# Errata in ISO 4869

> 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-4869/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iso-4869/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 4869: 3 defects in the printed text, with the evidence and what phonometry does about each.

---

These are the 3 entries of the [errata registry](https://docs.jmrp.io/phonometry/reference/errata/) on ISO 4869-2:2018, ISO 4869-6:2019 and ISO 4869-3:2007, 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 4869-2:2018, Table C.1 (the reprint that disagrees with the table it reprints)

- **Location:** Annex C (informative), Table C.1, "A-weighted octave-band sound
  pressure levels, $L_{p,\mathrm{A}f(k)i}$, **from Table 2**", PDF page 17
  (printed p. 11), against the normative Table 2 it names, PDF page 11
  (printed p. 5).
- **The print:** the two tables carry the same eight reference noises over the
  same seven octave bands, and seven of the eight rows agree digit for digit.
  The sixth reads 82,0 / **89,3** / **93,3** / 95,6 / 93,0 / 90,1 / 83,0 in
  Table 2 and 82,0 / **89,4** / **93,5** / 95,6 / 93,0 / 90,1 / 83,0 in
  Table C.1. The 250 Hz and 500 Hz cells differ; nothing else does.
- **The problem:** Table C.1 states in its own caption that it comes from
  Table 2, so one of the two is wrong, and the annex's own results say which.
  Formula (15) applied to the sixteen attenuation values of Table A.1 with
  Table 2's row reproduces all sixteen $PNR_{j6}$ of Table C.2 exactly; with
  Table C.1's row, thirteen of the sixteen fall 0.1 dB short. Table 2 is
  therefore the reading the worked example was computed from, and it is also
  the normative one, Table C.1 being an informative reprint. An implementer
  who takes the reference spectra from Annex C, where they sit next to the
  worked example, gets a protector's $H$ and $M$ values a tenth of a decibel
  low.
- **Evidence:** Formula (15), PDF page 11 (printed p. 5), evaluated on
  Table A.1, PDF page 15 (printed p. 9), against the sixth row of Table C.2,
  PDF page 18 (printed p. 12), all of ISO 4869-2:2018.
- **Library behaviour:** `HML_REFERENCE_NOISES` carries Table 2. The test
  suite computes the same row from Table C.1's values and asserts that it
  misses thirteen of the printed sixteen, so the two readings can never be
  silently swapped.
- **Status:** unreported.

## ISO 4869-6:2019, Table A.3 (the uncertainty rows are formed from the rounded row above them)

- **Location:** Annex A (normative), Table A.3 "An example of ANR earmuff
  active insertion loss test data in dB for a given laboratory", the rows
  "Combined standard uncertainty, $u$, ($\sigma/\sqrt{N}$)" and "Expanded
  uncertainty, $U_{95}$", PDF page 16 (printed p. 10), against the definitions
  of A.1 and A.2 on PDF page 14 (printed p. 8).
- **The print:** the table gives the active insertion loss of sixteen subjects
  at the octave frequencies 63 Hz to 8 kHz, then their mean, their standard
  deviation $\sigma$, $u$ and $U_{95}$. The $u$ row reads 0,5 / 0,2 / **0,4** /
  0,5 / 0,4 / 0,4 / 0,4 / 0,2 dB and the $U_{95}$ row **1,0** / 0,4 / **0,8** /
  **1,0** / **0,8** / 0,8 / **0,8** / **0,4** dB. A.2 defines $u$ as "the
  standard deviation of the individual active insertion loss data divided by
  the square root of the number of test subjects, i.e. $\sqrt{16} = 4$", and
  A.1 defines $U_{95}$ as $u$ multiplied by the coverage factor $k = 2$. The
  table carries no note on how it rounds.
- **The problem:** from the sixteen printed rows, the mean and $\sigma$
  reproduce in all sixteen cells, but $u$ at 250 Hz is 0.348 dB, which rounds
  to 0.3 and not 0.4, and $U_{95} = 2\sigma/4$ is 0.935 / 0.355 / 0.697 /
  0.940 / 0.711 / 0.839 / 0.720 / 0.325 dB, which rounds to 0.9 / 0.4 / 0.7 /
  0.9 / 0.7 / 0.8 / 0.7 / 0.3: six of the eight printed cells are 0.1 dB high.
  Every printed cell is instead the formula applied to the **rounded** row
  above it: $1.4 / 4 = 0.35$ prints as 0.4, and each $U_{95}$ is twice the
  $u$ printed over it. The same table in ISO 4869-1:2018 (its Table A.3, same
  layout) computes at full precision and says so in its NOTE 2, "All
  calculations are made with full precision before rounding to one decimal",
  and all 28 of its derived cells reproduce that way. A reader who applies A.1
  and A.2 to the printed data of ISO 4869-6 gets an expanded uncertainty a
  tenth of a decibel below the printed one in six bands of eight, and cannot
  tell from the page why.
- **Evidence:** the sixteen rows and four derived rows of Table A.3, PDF page
  16 (printed p. 10), recomputed with $u = \sigma/4$ and $U_{95} = 2u$ once at
  full precision and once from the printed $\sigma$ and $u$; A.1 and A.2, PDF
  page 14 (printed p. 8); all of ISO 4869-6:2019. For the contrast, ISO
  4869-1:2018 Table A.3 and its NOTE 2, PDF page 19 (printed p. 13).
- **Library behaviour:** `hearing.active_insertion_loss` returns $u$ and
  $U_{95}$ at full precision from the data, as A.1 and A.2 define them. The
  conformance row and `tests/hearing/test_active_noise_reduction.py` pin the
  mean and $\sigma$ rows as printed, reproduce the $u$ and $U_{95}$ rows the
  way the table forms them, and assert that full precision differs in exactly
  the seven cells named here, each by 0.1 dB.
- **Status:** unreported.

## ISO 4869-3:2007, 5.2.2 and Table 1 (a microphone index of exactly 5 dB in two rows)

- **Location:** 5.2.2, the paragraph on the directional test, and Table 1,
  printed p. 4.
- **The print:** 5.2.2: "the range of sound pressure levels at the reference
  point shall be within 5 dB for any two directions of measurement of the
  incident sound energy when measured with a directional microphone with a
  front-to-random sensitivity index of at least 5 dB. For other directional
  microphones, the relationship between the front-to-random sensitivity index
  and the allowable field variations is given in Table 1." Table 1 prints the
  rows "> 5" (5 dB), "4 to 5" (4 dB) and "< 4" (microphone not suitable).
- **The problem:** a microphone whose index is exactly 5 dB may read 5 dB by
  the text, which asks for "at least 5 dB", and 4 dB by the table, whose
  first row excludes 5 and whose second includes it. The allowable variation
  of the one microphone the text names is therefore not decided by the page.
  ISO 8253-2:2009, which prints the same directional test in 5.3 b), writes
  the first row of its own Table 1 as "≥ 5".
- **Evidence:** the table read against the sentence that introduces it.
  Verified on PDF page 8 (printed p. 4) of ISO 4869-3:2007, and on PDF page
  13 (printed p. 7) of ISO 8253-2:2009.
- **Library behaviour:** follows the text.
  [`RANDOM_INCIDENCE_VARIATION_LIMITS`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/hearing/_earmuff_insertion_loss.py)
  allows 5 dB from an index of 5 dB and 4 dB from 4 dB up to below 5 dB, and
  the conformance check "ISO 4869-3:2007 5.2.2, Table 1" pins the row for
  5 dB. A field that meets 4 dB passes either reading.
- **Status:** unreported.
