Voynich phonetics (June 2015 version) by Derek Vogt


Addendum from Derek, April 2016:

New version of the list… no new plants or astrological items to add, but /y/ is now included as a secondary interpretation for the letter ^r^ in a few cases, and the definitions of phonetic symbols are now collected near the top instead of given individually with their first applications in the tables, and a few old entries have some bits of new information added, like folios 27r & 28r. The slight shift in its overall appearance/style is a side effect of a switch from one file format to another (HTML to word processor).

The most significant difference now is the new section after the herbal & astrological tables but before the Notes: a few words that are not herbal or astrological names and, unlike them, should be in the authors’ native language, with much lower odds of being imported from another language coolpack 다운로드. And not only does each one of them have a plausible Romany connection, but some of them don’t work very well or even at all in any other language I know of except Romany.

The one that stands out the most to me is “ges” meaning “day”. It’s thoroughly out of the usual pattern for words meaning “day” in Indo-Iranian languages or any other Indo-European languages; so far off that, without knowing about it ahead of time, any grounds for thinking that the manuscript’s word for “day” sounded like “ges” would normally be grounds for thinking the language must not be Indo-European Windows 10 Educational Download. But there just happens to be one language in the family with that strange word, and it just happens to be the same word & language that were predicted from the manuscript.

[Here is the relevant new section]


Original post (June 2015):

Here (below) is an updated version of Derek Vogt’s provisional scheme for identifying elements of the Voynich manuscript, along with sound-sign correspondences 원펀맨 리메이크 다운로드. It is an update of his previous version which you can see here. I thought it was worth dedicating a new page to it, to make discussion easier.

Here are some notes about it which Derek posted as comments elsewhere:

“Entirely new entries since the last version include plants 4v, 7r, 9r, 17v, 25v, and 43r, the blue text for 2v, and stars 16 and 21 g402. One of those includes EVA-q, so that symbol has been added to the table of symbols and their sound values at the top. Plant 15r has been retracted.

Some “old” entries have had just a few new cognates added, or had old cognates that were only available as transliterations added now in their native alphabets, including plants 5v, 6r, 14v, 16v, 20r, and 39r, 2v’s red text, and stars 7, 26, 37, 38, 45, and 53.

Old notes 9 and 13, on the sounds /ɣ/ and /ʕ/, have been merged so that note 9 alone now address both sounds. A new note on the use of ^h^ after plosives has been added as #13, so later notes are not affected.”


Derek has also offered some useful comments on the process as a whole:

“The purpose of working out the phonetic system is to find a relationship between Voynichese and a known language or family of languages which can be used as a model for translation (presuming it has any relatives that are known) 스타듀밸리 한글 다운로드.

One way that could happen is by comparison of their phoneme inventories. Closely related languages usually use similar sets of phonemes, and very different sets of phonemes are used by more distantly related or unrelated languages. Fortunately, Voynichese’s inventory seems to be rather unusual, so a good match would stand out from the crowd. Even if we never found a perfect match, at least this would be a way of narrowing down the list of candidates to the ones that come closest 퍼피 구조대 다운로드. Wikipedia’s articles on languages usually have lists/tables of phonemes, like this one, but not all, and some of the languages/families in the same geographic region as Voynichese don’t even have Wikipedia articles. So information on the phoneme inventories of some candidate languages/families would need to be found somewhere else.

Another way would be by matching some of the rest of the words in the text, the ones that form sentences in the author’s language, not just things that are likely to come from other languages like the plant & constellation names 짱구는 못말려 극장판 쿵푸 라면대란 자막 다운로드. This requires some sign of what some of the other words should mean first. For example, one of the plants in my list produces an oil that induces vomiting, so another word on its page that isn’t used on most other pages should be equivalent to “vomit”, and others that are somewhat more common would probably equate to “induce” and “oil/juice/extract”. The problem with this is that you can only try one language or language family at a time and there are a lot of candidates, many of which don’t appear in online translators Windows Subscribers.

So again, the same issue applies: information on some of the candidate languages would need to be found somewhere else. We need a way to look up translations, or at least sound inventories, for lots of languages, even the obscure little ones for which such information is hard to get.”

Vogt - june 2015 scheme - part 1b

Vogt - june 2015 scheme - part 2b


For the plants, the sources I’m listing below are for the botanical identifications. The plants’ names in various languages are from the standard online translators I named above, plus Wikipedia in a few cases and a general web search in the case of sweet basil 직진게임 다운로드. All plant identifications except four were based on only the drawings. The exceptions are that Darren Worley also used some cultural information for the olive identification and my three were influenced by my phonetic reading of their names.

For the two astrological items on 68r3, the sources I name below are for both the names and the idea of what astrological entities the Voynich drawings represented 3m post-it downloads.

For all of the labeled stars on 68r1 & 68r2, the sources I name below are for the names/words for the astrological entities or the things they represent. The idea of which astrological entity to associate with each individual star on these pages is from me, using the phonetic system I had already developed using plants and Frederik de Wit’s Planisphere Celeste.


Bax’s website: anonymous Finnish biologist:
11v Daphne mezereum (no English name)
11v genus Salix, osiers and sallows
16v-a Nigella damascena or arvensis, breadweeds, fennels
16v-b Illicium verum, bedian star-anise
17v genus Dioscorea, yams, sweet potatoes
20r genus Satureja, savories
93v Bryonia cretica or dioica, bryonies
95v1 Fumaria officinalis, fumewort, earthsmoke

Bax’s website: anonymous Finnish biologist & Steve D:
04r genus Linum, flaxes

Bax’s website: Darren Worley:
01v-a-1 Olea oleaster, wild-olive

Bax’s website: Deyan:
66v-a-1 genus Aloe, aloes

Bax’s website: Hans Adler & Daniel Myers:
21r Portulaca oleracea, purslane

Bax’s website: Hans Adler, Jan M, Peter Ole Kvint, & Labyrinth:
06r genus Papaver, poppies

(reported by Stephen Bax in 2014):
16r Juniperus oxycedrus, sharp juniper

Edith Sherwood:
02r genus Centaurea, knapweeds
03v genus Helleborus, hellebores
05v Malva officinalis, marsh mallow
14v genus Stachys, betonies
29v Nigella sativa, black caroway, black cumin, Roman coriander
41v Coriandrum sativum, coriander, cilantro, Chinese parsley

Ellie Velinska:
08r genus Cucumis, cucumbers and cantaloupes
09r Vitis vinifera, common grape
17r Artemisia dracunculus, tarragon
37r genus Chenopodium, goosefoots
43r genus Diospyros, persimmons
55v Allium ursinum, wild garlic, bear’s garlic, bear leek
87v-b genus Pistacia, pistachios, terebinths

Ethel Voynich:
02v genus Nymphaea, water-lilies
04v Ipomoea aquatica, water-spinach
25v genus Plantago, fleaworts, psylliums, “plantains”
28r genus Rumex, sorrels, docks
38v genus Cynara, artichokes
39r genus Colchicum, meadow-saffrons, autumn-crucuses, colchicums

Ethel Voynich & Theodore Petersen:
06v Ricinus communis, castor oil plant

Theodore Petersen:
01v-b-1 Solanum melongena, eggplant

Derek Vogt (plants):
24v genus Aquilaria or Gyrinops, agar, agarwood, gaharu
27r-a Ocimum basilicum, sweet basil
31r genus Croton (tiglium?), rushfoils

Derek Vogt (astrology, using an online translator):
68r1-01&02 spring equinox; Aries; “beginning” and “end” in Hebrew
68r1-07 Pegasus & Equuleus; “horses” in Indo-Iranian languages
68r1-21 Ophiuchus; “snake-charmer” in Marathi
68r1-27 part of Serpens; same word ^artw,ãtw^ in 27, 32, 47, & 48
68r2-32 part of Cetus; same word ^artw,ãtw^ in 27, 32, 47, & 48
68r2-36 Eridanus/Fluvius; “stream/current/flow” in Arabic, Persian, & Urdu
68r2-38 Corona; “wheel/circle/spinning” in Indo-Iranian languages
68r2-45 Lupus “wolf” in Indo-Iranian languages
68r2-47 part of Argo; same word ^artw,ãtw^ in 27, 32, 47, & 48
68r2-48 part of Hydra; same word ^artw,ãtw^ in 27, 32, 47, & 48
68r2-53 Crater “cup/bowl/pitcher/jug/urn” in Indo-Iranian languages

Bax’s website: Darren Worley:
68r2-53 Crater “cup/bowl/pitcher/jug/urn” in Hebrew

René Zandbergen:
68r3-55 Pleiades (in Taurus)

Bax’s website: Marco Ponzi:
68r1-16 Alpha Draconis (Abʰaya)
68r3-55½ dragon of the eclipse (Gauzahar)
•For the first, he linked to a paper by R N Iyengar (Raja Ramanna Fellow at Jain University in Bangalore); for the second, he linked to multiple sources, so I can’t pick one to name here.

68r1-22 Beta Virginis (Auva, strangely unmentioned by Richard Hinckley Allen)

Richard Hinckley Allen:
all other star & constellation names on 68r1 & 68r2, including “Kurtos” for star 01 (alternative to Hebrew for “end” in the same label) and “Argo” for parts of the labels for stars 44 & 47








  1. Aljosa Puzar

    Did anyone (hint: Derek Vogt) look into 1910 description of Domari language? https://en.wikipedia.org/wiki/Domari_language

    Also, Romany-Armenian mixture of Lomavren language came up in on-line comments.

    Could these “close cousins” of Romany in Semitic and Caucasian etc. environments be the closest “suspects” according to Derek’s phonetics?

    All this is beyond fascinating. I am following your discussion with excitement.

  2. MarcoP

    While the Voynich ms doesn’t include a fully anthropomorphic root similar to the typical illustrations of Mandrake (Mandragora), it does include several vaguely anthropomorphic roots. Possibly, the one which comes closer to a full human figure is the plant at the center of f89r2.

    f89r has been connected with Mandrake by user Frater210 on this forum. I subscribe to his statement: “The only thing that is throwing me off is the differing leaves”.

    The plant appears to be labeled EVA:okol.shol.dy.

    According to Derek Vogt’s extension of Stephen’s phonetics, this could read:

    The Indian Materia Medica (Nadkarni, 1926) provides a few names that seem comparable with the hypothetical reading:

    Arabic: Ḍastam Haryah – the first part of the name could be close to the second part of the Voynichese label xaš-tm;

    Tamil: Katav-jate – somehow comparable with the whole label; according to Derek, an initial consonant before ‘a’ is often dropped, so the Voynichese akaš-xaš-tn(m) should be compared with atav-jate;

    Sanskrit: Lakshamana – considering the dropping of the initial “L”, this seems like a good match at least for the first part of the Voynichese label.

  3. Derek Vogt

    I’ve been considering pruning the least likely examples from my word list, thinking about ways for parts of my ideas to be wrong, and checking what would be left of the big picture if certain parts were wrong. I’ve found that it’s easier to find excuses to delete plant names from the list than star names, for more than one reason:
    •Coincidental false positives are an omnipresent threat in historical linguistics even when the sounds are all identified correctly, even more so when there’s any doubt about the sounds… and isolated individual words could be explained away as routine linguistic coincidences more easily than a collection of words that all belong together, such as points on a sky map.
    •There are so many kinds of plants out there that there’s always room for doubt about which one a particular drawing represented and whether it could be another one (possibly even one you didn’t think of at all), but a star in a particular location in a sky map can only be one of a few nearby stars in the real sky, allowing for a bit of possible distortion in the map.
    •There’s usually room for doubt about whether a particular word on a plant page is even the plant’s name at all, but there’s no such doubt that the star labels on f68r’s sky map are labels for the stars.
    •Pages with a bunch of text, not just pure labels, are more likely to have more features related to grammar or sentence structure, like suffixes, prefixes, and proclitics, which can be mistaken for part of a root word. For example, if the ^b^ at the beginning of many sentences’/paragraphs’ first words is a proclitic indicating the beginning of a new statement, then matching such a word with a plant name that starts with /b/ is not reliable, and this isn’t a problem for the star labels because they don’t have it.

    So even though this phonetic system was built up mainly using the plant pages at first, I’ve been looking at the sky map as the stronger support for it… which leads to the question: using only the stars and no plants at all, as if 100% of the plant names we thought we had were invalid for various reasons or we just hadn’t even tried it with them, would we still have built up the same phonetic system, or would we have gotten a different one, or nothing coherent at all? If the stars led us to a different one, that one would seem more likely to be right than the current one.

    Fortunately, f68r alone still does give us a pretty solid place to start from nothing: 11 EVA-cphocthy and 22 EVA-ocphy, because of their short lengths (four and three characters), positions in the sky map, and use of the same fairly uncommon Voynich letter (EVA-cph) in the same positions in the words as the [v] in similarly short Latin names for stars in roughly the same parts of the sky, Vega and Auva. Those would also immediately lead to 49 EVA-opocphor, because it’s the only other star using ^v^ and it also has EVA-p, which looks like it should be related, so it would have been relatively easy to spot the name El Bezze and read the Voynich name as ^ebeve-^ (^ẽbeve-^), with /v/ replacing the original /z/ (as if the language didn’t have /z/) and no written letter for the /L/ (already our first sign of the ^ẽ,ã^ phenomenon).

    So those words’ letters would have been our anchor letters, without an EVA-d (^t^) or EVA-k (^k^) in sight… whereas in real life ^k^ and ^t^ were among the first and ^v^ and ^b^ came along later. And that different starting point would also have us looking at EVA-o as primarily ^e^ instead of ^a^, and EVA-cth as /g/ instead of a fricatized counterpart. More significantly, our first impression of EVA-y would have been /a/, the thing at the end of “Vega” and “Auva”.

    The only way we could end up reading EVA-y as ^n,m^ instead of ^a^ would be to decide later that the interpretation as ^a^ was just wrong and ditch it (which would be easy because it is just a suffix in the two examples so far, but would still require evidence we don’t have yet in this scenario). Or we could find that the rest of the evidence in this new approach supports it and decide that the reading as ^n,m^ has been a mistake all along. And there are words on the page that could tip us toward the former, like 40 EVA-oydchy, 52 EVA-shchy, and 9 EVA-otydy, where that letter fits as ^n,m^ but not as ^a^, but only if the other letters in those words have the sounds we’ve already assigned to them so far. If the other letters turn out to have different sounds, then maybe /a/ fits in with them as part of a completely different set of names. We won’t know until we build up a system filling in the remaining missing letters properly.

    But there is a problem with trying to identify more letters’ sounds in the remaining star names: we should be going through words with only a single previously unknown letter as much as possible, but there’s not a single word like that left on this page; they all have two unknowns or more. For example, with 42 EVA-chodar, again skipping the suffix, it might seem clear that the first three symbols should be ^hes^ to match Greek “Hestia” and “Hesxara” if we already had ^he-^ or ^-es^; the consonant we already had would tell us what would fit the other spot. But at this point, we don’t have either; we have ^-e-^. And there are so many ways to fill in something like that that we could probably always come up with multiple wrong ones without some way to narrow it down.

    But I do see three ways we could have gotten past that ostacle.

    1. Take some letters from that “Taurus” label on f68r3. This breaks the “only use the sky map” rule, but, as was already noticed and used in real life , there’s only one cluster of seven stars, so it’s still a solid unique identification with an unmistakable label, not subject to the problems of identifying plants and finding their names in paragraphs. The initial ^t^ (a single consonant before the vowel being unlikely to correlate with the sequence /pl/ from “Pleiades”) would have unlocked several names in the map where the pair of blank spots we had were a ^t^ and something else, which would then unlock the other letters in those names: ^s^ in ^tẽs,tẽz^ for 29 Terazu, ^x^ in ^xt^ for 34 Xut, and so on. This would also lead us to the words where EVA-y appears in the middle, as part of the root word, instead of as an affix, and only works there as /n/ or /m/, not as /a/.

    2. Allow the use of the few most secure plant names like centaurea and ciliander/corantro instead of the draconian “no plants at all” rule. That would have added a few more letters to get the same star map domino effect going.

    3. Infer from ^v^ and ^b^ that gallows ligaturization could indicate fricatization, and thus that the second ligature in star 11 “Vega” is not /g/ but has shifted in spoken Voynichese from /g/ to its fricatized counterpart. That would mean that the un-fricatized plosive sound /g/ was represented not by the ligature in star 11 but by its non-ligature counterpart (EVA-t), and there are enough uses of that letter on the page that we would be able to test it and see that there are names for the right stars/constellations which generally have /g/ (or something close like /ʕ/) in the right spots. Not only would that give us /g/ and some other letters from those words, which would start the same kind of domino effect as above, but the newly corroborated theory of gallows letters would also give us a way to fill in the rest of the gallows pattern, plus the concept of EVA-ch as a neutral/unarticulated fricative, which means /h/ (or at least /ɦ/).

    The result of breaking that impasse in any of those three ways ends up the same: a phonetic system derived entirely (or, with #2, almost entirely) from stars rather than plants, which is mostly the same as the one we already have. The basic gallows letters and the meaning of ligaturaziation would be generally the same, as would the letters ^n,m^, ^t^, ^w^, ^h^, ^x^, and probably even ^o^. But there are a few differences…

    A. We would be thinking of EVA-o as primarily ^e^, with ^a^ as only a secondary option when it’s needed, instead of the other way around. Sometimes I’ve thought we should think of it that way anyway and it was more likely to sound like “e” than “a”, but for separate reasons; either way, it appears to correlate with both kinds of sound, as is typical of the aleps/alefs in most Semitic languages because these aren’t really two separate sounds in them anyway, so calling one the basic interpretation and the other just a backup doesn’t really matter.

    B. There would be no sign of what to do with EVA-S. It exists in the sky map, but only in two or three of the most troublesome words on it, none of which are clear about what sound belongs there (and one of which, 13, isn’t even clear about which letter is written). This letter-for-sound association was already the weakest case anyway; this alternative approach just brings it down from the weakest to practically nothing. Even coming up with the same idea that this alphabet was derived from the Syriac alphabet, and seeing this letter’s similarity with Syriac ade, would not really help us sound out any of the Voynich words this letter appears in on these pages, unless you accept the ^$on^ at the end of star 37 as matching Hebrew “$avaron” for “collar” (as in “the bull’s collar”), and I have serious doubts about that, with its missing “ava”. For all we’d know, this letter could be almost any other sound that some other letter doesn’t already have.

    C. The available examples of EVA-L would correlate mostly with /s/, not /ʃ/ (š), so the idea that it’s really split between the two might not have been apparent; we’d probably just stick with calling it /s/ by default and thinking of the few cases where it correlates with /ʃ/ instead as “close enough”.

    D. It would be a toss-up whether to think of EVA-y as more likely to sound like /n/ or /m/; our concept of it would be a vague idea of “some kind of nasal” instead of “[n] which occasionally seems to substitute for [m]”.

    E. We would probably be reading EVA-r like English “y” (which, like any other consonant in a system that’s inconsistent about vowels, can sometimes imply an unwritten following vowel). Examples of “y” fitting better than “r” on the star map include 16 ^aphoayᵃ^ for “Abʰaya”, 16 ^ãgayᵃ^ for “Alkaia”, 26 ^ãkxay^ for “Alḥai“, and 49 ^ẽbevey^ for “El Bezze”. This general impression would even remain the same if we included some plants, particularly ^kõoyatws^, where EVA-R correlates with the “y” or “i” in most versions of “coriander” while the preceding ^õo^ covers the “r” before it. Exceptions could (without invoking ^ẽ,ã^) include stars 14 ^eser,eʃer^ for “Ser, Sher, Shīr” and 38 ^akhar^ for “kekro, čakara, čaxra, črxs”. Others could go either way. It’s not clear whether these exceptions would have made us think the letter has a general equivalency with “r” just as well as with “y”, or that the letter is simply “y” and those “r”s had just gotten converted to that.

    • MarcoP

      Hi Derek,
      I would like to comment about the reliability of the f68r1/r2 sky maps as support for a phonetic system. The work you have done in this area is extremely valuable and the interpretation of the star labels you have so clearly presented is impressive. With the contribution of several users on this site, the maps have been interpreted as consistent with later printed planispheres, e.g. de Wit’s Planisphaeri Coeleste. I think this is a good explanation for the structure of the two images, but in my opinion there are a few aspects that might cause some skepticism in the general public:

      * All ancient Western planispheres included images of the constellations. I think the first star maps without constellation images were published by Piccolomini, in the second half of the XVI Century. Yet Piccolomini’s book only includes maps of small portions of the sky and stars are not labeled but only indexed with single letters. I think that planispheres representing the whole sky with no constellation images were first published in Europe by Argelander (XIX Century).

      * Planispheres without constellation “figures” appeared in China at a much earlier time (e.g. the Suzhou planisphere, XIII Century). But in those illustrations constellations were still represented as lines connecting stars. Moreover, ancient Chinese maps do not split the sky in two halves, but project the whole heaven on a single circular area.

      * Islamic globes that only represented stars (with no constellation images) are documented since the XII Century. Astrolabes also represent the relative position of individual stars (see for instance the Islamic astrolabe with 53 stars described in History of Cartography, Celestial Mapping, fig.2.8). Yet planispheres do not appear in Islamic manuscripts.

      * With the exception of purely decorative objects, ancient scientific planispheres and globes included some kind of reference system (a radial/circular grid, or at least an ecliptic circle with degree marks).

      Apparently, these two Voynich illustrations differ from all known medieval objects. If they are scientific images, they are highly innovative, in some respect closer to modern sky maps than images from other European manuscripts of the time. Yet the absence of any reference grid might indicate that the two illustrations are not scientifically accurate or (in the worst case) that they are not scientific at all (maybe they represent actual stars, but not as a precise depiction of the sky). Many of the possible parallels I have seen are of this decorative kind, but I am not aware of any convincingly close parallel (see attached images).

      If the Voynich illustrations are scientific in nature, how do we explain the existence of these exceptional planispheres? Do they belong to the same culture as that of the Voynichese language? Or are they the product of the Western contribution to the creation of the manuscript (as the images of the zodiac signs and the zoomorphic and anthropomorphic elements of the herbal)?
      Since the practice of representing the sky as a pair of circles was common in medieval European manuscripts and illustrations of the sky with stars only (no constellations) occurred in Islamic globes, a possibility is that an Islamic globe was used as a model to draw the two planispheres, translating the names from the language used on the globe (likely Arabic) into Voynichese. This process is of course speculative and better support for these ideas, or better explanations, would be needed.

      On the strictly linguistic side, we have a mix of star and constellation names of Arabic, Indian, Greek and Latin origin; as you commented in footnote 14 above, some of the proposed matching words might not be documented in an astronomical / astrological context. We looked for matches with more than a thousand objects (stars + constellations) which often had multiple names in each language: considering the different language families, this gives several thousands of candidate names for the 53 labeled stars in f68r1/r2. There easily are many tens of candidate names for each label.

      Is it possible to find a single artifact (globe, astrolabe or planisphere) that provides parallels for a significant amount of the proposed names (I am thinking of something like 30% of the star labels)? This kind of specific evidence in primary sources would make the support of this analysis much stronger, but I feel my linguistic and astronomic competence are too limited to attempt the search for parallels with a minimum of confidence.
      As Stephen proposed a while ago, it would be very useful to have the help of an expert in ancient astronomy. In my opinion, solving some of the problems connected with these illustrations and pointing to specific ancient textual and visual sources, the support that the sky maps provide would be made considerably stronger.

  4. Derek Vogt

    I wasn’t sure at first whether to put this here, or on the page about the Syriac alphabet connection, or on the Pelling-response page where general updates seemed to be requested (first by Pelling to Bax and then by Bax to anyone everyone in response to Pelling)… but posting in response to anything started by Pelling would just be feeding a troll, and this is part 1 of something where part 2 really doesn’t really relate to the Syriac anymore…

    I’ve made a video summarizing the big general picture I’ve developed over a bunch of separate posts here. I think everything in it is stuff I’ve posted before, but it’s been scattered, not all brought together in one place. Actually, I still needed to split it into two because it got much longer than I realized it would even after I stripped out a bunch of stuff, and this is only part 1. Hopefully the second will be ready within the next couple of weeks. I expect the second part to be both shorter (this is almost 40 minutes alone, based on which I now guess part 2 will be 20-30!) and more interesting to more people, but it also really depends on the setup in the first part here:


    • MarcoP

      Thank you Derek, it’s certainly helpful to have your research presented in a single place!
      I am looking forward for part 2!

    • MarcoP

      Hello Derek,
      about the interesting comparison with other alphabets in your video, I would like to mention that Mary D’Imperio suggested a possible similarity of the Voynich alphabet with Latin abbreviations.
      User Davidsch on the voynich.ninja forum has discussed a 1470 ca German manuscript (WMS 55, Analytica Priora, by Johann Lindner of Monchburg, Leipzig). On f93, the manuscript presents a complex diagram of the functioning of human perception (I cannot get a direct link, you have to search “analytica priora” in the box at the right, then download the high resolution image).

      I attach a comparison of some Voynich characters with symbols from this single page. My comments about each symbol, left to right:

      EVA:o – Lower case Latin “o”.

      EVA:a – Lower case Latin “a”.

      EVA:e – Lower case Latin “e” (in this manuscript, “e” is usually not closed as a loop; “c” is of course also similar this Voynich character).

      EVA:t – An abbreviation that usually occurs without the upper dash, but not in this manuscript (for more examples, see this blog post about Cod. Sang. 839). I think the abbreviation could read “quis” or “quem”, when it appears without the upper dash. Here it could be something similar but more complex, maybe “quamvis”.

      EVA:k – A rather common abbreviation, also mentioned by D’Imperio. It occurs as a suffix, reading “-tis”.

      EVA:p – Lower case Latin “p”.

      EVA:ch – A ligature between two characters: “ti”.

      EVA:g – An abbreviation possibly reading “dis”.

      EVA:y – One of the most common Latin abbreviations, reading “cum-” or “con-” as a prefix and “-us” as a suffix. Of course, the same symbol was used for number 9 (when using the “Arabic” symbols).

      EVA:iin – I am unsure about the reading of this suffix. It could simply be “-um” or maybe “-orum”.

      EVA:r – An abbreviation reading “re”. “r” is also similar, but without the loop.

      EVA: d – Lower case Latin “d”. In other manuscripts (e.g. Cod. Sang. 839 mentioned above), this letter also has an upper loop, but never as symmetrical as that of the Voynich symbol. Of course, number “8” in medieval manuscripts was written using exactly the same symbol as the Voynichese letter.

      My overall impression is that these symbols are graphically very similar to the Voynich letters, closer then the various Eastern examples you discuss. In my opinion, this can be the effect of at least two different phenomena:

      1. A scribe trained in a Latin script copying from a manuscript written in a different alphabet. The original alphabet (possibly related with Syriac, as you discuss) was distorted to fit the symbols the scribe was familiar with.

      2. The Voynich alphabet was actually designed on the basis of this kind of Latin script. In this case, the evidence you and Stephen have gathered suggests that the symbols were not matched to the closest sounds. Since this seems difficult to explain, I guess that this possibility should be discarded.

      • Derek Vogt

        The most likely explanation to me is #3: no connection at all, just superficial coincidence.

        However, on #2, the fact that they don’t match up with the sounds in the system we’ve been using here isn’t necessarily a problem. They could just be part of a different phonetic system which has little or nothing in common with this one, so whoever proposes any real connections with the Latin symbols here would simply be saying this phonetic system is wrong and proposing a new to replace it.

        The real problem for #2 is that abbreviations subtract from the already-small number of symbols that would be available to assign individual sounds to. With such a limited set of symbols to work with, anything that reduces it even more makes it less likely to fit a real language.

        • MarcoP

          Hello Derek,
          In the attachment I add examples from Cod. Sang. 839 (mentioned on voynichanalysis.wordpress.com) and Cod. Sang. 610 (discussed by J.K.Petersen). The last four from Cod Sang 610 are digits.

          I am undecided about the possibility that these similarities are coincidental. On the other hand, I think it is clear that the symbols that look like Latin abbreviations cannot be read as such: if there is a relation, it could be the reuse of existing symbos with a different function. This can be seen in most cases in which medieval alphabets were invented: Glagolitic, Hildegards’ Lingua Ignota, Foxton’s 1408 cipher alphabet all assign different sounds to previously existing symbols and add some newly created characters (many of which are alterations of previously existing symbols).

        • Darren Worley

          Derek – thanks for a great piece of work. Its really good to be able to hear the spoken sounds to accompany your proposed identifications. I look forward to watching part 2.

          I think Marco makes a valid comment about the possible use of Latin abbreviations. It seems plausible that a traveller might invent an alphabet based on Syriac and Latin symbols to transcribe a newly encountered language.

          As Marco rightly says there are several examples of medieval invented alphabets. There are also instances of travelling missionaries in the medieval period documenting newly-found languages eg. the Codex Cumanicus so extending this idea by introducing familiar symbols seems possible.

          The historic record also describes Europeans travelling through Syriac-speaking regions in this period. As well as the famous Marco Polo (and his companions), another was the Italian Dominican, Ricoldo da Montecroce (who I’ve mentioned before) he was the first Westerner to make contact with the Mandaeans whilst travelling through Mesopotamia in the 13th-century. In earlier times, during the Crusades when there was a continuous European presence (for example in Acre) there would have been contact between Europeans and Syriac speakers.

          Furthermore, there is another class of Latin abbreviations, that I’ve seen that might be relevant, and that is the use of number-prefixes to represent dates and text.

          eg. 8 to mean “octo” etc.

          October was written as 8bris [octo]
          September as 7bris [septem]
          November as 9bris [novem]
          December as 10bris [decem]

          Presumably this idea could be extended to use the other numbers too, but I haven’t seen any actual examples.

          I first came across this type of Latin abbreviation in a church inscription in Sestri Levante near Genoa. Its describes the consecration of a church on October 11, 1118 by Pope Gelasius II. I’m not sure how widespread the use of this shorthand would have been.

    • Derek,
      Have you any data on knowledge of Syriac in fifteenth century Europe? I realise that Serapion’s text, so often cited as an authority in the illustrated herbals, was originally written in Syriac, but the version known in Europe came from one which was evidently in Spanish, or in Hebrew, (scholars differ and the evidence is scanty) and which was translated by the Toledo school, being issued under Gerard of Cremona’s name. So it’s unlikely that any knowledge of Syriac accompanied it.

      The Nestorians, of course, retained Syriac as their liturgical language, and its near relative, Aramaic, remained a living language in Europe because it was, with Hebrew, the language of Jewish scholarship and commentary.

      Otherwise, I haven’t found much record of it being known – have you?

      • Derek Vogt

        No, nor have I looked. I have no theory about, or way to determine, where the Voynich authors were located when they wrote the manuscript, or where they or their teachers or their teachers’ teachers and so on were when they acquired or invented the script. I have figured all along that the simplest explanation would be that, if it was derived from another, they were most likely to have been in the territory where that other system was most common, when they adopted it. Then they could have moved on afterward, and the manuscript could have been written before or after the move.

        So, based on the idea that the writing system is derived from the Syriac one, I expect that they got it where Syriac writing was common, which would generally be the Middle East, not Europe. This is compatible with actually writing the manuscript either in the Middle East or in Europe.

    • Derek Vogt
      • Matthew Dean

        I like that your explication follows the evidence instead of shoehorning it into a pre-judged conclusion; predictive and replicable work is what is needed to bridge the gap between internet treasure hunt and culturally meaningful scholarship. Thanks for all of the effort put into both of the videos and the many open and reasonable posts here.

        Hoping some Romani linguists may pick this up (I’ve written one already), I’m also moved to ponder: in what contexts was the effort to compile and translate herbals and zodiacal-medical treatises in Europe or the Eastern Mediterranean worthwhile? Who was the intended audience in established cases? Where might there have been a Roma (or descendant culture of a shared common ancestor with Roma language) court or patron or school with resources to commission and utilize such a compendium?

        And more immediately: gathering (or perhaps hoping) that the manuscript as-is reflects both translations of pre-existing treatises and some original (local) meditations, can your sound inventory (once longer passages are rendered) find us any match to the existing texts behind other herbals we’ve examined, just as the iconographers have been seeking? Perhaps if the visuals are as innovative (idiosyncratic) as they seem to preclude perfect matches in a known inventory, the textual content or format (more easily parsed when sounded out) without those visual distractions may disclose a close correlation.

        I’m personally looking next at Johan van Ewsum, Job Ludolf, and Jacob Bryant to see if any almanackish vocabulary was indexed in some of the earliest Romani transcriptions.


        • Derek Vogt

          I figure the best way to find out that kind of cultural contextual stuff is to read the book, so the linguistic stuff must come first. So my main concerns right now are obstacles that could still get in the way of reading it even if I’m right about its phonetic system and origin. (Obviously my being wrong about the whole thing would also create a bit of an obstacle; sadly, it might even become a disincentive for others to try to come up with a better one later even though the method is still sound.)

          1. Although my ideas, particularly on the last slide of the second video, include solutions to most of the statistical oddities that have made people (generally non-linguists) think it doesn’t seem to be any real language, there is at least one left: why some gallows letters are several times more common than others. If they represented sounds, that would mean some sounds are represented many times more often than others. Why would that happen? It’s likely to make ANY possible theory of a real Voynich language (not just mine) look wrong, because languages don’t normally have that kind of phoneme distribution and nobody’s ever seen (or can imagine a reason for) literature that goes out of its way to avoid words containing sounds that would be used more in speech.

          2. As I mentioned in the second video near the end but didn’t lay out an argument for or against to save time, it’s possible that the Voynich language is, to one extent or another, a combination of two: the more easily visible Indic component, and something else non-Indic. (I would suspect Hebrew.) If another language has enough influence, then it would be nearly impossible to read even for an expert in Indic languages, which might even lead to the conclusion that the Indic/Romany theory was wrong and nobody even trying it again even if it’s right.

          3. I fully expect half of the plant names to be wrong. I could even list which ones I have greater or less confidence in and why, but in general, the point is that trying to find isolated cognates based on similar sounds & meanings is very prone to false positives, and even more so when either the sounds or the meanings are still in any doubt. At this point the likelihood that my overall theory is generally right relies mostly on bulk: take away dozens of words and dozens still remain. But even if most parts of the theory are right, taking away the wrong combination of individual examples would still undermine specific parts of the theory. For example, take away the basis for letter ^$^ as /ᵗs/, and you take away other words I thought I had found using it as /ᵗs/. Take away the basis for the de-repetition rule, and you take away other words I thought I had found that rely on de-repetition. If one or more of the parts of the theory get knocked out like that, then what’s left, even if it’s right, will appear less worth even trying to work on after that.

          * * *

          About who would have invested the resources to make something like this in such a language: I am not at all convinced that it needs to have been sponsored by the kind of large institution/house that people seem to have in mind. The materials might be materials that are often associated with books commissioned by noble houses or universities, but the artwork does not look like it was not drawn by a professional, so why would the usual kind of investor have skimped on the cost of hiring a reputable artist? Maybe because it was for someone who couldn’t pay the full price that usually went into otherwise comparable books.

          This makes me think of someone outside the usual scholarly circles of the time, trying to be like them by independently producing something like what they would produce. It reminds me of something in feudal Japan called “merchant swords”. Swords were worn in public as part of the daily routine, but seldom unsheathed & used. But only the nobility were allowed to have full-length swords. So relatively wealthy merchants would buy something with a full-size sheath & handle but a shorter blade under the legal limit that doesn’t fill the sheath. (This also influenced the proportions of the stereotypical image of ninja swords later.)

      • Alex

        hey derek, i started to research the voynich phenomenon only tonight and came across your two videos on youtube. thank you very much: they were very inspiring. im a linguist based in berlin (germany), not so much into phonetics, but your line of argumentation seems convincing (as far as i can judge the matter), maybe ill check up on the details sometime in the future. my questions are: 1. have you really stopped working on your account – as indicated in the end of your second video – waiting for others to complement it now? why not go on with it and try and decipher also the contents yourself? 2. have you tried to get in touch with people who know romani? i know some linguists in hungary who most probably know romani speaking scholars too. best, alex

        • Derek Vogt

          1. It’s not a matter of choosing to stop doing anything. It’s a matter of what is even possible to do anymore; anything else I can do at this point would be hardly any different from doing nothing anyway. If any more individual words present themselves to me because something about the context in which they appear tells me something about what they probably mean so I know what to look for in a translation dictionary, like Voynich ^eges^ a few months ago as a cognate of Romany “(e) ges” for “(the) day”, that would be interesting, but, on the scale of learning a language or translating a mystery manuscript, it’s really no progress at all. Real progress would be reading Voynich lines like ^btwr as xtw-r -abhootn tws htws har xopxara rw— twsar w— abwkwr gtw– akoatn shtn skar aoo$owrn twr xotw— aga tw— t*wr wr^, recognizing that as a sentence in an Indic language with funny spelling, and knowing what it means. I can’t do that. Only someone who knows a language that’s similar enough to this one could. (And even if I’m right about Voynichese being a cousin of Romany and other Indic languages, that doesn’t necessarily mean they remained similar enough overall for this to go smoothly even for an expert in those languages; English, for example, would be a big problem for an expert in other Germanic languages who didn’t already know English, because it’s full of quirky deviations from the group and imports from some other group.)

          2. I don’t know any such people. I know they exist, but one of them would need to take this seriously enough to bother with it, and I can’t make them do that. And the Voynich Manuscript has a reputation as something that only kooks and quacks waste their time on and claim to have achieved anything with, not anyone with a worthwhile scholarly approach, so I know that the wait to be taken seriously by the right strangers is not likely to be a short one.

          * * *

          I am thinking of doing one more short video on the idea that the language might have had a substantial amount of input from Hebrew, like English has from French, which I mentioned near the end of the second video but didn’t explain… and possibly including, or just making another separate video about, a collection of leftover bits about individual words that don’t affect the big picture much, and/or some discussion of the possibility that parts of my theory are wrong even if the rest isn’t… but that video or those videos would not be an attempt to begin translating the whole thing. That’s just not within the abilities of someone who doesn’t know the right other language(s).

          • Zodiac


            You have a list of rules to go from Voynich -> phonetics. Presumably with a Romani dictionary you could do the same. would it not be possible to computationally apply some phonetic similarity measure (e.g., Google “phonetic similarity of words” for some papers on the subject) to score each of your transliterated words against the dictionary, take the top score and see if the result makes any sense? Would that method have any merit?

          • Neticis

            Derek, can you provide complete table from EVA writing to your proposal? Then I could do automatic translation of pages from http://www.voynich.nu/ (e.g. http://www.voynich.nu/q01/f001v_tr.txt) and then we would have some samples to work with other (translation) tools.

            • Derek Vogt

              If the idea is to use “Find & Replace All” to replace EVA characters with my phonetic characters throughout the whole book, then that’s something I’ve already put some thought into before, because the substitutions would need to be made in the right order. For example, if you replace all EVA-d with ^t^ before replacing EVA-t with ^g^, then you’ll end up with a document in which two different original letters are both “t” and there’s no way to separate them anymore; replacing “t” with ^g^ then will replace them all, including the ones that were originally EVA-d instead of EVA-t. So just listing the letter-to-sound relationships isn’t good enough; a plan needs to be laid out.

              The simplest way to distinguish between symbols that don’t have sounds in my system yet & haven’t been replaced, from those that do and have, would be with font format changes in the word processor you’re doing this in. For example, if you start by making everything red and bold, and then each time you do a find-&-replace for a particular letter you also make the new one non-bold and black, the end result will be a document mostly full of plain black text with isolated bold red characters peppered throughout it, and those will be the unchanged EVA symbols.

              Better yet, you could use a font face that actually looks like what’s written in the manuscript. If you download the EVA font from René Z’s website and save it in your C:\Windows\Fonts folder or the non-Windows equivalent, then the EVA option will appear in the list of choices in every program you have that has a font selection, right along with the usual ones like Arial and Roman and Courier. Save the whole Voynich EVA transcription file in that font, and it will all look like Voynich text on your screen, and can have the usual changes applied like bold, colors, and sizes. Then while you’re using find-&-replace to change EVA to my phonetics, just make sure it replaces the red Voynich-font letters with black ones in your usual font. The result then will be a document full of mostly plain black letters in my phonetic system with just a few red ones standing out which still look the way they actually look in the manuscript.

              If the only program you have access to is a plain text editor so you can’t change colors or font faces like that, you would need some other way to distinguish the letters in my phonetic system from the letters that I have no sound for but are in the EVA. For example, there are EVA-b,g,j,v, and EVA-x, which are rare and haven’t been assigned sounds in my system, but there are also letters that I assign the sounds /b/, /g/, /j/, /v/, and /x/ to. And there are others which technically don’t contradict my system in any way and technically could be left unchanged, but then they would look like they were meant to be phonetic (EVA-q,u,z). Both of these groups would need to be gotten out of the way by conversion to something else that could not be mistaken for the letters that I assign sounds to; I picked these mostly based on what these symbols look like plus their lack of other uses in EVA or my phonetics:

              Replace all EVA-b with 6
              Replace all EVA-g with 9 or %
              Replace all EVA-j with 8
              Replace all EVA-v with ^
              Replace all EVA-x with π (Greek pi)
              Replace all EVA-q with + or 4
              Replace all EVA-u with @
              Replace all EVA-z with κ (Greek kappa, because it looks like a small ^k^)

              …But, like I said, that’s only necessary in a plain text editor. In a word processor, you could leave those as EVA letters, in the Voynich font and red or such, and then they’d stand out from the rest anyway.

              Now, on to the bulk of the book; the letters I’ve actually assigned sounds to! Remember what I said about putting the changes in the right order; for example, don’t convert anything else to “p” until the original “p” have all been converted to something else and are gone…

              p→b… then cBh→v… then f→p… then cPh→f

              t→g… then cGh→j or ǧ… then d→t


              But then we run into another complication. My system only has sounds for gallows letters standing alone or ligaturized with EVA-ch, such as EVA-p and EVA-cPh. But sometimes, depending on which transcription you start with, either the EVA-c or the EVA-h seems to be lost or replaced in some examples with something else like EVA-i or EVA-a. In my interpretation, no gallows combinations like EVA-i_h or EVA-c_a and so on exist, and these are simply EVA-c_h which got squished or smudged. So after each EVA-c_h has been replaced, its stray counterparts with the first or last part missing or replaced would need to be found and fixed one by one, by searching for fragments like EVA-c_, EVA-_h, EVA-i_, and EVA-_i (with B, P, G, or K in the open spot) to see what’s there in each case and correct it with ^v^, ^f^, ^j^, or ^č^. Fortunately, if the file you’re starting with used capitalization the way it’s meant to be used, so the central letter is capitalized if it gets a crossbar and lowercase if it doesn’t, you should be able to find all of these in just a total of four searches by just searching for capital letters. Then, once you’ve replaced all of those misspelled gallows ligatures, you can go on…

              a→w… then o→a… then e→o


              Sh→x… then s→Ṣ… then L→s

              I’ve been using “$” for EVA-s, but some EVA transcriptions might contain that symbol for another function, so I would switch to the more conventional way of handling Semitic emphatic consonants: the dot below the letter. Otherwise, the two uses of “$” would not be distinct from each other and could interfere with each other in later searches you might do.


              n→•… then y→n

              I have been using hyphens for both EVA-i and EVA-n just as place-holders because I have no sounds for them, but I’d switch to these two different sizes of dots here for two reasons: this way preserves the difference between the two characters just in case it might matter later, and some programs merge consecutive hyphens into a longer hyphen so you can’t see how many there were.

              Finally, although I have used EVA-m the same way as EVA-r a few times, it would be better, in a full replacement scheme for the whole book, to keep them separate, in case the difference turns out to matter later. Even sticking to the idea that it is still a form of the letter ^r^, you could keep it distinct and easy to find & change later by replacing EVA-m with a different version of “r”, such as:

              m→ʀ or ṟ

              EVA-r and EVA-k represent ^r^ and ^k^, so they can just be left as they are, unless you need to replace them just to make the font & color match the other replaced letters.

              • Zodiac

                All these steps are pretty easy to implement programmatically (no manual find/replace). I’ll try and have a go, did you use Takahashi transcription with capitalisation from http://voynich.freie-literatur.de/index.php?show=extractor ? Or is there a better source?

                • Derek Vogt

                  I’ve never used any transcription. When I look at Voynich text, I don’t first see it in a non-phonetic system like EVA and then convert to phonetics; I literally first see it phonetically, so converting it from that to EVA, not the other way around, is where there’s an extra step of work for me if I want to use EVA. And, other than some searches at voynichese.com, I’ve only worked with single words or short word strings at a time, not the whole book, so I just look at images of the pages from voynichese.com or https://www.jasondavies.com/voynich/ and type what I see myself, in either my phonetics or EVA.

                  I’m aware of only one specific difference between different transcriptions: the downloadable EVA font uses “b” for a symbol which voynichese.com calls “n”. The bottom part of it is curved like EVA-e instead of straight like EVA-i and the standard form of EVA-n, but the only way to find examples of it in a search there is by searching for “en”. Based on that, I presume that the differences are always over rare cases that are hard to figure out what to do with, like the last letter of Bax star 13 near the left edge of f68r1 (otnSh? otnr? otns?). In that case, there could be some transcriptions out there which, for example, already treat all gallows ligatures as “c_h” instead of allowing oddballs like “c_i”, which would simplify the conversion process, but I don’t know which ones are which.

                  The use of capitalization to add the horizontal connecting bar for gallows ligatures and EVA-s is something that the downloadable EVA font does, but I don’t know which Voynich transcribers have and haven’t used it. If you want to use that font, you’ll need to use capitalization to make your own Voynich writing look right.

                  • Derek, I’d be happy to implement it myself, but as I’m busy with other things, I’ll try to get some student, who will implement these rules as his academic work. I’ll try to keep you notified if I’ll have some new information.

  5. Derek Vogt

    [Note: this has been added to the main Post above, with the image. I leave it here so people can respond to it. SB]

    New version of the list… no new plants or astrological items to add, but /y/ is now included as a secondary interpretation for the letter ^r^ in a few cases, and the definitions of phonetic symbols are now collected near the top instead of given individually with their first applications in the tables, and a few old entries have some bits of new information added, like folios 27r & 28r. The slight shift in its overall appearance/style is a side effect of a switch from one file format to another (HTML to word processor).

    The most significant difference now is the new section after the herbal & astrological tables but before the Notes: a few words that are not herbal or astrological names and, unlike them, should be in the authors’ native language, with much lower odds of being imported from another language. And not only does each one of them have a plausible Romany connection, but some of them don’t work very well or even at all in any other language I know of except Romany.

    The one that stands out the most to me is “ges” meaning “day”. It’s thoroughly out of the usual pattern for words meaning “day” in Indo-Iranian languages or any other Indo-European languages; so far off that, without knowing about it ahead of time, any grounds for thinking that the manuscript’s word for “day” sounded like “ges” would normally be grounds for thinking the language must not be Indo-European. But there just happens to be one language in the family with that strange word, and it just happens to be the same word & language that were predicted from the manuscript.

    • Peter

      my key give me this

    • Christian

      Could the german word „gestern“ (yesterday) stand for “the day behind”
      ges – (ar)ter ?
      I am thinking about a synonymous usage of “day” and “sun” (as well “night” and “moon”) . On f93r the second word begins with “eges-”. Could this word be the name of the plant, even though the real sunflower came to europe after 1550 ?

      • Derek Vogt

        German “gestern” and English “yester-” in “yesterday” both come from Proto-Germanic “ɣestra”. The “ɣ” (the voiced counterpart for modern German “ch”) hardened into a plosive in German, but palatalized and softened into a glide in English. Proto-Germanic dates back well over two thousand years. So, German didn’t get this word from Voynichese.

        That alone doesn’t eliminate the possibility of Voynichese getting it from German more recently instead, but that would require that the speakers of Voynichese somehow interpreted it as a compound word, assigned separate meanings for each piece, and adopted each piece for general use instead of just as a quote of a foreign term. I don’t think that ever happens.

    • MarcoP

      Hello Derek,
      I see that EVA:okol “e ges” also appears in the Sun circle in f68r2
      This clearly fits well with your hypothesis.

      Could the Moon circle contain a word for “night”?

      • Derek Vogt

        That’s ^akaš^ or ^ekes^; ^eges^ would be EVA-otol, with two loops on top instead of a loop and a spike. It could have the same origin if there was some kind of devoicing effect, but in this case I don’t know why that would happen (unlike the case of a word starting with ^eks^, where the dropped vowel and resulting adjacency with ^s^ would explain it).

        It did make me notice a funny thing about f68r1, though. Star 18 is also EVA-otol, which I’ve already equated Al Gau and Al Qaus, for a long time, from Arabic القوس “alqws”, meaning “the bow” of Sagittarius. And I still do think that’s what it is; in a system with so few letters, and some letters forced into double duty, some coincidental cases of different words with the same spelling are bound to happen. But it does bring a second coincidence with it: Sagittarius was apparently also known by a few names related to the sun, as Richard Hinckley-Allen put it, “because our Archer was a type of the rising sun” (whatever that means).

        But that’s already the background behind the name “χut”, apparently meaning “Dayspring” (presumably a poetic expression for “dawn”), which is essentially identical to the label on star 34, ^xtwr^. And I can’t picture the same sky map having two star names with the same meaning in different languages for two stars side by side, so reading star 18 as related to “the day” would be redundant.

        • MarcoP

          Thank you for correcting me, Derek!

          Ellie Velinska pointed out another detail about EVA otol: it seems to label one of the “elements” at the top of f77r.
          Back in 2014, Stephen proposed that EVA:okos could mean “akash”. It’s difficult to understand if EVA:otol, okol, okos are related or not. The conceptual relationships between bows, days, sun and heaven contribute to make the matter both complex and interesting.

          I guess that Allen could be referring to the fact that the Sun was represented as an archer, with arrows symbolizing rays of light. In Western art, the bow is an attribute of the Sun personified as Apollo. The symbolism is obvious, and Allen seems to be aware of analogous Eastern personifications.

  6. TC

    I was wondering if you considered using your key to transliterate whole paragraphs, and see if anything can be made from it. For example, if I use your key on f1v (skipping the first line with those tricky gallows), the remaining lines would read something like this if I haven’t transcribed falsely. Does this ring any linguistic bells?

    • Derek Vogt

      That’s the idea. But first, we would need to know what language it is and have it read by somebody who knows that language. (And that only works if it is a language somebody somewhere knows!) Til then, it’s just like hearing someone speak a language that’s unknown to the listener(s): gibberish, no matter how clearly pronounced & heard. I have a theory about the language’s identity, but even that still creates challenges because it’s obscure; it would have been extinct for centuries, the nearest modern relatives are known only to a small number of people most of whom aren’t linguists and aren’t looking at this subject, and it could even be a creole mixed with something else that’s unrelated so even an expert in one contributing language wouldn’t recognize the parts that came from the other one. It’s still a long path ahead; all I can hope for with this phonetic reconstruction is to have given us a few steps along the way.

  7. MarcoP

    Hello Derek,
    it seems that the Ethiopic Ge’ez script adds a horizontal line to consonants to derive new sounds, including the derivation of “v” from “b” and “č” from “t”. According to this paper, the “butterfly bar” was introduced in the XVI Century, which would exclude direct relevance for the Voynich manuscript. Still I find it a potentially interesting parallel for your discussion of gallow ligatures.
    Another possible coincidence is the shape of the character for “ṣ”.

    One of the languages written in this script (Amharic) has been previously mentioned by Darren. Both Amharic and Ge’ez were mentioned by Stephen in his 2014 paper.

    • Stephen Bax

      I love those two scripts, which I have studied briefly. The older one (for Ge’ez) was apparently invented deliberately and systematically, instead of being derived from another script, and is accordingly very systematic, in the sense that each base consonant has a small line or stroke or similar added to it to make a consonant-vowel union. So ‘ba’, ‘be’, ‘bo’ have the same basic shape but will small vowel additions…. very neat.

      Alas, the Voynich script shows no sign of such a regular patterning running through all the script. A pity, as we could have deciphered it before breakfast.

  8. Hi Derek, can I ask what evidence you feel there is for [l] being the sound /s, $/? I’m intrigued by this sound assignment, and would like to know what you feel supports it.

    • Derek Vogt

      Your question mixes two letters: EVA-L and EVA-S.

      I believe EVA-L represented /š/ (English “sh”) or /s/ or both and sometimes replaced /z/ in words imported from other languages. The direct evidence is simply the 24 words in my list where I found it in the same position as /š/ or /s/ or /z/ in other languages’ words for (apparently) the same things.

      I believe EVA-S (which I transliterate as ^$^) represents a different sound which English doesn’t have. It’s an affricate, meaning it begins like a plosive and ends like a fricative, like English “ch” and “j”, but with the tongue in the same position as for /s/. That means the closest thing English has would be the sequence “ts”, but treated as one sound (just like we don’t think of “ch” as “tsh” or “j” as “dzh”). It is represented by a single letter in some languages, such as German & Italian [z], various Slavic languages’ [ц], and Hebrew [צ]. For its Voynich letter, the direct evidence is simply the 8 words in my list where I found it in the same position as /$/, /t/, /s/, or /j/ in other languages’ words for (apparently) the same things. The case for this one is weaker than for the one above, because there are fewer examples and their correlated sounds are less consistent.

      In both cases, there is also indirect evidence if we indulge in a bit of inference: their resemblances to letters for the same or very similar sounds in the Syriac alphabet, which the Voynich alphabet resembles in its other letters as well, as if it were derived from it (The similarities are close enough that the differences can be almost eliminated with just a few simple rules for how to change the letters.) In fact, based on the Voynich-Syriac similarities for ^w^, ^o^, ^n^, ^r^, ^t^, and ^k^ as given by Professor Bax a couple of years ago, the sounds of the Voynich letters ^š^, ^$^, and ^h^ were predicted based on their similarities to the Syriac letters šin, ade, and he before examples of words using them were even found. (And ^p^ from Syriac pe could have been, too, if we had looked at the graphical conversion rules closely enough at the time.)

      • Darren Worley

        Hi Derek – this isn’t the only evidence linking Voynichese with a Syriac-related alphabet.

        On f57v, I previously reported a symbol, the halqa, which is a comprised of a triangular shape, a circle and the vertical dash. It can be seen on the diagram at the 12 o’clock position in the 3rd ring from the centre (and in each of the other quarters in the diagram).

        It is reversed from its normal orientation, as it normally appears as the Mandaic alphabet as the letter “ga”.

  9. MarcoP

    Hi Derek, I am too ignorant to reliably tell, but it seems to me that gallows featuralism, recently discussed by Emma May Smith, has interesting analogies with your analysis of gallows letters:
    If so, this convergence of ideas seems promising to me.

    • Derek Vogt

      That blog entry boils down to two points: one that matches what I was thinking when I derived the phonetic relationships for the gallows letters, and one that I was unaware of but fits my phonetic thinking quite well.

      The first and more basic point is that the gallows characters form such a neat, perfect 2×2×2=8 pattern, with no gaps where a letter you might predict from the others is missing or any extra letters with one more feature from outside that pattern. From this, as that author and I both thought, the inference that they were set up that way deliberately and there would be logical meaning to it (which is usually absent from natural unplanned alphabet evolution) made sense and was just waiting for phonetic evidence for or against it.

      The second, which I hadn’t thought of or examined, was about how common each gallows letter is and what part of a word or page they are found in. And each of the rules for gallows letter distribution that she pointed out fits with the idea that the features indicate aspects of a sound, because types or categories of sounds in any real language are not found randomly distributed anywhere & everywhere, but predisposed to show up in certain places relative to each other.

      The most interesting part of that second point to me was that the plain gallows letters are about 10 times as common as their “pedistal” counterparts. In my phonetic system, that means the letters for plosives are about 10 times as common as their ligatures with ^h^ as a fricatizer. In Semitic alphabets, there are several letters for which the original sounds were plosives, but which started taking on affricate or fricative sounds part-time. And extra marks that can be added to specify one of a letter’s two possible sounds are not always used, even when the letter does represent the sound that the added mark would specify, so the original form of the letter remains ambiguous, possibly representing either its original sound or the newer derived one. For example, you can never tell, without using an additional outside source of information, whether an unmarked ב represents /b/ (the original sound) or /v/ (the later derived one). In Voynich context, that would mean that although the sounds of what I call ligatures (^f,v,č,ǧ^) are secure and should never represent their plosive counterpart sounds, the plain original letters (^p,b,k,g^) might be expected to sometimes represent their fricatized counterpart sounds. Indeed, there’s already some sign of that, with some Voynich words seeming to use a plain letter where the fricatized sound would be expected, especially with ^g^ and /ǧ/ (English “j”). And that kind of ambiguity between an original letter’s original and derived sounds, while a derived letter to specify just one of those sounds is inconsistently applied, would make the original letter more common and the derived letter less common, exactly as that blog author said they are.

    • Darren Worley

      The fact that the gallows characters form such a neat pattern seems too to deliberate to have occurred in a natural language. Are natural languages ever written so logically? This isn’t intended as a rhetorical question – I’d be interested to know of any examples. In my (limited) experience written languages are characterised by exceptions and non-conformance to orderly rules. Could this be read as a hint that Voynichese could be an invented transcription of a natural language?

      I favour the idea that Voynichese is some kind of proto-Yiddish, possibly Zarphatic (Judeo-French). I was surprised to learn during some background reading on this topic, that Roger Bacon (1214 – c. 1292) devised his own transcription method for writing Hebrew or Judeo-French (since this is what would have been the Jewish dialect spoken in Northern France, when he lived there). [Yes, that Roger Bacon, the one that was thought to have been the VM author by Jan Marek Marci back in 1666, and was also repeated Wilfrid Voynich…]

      Bacon wrote a couple of texts on this topic “Cambridge Hebrew Grammar” and “Linguarum Cognitio”, in the 13th-century, and his approach is described in a couple of recent papers: ROGER BACON’S POLYGLOT ALPHABETS by Horst Weinstock (1992) and ROGER BACON AND PHONETIC TRANSLITERATION also by Horst Weinstock (1991).

      Although Bacon might be too early to be responsible for the VM, it seems quite plausible that a later natural philosopher/polymath might have duplicated his method, or re-invented a transcription method for Hebrew or Judeo-French.

      • Darren, that is the undertone of my thinking: the script was designed. Maybe it was invented wholecloth, maybe it was a deliberation alteration of something already existing.

        • Derek Vogt

          Given the histories of other alphabets, the most likely origin for the Voynich alphabet is mostly unguided evolution for most of the alphabet overall, modified with deliberate tweaks to only the ones that seemed to need it. As a general rule, it’s the most likely because it’s how most other writing systems got the way they are, and because the patterned look of the gallows letters does not extend to the non-gallows letters.

          I have a more specific theory of that type, but, even if these details are wrong, that still would not affect that general rule.

          My theory is that they adopted the Syriac alphabet, with the following changes:
          •The letter ^a^ might be from Mandaic alap instead of Syriac waw. Or not. It works either way. It behaves more like an alap than a waw, but if we figure it came from the latter, then there’s no need to involve a second alphabet for just one letter.
          •Reversal of direction of writing, which brought with it the left-right reversals of some letters or parts of letters (which single-handedly explains most of the differences between Voynich and Syriac letters for the same sounds; they’re mostly nearly identical when this one change is accounted for)
          •Replacement of gamal and bet with symbols derived from their voiceless counterparts, kap and pe, thus creating the basic set of four gallows letters
          •Ligaturization of those letters with ^h^ to convert them from plosives to affricates or fricatives
          •Use of gallows letters with ^h^ after them as digraphs for aspirated plosives
          •Addition of a mark above ^h^ to create a symbol for the fricative ^x^

      • Darren Worley

        Thanks Emma May – I appreciate the reply.

        I just wanted to add that the paper ROGER BACON’S POLYGLOT ALPHABETS by Horst Weinstock can be download freely from the Web. I haven’t read the other one yet.

      • Derek Vogt

        Historically, the pattern in letter development is that the occasional changes which need to be made deliberately are done with some logic behind them, and the ones that just evolutionarily drift on their own due to imperfect handwriting and replacements of writing technology are more random, even prone to occasionally randomly creating the imperfect appearance of a partial logical pattern.

        For example, if you knew the sounds of our “B” and “P” and “D” but not much else, you might think a straight vertical line with a half-circle or two stuck to it was the general form for plosives, and you might even think there’s another pattern that the half-circles run top-to-bottom for voiced ones and leave half of the vertical line bare for unvoiced ones. With slightly looser rules for the system, you could even think “C” and “G” are related to the above, with the half-circle oriented the other way and the vertical line lost or altered, but voicing/voicelessness still indicated by a difference in the lower right section. But you’d be out of luck trying to find a symbol for the sound of “T” to complete the pattern with another letter featuring a half-circle. You might even be driven to the erroneous conclusion that “R” was your missing “T”.

        What really happened with those letters is that the Greeks & Romans had a tendency to round off prominent corners in Phoenician letters, and Phoenician letters had been simplified from pictographs, not designed based on types of sound. But there are two little non-coincidences in there: “G” was derived from “C” by conscious decision to distinguish two similar sounds that previously shared a letter, and “R” was modified from an older form that looked more like “P” but had to be distinguished from it as “P” evolved from “Π” toward its current form (again by rounding off a prominent corner).

        Then there’s another layer after that: you could think, based on what really did happen with C/G and P/R, that when letters that are otherwise too similar need to have a distinction created, an extra mark is added in the bottom area of the letter, and/or possibly to the right. And there are a couple more examples like that: “J” really was created by adding a hook to “I”, and “U” really was created by rounding off the corner of “V”. But this would also lead to you think that there are relationships like that between “O” and “Q”, “E” and “F”, “X” and “Y”, and possibly “N” and “M”, which there aren’t.

        The Arabic alphabet is even more full of this kind of stuff. All of those dots you see above & below the main bodies of the letters aren’t diacritical marks. They’re parts of the letters, and the only way to distinguish otherwise identical letters. But historically, they came in two different layers. The older ones got added to distinguish between previously different-looking letters, representing completely unrelated sounds, that were just coincidentally converging on each other because people’s handwriting was getting sloppy (like our modern “P” looking like the older form of “R” so they were hard to tell apart for a while; just imagine if issues like that one applied to about half of our letters!). So the dots, although they were phonetically random, served the purpose of fixing that problem. But later, as writers saw an increasing need to distinguish between similar but just slightly different sounds, they created more letters, this time by adding (or, in one case, removing) dots to the old letters with sounds that were most similar to the new sounds (such as the letters for /d/ and /t/ getting dots to convert them to /ð/ and /θ/). And when people speaking other languages like Persian and Urdu got the alphabet, they did more of the same for their sounds that Arabic didn’t have: fiddling with dots (and a slash in one case) on the nearest thing Arabic did have (such as the letters for /b/ and /ǧ/ (“j”) getting dots to convert them to /p/ and /č/ (“ch”)). So now the flurry of dots includes a mix of cases that were deliberately and logically planned with specific phonetic meanings for the dots, and others that weren’t and mean nothing except “these letters look too similar otherwise”.

        Korean hangul is the only writing system I know of in which such patterns extend throughout the whole system and really do mean what they look like they should mean, and the reason is that it didn’t evolve; it seems to be the only one, in standard use by a modern language & country, that was entirely invented. And even that one has had some blemishes sneak in after a few centuries of sound shifts without corresponding adjustments to the written forms.

      • D.N. O'Donovan

        Thank you for posting this. I’d been hunting for my own post in which I spoke about Bacon and his Hebrew studies, but couldn’t for the life of me recall the name of that author. Much frustration is relieved.


  10. Koen Gheuens

    Hi Derek

    I have been studying another possible label identified by Diane O’Donovan as a piece of cloth died Indigo: https://voynichimagery.wordpress.com/2013/12/03/im-certain/

    Too bad the cube of indigo isn’t labelled (I take the label to belong to the plant to its left), but the ‘cloth’ certainly is.

    (I tried to add an image but couldn’t figure it out, apparently it appeared (twice) under my post.
    I’ve been looking into indigo names for a while to see if I can make sense of the bottom label. It reads EVA: okoroeey.

    An indigo-root seems unlikely. That was used by mainly Europeans to refer to the Indian origin of the material, and Voynichese is not European *dodges crossbow bolt*. Also, I don’t see a credible way to read “indigo” in the label.

    It is also difficult to read the Arabic word “annil”.

    Last remained Diane’s suggestion of Thai “Khrām”. She referred to Old Thai script, which gave me a headache. It surprised me, however, that it is easy to read the label as “(a)Karaun”. I agree with the general feeling that Voynichese doesn’t differentiate between “m” and “n”. I also think that initial “o” is often, if not always, best ignored as it’s probably an article. The “a” in “Khrām” is long, so gets replaced by Voynichese “au” diphthong. Add to that the fact that Voynichese does not allow the “kr” cluster, word-initially and elsewhere, and thus needs to add an “a” in between, and we have a perfect match: K(a)raun/m = “Khrām”.

    • Derek Vogt

      Why is the cloth sample cut to that complicated shape?

      • There is an alternative explanation as well. What we see is a small block among fragments of herbs. Something similar happens in at least two herbals: Casanatense 459 (Tacuinum Sanitatus / Historia Plantarum) and Sloane 4016 (Tractatus de Herbis), both part of the same herbal tradition and approximately of the same time as the Voynich MS. Here, what is represented is the herb ‘Papirus’, in the first case as a stack of papers (3D) and in the second as a single sheet (2D – rectangle).

      • I found the following possible explanation for the shape in another post on Diane’s blog:

        “One of my very few certainties with regard to the written text is that the following detail shows (top) indigo, sold as it was as a pressed block with a stitched cloth cover and (below) indigo-dyed cloth. I think it significant that the cloth is depicted in a way resembling flags and canopies’ depiction in southern India. ”

        I agree that one can argue about the bottom drawing being a piece of fabric, but it is definitely *something* dyed indigo. A blue cube is natural indigo. Even today it is still traditionally sold like that in some places.

        By the way, I think I was mistaken about the cube label belonging to the plant on its left. So the cube label could also be interpreted in an indigo-related way, but I don’t know how (yet).

      • D.N. O'Donovan

        Derek (re your comment of Jan 23rd 12:41)

        That shape is conventional in southern India (and perhaps elsewhere) for flags and pavillion covers. I suppose it’s an alternative to the boring sort of fringe.

        May I also say that I wouldn’t have presumed to suggest Thai or any other language – not my field. I simply noticed that the block was one of the very few examples of a closed rectangular form, realised the maker had no choice but to depict it so, and then read in the historical documents that this is how indigo was prepared for sale – as a block stitched into a white cloth cover.

        Since the forms for indigo are so few, and Thai was one exception to the usual “nil’ or ‘indigo’ forms, I illustrated it. But the work here is all Koens’. I deserve no credit for suggesting Thai; all I did was illustrate its script.

    • Ellie Velinska

      I think the ‘O’ at the beginning of the VMs words is a nil. I base this opinion on the writing on the last page – oladabas= nil+laudābās (Latin for praise). The author liked word-play so it makes sense to come up with 0 as nil. The picture looks like crown so if I have to make-up words, I would go with nil+coron+9 (9 is standard Italian abbreviation at the end of the words). So my vote is for corona – Latin for crown 🙂
      The crown is symbol for costmary (Herba Sancta Maria in the old herbals)

      • Interesting observation. Do you mean nil because it shouldn’t be pronounced? But then what could be its purpose? To have a null character at the beginning of some words? Or do you mean as “should always be pronounce as nil”, in which case Voynichese has a whole lot of words starting with nil 🙂
        Unless you read it as “not” or “no”, in which case this plant would be labelled “no crown”, which is just silly.

        I like your crown example though. But if I let my imagination go wild, I’d say it rather looks like a ship with four masts 🙂 But as far as I can see four-masted galleons were post-Voynich.

        • Ellie Velinska

          Hi Koen. I mean the o at the beginning of the words is nil (void). Also some words are probably split into two strings – this again is also based on the marginalia on the last page (which I believe to be by the author). Words like carcere and Maria are split. If I understand correctly, statistically the VMs words are shorter than those in average European natural languages. Maybe some words are divided into two or more.
          It is my opinion that the VMs was written in the 1490s – at the time Trithemius was publishing his Steganographia. The ciphers in it contain some very short messages disguised in large bulks of symbols. He extended single letters to be written as words. Trithemius did tutor physicians like Agrippa and Paracelsus. I think the VMs case is something like that – German or Swiss-born physician who was festering in the writing environment influenced by Trithemius in the late 15th -early 16th century.

          • Darren Worley

            Hi Ellie – I think there’s a lot of merit in these ideas. I’ve been thinking along these lines too, I’d suggest that the VM is the product of a figure like the Lombardian/Italian philosopher, astrologer and professor of medicine Pietro d’Abano (c.1257 – 1316) [Peter of Abano, or Peter of Padua], a later follower or close-contemporary.

            I was surprised to read that Abano’s Latin translation of Abraham ibn Ezra astrological works were translated back into Arabic and European vernaculars. [ref: Thorndike, A history of Magic and Experimental Science” vol2]

            I think the VM originates in the border regions between Germany and Italy and was compiled/copied slightly earlier than you suggest, closer to 1400. Abano was from this area, and his earliest works were first printed here long after his death.

            Pietro d’Abano knew Arabic and Greek (translating many scientific texts) and possibly Hebrew. He would also have presumably spoken his native Lombardian language and possibly Old French too. He was known to have lived in Greece, Sardinia, Constantinople and Paris during his lifetime. (He also knew and interviewed Marco Polo).

            I’ve previously written elsewhere that I think the VM is the product of someone who seems to be attempting to reconcile Jewish/Old Testament theology with the pre-scientific ideas of the Middle Ages, and reading more about him, this does appear to be the case.

            A short biography available here says as much : [He] worked towards a synthesis of medieval, classical, Arabic and Jewish philosophy. A longer biography is available in Thorndike’s “A History of Magic and Experimental Science” vol2 pp. 874-947. (Its available online too.)

            Abano wrote or translated many texts, and several appear to have parallels with themes found in the VM – for example : the medical application of plants, on bathing, on medical astrology, on the Astrolabe etc.

            Here is an incomplete list:

            DE VENENIS – first pub. Padua, 1473 : a work on plants and poisons
            DE BALNEIS – first pub. Venice, 1553 : a work on bathing
            DE MOTU OCTAVAE SPHAERAE – first pub. 1988 : Astronomy – Motion of the Eighth Sphere
            DE IMAGINIBUS – first pub. Venice 1488 : on the astrolabe
            PROBLEMATA DI ARISTOTELE – first pub. Mantua, 1475 : Problems of Aristotle
            DE NATURA MEDICA DI DIOSCORIDE – first pub. Colle, 1478 : Translation of Dioscorides Natura Medica

            Later magic and occult works were credited to him, but might not have actually been by him. For example, Agrippa’s Fourth Book contains sections attributed to Abano.

            For me, knowing that medieval Latin European scientific works were translated into Arabic (and other European scripts), is quite enlightening, as I had thought that it had always been the other way around.

    • MarcoP

      A couple of herbal illustrations of Larkspur (Delphinium)


      Indian names: kheerga davai, lande-kaown, tagara

  11. Derek Vogt

    A funny thing about the letter ^r^ came up while I was reviewing my list to see how often it’s used in each of two ways:

    1. As part of a suffix ^ar^ or ^wr^, with no corresponding part in the apparent cognates, like ^xtwr^ for Xut (*34 Sagittarius)

    2. As part of a root word, corresponding to /r/ in apparent cognates

    I had recently (coincidentally) looked at examples of type 1 so much that I had started to wonder how many of type 2 there really were. And this is important for the accuracy of the reconstruction, because only type 2, not type 1, can tell us anything about the letter’s sound: the fewer of type 2 there are, the weaker the case is for the idea that the letter ^r^ represented /r/. (In fact, there are more examples of /r/ correlating with ^õo^ or ^õa^ than with ^r^… and still more with ^ã^ than with those!) It turns out not to be a problem; although how to score them isn’t so simple in some cases, there are roughly a dozen of each type, and the type 2s just hadn’t had my attention as much lately.

    But while I was confirming that, another half-dozen uses of the letter ^r^ fell into a third pattern: they are in positions where the apparent cognates have “i” or “y”. Those are essentially the same thing, just manifested in one case as a vowel and in the other case as a consonant (same category as “r” and “w”: approximants/glides). I’ll generally interpret these Voyich cases as the latter for two reasons. First, that fits together with /r/ better than a vowel would as two sounds that a single letter could represent (both approximants/glides done with the tip of the tongue). Second, there’s a vowel after it in the cognates in most of these examples, and a vowel “i” followed by another vowel forces the “y” glide as a transition between them anyway, whether it’s written separately or not. So, when this letter is at the end of a Voynich word, it would seem to have implied an unwritten subsequent vowel, which is written in the cognates. (Or, that final vowel was dropped when the words were imported to Voynichese from other languages.)

    *16 Alpha Draconis
    ^aphoar^→^aphoay^ (implied “aphoaya”)?

    *19 Beta Leonis
    ^ãgar^→^ãgay^ (implied “ãgaya”)?

    f5v marsh mallow
    ^kahoar^→^kahoay,kahoai^ (implied “kahoaya”)?

    *26 Alpha Serpentis
    “Alḥai” in “Unuk-Alḥai

    f41v coriander (cilantro)

    f6v castor oil plant
    kharvay (-xrwʕ)

    And not only does reading ^r^ as /y/ make those matches all work better than they do with /r/ there, but it even makes a new one possible. Star 31 should be Aquarius, but the only words I could suggest might be connected to ^hõaras^ before were Greek names with some extra sounds in the middle which lacked clear separate counterparts in the Voynich word: Hudroxóos, Hudor, and Xúsis húdatos. (Keep in mind the tendency for ^h^ to sometimes correlate with /x/ or /k/.) But using ^y^ for ^r^ gets us closer to the Latin name than to any of those…

    *31 Aquarius

  12. Derek Vogt

    For the foreseeable future, the table isn’t likely to change much, because I don’t know where many more changes to it would come from, so here is the latest form. The most recent thing prompting this is the addition of possible cognates for ^ãtw,artw^ in some of the star labels. Other changes include:

    ▶Four new plant identifications from Peter (11v, 13v, 22v, and 95v); two contradict earlier identifications which still appear just before or after the new ones
    ▶Two additional astrological identifications from folio 68r3 (55½, identified by Marco Ponzi, and 55 from Stephen Bax’s paper last year, which I didn’t address before but add now because 55½ relates to it)
    ▶New possible cognates for plants 2v, 3v, 5v, 27r, 31r, 43r, & 93v, and stars 8 & 36, fitting in with the cognates that were already there
    ▶Deletion of some cognates or apparently false cognates that I decided weren’t contributing anything to the theory or its presentation for various reasons
    ▶Original forms, in their native alphabets, for a bunch of words scattered all over the table that had previously only been available already transliterated into the Latin alphabet (mostly the Arabic star names); stragglers that I still can’t find in their original forms yet appear in rectangular brackets now
    ▶Improvements & corrections on scattered small phonetic details, especially with Indic vowels and a couple of Greek word-endings, but with no practical effect on whether or not the apparent conates are true cognates or the identifications are correct
    ▶Various small formatting changes (punctuation, spacing, colors) for internal consistency & clarity
    ▶A shorter “Notes” section at the end; the one about what EVA-n is not doesn’t seem important anymore, and the information in some others is now presented elsewhere on the page in other forms; the remaining items’ numbers have changed as a result

    * * *

    The added cognates for 27r and 93v are the latest things I’ve found in trying to find cognates for those words, but I am not as convinced of them as I am of most other entries in the table. I just leave them in there for now because they’re interesting and the closest I can get so far.

    In the case of 93v, it requires supposing that there was a a typographical error: a diacritical mark that was in the original words getting missed/dropped. In the Devanāgarī alphabet, a short /a/ is assumed after every consonant by default, so adding a diacritic is necessary to specify anything else such as /ī/ instead of /a/. Thus, the “ba” (ब) as reported at the plant-name website I’ve been using is what would result if the diacritic were lost from “bī” (बी). Although I can’t find any sign of a Sanskrit or Hindi word equivalent to “baja” in reference to plants, there are plants called “bīja”, and I have seen larger discrepancies than this in other alphabet-to-alphabet conversions before. So the idea fits together rather well, but I just don’t like posting a theory that depends on postulating otherwise unindicated errors in my sources.

    For 27r, ^k$ar^ seems to fit “kasar”, which shows up as an alternative name for basil on English botanical websites, but I still can’t find any good sign of where English got that word from. I have found that the syllabic writing system for ancient Mycenaean, Linear B, referred to basil with symbols that we transliterate as “qa-si-re-u”, but there are three catches with the idea that that’s related to “kasar”. First, the same symbol “qa”, was used for both /kʷa/ and /gʷa/, with no way to tell which was meant without using external clues in each individual case. Second, the same trouble exists for “re”, which could have been either /re/ or /le/. And there is good reason to think both of those actually go the wrong way, making this word /gʷasileu/ instead of /kʷasireu/ in Mycenaean: a well-establish sound shift in ancient Hellenic languages converts Proto-Indo-European /gʷ/ to /b/ and gives us “basileu”, the obvious ancestor of the word “basil”. Third, even if Mycenaean “gʷasileu” was related to a word with a /k,kʷ/ and an /r/, that word would need to have survived with little change for many centuries before either Voynichese or English could pick it up, without leaving any other trace I can find in any language. Or it would need to have been in English all along since it was Proto-Germanic; Proto-Indo-European /gʷ/ does tend to end up as /k/ in PG, but I don’t know how old the word’s oldest known occurrence is. Still, I haven’t found anything better for it, just its use in modern English websites.

  13. Derek Vogt

    I have a theory about spoken Voynichese’s identity. Until yesterday, I did not consider the Indic language family a good candidate, because they follow a phonetic template which is very different from the Voynich phonetics I’ve been using, and even if we were to try to rearrange the Voynich sound assignments, it has just too many phonemes for Voynichese’s limited letter set to cover. In order for a language to get from the Indic phonetic starting point to the current model of Voynich phonetics, it would have needed to make all of the following changes:

    1. Loss of retroflexes (ʂ-ɳ-ɽ-ɖ-ʈ-ɖʰ-ʈʰ)
    2. Loss of extra variations on “n” (ñ-ɳ-ŋ)
    3. Loss of second type of “r” or conversion to ^õa^ & ^õa^
    4. Loss of vowel length distinction
    5. Loss of any extra types of “L” if there originally were any extras
    6. Gain of “$”
    7. Loss of /d,dʰ,ɖ,ɖʰ/
    8. Gain of /x/
    9. Merger of /a/ and /e/, or at least allowance of a single letter for both
    10. Merger of /s/ and /š/, or at least allowance of a single letter for both
    11. Loss of /m/, merger with /n/, or at least allowance of a single letter for both
    12. Loss of either aspirated or unaspirated “t”, or the distinction between them
    13. Loss of “i,y”? (Merger with ^a,e^?)
    14. Loss of even a single distinct “L”

    That’s just too much change to say a language went through without some more direct signs of it.

    But now I’m aware of an Indic language (or small set of more closely related ones) which deviated from the standard Indic phonetic pattern, substantially reducing those requirements: Romani. It is already known to have done 1-6, apparently did something at least comparable to 7, and still fits other theories I’ve set out before which would also explain 8-10. Also, 11 resembles one way that those sounds are handled in the Brahmic alphabets (as I’ll describe below), which means they could both reflect the same underlying conceptual equivalency between those sounds among native Indic language speakers which would have allowed them to share a letter. Even 12, while still lacking a specific explanation that I know of, would certainly be understandable as part of the big picture, given whatever was going on with the other sounds near it in 1, 3, 6, and 7.

    That only leaves 13 and 14 as pure Voynichisms that have nothing to do with anything that I can connect with Romani. I think 2 changes are less of a stretch of the imagination than 14. 🙂

    Romani also happens to be the only Indic language which we already know migrated from India to Europe, sometime between about 1000 (based on timing of historical sound shifts other Indic languages) and 1542 (the earliest known written attestation of Romani). Along the way, it would have gone through areas where the locals spoke various languages including Persian, Arabic, Hebrew, and Greek, and wrote with various alphabets including Syriac.

    A few more-specific notes on the numbered list above:

    7. This was something I wasn’t even looking for because I had no hope of finding an explanation for a language with letters for /k,g,p,b,t/ but not “d”, but, after I made the Romani connection, a bonus came from it: I noticed a pattern that fits this phonetic oddity in a brief list of Romani words and their origins. The ones that came from Sanskrit or Persian and had a “d” in them have lost it in Romani. There are Romani words in that list with a “d”, but they came from Ossetian (in the Caucasus mountains) or Greek. This is the pattern we would see if, first, the language did something to its “d” either at home in India or at least by the time it got through Persia/Iran, and then, it started picking it back up again when it moved on just a bit farther west. Obviously, that list of words is not comprehensive, so I can’t say Proto-Romani totally ditched “d” rather than just losing certain instances of it, but I can say the pattern looks like something unusual did happen concerning “d” at that time in its history. And that would also happen to be when the language was exposed to the Syriac alphabet.

    8. Indic languages generally don’t have /x/, and Romani doesn’t either, but, during the migration, it would have been surrounded mostly by other languages that did, and there does appear to have been an unconventional relationship between /k/, /x/, and /h/ in Voynichese that I’ve commented on before. It appears as if one of two things happened, one option being a simple k→x→h sound shift chain followed by the loss of the middle step in modern Romani. The other would be that, when adopting words with /x/, the language converted many of them to /h/, but not all, and the mixing of those two phonemes also dragged in /k/ for a few of its own conversions to /x/ or /h/, before the dust settled with just /k/ and /h/ left again in modern Romani. This temporary instability around /x/ would also explain why the Voynich letter sequences ^bx^, ^gx^, ^px^, and ^kx^ have a similar, although milder, distribution compared to that of their ^h^-based counterparts, ^bh^, ^gh^, ^ph^, and ^kh^, which I take as aspirated plosives; it’s what you’d get if the people used ^h^ to indicate aspiration but secondarily saw ^x^ and ^h^ as two forms of the same thing, so ^x^ would also seem appropriate for aspiration. (See here for what I wrote about this before when I hadn’t thought of the Romani connection yet.)

    9. It’s common in Semitic languages to have just three vowel phonemes, one of which covers what speakers of other languages with more vowels would count as two or more separate sounds, such as /a/ and /e/. This is why, for example, some Arabic words & names get transliterated as beginning with “El” as well as “Al”, where both the [a] and the [e] correlate with the Arabic letter alif. When such an alphabet is adopted for a language with more vowel sounds than that, the people adopting it could tinker with the writing system to come up with a way to distinguish the sounds in writing, but sometimes, as in Persian and Urdu, they just accept that one letter has both sounds in different cases. So it would fit the historical trend perfectly for another Indo-Iranian language to do the same thing as those, adopting a version of a Semitic alphabet and ending up with a letter carrying two vowel phonemes. No merger of the sounds in the spoken language is necessary; the letter’s sound range already included both before the spoken language even got there.

    9+. The thing that actually stands out the most about Voynich letter ^a^ would be not that it has those two sounds, but that it might not have come from Syriac alef. It looks more like Syriac waw, which is also used as a vowel (covering both /u/ and /o/)… but it’s also graphically identical to Mandaic alef. Did they adopt the rest st of their alphabet from Syriac but just alef from Mandaic? (It happens to be the one letter that the Mandaeans attached the most symbolism to.)

    10. When the Aramaic alphabet was popular, the Aramaic language was experiencing a sound shift from /š/ to /s/, so the Aramaic-derived alphabets inherited a system where the letter for /š/ also sometimes represented /s/, even though there was already another letter for /s/. So, just like with /a/ and /e/, no merger of the sounds in the spoken language is necessary to explain the single letter correlating with both sounds; it already represented both before the speakers of Proto-Romani/Voynichese got there.

    11. In the Brahmic alphabets, although there are separate letters for /m/, /n/, /ŋ/, /ɳ /, and /ñ/, there is an alternative way to represent any of them. When a word has any of these sounds followed by a plosive, whatever other letter comes before them can have a dot added above it, and then the plosive letter immediately after it. The dot indicates that the plosive is preceded by whichever nasal sound fits: /m/ before /b/ or /p/, /n/ before /d/ or /t/, and so on. For example, the Nepali word for “grapevine” starts with अङ्गुर “aŋgur”, and the Hindi one is अंगूर “aŋgūr”. The dot above the अ for /a/ in Hindi does the same thing as the Nepali word’s letter ङ् for the /ŋ/ sound before the ग for /g/. So the dot is a single symbol representing both /m/ and all of those variations on /n/, illustrating that the people who created and use that writing system think of those sounds as all the same in a way, at least enough to all be representable by the same symbol.

    12. It’s not clear whether Voynich ^t^ was /t/ or /tʰ/ or even both, but it is interesting that the only plosive for which this ambiguity appears is alveolar, because there seems to have been so much other remodeling of alveolar or near-alveolar consonants in the history of Romani, of which this would be just one more part. On top of the apparent oddity with /d/ (#7), Romani evolution also includes losing all retroflex consonants (#1), losing one version of /r/ (#3), and somehow picking up both a voiceless alveolar affricate (#6) and a voiced alveolar fricative. And yet, as weird as it is, I didn’t even make up that history ad hoc just for Voynichese; it’s what professional historical linguists tell us about Romani already!

    * * *

    A couple of things I would have already said were in favor of an Indic Voynich theory are the fact that they both have a certain unusual combination of phonetic traits, including both voiced and unvoiced aspirated plosives together instead of just one or neither, the lack of a /z/ (or /ž/), and the lack of /θ/ and /ð/. Romani might seem at first to go against this, because, although it still lacks /θ/ and /ð/, it has lost its voiced aspirated plosives and gained /z/. The current concept of the Voynich phonetic system does not reflect these changes. But there must have been a time early in the history of Romani or Proto-Romani when they had not happened yet, and the carbon date of the Voynich Manuscript does predate the earliest known attestation of Romani.

    Other people have suggested a relationship with Romani before based on something else, but I believe this is the first time somebody has based that conclusion on a phonetic system, which would make it the first time a Voynich-Romani (or Voynich-Anything) connection can be tested, by an expert in Romani trying to read from the manuscript using the given phonetics…

    • MarcoP

      Thank you, Derek! An Indo-Aryan language, spoken in Central Europe in the xv Century, with no known written form at that time… Sounds like a great candidate!

    • MarcoP

      Hello Derek,
      in 2010, Georg Nicolaus Knauer published a paper that pushes the date of the first written Romani about 30 years back: “The earliest vocabulary of Romani words (c.1515) in the Collectanea of Johannes ex Grafing, a student of Johannes Reuchlin and Conrad Celtis”. I have been unable to download and read the paper, but people with an academic affiliation should be allowed to do so.
      The original XVI century Bavarian manuscript page can be seen here (Bayerische Staatsbibliothek in Munich -BSB- : Codex graecus Monacensis -Cgrm- 582a).
      The relevant part is the lower half. I can only make out a few of the Latin words whose Romani translation is provided. Some are astronomical: Aries, Taurus and other Zodiac signs (but not all of them I am afraid); the Sun and the Moon. Then there is a short list of animals (man, horse, mule…). Food and drink (bread, coffee, water…). Anatomical parts (head, hears, tongue…). Numbers from one to twelve.

      • MarcoP

        Not coffee but cheese (caseum)…. 🙂

      • Darren Worley

        Marco – great comments. The fact that Romani was first recorded in a Bavarian manuscript is intriguing given the recent evidence linking the VM with this region. [Because some unique features of the VM Zodiac (eg. Sagittarius Crossbowman) have only been found in manuscripts from German-speaking areas, in particular from Southern Germany.]

        If the BSB-Hss Cod.graec. 582a contains some Romani terms for Zodiac signs (Aries, Taurus) as you report, then maybe these could be correlated with some text in the VM?

        Comparison of the zodiac signs in the VM with Pal.Lat.1369 suggests that the text on the inner-most ring of the VM Zodiac pages likely contains text labelling the Zodiac sign they appear alongside. By applying Derek’s phonetic rules to this Voynichese text it might be possible to correlate this with the corresponding Romani zodiac terms.

        Derek/Marco – what do you think? Would this help test your theories?

        • Derek Vogt

          What Voynich folio are you talking about an inner ring of?

          I can’t read anything in that image Marco posted. There must be better online Romani translation sources than that: not only more legible but also with a lot more words in them than could fit on any single hand-written page. I haven’t found one yet, though.

          I personally need to switch to the other accepted spelling, “Romany”. The “-i” ending is making me read it with emphasis on the second syllable even though I know it belongs on the first! 😮

        • Dear Darren,
          The manuscript has been linked with German-speaking regions only because those who wish to link it to German-speaking regions have produced evidence which relates to none but German-speaking regions.

          It is simply and plainly untrue that such imagery occurs no-where else, and in fact most of the examples produced are “fish-in-a-barrel” ones. When the ‘German’ theory manages to find examples (as I have done) for the more difficult folios, or manages to explain exactly why no analysis of the written part of this text has ever suggested its underlying language is German, then the argument might be better weighted.

          I’d suggest too, that most of the suggested comparisons are spurious, since (for example) they fail to explain why the Voynich archer bears no relation in its stance, clothing or even its type of crossbow to those German examples, the great majority of which come after 1440, while supposedly explaining content fairly obviously gained from much earlier sources – as Tiltman reminded his readers in repeating the opinions of Irwin Panofsky and the keeper of manuscripts at the British Museum – viz that the manuscript ‘may be a copy of one much older’. It most certainly is, and the content is demonstrably older still.

          That is why we have back-to-front arguments, with late and imitative forms from German manuscripts being used to argue that the matter in MS Beinecke 408 is German.

          When we see a full and objective comparison with (e.g.) crossbowmen figures from Spain, France, England, Germany and from Jewish and Islamic manuscripts all compared, then and only then is any conclusion of German provenance acceptable. When the ‘German’ theory can account for imagery which shows the sun being reborn from a flower, or an image of two distinctly different creatures united by one head.. both ideas totally alien to the Latin European way of thought – then the story might be worth listening to. In fact such ‘Gemini’ figures, like the Leo, or the ‘fairy with wand’ can be traced directly to Hellenistic originals. Which reminds me – has any devotee of the German tale yet found an image of scales like those in the Vms? That is, with the crossbar threaded with a separate rod from which the scales hang? If so, I’d love to see it; I found none in any Latin European work.

          • Stephen Bax

            I am definitely a neutral on the German question, but I confess that the sheer number of crossbow connections with Germanic connections which have been referenced recently on these pages cannot, in my view, be dismissed so easily. Diane, you mention other crossbowmen from other contexts but where are they? I am not an expert on this at all, but I can’t remember seeing any.

            Of course, none of this means that the underlying language need be German.

          • Darren Worley


            Further to my earlier posts dated 14/September and the 27/September, could you post just one counter-example of a Sagittarius crossbowman from a non-German source? I’ve looked elsewhere and haven’t found any yet.

            You can dispute as much as you like, but you need to back up your argument with evidence.

            I look forward to your response.

            • Darren,
              I am not disputing that in manuscripts made in German-speaking regions from about the middle of the fifteenth century, it was a fashion to use the figure of a German man, bearing a German-style crossbow, dressed in German costume – usually hatless.

              What I am disputing is your effort to suggest that those images have any discernable relationship to what we see in MS Beinecke 408, or that you can expect to argue chronological order from frequency.

              Just as a thirteenth century French madonna is a woman holding a child, and a fourteenth century German madonna is a woman holding a child, so your examples of man-with-crossbow (no matter how many produced) cannot be used to argue either provenance or precedence in relation to the Voynich archer.
              The parameters which you insist upon are entirely artificial and, I should say, equally inappropriate. Expecting parallels only from manuscript art is a nonsensical habit which makes sense no-where except in the peculiar atmosphere of the German-theorists in Voynich studies.

              In fact, the closest forms to those used for centres of the month-roundels in the Vms are not in Latin manuscripts at all, but on coins, and Hellenistic at that.

              The only reason for insisting that none but fifteenth century Latin manuscripts be surveyed is a bit of a mystery to me. I don’t know whether it’s because people have no time to research iconography, or whether they have no interest in it, or whether the aim is simply to eliminate all sources and media in which some objection to the ‘German’ thesis might be plainly seen.

              In this case – it is just not good enough to announce that the only relevant aspect of the Voynich image is the sex of the figure and a vague description of its weapon – ignoring all other details, all other indications of where and when it was gained, and all analysis of its intended meaning.

              Setting up so narrow a limit for debate, effectively denies any real debate at all. It’s like permitting an atheist only to argue with quotations from the epistles of Paul!

              Who says that the imagery in any manuscript only came from other manuscripts? Are manuscript illuminators imagined locked in some dungeon, never permitted to see any other form of art, sculpture, wood-carving, coinage, or embroidery?

              There are both technical reasons (such as the style of crossbow), stylistic reasons (such as the method used to produce the image), reasons of detail (including costume) to assert that while it is possible that the German manuscripts naturalised the same, or similar forms, from about the second or third quarter of the fifteenth century (though mostly later still), that in the Voynich manuscript belongs to a different tradition, an earlier set of exemplars and originated a very long way in both time and distance from Lake Constance.

              It wouldn’t matter how many pictures one produced from within your preferred, and unintelligible limits of “fifteenth century Latin Christian manuscripts” – they would have little (I won’t say “NO” relevance) unless the Voynich archer had originated in the mind of a Latin manuscript artist … which I have already proven, with ample evidence, that it does not.

              If you want the thing translated, I read it as “Archi-pelago”, where the “arc-” refers at one to the bow of islands, and the Islands’ bow(man).

              But I’ve spend enough time and effort on the topic. How about you try to explain why you think none of the usual rules of provenancing art apply to arguments hoping to provide post-hoc justification for that long-held theory, Rene’s original brain-child?

              • All these references to what I am supposed to think or whom I am supposed to be influencing are both spurious and annoying. It’s probably even more annoying to Marco and Darren, who have been doing their own excellent work, not needing me for anything.

                • Stephen Bax

                  As I said, I am a determined neutral in this, not having any Germanic axe to grind.

                  However, I do consider it potentially significant that the VM Sagittarius is carrying a particular kind of crossbow. I think it highly unlikely that the artist(s) would draw this kind of weapon out of thin air. It is highly likely, in fact, that the image is based either on weapons which were typical in the artist’s vicinity, and/or from manuscript traditions known to the artist.

                  This surely means that it is highly relevant that the examples found by Darren, Marco, Rene and others seem to come from one geographical area, while we seem to have none from anywhere else?

                  Of course we should not ignore other aspects of the image or other images. Nor should we assume that the artist was necessarily actually working in that geographical area (as s/he could possibly have been copying from a manuscript which had travelled).

                  Nonetheless, to my mind the weight of evidence is now tending to suggest an influence on the Sagittarius crossbowman from that geographical region. That is a fascinating step forward, in my view!

                  However, let’s keep searching for more examples and expand the research to cover other elements, to see if they give corroboration!

      • Marco,
        The introduction to that paper makes the mistake of supposing that a Benedictine was merely an interviewer and scribe, and that his scholarly approach must rely on secular and ‘important’ German scholars.

        The Benedictine himself was plainly a scholar, well educated, clear-thinking and ready to engage in dispassionate study before (if ever) forming personal theories.

        The paper is otherwise well-written. Thanks for the reference.

    • Darren Worley

      Great work Derek. I like the scientific method you’ve outlined. Constructing a theory that can be tested is the right way to proceed.

    • Derek Vogt

      I did not distinguish between two forms of this theory. One is that Voynichese is the ancestor of modern Romani, an early stage in Romani’s development. The other is that Voynichese and the contemporary form of Romani back then had already split into two dialects or languages, making Voynichese not an ancestor but a cousin, with no living descendants. This version of events might not be the first one people think of when a relationship between an old language and a modern one is proposed, but in this case, it does simplify a few of the required phonetic transformations.

      • Derek Vogt

        It just occurred to me that some people might think my lists of phonetic changes here looks long & complicated, especially if they haven’t read about phonetic shifts in language evolution before. To put it in context, here is the accepted list of phonetic shifts to get from Proto-Indo-European to Proto-Germanic.

      • Derek Vogt

        Another discovery on the Romany front: Somewhere around here, I’ve posted before about a word (allowing for some spelling inconsistency) that appears four times on folio 68r in the sky map, always as the secondary one of two labeled stars that seem to be in the same constellation, which is named in the other star’s label.

        27 ^artw—^: Serpens
        32 ^ãtw—^: Cetus
        47 ^ãghoãtwr^: Argo; “Argo’s ^ãtwr^”
        48 ^ãtw-rhaš^: Hydra; “ãtw- of the snake”

        I expected this word to be relatively important because it’s the only one I can ascribe a meaning to that should be just a plain native Voynichese word, instead of an astrological or botanical name that’s likely to have been imported from another language. So the right language identification, in addition to having a reasonably close sound inventory, should also have a word in it meaning something like “secondary, the rest, continuation, extension, back, behind, rear”, which sounds close to “artw” (or “artr”, given the fact that Voynichese ^w^ has a bit of a tendency to show up at the ends of root words whose cognates have “r” there).

        I could not find any such word at Google Translate in Greek, Arabic, Hebrew, Bengali, Gujarati, Hindi, Marathi, Nepali, Persian, Punjabi, or Urdu. The best I could get were in the Turkic family:
        Azeri: ard, arxa
        Kazakh: артқы, артқа “artki, artka”
        Turkish: arka
        Uzbek: orqa

        But then, while trying to find a general online translator for Romany because Google Translate doesn’t include Romany, I ran into this page. It’s not a translator, but just a short list of Romany words and their English meanings… which includes “arter”, translated as English “after”. So now we have a Voynich word, the first one that would seem to be just a word rather than a name, for which the best match comes from not Indic languages in general, but only a specific one… and that one lone language is Romany.

        * * *

        Also, I somehow left out /f/ and /v/ in that image even though they are part of what originally drew me to Romany (because Romany and Voynichese have them but other Indic languages don’t), which is bugging me too much to not fix it, so here it is.

        • Hi Derek, I would be wary of leaning too much on “arter” as “after”. I do not know the value of the source you link to, but this Romany word may actually be English. In some English dialects “arter” is a pronunciation of “after”, and this could have been borrowed into English Romany speech.

        • Derek,
          re your Romani/y idea,
          I was very interested to read when exploring this possibility that one of the earliest of the ‘Gypsies’ to arrive in Europe was named Thomas, and claimed to have come from India – southern India from memory, though I can check back to the old blog if you want to be sure.

          The interesting thing is that we have an account of his going to the Basque country and speaking Basque with one of the natives. The latter, when asked how Thomas came to know Basque, said more or less, well of course he does, or “why would he not?”. Curious, isn’t it?

    • MarcoP

      These XIX Century books contain lists of Romany words.
      Die Zigeuner in Europa und Asien (1844).
      The dialect of the English gypsies (1875).

      I found some interesting comments by Reed Johnson about the Romany and/or Indian origin of the Voynich ms:

      The parallels for the woman eaten by a fish and the “nebula” diagram are particularly impressive. I am not sure of the relevance of these Indian images for the hypothesis of a Romany connection.

      By the way, in his 2012 post, Reed Johnson published the same Indian Gemini couple we used here.

      • Marco
        re your comment of October 31st.
        That figure with a fish for the lower body isn’t a woman, but a version of Matsya.
        A lot of readers will probably know this, because my post about it has been read pretty steadily since first written up in the old exploratory blog, and in the updated version published in voynichimagery on April 17th., 2013.

        The idea that the month-roundels reflect the present Greco-Indian system was entirely Reed’s idea, and he may have been the first to suggest it, too. I don’t think it correct, but the really interesting point for me was that when I looked at the way Kircher’s book represented the figure, I was immediately struck by a realisation that *that* is precisely the style and genuinely awful drawing which informs f.57v. I’ve been strongly inclined to the view, ever since, that the drawing on fol.57v may be Kircher’s work, or that of whoever made the drawings for his book. I do not think the image on f.79v refers to the same subject, but I may be mistaken there. (I wish it were possible to have the ink on f.57v tested. 🙂

    • Dear Derek,
      Do you happen to know the Berber terms for stars, as recorded by Michael Scot in Sicily?

      • Derek Vogt


    • Derek Vogt

      A simpler & quicker presentation:

    • Derek Vogt

      I had originally thought that one of the differences I needed to account for between Voynich and Romany was that the Voynich language has /x/ and Romany doesn’t. (Item #8 above) That was because Wikipedia’s phonetic inventory table for Romany doesn’t include /x/. But Wikipedia’s page on Romany includes a bunch of example words, some of which do have it, and it’s also one of the basic sounds in various words scattered all over my new English-Romany dictionary. So now it turns out that having /x/ is another thing both of these languages have in common. To be more specific, it’s another trait of theirs that’s different from other Indic languages, so it’s another Voynich deviation from Indic that would be explained by a Romany connection.

  14. MarcoP

    This post by Nick Pelling discusses what apparently is a Voynichese color annotation in f2r. Nick’s image processing makes the word more readable. Nick and Rene propose the possible transcriptions EVA:ios.an.on / ior.an.oin. If this is a color annotation, it should stand for green, since it is under a layer of green paint. So, this word seems to be well constrained and a good candidate for translation. On the right margin, there is another possible (but less reliable) color annotation: EVA:ytoailo.

  15. Drew Russell

    Recently, I have begun to transcribe entire pages from the manuscript into Roman characters, working based off research I have seen on this website – on this very page, in fact. I am doing this because my amateur experience does not extend sufficiently to allow me to identify plants, apply Middle Eastern star names, or even utilize the fascinating-looking website Negesta provided below. However, a few interesting patterns have come to my unskilled attention.
    First of all, I have noticed that short words beginning with the Voynichese letters for ‘h’ or ‘ħ’ often occur as or within a series. For example, the first line of the last paragraph of f1r, ‘va ħakhoon has għatooșn ħon bntoon hn ra t,’ and the last line of the same paragraph: ‘oa (indistinct symbol, perhaps smudge) has hak hagn hagon.’ While I lack a genuine professional theory about this characteristic, it is nonetheless interesting. Perhaps ‘and’ has a myriad forms in Voynichese, or else they simply sound similar due to coincidence, or perhaps complicated sound-harmony or grammatical rules.
    A side-note to the first line represented above is that each and every one of the words followed by a short-h word is a completely unique one-time occurring word.
    Also, there is a seeming tendency of the language to ‘build upon itself’ in the formation of words. For example, words incorporating the ubiquitous ‘twiir/twir’ word are common, with the same being true for other ‘-wiir/-wir’-terminated words. A significant percentage of one-time-occurrence words seem to be of similar nature, for example, on f1r ‘katħon,’ on f58r ‘tushar’ and ‘haskoon.’ In all three of these examples, each of the two components that make up the word is itself another word that occurs elsewhere in the manuscript. While this manner of word-building is common to many languages, it would ease translation of some of the one-time words, and for all I know it could provide insight into what in the world the language is.

    • Stephen Bax

      Thanks Drew. I would strongly suggest that you make use of the EVA transcriptions so that everyone else can easily follow what you are doing. No need to transcribe it all yourself again!

      For example if you start from Rene Zandbergen’s site, you can see a full discussion of f1r here. You can them find a full interlinear transcription here.

      There you can find an EVA transcription of the line you are referring to – I think it is


      For individual words you can use http://www.voynichese.com. For example here.

      If you use EVA then other people can easily use the same tools to follow what you mean.

      • Drew Russell

        I hadn’t thought of that with the EVA – mostly because I never learned it to begin with. I’ll try to remember to include EVA renderings in any further comments. Thanks!

    • Derek Vogt

      If the idea is to transcribe it phonetically, then the simplest method would be to download one of the non-phonetic transcriptions that I presume have already been done in systems like EVA and then apply a series of “find and replace” (as long as you do those in the right order so you’re not “replacing” with something you’ll “find” afterward). I’d like to do that myself, but I just don’t know where to get a whole-book transcription in any system.

      • Drew Russell

        As is, what I have been doing is simply enlarging the page images on http://www.Voynichese.com and writing the entire text down, with the Latin transcription added below the Voynich script. Although I have similarly had no luck finding a whole-book transcription of the manuscript anywhere, the above-mentioned website has something fairly close, with nearly all words identified and assigned EVA renderings – I say nearly all because I have encountered an instance or two where it does not register a word at all. The main disadvantage I have found in using the site’s page-images is that they only include the text, often omitting the margins of the page and therefore leaving out non-textual features such as plant roots. However, as noted above, there is the advantage of having what is in essence a combination of the original text and an EVA transcription.

      • MarcoP

        Hello Derek, a complete EVA transcription of the whole manuscript is available at http://www.ic.unicamp.br.

        voynich.freie-literatur.de provides separate EVA transcriptions of each single page.

  16. Negesta

    It’s great what you’ve done so far, imo.
    I think that’s the way. But it is a long way…

    1. I would suggest this database for protoforms: http://starling.rinet.ru/cgi-bin/query.cgi?root=config&morpho=0&basename=ALL&defpos=on
    It allows searching for matches on a larger scale.)

    2. Willing to test the partial decoding proposed by Stephen Bax, I made a model of how the word “dragon” would be written, according to some Latin, Ancient gr. and PIE roots, and I searched for matches on the first page (f1r). So, just before the (second) winged dragon image, at the end of the 5th row, I think I found the word “N DəRəiKhN”
    If “8” = “D”, or “TH”. If it’s T only there is a proto-germanic form “trak-an-“, *trakk-an”- m. for “dragon”. The initial “N” could be an article of some sort (un/an). It’s possible that “8a”=D, “8”=T

    Then I searched f25v where there is another (but different) dragon. I couldn’t find the same sequence, but found an interesting word: rows 3 and 5 “D/T KhON” . Possible abbreviation for the same word, where 8=”DRa”?

    • Hello! 8 – is a turkic sound Ƣ (Like R and Gh together)

  17. For completeness, could we include identifications by Fr. Theodore Petersen (whom many others have followed); also Dana Scott who made many acute observations, including the Rose (which I adopt). Also Wiart and Mazar who are experts in botany and identify plants from the eastern corpus, and just for the hell of it, why not list my identifications too? I have it on good authority that neither Gd nor any major Voynichero will actually strike you dead for mentioning these names and acknowledging their work. I know it will mess up the impression of uniformity, but balance is best, and the priority of the older researchers should not be waved away. IMO

    • Derek Vogt

      For a list of mutually supporting plant identifications and phonetic name-matches, adding plant identification sources whose identifications did not yield phonetic name-matches would not be completeness; it would be inaccuracy.

  18. Drew Russell

    This is pure unprofessional speculation, but it fits with an apparent pattern I have noticed.
    Perhaps EVA ‘g,’ one of the few remaining letters of unknown sound, is simply an embellished, predominantly final form of EVA ‘d’/Vogt ‘t.’ I base this idea off of the formation of EVA ‘m’, which Derek Vogt has identified as an alternate form for representing the ‘r’ sound; EVA ‘m’ appearing to be formed by attaching a descending ‘tail’ to EVA ‘r.’ Similarly, EVA ‘g’ closely resembles EVA ‘d’ with its own ‘tail’ attachment.
    Additionally, a short period of messing about with http://www.voynichese.com indicates that both EVA ‘m’ and EVA ‘g’ are most common as finals. Specifically, ‘m’ appears eleven times by itself, twice as an initial, fifty-two times as a medial, and a whopping one thousand fifty times as a final. ‘g,’ correspondingly, also appears eleven times isolated, then five times initially, seven times medially, and a not-so-whopping but still significant seventy three times.

    • only a exemple

  19. Beispiel

  20. That is interesting. But which translation you put here to reason? From Latin, I’m assuming that 1. it is a scientific book. 2. the origin must be based on the data around the Alps. 3. has set itself the highest success rate based my experiments. Similar systems have already been used in 1400. As discussed earlier, it is not by an alphabetical one to one application. Then the Rätzel would be solved long ago. Here an applied pattern of a code from the early 1500 century.

  21. Francisco pompeo

    Good to hear that finally someone is getting closer of what all that means!! I was reading a bit, and when it seems to be mixing india with europe, usually the answer is in georgia or close. It was once a kingdom formed by many people by the XIV century. Later, the mongols dominated the region and we know they are famous for destroying everything. We Got no clue probably because of that.

    • I think that it was an a Khwarezm’s or a Sasanides empire’s time of this language exist. Too many crosslinks with the nowadays languages on they territory… But mostly this language is turkic than iranian (persian/arian) in a division approx. by 70% to 30%.

    • I’ve recently conversated with one person who said me that he has seen the same symbols as voynichese scraped onto the Caspian seaside’s sand with a stick, near the Derbent city, approx. 30 years ago. He said that he will try to remember of what was inscribed there. So maybe a native speakers are still exist in the world…

  22. Hello all! Look please at what I’ve found yesterday…
    This is a latin transcrtption of an outer text line of this circular diagram (http://www.jasondavies.com/voynich/#f67r1_f67r2/0.253/0.492/2.60)

    “Boənroş hyiğu yököğu ukieğəş oş otöğən̡d yöhçieğu ot yitökən̡d yötieğu otor izyiw yizər ər hyieieb ğieğu bohtieğ ohthew boyiköt ow yötöku yöhçiw okieb bər ən̡d ğərəm byişib yilər ğər ğiebyiğu ğər ğər ər otər ğər ukərob”

    The language looks like azerbaycan or turkmen or another turkic language.

    • Hello Stephen! This is probably an old uyghur or chagatai language.
      I’ve found some words in the uyghur dictionary.
      “ğər” means “whore” (earlier I considered this word as a “woman”);
      “beehtiyar” means “unwilled”, “slave”;
      “ğərəm” means “haystack”, the same as “garem” or “harem”;
      “ğərəz” means “selfish purpose”, “evil intent”;
      I’m not sure in the last symbol of this word (m or z)
      The word I’ve written “yizər” can be similar to “hizər” as the first simbol is hamza. This means “one thousand”. Then, if this right, the last symbol of (EVA: daram) is still “M” and the whole word is still “ğərəm”.

      And I’m waiting for your verdict. Am I beat this Voynichese cipher? =) Will it be any prize for us both?

  23. Derek Vogt

    I believe I’ve figured out one more new thing about how written Voynichese worked and possibly one more new thing about its origin & culture. Some of what I’m about to talk about here is old stuff, looked at in a new light.

    I’ve said before that EVA-i (and EVA-n, which I think of as just a different look for the same entity) appeared to represent something non-phonetic, something quantitative or ordinal, and specified by how many of them are written in a row. Some possibilities I thought of at the time were about noun case and verb conjugation, but I now see it mainly as an indicator of emphasis/magnification/increase, so it could include things like plurals, countable & uncountable relative quantities, comparatives & superlatives, and the frequency or intensity of actions/events/descriptions, all depending on what word EVA-i/n is attached to. As I was recently reminded, some languages have a single general mechanism for intensification in all kinds of applications. Sometimes that is by repetition of part or all of a word, rather than by specialized words or affixes (prefixes, suffixes, infixes) like our “very” and “super-“. So EVA-i/n, with its conspicuous tendency to appear in repetition, would be like tally marks in a repetition counter, ticking off how many iterations of whatever else it’s attached to would apply. (Whether it indicates actual spoken repetition of what it’s attached to or is a purely written device like our bold & italic fonts doesn’t make a difference for now.)

    A later observation that I didn’t connect to that at first was that three of the four star labels on 68r1 and 68r2 which use EVA-i/n (27, 32, 48) were spelled similarly (^artw—^, ^ãtw—^, ^ãtw-^) and located where there should be a continuation of a constellation whose primary part is named nearby (26, 33, 51). So I inferred that they were different spellings of a word that meant something about being a secondary part of something bigger or following behind something else. But that was about what the whole word means, not thinking about its separate parts.

    With that word breaking down into two parts, these two ideas fit together perfectly; ^ãtw/artw^ would carry the basic meaning of one thing somehow connected to another, and EVA-i/n would quantify it, so together they would indicate something like how far back along the whole constellation the additional star is or how much more of the animal there is in that area of the sky. That allows another example to be connected to those three: star 47, which is in a similar position (the back half of Argo) and has ^atw/ãtw^ but no EVA-i/n. It looks like the same root word, and can now be read as having the same meaning, just without the quantification from EVA-i/n.

    • Derek Vogt

      What first made me think of repetition as an intensifier in Voynichese was that I was looking at things Voynichese has in common with pidgins & creoles, and that was one of them: pidgins & creoles tend to intensify by repetition more than most other languages do, some even exclusively.

      Another is that unique sounds in either parent language get dropped, and sounds that are distinct from each other only in small subtle ways get merged, so the sound inventory gets smaller than it is in either parent language and smaller than the average non-pidgin, non-creole language, reduced to a subset of the sounds that both parent languages have in common. And Voynichese’s inventory certainly looks small.

      Another is that grammatical functions such as verb tense and noun case are conveyed more with auxilliary words and word-order conventions, and less with affixes (prefixes, suffixes, infixes), so there are fewer different kinds of affixes or even none (although the few that remain, if there are any, can be very common). And Voynichese certainly appears to have a small number of very common affixes.

      Of course, regardless of how intriguing I find the possibility, none of that means Voynichese is necessarily a creole. It’s possible for a single language to undergo coincidentally pidgin-like simplifications that would yield similar results. But that doesn’t usually happen, so the similarities do shift the odds of Voynichese being a creole.

      The comparison between Voynichese and creoles is what made me think of the idea of EVA-i/n as an intensifier based on repetition or something like it, but the latter still works with or without the former; I’d still think that’s the best model for the use of EVA-i/n even if I found a good reason to think Voynichese was not a creole.

    • If I may also speculate a bit…..

      In various word paradigms and analyses of the word structure, the point about the strings of i’s followed by one of: n, r, l, m (all liquids, by the way – isn’t that curious), has been highlighted.
      I have been wondering if one should not concentrate on what follows the i’s, but on what precedes the i’s.
      Essentially always, this is ‘a’, and when not, it is usually ‘o’.
      So, what if, first of all, ‘a’ and ‘o’ aren’t really different, but just scribal variations or inaccuracies, and secondly, the four-column table of i’s followed by the above characters should be replaced by a single-column table of ‘a’ followed by 0-3 i’s.
      ‘a’ could be one vowel,
      ‘ai’ another vowel
      ‘aii’ yet another one, etc.
      whether a final ‘n’ should be read as ‘i’ remains an open question.

      Just a thought…

      • In principle, I note that no E is present in most EVA translations, which in turn makes me very skeptical of the system. Since especially the vowels are an integral part of a language.

  24. MarcoP

    Hello Derek,
    thank you for sharing your amazing work! I read here that you are considering the next steps. On this subject, I largely agree with much of what Darren wrote about structured data sets. In order to put to test a linguistic hypothesis, I think that “structured data sets” are worth trying. Darren’s specific suggestions also make sense to me: (i) names of the zodiac signs (ii) names of months or days [some cultures have different names for each day of the year] (iii) numerals or (iv) the direction of winds or compass-points. I would possibly add the names of the seven planets (or nine planets, in the Indian culture) and the four seasons.

    Collections of circular diagrams similar to the “cosmological” section of the Voynich manuscript (e.g. Walters ms W73) often include all the elements suggested by Darren. When each zodiac sign / month appears so prominently in the manuscript, I think it is reasonable to assume that the same information appears in one or more of the twelve-segments diagrams.
    It would also be interesting to systematically look for internal correspondences. Are the zodiac constellation in the f68r1/2 “maps” identifiable on other cosmological diagrams? I have made a few attempts of this kind, but I could not come up with anything worth sharing.

    • Derek Vogt

      The idea makes sense, but is only usable if we can find such data sets in the book and be sure from non-alphabetic symbols that the data set and the items in it are what we think they should be. And I’m not aware of any exampes that work so cleanly in the Voynich Manuscript. (For that matter I would think that if there were any then somebody would have used them for this before, but then again, I would have thought that Bax’s duplication of the Rosetta Stone method would have been tried before, and it apparently hasn’t.) For example, in the set that almost looks like a list of drawings of Zodiac people & animals, aren’t there one or two missing, and some kind of extra or one that appears twice?

      I did recently try going through the “12 winds” suggestions here, but most of those Greek & Arabic wind names didn’t come anywhere near the Voynich words in question as I read them. (At least that means I’m probably not just stretching my imaginaton to invent matches even when they’re not really there!)

      The only comparable Voynich images I’ve looked at closely and found decent phonetic matches in are the T-O maps/diagrams on f68v and f86-rosettes, but there are still problems with them. For one thing, they don’t even look like they could have the same overall meaning & significance. One is like a rough conceptual map of the world, while the other is tacked on to a corner item like an afterthought, the tenth or eleventh thing on a page meant for nine items, where some or all of those main nine already look like they represent the world or parts of it, so the T-O thing squished into the margin clearly can’t be the whole world there.

      And the two sets of labels don’t match each other very well. The ones where Africa would be expected say ^aka$^ and ^agataš/agatas^. We can connect those with what would seem like relatively little change if they were in different languages: a compression of ^t^ & ^s^ to ^$^ or splitting of ^$^ to ^t^ & ^s^, and a voicing or devoicing between ^k^ and ^g^… but those seem like large differences in what should be one word in one language at one time. And the ones where Europe would be expected are worse: at best, ^ãbãon/ẽbãon^ and ^abhãstn^, comparable to two different ancient names for (loosely) Europe or (more literally) western lands: Erebu and Baran. But the two Voynichese words are nothing like each other past the first two letters, so trying to connect them both to the same meaning in other languages can’t yield results that are more than half right, or give us any way of knowing whether we’ve guessed half-right or all-wrong. And for the ones where Asia would be expected, ^a$wš^ in one is decently close to an ancient name for Asia, “asu”, but its counterpart in the other T-O image has a bunch of words, none of which come close to that again at all… and worse yet, the one with a fairly simple, clear resemblance to a name for Asia is the one that looks less overall like it could be a world-map!

      This makes it look like the two T-O images represent two different ideas, and that kind of uncertainty about the meaning of an image is exactly what we’re supposed to be avoiding by trying to find things like this in the first place! These data sets just don’t want to cooperate by being very “closed”.

      • MarcoP

        Thank you for your reply, Derek! Of course, it would be very nice if we could “be sure from non-alphabetic symbols that the data set and the items in it are what we think they should be”. I agree with you that this is not the case.
        I also agree that my proposed interpretation of a set of labels as a wind rose is extremely unsatisfying. Yet I think that providing a satisfying interpretation for one of those diagrams still is the most viable way to prove a phonetic reading right and, since structured data set have a fixed order, the correct reading will likely be self evident. Of course it is possible that all the “structured data sets” that Darren mentioned are absent from the Voynich manuscript. If this is the case, the alternative is what you mentioned above: looking for low frequency words in the herbal pages, and trying to correlate them with plant properties. We have so much uncertainty about plant identifications, phonetic values, the nature of the language and actual or mythical plant properties in different traditions: the task of spotting meaningful words in random positions of the herbal pages seems to me extremely difficult. How can we avoid false positives?

        I would also like to address your question: in the set that almost looks like a list of drawings of Zodiac people & animals, aren’t there one or two missing, and some kind of extra or one that appears twice?
        The anomalies you mention are there, yet that cycle of images is still clearly recognizable as a zodiac. Actually, it has been read as such by whoever put the Latin alphabet month names on those pages a few centuries ago. Quoting René Zandbergen: “The illustrations for Capricorn and Aquarius are missing, but there is a missing folio (f74) exactly at the point where they should have appeared”. It is a strange zodiac, but it still is a zodiac. The names of zodiac constellations are interesting also because they are usually translated in different languages: the Greek ζυγός, zygos, the Persian tarazug, the Latin libra, the Hindu tuladanda are all related to the concept of “scales”.
        From Spain to India, different cultures shared the same set of symbols for the Zodiac signs: the borrowing has been semantic, not phonetic. The Zodiac images in the Voynich manuscript are strongly suggestive of the fact that the Voynichese culture used the same set of Zodiac symbols as European and West-Asian cultures of the time. The fact that the names of the zodiac signs are semantic increases the variability between different languages, likely making the search harder (wind and month names apparently were sometimes carried over phonetically between languages). But if the names of the zodiac signs could be found, we would have very much identified the language as well. Maybe in the future we will manage to restrict the possibly related languages and this will make the search for these structured data sets more manageable.

      • Derek,
        Sorry I came to this so late. I don’t know if it will assist you, but the so-called T-O diagrams have different intentions. That used for the north-marker on folio 86v is meant for a physical place – the furthest place known to the original maker of that map, and conceived as a tripartite ‘city’ which is plainly joined by roads and ?steps? to the main part of the map – and (btw) the map isn’t particularly rough. It takes work to find the identity for the various land-forms and structures, but to do so is not impossible).

        In another folio the four-part division has been made tripartite with pigment, (to avoid confusion, I mean the folio on the left, here:

        It is evident that the painter – who was probably the overseer of the copying – tries in various places to force the original imagery into the mind-set of Latin Europe, and this is a case in point. There are numerous others through the manuscript..

        What fascinates me is that the ‘Asia’ part of that last ‘T-O’ shape is filled with a perfect, and realistic Asiatic face. The ‘crown’ he wears is that of the Babylonian kings, but since in the Yemen, and in southern India, such crowns were also traditional, it doesn’t matter so much as that the ruler of ‘Asia’ is probably meant for one of the Mongols. Nice cue to end-date, and possible cue to the language(s) used in the Vms. Underneath that late ‘revisionist’ painting, we have a quadripartite world, just as the world map itself is on f.86v. NOT a European conception at the time, but perhaps one as old as the Hellenistic period in the eastern regions. Thing is this ‘world in miniature’ is associated with the north of the sky: the location of the little paradise from the time of ancient Egypt to the present time. The ‘faces’ also represent the stars which marked the times and hours as they circled the northern Pole.

        I had hoped, when I explained all this initially, that it might help with identifying star-names, but I don’t recall anyone in particular who might have followed it up.

        The manuscript is a compendium of very practical information – but not European in content, or form before the painter got to it.


  25. D.N. O'Donovan

    Could you provide details of your sources, please?

    • Derek Vogt

      I identified a few plants myself; those are marked as note 11 above, so at least you already know which ones to be suspicious of because the person identifying them might have just done it too conveniently. 😀

      For the rest, identified by other people:
      …plus a few by other individuals right here at this website. When I first saved a single list to include all of these, I didn’t include which ones came from which people. My next little Voynich sub-project will be to associate each individual suggested identification with the individual who suggested it, at least in my spreadsheet at home. How to get stuff from that spreadsheet online is another issue. (I think Professor Bax has directly uploaded Word files here before, so maybe he can do the same with Excel files…)

      For the plants’ foreign names & translations:
      and in a few cases

    • Derek Vogt

      For the plants, the sources I’m listing below are for the botanical identifications. The plants’ names in various languages are from the standard online translators I named above, plus Wikipedia in a few cases and a general web search in the case of sweet basil. All plant identifications except four were based on only the drawings. The exceptions are that Darren Worley also used some cultural information for the olive identification and my three were influenced by my phonetic reading of their names.

      For the two astrological items on 68r3, the sources I name below are for both the names and the idea of what astrological entities the Voynich drawings represented.

      For all of the labeled stars on 68r1 & 68r2, the sources I name below are for the names/words for the astrological entities or the things they represent. The idea of which astrological entity to associate with each individual star on these pages is from me, using the phonetic system I had already developed using plants and Frederik de Wit’s Planisphere Celeste.


      Bax’s website: anonymous Finnish biologist:
      11v Daphne mezereum (no English name)
      11v genus Salix, osiers and sallows
      16v-a Nigella damascena or arvensis, breadweeds, fennels
      16v-b Illicium verum, bedian star-anise
      17v genus Dioscorea, yams, sweet potatoes
      20r genus Satureja, savories
      93v Bryonia cretica or dioica, bryonies
      95v1 Fumaria officinalis, fumewort, earthsmoke

      Bax’s website: anonymous Finnish biologist & Steve D:
      04r genus Linum, flaxes

      Bax’s website: Darren Worley:
      01v-a-1 Olea oleaster, wild-olive

      Bax’s website: Deyan:
      66v-a-1 genus Aloe, aloes

      Bax’s website: Hans Adler & Daniel Myers:
      21r Portulaca oleracea, purslane

      Bax’s website: Hans Adler, Jan M, Peter Ole Kvint, & Labyrinth:
      06r genus Papaver, poppies

      (reported by Stephen Bax in 2014):
      16r Juniperus oxycedrus, sharp juniper

      Edith Sherwood:
      02r genus Centaurea, knapweeds
      03v genus Helleborus, hellebores
      05v Malva officinalis, marsh mallow
      14v genus Stachys, betonies
      29v Nigella sativa, black caroway, black cumin, Roman coriander
      41v Coriandrum sativum, coriander, cilantro, Chinese parsley

      Ellie Velinska:
      08r genus Cucumis, cucumbers and cantaloupes
      09r Vitis vinifera, common grape
      17r Artemisia dracunculus, tarragon
      37r genus Chenopodium, goosefoots
      43r genus Diospyros, persimmons
      55v Allium ursinum, wild garlic, bear’s garlic, bear leek
      87v-b genus Pistacia, pistachios, terebinths

      Ethel Voynich:
      02v genus Nymphaea, water-lilies
      04v Ipomoea aquatica, water-spinach
      25v genus Plantago, fleaworts, psylliums, “plantains”
      28r genus Rumex, sorrels, docks
      38v genus Cynara, artichokes
      39r genus Colchicum, meadow-saffrons, autumn-crucuses, colchicums

      Ethel Voynich & Theodore Petersen:
      06v Ricinus communis, castor oil plant

      Theodore Petersen:
      01v-b-1 Solanum melongena, eggplant

      Derek Vogt (plants):
      24v genus Aquilaria or Gyrinops, agar, agarwood, gaharu
      27r-a Ocimum basilicum, sweet basil
      31r genus Croton (tiglium?), rushfoils

      Derek Vogt (astrology, using an online translator):
      68r1-01&02 spring equinox; Aries; “beginning” and “end” in Hebrew
      68r1-07 Pegasus & Equuleus; “horses” in Indo-Iranian languages
      68r1-21 Ophiuchus; “snake-charmer” in Marathi
      68r1-27 part of Serpens; same word ^artw,ãtw^ in 27, 32, 47, & 48
      68r2-32 part of Cetus; same word ^artw,ãtw^ in 27, 32, 47, & 48
      68r2-36 Eridanus/Fluvius; “stream/current/flow” in Arabic, Persian, & Urdu
      68r2-38 Corona; “wheel/circle/spinning” in Indo-Iranian languages
      68r2-45 Lupus “wolf” in Indo-Iranian languages
      68r2-47 part of Argo; same word ^artw,ãtw^ in 27, 32, 47, & 48
      68r2-48 part of Hydra; same word ^artw,ãtw^ in 27, 32, 47, & 48
      68r2-53 Crater “cup/bowl/pitcher/jug/urn” in Indo-Iranian languages

      Bax’s website: Darren Worley:
      68r2-53 Crater “cup/bowl/pitcher/jug/urn” in Hebrew

      René Zandbergen:
      68r3-55 Pleiades (in Taurus)

      Bax’s website: Marco Ponzi:
      68r1-16 Alpha Draconis (Abʰaya)
      68r3-55½ dragon of the eclipse (Gauzahar)
      •For the first, he linked to a paper by R N Iyengar (Raja Ramanna Fellow at Jain University in Bangalore); for the second, he linked to multiple sources, so I can’t pick one to name here.

      68r1-22 Beta Virginis (Auva, strangely unmentioned by Richard Hinckley Allen)

      Richard Hinckley Allen:
      all other star & constellation names on 68r1 & 68r2, including “Kurtos” for star 01 (alternative to Hebrew for “end” in the same label) and “Argo” for parts of the labels for stars 44 & 47

      • Stephen Bax

        I have also transferred these comments to the main posting at the top, for greater visibility.

  26. Derek Vogt

    The f68r sky map that goes with this: https://stephenbax.net/wp-content/comment-image/163448.jpg


  1. Voynich Manuscript Inspired Zentangle Patterns | Dank Doodles

Leave a Reply to D.N. O'Donovan (Cancel Reply)

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

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>