Thanks Abu Ali, that's extremely interesting and pretty consistent with how I had understood the scope of the variations (but hard to get a sense of it when you can't read it directly). They seem to be relatively minor explanations, clarifications, and additions, and both texts seem to show similar levels of later editing from an earlier textual source.
But perhaps equally important are the *orthographic* variations, or what Sadeghi takes to be orthographic variations, which are not listed in that wikipedia link. These may give us fascinating insights into the linguistic climate when these texts were being written down, and also one man's 'orthographic variant' is another man's 'different word.' Sadeghi, being a neotraditionalist, is inclined to downplay the scope and nature of the orthographic variation. Unfortunately analyzing the significance of such orthographic variation in early manuscripts is practically impossible for non-specialists.
Looking at the more major variants, it seems that the scope of permissible scribal interpolation at this particular point in time was already relatively limited ... clearly the differences between the Sanaa I text and the so-called Uthmanic rasm are of the type that were still being worked out after the main text had been written. Additional material was used to sort of spruce things up and explain really baffling points, or to add more emphasis (like the extra 'hellfire' content in the Cairo edition 9:74, where a later scribe evidently added 'and in the hereafter' to make sure nobody misunderstands Allah's punishment as being limited to this world, but failed to modify the surrounding text to make it more grammatically and semantically coherent with the change.).
If you know Deroche's work on the other earliest Qur'anic manuscript, BNF Arabe 328, it shows somewhat similar rasm variants, although still much less than those in Sanaa I.
http://en.wikipedia.org/wiki/Codex_Parisino-petropolitanus The level of variation shown by the Cairo text and the Sanaa I text from a presumed ancestral text suggests that they were both deviating from an earlier canonical text that had not been written much earlier, and which was already seen as 'holy scripture,' but which lacked very strict control over its modification and textual supplementation. My personal guess is that the timeframe was consistent with what traditional Islam gives, meaning the main Qur'an rasm was basically finalized as a state project in the 650/660 timeframe, with what we call the Uthmanic Qur'an and the Sanaa I codex, as well as other 'companion codices,' being generated shortly after that timeframe, and then one of them (the 'Uthmanic') becoming progressively dominant and with an increasingly defined orthography, while the others were attributed to 'companions' and defined as being outside of orthodoxy. Probably the earliest Qur'an manuscript (in the sense of a true compilation of many surahs) had different surah orders and verse divisions than the Cairo text ultimately ended up with. It would have looked like a 'companion codex.' But the contents of each surah would not have looked much different at all. In so many respects, the Qur'an is remarkably conservative, and seems to lack much modification to reflect later Islamic history. That is part of what makes it so interesting.