Top-down structural analysis of the Voynich manuscript

Nothing seems to be known.

There is a rather sharp division into sections (see below), and the section boundaries fall on quire boundaries; so it is possible that each section was originally composed as a separate book.

However, the VMs is fairly homogeneous on several counts, including alphabet, language, calligraphy, and text/graphics layout. So it is almost certain that the book was written by a single person (or, at most, a handful of persons in close collaboration).

The book consists of 18 quires with varying numbers of leaves, some with large fold-outs.

The folios are numbered 1-116, with ordinary arabic numberals. From the handwriting, these numbers seem to be latter additions. The handwriting has been identified by experts as that of John Dee.

Some quires are numbered too, at the bottom of the first or last page.

The folios and quires are numbered in the order in which they occur in the bound book.

However, since the numbers are a later addition, it is not known whether that order reflects the author's intention. The quires and/or bifolios may have been scrambled before being bound and numbered.

Overall, the folio order seems roughly consistent with the division into sections according to the image contents. However, quires XV ("herbal" and "pharma" contents) and XVI (all "herbal") seem to be out of place. Interestingly, quire XV can be split into two "pure" bifolios, f87+f90 (all "herbal") and f88+f89 (all "pharma").

Also, the interleaving of the A and B hands seen in the herbal section could be taken as evidence of bifolio scrambling.

The numbering indicates that a few folios have been removed or lost after the folios were numbered:

• Folio 12 and possibly folio 74 appear to have been cut out.
• Folios 59-64 and 109-110 are two groups of bifolios that are missing from the center of the respective quires.
• Folios 91-92 and 97-98 appear to be a pair of bifolios that are missing from the *outside* of a quire.

[Rene: this is what the Beinecke catalog entry says about quire XVI. Your diagram shows instead two short quires, 16 and 18 in your numbering. Which is right?]

An arbitrary amount of material could have been lost before the folios were numbered. One evidence for this hypothesis is that there is no cover, title page, or preamble: folio 1 has the same overall character as the following pages.

[Folks: I am guessing blindly here, please correct me.]

There is evidence that the Voynich manuscript now at Beinecke is not the original work, but a copy. Here are some arguments:

• An "a priori" argument: books of that period typically existed in more than one copy, and each copy was just as likely to survive as the original. Many books of that period are known only through a single surviving copy.
• There is a surprising lack of marginal annotations and corrections in the VMs. This neatness seems unusual for an original work, even assuming that the author made a "fair copy" from a draft.
• The folios can be partitioned into two or more distinct subsets or "hands", on the basis of calligraphy and statistics (see below); and folios by two distinct hands occur interleaved in the Herbal section.
• The peripheral details in the drawings generally have a familiar European look. That applies to faces, hairdos, hats, castles, weapons, animals, etc. On the other hand, the themes and significant details look utterly alien.
• The layout and basic shapes of the plant drawings seem the work of a skilled artist, but that skill is laking in the strokes and other minor details. Contrast for example the 3/4 perspective view of the topside of the "sunflower" in page f33v (correctly rendered as an ellipse) and the childish treatment of its underside and stem.

If the VMs is a copy, it may have been copied by the author himself, by the reader who wanted the copy, or by one or more professional scribes.

Considering the size of the book, the neat lettering, the colored drawings, and the evidence of two or more "hands" (see below), the last alternative seems the most likely.

From the drawings, and partly also from the tex layout, the VMs appears to be a "technical" book, specifically about plants and herbal remedies, astrology, and related matters.

Barring intentional concealment or minor inclusions, there are a few subject matters we can exclude with some confidence:

• chronicle, history, legend, genealogy, novel, etc.;
• religion, philosophy, laws, moral precepts, etc.;
• geography, ethnography, travel account, etc.;
• mathematics, physics, astronomical calculations, etc.;
• accounting, census, etc.;
• almanac, farmer's calendar, etc.;
• diary, travel log, lab notes, etc.;
• poetry, songs, hymns, etc.;
• alchemy, etc.;
• magic, etc.;
• code book.

Magic and alchemy would actually fit the general looks of the book; but there is no trace of the symbols usually found in other alchemical books, or

Basically, nothing.

A distinctive feature of the VMs is the thoroughly "alien" character of both text and figures. Except for the most superficial details, like the nymph's hairdo, the shape of the letter strokes, there is very little than can be recognized with certainty.

There are many pictures of plants, arranged as in a typical herbal; but not one of the plants has been positively identified. There are diagrams associated with signs of the zodiac; but the divisions and pictures therein do not fit any standard astrological schema.

The pictures in the "biological" section (folios 75--84) are quite bizarre. They could be schematic drawings of body organs, or alchemical apparatuses; we can't tell for sure.

The topic of the "cosmological" section (folios 85 and 86) is a complete mystery.

From the placement and bending of the text lines it is obvious that a large part of the text was written after the figures were drawn.

However, it is also fairly obvious that the two were conceived together. The figures have labels in Voynichese, and were laid out on the page so as to leave space for the text. Moreover, they clearly seem to be dependent on the text for meaning.

There are places where the text seem to have been writtn first, leaving space for the figure (that sometimes didn't quite fit). Page f79v is a possible instance: note that the left margin is indented in steps, whereas in page f77v the margins follow the drawing's outline. Note also the various places where the text actually touches the drawing.

It is conceivable that the figures were added to disguise the true subject of the text. If that is the case, we would not know anything at all about the book's contents...

Fortunately, there are some arguments against this theory:

• There are lots of figures, some quite complex, some requiring complicated fold-outs. That is too much work for a mere decoy.
• The labels attached to the figures can often be found in the text of nearby pages, sometimes with minor changes in spacing or spelling. (It would be very difficult to find these occurrences by hand, without knowning the VMs language; so it is unlikely that the labels were intentionally planted as part of the disguise.)
• The pictures themselves are quite bizarre, and would hardly fool a hostile inquisitor; on the contrary, they would only draw his attention to the manuscript, and fuel his suspicions.
• The the themes of the pictures change from section to section, together with the paragraph structure of the text.

The author clearly put a lot of work and care into the aesthetic aspects of the book. Plentiful colored drawings, neat and consistent layout, clear handwriting --- all this strongly suggests that the book was meant to be seen, if not read, by other people besides its author.

From what we can discern of the topic, the intended reader of the book was some kind of "doctor" in the general sense---astrologer, alchemist, healer, druggist, philosopher, or something along those lines.

Besides the practicing "doctors", we must count as possible readers also the student aiming to become a "doctor", and the educated laymen interested in "doctoral" things.

The following possibilities seem consistent with the book's structure:

There is reason to believe that the text is meaningful:

• The letter, n-gram, and word statistics show a mixture of regularity, arbitrariness, and randomness generally comparable to what we see in natural languages.
• Occurrences of the most common "words" are distributed throughout the text, fairly uniformly. Relatively rare "words" tend to occur in clusters, and similar words often have similar distributions.
• The labels seen in the illustrations can often be found in the text, close to the illustration.
• The text is often visibly squeezed or stretched to fill the available space.

It is not clear how the VMs author could have produced a random meaningless text that long, with such "natural-looking" statistics.

Even less clear is why he may have wanted to do so. If his goal was to fake an arcane-looking book for sale, he would certainly have sprinkled at least a few tantalizing hints that the reader could recognize: Hebrew and greek letters, alchemic symbols, magic squares, pentagrams, etc. The lack of such adornment indicates that the book was written not for its looks but for its contents.

There is some evidence that the person(s) who wrote our copy of the Voynich manuscript did not understand it:

• In spite of the neat handwriting, some statistical anomalies of the text suggest a high frequency of errors, especially confusion of letters with similar shapes.
• The two "hands" observed in the herbal section have different statistics at the character, n-gram, and word level. A good part of the difference seems to be due to confusion between similar letters, each "hand" making each error with different probabilities.
• The spaces between "words" seem to be wrong (see below). The statistics seem consistent with them having been inserted by the scribe without regard for the "true" word boundaries.
• The VMs alphabet seems rather small, the set of "words" rather limited, and there are many places where the same "word" is repeated two or three times in sequence. These facts could be explained as the result of an ignorant scribe missing subtle but essential details of the orignal letters (like the difference between "P" and "R" in English, or "O", "Q", and "D", etc.)
• In the herbal section, when a paragraph of text is interrupted by a plant stem, the half-lines on either side of the stem are sometimes misaligned---as if the paragraph had been written as two columns, rather than line by line.
• The drawings seem to have been copied by someone who did not understand them. For instance, the center right illustration of page f77v probably depicts a tube coiled into a spiral, with shading strokes added for clarity. However, the bends were drawn the wrong way, so that the coil looks flat; and it looks as if the shading strokes were mistaken for a serrated edge.

The bulk of the text is structured into prose-like paragraphs of varying length.

Most paragraphs have between 5 and 15 lines. The lines generally extend from margin to margin, except for the bottom line of each paragraph, which is generally shorter and aligned only at the left margin. The first line starts flush on the left margin.

The margins are roughly parallel, but usually bend to make space for figures. The left marging is distinctly more even than the right one.

Lines are slightly bent and tilted in random ways; apparently, the scribe did not use any sort of guiding ruler. Line spacing is fairly even locally, but varies from page to page. There is usually a bit of extra space before the first line of each paragraph.

Some paragraphs in the Herbal section are split in two or more columns by intruding stem plants. Sometimes the two halves of a lines are visibly mis-aligned.

There are practically no diagrams or other obviously non-textual constructs ("displayed equations") embedded in the text.

There are no visible puctuation signs or paragraph marks of any sort.

[ Don't know this one ]

The first line of each paragraph often displays one or more peculiar letters, often (but not exclusively) at the left margin of the line

The peculiar letters look like exaggerated "P"s with an extra loop or two. They are characters "P", "F", "PZ" and "FZ" in the FSG notation. We call them collectively "the P-gallows",

Statistically, these P-gallows seem to be ornated alternatives of other "ordinary" Voynich letters or letter groups. They seem to be used only in the first line of each paragraph, especially where there is some space avalable above the line.