The whereabouts of the manuscript for Schubert's String Quintet, D956 are unknown and there are apparently no surviving sketches either. The earliest source, therefore, is the first edition which was published in 1853 by Spina, 25 years after the work's composition. Two new recordings of this work, together with my chosen comparison, offer three very different versions which derive principally from choices concerning the exposition repeat in the first movement and the repeats in the Scherzo, which the New Schubert Edition does little to clarify. In the absence of a primary source or other documentary evidence, then, the evaluation of these differences must be made wholly on musical grounds.
The Takacs Quartet take the least controversial approach: they repeat everything as directed. Their performance, which is fluent and spontaneous, has some beautiful moments such as the first movement retransition, or the return to the elaborated opening theme in the Adagio. However, the smoothness of their phrasing, which does have its virtues, sometimes obscures the music's rhythmic definition. The rather hollow recorded sound does not help, and there is a disturbing blemish in bar 361 in the first movement.
The Brandis benefit from better recording, and their stronger rhythmic characterization generates much greater drive and energy. In the first movement exposition repeat, on the second statement they omit the dominant seventh chord in bar 154 which surely does lead the music back to the C major opening, rather than towards the A major of the beginning of the development. Their quicker tempo in the slow movement, one and a half minutes faster than the Takacs, also gives this movement more animation. Moreover, the Brandis's decision to repeat from bar 186 in the second half of the Scherzo also makes good sense, since the remaining bars clearly function locally as a coda.
The Alban Berg's 1983 recording was chosen for comparison not only for its excellent performance and recorded sound, but also for its strategy with respect to repeats which is different from both the versions here under review. The Berg omit the first-movement exposition repeat entirely, but retain the C major dominant seventh chord and, in the Scherzo, their decision to omit the repeat of the second half as the solution of the 'coda problem' seems less compelling than the Brandis's insightful version.
On grounds of structural logic, then, the Brandis's account comes out best. However, the Takacs disc has the additional lure of the Quartettsatz which is superbly played and excellently recorded. Ultimately, choice will be determined by preference for an interpretation in which either the heart or the mind have the greater influence.'