption in favor of taking descendant traditions seriously. That presumption is correct and can be exploited. A framework that treats skepticism as complicity in the original erasure will accept anything.

3.3 The two failure modes are not symmetric in kind

Over-restriction perpetuates a documented injustice and suppresses true claims that cannot meet a standard calibrated to property-owning lines. Over-admission corrupts the evidentiary record, and corrupted records are far more expensive to clean than to fill.

But the asymmetry cuts a second way that favors caution. A true claim wrongly excluded remains true and can be admitted later when evidence improves. A false claim wrongly admitted becomes a citation, then a secondary source, then a premise. The West Ford tradition and the Woodson tradition both illustrate how a claim, once in circulation, persists after the evidence against it is published (Foster et al., 1998; Wiencek, 2003; Thompson, 2019).

The design conclusion is not that caution should dominate. It is that the two failure modes require different instruments, and that a single relaxed-or-tightened threshold cannot serve both.

4. The Severance Principle

The core proposal of this paper is that discoverability and admissibility are severable and should be severed explicitly.

Discoverability is the question of whether a person, a question, or a body of evidence can be found by someone looking. It should be expanded to the maximum extent the sources permit. There is no fraud risk in findability as such: a placeholder record makes no claim, and an index point asserts nothing.

Admissibility is the question of what a body may accept as established — a documentary edition’s annotation, a certifying body’s credential, a lineage society’s membership, a museum’s interpretive text. It should not be relaxed at all. It should, however, be made tiered and transparent rather than binary, which is a different reform and the subject of §5.

The severance is not merely conceptual. It is implementable, because the institutions that control discoverability (archives, documentary editions, database projects) are largely distinct from those that control admissibility (certifying bodies, lineage societies, peer-reviewed journals, interpreting institutions). The two reforms can proceed independently and be evaluated separately.

5. Four Controls

Control 1 — Existence records and identity claims are structurally distinct

A placeholder authority record asserts that a person existed. It must be structurally incapable of asserting who that person was.

Implementation: the schema separates an existence record, which carries only attributes derivable from the documenting source, from candidate identifications, which are linked child records carrying a proposed name, its evidentiary basis, its tier, and its status (proposed, contested, corroborated, refuted). No candidate identification may be merged into the existence record. A record with six candidates displays six candidates.

This preserves the accumulation benefit — every candidate ever proposed is visible and attached — while making the record incapable of resolving prematurely. It also creates a home for refutations, which currently have none.

Control 2 — Tier declaration is mandatory and public

Every claim entering the system declares its tier under the typology established in the second paper of this series:

Tier I — Principal’s acknowledgment
Tier II — Contemporary third-party documentation
Tier III — Genetic evidence with circumstantial support
Tier IV — Testamentary or behavioral anomaly plus contemporary rumor
Tier V — Family tradition alone
Tier VI — Modern assertion without traceable tradition

The tier is a description of the evidence, not a verdict on the claim. A Tier V claim may be true; the second paper found that several universally accepted paternities rest on Tier V evidence while the most contested rests on Tier III.

Two rules follow, and the second matters more than the first.

Rule 5a — No tier upgrading by accumulation. Three Tier V sources reporting the same tradition remain Tier V unless independence of transmission is demonstrated. Traditions consolidate; convergence is weak evidence when the sources share a channel. The Wetmore-edited Pike County Republican series is the canonical illustration: two accounts, one editor, one project, and therefore not two sources (Hemings, 1873; Jefferson, 1873).

Rule 5b — Prominence is not evidence. Where a patriline contains multiple plausible men, the tradition’s naming of the most famous receives no weight from his fame. This rule is what distinguishes the Eston Hemings case from the West Ford case, and it does so without prejudging either: the Jefferson claim answers the objection through thirteen years of presence correlation, and the Ford claim cannot answer it at all (Neiman, 2000; Wiencek, 2003; Thompson, 2019). The rule is neutral in form and discriminating in application, which is what a good rule looks like.

Control 3 — Pre-registration for genetic claims

Any genetic claim entering the record must declare, before analysis of results:

the segment-length threshold and the rationale for it;
the triangulation requirement (minimum independent descendant branches sharing a segment at the same location);
the endogamy correction applied to the relevant population;
the panel composition, including subtraction references;
the conditions under which the hypothesis would be abandoned.

Post hoc threshold selection — trying successively lower cutoffs until a favored candidate matches — is the characteristic failure of genetic-genealogical claims about historical figures. Pre-registration is inexpensive, is standard in adjacent fields, and is the single highest-yield control available.

Negative results must be publishable and indexed. A documented failure to identify, reporting field construction, panel composition, recovery statistics, and thresholds, is a contribution. Its absence from the literature is part of why the difficulty of this research is assumed rather than measured — and it is precisely why the failure-record chapter of a serious methodology is not an ornament.

Control 4 — Tractability gating

The third and fourth papers established a tractability typology governing which cases genetic methods can address:

Tier A — Mother identified; requires archival work, not genetics
Tier B — Moderate genealogical depth; autosomal or mitochondrial methods reliable
Tier C — Deep descent; panel-dependent; recovery stochastic
Tier D — No traceable descent from the candidate; unresolvable in principle by matching

The control: a claim’s evidentiary tier may not exceed what its tractability tier can support. A Tier D case cannot generate a Tier III claim, because there is nothing to match against; any genetic assertion in such a case is necessarily unfounded. A Tier C case may generate a Tier III claim only with a declared multi-branch panel and pre-registered thresholds.

This control is the most technically restrictive of the four and the most useful, because it can be applied mechanically at intake and disposes of a large class of unfounded claims without any judgment about their content.

6. Institution-Specific Application

The four controls do not apply uniformly. Each institution has a different function, a different failure cost, and a different appropriate threshold. Treating them as one institution would produce a proposal that fits none.

6.1 Archives and documentary editions

Function: description and access. Failure cost of over-admission: low — an existence record asserts nothing. Failure cost of over-restriction: high — this is where the erasure is reproduced.

Application: maximal discoverability. Create placeholder records for every documented unnamed person. Accept and display candidate identifications at all tiers, including Tier VI, with tier declared and status marked. Controls 1 and 2 apply; Controls 3 and 4 apply only to claims asserting genetic support.

The strong recommendation here is that documentary editions add index terms for unnamed persons. The Franklin Papers pilot proposed in §2.4 would demonstrate feasibility at modest cost.

6.2 Peer-reviewed genealogical and historical scholarship

Function: establishing findings. Failure cost of over-admission: high — published findings become premises.

Application: all four controls. Pre-registration required for genetic claims; tier declaration required in abstracts; negative results explicitly welcomed. The National Genealogical Society Quarterly and The American Genealogist already operate at approximately this standard and would require adjustment rather than reconstruction.

6.3 Certifying bodies

Function: credentialing practitioners. Application: the Genealogical Proof Standard needs no relaxation. It needs an explicit statement that a soundly reasoned conclusion may be probabilistic and tiered rather than binary — that “most probably X, at Tier III, with these alternatives not excluded” is a proper conclusion rather than a failure to reach one. This is arguably already implicit in the standard’s language and would benefit from being made explicit.

6.4 Lineage societies

Function: membership on documented descent. Failure cost of over-admission: high and irreversible in reputational terms; the restrictive standard is load-bearing.

Application: all four controls, and no relaxation of documentary requirements. What can be recommended is procedural rather than substantive: that criteria be published, that decisions state which criterion failed, and that a distinct associate category exist for documented Tier IV and Tier V descents that do not meet the full standard — recognizing the tradition and the family without asserting the descent.

This is the paper’s most politically exposed recommendation, and it should be read narrowly. It is a proposal about criteria and their transparency. It takes no position on the merits of any particular family’s claim, and it is offered on the general ground that a body applying unpublished criteria to contested claims invites the inference that the criteria are the claim.

6.5 Interpreting institutions

Function: public presentation. Failure cost of over-admission: high — museum text becomes public fact.

Application: tier declaration in interpretive language, using calibrated formulations. A Tier III finding warrants “most probably”; a Tier V tradition warrants “family tradition holds, recorded by X in year Y.” The failure the first paper documented was precisely the transfer of confidence from a rigorous narrow finding to propositions the finding never addressed — and interpreting institutions are where that transfer reaches the public.

6.6 Research ethics review

Function: protecting participants and descendant communities. Application: Control 3 as a condition of approval for genetic-historical protocols; descendant communities as participants in design rather than sources of samples; explicit consent covering disclosure of unexpected relationships in participants’ own recent ancestry (Kennett, 2019; Nelson, 2016; Erlich et al., 2018).

A consideration specific to this domain: identifying a woman as the mother of a prominent man’s illegitimate child assigns her, posthumously, to an event she may have neither chosen nor survived well. Where the circumstances involved enslavement, indenture, employment, or extreme youth, the research should proceed on the working presumption that she was the less powerful party.

7. What This Framework Does Not Solve

Tier D remains Tier D. Most eighteenth-century candidate women left no traceable descent. No indexing reform and no control regime changes this. A substantial share of the cases motivating the whole series are unresolvable in principle by matching, and the framework’s honest yield in those cases is a well-described placeholder record and nothing more.

Placeholder records may attract more claims than they resolve. The empty slot beside a famous name is a real hazard and Control 1 mitigates rather than eliminates it. This is a testable prediction and should be monitored in any pilot: if candidate identifications accumulate faster than they are adjudicated, the design is failing and should be revised.

Tier declaration can become a formality. Any mandatory-disclosure regime degrades toward box-checking. The countermeasure is that tiers be assigned in review rather than self-reported, which raises cost.

The controls cannot make an under-evidenced field well-evidenced. They can only prevent claims from exceeding their evidence. The recommendation that fastest addresses the original injustice remains the one requiring no laboratory and no reform at all: several of these women are already named, already documented, and still unwritten. Carrie Butler and Julia Chinn are not evidentiary problems. They are unattempted biographies.

8. Recommendations

Pilot placeholder authority records at one documentary edition, producing a publishable schema. Lowest cost, highest demonstration value.
Adopt the severance principle explicitly in professional guidance: expand discoverability without limit, hold admissibility fixed, and state that the two are different questions.
Require tier declaration in scholarly publication, archival description, and interpretive text, with tiers assigned in review rather than self-reported.
Require pre-registration of genetic-historical protocols, and establish an indexed venue for negative results.
Apply tractability gating at intake — the mechanical control that disposes of the largest class of unfounded claims at the least cost.
Recommend published criteria and a distinct associate category to lineage societies, without relaxation of documentary standards, and without adjudicating any particular claim.
Fund the Tier A biographies first. They need historians, not laboratories, and they need them now.

9. Conclusion

The four preceding papers established that a category of persons was lost, that the loss was produced by how the field of inquiry was defined rather than by what the archives contain, and that the loss is partially recoverable. This paper argues that recovery requires building something, and that what must be built is narrower than an advocacy program and more specific than a call for attention: an index point where a name is missing, and a discipline about what may be attached to it.

The design tension is real and does not dissolve. Making these women findable makes their slots claimable. But the tension is between two different institutional functions, held by two largely different sets of institutions, and it can be managed by separating them rather than by trading one against the other. Discoverability costs little and is nobody’s verdict. Admissibility is a verdict and should stay expensive.

The general caution belongs at the end because it is the one most easily lost. The finding that others were credited too easily does not license crediting anyone. That the standard applied to Jefferson was anomalously strict is a finding about the discipline’s consistency; it is not additional evidence in Jefferson’s file, and a framework built on the first proposition while forgetting the second would reproduce, with better intentions, exactly the failure it was designed to correct.

References

Antracoli, A. A., Berdini, A., Bolding, K., Charlton, F., Ferrara, A., Dickson, V., & Rawdon, K. (2019). Archives for Black lives in Philadelphia: Anti-racist description resources. Archives for Black Lives in Philadelphia.

Ball, C. A., Barber, M. J., Byrnes, J., Carbonetto, P., Chahine, K. G., Curtis, R. E., Granka, J. M., Han, E., Hong, E. L., Kermany, A. R., Noto, K., Ohla, J., Rañola, J. M., Sotoodeh, M., & Wang, Y. (2016). AncestryDNA matching white paper: Discovering genetic matches across a massive, expanding genetic database. Ancestry.com DNA.

Board for Certification of Genealogists. (2019). Genealogy standards (2nd ed.). Ancestry.com.

Boyd, J. P., Cullen, C. T., Catanzariti, J., Oberg, B. B., et al. (Eds.). (1950–). The papers of Thomas Jefferson. Princeton University Press.

Durand, E. Y., Eriksson, N., & McLean, C. Y. (2014). Reducing pervasive false-positive identical-by-descent segments detected by large-scale pedigree analysis. Molecular Biology and Evolution, 31(8), 2212–2222. https://doi.org/10.1093/molbev/msu151

Eltis, D., & Richardson, D. (2010). Atlas of the transatlantic slave trade. Yale University Press.

Erlich, Y., Shor, T., Pe’er, I., & Carmi, S. (2018). Identity inference of genomic data using long-range familial searches. Science, 362(6415), 690–694. https://doi.org/10.1126/science.aau4832

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.

Hemings, M. (1873, March 13). Life among the lowly, No. 1. Pike County Republican.

International Council on Archives. (2004). ISAAR(CPF): International standard archival authority record for corporate bodies, persons and families (2nd ed.).

International Council on Archives Expert Group on Archival Description. (2021). Records in Contexts: A conceptual model for archival description (Consultation draft v0.2).

Jacobus, D. L. (1968). Genealogy as pastime and profession (2nd ed.). Genealogical Publishing. (Original work published 1930)

Jefferson, I. G. (1873, December 25). Life among the lowly, No. 3. Pike County Republican.

Kennett, D. (2019). Using genetic genealogy databases in missing persons cases and to develop suspect leads in violent crimes. Forensic Science International, 301, 107–117. https://doi.org/10.1016/j.forsciint.2019.05.016

Labaree, L. W., Willcox, W. B., Oberg, B. B., et al. (Eds.). (1959–). The papers of Benjamin Franklin. Yale University Press.

Mills, E. S. (2017). Evidence explained: Citing history sources from artifacts to cyberspace (3rd ed.). Genealogical Publishing.

Myers, A. C. (2023). The Vice President’s Black wife: The untold life of Julia Chinn. University of North Carolina Press.

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

Nelson, A. (2016). The social life of DNA: Race, reparations, and reconciliation after the genome. Beacon Press.

Skemp, S. L. (1990). William Franklin: Son of a patriot, servant of a king. Oxford University Press.

Society of American Archivists. (2022). Describing archives: A content standard (Version 2022.1).

Stanton, L. (2012). “Those who labor for my happiness”: Slavery at Thomas Jefferson’s Monticello. University of Virginia Press.

Tang, D., Alonso, W., Zwaard, K., & the Enslaved.org Project Team. (2020). Enslaved: Peoples of the historical slave trade. Michigan State University Matrix and the Mellon Foundation.

Thompson, M. E. (2019). “The only unavoidable subject of regret”: George Washington, slavery, and the enslaved community at Mount Vernon. University of Virginia Press.

Washington-Williams, E. M., & Stadiem, W. (2005). Dear Senator: A memoir by the daughter of Strom Thurmond. ReganBooks.

Wiencek, H. (2003). An imperfect God: George Washington, his slaves, and the creation of America. Farrar, Straus and Giroux.