So I’ve calculated the E-family in terms of length. The table shows the percentage share.
The first row is: e alone (alone)
Anchor is the letter before the ‘e’. Run is from ‘e’ to ‘eee’.
We can, of course, see a ‘step’ here, as the length of the ‘e’ runs is naturally part of the token’s length.
I then removed this by defining e = e1 / ee = e2 / eee / e3 as an atom (just as qo is an atom for me)
then the staircase disappears and the subsequent pattern becomes homogeneous
That was, of course, to be expected. But it shows something:
1. The functional unit e is a kind of ‘composition symbol’. As soon as one counts runs as a single symbol, all minimum lengths collapse to two values and all building blocks lie within the same band L*3–6.
It almost looks as though the EVA transcription has transcribed the e-runs incorrectly, they are atoms. Unless – and this is the theory behind it – the cipher is length-dependent.
2 The token architecture thus appears considerably more structured than it would be without this ‘trick’.
Now the crucial question: why didn’t the writer simply choose three different glyphs: namely e1, e2 and e3?
Because these would not have altered the length!
The same structure is found in the aiin Family...
But it gets even better...
The first row is: e alone (alone)
Anchor is the letter before the ‘e’. Run is from ‘e’ to ‘eee’.
We can, of course, see a ‘step’ here, as the length of the ‘e’ runs is naturally part of the token’s length.
I then removed this by defining e = e1 / ee = e2 / eee / e3 as an atom (just as qo is an atom for me)
then the staircase disappears and the subsequent pattern becomes homogeneous
That was, of course, to be expected. But it shows something:
1. The functional unit e is a kind of ‘composition symbol’. As soon as one counts runs as a single symbol, all minimum lengths collapse to two values and all building blocks lie within the same band L*3–6.
It almost looks as though the EVA transcription has transcribed the e-runs incorrectly, they are atoms. Unless – and this is the theory behind it – the cipher is length-dependent.
2 The token architecture thus appears considerably more structured than it would be without this ‘trick’.
Now the crucial question: why didn’t the writer simply choose three different glyphs: namely e1, e2 and e3?
Because these would not have altered the length!
The same structure is found in the aiin Family...
But it gets even better...