Piano Forum

I think this is actually the most "strange" and perhaps also most interesting part of this movement. At the point you mention, m. 31-32 Mozart suggests us by using the dominant 7th chord B^b7 that we come back to the home key, E^b major, but he surprises us completely with this progression to G^bmajor, and, as you state, afterwards destabilizes the whole part through a series of sequencing modulations in the style of a development section. The relation between B^b7 and G^b is what I know as "third relationship", i hope this is the correct English term. B^b and E^b serve as pivot notes. This turn to G^b major has to me somehow the character of a deceptive cadence. M32, last quarter: G^b: V7.  M 34: G^b. I-(V7): (M.35) II (V7): (M.36) III Now instead of going on with the sequence (the next step would be (V7):IV C^b major, Mozart leads us to c minor by using the dominant G major.
M. 37.  c: i- V- G: ii7- #IV⁶ with a diminished third(A^b) and
                                    ⁵
diminished 7th (E^b)
So in m. 38 we are in the dominant of c minor, G. but since now he wants to go back to E^b major he uses in m. 39 c: (VII⁷) i⁶ but this i⁶  does not appear, instead it appears a E^b: (V²) (F7) : V7(B7), the thirtysecond run in m. 40 is on the dominant of E^b major.

I hope this makes at least a bit of sense, it is pretty complicated The chord on the last quarter in m. 37 is a typical Mozart chord which even has a particular name but I forgot this And the English name might be different anyway.

Okay hope this helps, it's my first attempt of doing such a complex harmonical analysis in English. It's fun hopefully perhaps someone else will correct what I said, if it's wrong.

Yikes.  I love analysis but I had a hard time parsing this one out.    I think your English is fine, and I am sure you are more or less right about everything.  I will try and state it in simpler words.

But first the last chord in m.37 is an augmented sixth chord, which almost always prepares a cadence.  In this case, Mozart frustrates a possible cadence to c minor by having a G major (V of c) flirt with a diminished-7 on F# (vii/V in C), finally dispelling the tension by lowering the G pedal point until it reaches the B-flat, the true dominant.

measures 24 - 40 are the "B" section, with a beautiful melody that foreshadows the second movement to the "Pathetique" sonata (even in the same key in m.24).  Harmonically, these bars explore the chromatic tensions implied in the ending of the first phrase (m.6). 

M.29 should be understood in the key of A-flat major as V/V, rather than V/I as if in E-flat major.  The reason is, when Classical composers wanted to modulate to a new key, they always used V/V to strengthen the modulation.  Although m.24 started clearly in A-flat major, that is not enough for Classical composers, who have to reaffirm it with cadences, departures and returns.

In this case, he continues the movement within the circle of fifths; bar 30 is, in A-flat major, V/V/V (F/B-flat/E-flat).  The seeming resolution in the middle of the bar is undermined with the iv-I figures which mirror the uncertain hesitations of bar 6, and suggest that e-flat minor is coming up.

G-flat major, in 32, comes as a deceptive cadence to e-flat minor, if only because of the minor Plagal cadences that happened for 3 beats (not really enough time  in this case to modulate).  But look at it this way: in m.24, he didn't bother modulating either; the beautiful effect is gained by the suddenness of the new harmony, not the preparation.

Therefore I think you are justified in calling m.32 I in Roman numerals; also you can call m.24 I.  It's not so common for Mozart, but there you have it, and the effect is positively Schubertian. :>

Starting in m.34, Mozart uses angstlich chromaticism to modulate up whole steps, creating a wonderful tension through paradox (bass moving resolutely down in chromatic steps, while harmonies are going up in whole steps).  This is sort of where the notion of key center breaks down.  You could concievably notate these harmonies all according to e-flat minor, the key that was supposedly frustrated earlier by the deceptive cadence, but that is unsatisfying, because of the G-natural in m.36, and the incompatibility of c minor in m.37 (it could be notated as v/ii in e-flat, but what's the point?  Audibly, it doesn't suggest that at all.)

My advice is to notate these all according to E-flat major.  They don't suggest any tonal center at all, and they don't have a harmonic direction in any normal sense.  Mozart ingeniously composed these rising-step modulations like a long passage, perhaps a long tunnel, or even a long staircase, that one must pass through, suggesting poetically a time of tribulation.  The flirtation of G major and f-# diminished chords over a G pedal further increases the nature of the trial, which is finally dispelled by the cadenza-like passage in m.40, bringing us safely back to E-flat major.

These then would be all modified harmonies according to E-flat major: m.34 ^bIII, m.35 ^biv, m.36 v, m.37 vi.  M.38-41 are easy to notate in E-flat major.

I hope that is clear.  To sum up, I don't believe there is a tonal center per se, and so if you want to use Roman numerals (you don't have to for this section), keep them only according to the home key.

Walter Ramsey