After the double bar, the same motifs are heard in contrary motion in the two voices (see the bottom line of ex. 6). Does the top C in bar 12 relate to the voice noted as top voice in the previous bar (stems up), but appearing inside the octaves of the other voice, or does it, on the contrary, come from this other voice? The last option is the more obvious, but may not be the intended one. On the other hand, the octave A in bar 12, played with the left hand, has a sforzato and could be the one related to the previous ‘top’ voice, where the actual motif a was heard, whereas the one in octaves is a chromaticised inversion. Four bars later the voicing is reversed.
It seems an impossible choice, and the solution might be that the two motifs are indeed to be considered separate. Again, a clear difference is to be heard between the two recordings. In the late recording (at 0’44), the chord is clearly articulated, the left hand slightly more accentuated, and the inner voice in bar 11 more audible. In the early one, I was still sticking to the standard choice: to bring out G-G#-A-C as the main melody. There had been some reflection going on here in the 34 years between the two performances and, to my mind, a definite improvement.
Comparable problems of voicing are abundant in all the pieces. The most striking ones are in no. 4, described in the above article as a double canon. The ’79 recording (at 2’32) shows that I was already aware of this. Compared to the 2013 recording (at 11’22) there is not so much difference, except for a slightly different rubato, but since the latter is recorded live in a concert hall, the sound is slightly less direct. The pedalling is essential for making heard the canon between the dotted crotchets (see Ex. 3a). In the middle part of the piece, this becomes crucial to the whole texture. Many pianists, of whom a couple are heard in the seminar, make the pedal last for two bars, instead of obeying the change every bar prescribed by Brahms (see ex. 3c).
The last part of the Intermezzo in F minor is a pianistic challenge, certainly in view of the canon that goes on until the end (see ex. 3d). It is important here to get the second voice to sound. From bar 99, the opening of the piece is reiterated but in a thicker setting with crossing of the hands. The dotted crotchets in both voices still have to be brought out, which requires strength in the fifth finger of the right hand crossing over. From bar 111 (repeated in bar 119), we get another setting of this canon, with both voices in the right hand. It is still the same canon of motif b, although for practical reasons not in dotted but in plain crotchets. Finally, the ending (ex. 3e) brings precipitated statements of the inverted a2 motif that has been liquidated in the four previous bars, and the setting requires an unorthodox fingering, indicated by Brahms. The four last chords are, of course, also in canon, with the C’s as main notes.
These examples should suffice to show that the understanding of the musical text and its structural characteristics will have an impact on the performance. Therefore the excerpts from my ’79 recording contain only nos. 1 and 4. But I will mention some other examples as well.
The polyphonic character of no. 2, the Intermezzo in A major, is obvious at a first glance. Ex. 7 shows some of its most characteristic examples of advanced contrapuntal techniques: