The idea is that these could be created by the following cipher:

1. Transpose the plaintext
2. Encode the plaintext, top and bottom areas with one Caesar shift, and the middle with another Caesar shift, symbols look just like ordinary letters
3. Encode again with homophonic symbols

Hey smokie,

An interesting hypothesis for the observation! I wonder if it could coexist with transposition. In other words, would it diffuse the bigrams or not?

So alberti can seen as a caeser shift. I think this could also be regarded as vigenère by rows instead of columns. AZdecrypt’s substitution + vigenère solver is flexible enough to solve it since you can submit word lists and the words can be as long as the cipher.

Since AZdecrypt does not care about a caeser shift perhaps we could get away by only caeser shifting the middle part? I would also consider a 7, 7, 6 row division.

Looking forward to your cryptograms.

https://itunes.apple.com/de/app/ifontmaker/id377381670?mt=8
The only disadvantage: Now my handwriting is public

Thanks! Here’s what your Z408 font looks like on the web with Z408:

http://zodiackillerciphers.com/z408-font/z408.html

O.k., so here is the basic idea. The top 6 and bottom 6 rows are 60% of the message and the middle 8 rows are 40% of the message. I calculated the frequency of all plaintext in Brave New World. At top is the frequencies * 60% and bottom 40%. If I Caesar shift the middle different than the top and bottom, I have to slide the frequency chart, and the result with be new frequencies totaled up as Caesar symbols. There are 25 different combinations of shifts, making 25 different possible Caesar symbol frequency charts.

The first question is, does any particular shift difference result in more or less mutually exclusive Caesar symbols in the top / bottom or middle like we see in the 340?

The answer is yes because some plaintext only appear in the top / bottom, and some plaintext only appear in the EDIT: middle. If I use a Caesar shift, these will become new Caesar symbols, but only appearing in the top / bottom.

With my new spreadsheet that has the entire book in it, I transposed the plaintext with a 17 x 20 inscription rectangle LRTB and a 17 x 20 transcription rectangle TBLR for period 20. Every chunk of 340 for the book with this matrix.

1 18 35 52 69 86 103 120 137 154 171 188 205 222 239 256 273
290 307 324 2 19 36 53 70 87 104 121 138 155 172 189 206 223
240 257 274 291 308 325 3 20 37 54 71 88 105 122 139 156 173
190 207 224 241 258 275 292 309 326 4 21 38 55 72 89 106 123
140 157 174 191 208 225 242 259 276 293 310 327 5 22 39 56 73
90 107 124 141 158 175 192 209 226 243 260 277 294 311 328 6 23
40 57 74 91 108 125 142 159 176 193 210 227 244 261 278 295 312
329 7 24 41 58 75 92 109 126 143 160 177 194 211 228 245 262
279 296 313 330 8 25 42 59 76 93 110 127 144 161 178 195 212
229 246 263 280 297 314 331 9 26 43 60 77 94 111 128 145 162
179 196 213 230 247 264 281 298 315 332 10 27 44 61 78 95 112
129 146 163 180 197 214 231 248 265 282 299 316 333 11 28 45 62
79 96 113 130 147 164 181 198 215 232 249 266 283 300 317 334 12
29 46 63 80 97 114 131 148 165 182 199 216 233 250 267 284 301
318 335 13 30 47 64 81 98 115 132 149 166 183 200 217 234 251
268 285 302 319 336 14 31 48 65 82 99 116 133 150 167 184 201
218 235 252 269 286 303 320 337 15 32 49 66 83 100 117 134 151
168 185 202 219 236 253 270 287 304 321 338 16 33 50 67 84 101
118 135 152 169 186 203 220 237 254 271 288 305 322 339 17 34 51
68 85 102 119 136 153 170 187 204 221 238 255 272 289 306 323 340

Some plaintext only appear in the blue areas, and some only in the red areas, resulting in Caesar symbols that only appear in the blue or red areas.

There are a lot of exclusive Caesar symbols in every message, but I wanted to isolate certain shifts that result in only a few Caesar symbols exclusive to the red. I set an artificial range of 2 to 6 symbol positions ( the 340 has 4 ). There are 860 messages in the book 340 long * 25 possible shift differentials = 21,500 messages. Of the 21,500, there were 759 that had only 2 to 6 Caesar symbols exclusive to the red area.

X axis is the shift differential, y axis is the count of messages that have 2 to 6 Caesar symbols that only appear in the red area. Shift differentials of 7 and 19 result in a lot more than the others.

But, basically, there are 759 / 21,500 = 3.5% of the possible message and shift combinations that can result in what we see in the 340 middle 8 rows. Not much, really, but I will continue.

EDIT: If a Caesar symbol only appears in the red area, then after homophonic encoding, the homophone will also only appear in the red area.

Here is the shift differential of 7. The top chart is the 60% top / bottom frequency, the bottom is the middle 40% frequency. The lowest chart shows the totals. With shift differential of 7, low frequency plaintext pairs like C-V, J-C and Q-J line up vertically.

If there is a C in the red area, but not in the blue areas, then it will become a Caesar symbol V in the red area but there will not be any Vs in the blue areas. If there is a J in the red area but not in the blue areas, then it will become a Caesar symbol C in the red area but not be any Cs in the blue areas. Etc.

Since high frequency plaintext like E appear in both areas, that will appear as an E in the blue area and an X in the red area. But there will be a lot of Es and Xs, unlike the middle area of the 340. Hopefully that makes sense. Basically, low frequency plaintext that appear only in the middle or red area of the message show up with a Caesar shift of 7 in the middle or red area of the message.

That is why there is a spike at shift differential 7.

Here is the shift differential of 19. C-J line up, J-Q line up, and Q-X line up. Other low frequency plaintext line up also like V-C, etc.

So that is how only four symbol positions could appear exclusive to the middle or red area of the 340.

Here is the table that I use. For example when I say shift of 7, the top / bottom could be encoded with the blue, or 5th row in the table. And the middle would then be encoded with the red, or 12th row in the table.

Using a shift normally makes a lot of Caesar symbols that are exclusive to the top / bottom and middle. Here is the data for all 860 chunks and 25 shifts. X axis is the shift, Y axis is the average count of Caesar symbol positions that appear in the middle, but not the top or bottom.

When I make my message I will show them in Caesar before homophonic encoding. Here is the top / bottom. The 340 has 39 positions exclusive to the top / bottom, which is very normal for this data. For example with a shift of 7 ( x axis ), the top / bottom will have over 30 total positions occupied by symbols that do not appear in the middle. That’s what made me think of the Caesar idea.

In other words, the symbols that only appear in the top and bottom 6 rows of the 340 could easily be created by some type of regional Caesar encoding, but the four symbols that only appear in the middle maybe only about 3% to 4% chance. And some shifts make the chances higher than others.

EDIT: When I make my message, I will start with a message that, with whatever shift, has about 39 positions in the top / bottom and about 4 positions in the middle.

To the question of whether transposition can co-exist with regional Caesar encoding. It can, but obviously the Caesar encoding will diminish the repeats.

Remember I transposed every chunk of the book with the matrix a few posts above to make a spike at P 20. X axis is the shift, Y axis is the count of repeats. With a shift of 0 ( no Caesar encoding ), the average is 236 repeats. That will easily create a spike at P 20 after homophonic diffusion. With other shifts, the average is about 194 repeats, or a reduction of about 18%. That would be more difficult to make a P 20 spike after homophonic diffusion, but there will be some messages that have starting repeats with a higher number.

I think that 18% reduction seems low, but there will be high frequency bigrams in the top / bottom, that after regional Caesar encoding, will have the same Caesar symbols as other high frequency bigrams in the middle. So there is some offset. In other words, maybe a bigram like TH in the top / bottom will show up as XL after Caesar encoding, and MA in in the middle will also show up as XL after shifted Caesar encoding in the middle. You will then have two XLs, causing a repeat even though they wouldn’t be repeats before Caesar encoding.

Slicing the bigrams between rows 6 and 7 and rows 14 and 15 will also cause a reduction. But there will also be some random offset.

To make what we see in the 340, you could start with a 19 x 18 grid, and use regional Caesar encoding, then redraft the message into 17 x 20 and encode homophonic.

Another weekend is shot and it is almost certainly not regional Caesar encoding. But I have a new spreadsheet that I can put books into, break into chunks of 340, and transpose and encode every chunk with just one click.

Here is smokie 64

The plaintext, from Brave New World, which, by the way, I had to cherry pick to get the P 20 repeats.

H T T H E L I F T M A N L O O K E
D A F T E R T H E M R O O F H E S
A I D O N C E M O R E Q U E S T I
O N I N G L Y T H E N A B E L L R
A N G A N D F R O M T H E C E I L
I N G O F T H E L I F T A L O U D
S P E A K E R B E G A N V E R Y S
O F T L Y A N D Y E T V E R Y I M
P E R I O U S L Y T O I S S U E I
T S C O M M A N D S G O D O W N I
T S A I D G O D O W N F L O O R E
I G H T E E N G O D O W N G O D O
W N F L O O R E I G H T E E N G O
D O W N G O T H E L I F T M A N S
L A M M E D T H E G A T E S T O U
C H E D A B U T T O N A N D I N S
T A N T L Y D R O P P E D B A C K
I N T O T H E D R O N I N G T W I
L I G H T O F T H E W E L L T H E
T W I L I G H T O F H I S O W N H

the matrix

1 18 35 52 69 86 103 120 137 154 171 188 205 222 239 256 273
290 307 324 2 19 36 53 70 87 104 121 138 155 172 189 206 223
240 257 274 291 308 325 3 20 37 54 71 88 105 122 139 156 173
190 207 224 241 258 275 292 309 326 4 21 38 55 72 89 106 123
140 157 174 191 208 225 242 259 276 293 310 327 5 22 39 56 73
90 107 124 141 158 175 192 209 226 243 260 277 294 311 328 6 23
40 57 74 91 108 125 142 159 176 193 210 227 244 261 278 295 312
329 7 24 41 58 75 92 109 126 143 160 177 194 211 228 245 262
279 296 313 330 8 25 42 59 76 93 110 127 144 161 178 195 212
229 246 263 280 297 314 331 9 26 43 60 77 94 111 128 145 162
179 196 213 230 247 264 281 298 315 332 10 27 44 61 78 95 112
129 146 163 180 197 214 231 248 265 282 299 316 333 11 28 45 62
79 96 113 130 147 164 181 198 215 232 249 266 283 300 317 334 12
29 46 63 80 97 114 131 148 165 182 199 216 233 250 267 284 301
318 335 13 30 47 64 81 98 115 132 149 166 183 200 217 234 251
268 285 302 319 336 14 31 48 65 82 99 116 133 150 167 184 201
218 235 252 269 286 303 320 337 15 32 49 66 83 100 117 134 151
168 185 202 219 236 253 270 287 304 321 338 16 33 50 67 84 101
118 135 152 169 186 203 220 237 254 271 288 305 322 339 17 34 51
68 85 102 119 136 153 170 187 204 221 238 255 272 289 306 323 340

The Caesar encoding. This is where I created the symbols that only appear in blue or red areas, like with the 340. Upper left is the transposed plaintext.

See letter C, heavy box, encodes to L. But to get an L symbol in the blue areas, you would have to have a letter J in the blue areas, and there aren’t any.
See letter S, heavy box, encodes to B. But to get a B symbol in the blue areas, you would have to have a letter Z in the blue areas, and there aren’t any.
See letter Q, heavy box, encodes to Z. But to get a Z in blue, you need an X in blue and there aren’t any.

That’s how Caesar encoding can be similar to the 340 top / bottom and middle symbols.

The key.

A 1 2 3
B 4
C 5 6 7 41
D 8
E 9
F 10 11 12
G 13 14 15
H 16 17 18
I 19 20 21
J 22 23 24
K 25 26 27
L 28
M 29 30 31
N 32 33 34 41
O 35 36 37
P 38 39 40
Q 41 42 43
R 44 45 46
S
T 47 48 49
U 50 51 52
V 53 54 55
W 56 57 58
X 59 60 61
Y 62 63 64
Z 65

After homophonic encoding. There are 65 symbols and you should like the Raw IOC. But there are five symbols where there are only one of that symbol.

22 10 5 43 5 25 50 42 45 54 54 26 62 11 32 9 55
27 32 53 54 7 25 38 39 40 45 16 15 52 52 19 38 41
7 24 5 39 27 63 55 17 11 27 20 21 15 53 47 9 7
24 18 62 35 15 40 54 19 26 22 55 41 38 7 42 41 34
26 43 27 53 41 39 36 10 54 42 23 32 13 14 40 20 38
16 29 1 42 37 11 15 43 21 13 5 33 55 53 25 34 48
28 50 30 6 41 22 8 54 39 32 59 60 31 26 17 41 61
40 46 7 33 18 35 42 2 56 4 22 59 57 1 41 5 8
31 34 36 41 37 42 53 6 1 41 27 30 51 58 31 38 32
42 41 7 1 31 41 5 6 33 59 42 61 52 34 18 17 30
59 60 44 41 32 7 61 3 42 61 53 54 1 33 55 45 39
34 7 4 11 30 40 50 38 59 62 61 41 35 23 1 32 56
5 36 23 6 59 39 57 60 43 45 22 56 63 56 10 41 57
61 65 23 41 7 58 9 44 59 37 11 41 35 5 24 33 45
14 25 34 43 56 8 15 6 60 13 51 12 41 38 14 54 14
38 10 39 32 52 41 18 13 14 9 33 15 49 51 42 43 19
13 36 51 11 8 20 34 41 42 22 51 41 14 42 47 3 57
64 41 42 40 6 54 26 41 55 53 62 29 15 54 32 27 58
1 25 13 38 48 12 21 39 42 40 9 63 23 38 14 52 27
49 33 10 51 36 27 25 15 41 42 52 56 50 30 26 15 22

It has a spike at P 20, and here are the repeats. Most of the repeats are within their regions, but there were a few that existed in different regions. They just happened to be bigrams that Caesar encoded to the same symbols.

And here is the epic fail. The homophonic encoding with cycles, at about 25% random symbol selection, actually created a lot more symbols that were unique to the top / bottom and middle. The Caesar symbols that were unique to the top / bottom or middle are guaranteed to be represented here, but because of the homophones, there were a lot more.

For example, you could have the Caesar symbol A with homophones 1,2,3,4. Caesar symbol A could appear in all regions of the message, but after encoding symbols 1 and 2 may appear only in the blue areas, and symbols 3 and 4 only appear in the red area.

The 340 has 39 in top / bottom and only 4 in the middle. This one has 43 in the top / bottom, but 25 in the middle. I could not recreate the 340 with the middle region.

So that effort did fail. But I always wondered if he could have used multiple keys, or cycled his homophonic key in different region, keeping all of the same symbols. It was tough to make this message and get the P 20 spike, so transposition and regional Caesar encoding don’t co-exist well together. The regional biases of those symbols in the top and bottom rows, is very strange, but it seems stranger to have only four symbols unique to the middle 8 rows.

Maybe he just used a different homophonic key in the top / bottom 6 rows, but he didn’t Caesar shift.

And here is the epic fail. The homophonic encoding with cycles, at about 25% random symbol selection, actually created a lot more symbols that were unique to the top / bottom and middle. The Caesar symbols that were unique to the top / bottom or middle are guaranteed to be represented here, but because of the homophones, there were a lot more.

Hey smokie,

Are you saying that a caesar shift by thirds generally overshoots the observation in the 340? What about just homophonic substitution?

The by thirds pattern (top/middle/bottom) in the 340 is a tricky interpretation I think. It is a distribution and other distributions could be equally significant. Unless, you consider the observation that it was derived from, the high unigram distance, in what seems to result in unusually large gaps between symbols. That said, I would not look at it as a distribution like thirds or 6-8-6, but rather as the distance between symbols. Your cipher has a unigram distance of 13227 while the 340 sits at 15034.

And here is the epic fail. The homophonic encoding with cycles, at about 25% random symbol selection, actually created a lot more symbols that were unique to the top / bottom and middle. The Caesar symbols that were unique to the top / bottom or middle are guaranteed to be represented here, but because of the homophones, there were a lot more.

Hey smokie,

Are you saying that a caesar shift by thirds generally overshoots the observation in the 340? What about just homophonic substitution?

The by thirds pattern (top/middle/bottom) in the 340 is a tricky interpretation I think. It is a distribution and other distributions could be equally significant. Unless, you consider the observation that it was derived from, the high unigram distance, in what seems to result in unusually large gaps between symbols. That said, I would not look at it as a distribution like thirds or 6-8-6, but rather as the distance between symbols. Your cipher has a unigram distance of 13227 while the 340 sits at 15034.

I just picked 6-8-6 because that is what it looks like, but if I had been correct with my idea, then it wouldn’t necessarily have to be exactly 6-8-6. It was difficult to make transposition + Caesar + homophonic all co-exist. It was not too difficult to make it so that a lot of symbols only appeared in the 6-6, while only a few symbols appeared in the 8 with transposition + Caesar. But I was not able to make that happen with transposition + Caesar + homophonic.

I think that shifting the homophonic would be exactly the same as Caesar shift + homophonic. The Caesar shift basically shifts the homophonic key. Same thing I think. But one important takeaway for me is that he probably did not change his key throughout the message. It was difficult for me to do it and make the P 20 repeats.

Largo, I adapted your wonderful fonts and combined them into a single font, and made each symbol a constant width, making it easier to display cipher texts in a grid format. I also drew in the missing symbols from Z13 and Z32, and moved the symbols around to match the webtoy transcription.

Here is a sample of what the ciphers look like in this font:

Click here for a live version: http://zodiackillerciphers.com/combined … phers.html
And download the TTF font file here: http://zodiackillerciphers.com/combined … ebfont.ttf

Damn good show old chap(s).

This is so awesome. Thank you so much!

For some reason I can’t get to the link that shows Moonrock’s different cycle types.

So what about this kind of cycle to explain things. Maybe he just used a pattern and if we can find it…

A B C D E

B C D E

C D E

D E

E

A B C D E

B C D E

C D E

D E

E

That would make some symbols get closer and closer together, but also some symbols farther and farther apart.

Or like this.

A

A B

A B C

A B C D

A B C D E

A

A B

A B C

A B C D

A B C D E

That would make symbols get farther and farther apart.