Is that normal?

[JoJo_Jost]

Anagrams of each line? Or of each page?

Global

Word shuffled-lines (same words, different order) would test this hypothesis. Not exactly because we already know that there are positional preferences for glyphs in lines that can nudge a little the probability of a vertical alignment. If the word structure does not matter then it's the next thing to test and that's more difficult.

Yes, I’d already thought of checking that too, but I was just too tired yesterday. Here’s the revealing result:

   

(It’s worth noting, however, that columns 1–4 are also influenced by the labels in Astro.)

So, strictly speaking, these graphs demonstrate once again – in a completely different way – just how highly structured the tokens in VMS are. Anyone who assumes a simply substituted natural language is refuted by this alone

But whilst I was doing that, I noticed something else that’s far more interesting and which has to do with the effect. More on that later.