Philip III to the Count of Añover, Lerma, 20 May 1610

01 The letter

The Biblioteca Histórica Municipal de Madrid holds a series of royal orders in cipher from Philip III to Rodrigo Niño Laso, Count of Añover, of the Council of War, of 1610 and 1611. Eight of them have the decipherment of the time bound in on a slip. Four leaves have none, and I 242,29 is one of them: that is why the series entered this project’s catalogue as entry 1988. The digitised item has five pages; the cipher is on the second, f. 53r, and closes with the date in clear, “de Lerma a 20 de Mayo de 1610”. The reverse carries an archival note about Angola.

The page shows no image of the leaf. The images are on Memoria de Madrid, whose viewer answers scripted requests with an “Access Denied” page; the project did not go round that check, and did not check the terms on which the images may be reproduced. Its only sight of the manuscript is the second page, opened once in a browser when the catalogue entry was made on 2 November 2026: no copy was kept, and the opening groups noted then agree with the first line of the transcription used here. The transcription itself is the contributor’s, made from the images with AI assistance.

02 The key

The cipher mixes numbers (10 to 114, and 241 and 242) with small letter-syllable signs, written here as ba, cle, je, fal, fol, xol and the like; these spellings are typographic approximations of the signs, not their values. A group may stand for a letter (25 = e, 28 = n, 32 = r), a syllable (fal = la, cla = ta, 54 = de, 73 = re) or a word (242 = tregua, 102 = avisareis, cel = effecto), and a value may have several groups (10 and 20 for a). Groups run on without word division.

Alberto Garvía García rebuilt the table from four letters of the same series that carry both the cipher and its clear text: I 242,42 (San Lorenzo, 21 February 1610), I 242,47 and I 242,48 (both of 20 May 1610) and I 242,33 (Lerma, 19 June 1610). He matched cipher sequences with the clear words and kept a reference for each value. A fifth letter, I 242,71 (Madrid, 20 February 1611), was kept apart as a negative control. The clear texts of those letters are decipherments of the time; nothing in them is claimed as new.

The table for this letter has 74 groups. Sixty-six are matched with a clear text. Eight have no sure value: the sign called U01 (read buena) and the numbers 101 (aviso), 114 (bien) and 241 (trato), which have no parallel and take their value from the context; e = mu, also from the context; fo = so, a revisable transfer (grade M; the contributor’s table gives no direct parallel); and two looped signs written 28~ and 58~, taken as nulls. The key is therefore partial. The whole table is in the repository (key.tsv).

03 The checks

The project re-derived the reading from the contributor’s transcription and table with a parser of its own, which tries every way of cutting the digit runs: it gives his ten lines, except at two signs written differently in the transcription than in the table (gel{b?} = mando, bum = pues), which it cannot parse as written. All 211 groups of his own division have a value in the table and decode to his reading letter for letter. This shows that the table reads the transcription; it does not show that the transcription is right.

The transcription has outside support. Brice Dudout had transcribed the same leaf independently, also with AI agents (on 3 November 2026, by his record), and his digit strings agree with the contributor’s on every one of the ten lines. They differ on one sign of line 5, bo (which gives por) against 60; the contributor looked again at the images and keeps bo, and Dudout marks his own reading doubtful. The other differences are spellings of the same signs.

Dudout also sent his alignment of a further glossed letter of the family, I 243,18 (Lerma, 30 April 1610): 114 pairs of cipher group and gloss. Of the pairs whose group is in the table, 71 agree with it and one does not (a group glossed or once and er beside it, which looks like a slip of the alignment); 24 groups of that letter are not in the table. None of the open groups occurs in it, so it supports the table and closes nothing. The alignment was not compared with the image here. Dudout reports as well that a test he had declared for a table of the whole family, on I 242,42 held out, failed his own bar (0.640 against a bar of 0.85); that test is of his table, not of the one used here, and I 242,42 is one of the letters this key was built from.

The contributor’s control windows on the sibling letters (27 of 34 earlier ones agree, and 4 of 4 added later) are checks of a key against the passages it was built from, not blind tests. The project’s statistical check scores the reading at −1.948 per letter in a Spanish 4-gram model, against −3.627 on average for 200 random keys of the same structure (z 7.7) and −2.176 at best for a key fitted to the shuffled ciphertext (z 10.4): verdict pass, on 205 groups, with a key that came from the sibling letters and not from this text.

No earlier reading of the letter was found. The contributor searched S. Tomokiyo’s pages, DECODE, Retortillo Atienza (2019) and the Correia thesis, and a web search here found nothing. He opened Monumenta Missionária Africana, first series, vol. V, p. 587: it prints a different dispatch on Pinda, of 31 May 1610, not this letter. Devos, Les chiffres de Philippe II (1950), was identified and not seen by anyone.

04 The reading

De [buena] parte se ha tenido [aviso] que olandeses han ydo a asentar fatoria al [puerto] de Pinda en la costa de Angola y que allí y en [otras] partes van [entablando] su [trato] contra lo [capitulado] en la tregua, y [porque] no es [bien] pasar esto en [disimulación], os encargo y [mando] procuréis que se remedie por los medios que os pareciere, sin que vaya [persona] a este effecto a las Islas, pues no convendría que fuese por la reputación, y avisaréis me de lo que se hiciere.

Square brackets mark the words that are not sure. In English: “From a good quarter word has been had that Dutchmen have gone to set up a factory at the port of Pinda on the coast of Angola, and that there and in other parts they are entering upon their trade contrary to what was agreed in the truce; and because it is not right to let this pass in dissimulation, I charge and command you to see that it is remedied by whatever means seem fitting to you, without a person going to the Islands for this purpose, since it would not be fitting that he should go, for reputation’s sake; and you shall inform me of what is done.”

Tregua has its value from the sibling I 242,48; that it is the Twelve Years’ Truce is the contributor’s interpretation, and the text names no article. Islas is kept as deciphered: which islands are meant is not in the text. Fatoria is the spelling the groups give.

Counted by the README rule, strictly: the ten lines hold 89 words, and 77 are sure, which is 86.5%, under the 95% bar, so the extent is partial. Ten of the twelve holes are bracketed in the contributor’s own reading (the eight groups without a sure value, and otras, where a sign gives “otraas”); the other two, mando and porque, are words he reads as firm that stand on a sign in question. His own count is 78 of 88 words (88.6%). Read here as sentences against the translation, every clause of the order holds (grade A).

05 What remains uncertain

• The transcription has not been compared with the manuscript by the project: the Memoria de Madrid viewer refuses scripted requests, and it needs a person at a browser. The contributor’s zip attachments on the issue could not be downloaded (403) and were not seen: the six of 12 November (his dossier) and the two of 14 November (a publication package and a complement of controls). The page rests on his Zenodo report and his issue comments only.
• buena (U01), aviso (101), trato (241), bien (114): codes with a value from the context only (grade I). disimulación: the sign e = mu, from the context (I). persona: fo = so, a revisable transfer with no direct parallel in the contributor’s table (M).
• puerto, entablando, capitulado: each has a looped sign (28~ or 58~) inside it, taken as a null without a parallel (M). otras: the sign m = as gives “otraas”, not explained.
• mando (line 6): the sign is printed with a query in the transcription; its value comes from a sibling (C on the value, the sign doubtful). porque (line 5): bo or 60, as above; por fits the sense.
• An open question on the last sign of line 4, which gives the tu of capitulado: the contributor writes it clu, Brice Dudout writes it du, and in I 243,18 the sign Dudout writes du is glossed conde twice. Either these are two signs in two hands, or the value in line 4 needs a second look on the images. It has been asked of both and is not answered.
• The open groups can still be worked: further letters of the series (I 242,66 with a gloss; I 242,35, 59 and 60 without) have not been used: no transcription of them has reached the project (the contributor reports work on 35, 59 and 60 in his dossier, not seen here, which settles no value of this letter), and a glossed one may carry them. The three other leaves without a decipherment stay in the catalogue as entry 1988; whether I 242,35, 59 and 60 are those leaves is not mapped.
• No human palaeographer has checked the signs, and the sibling clear texts behind the key values were not seen by the project.

Full texts

The transcription and the decryption of every document on this page, so that the reading can be used and checked. Short files are shown here; all are in the repository.

Philip III to the Count of Añover, 20 May 1610 (I 242,29): the ten transcribed lines read in part

Philip III learns that Dutchmen have gone to set up a factory at Pinda on the coast of Angola, against the truce, and charges the Count of Añover to remedy it without sending anyone to the Islands, and to report.

Builds coherent sentences, but under the read bar of 95% of words read as sense (86% here); the holes are marked in the reading.

# Philip III to the Count of Añover, Lerma, 20 May 1610: Biblioteca Histórica Municipal de Madrid, I 242,29 (Memoria de Madrid 424541), f. 53r, the ten lines of the ciphered body.
# Copied from section 4 of Alberto Garvía García's report (Zenodo 10.5281/zenodo.24174363, CC BY 4.0) on 13 Nov 2026. Not checked against the manuscript images (Memoria de Madrid refused the project's requests, 403).
# Notation as in the report: ? = doubt; G01 = an unidentified sign (read U01 in the key table); ^{+} = a cross after the number; ^{bucle?} = a possible loop (null?); the Greek delta, R, B and ji are typographic approximations of signs; spaces are regularised and are not cryptographic boundaries. The contributor transcribed the five pages in full (his dossier, not seen here); the report prints these ten lines.
53r.L01 5411{G01?} ba32 cle je2110 cle6857101 δ.55 fal2854 je332110283757
53r.L02 10^{+}je28cla32 1810clo7410^{+}26 bu2528^{bucle?}32 clo54294028532528 fal
53r.L03 5133cla541028 xol fal37 δ1026^{+}40372528553472 m ba32 cle33
53r.L04 7010282528 cla14 fal2858^{bucle?}57 ſu24151283472 fol47 bi clu
53r.L05 fal58^{bucle?}572528 fal24237bo32 δ69 n114 baſa32 nclo252856
53r.L06 ji e fal495528p25284732 xol37gel{b?}29765273O R je73
53r.L07 B5625bo32 fol33 B56p δp ba7349mel25 ji28 R7010
53r.L08 3720 be32fo6610n cle cel10fal33 O fal33bum695128702528
53r.L09 137410R 1870n25bo32 fal73 bucla49552837102 B54 fol
53r.L10 δ je24403840mel25,

# Mechanical decode of transcription.txt with key.tsv by an own parser (all segmentations of the digit runs by backtracking; the one with the fewest unparsed signs and then the fewest tokens is kept). Values as in key.tsv; '?' marks a sign not in the table (the 'gel' of line 6 and the 'bum' of line 8 are written differently in the transcription than in the table).
53r.L01  54 11 ba 32 cle je 21 10 cle 68 57 101 δ 55 fal 28 54 je 33 21 10 28 37 57
         debuena?partesehatenidoavisoqueolandeseshanydo
53r.L02  10+ je 28 cla 32 18 10 clo 74 10+ 26 bu 25 28~ 32 clo 54 29 40 28 53 25 28 fal
         aasentarfatoriaalpuertodepindaenla
53r.L03  51 33 cla 54 10 28 xol fal 37 δ 10 26+ 40 37 25 28 55 34 72 m ba 32 cle 33
         costadeangolayquealliyenotraaspartes
53r.L04  70 10 28 25 28 cla 14 fal 28 58~ 57 ſu 241 51 28 34 72 fol 47 bi clu
         uanentablandosutratocontralocapitu
53r.L05  fal 58~ 57 25 28 fal 242 37 bo 32 δ 69 n 114 baſa 32 nclo 25 28 56
         ladoenlatreguayporquenoesbienpasarestoendi
53r.L06  ji e fal 49 55 28 p 25 28 47 32 xol 37 ?g e ?l 29 76 52 73 O R je 73
         simulacionosencargoy[g]mu[l]procureisquesere
53r.L07  B 56 25 bo 32 fol 33 B 56 p δ p ba 73 49 mel 25 ji 28 R 70 10
         medieporlosmediosqueosparecieresinqueua
53r.L08  37 20 be 32 fo 66 10 n cle cel 10 fal 33 O fal 33 bu m 69 51 28 70 25 28
         yapersonaaesteeffectoalasislaspuasnoconuen
53r.L09  13 74 10 R 18 70 n 25 bo 32 fal 73 bu cla 49 55 28 37 102 B 54 fol
         driaquefueseporlareputacionyavisareismedelo
53r.L10  δ je 24 40 38 40 mel 25
         quesehiciere

# Philip III to the Count of Añover, Lerma, 20 May 1610 (Biblioteca Histórica Madrid I 242,29): the reading, line by line

Alberto Garvía García's cautious reading (Zenodo 10.5281/zenodo.24174363, section 5), checked here only against his own transcription and key table (`transcription.txt`, `key.tsv`, `decryption.txt`). Square brackets mark a value that is uncertain or read from context only; they are part of the result. Not checked against the manuscript.

| line | reading |
|---|---|
| 53r.L01 | De [U01: ¿buena?] parte se ha tenido [101: ¿aviso?] que olandeses han ydo |
| 53r.L02 | a asentar fatoria al [puerto?] de Pinda en la |
| 53r.L03 | costa de Angola y que allí y en [otraas: ¿otras?] partes |
| 53r.L04 | van [entablando?] su [241: ¿trato?] contra lo capitu- |
| 53r.L05 | la[58~: ¿nula?]do en la tregua y porque no es [114: ¿bien?] pasar esto en di- |
| 53r.L06 | si[e: ¿mu?]lación os encargo y mando procuréis que se re- |
| 53r.L07 | medie por los medios que os pareciere sin que va- |
| 53r.L08 | ya [persona?] a este effecto a las Islas pues no conven- |
| 53r.L09 | dría que fuese por la reputación y avisaréis me de lo |
| 53r.L10 | que se hiciere. |

Normalised: "De buena parte se ha tenido aviso que olandeses han ydo a asentar fatoria al puerto de Pinda en la costa de Angola y que allí y en otras partes van entablando su trato contra lo capitulado en la tregua, y porque no es bien pasar esto en disimulación, os encargo y mando procuréis que se remedie por los medios que os pareciere, sin que vaya persona a este effecto a las Islas, pues no convendría que fuese por la reputación, y avisaréis me de lo que se hiciere."

## Translation

From a good quarter word has been had that Dutchmen have gone to set up a factory at the port of Pinda on the coast of Angola, and that there and in other parts they are entering upon their trade contrary to what was agreed in the truce; and because it is not right to let this pass in dissimulation, I charge and command you to see that it is remedied by whatever means seem fitting to you, without a person going to the Islands for this purpose, since it would not be fitting that he should go, for reputation's sake; and you shall inform me of what is done.

## Notes

- "Fatoria" is kept as deciphered (not modernised to "factoría"); "tregua" is read from the sibling I 242,48 (242 = tregua); that it is the Twelve Years' Truce is the contributor's contextual interpretation, the body names no article. "Islas" is kept as deciphered; reading it as the Dutch provinces is an interpretation. The group "je" ends "olandeses"; it does not license adding "se" before "han ydo".
- Open signs, by the contributor's own table: U01 (buena), 101 and 114 and 241 (contextual codes, no direct parallel), e = mu, fo = so (persona), the nulls 28~ and 58~, the anomaly m = as giving "otraas".
- The contributor's figures, as his (not the README read bar): 173 of 211 cipher-group occurrences (81.99%) belong to firmly readable words; 202 of 211 (95.73%) have an equivalence at table level; 78 of 88 words firm (88.6%); 82 of 88 if the grade-M values are accepted.
- The share recorded here is the project's strict recount of 14 Nov 2026 by the README rule: 89 words, 77 sure (86.5%). The holes are the ten bracketed words above and two words on a sign the transcriptions do not settle (*mando*, the sign printed gel{b?}; *porque* in line 5, bo or 60). It is under the 95% bar, and the transcription is unchecked against the images, so this is read in part (NOTES.md, Strict count).

# English translation: Philip III to the Count of Añover, 20 May 1610

Partial English translation of `reading.md`. Each row corresponds to the same numbered line in the Spanish reading. Square brackets preserve unresolved cipher groups, uncertain readings and editorial doubts. English phrases may cross line boundaries where the Spanish original splits a word.

| Line | English translation |
|---|---|
| 53r.L01 | From [U01: a good?] quarter word [101: report?] has been had that the Dutch have gone |
| 53r.L02 | to establish a trading post at the [port?] of Pinda on the |
| 53r.L03 | coast of Angola, and that there and in [otraas: other?] places |
| 53r.L04 | they are [establishing?] their [241: trade?] contrary to what was stipu- |
| 53r.L05 | la[58~: null?]ted in the truce; and because it is not [114: right?] to let this pass in dis- |
| 53r.L06 | simulation [e: mu?], I charge and command you to see that it is re- |
| 53r.L07 | medied by whatever means seem appropriate to you, without any- |
| 53r.L08 | one [person?] going to the Islands for this purpose, since it would not be advis- |
| 53r.L09 | able for that person to go, for reasons of reputation; and you shall inform me of |
| 53r.L10 | what is done. |

## Editorial notes

- The translation is provisional and follows the line numbering of `reading.md`.
- Square brackets preserve unresolved readings and are not evidence that the proposed words have been deciphered.
- The uncertain sign `58~` and the proposed value `e = mu` remain unresolved. Their positions within the Spanish words are documented in `reading.md`.
- The identification of the truce as the Twelve Years' Truce is historical interpretation, not an explicit statement in the deciphered text.
- "Islands" translates *Islas* without assuming a specific geographical identification.
- No independent palaeographic verification is claimed.

Key: key.tsv

06 Sources

• Madrid, Biblioteca Histórica Municipal (Ayuntamiento de Madrid), I 242,29, f. 53r: Memoria de Madrid 424541; Hispana record 58678726. Sibling letters I 242,33, 42, 47, 48 and 71, and I 243,18 (Memoria de Madrid 425791), in the same library.
• Alberto Garvía García, report with the transcription, the key table and the reading, Zenodo, doi 10.5281/zenodo.24174363 (CC BY 4.0; second version 13 November 2026), and issue #499; English translation in pull request #547.
• Brice Dudout, comments on issue #499 of 13 and 14 November 2026: the transcription comparison and the alignment of I 243,18.
• Brásio, Monumenta Missionária Africana, first series, vol. V (1955), p. 587, opened by the contributor. Devos, Les chiffres de Philippe II (1950), not seen.

The transcription, the key table, the decode, the reading, the translation and the contributors’ files are in targets/anover1610.