Tuned to Itself

There is no such thing as being in tune with a gamelan in general. You can only be in tune with a particular gamelan.

This is not a quirk. It is the organising fact of the practice, and it is worth sitting with, because almost everything a European musical education installs assumes the opposite.

A gamelan is a set of tuned bronze instruments, built together, tuned together by its maker, and treated as a unit. There is no external standard pitch to tune to. The maker tunes the set to a version of the scale system that satisfies the maker, and that version is the set's own. Take one instrument out and put it in another gamelan and it will be out of tune with everything around it, and the other gamelan is not more correct.

The reference accounts put this plainly: no two gamelans are tuned precisely alike, and each instrument is tuned to match the ensemble it belongs to rather than any outside pitch.

Two systems are in use. Slendro divides the octave into five steps that are roughly, though not exactly, even. Pelog has seven notes with markedly uneven intervals, and in practice it functions less like a seven-note scale than like a pool from which overlapping five-note subsets are drawn, with a given piece composed on one subset. Large Javanese sets are double gamelans, carrying both systems, usually sharing a pitch or two between them.

Now the part that makes the whole thing make sense.

In Bali most of the metallophones come in pairs. The two partners are not tuned to the same pitch. They are tuned a few hertz apart, deliberately, the lower called pengumbang and the higher pengisep. Struck together, as they are meant to be, they beat against each other, and the resulting shimmer is called ombak, which means wave.

The shimmer is the sound. It is designed, it is controlled, and an ensemble tuned to perfect unisons would strike its builders as dead. Most instruments are paired this way, with the ugal commonly an exception.

So the detuning that European tuning theory treats as the error to be eliminated is here the effect being engineered.

How much variation is there, really? This has been measured. A 2023 study by Sethares and Vitale built a tool from data on forty-seven complete gong kebyar sets, most of it from measurements Andrew Toth made in the 1970s. The result is not a cluster around one ideal with measurement noise. It is a spread: a family of distinct tunings, each internally coherent, none of them a failed attempt at the others.

Set that beside the European story told in the tuning lesson of this subject. Centuries of argument about whether meantone or one of the irregular temperaments or equal division was the right answer, conducted on the assumption that there was a right answer and that one tuning would eventually win. Equal temperament did win, and the cost was the thing it was always known to cost: no interval except the octave is exactly in tune, and every key sounds the same.

The gamelan evidence does not settle that argument. What it does is show that the question's central assumption, that a correct tuning exists to be found, was a choice. Forty-seven sets, forty-seven tunings, none wrong, and an entire repertory that depends on the gaps between them.