# Errata in ISO 3743

> 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-3743/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iso-3743/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 3743: 15 defects in the printed text, with the evidence and what phonometry does about each.

---

These are the 15 entries of the [errata registry](https://docs.jmrp.io/phonometry/reference/errata/) on ISO 3743-1:2010 and ISO 3743-2:2018, 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 3743-1:2010, 9.1, Equation (21) (the total standard deviation written $\sigma_\mathrm{TO}$)

- **Location:** 9.1 "Methodology", Equation (21).
- **The print:** $u(L_W) \approx u(L_J) \approx \sigma_\mathrm{TO}$.
- **The problem:** the subscript is "TO", in capitals. The sentence before
  the equation calls the quantity "the total standard deviation" and the
  sentence after it "This total standard deviation"; Equations (22) and (23)
  on the next page and the heading of 9.5 write it $\sigma_\mathrm{tot}$, and
  no $\sigma_\mathrm{TO}$ is defined anywhere in the part. ISO 3743-2:2018
  prints the same equation for its own methods, Formula (11), as
  $u(L_W) \approx \sigma_\mathrm{tot}$.
- **Evidence:** Verified on PDF page 25 (printed p. 16) of BS EN ISO
  3743-1:2010, against Equations (22) and (23) on PDF page 26 (printed p. 17)
  and the heading of 9.5 on PDF page 28 (printed p. 19) of the same
  document, and Formula (11) on PDF page 19 (printed p. 13) of ISO
  3743-2:2018.
- **Library behaviour:** the total standard deviation is the `sigma_tot` of
  `HardWalledSoundPowerResult`, $\sqrt{\sigma_{R0}^2 + \sigma_\mathrm{omc}^2}$
  by Equation (22). No change was required.
- **Status:** unreported.

## ISO 3743-1:2010, Table 3 (a one-third-octave range under an octave heading)

- **Location:** 9.4, Table 3 "Typical upper bound values of the standard
  deviation of reproducibility of the method, $\sigma_{R0}$, for A-weighted
  sound power levels and sound energy levels determined in accordance with
  this part of ISO 3743", the third row of the octave block.
- **The print:** under the heading "Octave mid-band frequency, Hz" the rows
  read 125, 250, "400 to 5 000" and 8 000, with $\sigma_{R0}$ = 3,0, 2,0, 1,5
  and 2,5 dB.
- **The problem:** 400 Hz and 5 000 Hz are not octave mid-band frequencies;
  they are one-third-octave centres. Every other row of the block is an octave,
  the whole determination is made in octave bands (3.11, 7.5), and the octaves
  between 250 Hz and 8 kHz are 500 Hz, 1 kHz, 2 kHz and 4 kHz. The table is
  ISO 3744:2010 Table 2 converted to octaves: that table is headed
  "One-third-octave mid-band frequency" and prints the same four values
  against "100 to 160", "200 to 315", "400 to 5 000" and "6 300 to 10 000".
  The first, second and fourth labels were rewritten as the octaves 125 Hz,
  250 Hz and 8 000 Hz; the third kept its one-third-octave range. ISO
  3743-2:2018 Table 5, the same table for Part 2 with its own values, labels
  that row "500 to 4 000". The title carries the same slip: it announces the
  table "for A-weighted sound power levels and sound energy levels", yet four
  of its five rows are octave bands and only the last is A-weighted. 1.4 of
  the part gives the uncertainty "for measurements made in frequency octave
  bands and for A-weighted frequency calculations performed on them"; ISO
  3744:2010 Table 2 is titled "for sound power levels and sound energy
  levels", and ISO 3743-2:2018 Table 5 "for octave band and A-weighted sound
  power levels".
- **Evidence:** Verified on PDF page 28 (printed p. 19) of BS EN ISO
  3743-1:2010, against 1.4 on PDF page 10 (printed p. 1) of the same
  document, Table 2 on PDF page 38 (printed p. 29) of BS EN ISO 3744:2010 and
  Table 5 on PDF page 22 (printed p. 16) of ISO 3743-2:2018.
- **Library behaviour:** reads the row as the four octaves it spans, 500 Hz to
  4 kHz, each with 1.5 dB
  ([`_sound_power_hard_walled.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_hard_walled.py)).
  Pinned by `test_table3_sigma_r0_per_band_and_a_weighted` in
  [`tests/emission/test_sound_power_hard_walled.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_hard_walled.py)
  and by the conformance check "ISO 3743-1:2010 Table 3".
- **Status:** unreported.

## ISO 3743-1:2010, C.4.2.5 (the altitude correction of another standard)

- **Location:** Annex C (informative), C.4.2.5 "Radiation impedance
  correction, $C_2$", the sentences that size $s_{C_2}$.
- **The print:** "For altitudes less than 500 m, no meteorological correction
  is required. At 120 m altitude and 23 °C the correction is zero and at 500 m
  altitude the correction is 0,4 dB. Assuming a triangular distribution for
  this uncertainty, the standard deviation is
  $s_{C_2} = 0{,}4/\sqrt{6} = 0{,}2\ \mathrm{dB}$."
- **The problem:** the correction of this part is the $C_2$ of Annex A,
  $C_2 = -10 \lg(p_\mathrm{s}/p_{\mathrm{s},0}) + 15 \lg[(273.15 +
  \theta)/\theta_1]$ with $\theta_1$ = 296 K and the static pressure of
  Equation (A.2). At 23 °C it gives 0.07 dB at 120 m ($p_\mathrm{s}$ =
  99.89 kPa) and 0.26 dB at 500 m ($p_\mathrm{s}$ = 95.46 kPa), not zero and
  0.4 dB. The two printed numbers are those of the sum $C_1 + C_2$ of ISO
  3741:2010 clause 9.1.4, whose $C_1 = -10 \lg(p_\mathrm{s}/p_{\mathrm{s},0}) +
  5 \lg[(273.15 + \theta)/314]$ brings the sum to 0.000 dB at 120 m and
  0.394 dB at 500 m. A comparison method has no $C_1$: it cancels between the
  source under test and the reference sound source, which is why Annex A
  prints $C_2$ alone. The divisor is right, $0.4/\sqrt{6} = 0.16$, which
  rounds to the 0.2 printed.
- **Evidence:** recomputation of Equation (A.2) and of $C_2$ from the
  constants Annex A prints ($a = 2.256\,0 \times 10^{-5}\ \text{m}^{-1}$, $b$ = 5.255 3,
  $p_{\mathrm{s},0}$ = 101.325 kPa, $\theta_1$ = 296 K), and of $C_1 + C_2$
  from ISO 3741:2010 clause 9.1.4. Verified on PDF page 41 (printed p. 32)
  and PDF page 31 (printed p. 22) of BS EN ISO 3743-1:2010.
- **Library behaviour:** implements Annex A as printed: a site at 500 m and
  23 °C gets $C_2$ = 0.26 dB in `c2` and in `sound_power_level_at_reference_conditions`
  ([`_sound_power_hard_walled.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_hard_walled.py)).
  The Annex C budget is not modelled. Pinned by
  `test_annex_a_c2_at_500_m_is_0_26_db` in
  [`tests/emission/test_sound_power_hard_walled.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_hard_walled.py)
  and by the conformance check "ISO 3743-1:2010 Annex A".
- **Status:** unreported.

## ISO 3743-1:2010, C.4.2.8 and C.4.2.9 (the humidity band of ISO 3741 quoted at 30 % instead of 50 %)

- **Location:** Annex C (informative), C.4.2.8 "Temperature, $\delta_\theta$"
  and C.4.2.9 "Humidity, $\delta_H$", the last paragraph of each.
- **The print:** C.4.2.8 "Recommended temperature and humidity ranges given in
  ISO 3741 are ±1 °C and ±3 % $H$ below 20 °C when below 30 % $H$, to a maximum
  of ±5 °C and ±10 % $H$ above 20 °C when above 30 % $H$"; C.4.2.9 "The
  recommended humidity ranges given in ISO 3741 are ±3 %, if $H \leqslant$
  30 %, to a maximum of ±10 %, if $H$ > 30 % $H$".
- **The problem:** ISO 3741:2010 Table 3 has three humidity columns, below
  30 %, 30 % to 50 % and above 50 %, and allows ±10 % only in the last; from
  30 % to 50 % it allows ±5 %. Both clauses put the ±10 % above 30 %, so the
  middle column of the table they quote is lost. ISO 3743-2:2018 prints the
  same two sentences in D.4.2.8 and D.4.2.9 with "above 50 % $H$", which is
  the table.
- **Evidence:** Verified on PDF page 42 (printed p. 33) of BS EN ISO
  3743-1:2010, against Table 3 on PDF page 18 (printed p. 9) of BS EN ISO
  3741:2010 and D.4.2.8 and D.4.2.9 on PDF page 39 (printed p. 33) of ISO
  3743-2:2018.
- **Library behaviour:** the Annex C budget is not modelled, and nothing reads
  the ISO 3741 tolerances through it. No change was required.
- **Status:** unreported.

## ISO 3743-2:2018, D.4.2.5 (the altitude correction of another standard)

- **Location:** Annex D (informative), D.4.2.5 "Meteorological correction,
  $C_2$", the sentences that size $s_{C_2}$.
- **The print:** "For altitudes less than 500 m above sea level, no
  meteorological correction is required. At 120 m altitude and 23 °C the
  correction is zero and at 500 m altitude the correction is 0,4 dB. Assuming
  a triangular distribution for this uncertainty, the standard deviation is
  $s_{C_2} = 0{,}4/\sqrt{6} = 0{,}2$ dB."
- **The problem:** the same sentence as ISO 3743-1:2010 C.4.2.5, and the same
  defect. The correction of this part is the $C_2$ of Formula (E.2), which at
  23 °C and the static pressure of Formula (E.3) is 0.07 dB at 120 m and
  0.26 dB at 500 m; zero and 0.4 dB belong to the sum $C_1 + C_2$ of ISO
  3741:2010 clause 9.1.4, and neither method of this part applies a $C_1$
  (Formula (9) and Formula (10) carry none, and Formula (D.2) lists $C_2$
  alone).
- **Evidence:** recomputation of Formulae (E.2) and (E.3) from the printed
  constants. Verified on PDF page 38 (printed p. 32) and PDF page 42 (printed
  p. 36) of ISO 3743-2:2018.
- **Library behaviour:** implements Annex E as printed, with the same
  $C_2$ as Part 1
  ([`_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_special_room.py)).
  The Annex D budget is not modelled. Pinned by
  `test_annex_e_c2_matches_part_1` in
  [`tests/emission/test_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_special_room.py)
  and by the conformance check "ISO 3743-1:2010 Annex A".
- **Status:** unreported.

## ISO 3743-2:2018, 11.3.2, Formula (14) (the total and the operating deviation added instead of subtracted)

- **Location:** 11.3.2 "Round robin test", Formula (14).
- **The print:** $\sigma'_{R0} = \sqrt{{\sigma'_\mathrm{tot}}^2 +
  {\sigma'_\mathrm{omc}}^2}$.
- **The problem:** the sentence that introduces the formula says the total
  standard deviation of a round robin "includes the standard deviation
  $\sigma'_\mathrm{omc}$" and "allows $\sigma'_{R0}$ to be determined", which
  is Formula (12), $\sigma_\mathrm{tot} = \sqrt{\sigma_{R0}^2 +
  \sigma_\mathrm{omc}^2}$, solved for $\sigma_{R0}$: the operating deviation
  comes out by subtraction. With the plus sign the method's share would exceed
  the total it is extracted from. The paragraph after it only makes sense with
  the minus sign too: Formula (14) "is imprecise if $\sigma_\mathrm{tot}$ is
  only slightly higher than $\sigma_\mathrm{omc}$" and then "provides a small
  value of $\sigma_{R0}$", which a sum can never do. ISO 3743-1:2010 prints the
  same formula, Equation (24), with the minus sign.
- **Evidence:** Verified on PDF page 20 (printed p. 14) and PDF page 21
  (printed p. 15) of ISO 3743-2:2018, against Equation (24) on PDF page 27
  (printed p. 18) of BS EN ISO 3743-1:2010.
- **Library behaviour:**
  [`reproducibility_from_round_robin`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_hard_walled.py)
  evaluates the difference for both parts, and warns when
  $\sigma_\mathrm{omc}$ exceeds $\sigma_\mathrm{tot}/\sqrt{2}$ as both parts
  ask. Pinned by `test_round_robin_reproducibility_is_the_quadrature_difference`
  in
  [`tests/emission/test_sound_power_hard_walled.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_hard_walled.py)
  and by the conformance check "ISO 3743-1:2010 Eq. 24 / ISO 3743-2:2018
  Formula 14".
- **Status:** unreported.

## ISO 3743-2:2018, D.4.2.4 and D.4.2.12 (a background contribution that is not the product it names, and a budget that does not use it)

- **Location:** Annex D (informative), D.4.2.4 "Background noise correction,
  $K_1$", the extreme and the typical contribution, and D.4.2.12 "Typical
  value for direct method, $\sigma_{R0}$", the third term of the sum.
- **The print:** "The worst case, $\overline{L'_p} - \overline{L_{p(\mathrm{B})}}$
  is 4 dB. This results in a sensitivity coefficient of $c_{K_1}$ = 0,7 and a
  total contribution to uncertainty of 1,7 dB. Typically this contribution
  will be closer to 1,0 dB due to better control of the background noise.",
  after "high background noise levels with a standard deviation of 3 dB are
  assumed"; and in D.4.2.12, $\sigma_{R0} = \sqrt{0{,}3^2 + 0{,}4^2 + 0{,}4^2 +
  0{,}2^2 + 0{,}5^2 + 0{,}8^2 + 0{,}2^2 + 0{,}2^2 + 0{,}04^2 + 0{,}1^2} =
  1{,}2$ dB.
- **The problem:** two defects. (a) The contribution is the sensitivity
  coefficient times the standard uncertainty, and the clause gives both: at a
  4 dB margin $c_{K_1} = 1/(10^{0.4} - 1) = 0.66$, printed 0.7, and
  $u_{K_1}$ = 3 dB, so the extreme contribution is 2.0 dB (2.1 dB with the
  rounded 0.7), not 1.7 dB. 1.7 is the $c_{L'_p} = 1 + c_{K_1}$ that D.4.2.3
  quotes for the same scenario. ISO 3743-1:2010 C.4.2.4 runs the same
  arithmetic at its 6 dB margin and closes it: 0.3 × 3 dB = 1.0 dB.
  (b) The third term of the D.4.2.12 sum, which stands in the position of the
  $K_1$ row of Table D.2, is 0.4 dB, the typical value ISO 3743-1:2010
  C.4.2.4 gives for Part 1; the clause of this part puts the typical
  contribution at 1.0 dB. With 1.0 dB the sum is 1.5 dB rather than the 1.2 dB
  printed.
- **Evidence:** the clause's own inputs, recomputed, read on PDF page 37
  (printed p. 31) and PDF page 40 (printed p. 34) of ISO 3743-2:2018, against
  C.4.2.4 on PDF page 40 (printed p. 31) of BS EN ISO 3743-1:2010.
- **Library behaviour:** the Annex D budget is not modelled; the library
  reports the typical upper bounds of Table 5. No number changes.
- **Status:** unreported.

## ISO 3743-2:2018, D.4.2.11 and Table D.2 (a reverberation-time coefficient that is the derivative of neither)

- **Location:** Annex D (informative), D.4.2.11 "Reverberation time,
  $T_\mathrm{nom}$", the worst-case example, and the $T_\mathrm{nom}$ row of
  Table D.2.
- **The print:** D.4.2.11: "The sensitivity coefficient, $c_T$, due to
  reverberation time is obtained from the derivative of $L_W$ [see Formula
  (9)] with respect to reverberation time. The worst case example assumes a
  source producing dominant noise at about 500 Hz. Using the minimum 0,5 s
  nominal reverberation time from 6.3 and assuming a standard deviation
  $s_T$ = 0,3 s at 500 Hz, the sensitivity coefficient is −5 dB/s and the
  worst case uncertainty contribution $u_T c_T$ = 1 dB." Table D.2:
  $c_T = -4{,}3/T_\mathrm{nom} - 240 \cdot V/(T_\mathrm{nom}^2 \cdot S \cdot c)$.
- **The problem:** Formula (9) depends on the reverberation time only through
  $-10 \lg(T_\mathrm{nom}/T_0)$, whose derivative is
  $-10/(T_\mathrm{nom} \ln 10) = -4.34/T_\mathrm{nom}$: −8.7 dB/s at 0.5 s,
  not −5 dB/s. Table D.2 adds a second term Formula (9) does not have: it is
  the derivative of the $4.34\,A/S$ of ISO 3741:2010 Equation (20) with
  $A = 55.26\,V/(c\,T)$ ($4.34 \times 55.26 = 240$), so the table
  differentiates ISO 3741 rather than the formula the clause names; for a
  $70\ \text{m}^3$ cube at 343 m/s it takes the coefficient to −10.5 dB/s. And the
  worst-case contribution does not follow from the table's own $u_T =
  \sqrt{2.42\,T_\mathrm{nom}/f + s_T^2/N_\mathrm{decays}}$ with the 120
  decays the clause calls typical: $u_T$ = 0.056 s, and 0.056 s × 8.7 dB/s =
  0.5 dB.
- **Evidence:** Verified on PDF page 40 (printed p. 34) and PDF page 36
  (printed p. 30) of ISO 3743-2:2018, against Formula (9) on PDF page 18
  (printed p. 12); the ratio of the derivative to the printed coefficient,
  8.69/5 = 1.74, was read against the page, which prints −5 dB/s.
- **Library behaviour:** the Annex D budget is not modelled. No number
  changes.
- **Status:** unreported.

## ISO 3743-2:2018, Formula (D.1) (a sum that starts at $j = i$)

- **Location:** Annex D (informative), D.3, Formula (D.1).
- **The print:** $\sigma_\mathrm{omc} = \sqrt{\frac{1}{N-1} \sum_{j=i}^{N}
  (L_{p,j} - L_{p\mathrm{av}})^2}$ dB.
- **The problem:** the lower limit of the sum is $j = i$; no $i$ is defined in
  the clause, and the $N - 1$ in the divisor is the sample standard deviation
  of all $N$ repetitions, which starts at $j = 1$. ISO 3743-1:2010 prints the
  same formula, Equation (C.1), with $j = 1$.
- **Evidence:** Verified on PDF page 32 (printed p. 26) of ISO 3743-2:2018,
  against Equation (C.1) on PDF page 35 (printed p. 26) of BS EN ISO
  3743-1:2010.
- **Library behaviour:** the standard deviation of the operating and mounting
  conditions is taken from the caller as `sigma_omc_db`; where the library
  computes one from repeated readings it sums over all of them
  ([`operating_standard_deviation`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/workstation.py)).
  No change was required.
- **Status:** unreported.

## ISO 3743-2:2018, D.4.2.8 (the frequency of the formula defined as $F$)

- **Location:** Annex D (informative), D.4.2.8 "Temperature, $\delta_\theta$",
  the where-list under the formula for $c_\theta$.
- **The print:** the formula carries $\lg(2{,}6 f)$, and the where-list
  defines "$F$ is the highest frequency significantly affecting the A-weighted
  levels".
- **The problem:** the symbol is $f$ in the formula, in Table D.2 and in the
  where-list of D.4.2.9 a few lines further on, which defines the same
  quantity as "$f$ is the highest frequency significantly affecting the A-
  weighted levels"; ISO 3743-1:2010 C.4.2.8 writes $f$ in both places.
- **Evidence:** Verified on PDF page 39 (printed p. 33) of ISO 3743-2:2018,
  against C.4.2.8 on PDF page 42 (printed p. 33) of BS EN ISO 3743-1:2010.
- **Library behaviour:** the Annex D budget is not modelled. No change was
  required.
- **Status:** unreported.

## ISO 3743-2:2018, 3.1 (the designation written "ISO ISO 3743-2")

- **Location:** Clause 3 "Terms and definitions", 3.1 "special reverberation
  test room".
- **The print:** "room which meets the requirements of Clause 6 of ISO ISO
  3743-2".
- **The problem:** "ISO" is printed twice. The definition names the document
  it stands in, which the rest of the edition calls "this document".
- **Evidence:** Verified on PDF page 8 (printed p. 2) of ISO 3743-2:2018.
- **Library behaviour:** a label the library never reads. No change was
  required.
- **Status:** unreported.

## ISO 3743-2:2018, 6.3, Figure 1 (a curve that runs above Formula (1) at the low end)

- **Location:** 6.3, Figure 1 "Values of $R$ at the one-third-octave-band
  centre frequencies for $V$ = $70\ \text{m}^3$", and the sentence before it.
- **The print:** 6.3 gives $R = 1 + 257/(f V^{1/3})$ as Formula (1) and says
  "For a room volume $V$ of $70\ \text{m}^3$, the value of $R$ is determined from Figure
  1". The curve of the figure, read against its own grid (0,2 per division),
  passes 1,66 at 100 Hz, 1,54 at 125 Hz, 1,43 at 160 Hz, 1,34 at 200 Hz,
  1,06 at 1 000 Hz and 1,02 at 2 000 Hz.
- **The problem:** Formula (1) for $70\ \text{m}^3$ gives 1.62, 1.50, 1.39, 1.31, 1.06
  and 1.03 at the same frequencies. The figure agrees with it at 1 000 Hz, the
  frequency Formula (B.2) uses, but runs 0.03 to 0.04 above it from 100 Hz to
  160 Hz and about 0.01 below it from 1 600 Hz up, four to five times the
  width of the drawn line at the low end. The figure the clause sends a $70\ \text{m}^3$
  room to is therefore not the formula the same clause states for every
  room, and a room whose reverberation time follows the figure is measured
  against limits 2.5 % higher at 100 Hz than a room that follows the formula.
- **Evidence:** Verified on PDF page 10 (printed p. 4) of ISO 3743-2:2018,
  the curve read at each one-third-octave grid line against the horizontal
  lines at $R$ = 0 and $R$ = 2, against Formula (1) on PDF page 9 (printed
  p. 3) of the same document.
- **Library behaviour:** `reverberation_parameter` and
  `check_special_room_reverberation` evaluate Formula (1) for every volume,
  $70\ \text{m}^3$ included, and take nothing from the figure
  ([`_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_special_room.py)).
  The difference at 100 Hz is a quarter of the ±10 % tolerance of 6.3.
  Pinned by `test_formula1_gives_the_1_06_of_formula_b2_for_70_m3` in
  [`tests/emission/test_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_special_room.py)
  and by the conformance check "ISO 3743-2:2018 Formula 1 / Formula B.2".
- **Status:** unreported.

## ISO 3743-2:2018, Figure B.3 (the limiting curves widened at 6.3 kHz, where 6.3 widens them above it)

- **Location:** Annex B (informative), B.5, Figure B.3 "Limiting curves for
  the ratio of the reverberation time $T$ to the nominal reverberation time
  $T_\mathrm{nom}$ for a $70\ \text{m}^3$ room".
- **The print:** 6.3 sets the limits at $0{,}9\,R\,T_\mathrm{nom}$ and
  $1{,}1\,R\,T_\mathrm{nom}$ and says "For frequencies above 6,3 kHz,
  constants 0,9 and 1,1 shall be replaced by 0,8 and 1,2 respectively". The
  two limiting curves of Figure B.3, read against the figure's own grid
  (0,2 per division), begin to part above 4 kHz and pass 1,22 and 0,84 at
  the 6,3 kHz band, where they have nearly reached the plateaus of about 1,23
  and 0,82 they keep to 10 kHz.
- **The problem:** at 6.3 kHz the text still holds the band to 0.9 and 1.1:
  with $R$ = 1.01 for $70\ \text{m}^3$ the limits are 0.91 and 1.11. The figure draws
  the wide pair at that band already, 0.1 wider on each side, so a room whose
  6.3 kHz band reads 1.15 $R$ passes against the figure and fails against the
  text. The widening belongs to the bands above 6.3 kHz, the first of which is
  8 kHz.
- **Evidence:** Verified on PDF page 28 (printed p. 22) of ISO 3743-2:2018,
  the curves read at the one-third-octave abscissae against the grid lines at
  0 and 1.8, against 6.3 on PDF page 9 (printed p. 3) of the same document.
- **Library behaviour:** follows the text. The wider tolerance starts at the
  upper edge of the 6.3 kHz band, so 8 kHz and 10 kHz take 0.8 and 1.2 and
  6.3 kHz keeps 0.9 and 1.1, and `SpecialRoomReverberationCheck.plot` draws
  the curves that way
  ([`_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_special_room.py)).
  Pinned by `test_bands_above_6_3_khz_take_the_wider_tolerance` in
  [`tests/emission/test_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_special_room.py)
  and by the conformance check "ISO 3743-2:2018 6.3".
- **Status:** unreported.

## ISO 3743-2:2018, B.5, EXAMPLE (a centring that puts the curve of Figure B.4 outside the limits)

- **Location:** Annex B (informative), B.5 "Example of determination of the
  nominal reverberation time of a room", the EXAMPLE, and the NOTE to
  Figure B.3.
- **The print:** "When the data of Figure B.4 are centred within the limiting
  curves of Figure B.3, it is found that, at 1 000 Hz, the ratio
  $T/T_{1\,000}$ = 1 corresponds to $T/T_\mathrm{nom}$ = 1,09", and so
  $T_\mathrm{nom} = 0{,}8/1{,}09 = 0{,}73$ s; the NOTE to Figure B.3 reads
  "The data of Figure B.4 are centred within the limiting curves."
- **The problem:** the curve of Figure B.4, read against its grid, gives
  $T/T_{1\,000}$ = 1.40 at 100 Hz, 1.34 at 200 Hz, 1.30 at 250 Hz, 1.24 at
  315 Hz, 1.00 from 1 000 Hz to 2 000 Hz and 0.80 at 10 kHz. Scaled by 1.09
  and divided by the $R$ of Formula (1) for $70\ \text{m}^3$, the bands from 200 Hz to
  400 Hz exceed 1.1, the 250 Hz band reaching 1.13: centred as printed, the
  example room fails the requirement of 6.3 it illustrates. The 1.09 is the
  midpoint of the extreme ratios $q = (T/T_{1\,000})/R$ taken as if 0.9 and
  1.1 held at every band, $2/(q_{\max} + q_{\min})$ with the 0.80 of 10 kHz as
  the minimum, although 6.3 allows 0.8 there; the centring that uses each
  band's own limits puts every band inside at $T/T_\mathrm{nom}$ = 1.05,
  $T_\mathrm{nom}$ = 0.76 s. Figure B.3 does not draw Figure B.4 scaled by
  1.09 either: its data curve sits at 1.09 from 1 000 Hz to 2 000 Hz but at
  1.46 at 100 Hz, 1.37 at 250 Hz and 0.95 at 10 kHz, where 1.09 times
  Figure B.4 gives 1.52, 1.41 and 0.87.
- **Evidence:** Verified on PDF page 28 (printed p. 22) and PDF page 29
  (printed p. 23) of ISO 3743-2:2018, both curves read at the
  one-third-octave abscissae against their grids, with $R$ from Formula (1)
  on PDF page 9 (printed p. 3); the ratios are a recomputation from those
  readings.
- **Library behaviour:** `check_special_room_reverberation` centres
  $T_\mathrm{nom}$ by minimising the largest deviation of any band from the
  ideal curve, each measured against its own tolerance; on the curve of
  Figure B.4 it returns 0.76 s and qualifies the room, 0.17 dB from the
  printed 0.73 s in the $L_W$ of Formula (9). A value given in
  `nominal_reverberation_time_s` is checked as given, and the printed 0.73 s
  fails at 250 Hz
  ([`_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_special_room.py)).
  Pinned by `test_the_b5_example_centres_the_room_within_its_curves` and
  `test_the_printed_1_09_of_b5_is_the_narrow_midpoint_and_fails_250_hz` in
  [`tests/emission/test_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_special_room.py)
  and by the conformance check "ISO 3743-2:2018 B.5 EXAMPLE".
- **Status:** unreported.

## ISO 3743-2:2018, F.1 (Annex F tied to the direct method while 10.4 sends the comparison method to it)

- **Location:** Annex F (normative), F.1 "A-weighted sound power levels", the
  sentence above Formula (F.1) and the where-list under it.
- **The print:** "The A-weighted sound power level, $L_{WA}$, determined from
  octave band sound power levels according to 10.2 shall be calculated from
  Formula (F.1)", and "$L_{Wk}$ is the sound power level in the $k$th octave
  band determined according to 10.2".
- **The problem:** 10.2 is the direct method. The only clause of the document
  that calls Annex F is 10.4, "A-weighted sound power levels determined by the
  comparison method", which sends to it the octave bands "made ... according
  to 10.3". Read as printed, the annex covers the method that never refers to
  it and excludes the one that does. Annex E, the next one, writes "from the
  formula in 10.2 or the formula in 10.3".
- **Evidence:** Verified on PDF page 43 (printed p. 37) of ISO 3743-2:2018,
  against 10.4 on PDF page 19 (printed p. 13) and Annex E on PDF page 42
  (printed p. 36) of the same document.
- **Library behaviour:** Formula (F.1) forms `sound_power_level_a` from the
  octave bands of both methods, as Annex E is written for both
  ([`_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/emission/_sound_power_special_room.py)).
  Pinned by `test_comparison_a_weighted_total_is_annex_f` and
  `test_a_weighted_level_by_formula9_directly` in
  [`tests/emission/test_sound_power_special_room.py`](https://github.com/jmrplens/phonometry/blob/main/tests/emission/test_sound_power_special_room.py).
- **Status:** unreported.
