Excavation · The Casket Letters (Mary Queen of Scots, 1567 to 1587)
Area: History · Question: Is the document or object genuine?, Who wrote or made it? · Starts from a widely shared claim
The short answer
Unsettled, leans slightly toward yes
Current assessment · as of 2026-10-02
Unsettled. The evidence that would decide it does not survive.
Which way the evidence leans: slightly toward Yes, genuine: written by Mary. This is not a finding.
Central claim: The Casket Letters are Mary's genuine letters to Bothwell.
Evidence ●●○○○ Public belief ●●○○○ Source checked: no
What would settle it. An authenticated original, or a measured comparison on enough text, with a control of the same kind of writing, that survives review by someone outside the project.
The originals are lost (provisional confidence, because that rests on absence). What survives is a few French copies of four letters, about 1,750 words in all, which is too short for word-frequency tests, and Latin and Scots translations of the rest. The one measured comparison we know of leans slightly toward Mary's own phrasing, but it is our own unreviewed result and a difference in kind of writing could explain it. Many modern accounts treat the letters as doctored (National Museums Scotland says 'widely thought'), but that is adoption, not a test. The 1586 forgery of a different letter shows forgery was possible; it adds no evidence about 1567. We therefore do not say 'genuine' or 'forged'.
The one measurement we have, on the four surviving French letters, finds more shared phrases with Mary's authenticated French letters than with Catherine de Medici's, at levels typical of Mary's own letters (claim casket-french-lean-mary).
Why the lean is not a finding. It is our own result and no one outside the project has reviewed it. A difference in kind of writing (private letters against formal papers) could explain it and has not been controlled for. The text is about 1,750 words, too short for the stronger tests. The surviving French copies may themselves be back-translations from Latin (claim casket-french-printed-retranslation, proposed). A lean is not a finding: it stays under 'Unsettled' until a basis claim reaches moderate confidence.
Based on: casket-authenticity-contested casket-originals-lost casket-too-short-for-wordfreq casket-french-lean-mary casket-babington-postscript-forged-1586
We give no probability figure. Confidence words describe how sure we are of the basis, and no source supplies a number that would not be invented.
Challenge this assessment A challenge needs a specific claim and evidence we can check. How challenges work.
established confidence provisional · what a source says
The original Casket Letters are lost: only copies, translations and printed versions survive.
Anchor. Henderson (1889), ch. II-IV [read in OCR]: Morton kept the casket from 20 June 1567; Regent Moray received it from the Privy Council on 16 Sept 1568 and took it to York and Westminster; Morton had it back, and 'after Morton's execution they were delivered by his natural son, James, Prior of Pluscardin, to the Earl of Gowrie', then probably passed to Arran; the 'original copies of the Scottish translations disappeared completely from human ken'. No original has been traced since.
Would change if. an original surfaced and was authenticated
Next step. Read Henderson ch. IV in full and the Bowes correspondence he cites; separate the loss (documented) from who destroyed them (proposed, below).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
Assumptions this rests on.
Alternatives considered.
Why this confidence. Starts at moderate (The anchor was checked against a secondary source (Henderson 1889, read in OCR)).
These reasons were drafted by the project and have not yet been reviewed.
In June 1567 Mary's enemies said they had found a silver casket with letters and sonnets from her to the Earl of Bothwell. Read as genuine, the letters tie her to the murder of her husband Lord Darnley. They were shown to Elizabeth I's commissioners in 1568 and the originals later vanished. This dig keeps four things apart: that the originals are lost (documented), what the surviving copies and translations are (copies of unknown fidelity), whether the letters are genuine (contested; no independent test has decided it), and the 1586 forgery in Mary's correspondence (documented, but a different case that lends the 1567 letters no weight). A narrow computed test of the four surviving French texts is reported with its limits.
Evidence and belief diverge in two places. First, the composite view (real letters altered or spliced) is the most adopted among modern historians but, like the genuine and forged views, rests on interpretation, not an independent test. Second, the popular 'Mary was framed' framing borrows credibility from the documented 1586 forgery, which says nothing about the 1567 letters.
Every point on the timeline links to its source. Open the timeline.
Objections raised against this excavation, in the form people raise them, with the test we ran and where it stands. Open ones are shown as plainly as answered ones. To add a challenge or evidence, use the button on the claim: it must name the claim and give a source we can check, because disagreement, popularity and repetition carry no weight. How challenges work.
Mary was framed by the state.
partly answered raised by: Popular framing; no single source
Documented (1586): English intelligence forged text in Mary's correspondence. Unproven (1567-68): that the Casket Letters were forged by the Scottish lords. The sources for the 1586 forgery have not been read directly in this project.
Separate the documented 1586 forgery from the 1567 letters.
casket-babington-postscript-forged-1586 casket-authenticity-contested
Since the 1586 postscript was forged, the 1567 letters were forged too.
answered: finding stands raised by: Common argument by analogy
It shows forging Mary's letters was possible and was done. It adds no evidence that the 1567 letters were forged.
Ask whether a forgery by different people, in another country, 19 years later is evidence about the 1567 letters.
casket-babington-postscript-forged-1586 casket-authenticity-contested
The letters are Mary's genuine letters to Bothwell.
open: not yet testable raised by: Traditional view
No originals survive and only French copies of four letters exist. Contested, low confidence. No computational comparison against Mary's authenticated letters was found.
Look for surviving originals and for any published measured comparison.
James VI destroyed the originals.
open: not yet testable raised by: Historians' proposal, around 1584
Proposed at low confidence. That the originals are lost is established at provisional confidence only, because it rests on absence.
Check whether any primary document records the destruction.
A word-frequency study can settle who wrote the letters.
partly answered raised by: Common expectation
About 1,750 words survive across the four French letters, well below what word-frequency methods need. A phrase-sharing comparison leans toward Mary at low confidence only, and no published study exists.
Measure how much French text survives against what the method needs.
casket-too-short-for-wordfreq casket-no-computed-study-found
Status: published before the review gate existed; partial review, no independent reviewer · reviewed 2026-10-02. Published by the owner before the gate existed. Read directly: Henderson (1889) in OCR text. The 1586 forgery, the French copies' holdings and the contested-authenticity positions rest on Henderson or on the Q-001 page and were not each re-read; the NSA account was not read (nsa.gov blocks this environment). Computed results were reproduced from the published code. Headline status is draft. The review record.
established confidence provisional · what a source says
The original Casket Letters are lost: only copies, translations and printed versions survive.
Anchor. Henderson (1889), ch. II-IV [read in OCR]: Morton kept the casket from 20 June 1567; Regent Moray received it from the Privy Council on 16 Sept 1568 and took it to York and Westminster; Morton had it back, and 'after Morton's execution they were delivered by his natural son, James, Prior of Pluscardin, to the Earl of Gowrie', then probably passed to Arran; the 'original copies of the Scottish translations disappeared completely from human ken'. No original has been traced since.
Would change if. an original surfaced and was authenticated
Next step. Read Henderson ch. IV in full and the Bowes correspondence he cites; separate the loss (documented) from who destroyed them (proposed, below).
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
Assumptions this rests on.
Alternatives considered.
Why this confidence. Starts at moderate (The anchor was checked against a secondary source (Henderson 1889, read in OCR)).
These reasons were drafted by the project and have not yet been reviewed.
established confidence high · our conclusion
In 1586 Walsingham's decipherer Thomas Phelippes forged an enciphered postscript to Mary's letter to Anthony Babington, asking for the names of the conspirators.
Anchor. Surviving at the UK National Archives and described consistently by the NSA's history and archive sources (per open question Q-001 (formerly bounty B-001); not read in this dig). A DIFFERENT case from 1567: it shows forging Mary's letters was possible and was done, and adds no weight to the claim that the 1567 letters were forged.
Would change if. the archived document being shown to be in another hand or of another date
Next step. Read the document at the UK National Archives (or its published catalogue entry) and the NSA history 'To Catch a Queen'; nsa.gov was blocked from the cloud session.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
established confidence moderate · what a source says
French copies of only four of the letters survive: Nos. 3 and 5 at the Record Office and Nos. 4 and 6 at Hatfield. Letters 7 and 8 exist only in print, and Letter II survives in French only as its opening sentence and a phrase Cecil wrote in the margin.
Anchor. Henderson (1889, p. 68), as quoted on the open question page: 'The copies of the original French versions of the letters now known to exist are Nos. 3 and 5 in the Record Office, and Nos. 4 and 6 at Hatfield.' The book was read in OCR in this dig; the archive holdings themselves were not seen.
Would change if. another French copy being found, or Henderson's list being corrected
established confidence moderate · our conclusion
The surviving French text of the four letters is about 1,750 words (about 630, 500, 240 and 400 for Letters III-VI, +/-20%), well below the length at which word-frequency authorship methods are reliable (Eder 2010: 'simply disastrous' below about 3,000 words for novels; about 2,500 for Latin prose at best).
Anchor. Line count of Henderson's appendix by the open question author (OCR, +/-20%); Eder (2010) as cited on the open question page, not re-read here.
Would change if. a recount from the archive copies, or evidence that a word-frequency method works on texts of this length
contested confidence low · our conclusion
The Casket Letters are Mary's genuine letters to Bothwell.
Anchor. Positions as summarised in open question Q-001 (formerly bounty B-001) from the sources it lists; not each re-read here. Authenticity of the SOURCES (the copies) is tracked separately in sources/MANIFEST.yaml; 'genuine' here means authored by Mary as letters to Bothwell.
Would change if. an independent test (handwriting of an original, or a validated computed comparison on enough text) that discriminates the positions, or the text being found to contain anachronism
Disputed by.
proposed confidence low · our conclusion
James VI destroyed the originals (around 1584).
Anchor. Henderson (1889) lists three candidates and argues from the Bowes correspondence that the box passed to Gowrie and then probably Arran; he does not report a record of the destruction. Held apart from the loss itself.
Would change if. a contemporary record of the destruction, or of their later whereabouts
Next step. Read Henderson's chapter 'By whom were they destroyed?' (By Mary's accusers? the Marian party? James VI?) and the Bowes correspondence.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence moderate · our conclusion
The French versions printed in the 1570s are not copies of the originals: they were translated back into French from the Latin, which was itself translated from the Scots.
Anchor. Argued by Henderson (1889-90). Henderson's account of publication: the Latin and Scots Detectio of 1571 and a French translation of 1572 (ch. IV, read in OCR).
Would change if. a close comparison of the printed French with the French copies at the Record Office and Hatfield showing the printed French derives from them
Next step. Read Henderson's appendix (the Latin and French versions) side by side with the copies' texts.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
proposed confidence low · our conclusion
The four surviving French casket letters (III-VI) share three-word phrases with Mary's authenticated French letters more than with Catherine de Medici's, at levels typical of Mary's own letters.
Anchor. Test DT-002 (see tests.yaml). Median Mary:Catherine ratios: Letter III 1.59, IV 1.49, V 1.55, VI 1.35 (90% ranges 1.25-1.78, 1.23-1.88, 1.10-2.66, 0.94-1.64). Calibration: held-out Mary letters placed correctly 57/60, Catherine's 58/60. The pre-set bar of 3x was unusable and is reported (finding F10). The lean may reflect personal-letter language, not authorship (F11). It fits a genuine letter and a composite built from real letters equally; it does not show guilt.
Would change if. a genre-matched control (G-05) erasing the lean, a reviewer finding an error, or a rerun on cleaner texts
Next step. Control with personal letters by another French-writing woman of the 1560s (G-05); outside review before leaving proposed.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
searched gap confidence low · result of a search
No computational style comparison of the casket texts against Mary's authenticated French letters has been published, in two web searches.
Anchor. Two web searches by the open question author found no such study; no wider search yet.
Would change if. a published computed comparison being found
Next step. Search Google Scholar, Cryptologia and Digital Scholarship in the Humanities; ask the 2023 decipherment team. NARROWED (finding F7): manual comparisons do exist (Bresslau 1882, Sepp); what is unfound is a systematic computed comparison across all surviving French text.
Start this step → (signs you in with GitHub; trusted contributors can send it to the dig agent)
These pages are generated from plain YAML: claims.yaml · sources manifest. If you can show a finding is wrong, challenge it.