Im sure you guys have thought of this so, im just looking to get a feel for what you think of it but, what about a double play fair with a route? Would this type of cipher conform to the realizations made by Dan Olson and company?
Also, is it required that the second keyword be different?
Thanks.
Soze
I see the name Robert "Bobby" Prince in the word scramble. They show a listing for a older man by that name living in Sacramento, and he previously resided in Lower Presidio Heights, in San Francisco.
It is just a plaintext I used for one of my ciphers, I don’t wish to connect this person to the Zodiac case at all, he’s one of my hero’s. 
No. I don’t place any value on his claims. I had a brief correspondence with him to try to get more info. He claims his decipherment implicates Arthur Leigh Allen, but of course won’t offer any details or evidence of any kind.
Thanks for the update. Judging from his e-mail reply I have to agree.
Changing topics: How far into the period 15/19 scheme have you explored with regard to column ordering resulting from columnar transposition induced by a keyword? Maybe you are close but just haven’t found the right permutation of columns.
Columnar transposition with a keyword should not keep us from getting higher scores unless the 340 has a very low scoring plaintext or a huge amount of columns were used. Also smokie made me a test cipher were he randomly distributed his message accros 55 period 19 lines (45 degrees diagonally). By randomly rearranging these lines I was able to score smokie’s cipher up to 21700 and the 340 was stuck at 21000 (including all orientations).
We are currently working on this hypothesis:
1. The inscription could exist in any set of 2 dimensions.
2. The transcription could exist in any set of 2 dimensions.
3. A contiguous inscription direction, normal, flipped, mirrored or reversed.
4. A contiguous transcription direction, any valid columnar or diagonal transposition (24 total).
5. Inscription nulls.
6. Transcription nulls.
7. A secondary transposition with a contiguous direction in any set of 2 dimensions, normal, flipped, mirrored or reversed.
Also, as you guys are exploring different transposition schemes, which of them would be compatible with the resulting pair of pivots we see in the 340? It seems to me that whatever scheme is in play would tend to produce those pivots, because they seem so improbable. What kind of schemes would create period 15/19 ngram peaks AND those pivots?
Good question. I have to admit that I haven’t done much work in the direction of the pivots other than thinking about it. It seems hard to connect them to the periods. I could speculate that they appear around a local misalignment zone (period shift). Smokie has done work to show what happens to a period 19 cipher with various types of misalignments going on, you get these horizontal regions that shift the period by 1 in that zone. That would at best classify the pivots as an artifact of such zones interacting, sketchy.
Smokie’s misalignment zone example cipher:
Such a scheme fits into the hypothesis we are currently working on and is very hard to crack.
Recently I thought that perhaps vigenere was applied to the encoding (1 to 63) instead of the plaintext (1 to 26) but attempts to recreate such a thing totally randomized the encoding and no cycle was left standing.
The repeats show up primarily as diagonal rows going in one direction, and I suspect a knight’s tour scheme would make it difficult to make so many period 19 repeats because the repeats would be diffused in other directions.
Good thinking. That is my conclusion also. Sometimes the Zync gets at me.
It looks like a 16 x 21 grid, inscribed vertically and then transcribed right to left horizontally. Or a 21 x 16 grid inscribed horizontally and transposed vertically. Somewhere inscription or transcription is backwards. I am going to keel working on my spreadsheets, which are pretty crude. But even with a far from perfect untransposition, I was able to get a decent score and find a lot of words.
Yes, you solved it rather decently. Let me know when you want to know what exactly I did to create these ciphers.
etom hall compile ds
a room hd robert bobb
y prince hate waater
the score for boy sos
pooom and roos ii pri
nc cts his s never ani
n house crom these r f
orth e prove chimtoe
s being isolated few
it hout office allow
ede develop mplore t
hescor caused his ti
n put from th inspira
tion ent teas anroos
bible the do find muc
h cument written fro
m the dall what help e
rte sign do t with the
sound in by tosh was t
he doom
etom hall compile ds
a room hd robert bobb
y prince hate waater
the score for boy sos
pooom and roos ii pri
nc cts his s never ani
n house crom these r f
orth e prove chimtoe
s being isolated few
it hout office allow
ede develop mplore t
hescor caused his ti
n put from th inspira
tion ent teas anroos
bible the do find muc
h cument written fro
m the dall what help e
rte sign do t with the
sound in by tosh was t
he doom
To reiterate, this is not a solution to the 340. It’s a solution to a cipher of mine, with a plaintext that I put together from the internet.
Yes that’s one that you made for Smokie? . I like to break them up to see if I can work out how to unjumble them
e tom hall compile dsa room hd robert bobby prince hate waater the score for boy sos
pooom and roos ii princ cts his s never anin house crom these r forth e prove chimtoe
s being isolated few it hout office allow ede develop mplore thesc or caused his ti
n put from th inspiration ent teas anroos bible the do find much cument written fro
m the dall what help erte sign do t with the sound in by tosh was the doom
I had the privilege of meeting the gifted programmer and codebreaker George Lasry at the crypto symposium. He recently co-authored a paper about a hillclimber that breaks columnar transposition, even with long keys:
The classical columnar transposition cipher was the most popular type of transposition cipher. It was in use mainly during the second half of the nineteenth century and the first half of the twentieth century. It also served as a building block for more complex ciphers, such as the ADFGVX cipher and the double transposition cipher. Pen-and-paper as well as computerized methods for the cryptanalysis of the columnar transposition cipher have been published, but those apply mainly to the easier cases of short keys and complete transposition rectangles. In this article, a novel approach for the cryptanalysis of the columnar transposition cipher (when used with long keys) is presented. It is based on a two-phase hill climbing algorithm, a two-dimensional fitness score, and special transformations on key segments. This ciphertext-only method allows for the recovery of transposition keys with up to 1,000 elements, and up to 120 elements for worst case transposition rectangles.
Here is the paper: https://drive.google.com/file/d/0Bx2ZxP … pkNzg/view
Note that the last row of the rectangle is incomplete, and therefore the first 3 columns of the transposition rectangle, before transposition, are longer (by one row) than the other 4 columns. This case is referred as an incomplete transposition rectangle or Irregular Columnar Transposition (ICT).
The case where all columns are of the same length, and all rows are complete, is referred to as Complete Columnar Transposition (CCT).
Historically, several manual methods were developed to cryptanalyze columnar transposition ciphers, including for the CCT and the ICT cases. The best known example is the “strips method” described by Friedman, which can be applied to CCT cases with short keys. At first, the cryptanalyst arranges the ciphertext in columns, on paper. He then cuts the text into strips, each column into one strip. Next, he manually tries to match the strips against each other using those arrangements which create the most probable bigrams, or pairs of successive letters.
Thanks doranchak, interesting stuff!
Without keyword my solver can solve both ICT and CCT perfectly. But as he shows, and something I had not thought of is that with the use of a keyword, the irregular columns change positions. This could complicate solving the cipher immensly, I see a few options here. If we assume the 340 has 23 or less elements then the lengths of the elements could easily be brute forced or hill climbed. Or it can be tried to come up with some sort of cheat.
Anyone would like to make me a keyed ICT cipher with no more than 23 (period 15 to 19) elements, 340 chars and 63 symbols?
I think that Lasry may have been talking about this, which is on the book pages 21 and 22 of the NSA Military Cryptanalysis book:
The solution is on page 24:
It looks very similar to smokie9, but with smokie9 the columns are not transcribed according to a key. It seems to me that you would have to transcribe contiguous rows to create the period 15/19 repeats ( I think it is period 15 now by the way ). I can easily make a message like this if you want, but I don’t think that there will be very many period 15/19 repeats.
Let me know what you did regarding those two route cipher messages.
I have been doing a lot of thinking in the last 24 hours and will discuss soon.
It looks very similar to smokie9, but with smokie9 the columns are not transcribed according to a key. It seems to me that you would have to transcribe contiguous rows to create the period 15/19 repeats ( I think it is period 15 now by the way ). I can easily make a message like this if you want, but I don’t think that there will be very many period 15/19 repeats.
Cool, the hat diagram. Not sure how it would work though. I quickly tested it (keyed) with a few samples and there allot of repeats. I’m going to make a test cipher for myself. Period 15 is stronger yes.
Let me know what you did regarding those two route cipher messages.
Okay,
pr1_jarlve1: 22 by 16, corner 2, transposition.
pr1_jarlve2: 19 by 18, corner 4, untransposition.
Corners:
1 2 3 4
I think that basically they write the message on strips of paper and slide them up and down until they find frequent digraphs and trigraphs. But they also don’t have homophonic substitution.
Here is such a message if you want. The key at the top is made of four words. I transcribed the plaintext columns in alphabetical order. There are two A’s in the key, thus I transcribed the column under the leftmost A first and the rightmost A second. The next lowest letter in the key is an E, and there are a few of those. I started with the leftmost E and transcribed the E columns in same fashion. It is a traditional keyed columnar transposition. Then I encoded the transposed plaintext. If you want the plaintext version, I can give that to you also.
29 54 51 30 16 23 48 31 52 60 11 42 24 53 25 10 12
35 26 43 55 51 35 1 2 32 18 26 23 11 17 43 37 38
52 33 24 25 14 54 29 53 36 45 28 49 39 15 37 21 41
11 3 22 42 40 12 4 26 30 15 38 55 31 14 15 5 61
19 27 59 1 56 11 19 32 12 50 37 46 51 37 41 13 14
52 47 33 48 30 7 23 49 15 38 24 45 25 50 35 1 53
48 39 51 8 11 3 46 39 12 57 54 43 49 17 10 6 49
42 26 53 13 4 50 14 59 5 3 59 49 62 42 35 38 1
7 9 3 19 47 53 18 17 51 15 45 32 32 59 33 6 16
39 37 11 13 41 11 27 56 23 52 53 25 44 26 8 38 25
52 40 58 14 15 36 11 40 13 46 40 22 20 11 4 30 1
42 14 40 55 19 11 52 16 39 21 27 37 42 50 22 40 31
23 24 29 7 38 26 20 2 40 26 34 19 39 19 3 16 26
47 2 35 53 27 48 24 25 41 49 1 50 21 43 51 54 47
26 32 27 28 61 57 18 48 11 33 52 42 31 22 53 8 20
29 5 51 12 10 9 12 34 31 26 34 18 46 21 24 34 36
47 14 45 52 53 6 15 32 31 62 30 7 22 11 55 38 30
25 32 50 51 59 2 26 27 20 5 33 23 12 21 3 24 50
19 51 56 30 13 53 25 46 10 38 14 15 31 39 3 34 11
6 12 8 1 55 50 13 61 40 55 58 26 32 33 51 47 62
Thanks, what are the dimensions you performed the columnar transpositon in?
Sorry, I should have included that.
It is 17 columns. Most of the columns have 19 rows, but a few of the columns have only 18 rows. Very similar to the picture from the book. The last row plus a few symbols are actually gibberish, and you can take those out. The message is actually 319 plaintext long, so don’t include the last 21 symbols in your analysis.
Aha!
Was not expecting that.
I made a 15 by 23 keyed columnar transposition (so 23 elements) and the Period operation of my manipulation solver made quick work of it, scoring 23000+.
