More About Those Puzzling Pilcrows

Are There Pilcrows in the Voynich Manuscript?

PilcrowBlueIn the last article, I diverged from the Voynich theme to illustrate a brief history of the symbol we know as the pilcrow. I realized the article would be too long if I tried to tackle both the history of the pilcrow and its relevance to the VMS in one go, so this is a continuation of the previous blog.

What does a typographical symbol have to do with the Voynich Manuscript? Maybe nothing. The VMS has enough space on most pages to visually separate the paragraphs, and yet there’s something odd about the behavior of the tall glyphs, popularly called “gallows characters”, a clue that might be important in interpreting the text.

The Duplicitous Gallows

GallowsBeginThe first time I looked at the Voynich manuscript, I noticed a tall shape that looks like a Greek pi with a loop (or two loops) and one that looks like a P were often at the beginning of paragraphs. Sometimes they are embellished with extra loops that appear to be more decorative than meaningful (although that is not known for certain).

At the time, I knew very little about medieval scripts, nothing at all about capitula (although I was familiar with pilcrows from word-processing programs) and I further didn’t know that capitula could occur in the middle of a line, as well as at the beginning of paragraphs.

What I did know was that the VMS had been called a “cipher” manuscript and I noticed immediately that the gallows characters at the beginnings of sections would sometimes alternate (see Folio 3r as an example), so I entertained the notion that different gallows chars signaled a different encryption method, or perhaps a paragraph that required a different decryption key. It hadn’t occurred to me yet that the explanation might be simpler.

After paging through several of the VMS scans, it became apparent that gallows characters weren’t always at the beginning of paragraphs and didn’t all behave as pilcrows—some of them were midline and too close together to reasonably expect them to signify a new section.

This pattern prompted me to do some research on paragraph markers and I discovered capitula (section markers often used to separate units smaller than a paragraph) and noted that the C-shape in old manuscripts served two functions—it could represent a capital C or it could represent the capitulum symbol, depending on context.

Could some of the VMS glyphs have more than one purpose?


This example from folio 58r illustrates how gallows characters are frequently included within the main text, often close together, which provides an argument against them being section markers, but what if the gallows serve two purposes, as with medieval capitula, which can stand as section markers but also represent the letter C? Notice also how frequently EVA-k and EVA-t are followed by the glyphs that look like “ar” or “al” (more often than would be expected in natural languages and something I touched on briefly in another article). Is it possible that even the midline gallows is a marker rather than a letter?

Could the gallows characters be pilcrows or capitula in some situations and letter glyphs in others? If so, a computational attack would have to adjust for this possibility when estimating letter frequencies.

Maybe the glyphs aren’t doing double duty. Maybe they only represent section markers, proper names, or something else in the text that needs to stand out, but if that’s the case then there’s a problem… if the midline gallows are capitula or markers, it reduces the number of VMS glyphs that potentially correspond to an alphabet. The VMS character set is already rather restricted and reducing it further would make it even less like a natural language.

Counting the Pees and Ques Tees

In the EVA font, a character set that helps you write Voynich glyphs, the EVA-k and EVA-t stand for the pi-like gallows with one loop or two. Similarly, EVA-f and EVA-p represent the P-like gallows with one loop or two. There are also some glyphs that combine the gallows with the bench char and are numerous enough to be significant, but it’s a more complex subject, so I considered both benched and unbenched at the beginning of sections as gallows for the purpose of this tally.

I counted only the text groupings that could be identified as discrete sections, what we would call paragraphs. Most of them are fairly clear—either there is space between them or the last line is shorter and a new group begins. Here are the tallies for the first section from folio 1r to 57r, which consists entirely of plant drawings except for the first page.

Out of 219 paragraphs, two were preceded by the red “weirdo” characters that resemble seagulls on the first folio. There were 206 groups preceded by gallows characters, most of them without bench chars, but a few had a full or partial bench character attached. A very small number were preceded by a gallows character with a leading “o”. Including the very small number with a leading “o”, 94% of paragraphs began with a gallows character.

The gallows were distributed as follows:

  • EVA-p    85  (double loop P-shape)
  • EVA-t     66  (double loop pi-shape)
  • EVA-k     40  (single-loop pi-shape)
  • EVA-f      15  (single-loop P-shape)

Gallows with two loops thus occur more frequently than those with single loops at the beginnings of paragraphs, with the most visually ornate of the four being the most numerous (whether by coincidence or design is not known).

Of the 11 glyphs that were not red seagull-shapes or preceded by gallows, all were secondary paragraphs (not the first one on the page) and were distributed as follows:

  • EVA-q      6  (all were followed by the “o”)
  • EVA-y       2
  • EVA-ch     2 (both were bench chars with caps)
  • EVA-s        1

Whether the lack of a gallows character at the beginning of some paragraphs was intentional or accidental is difficult to know without further interpretation or decryption of the text. Most of the exceptions were “4o” combinations that almost invariably fall at the beginnings of word-tokens throughout the manuscript.

A simple count cannot reveal whether a gallows at the beginning of a section is a paragraph marker, especially when there are four different symbols used for this purpose, all of which also show up in the main text. It does seem unusual, however, to have only four letters of an alphabet at the beginnings of paragraphs for such an extended number of pages. Even in medieval books of lists, calenders, and indexes, there is more variation than this when the length of the document exceeds 50 or 100 pages.

On folio 58r, the first full page of text after the plant section, the pattern of leading gallows characters continues, as does that of gallows characters occurring midline.

Perplexing Paragraphs

PilcrowDoubleSlashIn many respects both styles of gallows characters (the ones with two legs and the P), resemble pilcrows. Look again at the marker on the right from a manuscript written in medieval Lombardy—it’s two slashes with a horizontal bar. It’s a bit like EVA-k or -t and the regular pilcrow that resembles a P or backwards-P could be represented by EVA-P. Maybe the VMS gallows characters are a hierarchy of pilcrows, like the red and blue capitula where one is used for greater emphasis than the other. If one were a pilcrow and the other a capitulum, you would expect the “capitulum” to show up more often and perhaps even double for a letter, as it did in most medieval western languages.

GallowsStretchBut what about the strange behavior of gallows glyphs where they stretch over several characters? Is each leg standing in for the same character so it doesn’t have to be written twice? Is it an embellishment? Is it a different letter-glyph, one that’s only occasionally needed? Capitula never stretch over letters like that, do they?

CapitulCrossWordI didn’t think so until recently, and then I found the capitulum illustrated on the right. The manuscript had many traditional capitula and a few like this, where the stem came down several letters later. I don’t know if it’s because the scribe didn’t leave enough room for the stem and compensated by adding it farther on or if it was to give the capitulum greater emphasis. Perhaps it was an embellishment, but whatever its significance, apparently capitula can stretch over several letters.


Could the gallows characters be capitula, name or title markers, letter-glyphs or possibly both? At least some of the glyphs behave like capitula or pilcrows. Assuming that a natural language were behind the VMS, it seems unlikely that so many sections would start with the same two (or four) characters unless their function went beyond an alphabetic character. Maybe they aren’t letters at all. Maybe some or all of them are intended to provide emphasis, serve as modifiers, or as some kind of semantic break between words and perhaps that’s why they’re never placed two in a row.

J.K. Petersen

© Copyright 2016 J.K. Petersen, All Rights Reserved

11 thoughts on “More About Those Puzzling Pilcrows

  1. D.N. O'Donovan

    JKP – could you clarify for me? One correspondent directed me here, saying that this introduction of the “pilcrow” theme into Voynich studies was an original proposition of yours; another tells me it was introduced by Thomas Spande, after a comment made by his daughter.

    So who should I credit in my footnotes? You, Thomas … or Ms. Spande?

    Those interested in the pilcrow might enjoy the very full treatment at a site called “Shady Characters”. Starting here:

    It continues over two later blogposts there, with interesting comments by readers.

    1. J.K. Petersen Post author

      I’m sorry, I have no idea who Thomas Spande is. Except for a few pages of the informative site maintained by René Zandbergen, the plant IDs of Sherwood & Sherwood, the published paper by Tucker and Talbert, the notes of E. Strong suggesting how he “decrypted” the VMS, a quick flyby of some of the blogs (very hit and miss), and recent participation on the forum, I have done almost no reading of other researcher’s ideas on the Voynich. I didn’t even find out about Friedman or Newbold until recently and haven’t had time to read them yet (but I plan to).

      99% of my research is based on studying Beinecke 408 and perusing digitized versions of old manuscripts. The idea that the gallows might be section markers (or decryption markers) occurred to me the first time I looked at the manuscript.

      Thank you for the reference to Shady Characters. I’ll take a look at that one.

  2. Koen Gheuens


    I’ve just read this post and found it very interesting. Your idea of having gallows double as paragraph markers and letter-glyphs answers a question I’ve been dealing with. If gallows roughly behave like upper case letters – as they appear to in my readings of plant labels – then why are there only four?

    The answer might just be that glyphs that were designed to mark paragraphs – or something similar – were also adopted to represent a sound in a more ornate way when desired. I like this idea because it would be a relatively “normal” way to explain a whole lot of Voynich peculiarities.

    1. J.K. Petersen Post author

      “The answer might just be that glyphs that were designed to mark paragraphs – or something similar…”

      Yes. If you look at the “o” character, it also exhibits interesting behavior in the sense of not always acting like a letter (in the natural language sense). It may also be a marker or modifier or possibly both a letter and a modifier, depending on the situation.

      Those looking for linguistic solutions need to study the behavior of the glyphs (studying statistics such as letter frequency, consonant/vowel balance, or the raw number of different symbols is helpful but not sufficient). The VMS script has “rules” for where and how a glyph may be used that are stricter than most natural languages and has characters that may perform specific functions in addition to or as opposed to representing specific sounds.

  3. Koen Gheuens

    The “o” is yet another tricky one. I’m not sure if we really *have* to assume a double function though. Often it is seen in the middle of words on places where one would expect a vowel.

    The best candidate for “other than sound” function is probably when it precedes a gallow word-initially. If we allow the possibility that in this case the “o” changes the sound value of the gallow, it would double the number of sounds that can be expressed by gallows, again solving part of the “not enough glyphs” problem.

    So many possibilities 🙂

  4. Lars Dietz

    The “o”s before gallows in the f58r excerpt pictured in this post seem to be written with darker ink than the rest of the word in many cases, especially in the first line. So they were probably added later, although it seems that enough space for them was left intentionally. No idea what that means, though.

  5. D.N. O'Donovan

    JK, If you thought of the ‘pilcrow’ as a new idea, it’s just another example of syncronicity, but I’m sure you’d like to know for form’s sake:

    Thomas Spande wrote to Nick Pelling’s blog, on Jan.13th., 2014
    ” There is in medieval ms, a marker (we use it now for paragraphs as a proofing mark) and this was called a pilcrow, if memory serves. I think the “&” might be used in the VM by both scribes in this sense ”

    On January 24, 2016 another person called Job mentioned the term but said nothing of substance.

    I myself much prefer Pelling’s site as a source for information, since comments can be made there which ensure that the information offered is correct, and correctly attributed. Not that that’s a problem with Pelling’s work. He is meticulous and when you go back to read the original, it always is the original source. Very helpful in undoing the chain of inadvertent ‘chinese whispers’ sort of mistakes that constantly occur. But each to his own.

  6. D.N. O'Donovan

    Sorry, here’s one I missed, and its the earliest. January 30th., 2010.

    A three-way conversation started between Rich Santacoloma, Rene Zandbergen, and a correspondent to Rich’s blog. The correspondent, William Puskarich asked if there was any example of the pilcrow in the manuscript, to which Rich replied:
    I suppose the Voynich “gallows” characters have some similiarity to these[i.e. pilcrows]. ..

    – so I guess the idea has been hovering about for quite a while, but you appear to be the first to really research it as far as I can tell.

    PS You should know of Thomas Spande. He developed my first mention of the Armenians, and did a great deal of good work, first finding some of the more-recently mentioned Armenian images that have appeared on Stephen Bax’ site. Again – on Pelling’s blog.

  7. D.N. O'Donovan

    Ah – this is what my correspondent must have been talking about.

    Again – ciphermysteries – under post called ‘Voynich Manuscript’

    Thomas Spande returned after a couple of years absence and wrote on February 4, 2016 at 7:27 pm :

    Nick, et al. I have been out of the loop for a bit as I closed up a lab… My daughter Helen, an art conservator has put forward to me in private conversations over the years that the gallows glyphs are pilcrows. Pilcrows for those of you not conversant with these, amount to start/stop symbols in renaissance manuscripts. Our paragraph symbol is the most commonly used one at this time but others existed. Anyway, no real role for those gallows has been widely accepted and I adopt my daughter’s idea that the four gallows serve the function of pilcrows. The pilcrow was adopted from the French in 1440 (Wiki). It had nothing to do with a printer’s mark as some assumed. The two most common, in fact dominant ones are the two that have two ascenders and one loop (I will refer to this as 1) and the one with two ascenders as 2. Glyphs 1 and 2 abound in the herbal section but vary according to whether the loose or tight scribe is at work. Often 2>1 for the loose scribe and 1>2 for the tighter. Doing simple stats on the occurrence of repeats, one sees a pattern emerge. Often (>>50%) one gallows is followed by another of the same: not random at all. I have noted that some “rubbish” text like “898989” (etetet in Armenian) does not appear between the same gallows. I am going to toss a cat among the pigeons and guess that this will prove true of other Voynich gibberish. This leads me to postulate that only the text between the same gallows will be meaningful; the rest are nulls. Well that is a lot of nulls and I think others have proposed that a template or grill has to be placed on the text to get at the reality of the Voynich. The idea above may indicate that the grill is already in place and it is provided by the gallows which are being used in the manner of pilcrows, start and stop instructions. Cheers, Tom

    He then followed this with other posts, in the same place, on Feb.5th and 6th. this year, developing the theme.

    Good to have that cleared up.

  8. Hannah James

    I have been pondering the PI like symbols too.
    Something else which stood out was the symbol that looks like a number 8.
    Possibly, the number 8 could be a varied reference to infinity symbol.
    The reason I say this is because I read somewhere, that the VMS could be the works of Leonardo Da Vinci and we know he often wrote from right to left and mirrored his letters.

    There is no doubt in my mind that a lot of the images are self explanatory. A heavy presence of female birth and reproduction systems in human, plant and planetary systems. Cycles within cycles. As above, so below.

    One thing that caught my eye and pushed me further towards the infinity symbol was the image of the ladies in birthing tubs and one analyst Edith Heywood referred to something resembling ribbons being waved at stars.
    Originally, I thought they looked more like spinal cords or umbilical cords but upon searching ancient infinity symbols, I discovered the word ‘Lemniscate” which is an infinity symbol which means ‘Decorate with ribbons’.

    Going back to the symbols which resemble PI, I looked at the legs of the symbols and some to have the little foot that goes off to one side as on the PI symbol.

    Some of the PI symbols in the text look like the writer started to write the PI symbol in absent mindedness as it would normally be written but placed a curl on one top side. Or, it could be a whole new language.

    A lot of similarities, that’s for sure.

    1. J.K. Petersen Post author

      I haven’t seen any evidence that there is a connection between the Voynich manuscript and Leonardo da Vinci. That was one person’s idea that hasn’t received much support.

      As for the pi-like symbol, there are many ways it could be interpreted, as a symbol, a marker, a modifier, a ligature… a lot of possibilities. It’s similar to the abbreviation used for “Item” in many manuscripts. Some people interpret it simply as a letter in a substitution code, but I think the answer is not quite so simple as that.


Leave a Reply

Your email address will not be published. Required fields are marked *