Nobody Agrees How Fast It Goes

Maelzel's metronome arrived around 1815 and Beethoven adopted it immediately, going back to add markings to symphonies already written.

Almost nobody plays them.

Many of Beethoven's markings are substantially faster than performance tradition, some to the point where conductors have called them unplayable. Explanations have been offered for two centuries: a faulty machine, a misreading of the scale, his deafness, a habit of marking the tempo he imagined rather than one an orchestra could achieve. Recent work has proposed that he may have read the device wrongly. None of it is settled, and the alternative, that we have been playing Beethoven too slowly since the 1830s, is entirely live, and is what conductors who take the markings literally are testing in public.

The dispute is a good way into the larger point: tempo is not information carried by the score. It is a judgement.

A number of beats per minute says nothing about articulation, about how much time is taken at a phrase end, about whether the pulse is felt in twos or fours, or about how much the hall will smear fast passages. Two performances at identical markings can feel entirely different, and two performances at markings twenty per cent apart can both be convincing.

The variation in practice is large. Recordings of the standard repertoire differ in total length by well over twenty per cent, and there are long-term trends: nineteenth-century music has generally slowed across the twentieth century while eighteenth-century music has got faster, which suggests that what changed is taste rather than evidence.

Rubato is the other half of the subject. The word means stolen time, and in the classical conception you take time in one place and pay it back in another, so the underlying pulse survives the theft. Romantic practice is freer, and early recordings show conventions that have disappeared entirely: heavy sliding between notes, chords arpeggiated that are written together, the pianist's hands deliberately not synchronised, melody drifting ahead of accompaniment.

To modern ears much of this sounds like error. It was standard, taught, and admired, within living memory of the composers concerned. We know only because recording arrived in time to catch it, and that is worth pausing on, because it means performance conventions we experience as timeless may be a century old, and we have no equivalent evidence about any period before 1900.

Which is the interesting problem with historically informed performance.

The movement recovers real information: period instruments with gut strings and different bows, smaller ensembles, conventions of ornamentation and articulation drawn from treatises written by people who were there. The result changed the sound of eighteenth-century music profoundly: lighter, faster, more transparent, far less vibrato.

It is also a twentieth-century movement with tastes of its own, and the preference for clarity, speed and lack of vibrato coincides suspiciously well with modernist taste generally. The strong form of the claim, that this is how the music sounded, cannot be verified, and Richard Taruskin's argument that period performance is a thoroughly modern style in period costume has never been properly answered.

Both things are true. The evidence is real, the interpretation of it is contemporary, and a performance is always a document of the moment it was made as well as of the piece.