A private email made me aware of three new Vatican challenge ciphertexts posted by George Lasry at MTC3. I spent many days on the Part 4 and Part 5 ciphertexts, but my attempts with methods described in, e.g., "Ciphertext-only Attack on "Vatican Challenge" Ciphers (1625, 1628)" have been unsuccessful.
The challenge ciphertexts are as follows:
The Vatican Challenge - Part 3 ... a short message dated "Brusseles 9 Oct. 1721", consisting of one- to four-digit figures.
The Vatican Challenge - Part 4 ... a letter from the bishop of Senigallia to the Secretariat, consisting of figures. The cleartext seems to include the date "Marzo 1536."
The Vatican Challenge - Part 5 ... The cleartext seems to include the date "Aprile 1542." According to the introduction of the challenge, this letter is in a bundle of "Lett. Orig. e cifre del card. Farnese al nunzio, 4 oct 1539-24 nov. 1548. ff. 7-123."
My article "Variable-Length Symbols in Italian Numerical Ciphers" describes various types of Vatican figure ciphers.
I guess the Part 4 cipher is something like the cipher for Cardinal Savello (1583) in that the same digit may have a value of its own and may also be a part of a two-digit symbol with a different value (e.g., "3"="i", "1"="o", but "31"="f"). See pdf for contexts of some frequent patterns in KWIC.
I guess the Part 5 cipher is something like Cardinal of Carpi's cipher with Granvelle (1548-1549). If so, there will be a difficulty in telling if a two-figure combination has a value of its own or is a part of a three-digit symbol (e.g., "90"="s", "901"="g"). See pdf for contexts of a frequent pattern and "^".
At present, I feel the ciphers are not polyphonic. But a polyphonic cipher like that of the bishop of Nazareth (1585) cannot be totally excluded.
Any ideas?
some thoughts....
ReplyDeleteSuccession of "2" and "1" is characteristic of the Part 4 ciphertext. Especially, "2222" is relatively frequent and even occurs at the beginning of a ciphertext. If "22 22" at the beginning of a ciphertext is parsed as "22 22", it will be "v u", but "2222" is too frequent to accept "22"="u/v". Another possibility is the first "22" is a single letter word ("a", "e", "i", "o" in Italian). Another possibility is "2 22 2", where "2" is a symbol on its own.
There is even "2422222246", where "2" occurs six times in succession. Is this "24 22 22 22 46" or "2 42 22 22 24" or any other?
The Part 4 ciphertext has portions in cleartext here and there. Assuming the transcription is correct, we may find some clue in relatively short sequences sandwiched between cleartexts such as "1216017256212240" or "222122212132130171307424217616411116". The latter is also interesting in a long succession of "1" and "2." The simplest hypothesis "22 21 22 21 21" suggests "22" is a vowel and "21" is a consonant, but there are too many words matching this pattern.
Re "2422222246", what about "24 222 222 46"?
DeleteOne short sequence sandwiched between a word in clear and an end of page is "1362424712160136245561216019." This short sequence has "13624" twice and "12160" twice:
Delete13624 247 12160 13624 556 12160 19
Any clue?
This pattern (13624...) should have appeared in the list of my comment of 23/08/2019, 18:06 below.
DeletePart 4 has double space " " here and there, which reflects wider letter spacing than elsewhere. Such spacing may indicate the break in the sequence of symbols (e.g., "12 34" may actually represent two symbols "12" and "34" rather than consisting of "1 23 4", etc.). For what it's worth, here is my tentative breakdown based on such spacing. Although I tried to keep consistency in splitting frequent sequences, there remain many inconsistencies, and this parsing may be totally wrong.
ReplyDeleteA dotted digit may represent (i) a symbol on its own (either significant or not), (ii) the first (or last or even middle) digit of a three-digit symbol, (iii) a digit preceding a three-digit symbol, (iv) the first (or last) digit of a two-digit symbol used for some special function (e.g., nomenclature).
ReplyDeleteThe Part 5 ciphertext has an instance of two dotted digits in a row. Only the case (i) is consistent with this.
I hear hill climbing has been successful as a tool in codebreaking. The examples I've seen is to find out the assignment of letters to cipher symbols. I wonder whether hill climbing can be used to parse the ciphertext into one-digit and two-digit cipher symbols (e.g., break "123456" to "12 3 45 6", etc.). (As noted in a previous comment, here is my tentative parsing, which has no ground and may be totally wrong.) I think when the ciphertext is correctly parsed, the number of distinct two-digit symbols is relatively small (e.g., only 50 out of all the 100 combinations "01"-"99" occur). So, the number of distinct symbols may be used as a score function.
ReplyDeleteFor a change, I took a look at Part 3. No success yet, but some observations:
ReplyDelete(1) "9" only occurs at the beginning of a four-digit group.
"2" occurs only as "2", "22", and "222".
(2) Some groups are very frequent and may be letters or syllables rather than code words. Most, if not all, of the two-digit groups would be letters.
groups freq.
77 17x
16 15x
14 13x
17 13x
03 10x
04 10x
66 10x
38 9x
61 9x
06 7x
15 6x
05 5x
53 5x
58 5x
(3) Frequent patterns
trigrams
17 74 77 2x
17 66 71 2x
16 05 14 2x
bigrams
77 03 3x *77 is the most frequent; 03 is frequent (6th)
17 38 3x *17 is the third most frequent, 38 is the 8th
17 66 3x *17 is the 3rd; 66 is the 6th (tie)
05 14 3x *05 occurs 5 times, 14 is the third most frequent (13 timex)
03 04 3x *03 and 04 are the 5th most frequent (tie)
14 77 3x *14 is the 3rd, 77 is the 1st.
Assuming the most frequent 77 is not a word break, I tried hypotheses 77 03=ER etc. but without success.
(4) Interesting patterns
Pattern XX:
03 03
*This occurs in a tantalizingly repetitive pattern "04 66 03 17 04 14 17 74 77 03 03 04 66". But I could not find a word matching this. Although it may simply be a spelling or proper name not in my dictionary, it is more likely that this pattern spans more than one word.
The shortened pattern "03 17 04 14 17 74 77 03 03" has too many matching words.
Patterns XYX
38 53 38 (this occurs in a context: "2 67 14 77 03 38 53 38 77 22"; even if one assumes "2" and "22" are word breaks, there are still too many patterns. Of those, "PRESTATE" looks familiar and the assignment 2(-) 67(p) 14(r) 77(e) 03(s) 38(t) 53(a) 38(t) 77(e) 22(-) is generally consistent with letter frequencies.)
73 2 73 (this occurs in a symmetric pattern "77 73 2 73 77", but there are still too many words to fit this. If "2" is a word break, this pattern is not so significant.)
16 58 16
66 47 66
(5) I pursued the PRESTATE hypothesis a bit more. But there is no pattern fitting "03(s) 17 04 14(r) 17 74 77(e) 03(s) 03(s) 04".
The shortened pattern "04 14 17 74 77 03 03 04" gives some words: generalesse, barchessa, perplesse, etc., but they do not seem plausible.
(6) The PRESTATE pattern "2(-) 67(p) 14(r) 77(e) 03(s) 38(t) 53(a) 38(t) 77(e) 22(-)" is followed by "05 04 14(r) 40 18 77(e) 15 35 58 16 38(t) 77(e) 53(a) 9336 9485."
Assuming that "9336" starts a new word, we need to find a word ending in "tea." This does not sound Italian. If "PRESTATE" is incorrect but "22" is indeed a word boundary, we need to find an assignment that is consistent with two word endings "38 53 38 77" and "38 77 53." If "38 53 38 77" is CVCV, "38 77 53" is CVV. Here I came to a deadend.
About Part 4 (and Part 5):
ReplyDeleteThe above-quoted article "Variable-Length Symbols in Italian Numerical Ciphers" points out that variable-length symbols written continously may confuse even the intended recipient who has the key. At the beginning of the article, there is a ciphertext and a key for the reader to try deciphering this kind of cipher.
For the first time, I did try it myself to see what it is like to solve this kind of cipher. If one cannot solve with the key, it would be hopeless to do so without one.
I could read the first eight words, but then came across "3(i) 79(p) 37(h) 55(l) 75(o) 17(b) 13(a)", which I could tell was wrong even with my limited knowledge in Italian. According to the known plaintext, this should be "37(h) 93(u) 75(o) 57(m) 51(i) 71(n) 3(i)." In dealing with Cardinal of Carpi's cipher, I wrote that working with wrong reading frames sooner or later leads to a symbol not used in the cipher, but such an illegal symbol did not occur in this example. This shows that one can never tell whether "3793755751713" should be parsed as "3 79 37 55 75 17 13" or "37 93 75 57 51 71 3" without knowledge of the vocabulary. (With hindsight, when I saw "3(i) 79(p)", I should have tried a different reading frame.)
My little experiment did show that word boundaries indicated by "4" and "6" are very useful in deciphering. I wonder whether the cipher of Part 4 (and Part 5) reserve any digit as word boundaries.
I posted another specimen in "A Simple Specimen of Variable-Length Symbol Italian Cipher with Ambiguous Assignment ".
DeleteI posted another similar cipher (1551) here.
DeleteIf, for example, "4" is reserved as a word boundary, "4x4" means a single-letter word. If such a pattern is too frequent, "4" cannot be reserved as a word boundary.
ReplyDeletePart 4:
ReplyDelete*5-gram etc. twice in succession
42380 42380
222024 222024
*with an interval (gap) of one digit
02422 8 02422
*with an interval (gap) of three digits
47213 6 21 47213
62547 17 0 62547<CLEARTEXT
15761 3 62 15761
7022024 21 0 7022024
*with an interval (gap) of five digits
24487 41725 24487
242562 76280 242562
*with an interval (gap) of six digits
22105 602224 22105
246550242 216512 246550242<CLEARTEXT
*with an interval (gap) of seven digits
47017 14 6 70 21 47017
44252 22 24 6 22 44252
56276.6725072 56276
*with an interval (gap) of eight digits
21617 64024472 21617
72521 10702384 72521
56122 21222242 56122
24172 22024876 24172
*with an interval (gap) of nine digits
27017 220241327 27017
24052 178252174 24052
725244 876270122 725244
872224 136742462 872224
*with an interval (gap) of ten digits
24013 8556701722 24013
22405 0725425217 22405
Part 5:
DeletePart 5
57274.57274.5
503827 503827 503
89082 49 89082
265740 581 265740
45782 80825 45782
80647 0361736 80647
27497 2742829 27497
038275 97564795 038275
27497 0^.5810365 27497
495782 7^,2080457 495782
38035 40^.6032^.70 38035
806.7 2476.8104026 806.7
I posted a seemingly similar but simpler ciphertext (deciphered) at "A Simple Specimen of Variable-Length Symbol Italian Cipher with Ambiguous Reading Frames".
ReplyDeleteThis comment has been removed by the author.
ReplyDeleteThough I'm far from versed at old handwriting, my little experience here allows me to suggest some corrections in the transcription of the cleartext. (I cannot guarantee that these are correct, because I have not seen the manuscript and also because I cannot still make out the Italian context.)
ReplyDeletePart 4
v vs. r
pevsuadeve->persuadere
febravo->febraro
g vs. q
guale -> quale
long s
guelto->questo
propolto->proposto
indultria->industria
coftructo->constructo?
pretelto->pretesto
qcclesialtico->ecclesiastico
Some of "p" by itself may be "x" ("p di Marzo") or "et"
I found a very long recurring sequence in #IMAGENAME:024r.jpg.
ReplyDelete…
02622 012254713 2 13 0 2525 4 24 052 178256270 21 6 1647 14 4 25252 13 624 0
12160 67 1464713 244 2524 280242 174 24 052 17 825627 13 017 25825252
12 16 82 170556 0550 24 12272507252 13 013 4 24 17 6 17 0 17224 512 246 258
240 224 615674 17 6 162 24 052225252 2226 12160 67 1464713 24 4
25242 8024 2 174 24 052 17825627 13 017 25825252 176 4712160 13 4
….