There has been a bit of talk, again, about the possibility of Francis Bacon’s Biliteral Cipher being used in the Voynich. It is one of my personal top-three candidates, and has been… and it’s been a favorite of mine for some reasons unique to my New Atlantis Theory. In fact those things which are considered detriments… the dating it would imply, and the origins and influence… I consider additional assets.
Bacon devised his cipher sometime in the late sixteenth century, but it first appeared in print in the early seventeenth. It is basically a binary alphabet, using only two characters. He designated these “a” and “b”… which sometimes causes confusion as to the structure and use of this cipher. I’ve seen people actually use those letters, or look for them, or alternately use the Biliteral, while not using it when an a or b appears in the plain text… but the cipher character or character distinction in its place. But the system is the simplest of all, and these errors muddle this simplicity, and cause further misunderstandings and confusion.
The way the code works is like this: Each letter of the plain text alphabet is represented by a series of five characters. These characters are designated as either “a” or “b” characters. The first letter of the alphabet, A, is written like this: aaaaa. The plaintext letter B is written like this: aaaab. The plaintext C is written like this: aaaba. You see it is exactly the progression found in the binary number system. D is written like aaabb, and E like aabaa, and so on. Here is the list as published:
The message: “I am here” would be written:
abaaa aaaaa ababb aabbb aabaa baaaa aabaa
Now it is immediately apparent that simply sending a message in this way is not the point… it is instantly recognizable. This is where the beauty of the cipher comes in: The encipherer can chose any way imaginable to represent the a’s and b’s of the cipher text. Bacon used, as an example, two different typefaces. And this is another point where many make an error in understanding the biliteral: It is assumed that it relies on variations in typefaces, fonts or strokes. But this is not the point of the cipher… one can use any variation in the cover text to designate the a’s and b’s of the cipher text. In fact the cover text does not even have to be characters… they could be illustrations, or color variations. Here is my plain text, “I am here”, written as stars:
The stars have two distinctions: five point stars as the “a’s”, and six point stars as “b’s”. But there is a further issue which often crops up, which causes misunderstanding of the biliteral. The fact is that two distinctions are used in encoding, such as my star points, or standard and italic fonts, or serifs and sans-serif typefaces… but somehow when some are looking for evidence of the biliteral, they sometimes look for two frequent characters which re-occur. In the case of the Voynich for instance, one may look for “o’s” and “gallows”, assume these to be the a’s and b’s, and pull them out. But this misses the value of the biliteral… it can be some aspect of the characters which is in all the characters. I have been most interested in the differentiation in height, for instance. This is a large distinction in the Voynich, easily determined, and applies to all of the characters. What the character is would not matter, only that it is high or low. I like the idea, because it would be very easy to encipher and decipher… a decipherer could quickly run along the text of the Voynich and jot down the value for highs and lows… I’ve done it, and it is fast.
And this is one of the key reasons I like the biliteral in the Voynich, it’s ease. In addition to this, however, is the fact that it would answer many questions about the bulk of the text, with its seeming complexity. Some have suspected that the Voynich contains gibberish, that it’s cipher is meaningless. And this could be correct about the bulk of the text content, in the case of the biliteral… because there is one more error made in thinking of this code: That the cover text needs to be meaningful, in order to “throw off” anyone from suspecting that there is a code somewhere. Well yes, this is a tactic in some biliteral messages… Bacon used it as an example, and prisoner’s today still use the biliteral to transmit messages to one another. If a guard sees a message which says, “Did you get the cigarettes I paid for last week? Better hurry or I will take back my teddy bear”, he will not mind all that much, and may not notice that there is a code hidden in the two slightly different fonts used.
But as I demonstrated in my stars example, no meaning is needed in the cover text. And the same applies when it “almost just about looks like there must be some meaning there”… as in Voynichese. As I wrote on Nick Pelling’s blog, the seeming complexity of a fake, meaningless cover cipher of Voynichese, “…could be “there to confuse us”… useless complexity would be a fantastic, brilliantly diabolical cover for a simple cipher, and really very easy to do (not to see, however).”
So I feel the use of the biliteral would explain so much about our views and examination of the text of the Voynich: The inability to make any sense of it, for one thing, because there would not necessarily be any sense in the cover cipher; the concurrent feeling that there must be some sense to the Voynich Ms., causing an unfortunate rift between the proponents of sense and those thinking “nonsense”… both could be correct. And for those who rightly note that the biliteral would mean that there is only one-fifth of the characters in the plain text (there always is, with a five place biliteral), I would, and have, pointed out that we cannot assume the amount of information in the Voynich. It may well be one fifth of what is seen on the surface… for that matter, it could be a two hundred character sentence, with one page per plain text letter! To make assumptions about the level of content would be foundationless, and yet, the biliteral would allow quite a bit to remain.
I worked many hours on the cipher and the Voynich, and came up with some interesting observations. I cannot “read” any of it, so I cannot place myself in that rare, hallowed circle who think they can. But nonetheless I found some results somewhat promising… promising enough to continue when I have the time and patience. To begin with I used four different variables, and reduced the Voynichese four times using them, for comparison. The two main variables were: 1) All high characters were my a’s, and all lows my b’s, and 2) Reversed: All high characters were my b’s, and all lows my a’s. The second set of variables were: 3) All connected characters were considered separate, i.e., a string of three “c’s” was three low characters, and 4) A string of connected characters was considered one character, so that the same three “c’s” would be one low character.
Using the above four choices (and there are, of course, many more which may be tried), back in September of 2007, I wrote that I had, “…come up with some strings of letters… “POL…” was one which occurred. And “QUO…” also.”. But these sort of strings are in themselves possibly meaningless… well are meaningless, unless one can place them in some context. But what really struck me was that for certain combinations of my four variables, I came up with many useless strings, and for others, many more useful strings. By “useless”, I mean the strings did not correlate to any letters in the biliteral alphabet, as Bacon envisioned.” I later explained [with error corrected for this post]:
“I’ll tell you what I did, looking at f106, line 13 as an example:
1) Treating each character, even connected ones, as separate chars, and “A” as lower, and “B” as upper (the slash is before two from the next line, to
even out the group to five):
BAAAA ABAAA BABAB AAABA AABBA AAAAA BAAAA BAAAA BABAB BAABA BAB/AA
Which gives: RIXCGARRXTW
2) Treating connected chars as singles, but still with “A” as lower, and “B” as upper:
BAAAB BAABA BABAA AABAA ABBAA AAABA AABAA BABAB BAABA BAA/AA
Which gives: STWENCEWTR
3) Now with each char as separate, but “B” as lower, and “A” as upper:
ABBBB BABBB ABABA BBBAB BBAAB BBBBB ABBBB ABBBB ABABA ABBAB ABABB
Which gives: QZL???QQLOM You see? Three chars now do not work for any letter.
4) Treating connected chars as singles, again (as in #2), but with “B” as lower, and “A” as upper:
ABBBA ABBAB ABABB BBABB BAABB BBBAB BBABB ABABA ABBAB ABBBB
Which gives POM? U/W ??LOQ …and again, three chars fall “off the list”.
The point being, for the last two choices of variables, I had a much higher number of characters which did not correlate to the biliteral alphabet. A red herring? Perhaps. But all this recent resurgent talk fo the Biliteral has inspired me to get back to it. And again, considering that I am looking at the Voynich as a possible artifact of Bacon’s New Atlantis, and that he invented the Biliteral Cipher, and that both were known and used by the early 17th century, it certainly holds a special place of consideration for me.
All this being said, I feel there are at least two other code and cipher concepts which would also account for the seeming complexity of Voynichese, while being extremely hard to discern, and yet, surprisingly easy and fast to encipher and decipher. Like the biliteral, in my opinion, and just as rarely considered possible.