# The death of Meriwether Lewis, 11 October 1809

**Verdict: Lewis shot himself. Self-inflicted about 85 percent, homicide about 15 percent, accident under 1 percent. Evidence label: Probable.** The evidence clearly favours self-infliction over the other answers but does not prove it, and the percentage moves when the same evidence is weighed differently.

The full report is `report.html` in this folder, titled "The Death at Grinder's Stand". It is on the site at https://historical-mysteries.netlify.app/meriwether-lewis/report.

## What happened
On the evening of 10 October 1809 Meriwether Lewis, who had led the Lewis and Clark expedition, rode up alone to Grinder's Stand, a settler's cabin that took in travellers on the Natchez Trace, the old road from the Mississippi to Nashville, about 72 miles south-west of Nashville in what is now Lewis County, Tennessee. He was 35. He had led the expedition to the Pacific of 1804 to 1806, had been private secretary to President Thomas Jefferson before it, and since 1807 had been governor of the Louisiana Territory at St. Louis. He was riding to Washington to defend his public accounts: the War Department had refused to pay bills he had drawn as governor, his private creditors had closed in, he was at odds with his deputy Frederick Bates, and the book of the expedition was three years overdue. He had come down the Mississippi ill with fever, and Captain Gilbert Russell, commanding Fort Pickering (the army post where Memphis now stands), had found him deranged and kept him a fortnight. With him on the road were James Neelly, the government's agent to the Chickasaw nation, Neelly's servant, and his own servant John Pernier. Neelly had stayed a day behind to find two strayed horses. The innkeeper, Robert Grinder, was away; his wife Priscilla gave Lewis the house and slept in the kitchen cabin, and the servants slept in the stable. In the night she heard two pistol shots. The servants found Lewis shot in the head and in the body, and by some accounts cut with a razor; he said he had shot himself, asked for water, and died about sunrise. Neelly buried him near the stand and wrote to Jefferson that the governor had killed himself. Jefferson, William Clark (Lewis's partner on the expedition) and the newspapers accepted it. No doctor saw the body and no inquest was held.

## Why it is a mystery, and what each side says
None of the people who wrote about the night was in the room. Everything comes from Mrs Grinder, who told the story at least four times over 36 years and told it differently each time, and from two servants whose words survive only as others reported them. Nobody recorded powder marks, the position of the pistols or the state of his hands. The case for suicide was made by the people who knew him: Jefferson wrote in 1813 that Lewis had suffered "hypocondriac affections" (depression, in the usage of the day) from early life; Russell reported two attempts on his own life on the river; a letter written at Nashville thirteen days before the death says the same; Neelly found him "at times deranged in mind" on the road. Most historians since Dawson Phelps, the National Park Service historian of the Trace (1956), and Donald Jackson, editor of the expedition's letters (1962), have held that view. The case for murder began in Tennessee: in 1848 the state legislature appointed a committee to put a monument over the grave, and the committee's report to the legislature of 1849 to 1850 said that death "by the hands of an assassin" seemed "more probable"; local tradition accused the innkeeper and the Lewis family accused the servant. The novelist Vardis Fisher's *Suicide or Murder?* (1962) argued that every document was second-hand and every witness interested. Since then the surgeon E. G. Chuinard (1991) has said Lewis could not have survived and moved as described and named Neelly; the historian John Guice (1998 to 2006) has argued that the story never fitted, from the moonless night to the two hours Mrs Grinder waited; the forensic scientist James Starrs and the independent researcher Kira Gale (2009, 2016) have argued for a plot by General James Wilkinson, the army's senior officer, and called Russell's 1811 statement a forgery; and Tony Turnbow, a historian of the Trace, has argued for a robber and a cover story arranged by Neelly. Thomas Danisi, Lewis's biographer, holds that Lewis fired the shots in a malarial delirium and that "suicide" is the wrong word. A coroner's jury in 1996 recommended exhumation; a federal court in 1998 and the Department of the Interior in 2010 were asked to allow it; the grave has not been opened.

## How I judged it, and how sure
I read every document of 1809 to 1813 that bears on the death and the modern debate from Fisher to 2025 (in full where it is in articles, in fragments where it is in books I could not obtain), put the events in date order, laid Mrs Grinder's tellings side by side, checked her one visual claim against the moon, and compared the reported wounds with modern forensic case series. Then I scored sixteen evidence items in an explicit model (`code/evidence_model.py`, inputs `data/evidence_items.json`): each item's weight is entered as a range of how much more likely the evidence would be if Lewis shot himself than if someone else did, and the ranges are combined under four starting assumptions (priors), two sceptic settings that discount every item, and a leave-one-out check. Central run (even prior, conservative grouping): 92% self-inflicted, 90% band 36 to 99. Sceptic runs: 76% and 63%. A murder-leaning prior (3 to 1) with every item quartered gives 32%. No single item moves the central result by more than 9 points (dropping the valuables-left item gives 83%). Accident, entered as a simple discharge, is under 1% in every base run; the sceptic runs lift it to 2% under the even prior and to at most 2.5% under the murder-leaning prior; and it reaches 7 to 17% only under a flat prior that starts it at a third. Every figure depends on the sixteen items, their ranges and the priors chosen; it summarises judgments and is not a measurement. My headline of 85% sits below the central run because the model cannot allow for how badly the body was examined: nobody who saw it knew what to look for (range, soot, hands). Self-infliction is therefore the probable answer rather than a demonstrated one.

The case rests mainly on the month before the death rather than on the night itself, for which there is no reliable witness. Attempts on his own life were reported thirteen days before he died (James House to Frederick Bates, Nashville, 28 September 1809) and are attested through three chains that do not depend on Russell's disputed statement of 1811. Jefferson and Clark accepted the report at once. The two men who could have robbed him went straight to his patrons, and his pistols, rifle, watch, horses and two trunks of papers were all accounted for. The reported wound sequence, a head shot that missed the brain and then a shot to the trunk with hours of survival, is one that suicides with a pistol are documented to produce, and every forensic pathologist who has examined the surgical claim that it is impossible has rejected it. Every murder theory requires at least one of the three people at the stand to have lied about a man the servants found alive and talking, and the conspiracy version requires the record to have been false from a fortnight before the death. Murder keeps about 15 percent for two reasons: the body was never examined for the signs that would decide a shooting, and the model cannot allow for the possibility that the three people at the stand agreed on a false story.

"Self-inflicted" covers both a decision to die and a delirium in which the same acts were done without one. The record cannot separate them, and the verdict does not depend on which illness lay behind the derangement.

## Where historians stand, and what this adds
**Agrees with the majority view.** Suicide has been the majority position among historians since Phelps (1956) and Jackson (1962); Paul Russell Cutright, a historian of the expedition (1986), James Holmberg of the Filson Historical Society (1992), the physician David Peck (2009) and the National Park Service hold it; Danisi accepts that Lewis fired the shots and rejects the word suicide. The case for murder (Fisher 1962, Chuinard 1991 to 92, Guice 2006, Starrs and Gale 2009, Gale 2016, the Tennessee monument committee (appointed 1848, report 1849 to 1850) and the 1996 coroner's jury) is stated in its authors' words and answered by name in section 9 of the report. The field is not settled. Among authors who have stated a position in print since 2006, I count 13 for suicide, 6 for murder in three versions that exclude one another, 3 for a self-inflicted death that was not suicide, and 6 or 7 undecided (`notes/D_scholarship_theories.md`, section 7c). "Agrees with the majority" means that majority and the documentary case behind it.

What this page adds to that case is its basis. It moves the weight from testimony about the night, which is all second-hand, to two other things: the evidence of attempts and derangement in the month before the death, reported through separate chains (House's informant from the Bluffs, Russell, Neelly, Pernier as Clark reported him, and Mrs Grinder as Neelly and the ornithologist Alexander Wilson, who visited the stand in May 1810, reported her), one of them written down thirteen days before the death; and base rates for multiple-shot and complex suicides from small modern series, which show that the reported wound sequence is a recognised shape of a failed-then-completed suicide.

## What is new (as far as I can tell from what I read)
None of it is new evidence; every document was in print. What is new is the reading and the weighing.
- John Pernier's own petition to Jefferson of 10 February 1810 (Founders Online, Jefferson/03-02-02-0171), read in full and placed in the sequence of his last months: destitute, refused by the government, asking $271.50 in wages "to pay my just debts & resqued from prison", dead of laudanum eleven weeks later. Chalkley 2004, the fullest study of Pernier, did not use it; Gale 2016 used its $271.50 and its fear of prison as the record of a silenced witness; I read it as the record of a destitute one.
- Three attestations of the suicide attempts on the boat that do not depend on Russell's disputed 1811 statement (House to Bates 28 Sept 1809, written before the death; Russell's lost letters as summarised by Clark to his brother, 26 Nov 1809; the Washington newspaper, the National Intelligencer, 27 Nov 1809), as an answer to the claim that the statement is a forgery. Only House's chain is certainly independent of Russell as a source.
- Forensic base rates (how often each feature occurs in modern case series) applied to the wound pattern: multiple-shot suicides are 1.6 to 8% of firearm suicides, the first head shot missed the brain in 5 of the 11 cases in the largest series (Karger and Brinkmann), torso shots predominate (Jensen 2025), unplanned complex suicide is a named category, and survival with the power to act after a trunk wound is ordinary.
- The evidence model, with its sensitivity printed beside the headline.
- Small points about dates and documents: Wilson's visit was Sunday 6 May 1810, as the editions of his works by George Ord (1828) and Alexander Grosart (1876) already implied (the "May 28th, 1811" printed in the Port Folio, a Philadelphia magazine, misled Clay Jenkinson in 2011); Russell's $99.58 check "none of which it is said could be found" reached Washington (the editors of the Retirement Series of Jefferson's papers); the monument committee's "assassin" (report of 1849 to 1850) rests on two stated grounds, the loneliness of the place and a story that the servant fled to Natchez, which the documents refute; and two pieces of contemporary St. Louis testimony (the St. Louis Missouri Gazette, 2 Nov 1809: fever "never failed of depriving him of his reason"; Major Amos Stoddard, Lewis's friend, 28 Sept 1809: "under the hands of the surgeons mate Dr. Smith") are not used in any of the works I read.

## Whose lines this follows
Dawson Phelps 1956 and Donald Jackson 1962/1978 (the documentary case); James Holmberg 1992 (Clark's letters); John Guice 2002 (the moonless night, which my calculation confirms); Ann Rogers 2005 (the shot sequence is a presumption); Michael Carrick, a firearms historian, 2006/2009 (the pistols); Tom Chalkley 2004 (Pernier); David Peck 2009/2022 (the medical answer to Chuinard); Thomas Danisi 2013 to 14 and Tony Turnbow 2012/2015 (the Neelly dispute, left open); Clay Jenkinson 2011 (Mrs Grinder as a witness against her own interest).

## What I could not read, and other limits
- Not read in full: Fisher 1962, Guice ed. 2006, Starrs and Gale 2009, Danisi and Jackson 2009, Danisi 2012, Gale 2015, Dillon 1965, Ambrose 1996, Kushner 1981, Phelps 1956, Jackson's Letters of the Lewis and Clark Expedition (the inventory of Lewis's effects taken at Nashville on 23 November 1809, and the full Russell statement), Chandler 1994, Peck and Peck 2021, Stroud 2018, Morris 2004, Turnbow's two books, Jenkinson beyond chapter VI. I read everything on the death in We Proceeded On, the journal of the Lewis and Clark Trail Heritage Foundation, from 1974 to 2025, and the books only as complete sets of search snippets. The digital libraries JSTOR, HathiTrust and Google Books were out of reach, and lending copies gave snippets only. So the principal opposing works were read in full where they are articles and only in fragments where they are books.
- Not obtained at all: the full text of the 1845 "schoolteacher" account (a manuscript version is in the Draper manuscripts at the Wisconsin Historical Society, 29CC33); the 1996 inquest transcript; the Avery 1978 thesis; the print-only WPO issues of Nov 2025 and Feb 2026; House's letter beyond the quoted phrases; the Clarion's own issue (read in the Kentucky Gazette and Enquirer reprints); the 1849 to 50 House Journal appendix (read in the transcription Elliott Coues verified for his 1893 edition of the expedition's history); the Marks and Neelly letters of 1811 to 12 except as Fisher and Gale quote them.
- Forensic literature in abstracts only; modern base rates used as orders of magnitude.
- No archival work.
- The Kentucky Gazette and Missouri Gazette texts were read from page images and are not verified by machine text recognition (OCR).
- The notes on the documents (`notes/A_documents_report.md`) list 24 points on which the record of 1809 to 1813 disagrees with itself and 20 documents not obtained. One is left open in the report: Lewis's watch is "Silver" in Neelly's and Brahan's letters of 18 Oct 1809, "His Watch" reached Washington by 5 Jan 1810 (Coles), and an unsigned 1812 note reported by Gale (WPO 42:4, n. 30, not obtained) says Neelly still held a "gold watch"; whether that is one watch or two cannot be settled from the copies I read. It does not touch the verdict: no 1809 document puts a watch with Pernier.

## What would overturn the verdict
Exhumation showing a penetrating frontal entry from behind or above, a trunk trajectory a man cannot give himself, a ball of a calibre no gentleman's pistol fired, or a range beyond arm's reach; a record of an 1809 inquest naming a killer; the full 1845 text proving checkable; House's letter proving misdated; anything placing Neelly at the stand on the night of the 10th.

## Files
- `report.html` (the page), `report-src/report.html` (the same file), `report-src/DESIGN.md`
- `sources/` (48 primary documents and document notes with citation, URL, fetch date), `sources/medical/` (55 files), `sources/scholarship/` (109 files), `sources/setting/` (43 files)
- `notes/A_documents_report.md` (the record file by file, 24 discrepancies, 20 documents not obtained), `notes/B_medical_forensic.md`, `notes/C_timeline_setting_versions.md`, `notes/D_scholarship_theories.md` (the debate 1809 to 2026 work by work, seven homicide theories in strongest form, provenance disputes, the 2010 decision, a table of positions), `notes/ACCESS.md` (how each source was reached)
- `data/evidence_items.json`, `data/evidence_model_results_one.json`, `data/evidence_model_results_product.json`, `data/astronomy_1809.txt`, `data/fo_index_*.tsv`, `data/wpo_death_index.tsv`
- `figures/fig1_night_by_source.svg` and `.png` (the night as each source tells it), `figures/fig2_shares_by_prior.svg` and `.png` (the share for self-infliction under each starting assumption and discount)
- `code/evidence_model.py`, `code/astronomy_1809.py`, `code/wbfetch.py`, `code/fo_index.py`, `code/fo_source.py`, `code/make_figures.py` and `code/render_png.js` (the figures)