Just a hoax?

[Jorge_Stolfi]

Also vertical pair repeats occur statistically less that would be expected

Some of those "LAAFU" ("line as a functional unit") effects may be due to the word length bias created by the trivial line breaking algorithm.  

Basically, when any text, in any language, is formatted into paragraphs, the first word of each line is noticeably longer than average, while the last few words are noticeably shorter.

If the average word lengths are different, then the word frequencies at those places are different.  

Since character and digraph frequencies depend on their occurrence in the most common words, then character and digraph frequencies in the first word and in the last few words of the line will be different than average too.

This rather unexpected effect of paragraph formatting was observed here only recently.  Does it completely explain the "LAAFU" anomalies, or only part of them?  I don't know; but claims of LAAFU should check for that possiblity.

One possible check is to re-format Voynichese paragraphs with very different margin widths, and see whether the new lines still show the claimed LAAFU anomalies.  If the VMS actually has LAAFU, the anomalies should disappear.

All the best, --stolfi