Beneath the elegant, aristocratic surface of Saint-Saëns’ music there often run deeper currents: of wit and irony, tenderness and pudeur, even occasionally of despair and disappointment. The way through that surface is, first of all, to be absolute master of it technically, but then one has to identify the particular current in question. Not easy, because the music often changes mood almost from bar to bar. The six artists playing his music here and on the disc chosen for comparison have technique to spare. The currents prove more problematic.

Sarah Chang makes a strong, vibrant sound – no wishy-washy pseudo-classicism here. She’s sparing with her portato, using it for just a few special moments, her virtuoso bowing at the end of the First Sonata is very exciting, and the work as a whole comes over as a big, Romantic one. But I don’t find much subtlety here, few hidden currents. Ulf Wallin does not command Chang’s tonal warmth – in truth, there are quite a few moments of scratchiness, especially in the scales of the last movement – and again accuracy is not quite enough. Neither violinist can match Philippe Graffin’s way of caressing, even bending a phrase so that it seems to speak: to a large extent this is due to his wide range of attacks (though that’s entirely the wrong word), drawing our attention to key notes within the phrase. And the Hyperion disc also includes two pieces not found on the CPO one.

Graffin and Devoyon do take a few liberties with the tempo in the last movement of the First Sonata, but entirely to the music’s benefit (let’s face it, this movement is just a shade on the long side). What Saint-Saëns’s music does not need is inconsistency in dealing with textures. Why does Lars Vogt in the first movement produce sudden staccato bass notes under what may have been Proust’s ‘petite phrase’? And why in the second does Pöntinen fiddle about with the pedalling (0'17"), when the composer’s instruction is absolutely clear?

I take issue with Vogt again in the Franck Sonata, where he frequently interrupts the flow with ‘expressive’ rubato; and the second movement is marked merely Allegro, not Presto – this is scrambly, lacking either Ashkenazy’s poise, at a slightly slower speed, or Cortot’s miraculous ability to bring out the central voice at a practically identical one. In general, Chang and Vogt agree on a rather hard-driven interpretation, with little of the gentle grace that can, in particular, act as a foil to the rigour of the technically astonishing canon in the last movement.

The Ravel sonata shows Chang and Vogt at their best. The clean ambience suits this piece well (that on the EMI disc is too plummy and the piano at times too forward), as does the players’ straightforward approach. I don’t know that there are many undercurrents in this piece: Ravel was interested mainly in exploiting the differences between the two instruments and with that in mind actually tore up what he felt was a more ‘musical’ finale. But at the end of the first movement Chang’s vibrant tone makes the ultimate arrival of a long legato line a magical moment and her final high G is flawless. Pity about the Franck.