System A′: the Bavarian agents’ reports from Hungary and a letter of King John, 1534–35

TASK: take a partial decipherment further: System A′: the Bavarian agents’ reports from Hungary and a letter of King John, 1534–35
This page: https://dbourdeau.github.io/cyphersolver/further/kaa4591.html

You are helping a public research project finish reading historical cipher letters. This one is read in part:
the key works, but some pieces are still open. Your job is to close one or more of the open gaps below.

THE READING
  Write-up:   https://dbourdeau.github.io/cyphersolver/kaa4591.html (read it first: how the text was read, the key, the reading)
  Date:       1534–1535
  Outcome:    key from adjacent plaintext · partial
  Read now:   46.5% of the deciphered cipher tokens give sense (pieces below may be untouched)
  Files:      https://github.com/dbourdeau/cyphersolver/tree/main/targets/kaa4591 (NOTES.md, key, transcription, reading, scripts)

OPEN GAPS (from NOTES.md, "## Remaining gaps")
  1. [no-key-material: needs a new source] R9409 and R9427 decoded, not read (2,924 words; holes in most sentences) - blocker: no-key-material; re-read 23 Oct 2026 (b20): all DECODE files opened, no decipherment or edition, R9427's gloss illegible beyond the salutation; System A' has no key sheet and the per-letter keys do not settle the open signs (incl. the lone c+ 'da', 53 places in R9409, probably a title sign, flagged for review)
  2. [no-key-material: needs a new source] R9408 (5 Oct: 125 of 1,530 unread; second image pass 20/72 lines done): 170 of 1,528 words unread ({?} in `r9408/reading.txt`: P1.01-06 and P1.20-26, the P2.01 greeting, the ratschlag clauses P2.06/25-30, P3.10-16, P4.16-28, P5.01-08, P6.04-07/28-30, the signature forename) - blocker: no-key-material (workable); the signs are legible but take no value the letter settles; the y (ÿ/γ) and 4 (4/ч) look-alikes were split at the image 2 Oct 2026 without opening further words; next: a second reading round.
  3. [no-key-material: needs a new source] R9413 (5 Oct: 94 of 1,647 unread): 232 of 1,663 words unread (P1 opening and P1.19-26, P2.05-12, P3.04-12 the ore description, the 4o group on P5, P6-P7 short lines, P8 address) - blocker: no-key-material (workable); next: image check of the noisy P1 opening lines and of the 4o group, second reading round.
  4. [no-key-material: needs a new source] R9409 (5 Oct: 304 of 2,381 unread, after image re-transcription of 181 lines): 450 of 2,361 words unread (P1.01-03 salutation and date, P7 about half, names at P3.05-30 and P6.18/31, runs at P2.11, P3.04/07/34, P4.03/05/22, P5.18-19, P6.05/09; the F.G.-like sign read as a letter at P1.18, P5.23/26, P6.24) - blocker: no-key-material (workable); c+ checked at the image (one shape for da/ch/f; sch a separate sign); next: second reading round of P7.
  5. [no-key-material: needs a new source] R9410 f.244r, decoded, not read since 23 Oct 2026 (v3: 0.323 explicit-hole) (5 Oct: 43 of 303 unread): 49 of 304 words unread (P1.03-05, the a# group, P1.29-32, the place in the date line) - blocker: no-key-material; System A' has no key sheet, three reading rounds made
  6. [no-key-material: needs a new source] R9410 ff.246-247, decoded, not read since 23 Oct 2026 (v3: 0.368 explicit-hole) (the Latin letter, Buda 30 June 1534; 5 Oct: 47 unread): 72 of 422 words unread (the P3.01-02 opening, P3.13 and P4.09-11 clauses, the name groups e9vn8/e9v9q4 at P4.12 and P4.20, the a# group, P6.04-05) - blocker: no-key-material; System A' has no key sheet; transcribed from the images and read 2 Oct 2026.
  7. [no-key-material: needs a new source] R9427 (5 Oct: 95 of 545 unread; image re-transcription not yet done): 123 of 542 words unread (ends of P1.01-03, P1.09-14, names P2.03-23, the signature) - blocker: no-key-material (workable).

ALREADY TRIED (from NOTES.md, "## Escalation")
  [x] siblings: R9368 (glossed sibling of System A), R9407 (augurelio1535), R9411/R9412 (sperantio1534, system L, a different key on look-alike shapes) and the R9423 register checked.
  [x] clear-pages: glosses used on R9427 and R9410 (and the R9368 gloss key tried on R9410: failed).
  [x] known-keys: the R9368 gloss key, the R9427 gloss key and the improved R9407 key (augurelio1535 pass 2) tried; the last fits by sign shape.
  [x] print: web search for KAA 4591 / Augurelio ciphers found no edition.
  [x] key-rebuild: System A′ from the R9427 gloss, extended by constrained annealing over R9408+R9410 with the gloss values fixed (open signs settled: # t, o a, X e in running text, A u, S h, | e). Scoring E as d vs ch over the combined text: ch scores better (−3.458 vs −3.537 per char) but gives "charzu" for darzu, so E is two look-alike signs (d and ch) merged in transcription — needs a visual re-split on the images.
  [x] retry: System A′ letters re-decoded with the extended key and each E resolved as d or ch by de-1500s context (`sysA_decode.py`); decrypts in `r9408/decrypt.txt`, `r9409/decrypt.txt`, `sysA/r9410_decrypt.txt`, `r9413/decrypt.txt`. Remaining noise is transcription-level (merged look-alikes, doubtful signs), not key-level.
  [x] key-rebuild: /retry (2 Oct 2026, System A′): the improved R9407 key (augurelio1535 pass 2) carried to R9408, R9409, R9410, R9413, R9427 by sign shape, re-learned per letter by EM and word-beam re-estimation (`sysA2/`), R9408 look-alike split finished at the image (`r9408/split_p6p7.md`), hand reading rounds over every line of all five letters (`r94xx/reading.txt`), a supervised re-key from the readings (no gain). Readings 77-89% of words, 84.0% overall.
  [x] retry: (2 Oct 2026, second pass): R9410 ff.246-247 transcribed from the images and read (82.9%); the look-alikes R9408 y/4 and R9409 c+ checked at the images, split, re-decoded and regraded (no gain).
  [x] retry: (23 Oct 2026, b20): R9409/R9427 DECODE files and R9427 gloss re-checked; partial readings in brief notation; still decoded, not read
  [x] retry: (3-5 Oct 2026): every unread line re-transcribed at the image (R9408, R9413 twice, R9409), sign-constrained pooled-key candidates (fill.py) with second-occurrence checks, reading rounds; 83.9% -> 89.0% of words; R9427/R9410a image passes and an R9408 second pass still to do.

STEPS
0. Set up. If you can run a shell, fork and clone the repository and work inside it:
     git clone https://github.com/dbourdeau/cyphersolver.git
   Then read, in this order: CONTRIBUTING.md (how a contribution is sent and credited); README.md, sections
   "Repository layout" and "Conventions" (how readings are measured and named); targets/kaa4591/NOTES.md in full, and the
   key, transcription and reading files it names; lang/README.md (shared language models).
   Work in targets/kaa4591/. Add your findings to NOTES.md under a dated heading without rewriting what is there, and
   update the "## Remaining gaps" line of every gap you close. Keep WORKLOG.md there: one dated line per step
   (hypothesis, value tried, solver run, crib, control), failures included. Do not edit docs/ (the site is generated).
   If you cannot run a shell, fetch the files from the links above and keep your work in files you send with the issue.
1. Pick a gap. Start with the ones marked "doable now" (open-codes, not-attempted, left-open): the images and the key
   are online, so they need only work. A gap marked "needs a new source" (no-key-material, too-short, illegible,
   needs-physical-access) can be closed only by something new: a key
   or a deciphered sibling found elsewhere, a better image, a printed edition or calendar, a photograph from the archive.
   For those, search for that material first, and say where you looked even when you find nothing.
2. Start from the existing key and transcription in the folder. Do not rebuild the key from scratch, and do not
   re-read parts that are already read unless you find an error in them.
3. For a code group or sign value: list every place the group occurs, show how each reads with the value you propose,
   and say what still does not fit. For a passage: re-read it at the image and give the folio and line. For a missing
   letter or leaf: say where it is held and how it can be reached.
4. Measure again. Apply the key to the whole text, write the reading file, and count the share of cipher tokens that
   give sense (not an estimate). The reading is "complete" only when (1) at least 95% of its cipher tokens give sense;
   (2) no gap is left untried; (3) every letter in it is covered; (4) what stays open is scattered code groups or names,
   or pieces blocked from outside. Otherwise it stays "partial": say which gaps are still open and why.
5. Be honest. Mark every uncertain word with [?], leave unread groups as their numbers or signs, and never fill a gap
   with a plausible guess. A gap retried and still open, with what was ruled out, is worth reporting.

REPORT BACK
With a clone: open a pull request with your changes to targets/kaa4591/. Say in the description which gaps you worked on,
  the share read before and after, and how you want to be credited (name, initials or handle, and a link).
Without one: open an issue: https://github.com/dbourdeau/cyphersolver/issues/new?template=further-reading.yml&title=Further+reading%3A+kaa4591&page=kaa4591.html
  Give the write-up, the gap you worked on (its line in Remaining gaps, or the code group, folio and line), what you
  propose, the evidence (every occurrence, the images or sources), the share read before and after, your step log,
  and the tools and AI model you used.
If you cannot open issues or pull requests, write this report for your human to send. Credit goes to the person who sends it.