Apologies, I had omitted to factor in the shift test, which reduced the 4 x 10^3 down to approximately 2.8 x 10^3 under the broad field analysis. I’ll have to come back and correct this later.

Posted by: @shaqmeister

↑

Then, from the “broad field” analysis, we have, say, 4 x 10^3 matching stems, which corresponds to around 1 in every 3.8 x 10^4, or roughly 0.003%.

So, this needs correcting. It should be:

From the “broad field” analysis, we have 2.8 x 10^3 matching stems which also lead to at least one English solution. This, then, should correspond to approximately 1 in every 5.5 x 10^4, or roughly 0.002%.

Posted by: @lendor-77

↑

To explain more clearly: for each candidate word, we could check whether every letter can be assigned to a different page of the magazine, using only the actual available correspondences.

For example, if a word such as “DEMONI” emerges, the test would no longer be interpretative, but purely combinatorial:

does there exist, or not, a configuration in which:
D comes from one page,
E from another,
M from another again,
etc., without reusing the same page?

If I am not misunderstanding what you mean by “using only the actual available correspondences,” then I would be confident that this should be easily possible, as in the following alternative selection example:

Here, I have used only those pages which have been used on the envelope itself and, if I remember correctly, I don’t believe that I have even introduced any new words from any page, as I had not needed to consult the edition while I was putting this together.

Hopefully this is free of gross mistakes, but I am tired today, so we’ll have to see!

Hi @shaqmeister! I am reporting the results obtained with Codex on the analysis of the Latin words found within the envelope sequence, both in the broad field and narrow field searches, applying the constraint that every letter of the base word must be assignable to a different magazine page without reusing the same page.

In the broad field search, 4,002 Latin bases compatible with the envelope sequence were identified. Of these, 2,754 produce at least one English solution after the shift. Applying the additional combinatorial constraint that all pages must be different reduces the number to 1,444 bases, corresponding to 3,528 total shift solutions.

In the narrow field search, 753 complete Latin words compatible with the envelope sequence were identified. Of these, 505 produce at least one English solution after shift. Applying again the constraint of all different pages reduces the number to 281 words, corresponding to 655 total shift solutions.

So, clearly, this is quite an interesting statistical analysis, although I still need to better understand how to interpret it and especially how to use it correctly within the broader reasoning! To be honest, I am not even sure I have fully understood yet where exactly you wanted to lead me with your observation XD

There is also a further aspect that should always be emphasised regarding the analysis above: in the Latin words identified, the letters of the sequence do indeed come from different pages, but this does not necessarily imply that only one letter was used from each page. Multiple characters may still have been cut from the same page in order to construct the rest of the envelope text.

To be honest, I think that I have lost sight of what all this analysis is trying to move towards, so it’s probably best not to follow too closely anything I might have suggested. 🤣 😬