# Errata in ISO 3382

> One page of the phonometry documentation, as Markdown. Index: https://docs.jmrp.io/phonometry/llms.txt

Canonical: https://docs.jmrp.io/phonometry/reference/errata/iso-3382/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iso-3382/index.md
Version: phonometry 4.0.0
Install: pip install "phonometry[full]"
Author: José Manuel Requena Plens (https://orcid.org/0000-0003-1250-6212)
Published: 2026-10-28
Updated: 2026-11-16T21:36:14Z
License: MIT (https://github.com/jmrplens/phonometry/blob/main/LICENSE)
Summary: Errata in ISO 3382: 5 defects in the printed text, with the evidence and what phonometry does about each.

---

These are the 5 entries of the [errata registry](https://docs.jmrp.io/phonometry/reference/errata/) on ISO 3382-1:2009, in the order the registry lists them. The registry's introduction says how each claim is checked against the printed page and what the status of an entry means.

## ISO 3382-1:2009, A.2.1 (the same symbol names two different levels, a page apart)

- **Location:** Annex A (informative), A.2.1. The "where" list under
  Equations (A.2) and (A.3) on printed folio 13 (PDF page 21), and the "where"
  list under Equation (A.5) on printed folio 14 (PDF page 22).
- **The print:** folio 13 gives "$L_{pE}$ is the sound pressure exposure level
  of $p(t)$", with $p(t)$ "the instantaneous sound pressure of the impulse
  response measured at the measurement point", that is, the receiver in the
  hall under test. Folio 14, inside NOTE 1, gives "$L_{pE}$ is the
  spatial-average sound pressure exposure level measured in the reverberation
  room".
- **The problem:** one symbol, two quantities, same subclause, no
  distinguishing subscript and no note that the symbol has been reused. The
  second is a calibration of the source in a laboratory; the first is the
  measurement the whole annex exists to make.
- **Consequence:** substituting (A.5) into (A.1) as the symbols are printed
  gives

  $$
  G = L_{pE} - L_{pE,10} = L_{pE} - \left[ L_{pE} + 10 \lg (A/S_0) - 37 \right]
    = 37 - 10 \lg (A/S_0)\ \text{dB},
  $$

  in which the hall has vanished and the sound strength depends only on the
  absorption area of the reverberation room the source was calibrated in. The
  substitution is what the printed symbols invite, and it is nonsense.
- **Evidence:** Verified on PDF pages 21 and 22 (printed pp. 13 and 14) of
  BS EN ISO 3382-1:2009.
- **Library behaviour:**
  [`reverberation_room_reference_level`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py)
  names its argument `reverberation_room_level`, and the hall's own level
  never reaches it: it is measured by
  [`sound_strength`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py) from the response
  passed as `ir`. Nothing stops a caller writing the substitution out by hand,
  but no single variable plays both roles, and the two names say which is
  which.
- **Status:** unreported.

## ISO 3382-1:2009, A.2.1 (a directivity survey "at every 12,5 degrees" that does not close the circle)

- **Location:** Annex A (informative), A.2.1, the note immediately under
  Equation (A.4), printed folio 13 (PDF page 21).
- **The print:** "When making such a measurement in a free field, it is
  necessary to make the measurement at every 12,5° around the sound source and
  to calculate the energy-mean value of the sound pressure exposure levels in
  order to average the directivity of the sound source."
- **The problem:** $360 / 12.5 = 28.8$. There is no whole number of
  12.5° steps that closes a turn: 28 steps reach 350° and leave a 10° gap, 29
  steps overshoot to 362.5°. The instruction cannot be followed literally.
- **Consequence:** two laboratories that both "measure every 12,5°" can use
  different bearing sets, and for a source at the Table 1 directivity limit
  (±6 dB at 4 kHz) their energy means differ. The reference level $L_{pE,10}$
  that every route in A.2.1 leads to is therefore not reproducible from the
  printed instruction alone. The same standard's own source-qualification
  survey in 4.2.1 uses 5°, which divides 360 exactly into 72.
- **Evidence:** Verified on PDF page 21 (printed p. 13) of
  BS EN ISO 3382-1:2009.
- **Library behaviour:**
  [`directivity_energy_average`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py) takes
  the reading the note can support: a uniform sampling of the full turn no
  coarser than the printed step, so at least
  $\lceil 360 / 12.5 \rceil = 29$ bearings, combined as the energy mean the
  note asks for. Fewer bearings raise `ValueError` rather than averaging a
  turn that was never closed.
- **Status:** unreported.

## ISO 3382-1:2009, C.2.1 and C.2.2 (the prose of both stage supports leaves out an integration limit)

- **Location:** Annex C (informative), C.2.1 on printed folio 23 (PDF
  page 31) and C.2.2 on printed folio 24 (PDF page 32), each in the sentence
  that introduces its own equation.
- **The print:** C.2.1 defines the early support as "the ratio, in decibels,
  of the reflected energy **within the first 0,1 s** relative to the direct
  sound", and prints

  $$
  \mathrm{ST}_\mathrm{Early} = 10 \lg \left[
      \frac{\int_{0,020}^{0,100} p^2(t)\ \mathrm{d}t}
           {\int_{0}^{0,010} p^2(t)\ \mathrm{d}t} \right]\ \mathrm{dB}.
  $$

  C.2.2 defines the late support as "the ratio, in decibels, of the
  reflected energy **after the first 0,1 s** relative to the direct sound",
  and prints

  $$
  \mathrm{ST}_\mathrm{Late} = 10 \lg \left[
      \frac{\int_{0,100}^{1,000} p^2(t)\ \mathrm{d}t}
           {\int_{0}^{0,010} p^2(t)\ \mathrm{d}t} \right]\ \mathrm{dB}.
  $$

- **The problem:** neither sentence describes the equation beside it.
  Equation (C.1) starts at 0.020 s, not at the 0.010 s the direct-sound
  window ends at, so the interval between them is counted in neither the
  numerator nor the denominator and the prose never mentions the gap.
  Equation (C.2) stops at 1.000 s, where the prose puts no upper limit at
  all.
- **Consequence:** both move a number, and the first moves it further. On an
  exponential decay of $T = 2$ s, a reader who takes "within the first 0,1 s"
  to start where the direct-sound window ends collects the 10 ms to 20 ms
  interval as well, which is 17 % more energy and **0.68 dB** on
  $\mathrm{ST}_\mathrm{Early}$, against the 1 dB standard deviation C.2.4 estimates
  for a single reading. The missing ceiling of (C.2) costs 0.01 dB in the
  same hall, because a 2 s decay is already 30 dB down at one second, and
  reaches 0.2 dB at $T = 4$ s and 1.0 dB at $T = 8$ s: it is the cathedral,
  not the concert hall, that the second omission separates.
- **Evidence:** Verified on PDF pages 31 and 32 (printed pp. 23 and 24) of
  BS EN ISO 3382-1:2009.
- **Library behaviour:**
  [`stage_support`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py) integrates the
  printed limits, which are the ones in
  [`EARLY_SUPPORT_WINDOW_S`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py) and
  `LATE_SUPPORT_WINDOW_S`. `tests/room/test_auditorium_stage.py` drops an
  arrival into the gap and another past the ceiling and requires both to
  change nothing.
- **Status:** unreported.

## ISO 3382-1:2009, Table 1 and A.4 (the same limits are called maximum in one clause and minimum in the other)

- **Location:** the caption of Table 1 and the paragraph of 4.2.1 above it,
  printed folio 3 (PDF page 11), against the fourth paragraph of A.4,
  printed folio 19 (PDF page 27).
- **The print:** 4.2.1 says "Table 1 lists the **maximum** acceptable
  deviations from omnidirectionality when averaged over 'gliding' 30° arcs
  in a free sound field", and the table's own caption reads "Table 1 —
  **Maximum** deviation of directivity of source in decibels for excitation
  with octave bands of pink noise and measured in free field". A.4 says "If
  the source directivity is close to the **minimum** limits given in Table 1,
  the measurement should be repeated with the source turned in at least
  three steps totally."
- **The problem:** one table, two opposite words for what its numbers are.
  The values are ceilings, as their own caption and 4.2.1 both say, and A.4
  calls them a floor.
- **Consequence:** A.4's sentence is the one that tells a laboratory when to
  do extra work, and read as printed it says the opposite of what it means.
  A source "close to the minimum limits" would be a near-perfect one, which
  is the case that needs no repetition at all; what A.4 is asking for is the
  repetition of a survey whose source only just clears the ceiling, because
  that is where the orientation of the source starts to matter to the
  answer. A reader who takes the word literally repeats the measurement for
  the wrong sources and skips it for the right ones.
- **Evidence:** Verified on PDF pages 11 and 27 (printed pp. 3 and 19) of
  BS EN ISO 3382-1:2009.
- **Library behaviour:**
  [`MAX_SOURCE_DIRECTIVITY_DEVIATION_DB`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py)
  and [`source_directivity_limit`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py) carry
  them as the maxima their own caption makes them. The three-orientation
  repeat of A.4 is a procedure, not a computation, and the library does not
  implement it.
- **Status:** unreported.

## ISO 3382-1:2009, 4.2.1 (a gliding average whose window has no stated phase)

- **Location:** 4.2.1, the paragraph immediately above Table 1, printed
  folio 3 (PDF page 11).
- **The print:** "Table 1 lists the maximum acceptable deviations from
  omnidirectionality when averaged over 'gliding' 30° arcs in a free sound
  field. In case a turntable cannot be used, measurements per 5° should be
  performed, followed by 'gliding' averages, each covering six neighbouring
  points."
- **The problem:** six 5° points cover 30° of arc read as six sectors, and
  25° read as the span between the first and the last, so the two sentences
  agree only under the sector reading. More to the point, nothing says where
  those six points sit relative to the arc they average: a window may lead
  its bearing, trail it, or straddle it, and the clause does not choose. Nor
  does it say how the six are combined, although the reference they are
  compared with is explicitly "a 360° energetic average".
- **Consequence:** over a full turn the six-point windows are one cyclic set
  whichever end of its arc a window is pinned to, so the phase moves the
  bearing each deviation is reported against by up to half a window, 15° of
  the pattern, and leaves the deviations themselves alone. For a source near
  its Table 1 limit that is still what decides whether the largest deviation
  is reported on a lobe or between two of them, which is the orientation A.4
  then asks to be turned and measured again. The other two silences do move
  the number: the span reading and the combination law both change what an
  arc averages, so two laboratories that follow the clause can report
  different maximum deviations for one source, and the standard gives no way
  to tell which of them read it right.
- **Evidence:** Verified on PDF page 11 (printed p. 3) of
  BS EN ISO 3382-1:2009.
- **Library behaviour:**
  [`gliding_directivity_deviation`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/room/auditorium.py)
  takes the sector reading, averages the arcs energetically as the reference
  is, and starts each window at the bearing it is reported against, wrapping
  round the turn. Its docstring says all three choices are choices.
- **Status:** unreported.
