On a clear afternoon in Myrtle Avenue, Hounslow, a Qantas Boeing 747-400 crosses the rooftops at something like a hundred feet. Net curtains twitch. A sunflower nods in the wash. Then the next one comes, and the one after that, roughly every ninety seconds, for eighteen hours a day.

Everyone who lives under that approach knows Heathrow is quieter than it used to be. What almost nobody can explain is why — or what exactly stops an airline rolling a 1968 Boeing 707 down runway 27L at three in the morning. The answer is two completely different legal machines, built on opposite sides of the Atlantic, that have never agreed on how to solve the same problem.

One certifies the aeroplane. The other rations the night. Neither of them, it turns out, is a curfew.

The unit that is not a decibel

Start with the measurement, because everything downstream depends on it. Section B36.2 of 14 CFR part 36 is one sentence long: the noise evaluation metric is the effective perceived noise level expressed in EPNdB.

An EPNdB is not a decibel you could read off a phone app. It is a composite that weights the sound for how annoying the human ear finds it, penalises pure tones such as a fan whine, and integrates the whole event over the time the aeroplane is audible. A jet that is briefly loud can score better than one that is moderately loud for much longer. That is deliberate. The regulation is trying to measure irritation, not pressure.

Section B36.3 then nails down where you stand to measure it, and the three positions are oddly specific. The lateral point sits on a line 1,476 feet (450 m) from the runway centreline, wherever the noise after lift-off peaks. The flyover point is 21,325 feet (6,500 m) from the start of the take-off roll. The approach point is 6,562 feet (2,000 m) from the threshold, which on level ground puts the microphone 394 feet below a standard three-degree glideslope.

Three microphones, three numbers, one aeroplane. Every argument in this article is downstream of them.

Stage 1 to Stage 5, and the dates that emptied the fleet

Section 36.1(f) defines the ladder. A Stage 1 aeroplane is defined negatively — one that has never been shown to meet the Stage 2 or Stage 3 levels. Stage 2, Stage 3, Stage 4 and Stage 5 aeroplanes are each defined as having been shown not to exceed the corresponding limit. One aeroplane may not be identified as complying with more than one stage at a time.

Section 36.103 sets the windows by application date. Apply for a type certificate between 5 November 1975 and the end of 2005 and you must meet Stage 3. From 1 January 2006, Stage 4. From 31 December 2017, Stage 5 for anything with a maximum certificated take-off weight of 121,254 pounds (55,000 kg) or more, and from 31 December 2020 for everything lighter.

None of that, however, grounds a single aeroplane. Part 36 is a certification rule, and it says so with unusual candour in section 36.5: no determination is made, under that part, that the noise levels are or should be acceptable or unacceptable for operation at an airport. The stage is a label. The ban lives somewhere else entirely.

It lives in 14 CFR part 91, subpart I, and the wording is blunt. Section 91.805: on and after 1 January 1985, no person may operate to or from an airport in the United States any subsonic aeroplane covered by the subpart unless it has been shown to comply with Stage 2 or Stage 3 levels. That one sentence ended the Stage 1 era.

Section 91.853 did the same to Stage 2, for the big jets: after 31 December 1999, nothing over 75,000 pounds may operate to or from an airport in the contiguous United States unless it complies with Stage 3, 4 or 5. The smaller jets got sixteen more years. Section 91.881 closed that door after 31 December 2015, for civil subsonic jets of 75,000 pounds or less.

Those two dates are why the American business-jet market looked the way it did in 2014, with elderly Learjets and Hawkers suddenly unsellable, and why the second-hand 727 and DC-8 trade collapsed fifteen years earlier. Not fuel prices. Not fashion. A sentence in part 91.

The FAA’s own primer on aviation noise fundamentals, from its Noise Policy Review series.

Where the actual numbers live, and the awkward part

Here is something that catches people out. For Stage 2 and Stage 3, part 36 gives you real numbers you can read. Section B36.5(b) sets a Stage 2 flyover limit of 108 EPNdB at a maximum weight of 600,000 pounds or more, dropping 5 EPNdB for each halving of weight down to a floor of 93 EPNdB at 75,000 pounds. Section B36.5(c) does the same for Stage 3, this time splitting by engine count: 106 EPNdB flyover for four-engined aircraft at 850,000 pounds, 104 for trijets, 101 for twins, each falling 4 EPNdB per halving.

For Stage 4 and Stage 5, the numbers are not there. Section B36.5(d) says the Stage 4 limits are those prescribed in Chapter 4, paragraph 4.4 of ICAO Annex 16, Volume I. Section B36.5(e) sends you to Chapter 14, paragraph 14.4 of the same Annex. Both are incorporated by reference. The American rule for every modern airliner points at an international document that is not published in the Federal Register and that you have to buy from ICAO.

That cumulative-margin idea is the quiet revolution in the whole system. Under Chapter 3 you had to hit a number at each microphone. Under Chapter 4 and Chapter 14 you are given a total to beat and may choose where to spend it — trade a little approach noise for a better flyover figure, if that suits your wing and your engine.

The FAA’s own rule records how strict Chapter 14 became: the ICAO committee agreed an increase in stringency of 7 EPNdB cumulative relative to Chapter 4 levels. A footnote to the same rule adds the detail that matters to engineers — Chapter 14 also requires a minimum reduction of 1 dB at each individual point, so the total cannot all be won at one microphone.

Britain went the other way: a budget, not a ban

The United States rations noise by outlawing categories of aeroplane nationwide. Britain, at its three noise-designated airports, does something far stranger and rather cleverer. It gives each airport a nightly allowance of noise and lets the airlines spend it.

The currency is the Quota Count. The Department for Transport defines QC as the weighting attributed to the arrival or departure of a specified aircraft type by reference to its certificated noise performance, divided into 3 EPNdB bands. Read that twice, because it is the hinge of this entire article: the British night-flight system is denominated in the same certificated EPNdB figures that part 36 produces. The American machine certifies; the British machine then prices the result.

The bands run from QC0.125 at the quiet end up to QC16, and the structure follows the physics. Each band is 3 EPNdB wide, and 3 dB is a doubling of acoustic energy, so every step up the ladder doubles the QC value. A QC2 movement therefore costs the same budget as sixteen QC0.125 movements — which is precisely the arithmetic that local campaigners put to the Department during the last consultation, pointing out that a quota expressed in noise says nothing at all about how many times you are woken.

Crucially, QC is attributed separately to the arrival and to the departure. An aircraft does not have one rating; it has two.

Two different nights

British night-flight law runs on two overlapping clocks, and conflating them is the commonest mistake in the public argument. The night quota period, which the movement and QC limits actually police, is 23:30 to 06:00 — six and a half hours. The night period, used for noise-contour assessment, is 23:00 to 07:00 — a full eight.

That gap is not academic. When the Planning Inspectorate examined Stansted in 2021, it noticed that the government’s night restrictions simply do not reach into the shoulders of the night at all.

The Inspector’s remedy was a planning condition expressed as a noise contour rather than a movement cap: the area enclosed by the 48 dB LAeq 8-hour night contour at Stansted must be no more than 74.0 square kilometres. It becomes legally binding once the airport reaches 35 million passengers a year. A third mechanism, bolted on because the first two left a gap.

The actual allowance, airport by airport

The regime now in force runs from October 2025 to October 2028. It is a deliberate holding pattern: the Department chose a three-year bridging regime while it waits for the Aviation Night Noise Effects study, and kept every movement and quota limit exactly where it already was.

Look at Gatwick’s summer line. It is allowed 11,200 night movements but only 5,150 quota points, which works out at an average of well under half a QC point per flight. The airport’s own historic figures bear that out: its average quota count per movement has run around 0.35 to 0.52. At Gatwick in summer the binding constraint is not noise at all. It is the movement cap. The quota is slack.

Heathrow is the opposite story. Its quota limit was cut from 4,080 to 2,415 in winter, and from 5,100 to 2,735 in summer, between the 2017-18 and 2018-19 seasons. The movement limit did not move. The noise budget did.

Richmond Council sets out the noise and night-flight arguments around Heathrow expansion.

Heathrow does not have a curfew

This is the part that surprises almost everyone, including a good number of the people who live under the approach. During the last consultation, residents near Gatwick complained that the government had given Heathrow a curfew and denied them one. The Department’s response was that this is a misunderstanding: all three noise-designated airports are 24-hour airports, the government imposes no period of closure at any of them, and it prevents operations at no time of night — other than for QC4, QC8 and QC16 rated aircraft, which are barred outright.

So what is it that keeps Heathrow quiet between half past eleven and half past four? A voluntary agreement. The airport has undertaken that no flights are scheduled between 23:30 and 04:30, and that flights scheduled between 04:30 and 06:00 will not land before 04:30. It is an industry commitment, not a statutory curfew, and the distinction is the difference between something a minister would have to legislate to remove and something an airport could in principle renegotiate.

Restrictions on night flights at Heathrow go back to 1962. The airport now runs about 5,800 night flights a year, roughly sixteen a night. And on the nights when weather or a European air-traffic strike wrecks the schedule, the system bends: airports grant dispensations, which are counted and published, and which ran into the hundreds in the worst recent seasons.

Why America has almost no curfews at all

If the British system is a budget, the American one is a wall — built in 1990 and facing inwards. The Airport Noise and Capacity Act of that year, implemented through 14 CFR part 161, did not ban local noise restrictions. It did something more effective. It made new ones nearly impossible to obtain.

The pivot date is 1 October 1990. Part 161 applies to Stage 2 restrictions proposed after that date and Stage 3 restrictions that took effect after it, and it also catches any amendment to an older restriction where the change reduces or limits operations. Everything already running on 1 October 1990 was, in effect, grandfathered. Everything since has needed to run the gauntlet.

And the gauntlet is six statutory conditions, every one of which the airport must support with substantial evidence. The restriction must be reasonable, nonarbitrary and nondiscriminatory. It must not create an undue burden on interstate or foreign commerce — which requires a cost-benefit analysis showing the benefits have a reasonable chance of exceeding the costs. It must maintain safe and efficient use of the navigable airspace, must not conflict with any existing federal statute or regulation, must have given the public an adequate opportunity to comment, and must not create an undue burden on the national aviation system.

Condition 2 is the one that does the killing. An airport proposing a night curfew has to demonstrate, in money, that the quiet is worth more than the flights that would be lost. That is a hard case to make and an expensive one to assemble, and the practical effect is that the American curfews people can name are overwhelmingly old ones — restrictions that were already in place before October 1990 and were therefore never put to the test.

Two legal cultures, then, pointed at one problem. America regulates the machine and leaves the airport almost powerless. Britain regulates the night and leaves the machine to ICAO. Neither country has a national curfew. Both have got dramatically quieter anyway, because the thing that actually silenced the approach into Myrtle Avenue was not a statute at all. It was the high-bypass turbofan — and the certification ladder that made buying one compulsory.

Safran on how engine makers attack noise at source, which is what every stage on the ladder was designed to force.

Sources: 14 CFR part 36 (sections 36.1, 36.5, 36.103 and appendix B, sections B36.2, B36.3 and B36.5); 14 CFR part 91 subpart I (sections 91.801, 91.805, 91.853, 91.855 and 91.881); 14 CFR part 161; FAA, Stage 5 Airplane Noise Standards final rule, 82 FR 46123 (4 October 2017); Department for Transport, Night Flight Restrictions at Heathrow, Gatwick and Stansted from October 2025 — Decision Document, December 2024, and its Annex A; Planning Inspectorate Appeal Decision on Stansted Airport, 26 May 2021.