The Voynich Botanical Plant Names Decoded

“If you find from your own experience that something is a fact and it contradicts what some authority has written down, then you must abandon the authority and base your reasoning on your own findings.”
— Leonardo da Vinci

Ethel Voynich and the botanists Hugh O’Neill and Theodore Petersen were the first to identify some of the drawings in the botanical section of the Voynich Manuscript (VM). Modern botanists have ignored their work, dismissing these drawings as a mishmash of flowers, leaves and roots belonging to unrelated plants and not worthy of any attention. Initially I accepted their evaluation of these bizarre drawings, however, not all of them should be dismissed in such an arbitrary fashion. Ethel Voynich correctly identified folio 9v, as representing a viola and folio 2v, a waterlily, folio 56r is obviously a sundew. These drawings are not the product of a fanciful imagination. The manuscript is over 550 years old. Illustrations from some contemporary and earlier medieval herbals are just as bizarre. I have written two articles attempting to identify the plants in these drawings(i)(ii) but have never felt confident that I identified all of them correctly. This would require decoding the plant’s name in the accompanying text to establish identity with certainty.

As this appeared to be an impossible task, I lost interest in the drawings until I read that the little used spice, Sumac, was popular in 15th century Italian cuisine and wondered whether it was one of the VM plants. I did not have to look further than folio 4r. On the second line of the folio’s text is the following: ccam, a part of the drawing, an upside down u and a mirror image s. Written in reverse, ccamus, reads sumac or sumach depending on whether cc represents a single c or a ch. Both spellings for sumac are correct. See Figure 1. This was a surreal moment.

Decoding the plant names required the following information:

• The language used to write the manuscript
• The selected language’s names for medieval plants and herbs
• A comparison of the VM’s unique alphabet with our Latin alphabet
• The identity of the code
• A method for testing the validity of the above assumptions

A daunting task.

Language: Italian. Sergio Toresella, an expert on ancient herbals, suggested that the VM’s script and the style of the herbal drawings is consistent with documents from Northern Italy, written around the year 1460(iii).

Plant names: My Oxford Italian dictionary was not very helpful in providing the common names of Italian plants and herbs. While searching for other Italian dictionaries, I had the good fortune to find Florio’s 1611 Italian dictionary(iv). This dictionary provided exactly what I required, even the Italian name for the various “worts” that the English used to name their herbs (e.g. bloodwort, lungwort, colewort etc.). The dictionary has the disadvantage of only having an Italian/English version. The identification of over 100 of VM’s botanical and some herbal drawings is the result of examining over 600 of the dictionaries pages for common Italian plant and herbal names.

Alphabet: I was unable with EVA (European Voynich Alphabet) or similar alphabets(v), to decode many plant names in the herbal section of the VM. However, with the 8 symbols in EVA, that resembled lower case Latin letters, I was able to decode a few words in the herbal section of the VM(vi), Table 1.

a	e	o	ch	i	p	r	s

Starting with the symbols in Table 1, I developed the alphabet in Table 4, dubbed AVA (American Voynich Alphabet). It includes the eight symbols from Table 1, additional symbols that also resemble lower case Roman letters and a few symbols whose identity was determined by trial and error. The AVA alphabet was basically derived as follows:

• Table 2 shows the percentages for the vowels “a”, “e”, “i” and “o” calculated by combining the values given for Currier's A and B botanical folios(vii). These values are compared to the percentages of the same vowels for Italian, English and German. The percentages are closer to Italian than English or German where “e” has a higher percentage than the other vowels.

• The percentages for the two symbols on the right are 14.3% and 11.7% respectively. The “9”-shaped symbol is only found at the beginning or the end of a word. Trial and error indicated that it may also represent an “a” which combined with the 8.7% “a”-symbol percentage from Table 2 brings the total for the letter “a” to 24.0%. The “8”-shaped symbol represents the letter “b”, also determined by trial and error. Captain Prescott Currier subdivided the VM into A and B folios, suggesting that two different scribes wrote the VM in two different languages. This theory is based on observing that the combination of these two symbols occurs 8 times within 25 A pages and 554 times in 25 B pages(viii). Table 3 shows the decoding and translation of the two words “ba” and “baba” and indicates that the content of the text rather than two different scribes may account for the atypical presence of these two symbols on the B pages. Neither of the words, ab or abba, occur in a modern Italian dictionary. There appears to be a significant difference between modern Italian and the Italian used in the 16th century and earlier.

• Trial and error led me to conclude that most of the other symbols also represent the letters in the Latin alphabet that they most closely resemble. See Table 4 for additional examples.

• The symbols for “c”, “p” and “v” were determined by decoding anagrams for sumac, folio 4r; apio, wild parsley, folio 65r; and viola, folio 9v. The anagram for viola also established the symbol for “l”.

• Trial and error indicated that the “gallows” symbols represent the lower case letters, p, l and tl respectively.

• Some letters have more than one symbol.
• The Italian alphabet has only 21 letters. “j”, “k”, “w”, “x” and “y” are not assigned symbols.
• The Voynich text does not appear to use upper case letters or punctuation.

Table 4 provides a comparison of the symbols with lower case Latin letters and was used to decode only the plant names. It could, with some additional work, be used to decode all the text.

a

a1

b

c2

d3

e

e4

f

g1

h

i

i5

l6

l

m	n	o	p	p	r	s	t3	u	v7	tl

1. “a” and “g” are similar
2. The symbol for “c” may also represent “ch”
3. “d” and “t” are also similar, except the “d” lacks a line through the middle
4. Alternate symbol for “e”
5. A normal “i” is occasionally used (folio 50r, 2nd word)
6. This symbol for “l” may also represent two “l”s.
7. A “v” may replace a “u”, particularly at the beginning of a word (see Florio’s 1611 dictionary)
8. Many of the embelishments make the VM’s alphabet appear confusing. They are meaningless and should be ignored.

The alternative to AVA is EVA’s stroke based alphabet. EVA considers that the letters m and n may be represented by a series of sloping strokes with or without a curve on the final stroke and therefore represent different letters(ix). EVA’s and AVA’s interpretation of many of the VM’s symbols is very different and brings to mind Occam’s Razor. His principle states that when choosing between competing hypotheses that potentially may be equally useful, the simplest one should be selected. Simplicity is the key to an alphabet used to encode the text of a manuscript of 116 pages with no indications of any errors or corrections. The writing flows smoothly, giving the impression that encoding the text did not requiring much thought. Employing a code that used a number of symbols resembling lower case Latin letters could easily have become automatic and an occasional error, like substituting a normal i would go unnoticed.

The AVA alphabet should be considered in conjunction with an anagram based code. Neither the phonics of a symbol or its order in the word are important. Various symbols may always appear at the beginning of a word, it is up to the discretion of the scribe and you will discover as you read on some words are split into two or three anagrams. The code is flexible. The appendix demonstrates AVA’s authenticity by enabling the decoding of the names of 111 of VM’s plants.

Code: Attempts to decipher the VM’s text(x) include, frequency analysis, single letter substitution codes, phonetics(xi) and Gordon Rudd’s hoax hypothesis(xii). The VM’s alphabet complicates decoding. A simple substitution code fails to explain the arbitrary order of the symbols within the words. Elizabeth Friedman has stated that simple substitution codes do not work(xiii). A code based on anagrams can explain why some symbols frequently appear either at the beginning, middle or end of words. A sequence of words that differ by only one letter occurs fairly frequently in the VM’s text and cause single letter substitution codes to yield babble-like text. An anagram dictionary will show that decoding the anagrams of words differing by a single letter is not a problem. Anagrams are a simple way to code a lengthy manuscript, but have the disadvantage of being degenerate and when decoded may produce spurious results.

Method: The herbal section provided the initial key to unlocking the VM’s code. These drawings of single plants or parts of plants are labeled presumably with the plant’s name, i.e. the VM’s Rosetta stone.

Discussion: I suggested, in my article, The Voynich manuscript Decoded?(xiv), that the labels in the VM’s herbal section were coded in Italian, using anagrams and an alphabet similar to AVA. Slashdot selected this article for solving the VM’s code, but it was rejected by some of VM’s investigators who may have considered that the drawings were not identified correctly and a code based on anagrams could not result in the labels being decoded with any degree of certainty. The Latin anagram , that Galileo sent Kepler to inform him that he had discovered rings around Saturn, was quoted to discredit a cipher using anagrams. Kepler’s decoding of this anagram, SMAISMRMILMEPOETALEUMIBUNENUGTTAUIRAS, when translated into English, was. “HAIL, TWIN COMPANIONSHIP CHILDREN OF MARS.” It should have read “I HAVE OBSERVED THE MOST DISTANT PLANET TO HAVE A TRIPLE FORM”(xv)(xvi). Galileo wanted to safe guard his astronomical discovery prior to publication. He believed that Kepler might publish this discovery himself. William Friedman also disguised his solution to the VM’s code, a synthetic language, in the form of an anagram. “I PUT NO TRUST IN ANAGRAGATIC AND ACROSTIC CYPHERS, THEY ARE OF LITTLE REAL VALUE – A WASTE – AND MAY PROVE NOTHING – FINIS’(xvii). No one was able to decode his anagram, but he left his assessment of the VM’s code in a sealed envelope which when opened after his death was: “THE VOYNICH MANUSCRIPT WAS AN EARLY ATTEMPT TO CONSTRUCT AN ARTIFICIAL OR UNIVERSAL LANGUAGE OF THE A-PRIORI TYPE – FRIEDMAN”. Galileo’s anagram has 37 letters with 1.38* 10 43 permutations. A very large number making it improbable that anyone could decode it. The mathematical expression for calculating the number of permutations for an anagram is x!, where x represents the number of letters in the anagram, i.e. an anagram of 4 letters would have 4! or 1*2*3*4 = 24 permutations, a 5 letter anagram 120 etc. The number of permutations of anagrams does not limit their usefulness as a code. The order of the letters in a word is not random and an Italian anagram dictionary shows that few anagrams of 6 to 10 letters will represent more than 1, 2 or 3 words. The VM’s 7 letter anagram for the name of the plant from folio 37v is, (orcocor), when decoded produces 2 words, one being corcoro, the Italian name for the plant, Pimpernel. The 6 letter word, (coiban) from folio 47v, with 720 permutations, when decoded produces 3 words, one is bianco, the Italian name for the herb, lungwort.

Medieval scientists and inventors did not have a patent office to protect their intellectual property, getting work published was a slow procedure. Galileo used an anagram very effectively as a way of protecting his discovery prior to publication. Kepler was never intended to decode it and when Galileo was ready, he decoded the anagram himself. Roger Bacon claimed “that no scientist should ever write of his discoveries in plain language, but must resort instead to concealed writing”(xiii). At that time, a code based on anagrams, for all its shortcomings, was a popular and accepted method for protecting intellectual property. Only anagrams of single words, not complete sentences, were used to code the VM. Not all anagrams are degenerate and those that represent a single word may provide a clue as to the meaning of the text. The question is, was the VM’s author protecting his work from prying eyes, or was he hiding secret information? I agree with Wilfred Voynich’s assessment of the manuscript: “the drawings indicate it to be an encyclopedia work on natural philosophy”(xix).

Results: Armed with the name of a plant from Florio’s 1611 dictionary that resembled a drawing from one or more of the VM’s botanical folios, I proceeded to check to see if any of the text in these folios contained an anagram of the plant’s Italian name. Only if the VM’s drawing matched an image of the plant cited by Florio’s dictionary, and the anagram, when decoded, matched the Italian plant name, did I assume that I had correctly identified the folio’s drawing. I never found any folios with similar drawings with the same plant name. This procedure has identified over 100 of the VM’s plant drawings. The results are given in the appendix to this paper. They confirm that the VM’s language is Italian, the code is a modified anagram cipher and the symbols, in general, represent the letters they resemble.

While examining the text in the VM’s botanical pages, I made the following observations:

• The first word in the text of the botanical folios is not the name of the plant, its anagram may be anywhere within the text. Nick Pelling has pointed out that the first word generally begins with one of the four “gallows” symbols, making it unlikely that it represents the plant’s name(xx). As the large drawings serve to identify these plants, when text was added later, the position of the anagram/name was not important. The German herbals, printed later in the 15th century, first provide the Latin or German name of the plant, followed by text and a small illustration.

• The right and left hand margins of the VM’s text are even. This is due to the author writing from left to right as far as the right hand margin, stopping, even if he was in the middle of writing a word and continuing on the next line. As a result, one or two letter words are found at some right and left hand margins.
• Some words are written around drawings. The example representing sumac shows that the anagram for sumac is split into two parts. The words in the VM’s text are fairly short; lengthy words may be split into two or more anagrams.
• There are practically no words with fewer than two letters in the general body of the text. The Italian language uses 1 or 2 letter words for conjunctions, prepositions and definite and indefinite articles. Single letters like a (with the, for the), e (and, more over), i (I, they, the) and o (or) or double letters like ce, la etc., when used are incorporated into the anagram of the plant’s name. Florio’s dictionary provides a translation of these 1 and 2 letter words.

All the folios in the appendix use the same template and format, as illustrated for Folio 4r.

• The drawing from the VM’s folio and an authentic illustration of the plant cited ln Florio’s 1611 Italian dictionary, are compared. As an additional check, I also compared, when possible, the folio’s drawing with a drawing of the same plant from the 10th century herbal, Materia Medica(xxi). Many of the botanical folio’s drawings of the herbs, vegetables and flowers can be found in Materia Medica, this helps to confirm their identity. It is unlikely that the VM’s author ever saw this herbal. Materia Medica is based on the work of the 1st century Greek physician Pedanius Dioscorides and was an important and influential herbal until the end of the 15th century.
• The anagram of the plant’s Italian name is reproduced from the folio’s text and its position in the text is provided.
• The anagram is transcribed using the AVA alphabet (Table 4) and decoded. The spelling of the decoded anagram and Florio’s Italian herbal name may not match exactly, but if they are similar phonetically, I considered that the folio’s drawing was identified correctly. Italian is a phonetic language and spelling was not important(xxii) until the 19th century.
• Florio’s 1611 Italian dictionary’s Italian name for the plant with the English translation, reproduced from the dictionary’s page on the internet.
• A link to the plant’s illustration is given below the box.

I considered the VM’s folio drawing correctly identified, if the drawing, a picture of the plant cited in Florio’s dictionary, its Italian name and the decoded anagram are a reasonable match. About 86% of the plants in the botanical section have been identified by this process. A few folios like folio 53v, hawkweed and folio 56r, sundew, are not included because I have not found their Italian names. The same is true for the remaining missing folios.

Conclusion: I made the following analysis of the 112, or 86% of the plants identified in the botanical section:

• 65 or 57% of the 112 decoded anagrams have no additional symbols and vary from 3 to 8 symbols in length; 53 are single anagrams, the anagrams representing the names for the remaining 12 plants are a combination of 2 or more shorter anagrams.
• There are 56 anagrams that include additional symbols that represent prepositions, conjunctions or definite or indefinite articles. Such inclusions are consistent with the absence of one and two letter words in the VM’s text, one reason for making this manuscript difficult to decode. These identifications could be excluded, but they may provide additional insight into translating all the VM’s text.

This work depends on the code, language and correlation of the VM’s alphabet with the Latin alphabet being correct. It is possible to find fault with anyone of these assumptions. The best proof I can offer is that 112 of the VM’s botanical drawings are identified. They represent a cohesive group of plants, mainly herbs and vegetables grown in Italy in the 15th century. It was the use of symbols, that in general look like the letters they represent, the splitting of words into two or more parts, the combining of one or two letters, representing prepositions, conjunctions or a definite or indefinite articles, with the plant’s name, and a code based on anagrams, that makes the VM’s text so difficult to decipher.

Acknowledgements: I would like to thank Steve Bush for making Florio’s 1611 dictionary available on the internet, this dictionary made the above paper possible. A special thanks to Rene Zandbergen and Jason Davies for their very useful web sites, to Wikipedia for the information that enabled me to identify the “worts” and other plants mentioned in Florio’s dictionary and to the Beinecke Library for providing access to the folios of this mysterious manuscript. I am also grateful to Rene Zandbergen for providing me with the link to Materia Medica’s web site and some useful discussions. As always I would like to thank my daughter Erica for the time and effort that she spends on my web site.

Addendum: Dante’s Divine Comedy is one of the world’s greatest works in literature. It was written in the 14th century and was the first Italian literary work to use the dialect of the people of Florence and Tuscany. Prior to this epic poem, all intellectual works in Italian literature were written in Latin. Unlike the English alphabet which uses 26 letters, Dante used only 21 letters; the letters “j”, “k”, “w”, “x” and “y” were omitted. Italian, a major European language, has its origin in the “writings of Dante.” He is credited with standardizing the grammar and core lexicon of the Italian language, and thus the dialect of Florence and Tuscany became the basis for the official language of Italy.

A discussion of “Dante’s Italian dialect” (volgare) is important for the following reasons:

• Other regions in Italy during the Middle ages used different dialects. Some are still in use today.
• In this article “The Voynich Botanical Plant Names Decoded,”(xxiii) I used Florio’s 1611 Italian dictionary to decipher the VM’s anagrams representing the Italian names for many of the manuscript’s plant drawings. The dictionary uses Dante’s 21-letter alphabet, basically omitting the letters “j”, “k”, “w”, “x” and “y”. There are a very limited number of words beginning with the letter “x”. The Italian language and spelling used in the dictionary and therefore the Voynich Manuscript is compatible with both “Dante’s Italian dialect” (volgare) and modern Italian.

As a result of reading the above article, Robert Ward a student in Italian and of Italian decent, suggested that I investigate what Italian dialect Leonardo da Vinci used when he wrote his notebooks. This has proved to be a very valuable suggestion. I consulted J.P. Richter’s “The Notebooks of Leonardo da Vinci” to compare the dialect, alphabet and orthography of Leonardo with the dialect, alphabet and orthography used by the author of the Voynich Manuscript. The quotes in the following discussion are from Richter’s preface and pages 1-4.(xxiv) The following comparisons will show that the Voynich Manuscript is probably Leonardo’s first codex.

• Leonardo was born near Vinci, a little town in Tuscany not far from Florence. His father was a notary in Florence. Leonardo’s formal education was limited, he was not taught Latin; his illegitimate birth prevented him, as an adult, from entering any professional occupation. He probably wrote using the same Italian dialect and alphabet as Dante and Florio’s 1611 Italian dictionary and therefore the letters “j”, “k”, “w”, “x” and “y” do not appear in his script. You will find “an ‘x’ similar to the letter ‘x’ in his notebooks, but it represents an abbreviation for ‘ver’. He sometimes used a ‘j’ in place of an ‘i’. In his drawings he included the letters ‘k’, ‘x’, and ‘y’. This extended 21-letter alphabet and the use of numbers in his figures enabled him to identify parts of the figure with the corresponding text.”
• I used the 21-letter AVA alphabet that excludes the letters “j”, “k”, “w”, “x” and “y”, and Florio’s 1611 Italian dictionary to decode anagrams representing the names of 111 of VM’s plants(xxv); i.e. the dialect of the Florentine poet Dante.
• Leonardo was left handed, he wrote in his notebooks from right to left using mirror image writing. Vasari observed with reference to Leonardo’s writing: “he wrote backwards, in crude characters, and with the left hand, so that anyone who is not practiced in reading them, cannot understand them.” Occasionally Leonardo wrote from “left to right in the normal way.”
• The VM is written from left to right. Most investigators have assumed that its author was right handed. In the article, The Voynich Manuscript’s Script and Code(xxvi), I show that the VM’s author was probably left handed.
• Leonardo’s mirror image script uses normal Roman letters with a few symbols representing abbreviations. These symbols are similar to symbols used in the Voynich Manuscript. His notebook show “no punctuation whatever to regulate the division and construction of sentences,” and no capital letters to indicate the beginning of a new sentence or the end of an old one.
• The VM is written using its own unique alphabet and many of the letters are represented by symbols. It has no punctuation marks, full stops or capital letters to indicate the beginning of a new sentence or the end of an old one.
• Richter stated that he corrected spelling errors when he translated the text from Leonardo’s notebooks. Leonardo often spelt words phonetically.
• I found that the spelling of some of the decoded plant names in the VM matched that given in Florio’s 1611 dictionary, while the spelling of other names only matched phonetically.
• Leonardo’s notebooks date from about 1489 until his death in 1519. Richter states that “within this space of time his handwriting altered so little that it is impossible to judge from it the date of any particular text.” If the Voynich Manuscript was written during his youth, his style of writing may have been set at an early age. There are no known examples of his handwriting prior to 1473 and apart from having additional loops and flourishes, it is similar to his writing in later years.
• Robert S. Brumbaugh’s son and others have observed the similarity the VM’s script to that of Leonardo.(xxvii)
• Richter noted that one of the difficulties he encountered translating Leonardo’s notebooks was: “Leonardo made use of an orthography peculiar to himself; he had a fashion of amalgamating several short words into one long one, e.g. leforme, ditutti = le forme di tutti. Richter considered that “these combinations were intentional and indicated Leonardo’s desire of substituting a sort of phonetic writing for the rules in general use.” His doubling of the letters as e.g. “chessia” for “che sia” and “essella” for “e sel la”, he “believed was clear evidence of what may be called the orthography of Leonardo da Vinci.”
• The anagrams used to decode the names some plants(xxviii) included, in addition to the plant’s name, a preposition, conjunction or definite or indefinite article, i.e. combination of two or three words or anagrams into one longer word or anagram, e.g. the anagram “atleebar” for “albatre e” translated as “straberry tree and”.
• Another orthographic peculiarity of Leonardo’s writing was “that he would quite arbitrarily divide a long word into two separate halves.”
• In the Voynich Manuscript some longer plant names were split into two or three consecutive words or anagrams, e.g. the anagram “borari chcia” for “barracchio” translated as “buttercup”.

Over the last 15–20 years I have studied Richter’s translation of the notebooks of Leonardo da Vinci, but I never appreciated how important his description of Leonardo’s peculiarities in orthography would become in helping to confirm that Leonardo may have written the Voynich Manuscript. All the orthographic peculiarities Richter found in Leonardo’s notebooks are also found in the Voynich Manuscript. In addition, the orthographical peculiarities found in Leonardo’s notebooks, Richter states, are not found in the writings of other 15th century Italian authors.

Leonardo’s notebooks were ignored for several centuries, mainly due to the difficulty in reading his mirror image writing. Over a century ago, Richter undertook the monumental task of translating the 5,000 pages from these books and papers that have survived. They show that Leonardo was not only one of the world’s greatest painters and draftsmen, but that his intellect and understanding of the physical world was centuries ahead of his time. The Voynich Manuscript was discovered in a monastery outside of Rome, just over 100 years ago. Since then all attempts to decipher the manuscript’s code have been unsuccessful. What I have accomplished by presumably identifying and decoding the names of 111 plants in the VM’s botanical section merely scratches the surface. It does show that the manuscript was probably written using Dante’s Italian dialect, anagrams as the code using the 21-letter AVA alphabet. The above discussion, showing the similar and unique orthographical writing styles in Leonardo’s notebooks and the Voynich Manuscript helps to establish Leonardo da Vinci as the author of this manuscript.