# Errata in IEC 60534

> 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-60534/
Language: en
Alternate: https://docs.jmrp.io/phonometry/es/reference/errata/iec-60534/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 60534: 10 defects in the printed text, with the evidence and what phonometry does about each.

---

These are the 10 entries of the [errata registry](https://docs.jmrp.io/phonometry/reference/errata/) on IEC 60534-8-3:2010 and IEC 60534-8-4:2005, 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 60534-8-3:2010, Annex A (the piping geometry factor is printed rounded, and the annex did not use the rounded value)

- **Location:** Annex A (informative), A.2, the "Given data" block on printed
  folio 32 (PDF page 34) of BS EN 60534-8-3:2011, against the Equation (2) row
  of Table A.1 on the same folio.
- **The print:** the given data lists "Piping geometry factor: $F_\mathrm{p} =
  0{,}98$", under the heading "The following values are used in, or determined
  from, calculations based on IEC 60534-2-1." Table A.1 then prints
  $p_{vc} = 567\,787$ Pa for example 1 and five more values for the other
  columns, from
  $p_{vc} = p_1\left[1 - x/(F_{LP}/F_P)^2\right]$ with $F_{LP} = 0{,}792$.
- **The problem:** those two cannot both be right. Solving Equation (2) for
  $(F_{LP}/F_P)^2$ from each printed pair gives 0.647 829, 0.647 827,
  0.647 821, 0.647 829 and 0.647 833 in the five columns that print a value,
  which is $F_p = 0.984$ to four digits in every one of them. The printed
  0.98 gives 0.653 128 and $p_{vc} = 571\,294$ Pa, 3 507 Pa away from the
  printed figure. The value is a computed one, not a datum: the annex says it
  comes from IEC 60534-2-1, and the head loss coefficient it prints,
  $\Sigma\zeta = 0.86$, gives $F_p = 0.984$ for the DN 100 case. So the
  annex computed with three decimals and printed two.
- **Consequence:** every downstream quantity moves. With the printed 0.98 the
  four regime boundaries come out $x_C = 0.287$, $\alpha = 0.786$,
  $x_B = 0.578$ against the printed 0.285, 0.784 and 0.576, and example 1's
  sound power comes out 21.9 W against the printed 22.3 W. Nothing is far
  wrong, and nothing reproduces either: a reader checking their implementation
  against Annex A with the number Annex A prints will not match a single row.
- **Evidence:** the given data and the six $p_{vc}$ values read from the
  printed page. Verified on PDF pages 33 and 34 (printed pp. 31 and 32) of
  BS EN 60534-8-3:2011.
- **Library behaviour:**
  [`valve_aerodynamic_noise`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves.py) takes
  the ratio as an argument and does not hold a value of its own; the
  conformance rows and `tests/noise_control/test_valves.py` pass
  $0.792/0.984$ and say why in the fixture.
- **Status:** unreported.

## IEC 60534-8-3:2010, Table A.1 (an equivalent orifice diameter ten times too small, contradicted by the row below it)

- **Location:** Annex A (informative), Table A.1, the Equation (8c) row on
  printed folio 33 (PDF page 35), against the Equation (8a) row printed
  immediately below it.
- **The print:** all six columns of the (8c) row read $d_0 = 0.010$ m. The
  given data on folio 31 gives $N_\mathrm{O} = 6$ cage openings and
  $A = 0{,}00137\ \text{m}^2$ for one of them, and Equation (8c) is
  $d_o = \sqrt{4 N_o A/\pi}$.
- **The problem:** $\sqrt{4 \times 6 \times 0.00137/\pi} = 0.102$ m, not
  0.010 m. The two numerals are the same three digits in a different order.
  The row below settles which is meant: Equation (8a) is $F_d = d_H/d_o$, the
  (8b) row prints $d_H = 0.030$ m, and the (8a) row prints $F_d = 0.30$ in
  all six columns. 0.030/0.102 is 0.30; 0.030/0.010 is 3.0.
- **Consequence:** a reader who takes the printed $d_o$ gets a valve style
  modifier of 3.0, a jet diameter ten times too large from Equation (9), and a
  peak frequency ten times too low, which moves the internal spectrum of
  Equation (19) by more than three octaves. The rest of the table is computed
  with 0.102 m, so the error is confined to the one printed cell.
- **Evidence:** the (8b), (8c) and (8a) rows read from the printed page.
  Verified on PDF pages 33 and 35 (printed pp. 31 and 33) of
  BS EN 60534-8-3:2011.
- **Library behaviour:**
  [`valve_style_modifier`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves.py)
  implements (8b) and (8c) as printed and returns 0.296 for the annex's cage,
  which rounds to the printed $F_d$; the test named after this entry pins both
  readings so the printed $d_o$ cannot come back.
- **Status:** unreported.

## IEC 60534-8-3:2010, Table A.2 (two frequency factors whose exponent is one power of ten out)

- **Location:** Annex A (informative), A.3, the $G_x$ column of Table A.2 on
  printed folio 43 (PDF page 45), bands 5 and 10 of 33.
- **The print:** the column runs $G_{x,4} = 5.6 \times 10^{-9}$,
  $G_{x,5} = 1.4 \times 10^{-9}$, $G_{x,6} = 3.6 \times 10^{-8}$, and later
  $G_{x,9} = 5.8 \times 10^{-7}$, $G_{x,10} = 1.4 \times 10^{-7}$,
  $G_{x,11} = 3.5 \times 10^{-6}$.
- **The problem:** below the internal coincidence frequency Table 6 makes
  $G_x$ proportional to $f_i^4$, so the column has to rise monotonically, and
  it does everywhere except at those two bands, where it falls. Recomputing
  Table 6 for this pipe gives $1.4 \times 10^{-8}$ at band 5 and
  $1.4 \times 10^{-6}$ at band 10: the mantissa is right in both and the
  exponent is one too small.
- **Consequence:** none for the rest of the annex, and that is what settles
  it. The transmission losses printed two rows further down, $TL_5 = -86.1$
  dB and $TL_{10} = -76.2$ dB, are what Equation (20a) gives with the
  corrected factors; the printed factors would give $-96.1$ dB and
  $-86.3$ dB. So Table A.2 computed with the right values and printed the
  wrong ones, and anyone seeding an oracle from the $G_x$ column alone
  inherits a 10 dB error in two bands.
- **Evidence:** the $G_x$ column read from the printed page. Verified on PDF
  pages 45 and 46 (printed pp. 43 and 44) of BS EN 60534-8-3:2011; the 24
  printed transmission losses on the second of them are what the library
  reproduces to within 0.07 dB.
- **Library behaviour:**
  [`pipe_transmission_loss`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves.py)
  computes $G_x$ from Table 6, and the conformance row "Pipe transmission
  loss, example 7, 24 bands" reproduces every printed loss, which the printed
  $G_x$ could not.
- **Status:** unreported.

## IEC 60534-8-4:2005, Equation (12) (the Strouhal number is printed one way in the clause and another in the annex)

- **Location:** Clause 5.1, Equation (12) on printed folio 11 (PDF page 13) of
  BS EN 60534-8-4:2005, against the same equation restated in the Table A.1
  row (12) on printed folio 23 (PDF page 25).
- **The print:** the clause prints
  $N_{STR} = \dfrac{0{,}02\,F_L^2\,C}{N_{34}\,x_{Fzp1}^{1,5}\,d\,d_0}
  \left(\dfrac{1}{p_1-p_v}\right)^{0,57}$ and the annex prints
  $N_{STR} = \dfrac{0{,}036\,F_L^2\,C\,F_d^{\,0,75}}
  {N_{34}\,x_{Fzp1}^{1,5}\,d\,d_0}
  \left(\dfrac{1}{p_1-p_v}\right)^{0,57}$.
- **The problem:** they are two different functions of the valve, not two
  roundings of one. The annex carries a factor $F_d^{0.75}$ the clause does
  not have, and a leading constant 1.8 times larger. The exponents 0.57 and
  1.5, the square on $F_L$ and the product $d\,d_o$ are identical in both, so
  the difference is confined to the numerator. The annex's own numbers settle
  which one it evaluated: with $C_v = 90$, $F_d = 0.42$, $F_L = 0.92$,
  $N_{34} = 1.17$, $x_{Fzp1} = 0.2386$, $d = d_o = 0.1$ m and
  $p_1 - p_v = 997\,680$ Pa, the annex form gives 0.399 and the clause form
  0.425, and Table A.1 prints $N_{Str} = 0.399$ in two of its three columns
  and 0.243 in the third, both of which are the annex form to three digits.
- **Consequence:** the peak frequency of Equation (11), and with it the whole
  band spectrum of 5.4 and the transmission loss of Equation (16b), which is
  evaluated at that frequency. For Annex A's valve the two forms are 6 %
  apart; for a single-port valve with $F_d = 1$ the annex form is 80 % above
  the clause form, five sixths of an octave in the peak frequency. An
  implementation that follows the normative clause cannot reproduce a single
  frequency-dependent row of the informative annex.
- **Evidence:** both printings as they appear on the page. Verified on PDF
  page 13 (printed p. 11) and PDF page 25 (printed p. 23) of
  BS EN 60534-8-4:2005.
- **Library behaviour:**
  [`jet_strouhal_number`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  takes a ``form`` argument and implements both. The default is ``"annex"``,
  the form that reproduces the printed examples, and the conformance row
  "Strouhal number and turbulent peak (Eqs. (11), (12))" pins it; a test named
  after this entry pins the ratio between the two.
- **Status:** unreported.

## IEC 60534-8-4:2005, Table A.1 (a band transmission loss printed without its minus sign)

- **Location:** Annex A (informative), Table A.1, the Equation (22a) row on
  printed folio 25 (PDF page 27), all three columns.
- **The print:** the three cells read "TL (8 000 Hz) = 51,76 dB", with no sign
  before the 5.
- **The problem:** Equation (22a) is $TL(f_i) = TL_{fr} + \Delta TL(f_i)$, and
  the table prints both of its inputs one row above and two folios earlier:
  $\Delta TL(8\,000\ \text{Hz}) = -7.053$ dB in the (22b) row on the same
  page, and $TL_{fr} = -44.71$ dB in the (15) row on printed folio 23. Their
  sum is $-51.763$ dB. The row below settles it too: Equation (21) with
  $L_{pi}(8\,000\ \text{Hz}) = 116.3$ dB and the 12.67 dB spreading term
  gives 51.87 dB against the printed $L_{pe,1m} = 51.8$, and the unrounded
  116.252 dB of the chain gives 51.82; with $+51.76$ it would give
  155.4 dB.
- **Consequence:** none for the annex, which computed with the right sign and
  printed the wrong one, and 103 dB for anyone seeding an oracle from that
  cell. Every other transmission loss in this document is printed negative
  ($-44.71$; $-29.56$; $-74.27$; $-62.917$; $-75.006$; $-7.053$),
  including the one directly above it.
- **Evidence:** the three (22a) cells and the (22b) cells above them, read on
  the printed page. Verified on PDF page 27 (printed p. 25) of
  BS EN 60534-8-4:2005: there is no hyphen, no minus and no leading dash in
  any of the three, and the (22b) cells beside them print theirs.
- **Library behaviour:**
  [`transmission_loss_correction`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  and the ``band_transmission_loss`` of ``valve_hydrodynamic_noise`` return
  $-51.76$ dB for this band, which is what the conformance row "Frequency
  route at 8 kHz, examples 1 to 3 (Eqs. (19) to (22))" pins together with the
  external levels the negative value reproduces.
- **Status:** unreported.

## IEC 60534-8-4:2005, 6.3.2 b) (a seat diameter formula whose constant is not in the unit its symbol is declared in)

- **Location:** Clause 6.3.2, item b), the unnumbered display formula on
  printed folio 15 (PDF page 17), against the Clause 3 symbol table on printed
  folio 6 (PDF page 8).
- **The print:** "$d_o = 5{,}2\sqrt{N_{34}\,C_n}$", with no equation number
  and no unit on the constant. The symbol table declares $d_o$ "Seat or
  orifice diameter", unit **m**.
- **The problem:** the two cannot both hold. For the last stage of any real
  multistage trim the formula returns tens: $C_n = 90$ as a $C_v$, with
  $N_{34} = 1.17$, gives 53.4, and a seat 53 m across is not a valve. Read
  as millimetres it is 53 mm, half the bore of the DN 100 valve of Annex A,
  which is what a last stage looks like. IEC 60534-8-3 gives the same quantity
  a second route, through its own Equation (27) and the total flow area, and
  for this stage that route gives 48.4 mm.
- **Consequence:** the result of this formula is fed to Equation (12), where
  $d_o$ sits in the denominator beside $d$ in metres. Taking the printed
  number as metres makes the Strouhal number a thousand times too small and
  the peak frequency with it, which moves the spectrum ten octaves.
- **Evidence:** the formula and the symbol table row as printed. Verified on PDF page 17 (printed p. 15) and PDF page 8 (printed p. 6)
  of BS EN 60534-8-4:2005. This entry rests on the arithmetic of the printed
  formula as well as on the print itself: nothing on PDF pages 8 to 20 states
  the unit of the constant 5.2.
- **Library behaviour:**
  [`last_stage_seat_diameter_mm`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  implements the formula as printed and carries the unit in its name, and its
  docstring says to divide by a thousand before passing the result to
  Equation (12). The test named after this entry pins both the value and the
  order of magnitude.
- **Status:** unreported.

## IEC 60534-8-4:2005, Equation (23b) (a stage inlet pressure computed from the next stage instead of the previous one)

- **Location:** Clause 6.2, Equations (23a) and (23b) on printed folio 13
  (PDF page 15) of BS EN 60534-8-4:2005.
- **The print:** (23a) is "$p_{1,i} = p_1$ for $i = 1$" and (23b) is
  "$p_{1,i} = p_{1,i+1} - \dfrac{p_1-p_2}{(C_{i-1}/C)^2}$ for
  $i = 2 \ldots n$".
- **The problem:** as printed, each stage's inlet pressure is computed from
  the **next** stage's by subtracting a positive quantity, so the sequence
  increases with $i$: the last stage would start at the highest pressure and
  the first at the lowest, which contradicts (23a) and reverses the flow. The
  index in the denominator is $C_{i-1}$, the stage *before* the one being
  computed, which is the recursion the equation is written for:
  $p_{1,i} = p_{1,i-1} - (p_1-p_2)/(C_{i-1}/C)^2$. Read that way the equation
  is the series law for flow resistances, $1/C^2 = \sum_i 1/C_i^2$: each stage
  takes a share of the differential in inverse proportion to the square of its
  own capacity, and the shares sum to the whole.
- **Consequence:** every per-stage quantity of Clause 6, since (24a) chains
  the outlet pressures to the inlets and (26) makes each stage's pressure
  ratio from both. Following the printed index gives a trim whose first stage
  sees the smallest differential, which is the opposite of every multistage
  design the clause describes.
- **Evidence:** both equations as printed. Verified on PDF
  page 15 (printed p. 13) of BS EN 60534-8-4:2005: the subscript is $i+1$ on
  the pressure and $i-1$ on the flow coefficient. This
  entry rests on the internal contradiction between (23a) and (23b) as well as
  on the print.
- **Library behaviour:**
  [`stage_conditions`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  implements the forward recursion and says so in its docstring; the test
  named after this entry pins that the inlet pressures fall along the trim.
- **Status:** unreported.

## IEC 60534-8-4:2005, Equations (18a) and (18b) (two conditions that do not divide the domain between them)

- **Location:** Clause 5.3, Equations (18a) and (18b) on printed folio 12
  (PDF page 14) of BS EN 60534-8-4:2005.
- **The print:** (18a) ends "for $x_\mathrm{F} \le x_\mathrm{Fz}$" and (18b)
  ends "for $x_\mathrm{Fzp1} < x_\mathrm{F} \le 1$".
- **The problem:** the two conditions are written against two different
  thresholds. $x_\mathrm{Fz}$ is the characteristic pressure ratio at the
  $6 \times 10^{5}$ Pa the estimate of Equation (3a) and the charts of Figures 4 to 9
  are drawn at, and $x_\mathrm{Fzp1}$ is that same ratio moved to the working
  inlet pressure by Equation (3c), so the two are equal only at that one
  pressure. Above it $x_\mathrm{Fzp1} < x_\mathrm{Fz}$ and the interval
  between them is claimed by both equations, the turbulent one by (18a) and
  the cavitating one by (18b); below it the same interval is claimed by
  neither. Everything else in the document tests the corrected ratio: the
  conditions printed above (7a) and (7b), the region printed for (9), the
  NOTE to (17), and the split of 6.3. (18a) is the only condition in the
  document that names $x_\mathrm{Fz}$, and the two do not even agree on the
  boundary itself, which (18a) includes with ≤ and (18b) excludes with <.
- **Consequence:** it grows with the inlet pressure, because the inlet
  pressure is what separates the two thresholds. At the 10 bar of Annex A the
  disputed interval is $0.2386 < x_\mathrm{F} \le 0.2543$ and the two
  branches differ by at most 0.03 dB inside it, because Equation (9) starts
  the cavitation term at exactly zero on the threshold. At 100 bar the
  interval runs from 0.179 to 0.254 and the two branches differ by up to
  8 dB; at 400 bar, by 10 dB. High-pressure liquid service is where this
  method earns its keep.
- **Evidence:** the two conditions as printed. Verified on PDF page 14
  (printed p. 12) of BS EN 60534-8-4:2005: (18a) reads "for $x_F \le x_{Fz}$"
  with the subscript Fz and no p1, above a (18b) that reads
  "for $x_{Fzp1} < x_F \le 1$".
- **Library behaviour:**
  [`valve_hydrodynamic_noise`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  decides the regime once, on $p_1 - p_2$ against
  $x_\mathrm{Fzp1}(p_1 - p_\mathrm{v})$, which is the test 5.1 prints for
  Equations (7a) and (7b), and the sound power, the transmission loss, the
  external level and the band spectrum all follow that one flag. The test
  named after this entry pins that a point inside the disputed interval comes
  out cavitating.
- **Status:** unreported.

## IEC 60534-8-4:2005, Equations (19a) and (19b) (a turbulent condition written against the wrong ratio, and two conditions that both claim the threshold and x_F = 1)

- **Location:** Clause 5.4.1, the conditions above Equations (19a) and (19b)
  on printed folio 12 (PDF page 14) of BS EN 60534-8-4:2005.
- **The print:** "For turbulent conditions $x_\mathrm{Fz} \le x_\mathrm{Fzp1}$
  or $x_\mathrm{F} \ge 1$):" above (19a), and "For cavitating conditions
  ($x_\mathrm{Fzp1} \le x_\mathrm{F} \le 1$:" above (19b).
- **The problem:** the first alternative of the (19a) condition compares the
  two thresholds with each other and names no operating point. Read as
  printed, that alternative holds exactly when $p_1 \le 6 \times 10^5$ Pa,
  because Equation (3c) raises $x_\mathrm{Fzp1}$ above $x_\mathrm{Fz}$ only
  below that pressure. Below $x_\mathrm{F} = 1$, where the second alternative
  does not apply, the turbulent spectrum of (19a) would then cover every point
  at or below 6 bar, cavitating ones included, and no point above it, where a
  turbulent point, $x_\mathrm{F} < x_\mathrm{Fzp1}$, would fall under neither
  condition; the intended condition is $x_\mathrm{F} \le x_\mathrm{Fzp1}$.
  The two conditions then share both of their ends: the threshold
  $x_\mathrm{F} = x_\mathrm{Fzp1}$, which (19b) includes with ≤ while the
  corrected (19a) would include it too, and
  $x_\mathrm{F} = 1$, which (19a) claims with ≥ and (19b) with ≤. Each
  condition also leaves a parenthesis unpaired, a closing one after (19a)'s
  and an opening one before (19b)'s. The rest of the document does not settle
  the threshold either: the test of 5.1 calls a point turbulent when
  $\Delta p$ "is lower than" $x_\mathrm{Fzp1}(p_1 - p_\mathrm{v})$ and
  cavitating when it "exceeds" it, the region of Equation (9) is printed
  $x_\mathrm{Fz,p1} \le x_\mathrm{F} \le 1$, and 6.3 applies the turbulent
  equations when $x_\mathrm{F,i} \le x_\mathrm{Fzp1,i}$.
- **Consequence:** none on the threshold, where Equation (9) returns exactly
  zero and (19b) reduces to (19a). At $x_\mathrm{F} = 1$ neither applies,
  because Equations (9) and (13) divide by $1 - x_\mathrm{F}$ there. Read
  literally, the (19a) condition would split the band spectrum below
  $x_\mathrm{F} = 1$ by inlet pressure instead of by regime.
- **Evidence:** the two conditions as printed. Verified on PDF page 14
  (printed p. 12) of BS EN 60534-8-4:2005: the turbulent condition reads
  "$x_{Fz} \le x_{Fzp1}$ or $x_F \ge 1$)" and the cavitating one
  "($x_{Fzp1} \le x_F \le 1$:", with Equation (3c) on PDF page 10 (printed
  p. 8), the 5.1 test on PDF page 12 (printed p. 10) and the 6.3 condition on
  PDF page 16 (printed p. 14).
- **Library behaviour:**
  [`valve_hydrodynamic_noise`](https://github.com/jmrplens/phonometry/blob/main/src/phonometry/noise_control/valves_hydrodynamic.py)
  picks (19a) or (19b) by the one regime of 5.1, cavitating when $\Delta p$
  exceeds $x_\mathrm{Fzp1}(p_1 - p_\mathrm{v})$, so the threshold itself is
  turbulent, and it stops below $x_\mathrm{F} = 1$. A
  `HydrodynamicValveNoise` built by hand reads its regime the same way and
  refuses a point at or past $x_\mathrm{F} = 1$;
  `test_a_differential_on_the_threshold_is_turbulent` pins the threshold.
- **Status:** unreported.

## IEC 60534-8-4:2005, Table A.1 (three printed intermediates its own equations do not reproduce)

- **Location:** Annex A (informative), Table A.1: the Equation (17) row on
  printed folio 24 (PDF page 26), columns 2 and 3; the Equation (20a) row on
  the same folio, column 3; and the Equation (11) row on printed folio 23
  (PDF page 25), column 1.
- **The print:** $TL_{cav} = -62{,}917$ and $-75{,}006$; $F_{turb}(8\,000\
  \text{Hz}) = -36{,}24$; $f_{p,turb} = 494{,}5$ Hz.
- **The problem:** none of the three follows from the values printed beside
  it. Equation (17) with the annex's own $TL_{turb}$, $f_{p,turb}$,
  $f_{p,cav}$ and efficiency ratio gives $-62.86$ and $-74.93$, which is
  0.06 and 0.08 dB away. Equation (20a) at 8 kHz with the column's own
  $f_{p,turb} = 397.93$ Hz gives $-36.18$; the printed $-36.24$ needs
  396.0 Hz. And the unrounded chain through Equations (12) and (11) gives
  494.64 Hz in the first column, where columns 2 and 3 reproduce their 654.35
  and 397.93 Hz to the last printed digit.
- **Consequence:** small and confined. None of the three reaches a printed
  result: the external levels of Equations (18a) and (18b) round to the same
  62.7 / 81.0 / 66.9 dB either way, and so does the band level of Equation
  (19a). It matters only to an implementer comparing intermediates, who will
  find three rows out of forty that cannot be matched exactly and no
  explanation on the page.
- **Evidence:** the three rows as printed, recomputed from the intermediates
  printed beside them. Verified on PDF pages 25 and 26
  (printed pp. 23 and 24) of BS EN 60534-8-4:2005. This entry rests on a
  recomputation as well as on the print.
- **Library behaviour:** the conformance row "Cavitating transmission loss,
  examples 2 and 3 (Eq. (17))" carries a tolerance of 0.1 dB and names this
  entry as the reason; the tests named after it pin what the equations give
  and record what the annex printed.
- **Status:** unreported.
