Monthly Archives: June 2018

The Chameleon Quality of Scribal Conventions

1 Reply

Medieval alphabets, numbers, and abbreviations are often the same shape. For example, the glyph identified in the VMS as EVA-l (ell) was used as both a number and as a scribal abbreviation. In the previous blog, I described the “is” glyph, which is used to create syllables such as ris, tis, or cis. This time I’ll illustrate the flexibility of the EVA-l shape.

Something I noticed, when reading early medieval texts, is that many basic abbreviation symbols were based on Indic-Arabic numbers, long before these shapes came into general use. I’m assuming this was to help distinguish abbreviations from regular letters.

Numerals used as scribal abbreviations

Thus, the number 1 (the old style with a slight wiggle), and the lightning-bolt style of 5 were used for er/ir/re/ri and other sound-alikes that usually include “r”. Number 2 represented ur/tur, number 3 (often written like a zee) was at the ends of words to stand for rem or us/um. Number 4 (as shown above) was a general-purpose abbreviation, 7 became et, and the number 9 was commonly used at the beginnings of words for con/cum, and at the ends for us/um. These conventions continued from the early medieval period until the 16th century. The only significant change was that the number 4 gradually fell out of use  by the 15th century.

The 4 had lapsed as an abbreviation by the time the Voynich Manuscript was created, but it had become common to use it as a numeral.

This clip from a legal charter illustrates the flexible nature of 4. It represents whatever letters are missing, similar to a short macron or curved macron in late medieval texts. It usually stood for one- or two-letter omissions (a bar was more commonly used if several letters were missing). Here it variously stands for m, n, and er:

Scribal abbreviation 4 symbol representing a number of different letters

Examples of the flexible nature of the “4” abbreviation standing in for several different letters or more than one letter.  [Image credit: Stiftsarchiv Reichersberg German legal document from 1231]

You might wonder why a single glyph would be substituted for another single glyph rather than writing the missing letter. Part of the reason is space. There is plenty of space between lines that goes unused, so substituting a superscripted letter shortens the overall length of the document (and the amount of parchment needed). Vanity may also play a part—those who could write and read abbreviated text probably moved up in the social hierarchy.

Paper began to replace parchment in the 14th century, and was less expensive, so some of the superscripted abbreviations, like 9, were lowered to the main text (some scribes wrote it both ways). The 4 continued to be superscripted until it became more strongly associated with numbers rather than with abbreviations:

Illustration of flexibility of scribal abbrevation symbols

I think it’s important to understand how scribes made the distinction between letters, abbreviations, and embellishments if one is to analyze anything written in the medieval period.

I still encounter considerable skepticism about the VMS glyph-shapes being inspired by Latin. They are not as unusual as many people have suggested.

You can search all over the world for that elusive “alphabet” without finding it. In fact, I did exactly that. Even though I recognized these shapes as Latin, I wanted to be sure I had not overlooked anything and spent two years learning dozens of foreign alphabets (Armenian, Syriac, Gujarati, Georgian, Sanskrit, Hebrew, Greek, etc., in addition to the ones I already knew… Korean, Japanese, Russian and a tiny bit of Chinese), well enough to read simple words… and then came back to where I started—almost all the VMS glyphs are normal Latin scribal repertoire, and the few that are questionable are similar to Greek conventions or could reasonably be constructed from Latin scribal building blocks put together in a slightly unconventional and yet acceptable way.

Understanding scribal conventions might help sort out which variations in VMS shapes are meaningful and which are not. For example, in Latin, you can draw the tail on EVA-m in any direction without changing the meaning, but if you change the left-hand side, it becomes another syllable. In Latin, the 9 shape can be drawn any way you want, as long as it is vaguely 9-shaped, but if you move it from the end of the word to the beginning, you change the meaning. VMS glyphs might follow similar concepts even if different meanings have been assigned.

Summary

If some of the VMS glyphs are abbreviations, it creates one-to many relationships of varying lengths. If this were a substitution code, this, in itself, would not be unduly difficult for cryptanalysts to unravel, but in medieval times there was another twist—scribal abbreviations commonly represented not only several letters, but often different letters in each word. In this way, scribal abbreviations diverge from typical one-to-many/many-to-one diplomatic ciphers, in that the interpretation of a specific shape can change from one word to the next.

In Latin, and possibly also in the VMS, two words can look the same, but mean something different.

J.K. Petersen

Copyright © 2018 J.K. Petersen, All Rights Reserved

Le’go my Lego

2 Replies

I’ve frequently posted that Latin scribal conventions are flexible. They are like Lego® blocks that can be taken apart and recombined in many ways. This is essential to understanding Voynich text. Looking for similar shapes is not enough, one has to consider the combinatorial and positional dynamics of medieval text in relation to the VMS.

Basic Building Blocks

I’ve already discussed these shapes in previous blogs, but I’m going to post again with additional examples. This is one of the most common Latin scribal conventions, frequently used to represent the syllables “ris”, “tis”, and “cis”.

The glyphs below are from two scribes with different handwriting styles so you can see how the abbreviations vary from hand to hand, but can still be understood by the reader:

Some parts of the ris/tis/cis abbreviation are meaningful and some can be changed to suit the person’s handwriting or aesthetic sensibilities. This is one of the reasons why it’s important for the dynamics of scribal abbreviations to be understood—meaningful and non-meaningful variations in glyph-shapes need to be differentiated to create accurate VMS transcripts.

For ris/tis/cis, the left side of the glyph defines the meaning (although ris and tis can sometimes be hard to tell apart in 15th-century documents and tis and cis can be hard to tell apart in 13th-century documents). In the example below, it’s quite clear that the left side is “r” for ris, but the shape of the right side can be varied, as long as there is a small loop with some kind of tail. In this 15th-century addition to an early medieval calendar, the tail is long and swings left:

 

A further example (c. 16th century), shows that the ris/tis/cis abbreviations were still used during the Renaissance and early modern period, and could vary in shape, as long as the meaning was clear from the context. Note here that “tis” looks like a small EVA-k at the end of the word hereditatis. It has a straight stem (no foot as in earlier script styles) and a short tail, and yet is perfectly understandable:

Here’s an example in a different 15th-century hand in which ris has a foot and so, to distinguish it from tis, the scribe has given tis a bigger foot and a subtly more rounded back. Also notice how cis is rounded to the point that it looks like EVA-d with a tail. This cis-variation is also found in the VMS and might be a hand-variant of the cis-shaped glyph or might be a separate glyph:

This example, from yet another 15th-century scribe, shows tis and cis in context. Even if you don’t know medieval Latin, you can probably figure out that the word on the right is “dulcis”. Note how the tis and cis are at the ends of the words. This is how they are usually positioned in languages that use Latin conventions and in the VMS, but they can be placed elsewhere if desired:

Similar Shapes in the VMS

Now I’d like to point out a snippet from the VMS that suggests the Latin “is” convention was probably known to the scribe or designer, because it is not always written like EVA-k and EVA-m in the VMS. A right-facing  loop with a stem is sometimes attached to other letters.

On the left, an “is” shape has been attached to EVA-a or, alternately, it might be a straight-leg EVA-d attached to the right side of a bench. Either way, EVA-d has a variant form with a straight leg that is similar to Latin short-leg cis. I don’t have a clip handy, but  “is” also appears next to other glyphs such as EVA-y.

In the example below-right, both ris and cis shapes are visible, but what you might overlook if you are not familiar with Latin scribal variations, is that the character in the middle of the second line is drawn a little differently. This might be a differently-written ris (different from others in the VMS), OR if one were reading this as a Latin scribal glyph, it could also be interpreted as EVA-i with the looped shape added:

Summary

After I wrote a series of blogs on Latin scribal conventions there was a mini-flood of Voynich “solutions” that made the headlines, all claiming that the VMS was abbreviated Latin. They looked suspicious to me because the researchers clearly didn’t understand how to use these conventions correctly and thus could not have come up with the idea independently.

Here are some of the major problems with these “solutions”:

  • Latin scribal abbreviations don’t automatically mean Latin language. Latin conventions were used in all major European languages. Even some of the west Asian languages apply some of the same concepts. The VMS might turn out to be Latin, but not in the simplistic way these researchers translated the VMS text. And it might not be Latin—the same scribal conventions are used in Spanish, French, English, German, Czech, Italian, and other languages. Sometimes the same shape means the same thing in different languages, and sometimes it is adjusted to fit the spelling or grammar of the host language (for example, a squiggle-apostrophe might mean -er in English and -re in French).
  • Most of the proposed “solutions” are nonsense words, not Latin. They might include one correct Latin word out of every 20 or 30 words (more by accident than by design) and they might look a bit like Latin, but it’s not Latin vocabulary, it is certainly not Latin word-order or word-frequency, and it definitely isn’t Latin grammar (not even note-form grammar).
  • Those offering the solutions expanded the abbreviations incorrectly—that was the give-away that they hadn’t really done their homework and probably got the idea from somewhere else. Saying that it’s Latin is easy, anyone can do it—demonstrating that it’s Latin when you don’t know scribal conventions makes it really obvious to those who can read medieval script that the person offering the solution is not even superficially familiar with the basics. One of the solutions, published in a prominent newspaper, expanded one of the most simple and common Latin abbreviations the wrong way (he forgot to take into consideration the position of the abbreviation in the word). Another solution didn’t recognize the fact that Latin scribal glyphs have many different meanings, not just one, and that the meaning is inferred from context. The solver expanded them all the same way which, once again, generates nonsense rather than natural language.

The VMS might be natural language and it might not.  If it is natural language, it’s possible scribal shapes are different from alphabetical shapes. The majority of computational attacks do not take this into consideration, or the fact that if there are scribal shapes, they might need to be expanded.

So, I will continue to post examples, as I have time. Perhaps a better familiarity with medieval conventions will help researchers understand which glyphs might be combination glyphs, which parts might need to be expanded (if the glyphs do, in fact, represent abbreviations and ligatures), and how VMS transcripts could best  be adjusted to provide a more accurate rendering of the text.

J.K. Petersen

© Copyright 2018 J.K. Petersen, All Rights Reserved

C-3P oh?

Leave a reply

C-3PO, of Star Wars fame, knows more than 6 million forms of communication, so maybe he can read Voynichese. Unfortunately, he’s signing autographs on distant planets, so I had to solve this puzzle for myself. Even if you can’t read abbreviated medieval script, you will probably notice this folio includes encoded data.

The following sample is from a 15th century text that deals mainly with astrology but I could see that the subject matter had changed (or, at least, the focus had changed) when I reached folios 160 to 169. I looked around the Web to make sure no one else had already posted about this section and couldn’t find anything, so here it is…

Each folio has two columns. Each column has text on the left and cryptic letters and numbers on the right. This manuscript (BSB CLM 667) is from the late 15th century, but I have seen diagrams in other manuscripts from the mid-15th century that represent information in somewhat similar ways.

As I glanced through it, I noticed these are lists of plant names written with common Latin abbreviations, including abrotanum, gariofilatum, anetum, berberis, tamarisci, malva, strignos, turbitus, costus, epithimum, polipodium, and others. The spelling is slightly unconventional, but the names are consistent with plant names in medicinal herbals.

So what is the encoded information next to the plant name?

I was intrigued because I’ve long suspected that at least some Voynichese might be expressed in novel ways. In fact, I’m hoping it is because it would be more satisfying to discover that it’s a terse code rather than nonsense text. So, several years ago, I labored for almost a year to create a color-coded “concordance” of every token in the VMS, looking for patterns that

  1. might be specific to certain sections of the manuscript,
  2. might link separate sections, or
  3. might recur on certain locations on each plant or other section-specific page. If such patterns could be identified, it might be possible to zero in on sections of codified information that occur on more than one page.

But back to analyzing the text to the right side of each column…

Years of looking at ancient and medieval herbals helped me puzzle out the CLM 667 text in a few seconds because the plant names gave me the context I needed to interpret the rest (I wish the VMS were as cooperative, but then I guess there would be no mystery to solve).

This is how it works…

You’ll notice in Clm 667 that the first glyph in each column is a letter, and is sometimes followed by a number or another letter.

Note that each sequence begins with http://garrygolden.com/2013/11/07/video-post/ c or order Aurogra online no prescription Pharma Life f. That instantly reminded me of Latin calidus and frigidus, properties or “temperaments” that the ancient Greeks associated with each kind of plant.

In ancient medicine, they believed that plants should be chosen to balance their properties against those of the illness. For example, if you had a fever (were hot and sweaty) then plants that were “cold” and “dry” might be suitable for “balancing” your humors. Thus, they felt it important to assign and record these properties.

So, guessing that the first letter represented hot or cold gave me clues to the rest of the sequence. If there was a number after the c or f, it indicated the degree to which this plant embodied the stated property. For example c 2 would represent calidus (hot) in the 2nd degree or f p’ (notice the cap, which is a Latin abbreviation symbol) would mean frigidus (cold) in the 1st degree, with p’ (which can also be abbreviated as p’° or p°) representing primo gradu .

I noted that if the next character was a letter rather than a number, it was always s or h. That confirmed my hunch about c and f. Plants are categorized as hot or cold and dry or wet. In Latin, dry is siccus and wet is humidus. So, if the annotation is f s 2, it stands for frigidus et siccus in secondo gradu.

In contrast, here is a more traditional example for absinthium in an herbal manuscript (Historia Plantarum), created c. 1400, in which the plant is described as calidum et siccum (hot and dry), followed by additional information that it is hot in the first degree and dry in the second. The verbose entry requires 49 characters (including the macron but not including the spaces):

In contrast, the writer of Clm 667 created a simple system for classifying properties of plants that can be expressed with four characters or less.

Now imagine if the lists in Clm 667 were converted to a cipher system. Like the VMS, the text would be extremely repetitious, and it would be very difficult to discern what kind of information was in the “properties” text (especially if the spaces were removed or represented by null characters). Also, like the VMS, the glyph positions would be more regimented than narrative text—certain glyphs would occur more often at the beginning, some more often at the end.

Summary

There are several reasons for posting this example…

  • It illustrates medieval evolution in representing information,
  • it provides a 15th-century example of codified plant data that is outside the mainstream (not everything was slavishly copied in the Middle Ages),
  • it demonstrates that VMS labels shouldn’t be assumed to be nouns (I’ve noticed this  is a very widespread assumption among Voynich researchers)—they may be abbreviated or encoded character traits (easy to say, but this example demonstrates how it might be done),
  • it expressly demonstrates that VMS text should not be assumed to be wholly linguistic. The text may be abbreviated in a number of ways… the VMS could include scribal abbreviations that are linguistic or symbolic, or an entirely different system of coding,
  • if there is codified data that includes numbers, then the majority of Voynich “solutions” are inadequate even if they turn out to be partially right—very few researchers include numbers in their proposed solutions, even though numbers are commonly found in medieval manuscripts, and
  • it provides an example of a system that might account for glyph-priority within tokens.

The idea that the VMS might be ultra-abbreviated is not new—the possibility has been mentioned by others. Highly verbose codes have also been suggested—neither contention has yet been demonstrated or proved.

I’ve been investigating the possibility of codified text for almost as long as I’ve known about the VMS because I’m familiar with data encoding in scientific papers and noticed many of the VMS tokens were quite short and formulaic, but finding a medieval example to confirm that this kind of thinking existed in the 15th century can be difficult if you are explicitly searching for it. Sometimes it’s better to wait until a suitable example comes along, as happened with Clm 667.

J.K. Petersen

© 2018 J.K. Petersn, All Rights Reserved