Atoms and elements of the Voynichese alphabet

[Torsten]

I believe that every ending ir is a rather common error by the Scribe, caused by him mis-reading an in in the Author's sloppy handwriting as r.  (But not every r is such a "quillo"!)

Claiming that one of the most common combinations is just a steady error while trying to write an even more common combination?  That would be an assumption I can't follow.

I have statistical and other reasons to suspect that ir is a variation of iin.  Starting from the fact that frequencies of the endings ir, iir, and iiir are roughly proportional to those of iin, iiin, and iiiin (the last in each set being zero).  

The proportional frequency relationship between "iir" and "iin" that you use as evidence for scribal error is not specific to this pair. It is a universal feature of the Voynich text. Every pair of similar frequently used words shows the same proportional relationship (see Timm 2014):

chedy (501) → lchedy (119) → olchedy (38) → qolchedy (10)
shedy (426) → lshedy ( 42) → olshedy (23) → qolshedy ( 2)

daiin (863) / dain (211)
daiir ( 24) / dair (106)
 aiin (469) /  ain ( 89)
 aiir ( 23) /  air ( 74)

chol  (396) / shol  (186)
cheol (172) / sheol (114)

If proportional frequency between similar words means "scribal error," then "dain" is an error for "daiin," "shol" is an error for "chol," "lchedy" is an error for "chedy," and so on for thousands of word pairs throughout the manuscript. The entire vocabulary becomes errors.

These proportional frequencies are explained by the observation that similar words co-occur within the same contexts in the Voynich text. If "shedy" is used on the same pages as "chedy", it is hardly surprising that their word frequencies correspond. This is also what quimqu's recent Levenshtein distance 1 analysis confirmed: similar words show +95% page similarity and +123% paragraph similarity over frequency-matched controls. Similar words appear together — and their frequencies reflect that co-occurrence.

"These observations make it possible to predict the occurrence and the frequency of similarly spelled words. For instance, if it is known that 'chedy' is frequent, it is possible to predict that 'shedy' is also frequently used although less frequently than 'chedy'" (Timm 2014, p. 6).