# Are You Tired of the Strokes yet?

You probably are.

Well, in this case, let me point you to a mostly overlooked gem in Voynich research, namely Sarah Goslee’s website. Not only is she a fellow SCAdian*) (Hail from Drachenwald!), but she has also put together a few nice statistical tests on the VM. As always, caveat emptor!, and honestly I haven’t figured out what “principal coordinates ordination on Euclidean distances of row-standardized frequencies” is supposed to be, but I’ve been in the game long enough to be suitably impressed by a procedure with a name of that length.

No, seriously, I’m still struggling to understand what exactly Sarah did and what the results mean, but this has all the appearance of a very interesting and competent piece of research which has up to now not received the mention it deserves, IMHO.

Hence, my usual piece of advice: Check it out, bros!

*) No, it’s not this.

## 21 thoughts on “Are You Tired of the Strokes yet?”

1. tonygdb

Two stroke theory – I suggest the VM was constructed using only the two stokes (‘c’ and ‘i’ without its tittle)
EVA n,r,l,m are all the ‘i’ stroke whilst EVA d,e,o,c,h,g are all ‘c’ stroke and EVA a,y are each where both a ‘c’ and ‘i’ fall next to each other
The rest are merely camouflage to make it resemble an old manuscript –
Constructed in the following manner – (lined up in Courier New font)
| | \ | \\ | \\\ | \\\\
c | a | f | g | h | i |
cc | b | j | l | m | n |
ccc | c | o | p | q | r |
cccc | d | s | t | u | v |
ccccc| e | w | x | y | z |

[IMG]http://i262.photobucket.com/albums/ii112/tony59b/how.jpg[/IMG]

In the first stage in the above image the author writes his message in cipher using only 2 letters the ‘c’ & ‘i’ –
in the second stage above (highlighted in red) the author cunningly disguised it to resemble an old manuscript – the gallows substitute for rubrication (this is why one appears at the start of each paragraph); the EVA q is a modification of the forward slash that appears in many old MS’s (I’ve used it as a word divider in the above but I think it was originally used more in the sense of a comma??); the other ascenders, serifs, ligatures etc. represent different abbreviations –
In the third section I’ve just removed the coloured highlights to show how much it resembles the VM!!!

In quires 1 to 4 – only about 6% of the groups do not conform to one of the above combinations

The following count is based on Takeshi transcript replacing EVA n,r,l,m,I with i; o,h,d,s,g,c, with c; a,y with ci; and all gallows k,t,p,f, with a blank space then counting the resulting combinations (don’t forget to separate the EVA q by putting a space behind it to ‘count whole words only’)-

quire 1 2 3 4 total
c 393 283 427 324 1427
cc 61 40 46 51 198
ccc 130 72 92 132 426
cccc 43 19 34 61 157
ccccc 11 4 11 22 48
ci 251 202 239 132 824
cii 36 18 28 24 106
ciii 11 14 14 5 44
ciiii 46 35 40 32 153
cci 203 155 157 113 628
ccii 68 18 63 33 182
cciii 36 25 19 31 111
cciiii 87 73 84 84 328
ccci 345 264 314 257 1180
cccii 45 10 30 18 103
ccciii 6 9 9 14 38
ccciiii 33 11 23 17 84
cccci 118 53 77 134 382
ccccii 22 1 8 11 42
cccciii 4 1 0 1 6
cccciiii 8 3 7 6 24
ccccci 55 20 30 58 163
cccccii 6 3 1 7 17
ccccciii 5 0 0 0 5
ccccciiii 12 5 8 6 31

This is remarkably consistent throughout the 4 quires – (I have only used the first 4 quires so as not to mix up Currier A & B which may be due to a switch of language)

Turning the above into percentages and putting in order we get –

VM german
21.3 18.5
17.6 11.5
12.3 8
9.3 7
6.3 7
5.7 5
4.9 5
2.9 5
2.7 5
2.4 5
2.3 4
2.2 4
1.6 3.5
1.5 2.5
1.2 2.5
0.7 1.5
0.6 1.5
0.5 1.5
0.5 1
0.3 1
0.2 0.5
0.1
0.1

Where German is the nearest match I’ve got but it’s not that convincing – anyone know of a better match? – and how do modern German frequencies compare with German from 4 centuries ago?? – more counting – Oh the joys of deciphering!

As I see it the main problem is accurately converting it back to the ci combinations –

Are all EVA Y ‘ci’ or are some just ‘c’
EVA ‘a’ and ‘o’ can be confused in some cases (the ‘o’ being formed by 2 downward penstrokes presumably to avoid the nib digging into the paper on an upwards pen stroke and splattering ink everywhere – try it and see!!!)
EVA Sh & ch – converting these in Takeshi always gives ‘…cc…’ but some (those with a long ligature should be ‘..c c…’ i.e. parts of separate combinations
Even some EVA ‘d’ and ‘g’ could be ‘ci’ combinations

Lesser problems – those combinations beginning with ‘i’ (these could simply be from a common word list or numbers)
Direction of writing – there are a few instances where it looks as if the previous line was written before the preceding one – i.e. a particularly clear instance is f.105r where the 4th group from the end of the second line is raised to avoid clashing with the group beneath and in the penultimate line on the same page the 3rd group onwards is raised to avoid the centred groups beneath.

This all resembles a version of a Baconian biliteral cipher – although not published until 1623 Bacon invented it in 1576-9 whilst in Paris – if the VM is by Bacon or a contemporary to whom he may have shown his cipher then the alphabet will be one used around 1600 and will not look exactly like the one given above the main difference being the last 5 letters – probably u doubling for v and w and depending on the authors native language.

All the above tends to indicate it was a fraud perpetrated on the King knowing he would pay a tidy sum for an old MS – which begs the question – what if anything could be the underlying plaintext that wouldn’t lead to a swift beheading if deciphered?

Still some way to go – I haven’t abandoned all hope just yet!

And God help us if somebody was ‘communicating with angels’ when they wrote it!

Seriously I’d be interested to hear any comments on the above.

2. Hi Tony,

It’s an interesting idea, but I’m not so sure the statistics are meaningful. After all, it’s not too surprising that after combining VM characters and then seperating the groups again you come up with a list of sequences of decreasing frequency…

Other questions are, what would the labels be good for? Apparently they would be too short to even hold a single plaintext character.

Furthermore, as Nick Pelling, who has devoted a post on his blog to the topic, pointed out: You’d arrive at an awful lot of ciphertext for very little content.

But I wouldn’t consider these knockout arguments. It’s a track worth pursuing, good hunting!

3. Tony

Hi Elmar,

I too am not at all sure how meaningful the statistics are, but cannot get away from the fact when you remove all the camouflage you are left with a set ‘ci’ combinations.

You must have misunderstood the system by your comment that the labels ‘would be too short to even hold a single plaintext character’ – the labels generally consist of one or two ‘ci’ combinations which would give one or two characters –
‘what would they be good for?’ – making it look like an old herbal etc. – they don’t have to say anything – they’re just part of the whole to make the duped purchaser buy.

Nick is good at writing limericks –

‘an awful lot of ciphertext for very little content’ – Disappointing as it may be to some, there need be no meaningful content at all.

4. proto57

Tony: F. Bacon’s Biliteral Cipher does not rely on finding unique characters for enciphering with, with the rest being “camouflage”. What you call camouflage may just as well use (whatever) distinction the author chose to designate the a’s and b’s. In fact this is far preferable… because they you have more cipher text to use to encipher with, and can have a much longer plain text. In fact that is the beauty of it… you could go as far as to say the “reason” for it.

That being said, I looked very carefully at your sample (http://i262.photobucket.com/albums/ii112/tony59b/how.jpg), and your description, and I don’t think you are using the Biliteral to begin with. I know you say it “resembles” it, but I can’t see how… maybe that is something I am missing.

Of course I agree that the Biliteral cipher could possibly be “the one” used in the VMs, but I don’t feel counts like the ones you have done will find them (even if it was the one being looked for, which in your example, I don’t think it is). Once you have picked a distinction in the Vms to count (meaning what you would assume to be a Biliteral “a”, and what, a “b”), you have to match the frequencies of them to a count of plain text which has been encoded in the biliteral… the number of times a’s and b’s would appear in such a plain text. The trouble is that by the time such a count is made, and the positioning of the less frequent b’s is lost, it would have little meaning. For instance, “A” encoded in the biliteral is aaaab and “D” encoded is aabaa, so they both have a count of 4 a’s and one b. A count would therefore tell you nothing about the relative frequencies of plaintext A’s and D’s. The only way (I think) the biliteral can be found is to count groups of five, and note the distinctions and their positions within each group. But taken apart with individual counts as you did, it dissolves in your hands.

My point is that a character count of a text encoded in biliteral would count very differently than the same text enciphered with a substitution or transpostion… AND… It would not be directly comparable in any way I can think of… and I don’t think, in the ways you have tried, to it’s plain text.

After that, I would suggest the danger of making so many subjective choices, then analyzing those. Once one has decided what we are going to replace with what, and decided on exceptions to those, and then counted based on those choices… we are already so far lost.

That’s not to say that some sort of count is the answer… only that if you allow for literally billions (if you are lucky) or an infinite (if you are less lucky, and guess better) number of possible ways to break down the counts you feed in the hopper… you will only get some vast multiple of that out the other end. No doubt they will align in seemingly meaningful ways… and encourage one to continue, with a tweek here and there.

And lastly, I think that your test has set up a situation, then found the situation you have constructed. An example of this is where you assume you, “cannot get away from the fact when you remove all the camouflage you are left with a set ‘ci’ combinations.” But the problem here is that you first pulled out the ‘ci’ combinations, and assumed the rest was camouflage. Then later, of course, looking at from the other end, if you “remove all the camouflage” you will be left with the ‘ci’s’ again.

I suppose I sound negative, and I’m sorry for that. I don’t usually comment on other’s cipher attempts, being very much an amateur myself. I mean to be constructive… and if you are looking for a biliteral, that you can find a way to pull it out that would work. And if you are looking for the other cipher you describe, that you find an approach to the count that would produce meaningful pointers for you. It’s just that I did not see them in there, and so, wanted to comment. If I got it all wrong (I’m perfectly capable of misunderstanding, as I’ve shown in the past!) then I apologize for the longest and worst post response in my history so far. Elmar, feel free to clip it… Rich.

5. proto57
6. Tony

Rich – It’s only resemblance to Bacon’s biiteral cipher is it uses only two letters – I did not say it was one – I was merely struck by the coincindence of it using only 2 letters like Bacons and the time frame.

“you first pulled out the ‘ci’ combinations, and assumed the rest was camouflage” –
An erroneous assumption – I was not looking for any particular type of cipher in it, merely trying to understand it, knowing that all the gallows symbols are probably the same (as what is high frequency following any particular one is also generally high frequency following the others – also in the circular diagram that gives 4 ‘alphabets’ the EVA P & F occupy the same position, hence, are interchangeable) and that 86% of all paragraphs begin with a gallows symbol – it being extremely unlikely that any letter (or 4 letters) will start 86% of the paragraphs – it must be something else –
Whilst looking at an old manuscript it became apparent to me that the gallows were nothing more than a sort of rubrication (just lacking the colour!) – see image –

[IMG]http://i262.photobucket.com/albums/ii112/tony59b/rubrication2.jpg[/IMG]

‘A’ is from MS Arundell 117 f.109 14th cent. and is fairly typical of that period – the initial letter ‘O’ is enlarged and rubricated – the fifth line is the start of a new paragraph and has the symbol marked ‘F’ before it – as does every paragraph!!

C D E F G are all variations of the same – all have the same highlighting function

Compare ‘A’ with ‘B’ – it is my contention that whoever wrote the VM they only intended to look something like an old MS – hence my use of the word camouflage
‘I’ shows the gallows for comparison (they even look similar)

‘H’ just points out that the rubrication was sometimes written before the letter, sometimes through it in old MS’s

It was by removing all the rubrication and flourishes because I thought/think that is what they are that I came to the ‘ci’ combinations – not the other way round.
The ‘ci’combinations should be just a simple sub – but it doesn’t seem to be – maybe I’m missing something else –

Anyway I thought putting the above out there, may be of some help – bring some people back to reality – I’ve read some real nonsense written about this cipher – I’ve also seen some incredible analysis which hasn’t advanced our understanding of it one bit (but then it never will unless you know what you’re looking at!)

Don’t worry about sounding negative (I’m quite used to that – i.e. ‘Dorabella’)
Tony

7. proto57

Apologies to Elmar for hijacking his post!

Tony: Don’t get me wrong, it’s all very interesting… and I will imagine that the answer will come from some place no one suspects, and certainly from someone trying something new… the latter we know, because the usual methods seem to have always hit dead ends. One thing that always struck me is that it existed before computers… then at the dawn of computers it was looked at by the first computer decipherers… and then while most other ciphers fell before the power of computers,and the dust settled… there was the Voynich, still as impossible as ever. And as other problems are solved by the increasing power of the machines, still… there it is, unsolved. With all the transcriptions, and all the counts and analysis…

And yet, many are resistant to trying new approaches. So don’t get me wrong in that regard… keep doing what you are doing, because “who knows?”, and how can I?

“…it is my contention that whoever wrote the VM they only intended to look something like an old MS”.

And ironically, in that, we completely concur. I find these tidbits scattered throughout the investigation. You do know my theory, correct? Since I totally stole Elmar’s post, I may as well plug my theory. At any rate, the idea that it is a construction, made to mimic the style of many older works… that is core to my theory. So go for it.

“I’ve also seen some incredible analysis which hasn’t advanced our understanding of it one bit…”

Yes I know what you mean… brilliant work I could never equal… analysis to the finest degree… and nothing has turned up… yet. I took a course in behavior modification years ago, as I was a supervisor in a group home for the mentally handicapped. We would never tell someone they were doing something wrong… only say to them, “Try another way”.

I hope I did not break my own admonition, and do encourage you of course. Not that it should matter to you… that’s part two of solving this. I have no doubt whatsoever they will have a very thick skin… that is requirement one.

(OMG Elmar I hope you don’t even hate me more, now for stealing your blogwidth… sorry… I’ll make it up to you, I swear…)

8. Rich, don’t worry, this is a place for discussion, so keep the posts flowing! (If WordPress charges me extra for the bandwidth used, I’ll forward the bill to you, but currently the hit counter shows there is still some leeway. ;-)

Tony — You had a hand in the Dorabella cipher as well? Pray tell!

9. Tony

Since you ask – you’ll find its solution here –

http://www.villageidiotvsworld.com/crypto/

(according to a panel of 3 judges led by Professor Kevin Jones it’s incorrect – I think they may have some musical qualifications – huh – idiots)
You’ve probably seen it under the name ‘Jean Palmer’ which was my pseudonym for ‘The Agony Column Codes & Ciphers’
I’m also the one who solved 6 of Bellaso’s ciphers recently – again on the above website.
I can’t play any musical instrument though.

10. Dorabella — So, essentially, you did a simple substitution and with liberal anagramming, “slang”, allusions, puns and alterations arrived at something which still doesn’t make coherent sense?

Have you ever double-checked your theory by using a different symbol/letter substitution table and see what happens? I’m pretty sure you’d arrive at results which are as comprehensible as the current “solution” is, making your solution somewhat arbitrary.

11. Tony

Infuriating – typical of the negative, ignorant comments it received –
Why do you say ‘liberal anagramming’? – it does not contain a single anagram
‘Backslang’ not ‘slang’ – totally different & in common use at the time
Etc. etc. etc.

12. Well, if this is the typical response, perhaps there is a reason to it… other than people being too ignorant to understand the full geniality of your “solution”.

I losely used the term “anagramming” for your method of quite arbitrarily reversing the flow of reading in the text.

Come on, face it, your “raw” plaintext is Bltaceiarwunisnfnnellhsywyduo, from this point you proceed with “Backslang” etc. to arrive at something only little more comprehensible.

But what makes you sure your “raw” plaintext is correct? It simply reads like a string of pretty much random letters to me, and I’m convinced if you used different substitutions in a second run, rendering a different “plaintext”, you could just as well construct a vaguely readable message out of the second plaintext string — How can you be sure your substitution table is correct then?

And even worse, if the above is possible, how can you be sure that your method “raw plaintext” -> “processed plaintext” (ie “Be Hellcat…”) is correct?

Just because you arrive at *any* solution doesn’t mean it’s the correct solution.

(I also have my doubts that “effin” is a word old enough, and that Elgar would have used it in correspondence, cipher or not.)

Elmar

13. Tony

“I’m convinced if you used different substitutions in a second run, rendering a different “plaintext”, you could just as well construct a vaguely readable message”

I would like to see you back this statement up with some examples.

14. proto57

Hi Tony: I’m afraid I have to agree with Mr.Vogt.

There are far too many subjective variables for this to be a solution. You made a choice of substitutions, a choice of breaks in the resulting string, a choice of rules for the reversal of letters, and a choice of the letters to add to fill it out. Each step means the possible outcomes are infinite… but, then, this still only results in what is a meaningless sentence, with several “non-words”:

“B hellcat ie a war using effin henshells why your antiquarian net dimineuendo am sorry you theo o tis god then me so la deo da aye”

For a solution to be correct, it would have to be repeatable by a second party, and then result in meaning, and also, the same meaning you came to. I think this never would. But if you are certain, then test it. Explain the system to someone, and give them the rules and substitutions, and see if they come back with the same answer you did. If you are correct, they will. Rich.

15. “I would like to see you back this statement up with some examples.”

Technically, you should have done that long ago yourself to test the validity of your hypothesis, and that you apparently didn’t do it casts a doubtful light on your method, if not on your results.

But I’m game, and I’ll make it easy for you. Toss me a string of characters of your choice (be fair, include some of the more frequent ones and a vowel or two), and I’ll synthesize a secret message from Churchill to Hitler from them. Just for you.

16. Tony

Elmar – “apparently didn’t” – did and know just how difficult it is to manipulate a string of 87 characters to say anything at all – let alone something comprehensible and relevant that the recipient would understand – I can easily understand how you think it incomprehensible but then I doubt if you have read Elgars notebooks, letters, etc. and are not familiar with his love of word play – also if your native language is not English it must be very difficult to appreciate.

kinljtvlaqoefmeueetiiemrqrsod
vevtrlalneentsfveoetwdfmraarod
ndtwd’nmwxsdnetwfdmlisdtlsrl

I can’t wait to see what you come up with –
if you find the above to difficult or think it an unusual string – I’ll make it easier for you – pick any normal route through the ‘clockdial’ to produce your own string of letters (the above is one)

Proto – “subjective variables” – I didn’t make those choices – I was led to them by logical processes of deduction – the choice of substitution by the letter frequencies & the ‘clockdial’ – the choice of breaks by the common words – I discovered the backslang because I had already guessed it to be there I did not make it up – the choice of letters to add to fill it out are fairly obvious having got that far.

I agree almost entirely with your last paragraph but believe a second party would come up with exactly the same solution – provided he was familiar with Elgar’s letters, writing, wordplay, the circumstances surrounding the message etc. – but this is not something for me to do – perhaps some cryptography Professor who has a new batch of students might like to test it out someday.

17. proto57

Hi Tony: You say,”Proto – “subjective variables” – I didn’t make those choices – I was led to them by logical processes of deduction”

It honestly surprises me that you can say, after working this out, that you “didn’t make these choices”. Who did? Using “logical processes of deduction” means you made the choices, of course. We all make choices, and this is how we do it. These choices cannot be anything but subjective, although the better ones might be based on better deduction.

“…perhaps some cryptography Professor who has a new batch of students might like to test it out someday.”

You would not need such an expert, I don’t think… you would only have to post your idea on a few forums, and see what even amateurs came up with. This because the cipher was not intended for a cryptography professor to understand… it was a young woman acquaintance of Elgar’s, whom, by all indications, he believed would easily read it.

But I do have another question for you… do you feel the resulting sentence makes sense, and if so, what do you feel it means?

18. Tony

Yes – it makes sense to me!
It is a friendly, teasing note accompanying a letter from Elgars wife to Dora after a short stay with Doras family –
What it says to me in plain English is –

Bella you are a fiery tempered woman – throwing eggshells at us – that is why we left – I wish we could spend more time at the piano together – alas.

(There is evidence for a friendly egg fight in a previous letter.)

Elgar probably scribbled this note down in less than 5 minutes –
I begin to wish he’d never written it at all.

19. proto57

Tony: If, “B hellcat ie a war using effin henshells why your antiquarian net dimineuendo am sorry you theo o tis god then me so la deo da aye”

Is interpreted by you as,”Bella you are a fiery tempered woman – throwing eggshells at us – that is why we left – I wish we could spend more time at the piano together – alas.”

Then I am afraid you have completely lost me. I wish you all the best… but I’m gone. There is no way on this earth you and I will ever understand each other… Rich.