Under a gray Midwestern sky on Tuesday morning, in the parking lot of the National Personnel Records Center annex off Goodfellow Boulevard, the National Archives and Records Administration and the Department of Veterans Affairs marked the 50th year since the fire with a ceremony, a 212-page report, and a small rule change with large implications. The pilot unveiled there allows sworn neighborhood and workplace affidavits to corroborate lost service when coupled with a concrete locator — a payroll stub, a ship log extract, or a geospatial placement on the long-running Muster Map.
Veterans groups applauded. Genealogists staged a protest at the outer gate, objecting to tightened redactions on public access copies also announced that morning. And on Capitol Hill, the House Committee on Oversight and Accountability posted a hearing for Thursday to probe fraud safeguards around the new affidavits. The anniversary became a live argument about proof.
The question is overdue and simple to state: when the government lost millions of service jackets, what evidentiary threshold should govern benefits, honors, and historical accountability? The record supports a threshold that leans further into structured alternate proofs — sworn statements paired with locators, geospatial placement, cross-checked unit records — and it points to a parallel obligation for agencies to publish clear fraud rates and invest in verification workflows, rather than shrink access behind heavier redactions or steep standards that punish families for a failure the state caused.
What changed this week
The pilot rule, signed jointly by VA and NARA, does three things. It states that a sworn neighborhood or workplace affidavit may serve as corroboration of service dates or locations when paired with at least one documentary locator; it sets a verification track that routes such affidavits through a cross-check against unit morning reports, payroll microfilm, and the Muster Map geospatial layers; and it caps the number of affidavits per claimant to two, to nudge quality over volume. The test will run 18 months, with quarterly public metrics.
“Reconstruction at Fifty,” the report released alongside, is plain in its appraisal. Of the service jackets destroyed between 1912 and 1968 across Army, Air Force, Navy, and Marine Corps holdings, the bulk have no surviving personnel forms. What endures are fragments and neighboring records — morning reports, ship deck logs, hospital admission cards maintained at other facilities, base newspapers, payroll rolls, and private material tucked in attics and union files. The report maps the last decade’s gains, including court acceptance of geospatial corroboration in Blue Water Navy decisions, and sets them against persistent bottlenecks. Mosaic cases tied to burned files average 182 days to adjudication at VA, compared with 104 days for intact jackets.
When jackets were lost in the 1973 fire, the bar to prove service should be lower under the affidavit‑pairing pilot, and the audit trail should be published with the program’s quarterly metrics.
The same morning, NARA finalized a redaction update for reconstructed files that adds masking for third-party addresses, certain medical indicators, and names in affidavits where the affiant has no documented consent on file. The Archives cites Privacy Act obligations and the history of PII leakage used in affidavit fraud. The National Genealogical Society calls the policy an overreach and is threatening litigation under FOIA.
How we got here
The fire that began late on July 12, 1973, burned through the night and into a second day, jumping firewalls and collapsing sections of the sixth floor. After the hoses were coiled and the smoke cleared, most personnel jackets for the four services from 1912 to 1968 were gone. Some microfilm and unit records on lower floors and offsite survived. The shock reorganized federal practice.
Congress responded in 1976 with the Veterans Proof Act, which allows alternate evidentiary paths when government loss is shown. The statute names categories of corroboration: unit morning reports, payroll and muster rolls, ship logs, base newspapers, hospital ledgers, union and employer records, parish registers, and sworn affidavits. In 1977, the Defense Mapping Agency and the Archives launched the Muster Map program to place units and ships in time and space. A decade later, the Court of Appeals for Veterans Claims (CAVC) affirmed what practitioners were already doing. In Klein v. VA (1990), the CAVC held that a “corroborative mosaic” could meet the benefit-of-the-doubt rule: no single alternate record controls, but consistency across sources and time satisfies the standard.
The mapping work matured as agencies did. The Defense Mapping Agency became the National Geospatial-Intelligence Agency in 1996, and the NGA–NARA memorandum has kept the Muster Map alive through paper atlases, CD-ROMs, and now digital layers. Deck logs for Navy and Marine Corps operations during the Vietnam era were digitized with geotags; Blue Water Navy cases in the late 2010s leaned on those layers to establish exposure zones, persuading the Federal Circuit to expand eligibility.
Maps as proof, and their limits
The mapping program’s advocates talk about it as a stabilizer. In a basement workroom at the annex, a conservator feeding a fragment through a scanner can pin a unit to coordinates and a date using a morning report and the Muster Map overlay, then send that packet straight to an adjudicator’s workstation. When it works, it reduces a claim to a set of cross-checks and an exposure determination.
People assume maps will save them, and sometimes they do. But these are mosaics with torn tiles. A ship track fixes a line; it doesn’t tell you who was on deck, and an affidavit can be coaxed into the gaps.
Friesz’s objection is not technical; it is philosophical. She argues for stricter standards even though her family would have benefited from looser ones. Her position has weight in this city, where the NGA’s West campus looms over construction cranes and the Archives has rebuilt a cold-storage annex and a Reconstruction Lab to stabilize charred paper. The risk, she says, is that public policy drifts toward forgiving narrative and away from verifiable event.
The counter-case is equally practical. The Court of Appeals for Veterans Claims did not ask for omniscience; it asked adjudicators to weigh a pattern. Deck logs and unit reports have wider coverage than any surviving personnel file. If the government’s loss made personal forms scarce, geospatial placement, payroll stubs, and sworn accounts from people who were there can fairly substitute, provided they are checked against each other and logged for audit.
At NPRC’s Verification Branch and the VA St. Louis regional office, outcomes turn on workflow, from two‑week affidavit cross‑checks to a 182‑day average on mosaic cases and the fields NARA’s July redaction bulletin leaves visible.
Fraud, with denominators
Fraud risk is real and should be counted plainly. A 2009 joint review by VA and the Government Accountability Office, “Reconstruction and Adjudication After the NPRC Fire” (June 23, 2009), credited the Veterans Proof Act with increasing approvals for burned-file veterans, and also documented a rise in organized attempts to exploit affidavits and fabricated unit newsletters. The report forced procedural tightenings, including affidavit notarization protocols and retrieval of base newspaper masthead records to check authenticity.
The numbers that followed did not justify a backlash against alternate proofs wholesale. VA Inspector General referral logs from the late 2000s through the mid‑2010s show suspected-fraud referrals in the hundreds per year across all benefits categories, with a small fraction tied to affidavit abuse. A June 7, 2013 briefing memo from VA’s Office of Field Operations to the House Committee on Oversight and Accountability majority staff, titled “Affidavit‑Linked Claims: FY08–FY12 Metrics” and reviewed by The Ledger, cites 1,240 suspected cases involving affidavits over five years ending in 2012, during which more than 500,000 burned-file claims were processed. That is a quarter of one percent flagged for suspicion, an even smaller share substantiated.
The 2021 Justice Department indictment of a multistate ring that forged workplace affidavits and doctored base clippings is the worst case on record. It rightly triggered a clampdown: stepped-up affidavit verification at VA, and stricter redactions at NARA to limit the too-ready harvest of names and addresses used by those forgers. The episode sits alongside five-year totals showing 1,240 suspected affidavit cases out of more than 500,000 burned-file claims, per the VA memo and Oversight records, a scale that should guide policy.
That obligation extends to capacity. Mosaic cases now take an average of 182 days, and some burn down statutory timetables. Adjudicators at VA regional offices and records technicians in St. Louis carry the weight without the staffing to match. A protested scanning contract for payroll microfilm — 4.2 million images of pay sheets and muster rolls that serve as corroborating locators — has delayed ingestion that the new pilot depends on. On this point, the Archives’ announcement fell flat: the report names the backlog but offers little on procurement fix or surge staffing.
Privacy and the public record
The other weight is privacy. FOIA and the Privacy Act pull in opposite directions at the NPRC. Reconstructed files are stitched from public and private pieces; an affidavit given for benefits becomes part of a file that researchers, family historians, and, yes, fraudsters may try to read. NARA’s July redaction update masks more fields. The Archives says it is aligning with the Privacy Act. To many on the outside, it looks like a retreat from transparency that makes verification harder.
I sell old unit yearbooks and base papers on Chippewa. Families come in with a shoebox and a hope. I want fraud locked out and I don’t want people’s medical business on a photocopier — I worked as a radiology tech, I know what that looks like. But when the redactions blot out names and dates that would prove a story, you’ve turned history into a rumor and knocked a zero off half my stock.
Dardano’s ambivalence is common. It is also the place where policy choices have stakes beyond sentiment. If agencies are going to keep names from public copies of affidavits absent written consent, then other paths must be strengthened: publish monthly fraud referral and substantiation rates; require that affidavit forms include opt‑in consent for name release; invest in a verification unit that can check affidavits against payrolls within weeks, not months, and report turnarounds openly.
Three files, three outcomes
One widow in Wyandotte, Michigan, sent VA a union stewards’ ledger, a parish register noting a funeral honor guard, and a typed statement from a neighbor who worked the same shift at the Packard plant and recalled weekend drills at Selfridge. The Muster Map placed the unit at Camp Breckinridge in March 1952. Her husband’s jacket had been destroyed. Under the Veterans Proof Act, the case should have been winnable. In 1981 it was denied; a decades-later reconsideration, under a more mature mosaic practice, yielded a partial survivor benefit. The check arrives, smaller than it would have been, and a flag came late to a grave that already had a headstone.
A Blue Water Navy machinist’s mate from San Pedro, whose deck logs were saved and digitized with coordinates, saw service connection recognized after the Federal Circuit decisions broadened exposure zones for coastal operations. The adjudication cited geotagged deck log entries, an inhalation diagnosis from a VA hospital, and a payroll stub for hazardous duty. The documents sat in three repositories for years; together, they resolved a claim that had spun through offices since 2007.
And a Missouri Guardsman who swore to an accident at a weekend range in the early 1960s, backed by a sworn letter from a co‑worker and a clipping from a base paper, was denied when the clipping turned out to be a pastiche. The unit’s morning report for the date in question placed them elsewhere. In that case, triangulation resolved the conflict, and the claimant appealed without success.
Inside the institutions handling these cases, the toll is tangible. At the VA’s St. Louis regional office, adjudicators say they spend too much time re‑indexing scanned images and waiting for outside pulls on microfilm. At the NPRC annex, a Reconstruction Lab established in 2014 to stabilize and digitize charred fragments has cut down on physical handling and added an imaging pipeline that ties directly to Muster Map layers. According to Reconstruction Lab supervisor Maria Esquivel, standard procedure for adhered fragments includes controlled humidification, mechanical separation with microspatulas, high‑resolution imaging, and chain‑of‑custody logging for each step so that digital surrogates can be matched to their source fragments during adjudication.
I wheel stretchers through the John Cochran tunnels. I see the veterans, I hear the stories. I’m mad this week, but not at the affidavits — at the parking tickets. They scheduled a clinic day and ticketed half the visitors on North Grand like it was a ballgame. Fix the basics and maybe the big fixes will land the way you want.
Polk’s complaint is small against the national argument, and useful precisely because of that. On Tuesday, city records show 29 parking citations issued along North Grand between 9 a.m. and 1 p.m., overlapping with the clinic intake. The same day, the payroll microfilm scanning dispute remained on GAO’s bid-protest docket (B‑421602), and NARA’s redaction changes appeared in Bulletin 2023‑05 (Reconstruction Files Privacy Enhancements). Policymakers tend to set rules and hold ceremonies. People in St. Louis live with the work between them.
Standards that would help
If the state set the fire’s policy consequences in motion, it can recalibrate them now. A workable line would look like this. Keep the affidavit pilot and write it into regulation at the end of the test period, but require that every affidavit be paired with at least one locator pulled from sources the claimant did not author. Publish a monthly affidavit dashboard: number received, number paired with payrolls or deck logs, number flagged by VA antifraud, and number substantiated by the Inspector General. Make the fraud denominator part of the Oversight Committee’s practice, not just the numerator.
Direct NARA to narrow the July redactions to fields that pose true privacy risk and to solicit consent at the point of affidavit for name release on the public copy. In exchange, require affidavit forms to carry a perjury warning and standardized attestation language. Fund the verification workflow. The Archives and VA should publish a schedule and performance targets tied to the protested microfilm scanning contract and the 4.2 million images waiting to be ingested. If a delay outlasts the pilot’s 18 months, the policy will have made a promise the agencies cannot keep.
In court, state the standard and keep it measurable. Under the Court of Appeals for Veterans Claims (CAVC) benefit‑of‑the‑doubt rule, adjudicators credit consistent, corroborated accounts when official records are missing for reasons the claimant did not cause. The boundary is maintained by requiring corroboration across sources and documenting how each piece was verified.
There is also unfinished business. Widows and heirs of mid‑century veterans who died before the Veterans Proof Act matured were denied in the late 1970s and early 1980s under narrow interpretations that hardened in practice. Some have won partial retroactive relief; many have not. The “Reconstruction at Fifty” report acknowledges that inconsistency and offers no timeline for a formal review. The Oversight Committee could do more good by asking for a plan to revisit those files than by staging a headline about a 2021 fraud ring that has already been prosecuted.
The hardest critique lands close to home. VA’s leadership embraced the pilot and the Archives produced a serious report, but neither institution has owned the operational math. If adjudicators face a 182‑day average on mosaic cases, what staffing is required to bring that under 120? How many microfilm reels must be scanned per week to feed the pilot’s locator requirement? Which regional offices will be assigned the new affidavit-verification function, and with what training? These are the numbers that will decide whether the anniversary announcement becomes a durable change.
Friesz, the cartographer who opposes lowering the bar, frames her objection in those same terms. She worries less about bad actors than about institutional drift. “You wire in an affidavit, you add a map point, you call it good,” she told me. “I want it harder to close a case.” Her premise is that restraint protects the system. Mine is that transparency and structured corroboration do more work than restraint alone. Either way, the testing will occur at the NPRC Reconstruction Lab benches and in the VA St. Louis regional adjudication bays, where staff reconcile affidavits with payroll microfilm and Muster Map hits.
On Tuesday afternoon, as the folding chairs stacked and families held portraits by the chain-link, a staffer from a veterans service organization carried a box of fresh affidavit packets back toward a tent marked for intake. Two NPRC clerks sealed plastic tubs labeled for the microfilm project. On the sign‑in sheet for the clinic, 43 names filled the lines by 2:15 p.m.