The sixteen waltzes of Brahms's op. 39 (1865) manage only six direct resolutions of the dominant ninth chord, four of them in nearly the final bars of the last number. The two earlier examples are both obscured by registral shifts. (Direct resolutions, recall, are external, that is, the ninth over V is resolved in the following I, not before.)
Here is waltz no. 16; the score here lacks only the last phrase and cadence to C# minor.
The voice-leading is impeccable, though I should note that most composers before the 1890s do try to avoid the parallel fifths that can easily arise in resolving the major dominant ninth chord. The invertible counterpoint (cf. bars 1-8 with 9-16) is Brahmsian, not Schubertian!
The resolution in no. 7--see the boxes in the third system--has to be called direct but F#4 goes to E5, not E4, and the contrast is strong: the phrase before is a long descent (starting in the middle of the second system) and the phrase beginning with E5 is an equally strong ascent. I've checked the other versions (piano four-hands and two pianos) and neither has a simple resolution, though it would have been very easy indeed to add the extra notes.
The F#4 is repeated in the left hand (second box) and does move to E4 but the effect is a muddle of A major and D major, made worse if the pedal is held down. The pedal marking, incidentally, is in the manuscript. Note that the composer deleted several before that (arrow).
Two other points of interest in this score: (1) In bars 4-5, what I call an "almost direct resolution", where unfolded thirds clearly show the voice-leading as C#6-B5 and A5-G#5; (2) an expressive highpoint in the final cadence--a cliché of the waltz going all the way back to Schubert's generation--and an internal resolution, if one regards the entire bar as V and, even better, holds the pedal down (which I have never done; the "clean" sound of the higher register ii6 is more appealing).
In a subsequent post I will provide examples of the other internal resolutions and evasions.
