It was indeed unclear, and I opened a thread over 10 years ago (missing an even older one on the same topic) to ask for a clarification. For some reason, I am convinced that a clarification has been made since then, in the sense that you can get spells that are more effective than otherwise possible, but a quick search cannot locate it and I am too lazy to do more 
Update: I could find indirect evidence in support of my thesis. It's in HoH:TL p.23. The example folio has versions of several corebook spells, such as Ball of Abysmal Flame, at the same level as in the corebook, that were included in the Folio "because the inventing magi used experimentation, and
each spell has a beneficial side-affect". Note that adding an extra effect to an Hermetic spell would normally add at least a magnitude, but that's not the case for these spells. To me this seems strong, though perhaps not definitive, evidence that when arcane experimentation modifies a spell, the level (and Arts) stay unchanged.
I still do have the nagging impression that everything was made clearer somewhere.