Twinned Glyphs

[Emma May Smith]

It might be worth, rather than trying to explain the whole script as having a single system, laying out the functional relationships between individual pairs. Some relationships may be strong, others may not exist. Those relationships which are strong can then become a base of evidence to build a more informed theory about the script as a whole.

You could build a "functional similarity index" which rates each glyph against another according to how similar they behaved within words. The index could be built on factors such as the glyphs which occur before and after, or whether replacement of one glyph with another results in a valid word, or a number of other measures. This could be displayed as a map/digram showing differing levels of similarity to compare with the graphical evidence, allowing revision of your theory. (This has, in fact, been done before. But revisiting the idea is worthwhile if it's a genuine attempt at collecting new insights.)

(By the way, I think that looking at the design of the script and the way the glyphs work and interact in the text is a very useful line of research. Please don't take anything I say as discouraging. Only that it's better to be able to say one small thing securely than lots of big things uncertainly.)