At a glanceExisting decipherment · partial · 2 of 3 telegrams (27 of 74 code items)
- The telegrams
- King Carlos I of Portugal to Queen Amélie, spring 1903: Lisbon to Paris (Hotel Bristol), 12 May; Lisbon to Tlemcen for the yacht Amélie at Oran, 4 March; and one undated. French clear text with five-figure code groups.
- Where
- Arquivo Nacional Torre do Tombo, Documentos de D. Carlos, D. Amélia e D. Manuel II, maço 12 (PT/TT/CAM/M12), images m0054–m0060 of 61. Catalogue entry 401.
- The code
- A code of five-figure groups, the same in all three telegrams (they share groups). It is not alphabetical: 38564 = ter stands below 58215 = com.
- How it was deciphered
- From the decodes of the time. A Hotel Bristol card (m0060) repeats the clear words of the 12 May telegram and puts a Portuguese sentence where its 14 groups stand; the undated telegram has its decode in ink on the form.
- What it says
- All are well at home; the Queen “may give” something, “but only after having arranged it with Rosa” and after the French Government has been spoken to; he has written to her at Malta.
- Still open
- The Tlemcen telegram of 4 March (47 items): no decode and no codebook. Which group carries which word in seven groups of the May telegram and in the undated one.
- Credit
- Transcriptions, dates and directions, the link between the May telegram and the card m0060, and the controls by WAINUT (27 October 2026), done with AI language-model agents; the transcriptions were made with a vision-capable language model and checked against the DigitArq images. The decodes are those made on receipt in 1903.
01 The telegrams
Maço 12 of the papers of King Carlos, Queen Amélie and Manuel II holds 61 images. Images m0001–m0053 are the King’s letters to “Ma chère Amélie”, in French and in clear, of 1904–1905. Only three telegrams are in code. The DigitArq record calls the unit not accessible; its images download all the same through DigitArq’s own JSON service.
| image | route and date | items | decode of the time |
|---|---|---|---|
| m0054 (cover m0055) | Lisbon → Paris, Hotel Bristol, 12 May 1903 | 14 groups among clear French words | the card m0060 |
| m0056 (back m0057) | Lisbon → Tlemcen, 4 March 1903, to the Queen “bord yacht Amélie, Oran” | 47 (44 groups and three “03”) | none found |
| m0058 (back m0059) | undated (spring 1903?) | 13 (12 groups and “03”) | in ink on the same form |
The May telegram is on printed strips. Its header, SS PARIS LISBOA 3428 33 12 12/40- HS, follows the rule printed on the form: the first number after the place of origin is the serial number and the second the taxed words. So it left Lisbon as no. 3428, 33 words, on the 12th at 12:40, and the Paris stamp reads 12 MAI 03. The first pass had it as Paris to Lisbon, 1903/04. The March telegram is written by hand on a French form: “Pour Tlemcen de Lisbonne No 1728 Mots 86 Dépôt le 4/3 1903 à 8h30”, stamped TLEMCEN 5 3 03 ORAN. Its sender is not named; it is filed with the King’s papers. The undated telegram is written on a blank form with no office data.
02 Lisbon to Paris, 12 May
The French words are in clear. The 14 groups carry a Portuguese sentence, which the card m0060, printed “Hotel Bristol, Paris”, writes out in its place. The card repeats the telegram’s clear words in order and puts the Portuguese where the groups stand, so it is the decode of this telegram, made on receipt.
Merci dépêche. Ici tout et tous très bien. Podes dar, mas só depois de ter combinado com Rosa e ella ter fallado com Governo francez. Petits baisent main. T’embrasse. Carlos.
“Thanks for your telegram. Here everything and everyone very well. You may give [it], but only after having arranged it with Rosa and after she has spoken with the French Government. The little ones kiss your hand. I embrace you. Carlos.” What is to be given is not said, and “ella” has no clear antecedent. Rosa is probably Tomás de Sousa Rosa, then Portuguese minister in Paris (an identification made here, not checked). Edward VII had visited Paris on 1–4 May 1903; whether the telegram has to do with that is not known.
The two repeated groups fix part of the alignment. 38564 (groups 5 and 11) and 58215 (groups 7 and 13) stand two places apart, as “ter … com” does twice in the decode (“ter combinado com”, “ter fallado com”). So 38564 = ter, 58215 = com, and the groups between them are 29955 = combinado and 49452 = fallado. The last group, 82042, is left for “Governo francez”. The 19 Portuguese words stand on 14 groups, so some groups carry more than one word, and groups 1–4 and 8–10 cannot be split word by word. The decoder’s hyphens suggest that the code has syllables as well as words.
03 The undated telegram
Ici tous et tout bien. T’ai écrit à Malte. T’embrasse, et les petits salue tous. Carlos.
“Here everyone and everything well. [I] have written to you at Malta. [I] embrace you, and the little ones greet everyone. Carlos.” “salue” is singular as written; “[I] embrace you and the little ones; greet everyone [from me]” is also possible. Malta fits the Queen’s Mediterranean cruise of spring 1903. The 12 groups and “03” do not map one to one onto the 15 words, so which group carries which word is not known.
04 The Tlemcen telegram
The telegram of 4 March 1903 reached the Queen on board the yacht Amélie at Oran: 44 five-figure groups and three “03”, 86 taxed words. No decode was found in the maço.
17582 92270 19652 10734 92270 43554 76138 78341 76342? 93434 94057 33725 52162 55820 03 68564 90387 59338? 91518 38965 03 57174 43554 78341 85328 57944 14492 95268 58215 46864 56994 26722 85387 39965 03 42618 23014 82927 81037 73374 17582 84437 32937 56484 57218 93057 61742
It shares 58215 and 91518 with the May telegram, and 10734, 43554 and 78341 with the undated one. Of these only 58215 (com) has a value. The fourth group of line 3 was taken as 19338 in one transcription and 59338 in the other.
05 One code, not a cipher
Two controls (groups.py). Random telegrams of 12 and 44 five-figure groups share two or more groups in 0 of 20,000 trials, so the shared groups show that the three telegrams use one code. In the undated telegram the decode has 60 letters and the 12 groups have 60 figures; taken letter for figure, the mapping has 37 conflicts, against 38.1 on average (minimum 32) for 1,000 shuffled orders of the figures. That is no better than chance: the 60 = 60 is a coincidence and the system is a code, not a letter cipher.
The code is not alphabetical (com has the higher number), unlike the code of Sabugosa’s telegrams of the same weeks (targets/sabugosa1910), so the four values from the May telegram predict nothing about the rest of the book.
06 What remains uncertain
- The Tlemcen telegram, all 47 items: no decode in the maço and no codebook. Only 58215 (com) has a value, and alone it gives nothing.
- May telegram, groups 1–4 (28548 40710 86415 23274 = “Po-des dar mas só depois de”) and 8–10 (18111 91518 34678 = “Ros-a e ella”): the words are given by the decode (grade H), but not which group carries which.
- May telegram: 38564 ter, 58215 com, 29955 combinado and 49452 fallado are fixed by the repeats; 82042 “Governo francez” is what is left over (grade I).
- The undated telegram: the words are given by the decode (grade H); the split over the 12 groups and “03” is not known, and the date is not on the form.
- Rosa as Tomás de Sousa Rosa, and the object of “Podes dar”, are not checked.
- Other units of the same collection and the Arquivo da Casa Real were not searched; they may hold a decode of the March telegram or the codebook.
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.
m0054, Lisbon -> Paris, 12 May 1903
King Carlos thanks the Queen, then in Paris, for her telegram and says all is well; in code he tells her she may give [something] but only after arranging it with Rosa (probably the Portuguese minister in Paris) and after the French government has been spoken to. The children kiss her hand.
Transcription
# ANTT PT/TT/CAM/M12, image m0054 (cover m0055). Telegram Lisbon -> Paris, 12 May 1903, King Carlos to Queen Amelie. # Printed strips pasted on a French receiving form. Header strip: SS PARIS LISBOA 3428 33 12 12/40- HS # (form's own rule: the first number after the place of origin is the serial number, the second the taxed words: # destination PARIS, origin LISBOA, no. 3428, 33 words, day 12, 12:40). Stamp: CENTRAL PARIS 12 MAI 03. # Address strip on the cover (m0055, read on a reduced image): REINE PORTUGAL HOTEL BRISTOL PARIS. # The message strips as printed. Clear words in capitals; the 14 five-digit groups are the code. # A receiver marked one digit in most groups with a short stroke above it (not reproduced). = MERCI DEPECHE ICI TOUT ET TOUS TRES BIEN 28548 40710 86415 23274 38564 29955 58215 18111 91518 34678 38564 49452 58215 82042. PETITS BAISENT MAIN . T EMBRASSE = CARLOS +
Decipherment of the time
# ANTT PT/TT/CAM/M12, image m0060: a card printed "Hotel Bristol / Paris", written in ink on the receiving side. # It repeats the clear words of telegram m0054 and puts a Portuguese sentence where the 14 code groups stand, # so it is the decode of m0054 made at the time. Line breaks as on the card. The hyphens in "Po-des" and "Ros-a" # are the decoder's. Merci dépêche. Ici tout et tous très bien. Po-des dar mas só depois de ter combinado com Ros-a e ella ter fallado com Governo francez. Petits baisent main T'embrasse. Carlos.
Decryption
# m0054 against the decode of the time on m0060. Clear words of the telegram as they stand; code groups with
# the words the decode puts in their place. The decode has 19 Portuguese words (with the decoder's hyphens) for
# 14 groups, so some groups stand for more than one word.
# Fixed by the two repeats: 38564 (groups 5 and 11) and 58215 (groups 7 and 13) stand two places apart, as
# "ter ... com" does twice in the decode ("ter combinado com", "ter fallado com"). So 38564 = ter, 58215 = com,
# and 29955 = combinado, 49452 = fallado. The last group, 82042, is left for "Governo francez" [inferred].
# Groups 1-4 and 8-10 cannot be split word by word.
MERCI DEPECHE ICI TOUT ET TOUS TRES BIEN (clear)
28548 40710 86415 23274 = Po-des dar mas só depois de [split unknown]
38564 = ter
29955 = combinado
58215 = com
18111 91518 34678 = Ros-a e ella [split unknown]
38564 = ter
49452 = fallado
58215 = com
82042 = Governo francez [inferred]
PETITS BAISENT MAIN . T EMBRASSE = CARLOS (clear)Reading
# m0054: King Carlos to Queen Amelie, Lisbon -> Paris (Hotel Bristol), 12 May 1903 Merci dépêche. Ici tout et tous très bien. Podes dar, mas só depois de ter combinado com Rosa e ella ter fallado com Governo francez. Petits baisent main. T'embrasse. Carlos.
Translation
# m0054: English translation Thanks for your telegram. Here everything and everyone very well. You may give [it], but only after having arranged it with Rosa and after she has spoken with the French Government. The little ones kiss your hand. I embrace you. Carlos. Notes. The telegram is in French with one coded sentence in Portuguese. The coded sentence is taken from the decode of the time on the card m0060; the French words are in clear on the telegram. "Podes dar" has no object in the decode; what is to be given is not said. "ella" (she) has no clear antecedent in the telegram. "Rosa" is probably Tomás de Sousa Rosa, then Portuguese minister in Paris [our identification, not checked].
m0058, undated (spring 1903?)
King Carlos tells the Queen that all are well, that he has written to her at Malta, and sends his love and the children's greetings. Fits her Mediterranean cruise of spring 1903.
Transcription
# ANTT PT/TT/CAM/M12, image m0058 (back m0059). A blank French telegram form (Telegramme, No 700); no office # data filled in. The code groups by hand; the writer makes 5 like a long s; a receiver underlined single digits # in most groups (not reproduced). The decode written below them is in m0058_contemporary.txt. Undated. 03 68692 39210 43554 93001 43294 10734 16425 88018 78341 88717 34057 55058.
Decipherment of the time
# ANTT PT/TT/CAM/M12, image m0058: the decode written in ink on the same form, under the code groups.
Ici tous et tout bien t'ai écrit à Malte t'embrasse
et les petits salue tous
Carlos.Decryption
# m0058 line by line against the decode of the time. 12 groups plus "03" for 15 French words and the signature; # the groups do not map one to one onto the words, so the split between the lines follows the layout on the form. 03 68692 39210 43554 93001 43294 10734 16425 = Ici tous et tout bien t'ai écrit à Malte t'embrasse 88018 78341 88717 34057 55058. = et les petits salue tous / Carlos
Reading
# m0058: King Carlos to Queen Amelie, undated (spring 1903?) From the decode of the time written on the form (French). Ici tous et tout bien. T'ai écrit à Malte. T'embrasse, et les petits salue tous. Carlos.
Translation
# m0058: English translation Here everyone and everything well. [I] have written to you at Malta. [I] embrace you, and the little ones greet everyone. Carlos. Note. "salue" is singular as written. The line can also be read "[I] embrace you and the little ones; greet everyone [from me]". The mention of Malta fits the Queen's Mediterranean cruise of spring 1903 (m0056 reaches her on board the yacht Amelie at Oran on 5 March 1903).
m0056, Lisbon -> Tlemcen, 4 Mar 1903 decoded, not read
Code telegram of 86 taxed words to Queen Amélie on board the yacht Amélie at Oran; no decode found.
A mechanical decode with holes in most sentences: shown so that the transcription and the key can be checked and improved, not as a reading.
Transcription
# ANTT PT/TT/CAM/M12, image m0056 (back m0057). French receiving form (Telegramme, No 700), by hand. # Pour Tlemcen de Lisbonne No 1728 Mots 86 Depot le 4/3 1903 a 8h30 m. # Address: Reine Amelie Portugal bord yacht Amelie Oran Tlemcen. Stamp: TLEMCEN 5 3 03 ORAN. # Sender not named on the form (filed with King Carlos's papers). The writer makes 5 like a long s. # A receiver underlined one digit in most groups (not reproduced). A struck-out figure (036?) at the start of # line 3 is left out. A trailing ? marks a group with a doubtful digit. # 47 items: 44 five-digit groups and three "03". 17582 92270 19652 10734 92270 43554 76138 78341 76342? 93434 94057 33725 52162 55820 03 68564 90387 59338? 91518 38965 03 57174 43554 78341 85328 57944 14492 95268 58215 46864 56994 26722 85387 39965 03 42618 23014 82927 81037 73374 17582 84437 32937 56484 57218 93057 61742
Decryption
# m0056: no decode of the time was found, and the code is not rebuilt. Not read. # Only one group has a value from the m0054 decode: 58215 = com [inferred there from a repeat]. # Groups shared with the other two telegrams (same code): 10734, 43554, 78341 (also in m0058); # 58215, 91518 (also in m0054). [..] = group without a value. [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] 03 [..] [..] [..] [..] [..] 03 [..] [..] [..] [..] [..] [..] [..] [com?] [..] [..] [..] [..] [..] 03 [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..] [..]
07 Sources
- Arquivo Nacional Torre do Tombo, Documentos de D. Carlos, D. Amélia e D. Manuel II, maço 12, PT/TT/CAM/M12, DigitArq record, 61 images: m0054 and its cover m0055 (12 May), m0056 and back m0057 (4 March), m0058 and back m0059 (undated), m0060 (the Hotel Bristol card).
- No printed decipherment of these telegrams was found (one web search). Related only: A. H. Norris, “A letter from Queen Amelia to her son King Manuel II”, BHS Portugal Report 4 (1977), p. 56, on a letter in which the Queen put certain words in cipher; a different document.
Files: targets/carlos1903/ (notes, transcriptions of the three telegrams, the decodes of the time, readings and translations, groups.py with its output). Contributed by WAINUT; reviewed for the site on 27 October 2026.