Monthly Archives: July 2018

Matching the Marginalia

go now It’s a challenge to find paleographic matches to the VMS. I’ve been hunting for almost eleven years, sampled more than 50,000 characters from scripts that were similar to the marginalia or to the main text, and developed tools to more objectively compare scripts based on slant, heaviness, spacing (between letters and lines), and the individual shapes of each letter.

http://vintagegoodness.com/thrift-haul-i-bought-some-stuff-hallelujah/ Evaluating the VMS main text is particularly difficult because some of the glyphs may not be letters, they may be symbols, numbers, or abbreviations, so I will describe the marginalia first, and then explain my strategy for locating scripts similar to Voynichese in a separate blog.

As a sideline, I also looked for matches to the column text on folio 1r, but it is not my primary interest, so I have presented it more simply as a chart, which you can see here.

Selecting the Marginalia

For sampling individual letters in the marginalia, I used these sections:

The more consistent letter-forms on folio 116v (some shapes look like they might be in another hand, although it’s hard to tell, so I gave precedence to those that matched several times on the page rather than those that were questionable), and
The strip of text at the top of folio 17r (which I believe is the same handwriting as most of the text on folio 116v):

Note that I have doubts that the first letter of “valden” is a “v”. For one thing, no one wrote v like this in the Middle Ages (in more than 1,000 samples of text in a similar style, only one v comes close to the shape in “valsen”—this is an unusually low hit rate), but more importantly, if you look closely at the scans, you might notice traces of tails and letters under the last line (to see it you have to ignore the bleed-through from the other side).

Also, the part of the first letter that is visible is a close match to the top of other “p” shapes on this page, which makes me even more suspicious about this letter being “v”. Judge for yourself, here are some examples of common “v” shapes from the Middle Ages (keep in mind that “v” and “u” were used interchangeably in the Middle Ages):

The “v” in “vix” is not common either, usually “v” was written with a curve, but this form is not rare, it is simply uncommon, so I have provisionally used this for “v”, but gave it lower priority than other letters that are written several times (and more clearly).

First Impressions

When I first saw the marginalia, it looked to me like hybrid text, Gothic with hints of both cursive and book hands (I’ll post details later). There are also hints of Italic/humanist hand (e.g., note the small, rounded, spaced-out m).

But… scribes and scholars moved around, so it’s hard to generalize, and there were blended scripts similar to the VMS marginalia in several multicultural areas, including 1) Lombardy, 2) the area encircling the Veneto, 3) some parts of Bavaria/Tyrol, and parts of 4) Alsace/Schwabia.

There’s no sign that the marginalia was influenced by Anglicana, but not all English scribes wrote Anglicana, some adopted the styles of St. Gall. It doesn’t seem as elegant as many of the Parisian scripts, but not all were elegant, and scholars brought many different writing styles to Paris, so… I tried to put my initial impressions out of mind as I searched for similar texts, because I didn’t want to narrow my search based on preconceptions. Thus, I searched digitized manuscripts from as many regions as possible.

Matching the Marginalia

My only requirement for collecting a sample was overall similarity to the basic forms in the marginalia. These scripts tend to score 50 or higher. A really good score is 73 or higher.

Language didn’t matter and, for the most part, dates were ignored (I was hoping dates would fall naturally into place). In general, scripts scored as follows:

I sampled the following characters:

letters of the alphabet,
common abbreviations, and
words like leben/leber/lebe, maria, and vix, when I could find them, or similar words if I couldn’t.

A full sample was typically about 40 characters but not all source texts are long, some are marginalia or short dedications and thus were sampled for as many characters as were available. If the text was heavily abbreviated, I sampled most of the common abbreviations.

Scribes often used more than one shape for certain letters, depending on their position in the word, so I included terminal forms and forms with tails, but “character ratings” were only given to letters that specifically matched marginalia. The rest were evaluated separately.

Here are some general observations from approximately 1,000 reasonably close samples…

Gothic cursive, book, and hybrid hands were common throughout Europe from the late 14th century until the early 16th century, so common that paleographers often have difficulty identifying place of origin.
I was hoping for scores of 95 or better (out of a possible 120 for individual character shapes), but found it extremely difficult to find hands with totals higher than 75. I only have a few. There were a number of interesting finds in the 60 to 75-point range, however. Even if they didn’t match overall, there is some useful information that can be gleaned from individual characters that scored high.
I was hoping high scores might help me geolocate the marginalia-writer’s origins, but many of the manuscripts that scored high are of uncertain origin.
Certain letter-forms are very common, while others are specific to the scribe. For example, thousands of scribes wrote p, r, s, and o like the marginalia, while other letters are unusual or more commonly found in 14th-century scripts or in book hands. I will describe these characteristics in separate blogs.

The marginalia is not abbreviated in obvious or traditional ways, but one gets the feeling that some of the words might be missing letters because the words don’t quite make sense. There are, however, languages that have implied vowels, like some of the Bohemian languages, so perhaps this mindset was familiar to the writer and influenced the way words were written.

Scoring Individual Letters

Each letter is scored differently based on how widely the letter varied in medieval scripts. The letter “o”, for example, is not going to vary as significantly as an ornate letter like “g” or “h-with-a-tail”.

A perfect match scores 6. Out of tens of thousands of individual-letter clips, only one 6 has been given so far. Very close matches generally score 5 because even close matches don’t quite look like the same handwriting. The overall shape might be the same, but there are usually subtle differences in pressure, stroke order, or the way the parts connect.

How objective and consistent are these scores?

The first (and more important) answer is that I have a list of criteria for rating each letter—these could be taught to another person as long as that person has experience in reading old scripts.

The second answer is anecdotal…

One day I noticed two sets of samples lined up next to each other with identical scores. I thought I had discovered two different manuscripts written by the same scribe. As it turned out, I had inadvertently sampled the same manuscript (different folios) almost four years apart. The results were exactly the same.

Why does it matter?

The marginalia may have been added long after the VMS was created, so why would I devote so many years to studying it?

It’s possible it was added later, but research into the letterforms has convinced me that the marginalia may have been written close to the time of the VMS, perhaps even at the same time. If so, then the marginalia writer was potentially a designer, advisor, or even illustrator.

Even if the marginalia was added later, anything that can be learned about the Voynich Manuscript brings us a little closer to understanding it, and if the source of the script style can be located (or connected with other documents in the same handwriting), it might help unravel the missing provenance.

J.K. Petersen

Multilingual Melting Pots

4 Replies

There is a tiny linguistic gem northeast of the Veneto—an alpine village nestled near an important mountain pass. The inhabitants speak Tischlbong, a language related to Carinthian, a south-Bavarian dialect spoken by Slovenes who inhabit the mountains of the southern Tyrol and parts of the Slovenian Styria. As a distinct dialect, Carinthian dates back to about the end of the 13th century.

Timau by Mikmaq, Wikipedia

The little village of Timau is known since the early 13th century, and was once part of the Lombardic-Venetian empire. Timau (once known as Teschelwang), was rebuilt after severe flood damage and became part of Italy in 1866. The younger people now speak Italian, but also generally understand Friulian, a Romance language with roots in Ladin.

It is the older people who are fluent in Tischlbong, a medieval dialect shared with Slovenian residents of the Tyrol. A similar situation exists in Lombardy—the Lombardic language is mostly spoken by the elders, with the younger generation moving over to Italian.

Talking Tischlbong

At first the Tischlbong language seems odd, but if you switch your brain to German and note the patterns, after a while it makes sense. Substitute “b” for “w”, sometimes “p” for “b”, and drop some of the endings and doubled letters, and one gets words like

is da (ist die/is the); af (auf/of)
Otobar (Oktober/October)
ausar (her ausser/but for)
varlosen (verlassen/leave)

which has some commonalities with Voynichese in its simplicity and emphasis on vowel forms.

Tischlbong caught my attention not only because of its unique characteristics of dropping characters and compressing words, but because the larger Carinthian-speaking population has multicultural connections to Graz, the Veneto, and Slovenia. Historically, the principality of Carantani extended as far as Salzburg, on the German border. Forerunners to the language may also have been spoken in Swabia.

Klenje, Wikimedia Commons

Also noteworthy is that many of the inhabitants of Timau communicate with neighboring villages in Friulian, a language that includes words from Latin, Ladin, Venetian, Lombardic, and even ancient Celtic, while still retaining some elements of French grammar. Thus, Friulian adds a strong Romance component to the local culture along with a number of German words.

As examples, we see bon viaç (French bon voyage) for good journey, buine sere (Spanish/Italian) for good evening, and cràmar (from MH German Kramaere) for pedlar or haberdasher.

Records of these languages and villages begin to emerge in the middle medieval period, early enough to precede the Voynich Manuscript.

There are several places where combinations of languages happened naturally. One I’ve mentioned several times is the Burgundian corridor bordering Italy, Provençe, and Switzerland. I’ve also blogged about blended Latin and German, but wanted to include the eastern side of Lombardy, southern Tyrol, and the Veneto, as well, and to mention Tischlbong and Friulian (and their sister dialects) in the context of blended languages.

Summary

The text on folio 116v of the Voynich Manuscript is somewhat Germanic on the top line, Romance on the second line (at least in structure and the balance of vowel sounds), Latin on the third line, and Voynichese and Germanic on the fourth line. If a resident in one of the borderland villages of mixed Germanic/Romance origin were writing something for his or her convenience, it might come out in a similar way.

As for the main text, it is very concise and sparse and, in some ways, reminded me of Tischlbong. I’m not assuming the VMS is natural language, but I do think it came from the mind of someone who liked to focus on what was important and who did not include a lot of unnecessary “extras”, and many Tischlbong words, when compared to their German analogs, defy convention and include only what is necessary to understand them and no more.

J.K. Petersen

A Line in the Stars

3 Replies

Recently Voynich Views posted a blog about the line (or lack of one) connecting the fish that represent the astrological symbol Pisces, paying specific attention to those that look like the VMS long-nosed fish. I have a dataset of more than 500 zodiac cycles, so I thought I would post them to help round out the picture.

Long-nosed fish are definitely in the minority. Only about 8% of the total samples in my files have long noses, and if you filter out those with a line between the fish, only 3% remain.

The majority of those with long noses originated in France, with a few from Austria and Germany. There were also single examples from Italy, Spain, Belgium, and the Netherlands. If you look back at the map of symbols similar to VMS Scorpio, you will see that several of those that resemble the VMS lizardy Scorpius are also from France.

The earliest examples, mainly from the 12th and 13th centuries are from France, England, Cologne, and Austria. Those enclosed in circles are from France, Austria, Germany, and the Hague.

Only two or three of the fish pairs have the same “wavy” shape as the VMS fish (a couple from Paris, possibly one from England, and one from Austria). Note that the 12th century zodiac from Seckau, Austria, may have had the drawings added in the 15th century, so I have sorted according to the later date [click to see larger]:

It’s interesting that so many of the VMS details are in the minority. Crossbow Sagittarius is relatively uncommon, as is lizardy Scorpius, and double-Cancer. The Gemini “courting couple” is also uncommon.

Pisces with a long nose is quite uncommon, especially those without a line between the fish. I sometimes wonder if the person who designed the VMS zodiac signs got the general idea from looking at zodiacs but inserted pictures copied from other sources. For example, in CLM 10072 and BNF Lat. 13025, we see a long-nosed fish in a pond, and a long-nosed fish with the scales emphasized in an embellished initial. The Book of Wonders and the Rochester Bestiary also have a variety of long-nosed fish. Plus, there are a couple of “zodiac man” images with long-nosed fish under the man’s feet that I didn’t include in the above compilation.

Another way that the VMS fish differ from many fish drawings is the way the scales cover the whole body. Many illustrators don’t put scales on the bellies:

The wavy form of the VMS bodies is also unusual. I found a “wavy” fish in an Arabic manuscript, but other than the undulating posture, it doesn’t strongly resemble the VMS fish. It has spots, three fins top and bottom, a pale belly, and a drawing style that is much more eastern:

There are also some long-nosed fish in the marketplace and in the water in Tacuinum Sanitatus:

This fish is scaly top to bottom, but the fins and nose are quite different from the VMS fish:

In the planetary influences section, Cod. Sang. 760 has a human Sagittarius and some scaly long-nosed fish, but the fish are roughly sketched and probably later than the VMS:

Morgan B.19 has a very cute drawing that looks like it was done by a child that includes a long-nosed fish with double dorsel fins:

In the Andalius De Sphaera (c. 1327), we see long-nosed dolphins. They’re not fish, they’re mammals but many old texts refer to them as fish and even today many people call them “fish”:

Family emblems, such as this one (which also features cloud-band imagery), sometimes include long-nosed fish:

The best match I’ve seen for the VMS fish outside of zodiac imagery was the one pointed out by MarcoP in the Sachsenspiegel. It has a long nose, a face quite similar to the VMS fish, and is scaly top to bottom. I’ve turned it sideways so it is easier to see:

Summary

Even though the VMS long-nosed fish are unique in a number of ways, there are several possible sources for the imagery, including zodiacs, bestiaries, and manuscripts unrelated to astrology.

The Parisian zodiac dating from c. 1350 (Morgan M.75) is the closest I’ve found so far as a Pisces symbol, in terms of style and features. Morgan M.75 also includes a four-legged Scorpius and crayfish Cancer, so perhaps there is some common source that connects them, but I’d like to gather more information and continue analyzing what I have before venturing any conclusions.

J.K. Petersen

[Postscript July 27, 2018] I found this picture after I posted the blog:

The drawing style is different from the VMS, but the features are notable. It has long-nosed double-dorsal fish that are scaly from top to bottom. Plus, something that is even harder to find in fish with these characteristics… the lower one has an undulating posture. The image detail is from a 14th-century hell-mouth fresco in the Timios Prodromos monastery in Serres, Greece.

Serres also has grottoes with fabulous stalactites. I’m always on the lookout for grottoes since some of the VMS bathing images have grotto-like structures.

J.K. Petersen

Postscript 30 August 2018: It has been suggested that the VMS Pisces symbol represents the New World alligator gar. Here is a picture of the gar. Note there is one fin top, two sets of fins bottom-front, and another fin back-bottom. Drawn from the side, it might look like one top and three bottom.

If the researchers who identified the VMS fish as alligator gar are going to ignore the mismatch in the number of fins, then any fish with a long nose could be said to be similar to the VMS fish, including the northern pike, which is indigenous to both Old and New Worlds:

The same researchers claim the fish with the nymph in its mouth on folio 79v is also an alligator gar. I can’t take that seriously. Look at the mouth, it’s very broad and blunt, more like a wide-mouthed bass:

In contrast, the mouth of the alligator gar is long, narrow, and toothy (especially this species):

Natural or Traditional?

Four years ago, I looked at hundreds of pictures of New World and Old World fish and there are several with long noses and two fins top and two bottom, but I have the feeling the VMS illustrator consulted other drawings rather than studying real fish because VMS Pisces resembles medieval illustrations more than it does naturalistic drawings of the time.

Here is a small subset of Pisces symbols with varying numbers of fins and long noses. They are all Old World illustrations:

I’ll leave it to the reader to decide whether the VMS fish look more like the New World alligator gar, or like fish in pre-conquest medieval illustrations. I’m partial to the fish in the Sachsenspiegel with its similar face and tail, and to the fish in the Greek fresco, which have the right number of fins, top-to-bottom scales, and narrow curvy bodies.

Small Clues in the Marginalia

4 Replies

I’ve mentioned that the person who wrote the VMS marginalia may have learned to write in the late 14th century or early 15th century. If so, he or she could have lived at the same time as those who illustrated and wrote the VMS, and might even have been involved in the creation of the manuscript in some way.

I’ve already posted some of the statistical properties of the marginalia “g” on the voynich.ninja forum. Here is a reprise of the comparison chart:

Samples of the letters “g” and “a” suggest the marginalia writer was familiar with older handwriting (e.g., 14th century), or with more formal book hands, and it appears that the writer may have learned to write in the late 14th or early 15th century. This suspicion is not based on just one set of statistics, however.

The Sample Set

To sample the marginalia, I used the text on 116v and 17r because I’m fairly confident it is the same handwriting. I did not use “ven mus mel” because there are too few words to be certain if it’s the same. Many scribes had similar handwriting. In fact, they were encouraged to copy each other so several scribes could work on the same manuscript without awkward transitions.

Below is the sample set I use for full comparisons. Note that the “z” is actually an abbreviation symbol, but most scribes wrote it like the letter “z” and sometimes also the number “3”, so I provisionally stored it in the “z” slot.

It’s not clear whether the last letter of “gaf” on the last line is “f” or “s” but since most scribes wrote them the same (except for the crossbar), it probably doesn’t make a big difference. When assigning values to similarity, I gave higher priority to letters that are repeated and clearly written. I consider the “u” and “v” shapes to be lower priority since “u” is not clear and the “v” appears to be in a different style.

Note that “n” has both a descending “tail” and a non-tail form. This is very common. Many scribes wrote n, m, and h both with and without a tail:

Now here’s an interesting puzzle…

Is the first “a” on f17r a double-chamber “a”, the style that was popular in the 14th century, but less so in the 15th century?

It’s hard to tell, but it looks like it might be, if the top is slightly fuzzed. It was very common for the two-chamber “a” to be taller than other characters nearby:

It’s impractical to post exhaustive statistics or lists of manuscripts that have different styles of the letter “a” in a blog, but to give the flavor of it, here are examples from the 14th century with only the double-chambered “a” (from scripts with some similarity to the marginalia):

In paleography, there are many hybrid scripts (scripts that blend different styles). Paleographers frequently argue about names and classifications (some have devoted whole books to this), but diversity isn’t a bad thing—it can help identify an individual’s writing.

The Significance of the Double-Story “a”

As the single-story “a” became more popular in the 15th century, some scribes mixed both singles and doubles, sometimes even in the same sentence. Here you can see both kinds of “a” in marginalia that was added c. 14th century to an early medieval calendar:

Sometimes the mixing was random, other times, the double-story “a” was used as an uppercase “a” or at the beginnings of words.

The double-story “a” was retained for a longer time in more formal book hands, but casual hands gradually abandoned it for the single-story “a”. By the late 14th century and early 15th century, many manuscripts included only the single-story “a”:

Some Statistical Properties of the Marginalia “a”

As mentioned previously, I set up a mathematical system to compare letters more objectively. A perfect match on higher-priority letters would be 120 (with slant, weight, connectivity, spacing, etc., assessed separately). After 10 years, 5 months, and 5 days of searching for a close match, I still find it quite difficult to locate scripts higher than 75, so a bit of luck will probably be necessary to do better.

It surprised me that a medieval “a” could be so varied. Out of a sample of more than 1,000 manuscripts specifically chosen for their similarity to the VMS marginalia, only 20% scored 4 or higher (out of 6) for letter-forms.

Note that the stem is fairly vertical, and the loop doesn’t quite close on the bottom. There is a small point at the top of the stem and the serif turns up at a fairly sharp angle.

Finding all these characteristics together is a challenge, but here are some of the better matches. Note that the overall score (listed under the letter) isn’t always high. In other words, even if the “a” was fairly close to the VMS script, sometimes other letters were not:

If the letter on 17r is a double-story “a”, then it has some physical and statistical commonalities with the single-story “a”. The stem is fairly upright and the bottom loop is not quite closed. Note that the bottom loop is larger than the top one (some scribes did it the other way around).

Here are some examples from handwriting that is fairly similar to the VMS (note that only one “a” is significantly similar to the VMS “a” in shape, proportions, and slant, and it lacks the sharp serif):

When the double-story “a” is substituted, the statistics for overall similarity come out very much like the previous search, as did the date range of most of the samples. In other words, scribes with handwriting similar to the VMS writer used either or both forms of “a” around the same time period as they wrote letter-forms similar to the VMS.

Summary

The greater popularity of the double-story “a” in the 14th century might suggest a date earlier rather than later in the 15th century, although there are a few areas where the double-story “a” was still common until the end of the 15th century. Some parts of eastern Europe, and many of the smaller more remote towns, lagged behind large urban areas in terms of the evolution of script. It does seem very likely, however, based on the paleographic evidence, that the marginalia was added after 1330 and before 1490. It’s not as precise as radio-carbon dating, but it does help when research from various disciplines tends to confirm rather than deny the existing data.

I have much more information on this topic, but will cover it in separate blogs, and I will leave it up to the reader to decide whether the letter on folio 17r is a double-story “a”.

J.K. Petersen