but if Scriabin or someone like him modulates following a path different from the functional one, then you can speak of the dissolution of tonality.
That is to say... by "functional path" we can mean a logical thread, conceived according to functional attractions. The typical one is tonic-dominant-tonic, which is the typical modulation prospect of ancient and classical forms. In theory, when you reach the end of the part in the dominant, the return to the tonic is considered a predictable reaffirmation and a stronger reinforcement of the home tonality, a sort of final victory of that tonality over modulating drifts; these paths are manifold, in reality Bach as well as others modulate even to distant keys, but generally they do so at precise points and never for too long, to avoid, precisely, an excessive "negation" of the home tonality. Just think about the fact that when in analysis you try to understand a modulation, or a modulating progression, for example, you do it by reasoning not so much in terms of near keys and far keys but in terms of functions, of relationships. And of zones of the piece. Beethoven's Waldstein...
is often cited by analysis books (De La Motte, for example) because in the first measures it performs an absolutely unexpected progression. Generally, progressions at the exposition of the first theme are to the dominant, or at most to the subdominant. The sonata is in C; in the first 4 measures we have I - dominant of the dominant - V, but in the fifth measure a Bb major enters, which relative to the home C is a minor seventh (!) and which a few measures later you will discover to be a subdominant of the subdominant, i.e., IV/IV; immediately after it becomes, moreover, a minor subdominant (fam at m. 8). The "function" of a minor subdominant of the subdominant does not exist, that is, it has no tension towards the tonic; as much as it is simply a tone below, its functional path is "centrifugal," it tends to move you very far away. And indeed when at measure 13 we return to C major, after having already passed through c minor and its dominant, the piece could almost close—the effect is disruptive, an incredible opening, as if you had already traveled halfway around the world.
in Beethoven, something like this is still a beautiful device. But if instead of returning to the thirteenth measure in C you had continued to wander through other keys, the fact that the piece had that key signature would not have mattered to anyone.
A tonality makes sense if you affirm its values; so when you deny them, the listener will notice. This at least in so-called "tonal" music. It goes without saying that the path of the weakening of tonal values has been long and has passed through the deepening of other functions. It is often said that in Romanticism third relations were used a lot, i.e., modulations to "distant" keys according to the circle of fifths but "close" according to a hypothetical circle of minor and major thirds. In reality, these "affinities" are very ancient, given that they follow the bimodality of the tonal system. That is, the close relationship between C major and a minor is a third affinity, not a fifth. Through this affinity it is possible, for example, to modulate to all the main functions of the parallel mode without excessive "logical" problems. I don't know, a modulation from C to E major would "simply" be the modulation to the dominant of a minor, a little nothing that can be done in two chords. These procedures, which seem obvious when we study harmony "synchronically," are not so obvious in historical perspective, because classical modulation is not done like that for beauty's sake, but should follow a rational path of moving away from and approaching the tonic. So much so that modulations by third affinity, like many other "Romantic" modulations, are analyzed very differently from theorist to theorist. Understanding why Brahms or someone like him modulates to that key, on that chord, at that moment becomes even more difficult if you no longer have the formal box that was modulation according to the circle of fifths. And from here concepts were created such as the superposition and subposition of thirds, for example, according to which some modulating paths are nothing more than the re-proposition of a chordal structure (I don't know, a piece in A major, modulations to F# and D, and then you notice that the model is nothing but a D major chord); or the birth, as was said before, of non-functional functions, such as subdominants of subdominants, which have no attraction toward the tonic; or modulations made on fixed sounds, like those of Liszt, in which one sound of the chord remains stationary and all others move, creating apparently non-functional modulating paths. For both the composer and the analyst, modulations have become creative, let's say