Transcriptions and manuscript decisions

Back to the article.

The manuscript is AAV, Segr. Stato, Spagna 1A, ff. 70r–73v, DECODE record 92. The two transcriptions represent different stages of reading the same letter.

File	Contents
public.txt	EHum's complete transcription, dated January 9, 2016, including its original header, cleartext, and comments.
transcription_ms.txt	The author's manuscript-corrected transcription, checked against the DECODE photographs on September 15, 2026.
corrections.tsv	Proposed corrections, with position, replacement, confidence, and reason; only sure or probable rows are applied.
decisions.tsv	Individual glyph-review decisions, including readings retained and cases left open.
dot_decisions.tsv	Review of the 45 places where the edition supplied or ignored an above-mark.

The ciphertext appendix presents the public transcription's cipher portions in a consistent notation. The edited Italian and alignment use the corrected transcription. Both transcriptions retain uncertain readings and gaps; neither is a transmitted plaintext.

Counts and notation

The public transcription contains 113 cipher lines and 6,577 digit-position tokens. A token can be a digit, an unreadable ?, or alternative readings such as 3/7. There are 44 unreadable tokens and 207 tokens marked above: 178 dots, 25 commas, and four dashes. Two-digit cipher codes still occupy two token positions. Above-marks and uncertainty marks do not add positions.

Folio	Public transcription	Corrected transcription
70r	183	184
70v	1,242	1,242
71r	1,246	1,246
71v	1,190	1,171
72r	1,298	1,296
72v	491	492
73r	927	931
73v	0	0
Total	6,577	6,562

The public file is an unchanged copy of the MysteryTwister transcription. Its SHA-256 is 5402202eebefdb8fb967c27491255e128bcbfec7bfa1d04fa45b8a2d300cec45. The corrected file adds a manuscript-check header and applies the recorded cipher corrections. It retains the public cleartext, including the insecure departure numeral XXV14^- on 70r.

Manuscript check

All 196 ranked classifier suspects and all flagged differences in digit count were examined against the ink. The decision tables preserve the author's reading and confidence, including decisions that support the public reading over the classifier. The check did not examine every glyph the classifier accepted. The manuscript photographs remain restricted; the article's illustrated details show selected comparisons.

The correction table has 129 rows: 88 substitution proposals, 29 deletions, and 12 insertion rows supplying 14 digits. Two substitution proposals remain open and are not applied. The corrected text therefore incorporates 86 substitutions, 29 deletions, and 14 inserted digits. Five duplicated groups account for 21 of the deleted digits:

Folio and cipher line	Original positions	Digits removed
71v L04	30–34	5
71v L05	21–23	3
71v L09	46–51	6
71v L20	42–45	4
72r L07	27–29	3

Thus 6577 − 29 + 14 = 6562. Substitutions include two swapped pairs. On 70r, the hooked 1 in Montepulciano supplies u. On 70v, a dot and two 1s give che vi va. Under the stamp on 72r, the corrected digits give la S.M.tà [h]a voluto, identifying the Emperor as the person who requested the public offer. The dot decisions distinguish real dots from neighboring strokes: the mark in risposta at 70v L05 p26–27 is the top bar of a 5; the dot in Marchese at 73r L01 p57–58 is real but misplaced.

Reading the tables

• folio, line, and pos identify the public transcription's positions, not positions renumbered after correction. Lines and positions are one-based; only cipher lines count. Folio labels retain their leading zero.
• In corrections.tsv, sub replaces a token, del removes it, and ins inserts the value before the stated position. A trailing ^ in a replacement retains the old above-mark, or supplies a dot if none was recorded. A replacement without ^ clears the above-mark; substitutions preserve following punctuation and remove the old uncertainty flag.
• Only changes judged sure or probable enter the corrected text. open decisions remain unresolved. Confidence is the reviewer's judgment.
• transcribed, machine, edition, and page distinguish the original label, classifier proposal, editorial requirement, and visual reading. Notes sometimes describe a reading before the subsequent correction.
• Identifiers such as dots_01 in the notes identify review-sheet panels. These panels used restricted photographs; the folio, line, and token positions are the locators for the published decisions.

The alignment before the check and alignment after it preserve the two states used in article table 5. Both the transcription and the editorial reading changed. Their counts therefore describe edition states, not an isolated measure of transcription accuracy.