Alexander Melnikov’s excellent Prokofiev sonatas series concludes with three of the most modest of the nine. Not that ‘modest’ is generally a good fit with Prokofiev. But it does at least apply to the eight-minute duration of the single-movement Nos 1 and 3, both of which are based on pre-Op 1 ideas. I was somehow expecting Melnikov to take a more headlong approach to the Scriabinesque flights of No 1, and he would obviously be fully capable of that. But his relatively restrained tempos, with plenty of subtle yielding in the lyrical contrasting sections, bring their own rewards. So, too, in No 3 the tumult of the tarantella-like main theme is never allowed to gain the upper hand over lucidity, and the benefits are felt when textures accumulate without detriment to clarity or the basic tempo.

The Fifth Sonata was the only one to be composed during Prokofiev’s years in the West, and it is in some ways the hardest to make convincing – whether in its 1923 original form or, as here, in the 1952 53 revision. Melnikov’s crispness again pays high dividends. He does not allow the first movement’s curious Allegro tranquillo marking to lead him into complacency, and the high level of intensity he sustains serves the music well. He clearly senses that the second movement emanates from the same piquant, stylised world as the ‘Blues’ middle movement of Ravel’s Violin Sonata, which was composed not long afterwards (and which is – Prokofiev lovers forgive me – a good deal more memorable). Without quite making the case for Prokofiev’s finale as a worthy companion to his Soviet-era sonatas yet to come, Melnikov does succeed in bringing out some of its proto-epic qualities.

Melnikov’s Visions fugitives are bursting with character, but perhaps not so conspicuously as to make you want to drop everything to hear them. I’ve heard more hurtling tempos in some of the faster pieces (such as the No 5 Molto giocoso) and more searching accounts of some of the more poetic ones (such as No 18).

The piano sound is not of the richest – anything in the top octave sounds rather tinny. All in all, this is a worthy conclusion to a fine series but perhaps not one that taken in isolation makes urgent claims on the collector.