Notes on the Voynich Manuscript - Part 13 [1992 January 30] ----------------------------------------- Another Test of the Prosodic Hypothesis In an earlier note, I said that a good test of whether the Voynich groups were true words would be to find in the text many of the single words in the illustrations. Likewise, if the groups are not words, but can be derived from them, one should be able to predict how the single words would be transformed into groups, and find those. That is what I tried to do. This experiment assumes most of the letter values that Mr Guy and I have speculated on; in particular, it assumes 'a', 'e', 'o', 'u', 'n', and that 's' is probably a sibilant. It assumes that Voynich 'a' and '9' are equivalent. I also added an assumption that needs to be justified elsewhere: that the gallows letters usually mark the stressed syllable. Well, here are the rules, roughly: 1. The groups are based on metrical considerations, and every group tries to contain a stress or intonation 2. A break is always written after a long vowel, a nasalised vowel, or a long diphthong, unless this is the beginning of the group 3. A break is written after a stressed vowel if the following syllable can be detached and fused with the next group. In particular, this means that a sibilant (or liquid, I suspect) will act as a glide, and hold the syllable; a dental or plosive will force a break, as will a vowel that does not form a diphthong 4. An unstressed word (probably a particle) will fuse with the following group 5. Group medial 'a' is usually written with 'a'; group final 'a' with '9' Indeed, I suspect the latter to be just the former with a flourish, and suspect the (final) 'iv' and 'iiv' to be some other letter with a flourish For the test, I chose f82v (p 104 of Brumbaugh), the one with the nymphs. The marginal words, taken clockwise from due west, are as follows 1. olpaxct9 2. oqpcct89 3. oqpc89 (just to the left of the cute one) 4. 8aiivoqp9 5. oqpc8ax 6. oxlpcc89 (tiny letters above the east one) 7. oqpox89 8. olpox 9. olpc8a2 (above rainbows) 10. oqpc8ax (between rainbows) 11. oqpcc89 (below rainbows) 12. olpaolp9 13. oxlpox 14. olpaiv Now, these "words" might not be pure nouns. We might have words like "alembic", "alkahest", or even "topneuma" (which would break, by my rules, into "-topneu/ma-") Well, all I have to do, is take each name and (a) decide where it will break, (b) decide where it can fuse, (c) construct the pattern to be found in the text, and, of course, (d) find it. The following list is in the order in which I found them 1. olpaxct9 "ax" is a long diphthong, "ct" is 't', so a break is forced. The left group can fuse to a particle; the right group must fuse, because it lacks a stress or a long vowel. The '9' will change to 'a'. We are looking for -olpax/cta?-, where the '?' implies that a letter must be found. Line 22, five lines below the picture and in the same paragraph, groups 3 and 4 are 4olpax ctax Score 1. 5. oqpc8ax This will not split, since there is no long vowel and the '8' is a sibilant. It can fuse to the left. So we expect -oqpc8ax/-, the break being forced. It's not there. However, what is there, in line 1, is oqp8ax. Now Voynich 'c' might be epsilon, the short 'e', so this looks suspicious. I think that vowel is the schwa, and that it is written in transcribed text but can be elided in dictated text. The sibilant would then join to the preceding consonant, and the stress would be thrown onto the next vowel. Is this plausible? Well, isn't that just how 'Caesar' turned into 'Tsar'? And, for the elision of unstressed vowels in metrical text, we need look no further than Virgil's famous "novus seclorum ordo". 10. oqpc8ax is the same as 5. 12. olpaolp9 Another crucial test, for, by my rules, this must split. One test, then, is to prove this word is not in the text (such are the paradoxa of Voynich). And it isn't. Indeed, no group on the page, except the very first, contains more than one gallows letter. Another brownie point for the hypothesis? What we would expect, clearly, is -olp9/olpa-. Now look at line 17, groups 4 and 5 4olp9 olpaiiv 6. oxlpcc89 This cannot split after the first diphthong. It can split after the "cc", but only if the split-off "89" can find a home to the right. So we expect either -oxlpcc89 or -oxlpcc/8a?-. We find, line 14 at the right: oxlpcc89 4olpcc89 olp9 That's our word. The stress probably falls on "cc", so "89" is marked by falling intonation. And the next group has an initial fused particle - the infamous "4" - which is probably associated with rising intonation. The break then comes at the right place. And note that the "89" of the next word can't detach, since it can't fuse with "olp9" - "ao" does not form a diphthong in this scheme. 3. oqpc89 Will not split. Found in line 2, third from right. There seems to be a half space in front of it, so maybe something fused on the left. 7. oqpox89 We expect either -oqpox89/- or -oqpox/8a?-, and find (line 20, 3 from r) 4oqpax 89qpa2 This was the biggest surprise of the night. Not only did I find a medial "9", I could explain it! The scribe (I conjecture) was consciously following the prosodic rules, but he was also listening to the meaning. And, when a word ended in the middle of a group, he sometimes used the final form, by mistake. Just possibly, those 500 medial "9" symbols might be the clue to the reconstruction of the actual words. [Note: but of course the word "found" was not the word looked for! My eye misread "pax" as "pox", and the brain never caught this mistake, even while typing the note. Sigh] 8. olpox We expect -olpox/-, and find in line 4, group 7, 4olpox. Note also that the initial word of the page is qpolpox. 9. olpc8a2 This can split after the "a", if it's a long a and the next letter won't glide. Pure conjecture, of course, but we do find, in line 16 group 5 4olpc89 2c'tcc8a2 aivox9 11. oqpcc89 Again, this may split after the long "cc", so we expect either -oqpcc89/- or -oqpcc/8a?-. Line 4 has 4oqpcc89 14. olpaiv Cannot split, and cannot fuse to the right. We expect -olpaiv/-, and find, in line 24, group 7, 4olpaiv That was done by eye, last night. This morning, I searched the computer form, and duly found 13. oxlpox Cannot split nor fuse to the right, so -oxlpox/-. And how could I have missed it: line 33, next to last line, group 6. 4. 8aiivoqp9 That must split after the "v", so we seek -8aiiv/oqpa- It doesn't occur on this page, but on f83r, line 22, appears 4oqpc8aiiv oqpaiiv The lonely nymph, No 2, oqpcct89, I can't find at all. If we elide the "c", then the group oqpct89 occurs a lot. That's a strange consonant cluster, though. It's as if we were to write ORETSA and read it ORTSA. That's 11 of the 13 different words, folks, by the rules, and all 13 if you allow me my elided 'c'. Maybe I spent most of yesterday evening fooling myself, but if not, that seems to be much better than anyone has ever done before. Karl Kluge just brought round some more folio copies, so it's back to the pictures again! Robert [Note: I still do not know whether I was completely fooling myself, but, in retrospect, this was not only the hardest work I ever did on the MS, it was the moment when I felt closest to unravelling the puzzle and piercing the veil. For the first time, I thought I could discern, however dimly, some scintilla of truth through the dark glass into which I was peering. I could almost hear the sounds, and they were the sound of language, of human speech - indeed, it seemed, of sonorous, chanted, ritual speech. If only, I thought, if only I weren't so wretchedly ignorant of languages, of scripts, of phonetics and linguistics. Perhaps somewhere during this work, I held the key in my hand, and didn't even know it. And the haunting words of the Epistle come to mind again and again: "blepomen gar arti dia esoptrou en ainigmati" - "for now we see as in a glass, darkly". The Voynich manuscript is a dark glass indeed.]