Enola & Facktor
Facktor Facktor
I just mapped the letter frequencies in the Voynich Manuscript and found some curious clusters. Want to take a look?
Enola Enola
Sure, share the data. I’ll line up the frequencies, look for repeating motifs, and see if they line up with any known cipher patterns or botanical diagrams. Let's see if the clusters reveal any hidden structure.
Facktor Facktor
Here’s the raw letter‑frequency list from the 1,500‑page sample (alphabetical order, counts per 1000 letters): A = 78, B = 15, C = 22, D = 31, E = 95, F = 18, G = 28, H = 40, I = 84, J = 3, K = 11, L = 55, M = 37, N = 62, O = 88, P = 25, Q = 2, R = 71, S = 66, T = 93, U = 21, V = 14, W = 6, X = 4, Y = 9, Z = 1. The bigrams that exceed 5 per 1000 are: TH = 12, RE = 9, IN = 8, ON = 7, AN = 7, AT = 6. If you want a visual plot or a comparison to Latin or botanical glyphs, just say the word.
Enola Enola
Interesting numbers. The high E, T, O, N, I, A values look like a language, not pure noise. The low J, Q, V, X, Y, Z are similar to English but a bit lower; maybe the manuscript has a different vowel‑heavy pattern. The bigrams TH, RE, IN, ON, AN, AT match English quite well. That cluster could be a clue that the text follows a Romance‑language structure, or at least uses similar digraphs. If you plot it against a Latin frequency chart, we’ll see whether the ratios are closer to Latin or to modern English. I'd like to overlay the Voynich frequencies with a 17th‑century Dutch sample next—just to see if the clusters match that period. Let me know if you have the plot ready.
Facktor Facktor
I pulled a quick bar chart script for you—just run it in Python and feed the two dictionaries I gave you earlier. The script normalizes each dataset to per‑10,000 letters so the ratios line up nicely. Once you plot them side by side, look at the relative drop from E down to Z; that slope will tell you whether it leans more towards Latin or 17th‑century Dutch. Let me know if the curve feels off and we can tweak the smoothing window.
Enola Enola
Got the script. I’ll run it and generate the two bars. The slope from E to Z should be steep if it’s Latin, flatter if Dutch. Once I see the curves I’ll note where the drop deviates from the expected trend. I’ll tweak the smoothing window if the curve looks too noisy. Let me know if you want me to compare it against any other corpus, like early modern English.
Facktor Facktor
Give it a try with early modern English too—those frequencies sit between Latin and modern English, so they might sit right on the edge of the curve you’ll see. That’ll help you spot if the Voynich data is drifting toward a later dialect or sticking with the older vowel‑heavy patterns. Just tell me what the curves look like.
Enola Enola
The Voynich curve drops sharply from E to Z, much steeper than early modern English, which is more gradual. Compared to Latin it’s closer to Latin’s drop but still a little flatter—so it’s not purely Latin. The early modern English curve sits between the two, with a moderate slope. The Voynich plot sits just above the Latin line but below the early modern English line, suggesting the manuscript’s language is more vowel‑heavy, like an older Romance, yet it carries some English‑like digraphs that make it drift a touch toward the later dialect. The curves overlap in the middle letters (N, O, R, S), but the tail end diverges the most. Overall, the Voynich data seems to lean toward an older, Latin‑influenced pattern, with some hints of English.