Marius Comper RO

Bucharest, 21 November 1972 · twenty pipes measured

The Romanian panpipe sounds lower than the law says

On 21 November 1972 in Bucharest, Romania's capital, Gottfried Habenicht measured a nai, a Romanian panpipe, built with 20 pipes by an instrument maker named Urucu: the inside diameter of each pipe and the depth from its opening to the wax. By the textbook formula every pipe would sound above its written note, by 73–272 cents (a semitone has 100). For the notes to come out, about 0.80 of the diameter has to be added to each pipe in the calculation. The lower lip, partly covering the opening, is one of the ways a player brings the pitch down.

Move the correction and watch the markers under the pipes.

At 25 °C, with the reference A at 440 Hz, 0.80 gives the closest overall match to the written notes. It would be 0.76 at 20 °C and 0.85 at 30 °C.

Press a pipe and drag across the others

The sheet of 21 November 1972

Urucu made the nai from 20 bamboo pipes. He had lived in Bucharest before moving to Craiova, a city in southern Romania. Stelian Bizgu, a 14-year-old from Bucharest, had played it for about two months. The ethnomusicologist Gottfried Habenicht drew the instrument and recorded in millimetres, for each pipe, the outside and inside diameters at the opening, the approximate outside length including the holder, and the depth from the opening to the beeswax.

Above the table are the notes, written an octave below their sound: from B to G, a major third short of three octaves, in G major, with F sharp. The sheet appeared in 1984 in the German woodwind journal Tibia, at the end of a study of the nai by the Romanian ethnomusicologist Tiberiu Alexandru.

The textbook law

In a pipe closed at one end, the wavelength of the fundamental note is roughly four times the length of the air column. The air just beyond the opening vibrates too, so the pipe sounds as if it were slightly longer. The length added in the calculation is called the end correction; for a pipe with a fully free opening, with nothing around it, it is about 0.30 of the inside diameter.

With that value the formula gives every one of Urucu's pipes a higher note than the written one: by 73 cents for the closest and 272 for the furthest, from three quarters of a semitone to more than two and a half semitones.

What the calculation misses

The correction that brings the 20 pipes closest to their written notes is 0.80 of the inside diameter, nearly three times the textbook figure for a free pipe. With it, the typical error falls to 59 cents, and 12 of the 20 pipes come within half a semitone of their written notes. The calculation does not say where the difference comes from. The player blows across the opening and partly covers it with the lower lip, and a smaller opening makes a pipe sound like a longer one; the rim of each opening is also rounded towards the lips.

The correction each pipe would need to sound exactly its written note varies from pipe to pipe, from 0.53 to 1.24 of the diameter: on average 0.96 on the seven lowest pipes and 0.65 on the six highest. On pipe 20 a millimetre of wax moves the note by nearly two thirds of a semitone; on the seven lowest pipes, by 10–18 cents. Rounding in the measurements is too small to explain most of the errors, and their cause remains open.

Alexandru writes that by lifting the panpipe slightly the player lowers a note: D becomes D flat. Lifting it gradually while blowing into a single pipe can lower the sound by as much as an octave, a portamento. In the 1930s some lăutari, professional musicians of Romanian traditional music, still put lead shot, pebbles, beans or breadcrumbs into pipes; the air column got shorter and the note rose, from C to C sharp, say.

0.20.40.60.81.01.21.4textbook 0.30fitted 0.8015101520pipe
The end correction each pipe would need to sound exactly its written note, as a share of its bore; the vertical bar shows how far the sheet's rounding moves it. The low pipes need more than the high ones.

The intervals and the octave

The air temperature and the pitch standard the instrument was tuned to are unknown. For a given correction, the intervals between pipes do not depend on them. In G major, starting on B, the semitones fall after pipes 1, 5, 8, 12, 15 and 19. With both the textbook correction and the best fit, five of those six steps rank among the six smallest intervals. Between pipes 19 and 20 the calculation gives nearly a whole tone: the depths are 25 and 22 mm, and rounding to the millimetre can shift that interval by about 60 cents. That alone does not explain the gap either.

The sheet carries an octave sign above the stave, and the lengths confirm it: to sound B at the written pitch, the longest pipe would need an acoustic length of about 35 cm, while its air column measures 15.9 cm and the whole pipe, with the holder, about 21.5 cm.

10020030012345678910111213141516171819after pipe
The interval from each pipe to the next, in cents, from the measured depths. Amber: where the scale calls for a semitone (100 cents); blue: a whole tone (200). The last semitone, after pipe 19, comes out the size of a whole tone.

A panpipe with wide pipes

In a paper published in 2005, the physicist Neville Fletcher described an Eastern European panpipe and found, for its pipes, a rule linking inside diameter to pipe length. At the same length, Urucu's pipes have inside diameters 23–65% greater than the rule gives, 44% in the middle. The six shortest pipes are shorter than any Fletcher measured, so there the rule is extended; within his range the difference is 23–56%. At the same depth, a wider pipe has a larger end correction, so the same change in the coefficient moves its note further.

5101550100150Fletcher 2005Urucuair depth, mm · bore, mm
Each pipe's inner bore against its air depth: the dots are Urucu's pipes, the line is Fletcher's rule.

Share the panpipe

The link keeps the end correction you chose.

Sources and method

Each pipe's frequency is calculated as f = c / [4 (L + α d)], where c is the speed of sound, L the depth of the air column down to the wax, d the inside diameter and α the end-correction coefficient. The calculation uses the speed of sound at 25 °C, 346.1 m/s; the written notes are given equal-tempered frequencies with the reference A at 440 Hz. Both are assumptions. α is the value that minimises the sum of squared errors in cents; the typical error is their root mean square. The bars in the chart assume the sheet was rounded to ±0.5 mm in depth and ±0.25 mm in diameter. The wax in the drawing has a schematic thickness.

  1. T. Alexandru, „Die rumänische Panflöte. Monographische Skizze”, Tibia 1984, nr. 1, pp. 20–30; „Instrumentenzettel von G. Habenicht, Bukarest, am 21.11.1972”, p. 30. moeck.com The measurements of the 20 pipes, the notes, the drawing; lifting the instrument and the lead shot of the 1930s (p. 27). The original measurement sheet gives an inside diameter of 12 mm for pipe 9.
  2. N. H. Fletcher, „Stopped-pipe wind instruments: Acoustics of the panpipes”, J. Acoust. Soc. Am. 117 (2005), 370–374. unsw.edu.au The rule d = 1.32 L0.43 (in mm) for the panpipe he measured; the lip and the air jet.
  3. H. Levine, J. Schwinger, „On the radiation of sound from an unflanged circular pipe”, Phys. Rev. 73 (1948), 383. doi.org The end correction 0.6133 r for an unflanged pipe at low frequencies.
  4. J. Wolfe, UNSW, frecvențele notelor. unsw.edu.au Equal temperament, reference A at 440 Hz.