# Errata in IEC 60268

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

Canonical: https://docs.jmrp.io/phonometry/reference/errata/iec-60268/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iec-60268/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 IEC 60268: 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 IEC 60268-1:1985, IEC 60268-3:2013, IEC 60268-16:2011 and IEC 60268-7:2010, 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.

## IEC 60268-1:1985, Appendix A, Figure A1 (last shunt capacitor printed as 41.47 nF)

- **Location:** Appendix A, "Noise weighting network and quasi-peak meter",
  Figure A1 "Weighting network" (printed p. 29), drawing 0641/85. The French
  print of the same artwork (printed p. 28) carries the same value as
  “41,47 nF”.
- **The print:** the last shunt capacitor of the ladder, the one across the
  600 Ω amplifier input, is labelled **41.47 nF**.
- **The problem:** it should be **31.47 nF**, which is what ITU-R BS.468-4
  Figure 1a prints for the same network. Every other element of Figure A1
  matches BS.468-4 Figure 1a exactly: 600 Ω source, 13.85 nF, 12.88 mH,
  26.82 nF, 33.06 nF, 9.21 nF, 26.49 mH and Z = 600 Ω. The intended reading is
  not in doubt, because the document contradicts itself: evaluated against
  Table AI, printed two pages earlier in the same annex, the 31.47 nF ladder
  reproduces all 21 rows to a maximum of **0.050 dB** and violates no
  tolerance, while the 41.47 nF ladder is out by up to **2.252 dB** (at
  31 500 Hz) with a root-mean-square error of 1.055 dB and **breaks Table AI's
  own tolerance column at seven frequencies**, every one from 8 000 Hz to
  20 000 Hz: −0.40 dB against ±0.40 at 8 kHz, −0.74 against ±0.60 at 9 kHz,
  −1.16 against ±0.80 at 10 kHz, −1.85 against ±1.20 at 12.5 kHz, −1.98
  against ±1.40 at 14 kHz, −2.05 against ±1.60 at 16 kHz and −2.12 against
  ±2.00 at 20 kHz. Sweeping the capacitor to minimise the error against
  Table AI lands on 31.4798 nF.
- **Evidence:** the two ladders evaluated independently by an ABCD chain
  product over the seven printed reactive elements between the printed 600 Ω
  source and load, normalised at 1 kHz, and compared row by row with Table AI
  and its tolerance column. Neither Amendment 1:1988 (which replaces Table AII
  only) nor Amendment 2:1988 (which replaces sub-clause 12.1, on producing a
  uniform alternating magnetic field) touches Figure A1, so the misprint
  stands in the current document as amended. Verified on PDF page 31 (printed
  p. 29) and PDF page 29 (printed p. 27), which carries Table AI, of IEC
  60268-1:1985, and on PDF page 1 (printed p. 1) of Recommendation ITU-R
  BS.468-4.
- **Library behaviour:** unaffected. The weighting network is built from the
  BS.468-4 Figure 1a component values in
  [`filters/weighting.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/filters/weighting.py), with
  31.47 nF, and the Table 1 rows are the oracle. The entry matters because IEC
  60268-3:2013 sub-clause 14.12.11 sends a reader to "a weighting network
  complying with Appendix A of IEC 60268-1", so a clean-room implementation
  started from IEC 60268-3 lands on the wrong capacitor.
- **Status:** unreported.

## IEC 60268-1:1985, Appendix A, Table AII (lower-limit row slipped one column)

- **Location:** Appendix A, Table AII, the tone-burst dynamic characteristic
  of the quasi-peak meter, "Limited values — lower limit" row (printed p. 31).
- **The print:** the lower limit (%) row reads
  `13.5 | 22.4 | 34 | 41 | 44 | 44 | 50 | 68` for the 1, 2, 5, 10, 20, 50, 100
  and 200 ms columns, while the (dB) row printed immediately beneath it reads
  `−17.4 | −13.0 | −9.3 | −7.7 | −7.1 | −6.0 | −4.7 | −3.3`.
- **The problem:** the 50 ms and 100 ms cells contradict their own dB cells.
  −6.0 dB is 50.1 %, not 44 %, and −4.7 dB is 58.2 %, not 50 %. The percentage
  row has slipped one column to the right from 50 ms onwards, carrying the
  20 ms and 50 ms values into the two cells after them; the dB row and the
  200 ms cell stayed where they belong. ITU-R BS.468-4 Table 2 prints
  `... | 44 | 50 | 58 | 68` for the same four columns.
- **Evidence:** the two rows of the same table read against each other, and
  against the corresponding row of ITU-R BS.468-4 Table 2. **Corrected by
  Amendment 1:1988-01**, whose English sheet is headed "Page 31 / Replace
  Table AII by the following:" and prints the lower limit row as
  `13.5 | 22.4 | 34 | 41 | 44 | 50 | 58 | 68`, matching BS.468-4; every other
  cell of the replacement table is identical to the base print, so this row is
  the entire substantive content of the amendment. Verified on PDF page 33
  (printed p. 31) of IEC 60268-1:1985, on PDF page 3 (printed p. 3) of IEC
  60268-1:1985 Amendment 1:1988, and on PDF page 4 (printed p. 4) of
  Recommendation ITU-R BS.468-4.
- **Library behaviour:** unaffected. The eleven acceptance windows are
  transcribed from BS.468-4 Tables 2 and 3 in
  [`tests/reference_data/`](https://github.com/jmrplens/phonometry/blob/main/tests/reference_data), which agree with the
  amended IEC table. Recorded because the unamended base document is the one a
  reader is likely to hold, and it widens the 50 ms and 100 ms acceptance
  windows by 1.1 dB and 1.3 dB at the bottom.
- **Status:** unreported (corrected by the issuing body in 1988).

## IEC 60268-1:1985, Appendix A, Table AI (16 000 Hz tolerance printed as ±1.65)

- **Location:** Appendix A, Table AI, the tolerance column, 16 000 Hz row
  (printed p. 27; the French Tableau AI on printed p. 26 prints the same
  value).
- **The print:** `±1.65 1)`.
- **The problem:** ITU-R BS.468-4 Table 1 and AES17-2015 Table 1 both print
  **±1.6** for the same row, and the table's own footnote 1) is what settles
  it: the marked tolerances "are obtained by a linear interpolation on a
  logarithmic graph on the basis of values specified for the frequencies used
  to define the mask, i.e. 31.5 Hz, 100 Hz, 1 000 Hz, 5 000 Hz, 6 300 Hz, and
  20 000 Hz". Interpolated on that rule between (6 300 Hz, 0 dB) and
  (20 000 Hz, ±2.0 dB), 16 000 Hz gives 1.6137 dB, which rounds to 1.6 at one
  decimal and to 1.61 at two. No rounding of the rule produces 1.65, and no
  alternative anchor pair does either: taking the 6 300 Hz to 31 500 Hz line
  instead gives 1.6216 dB. The value is also anomalous within its own column,
  which is quoted to one decimal everywhere else.
- **Evidence:** the footnote rule applied to all 14 marked rows of the same
  column, which reproduces every one of them (63 Hz 1.400, 200 Hz 0.8495,
  400 Hz 0.6990, 800 Hz 0.5485, 3 150 and 4 000 Hz 0.5000, 7 100 Hz 0.2070,
  8 000 Hz 0.4136, 9 000 Hz 0.6175, 10 000 Hz 0.7999, 12 500 Hz 1.1863,
  14 000 Hz 1.3825, 31 500 Hz 2.7865) and 16 000 Hz alone disagrees with what
  is printed. Not corrected by Amendment 1:1988 or Amendment 2:1988. Verified
  on PDF page 29 (printed p. 27) of IEC 60268-1:1985 and on PDF page 2
  (printed p. 2) of Recommendation ITU-R BS.468-4.
- **Library behaviour:** none needed. The tolerance mask is taken from
  BS.468-4 Table 1, and the realised digital curve is held to a far tighter
  bound than the mask anyway: the mask governs a measuring instrument
  comprising the amplifier and the network, not a filter's departure from the
  nominal curve.
- **Status:** unreported.

## IEC 60268-3:2013, clause 14.12.9.2 f) (DIM denominator)

- **Location:** clause 14.12.9.2, item f) (printed p. 39), the formula for the
  dynamic intermodulation distortion $d_\text{DIM}$.
- **The print:**
  $d_\text{DIM} = (\sum_{i=1}^{9} {U'_i}^{2})^{1/2} / U_2 \times 100\ \%$.
- **The problem:** the denominator is one of the nine terms of its own
  numerator. Table 2 of the same clause (printed p. 38) defines $U_2$ as the
  intermodulation component at $f_s - 2f_q = 8.70\ \text{kHz}$, and item d)
  defines $U_1, U_2, \ldots U_i$ as exactly those components, so the sum
  $i = 1\ldots9$ runs over $U_1 \ldots U_9$ and includes $U_2$. Meanwhile the
  defining clause 14.12.9.1 states the ratio of the r.m.s. sum of the Table 2
  intermodulation product voltages "to the amplitude of the output voltage at
  the frequency f_s", i.e. the 15 kHz sine component $U_s$, the Otala
  convention, and item d) measures "the amplitudes of the sinusoidal signal
  $U_s$" precisely so that it can be used, which the f) formula then never
  does. The denominator should be $U_s$. An earlier revision of this entry
  said that "U2 is used throughout 14.12 for the total output voltage"; that
  is false, in both the English and the French print.
- **Evidence:** Table 2, item d) and item f) read together in both language
  columns of the bilingual edition; the historical DIM literature (Otala)
  defines the ratio to the sine amplitude. Verified on PDF page 41 (printed p.
  39), PDF page 40 (printed p. 38), which carries Table 2, and PDF page 102
  (printed p. 100), which carries the same item f) in the French column, of
  IEC 60268-3:2013.
- **Library behaviour:** follows the 14.12.9.1 definition (reference = the
  output amplitude at $f_s$), with a code comment at the reference measurement
  in [`distortion.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/distortion.py).
- **Status:** unreported.

## IEC 60268-16:2011, Table M.1 (the beta row states the wrong redundancy term)

- **Location:** Annex M, Table M.1 "Example calculation", step 4, the row
  labelled "Sum of beta\*$MTI$ = $MTI_k$ $\times$ beta weighting" (printed
  p. 67), directly below the matching alpha row.
- **The print:** the label reads $MTI_k \times \beta_k$, and the seven cells
  of the row read 0,059 | 0,052 | 0,045 | 0,008 | 0,037 | 0,081 | 0,000,
  summed on the next page as $\sum \text{beta*}MTI = 0{,}282$.
- **The problem:** the label and the cells state different quantities, and the
  label is the one that is wrong. Clause A.5.6 of the same edition (printed
  p. 47) defines the index as
  $STI = \sum_{k=1}^{7} \alpha_k \times MTI_k -
  \sum_{k=1}^{6} \beta_k \times \sqrt{MTI_k \times MTI_{k+1}}$: the redundancy
  term is the *geometric mean of two adjacent bands*, not the band's own
  $MTI_k$. Read against the table's own MTI row, $\beta_k \times MTI_k$ gives
  0.062 | 0.051 | 0.044 | 0.008 | 0.036 | 0.076, summing to 0.277, which
  disagrees with five of the six printed cells and with the printed total.
  $\beta_k\sqrt{MTI_k MTI_{k+1}}$ reproduces all six cells and the 0.282
  total. The seventh cell is not part of either reading: the redundancy sum
  stops at $k = 6$ because the 8 kHz band has no band above it to pair with,
  so its 0.000 is the placeholder of a column with no redundancy partner, not
  a term. The alpha row above it, labelled the same way, is
  correct, because there the label and A.5.6 do agree.
- **Evidence:** both readings recomputed from the table's own step 4c MTI row
  and compared cell by cell with the printed row and with its printed total;
  A.5.6 read against the label. The defect does not move this example's
  answer, since $1.040 - 0.277 = 0.763$ and $1.040 - 0.282 =
  0.758$ both print as the STI 0.76 the table ends on, which is how a label
  contradicting the normative formula survives a worked example. Verified on
  PDF page 69 (printed p. 67) and PDF page 49 (printed p. 47) of
  IEC 60268-16:2011.
- **Library behaviour:** implements A.5.6 with the redundancy term as printed
  there, in
  [`_index_from_corrected_mtf`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/speech/sti.py); the pairwise
  weighting-factor test of A.2.2 pins it independently, and the conformance
  rows "IEC 60268-16:2020 A.2.2" and "IEC 60268-16:2011 Annex M" both read
  the index it produces.
- **Status:** unreported.

## IEC 60268-16:2011, Table M.1 (I_k tabulated a million times its neighbours)

- **Location:** Annex M, Table M.1, the row "Combined squared sound pressure
  $I_k$, MPa$^2$" of step 2 (printed p. 64) and of step 3 (printed p. 65),
  read with the $I_{am,k}$ and $I_{rt,k}$ rows below it.
- **The print:** for the 77,9 dB signal of the 125 Hz band, step 2 prints
  $I_k$ = 61,7, and four rows below it prints $I_{rt,k}$ = 40 000 for the
  46 dB reception threshold of the same band.
- **The problem:** two defects in one row. The unit is impossible: at 77.9 dB
  re 20 µPa the squared sound pressure is $0.0247\ \text{Pa}^2$, so the cell
  cannot be 61.7 MPa$^2$ under any reading of the prefix. What the row
  actually tabulates is the dimensionless intensity ratio
  $10^{L/10} = 61\,722\,596$ divided by $10^{6}$. And that divisor is not
  applied to the two quantities the standard adds to $I_k$ in the very next
  rows: $I_{am,k}$ and $I_{rt,k}$ are tabulated as the plain ratio, 40 000
  being $10^{4.6} = 39\,811$ rounded, undivided. A reader who forms
  $I_k + I_{am,k} + I_{rt,k}$ from the cells as printed understates its first
  term by $10^{6}$. The printed "adjustment to remove masking and
  threshold" row is the check: 1.019 at 500 Hz is
  $(I_k + I_{am,k} + I_{rt,k})/I_k$ only once $I_k$ is restored to
  26 305 192; formed from the cells as printed the same expression reads
  19 279.
- **Evidence:** every cell of both $I_k$ rows recomputed as $10^{L/10}$ from
  the combined levels printed above them, and every cell of the $I_{am,k}$ and
  $I_{rt,k}$ rows recomputed as $amf_k \times I_{k-1}$ and $10^{ART_k/10}$;
  the first set reproduces at $10^{-6}$ of the computed value and the second
  two at $10^{0}$. Verified on PDF page 66 (printed p. 64) and PDF page 67
  (printed p. 65) of IEC 60268-16:2011.
- **Library behaviour:** carries all three quantities on one scale, the plain
  ratio to $p_0^2 = (20\ \mu\text{Pa})^2$, in the correction of
  [`sti.py`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/speech/sti.py); the transcription in
  [`tests/reference_data/`](https://github.com/jmrplens/phonometry/blob/main/tests/reference_data) keeps the printed cells
  verbatim and names the $10^{6}$ it rescales them by, and the conformance row
  "IEC 60268-16:2011 Annex M" reads the adjustment they feed.
- **Status:** unreported.

## IEC 60268-16:2011, Table M.1 (step 3 I_am,k at 250 Hz)

- **Location:** Annex M, Table M.1, step 3, the $I_{am,k}$ row, 250 Hz column
  (printed p. 65).
- **The print:** 2 850 000, the same value as the 500 Hz cell beside it.
- **The problem:** the cell does not round from the quantity it names. With
  the operational levels printed two rows above, $I_{am,k}$ at 250 Hz is the
  auditory masking factor of the 125 Hz band times that band's combined
  intensity, $0.01463507 \times 195\,339\,273 = 2\,858\,804$, which at the
  three significant figures the row is printed to reads 2 860 000. The 500 Hz
  cell is correct: its 2 852 252 does print as 2 850 000. The two cells are
  reproduced together only by carrying the *rounded* $amf \times 1000 = 14.6$
  of the row above instead of the factor itself, and step 2 shows that is not
  what the table does, since its two corresponding cells are printed apart, as
  508 000 and 507 000, which only the unrounded factor gives.
- **Evidence:** both cells recomputed from the printed operational speech and
  noise levels, and the step 2 pair recomputed the same way as a control. The
  defect changes nothing downstream: the masking and threshold correction of
  the band is 0.985552 with the correct value against 0.985596 with the
  printed one, and the row prints 0.986 either way. Verified on PDF page 67
  (printed p. 65) and PDF page 66 (printed p. 64) of IEC 60268-16:2011.
- **Library behaviour:** computes $I_{am,k}$ from the unrounded masking factor.
  The transcription in [`tests/reference_data/`](https://github.com/jmrplens/phonometry/blob/main/tests/reference_data)
  keeps the printed cell and the test
  `test_annex_m_step3_masking_intensity_at_250_hz_is_the_printed_erratum`
  asserts the computed value against 2 858 804 and against the print, so the
  one cell of the table that is not an oracle cannot quietly become one.
- **Status:** unreported.

## IEC 60268-7:2010, 8.7.3.3, Formula (3) ($U_{470}$ for the product at 460 Hz)

- **Location:** 8.7.3.3 c), Formula (3), printed p. 28.
- **The print:** "$L_\mathrm{d3} = 20 \log \{(U_{470} + U_{740})/U_{600}\}$", one
  paragraph after item b), which reads "the third-order components are at
  460 Hz and 740 Hz".
- **The problem:** the test signal of 8.7.3.2 is the pair 70 Hz and 600 Hz, so
  the third-order modulation products are at $600 \pm 2 \times 70$ Hz, 460 Hz
  and 740 Hz, as item b) says. No product of the two tones falls at 470 Hz:
  the subscript of the first voltage should read 460.
- **Evidence:** item b) and Formula (3) on PDF page 30 (printed p. 28) of IEC
  60268-7:2010, Edition 3.0.
- **Library behaviour:**
  [`headphone_modulation_signal`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py)
  generates the 70 Hz and 600 Hz pair, and `modulation_distortion` reads the
  third-order products at 460 Hz and 740 Hz; the conformance row of 8.7.3.3 b)
  pins the four product frequencies.
- **Status:** unreported.

## IEC 60268-7:2010, 8.7.4.2 c), Formula (5) (a closing parenthesis with no opening one)

- **Location:** 8.7.4.2 c), Formula (5), printed p. 29.
- **The print:** "$L_\mathrm{dd3} = 20 \log \{U_{2f2-f1} + U_{2f1-f2})/2\,U_{f2}\}$".
- **The problem:** a brace opens before $U_{2f2-f1}$ and a parenthesis closes
  after $U_{2f1-f2}$ that was never opened, so the print leaves open whether
  $2U_{f2}$ divides both products or only the second. The characteristic
  8.7.4.1 refers to is the one of IEC 60268-3:2013, 14.12.8.1 b): "the ratio of
  the arithmetic sum of the output voltages at frequencies $2f_2 - f_1$ and
  $2f_1 - f_2$ to the reference voltage $U_{2,\mathrm{ref}}$ which is equal to
  twice the output voltage $U_{2,f2}$", that is
  $20\lg\{(U_{2f_2-f_1} + U_{2f_1-f_2})/2U_{f_2}\}$.
- **Evidence:** Formula (5) on PDF page 31 (printed p. 29) of IEC
  60268-7:2010, Edition 3.0, read against 14.12.8.1 b) on PDF page 38
  (printed p. 36) of IEC 60268-3:2013.
- **Library behaviour:**
  [`difference_frequency_distortion`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/intermodulation.py)
  divides the arithmetic sum of both third-order products by the sum of the
  two tones' amplitudes, which is $2U_{f_2}$ for the equal tones of
  [`headphone_difference_frequency_signal`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py).
  No change was required.
- **Status:** unreported.

## IEC 60268-7:2010, 8.7.4.2 a) (the test signal sent to 8.7.3.1)

- **Location:** 8.7.4.2 a), printed p. 28.
- **The print:** "The headphone is brought under standard measuring
  conditions, and then the input signal is changed to that required for the
  measurement (see 8.7.3.1)."
- **The problem:** 8.7.3.1 is the general subclause of modulation distortion,
  one sentence that refers to IEC 60268-2. The signal the difference-frequency
  measurement requires, two sinusoids 80 Hz apart, each giving half the rated
  input voltage, is specified in 8.7.4.1, a few lines above the item.
- **Evidence:** 8.7.3.1, 8.7.4.1 and 8.7.4.2 a) on PDF page 30 (printed
  p. 28) of IEC 60268-7:2010, Edition 3.0.
- **Library behaviour:** none needed.
  [`headphone_difference_frequency_signal`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py)
  generates the signal of 8.7.4.1.
- **Status:** unreported (cross-reference defect, no numerical consequence).

## IEC 60268-7:2010, Figure 3, NOTE (the head and torso simulator cited as IEC 60969)

- **Location:** the NOTE under Figure 3, "Illustrated measurement diagram by
  simulated programme signal", printed p. 16.
- **The print:** "Power summation of the 1/3-octave-analized data multiplied
  by filtering coefficients given by IEC 61672-1 and/or IEC 60969 gives the
  corrected voltage."
- **The problem:** the filtering coefficients the NOTE means are those of the
  two filters 7.4 lists above the figure, the A-weighting of IEC 61672-1 and
  the "free field compensating filter, that is the filter with inverse
  response of the free field response of the manikin specified in IEC 60959".
  The figure itself labels the manikin "HATS (IEC 60959)", and Clause 2 lists
  IEC TR 60959 among the normative references; it lists no IEC 60969. The
  digits are transposed.
- **Evidence:** 7.4 and Figure 3 on PDF pages 17 and 18 (printed pp. 15 and
  16), and Clause 2 on PDF page 9 (printed p. 7), of IEC 60268-7:2010,
  Edition 3.0.
- **Library behaviour:** none needed.
  [`program_signal_level`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py)
  takes the free-field response of the head and torso simulator as an input
  and compensates it.
- **Status:** unreported (cross-reference defect, no numerical consequence).

## IEC 60268-7:2010, 8.6.3 to 8.6.5 (cross-references that miss their subclause)

- **Location:** 8.6.3.2 a) and b), 8.6.3.3, 8.6.4.2 a) and b), 8.6.4.3,
  8.6.5.2 a) and b) and 8.6.5.3, printed pp. 24 to 27.
- **The print:** the free-field and diffuse-field comparison measurements
  bring the headphone under comparison conditions "(see 7.4 and Annex C)" and
  "(see 7.4 and Annex D)", and cover "at least the rated frequency range (see
  8.6.5)"; 8.6.3.3 uses a headphone "measured as described in 8.6.2.2" and
  8.6.4.3 one "measured as described in 8.6.3.2", with its results "calculated
  as described in 8.6.3.2"; 8.6.5.2 a) refers to "8.6.2.2 or 8.6.3.2 and Annex
  E" for the conditions, b) to 8.6.5 for the rated frequency range and reads
  the probe microphone "through a 1/3 octave filter with appropriate centre
  frequency (see 7.5 and Annex B)"; 8.6.5.3 is "the same as given in 8.6.4.2,
  except that the sound field is replaced by a headphone, previously
  calibrated by the method of 8.6.4.2, using at least 16 test persons".
- **The problem:** 7.4 is the measurement conditions for the simulated
  programme signal; the free-field and diffuse-field comparison conditions are
  7.5.2 and 7.5.3, and the ear canal measurement conditions, with "A very
  small microphone, in accordance with the requirements in Annex B", are 7.6,
  not 7.5, the loudness comparison conditions. The rated frequency range is
  8.6.6, which 8.6.2.2 b) cites correctly. The direct measurement a
  substitution reference has to come from, and the one whose results it
  calculates like, is 8.6.3.2 for the free field and 8.6.4.2 for the diffuse
  field; the conditions 8.6.5.2 a) borrows are those of the same two methods,
  8.6.3.2 and 8.6.4.2; and the direct ear-canal method 8.6.5.3 repeats is
  8.6.5.2. Each reference lands one subclause short of the one its text
  describes. The second 8.6.4.2 of 8.6.5.3 follows the same pattern: the
  headphone that replaces the sound field of an ear canal measurement has to
  be calibrated by the ear canal method, 8.6.5.2, since 8.6.4.2 is a loudness
  comparison against the diffuse field alone and cannot calibrate the
  free-field ear canal response that 8.6.5.1 a) also covers.
- **Evidence:** the references read against the subclauses they name, on PDF
  pages 26 to 29 (printed pp. 24 to 27), with 7.4 to 7.6 on PDF pages 17 to
  19 (printed pp. 15 to 17), of IEC 60268-7:2010, Edition 3.0.
- **Library behaviour:**
  [`field_comparison_response`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py)
  and `ear_canal_frequency_response` follow the text of each method; the
  16-person panel of a substitution reference is the one of 8.6.3.3 and
  8.6.4.3 (`FieldComparisonResponse.qualifies_as_reference`), and the one of
  8.6.5.3 is read as a panel of the direct ear canal method
  (`EarCanalFrequencyResponse.qualifies_as_reference`). The conformance row
  "IEC 60268-7:2010 8.6.5.3 with 8.6.5.2 h)" pins the 8 and 16 persons.
- **Status:** unreported (cross-reference defect).

## IEC 60268-7:2010, Table 1 (six clause numbers that name another subclause)

- **Location:** Table 1, "Classification of characteristics", printed p. 32.
- **The print:** the rows "8.3.4 Rated characteristics of protective
  devices", "8.5.1 Rated maximum or working sound pressure level", "8.6.5
  Rated frequency range", "8.7.1 Rated harmonic distortion", "8.7.2 Rated
  modulation distortion" and "8.7.3 Rated difference-frequency distortion".
- **The problem:** in this edition the protective devices are 8.3.6 (8.3.4 is
  the simulated programme signal characteristic voltage), the maximum and
  working sound pressure levels are specified in 8.5.2 (8.5.1 is the general
  subclause), the rated frequency range is 8.6.6, and the harmonic,
  modulation and difference-frequency distortions are 8.7.2, 8.7.3 and 8.7.4
  (8.7.1 is the general subclause).
- **Evidence:** Table 1 on PDF page 34 (printed p. 32) read against the
  contents list on PDF pages 4 and 5 (printed pp. 2 and 3) of IEC
  60268-7:2010, Edition 3.0.
- **Library behaviour:** none needed; the library does not read Table 1.
- **Status:** unreported (cross-reference defect, no numerical consequence).

## IEC 60268-7:2010, 8.3.4.2 NOTE 1 (a crest factor credited to IEC 60268-1)

- **Location:** 8.3.4.2, NOTE 1, printed p. 20.
- **The print:** "IEC 60268-1 specifies the spectrum, filtering circuit and
  crest factor of the simulated programme signal."
- **The problem:** IEC 60268-1:1985, which Clause 2 cites undated, specifies
  the spectrum (Table II, Figure 1) and the filtering circuit (Figure 2) and
  no crest factor: Clause 7 defines the signal as "stationary weighted
  Gaussian noise without amplitude limiting". Neither amendment of 1988 adds
  one; Amendment 1 replaces Table AII and Amendment 2 replaces 12.1. The only
  crest factor the two parts print is the "peak-to-r.m.s ratio between 1,8
  and 2,2" of IEC 60268-7 8.3.2.2 b) itself, for the clipped signal of the
  limiting voltages.
- **Evidence:** NOTE 1 on PDF page 22 (printed p. 20) and 8.3.2.2 b) on PDF
  page 21 (printed p. 19) of IEC 60268-7:2010, Edition 3.0; Clause 7 on PDF
  page 15 (printed p. 13), Table II on PDF page 23 (printed p. 21) and
  Figures 1 and 2 on PDF page 24 (printed p. 22) of IEC 60268-1:1985.
- **Library behaviour:** none needed.
  [`simulated_program_signal`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/program_signal.py)
  is unclipped by default, as Clause 7 asks, and clips to a peak-to-RMS ratio
  only when `peak_to_rms` is given, which the library credits to 8.3.2.2 b).
- **Status:** unreported (cross-reference defect, no numerical consequence).

## IEC 60268-7:2010, 8.5.2 b) and 8.5.3 c) (two settings of the e.m.f. of 1 mW)

- **Location:** 8.5.2 b), printed p. 22; 8.5.3 c) and d), printed p. 23.
- **The print:** 8.5.2 b) defines the working sound pressure level at "a
  sinusoidal voltage at 500 Hz, in series with the rated source impedance, of
  such value that 1 mW would be dissipated in a pure resistance equal to the
  rated impedance of the headphone, connected in place of it". The method,
  8.5.3 c): "The source e.m.f. is then adjusted so that the voltage across the
  input connector of the headphone is such that it would cause 1 mW to be
  dissipated in a pure resistance equal to the rated impedance of the
  headphone", and d) states the level "as the result b) in 8.5.2".
- **The problem:** the two set different e.m.f.s. With a rated impedance $R$,
  a rated source impedance $R_\mathrm{s}$ and 1 mW, the definition asks for
  $E = \sqrt{PR}\,(R + R_\mathrm{s})/R$, the e.m.f. that drives 1 mW into $R$
  in place of the headphone; the method asks for the e.m.f. that puts
  $\sqrt{PR}$ across the headphone itself, whose impedance $Z$ at 500 Hz takes
  its own share, $E = \sqrt{PR}\,|Z + R_\mathrm{s}|/|Z|$. They agree only
  when $Z$ is the rated impedance or $R_\mathrm{s}$ is zero. A 32 Ω
  headphone of 40 Ω at 500 Hz on the 120 Ω source of IEC 61938 reads
  $20\lg\{(152/32)/(160/40)\} = 1.49$ dB lower by the method than by the
  definition, and the rated impedance of 8.2.1 b) allows the modulus to sit
  anywhere above 80 % of $R$.
- **Evidence:** 8.5.2 b) on PDF page 24 (printed p. 22) and 8.5.3 c) and d)
  on PDF page 25 (printed p. 23) of IEC 60268-7:2010, Edition 3.0.
- **Library behaviour:**
  [`working_sound_pressure_level`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/electroacoustics/headphones.py)
  follows the definition of 8.5.2 b) to d) by default, and the method of
  8.5.3 c) when it is given the headphone's measured impedance at 500 Hz as
  `headphone_impedance_ohm`; the conformance row "IEC 60268-7:2010 8.5.2 b)
  and 8.5.3 c)" pins the 1.49 dB.
- **Status:** unreported.
