Rockets Don't Push Against Air

1 · The claim, stated fairly

Quoting the site directly: “Mainstream physics is completely wrong in applying Newton's third law to a pressurized gas release in a vacuum environment. Thrust must have something to push off, and an infinite vacuum does not provide that. If verified, this implies the law was abused to sell humanity on the idea of space travel.” The page also flags, in its own words, that a common way this kind of test goes wrong is using a vacuum chamber that's too small — worth taking seriously, and returned to with numbers in §9 and §10 below. As of this writing the referenced experiment by Alex Kampf is listed as pending upload, so this page argues from the claim as stated and from physics, telemetry, and independent data already public, not from a result that doesn't exist yet.

2 · What's worth conceding up front

The chamber-size caution is real, and it's good instinct to flag it before running the test rather than after. A vacuum chamber that's small relative to the thrust and burn duration will re-pressurize from the exhaust itself faster than it can be pumped down, especially with a valve or piston releasing compressed gas rather than a sustained combustion exhaust plume. A test like that can look like thrust “fails” in vacuum when what actually failed was the vacuum. That's a legitimate methodological trap, and it's exactly the trap that real propulsion tests — amateur and industrial alike — are built to avoid. More on that in §9, and the arithmetic for his own chamber in §10. Worth flagging now, though, that chamber size has a second face which cuts the other way and which neither side has raised: a chamber small enough for the exhaust to cross it almost instantly makes any timing comparison uninformative, whichever model is right. A test of Kampf's Law needs a chamber big enough to fail that trap as well as the pressure one.

3 · How a rocket actually pushes

Before any thought experiment, the plainest true statement, because it is the one the claim runs aground on and it is rarely said out loud.

The burning gas pushes on the inside of the engine. The back of the engine has a hole in it, so nothing pushes back.

A sealed pressure vessel goes nowhere: gas presses outward on every wall and the pushes cancel. Cut a hole in one end and the cancellation stops. Pressure still bears on the closed forward wall of the combustion chamber; across the open end there is nothing to bear on. That unbalanced push on the vehicle's own structure is the thrust, and it is present before a single molecule leaves the nozzle. What the exhaust does afterwards — whether it meets air, or a wall, or nothing at all — is no longer the rocket's business.

4 · First principles: a thrown chunk of mass is still mass

Newton's third law isn't a law about air. It's a law about how mass affects mass.

Strip away rockets, nozzles and exhaust entirely for a second. Take a large block sitting still. Bolt a small machine to one edge of it that continuously throws off tiny chunks of the block's own mass — say 1% of the block at a time — sideways, at some fixed speed. The moment the first chunk leaves, the block has to move the other way. Not because anything pushed on the block from outside; because momentum has to balance, and it balanced at zero before the machine ever fired. There's no step in that story that requires air, a wall, or anything else nearby. It works the same way sitting in the middle of empty space as it does sitting on a table.

Now keep the machine running at a constant rate — same size chunk, same throwing speed, over and over. Two things follow, and both are just F = ma rearranged: the force stays constant (same mass, same speed, thrown at the same rate), but the block's own remaining mass keeps shrinking as it gives pieces away. Constant force over shrinking mass means the block's acceleration keeps climbing the whole time the machine runs — slowly at first, sharply near the end, right up until the machine runs out of chunks to throw. That's not a rocket-specific fact. It's true of any object continuously throwing part of itself away at a steady rate, and you don't need to know a single thing about combustion chemistry or nozzle geometry to derive it.

The only place a gas exhaust could possibly be special is if gas were somehow exempt from having mass — and it isn't. A kilogram of hot combustion gas leaving a nozzle at 3,000 m/s carries exactly as much momentum as a one-kilogram block thrown at 3,000 m/s. The demo below runs exactly the thought experiment above: a shrinking mass throwing off small chunks at a constant rate, with the live numbers — mass remaining, velocity, and acceleration — updating as it runs so you can watch acceleration climb as the mass burns down, before a single real rocket number appears on this page.

5 · Version two: add resistance at both ends

Real rockets don't fly in true vacuum until they're well clear of the pad, so it's worth extending the same thought experiment honestly rather than pretending air isn't there. Put the same block-and-machine setup inside a column of air instead of empty space. The block now has to shove air out of its own way as it accelerates — that is drag, and it is a real loss, but it is a loss acting on the vehicle, not a reduction in the thrust.

That drag isn't fixed — it depends on how much air is actually there — so it shrinks as the block climbs into thinner air, and the same machine delivers a larger share of its effort into moving the block — simply because there is less air in front of it to push aside.

Drag is not, however, why a nozzle is rated for more thrust in vacuum, and it is worth getting that right — because the real reason cuts against the claim rather than merely failing to support it.

That is also the technical form quoted in the note below §6. The A_e(p_e − p₀) term is precisely this: pressure at the nozzle exit acting outward, minus the ambient pressure that would otherwise have acted inward across the same area. Nothing in it refers to what the exhaust meets after it leaves — and nothing could, since the vehicle's momentum is settled at the exit plane.

6 · What the standard model predicts — and what you can check against the webcast

This is the same shape as the demo in §4, at real Falcon 9 Block 5 numbers instead of toy units: 9× Merlin 1D, sea-level thrust 7,607 kN, vacuum thrust 8,227 kN, ~411 t propellant, engine cutoff ~162s, run against the U.S. Standard Atmosphere 1976 density model. None of that has to be taken on faith either — SpaceX webcasts every launch live with an on-screen speed/altitude readout, and every mission's recording stays publicly archived afterward, searchable by mission name, on SpaceX's own YouTube channel — for example, this Falcon 9 Starlink launch from Cape Canaveral shows exactly that readout in the corner of the feed. As §7 below covers in more depth, independent enthusiasts hand-log those same readouts frame by frame with no SpaceX involvement in the analysis. The chart below is a simulation, and should be labelled as one on a page arguing that you need not trust anybody's numbers: it runs those published parameters against the standard atmosphere, and plots the result against a hypothetical “air pusher” model where thrust is simply scaled by ambient air density — the literal version of “thrust needs a medium” — run through the identical fuel burn and mass-loss schedule, so the only variable that differs is what creates the thrust.

A few things worth reading directly off that chart rather than skimming past. First, the real curve only runs through T+162s, main engine cutoff (MECO) — that's not a cherry-picked cutoff, it's the window where the webcast overlay and independent frame-by-frame trackers are both actually watching the same first-stage burn; second-stage flight happens on a different engine in a different regime — the same physics throughout — and isn't part of this comparison. The plotted quantity is net acceleration: thrust less drag and the gravity component along the flight path, divided by the falling vehicle mass. The Max‑Q plateau is a throttle schedule fed into the model, not an emergent result. Second, the real curve isn't a smooth monotonic climb — there's a visible flattening between roughly T+50s and T+82s, where the simulated acceleration briefly plateaus around 1.4–1.9g instead of continuing to rise. That is not noise, and it is not the phenomenon this page is arguing against — it is Max‑Q, the peak of dynamic pressure ½ρv², which builds as the vehicle accelerates and then falls away as the air thins, topping out somewhere around 11–13 km. Falcon 9 is deliberately throttled down through that window to keep dynamic pressure on the airframe in check — a real engineering constraint on a real airframe, not a claim about what air can push. Compare that to the “air pusher” curve: its thrust has already fallen to roughly 60–65% of sea-level value by T+50s, purely because it's a few kilometres up and the model ties thrust directly to ambient density — well before the real vehicle even reaches the altitude and speed where the legitimate Max‑Q throttle-down happens. One is a deliberate, temporary, physically-justified pullback timed to airframe loads; the other is an unrecoverable structural collapse of the entire propulsion concept, and it starts almost immediately after liftoff.

The clearest way to see where those two predictions actually diverge is speed, not acceleration — because speed is the thing that determines whether a vehicle reaches orbit or falls back down:

The modelled curve keeps building speed the whole way to MECO, exiting the observed window at roughly 3,050 m/s in this run — higher than a webcast will show, because the simulation spends the full propellant load while a real booster reserves ten to fifteen per cent for boostback and landing, putting actual Starlink MECO nearer 2.2–2.5 km/s and continuing to climb from there toward orbital velocity (~7,800 m/s) once the second stage takes over — not shown here because it is a different stage and engine, though the physics is identical. The “air pusher” curve does something categorically different, and it's worth reading the speed line carefully rather than assuming it simply falls once trouble starts: altitude peaks at about 7.9 km around T+1:35 and sinks from there — but total speed keeps climbing well past that point, for a reason that's really just this model's core flaw playing out a second time. As the vehicle sinks back down it's descending into denser air, which in a thrust-follows-density model means more thrust, not less; by the time the shared pitch-over program (the same one steering the real ascent) has tipped that thrust to about 15° off horizontal, it's diving through increasingly dense low air at up to roughly 75% of sea-level thrust, pouring nearly all of it into horizontal speed instead of altitude. That doesn't stop until the same propellant load the real vehicle burns runs out — T+2:42, the identical instant as the real MECO, because both curves share one fuel schedule. Only then does thrust actually reach zero; freefall and drag take over, and about seventeen seconds later, around T+2:59 — call it the T+3 minute mark — it's back at sea level, crashed at over 1,000 m/s, having burned its entire propellant load without ever reaching orbital speed or getting anywhere near back to its own earlier peak altitude. That's not an ambiguous or marginal outcome; it's what “thrust needs a medium” actually predicts once you run the model instead of just asserting it, and it is not what any of the thousands of public orbital launches on record have ever shown happening.

For the technically inclined: the equation aerospace engineers actually use to size a nozzle is F = ṁ·V_e + A_e(p_e − p₀) (NASA Glenn, Thrust Equation). The first term is pure momentum thrust — medium-independent, exactly §4's block-and-machine result. The second term is the pressure-difference correction from §5: as ambient pressure p₀ falls, that term grows, adding thrust rather than removing it.

7 · You don't have to trust anyone's numbers to check this

Everything in §6 used SpaceX's own published thrust and mass figures, and it's fair to ask why those should be believed. They don't have to be. Altitude, speed, and time are visible directly on the public launch webcast, and acceleration is just the rate those change — nothing about deriving it requires trusting a single claimed thrust or propellant number from SpaceX at all. Independent, unaffiliated communities already do exactly this: enthusiasts on forums like NASASpaceflight.com have hand-logged on-screen velocity and altitude readouts frame by frame from Falcon 9 webcasts, and independent, community-built tools such as Flight Club — a physics-based launch trajectory simulator built and run by a private individual with no SpaceX affiliation, that integrates thrust, drag, and gravity itself rather than replaying SpaceX's numbers — visualize and cross-check that same public flight data with no launch-provider involvement in the analysis. Back out the acceleration curve from externally observed altitude-and-speed data alone, with no claimed thrust or mass figures anywhere in the calculation, and it lands on the same climbing shape in §6 — not because independent observers were told to expect it, but because that's what the video feed itself shows happening.

8 · Independent of thrust mechanism: the air is functionally gone long before it would matter

Even setting the thrust-mechanism question aside, “no medium to push against” isn't some exotic condition unique to a vacuum chamber or deep space — it's the normal state of the sky well below where second-stage burns and orbital insertion happen. The chart below extends the same atmospheric model up past the Kármán line (100 km, the conventional edge of space) to low Earth orbit altitudes like the ISS (~423 km).

Pressure at 80 km is already about 0.001% of sea level (1.05 Pa); at the Kármán line, 0.032 Pa — roughly three ten-millionths of sea level, a good laboratory vacuum rather than anything resembling interplanetary space, which is more than ten orders of magnitude lower again. Worth noting for §9: NASA Glenn's VF-5 chamber at 10⁻⁷ torr is a harder vacuum than the Kármán line by three orders of magnitude. If a rocket's thrust genuinely depended on ambient air, it would have already failed by the time a Falcon 9 second stage lights, tens of kilometres below where anyone claims spacecraft do their real work. The claim would predict engine failure at an altitude where every orbital mission on record is still climbing normally.

9 · This isn't just NASA saying so

It's fair to be skeptical of taking any single institution's word for something this consequential, and this claim in particular is often aimed squarely at NASA's credibility. So set NASA aside for a moment and look at the same physics at completely different scales, run by people with no stake in the space program at all.

At the small end, amateur and hobby rocketry has been measuring motor thrust on load cells for decades, with no institution required. A sea-level static fire does not by itself settle the vacuum question — nobody disputes that rockets work in air. What it settles is the momentum budget, and that is the check worth running: take any motor's total impulse from the database, divide by its propellant mass, and you get an effective exhaust velocity of roughly 1.8–2.3 km/s for ammonium-perchlorate composite propellant. That is what combustion chemistry and nozzle theory predict from the exhaust alone. Being careful about what that does and does not exclude: a 1.8–2.3 km/s spread is a wide band, so this rules out the atmosphere supplying a large share of the thrust — which is exactly what the claim requires, since “thrust needs a medium” needs the medium to supply essentially all of it. It does not resolve a contribution of a few per cent, and does not need to. If anything the measured figure sits at the low end of the theoretical one, which is the wrong direction for the claim: at sea level the ambient pressure term of §5 is subtracting from thrust, not adding to it. ThrustCurve.org is an independent, community-run database (maintained by a private individual, not any government or agency) of over a thousand certified motor thrust curves contributed by hobbyists and testing organizations with no connection to NASA or spaceflight programs. Building your own thrust-measuring rig is genuinely a weekend project: amateur rocket engineer Richard Nakka's guide to a DIY strain-gauge load cell puts the load cell itself at about $10–20 in parts (the rest of a test stand is simple hardware around it) and produces a real, directly measured thrust-vs-time curve on a home motor test stand — the same measurement this page is arguing about, done by one person in a garage.

At the large end, but still nothing to do with NASA, publicly documented missile intercepts show the same physics operating under combat conditions, verified by multiple independent militaries with every incentive to get it right and none to fake it: Arrow-2 interceptors reaching roughly 50 km altitude by design, Arrow-3 — whose exoatmospheric engagement ceiling runs over 100 km — used repeatedly since its first combat interception in November 2023, including during the April 2024 Iranian attack, and a first combat-verified THAAD intercept in December 2024, a system rated to engage threats up to 150 km. None of that hardware is run by a space agency selling the public on space travel; it's missile-defense engineering that has to work the first time, cross-checked by the intercepting country, the country that launched the threat, and outside observers, all in agreement about where these interceptions happened. Further detail on individual systems is catalogued at the non-governmental CSIS Missile Defense Project.

Only after that does NASA's own data belong in the conversation, and it belongs there as one more point on a curve that amateur and military sources have already traced — not as the source of the curve. NASA Glenn's Electric Propulsion and Power Laboratory has been testing thrusters in hard vacuum since the 1960s; its largest chamber, Vacuum Facility 5, is 4.6 m in diameter and 18.3 m long, pumped down to roughly 1×10⁻⁷ torr — sized specifically so a firing thruster's own exhaust can't re-pressurize the chamber faster than it's evacuated, precisely the failure mode §2 flagged. Ion and Hall-effect thrusters have been hot-fired there for thousands of hours with thrust measured directly on a stand — and if the objection is that those are not gas rockets, though the book files them under this same claim at p. 183, then chemical upper-stage engines are fired routinely in altitude cells outside NASA too: DLR Lampoldshausen’s P4.1 for the Vinci engine, the USAF Arnold Engineering Development Complex J-4 and J-6 cells, and JAXA Kakuda. Thrust in all of them is read at the engine mount with cell pressure logged, upstream of any diffuser. Back at Glenn, the longest of those runs is NASA's own NEXT ion engine, fired for over 48,000 hours — roughly five and a half years of continuous firing, the longest-duration test of any space propulsion system on record. What NASA's numbers add isn't a different claim than the garage load cell or the missile intercept — it's the same relationship extended to a bigger engine and a longer burn. The curve doesn't change shape depending on who's measuring it.

One gap remains in all of that, and it is worth closing rather than glossing: every source above except the missile intercepts is measured on the ground, where a chamber can always be accused of not being empty enough. Engines have also been measured firing in real space — and, more usefully, measured several physically unrelated ways at once, so the methods can be checked against each other. A 1970s ion thruster's thrust was determined from an onboard accelerometer, from the change in its orbit, and from beam telemetry calibrated on a ground stand, and the three agree to a few per cent. That record, and what it says about the size of the chamber artefact, is set out on its own page: Thrust, Measured in Flight. The short version is that the artefact is real, is about one per cent, and runs in the direction that makes chambers flatter engines slightly — two orders of magnitude away from what the claim needs.

9a · Somebody already built this, in 2018, and filmed every failure

Eight years before the book, a YouTube channel called Warped built a vacuum chamber for exactly this purpose and fired real solid rocket motors inside it — Rockets in a Vacuum Chamber — Newton’s Third Law of Motion Visualized, uploaded 17 February 2018, about sixteen minutes. He says at the outset that flat-earth material had got into his head and he wanted to challenge the third law properly, so he sourced actual solid propellant from an amateur rocketry association rather than using the black-powder motors of his earlier attempts.

The result matters here because of what Kampf’s Law asserts: that gas propulsion needs an external substrate to compress against, and that with the substrate removed “thrust will be zero.” A solid rocket motor ejects gas. So a solid motor burning in a chamber at vacuum, with visible thrust, is a direct counter-instance — and it was filmed, published and left standing for eight years before the claim was printed.

What that video hands the other side, stated first

It would be dishonest to cite this as a clean win, because three of its five tests failed and the creator says so plainly.

Attempts one, two and three did not ignite. Reviewing the first failure he says there is some truth to what the conspiracy theorists think — about ignition specifically, not propulsion. The fix on the fourth attempt was a rupture disk over the nozzle, holding a pocket of atmospheric pressure inside the motor casing until it lit. Read uncharitably, that is: he had to put the air back to make it work. He also gives no numeric thrust figure in the narration, and notes as a curiosity that the main ignition flash seemed to happen out in the chamber rather than immediately behind the nozzle.

Ignition and thrust are two different problems

Solid propellant needs chamber pressure to light and to spread flame across the grain. Near vacuum, the igniter’s gases expand away before they can heat and pressurise anything, so the motor hangs fire. That is a statement about starting a fire, not about what the fire does once it is burning.

The reason this is not a rescue is that the fix is standard flight practice. Solid motors intended to ignite in space are built with nozzle closures — a diaphragm sealing the nozzle that holds internal pressure until burn-through — for this exact reason. The rupture disk is an amateur version of hardware that exists on flight motors because ignition, and only ignition, needs pressure. If ambient gas were what produced thrust, no closure would help, because the closure seals the inside of the motor and does nothing whatever about the vacuum outside it.

Two things the footage settles, and one it does not

We went through the high-speed sequence of the fourth test frame by frame rather than relying on the narration.

The ignition jet is attached to the nozzle. On the on-screen high-speed clock, a small orange jet appears at the nozzle throat and, over the next fifty milliseconds, extends downstream as a narrow bright plume anchored at the nozzle. The large luminous fireball that fills the chamber arrives about a further hundred milliseconds later and sits well downstream. So the flash he found interesting is that second event, not the ignition — the thrust-producing jet is exactly where a nozzle should put it, and the page should not be read as conceding otherwise.

There is an instrument, not just an impression. The motor is mounted in line with a Pelouze spring dial reading to 50, so the rig does carry a thrust scale rather than relying on visible movement alone. What we could not establish is whether the needle deflects: our attempt to measure its angle across sampled frames kept locking onto the orange igniter lead crossing the shot, and three frames from a slow-motion replay are not enough to rule on it either way. So we claim nothing about what the scale read, and anyone citing this video for a thrust value is going beyond what it shows.

And the chamber stops being a vacuum during the burn. This is the caution that matters most, and it applies to every home version of this experiment including the book’s. A burn dumps combustion gas into a sealed volume. For a chamber of roughly this size and a modest solid grain — take 40 g of propellant at a mean molar mass near 25 g/mol, in about 0.2 m³ — the gas released is around 1.6 moles, which is 0.2 atm once cooled to room temperature and roughly half an atmosphere while it is still hot. Those are order-of-magnitude figures from estimated dimensions, not measurements. The point stands regardless of the exact numbers: only the first moments of a burn are a vacuum test, which is precisely the interval the creator points at when he says thrust was produced at initial ignition.

That has a direct consequence for the rig described in the book. Gas released into a small chamber raises its pressure, and moving the wall to change the standoff distance also changes the free volume behind it. Wall proximity and pressure rise are therefore confounded in any short chamber, and separating them needs either a much larger volume or a pressure trace during the run. That belongs in the list of things to fix before the result means anything, not after.

And the quantitative point runs the opposite way

Thrust is F = ṁv_e + (p_e − p_a)A_e. The ambient pressure enters with a minus sign. Removing the atmosphere therefore adds exactly p_atmA_e to the thrust, whatever the motor is doing internally:

An Estes D12 averages about 12 N of thrust at sea level. The same motor in vacuum gains roughly 8 N — about two-thirds more. This is why vacuum specific impulse is quoted higher than sea-level specific impulse for every engine ever flown, and why upper stages use large expansion ratios that would be useless in air. The sceptical prediction is not merely wrong in sign; the size of the effect it gets backwards is a routine engineering number.

The flash appearing out in the chamber is what an exhaust plume does with no back-pressure to confine it: it expands enormously the moment it leaves the nozzle. We have not measured that against the footage, and we flag it as expected rather than verified.

10 · What standard physics predicts for that rig, in advance

The book does not treat Claim #2 as pending: p. 82 names Kampf's Law as a law, and p. 183 says the experiment “conclusively proved these animations are nothing more than CGI.” The design is on p. 80 — a syringe on a linear track inside a tube, with the end wall set at two distances, scoring “a more delayed reaction… as we increase the length of the chamber” as the discriminator — but the run itself isn't in the book, so there is nothing yet to check it against. Rather than wait, here is what the mainstream answer expects from that apparatus, set down in advance so the result can be judged against it instead of interpreted afterwards.

One caveat on the three figures below: the book gives no volume, gas, pressure, mass or exhaust speed, so the 60 mL of air, 50 g carriage and few-hundred-metres-per-second exhaust are illustrative values of the right order, not his. Every prediction scales with them, and we would revise against his the moment they are published.

• — Expect a null result, on friction alone. Sixty millilitres of air at atmospheric pressure is about 73 milligrams. Expelled at a realistic few hundred metres per second, the entire impulse available is roughly 0.026 N·s — delivered over half a second, about 0.05 N of thrust. The static friction of a 50 g carriage on an ordinary track is about 0.05 N. The rig is at the threshold where it may not move at all, and a stationary syringe would demonstrate nothing about Newton. A pendulum or torsion balance is what every real thrust stand uses, for this reason — and, as it happens, what The Action Lab used in the video the book cites at p. 20, hanging its syringe on threads. That rig moved. The difference is the mounting, not the physics — and to be clear, the 60 mL and 50 g figures in this bullet are illustrative values for the p. 80 track rig, not measurements of that syringe.
• — Wall distance is the wrong discriminator — for how far it moves. Thrust is set at the nozzle, and nothing downstream of the exit plane reaches back to change it. A near wall can perturb that, but only by returning gas to the vehicle, and the perturbation has two parts of opposite sign: rebounding exhaust striking the base pushes forward, arriving late; accumulating exhaust raises ambient pressure and eats into thrust through the p₀·Aₑ term of §5, which in a small chamber is the larger. “The near wall changed the result” is therefore not a clean prediction in either direction. What would distinguish the models is thrust holding constant as the tube lengthens with chamber pressure logged throughout — and the onset timing, which the next section takes on its own terms.
• — And the chamber will not stay empty. Releasing 73 mg of air into a 20-litre chamber raises its pressure by about 0.3 kPa; into a 2-litre chamber, about 3 kPa. That is the quantitative version of the caveat conceded in §2 — and stating it in advance is what stops it becoming an all-purpose escape hatch for whichever side dislikes the answer.
• — And the chamber is unlikely to be long enough for a camera to see anything. What made The Action Lab's footage readable at all is that its chamber was large compared with its jet: at 19 mbar the coherent jet died within roughly 0.1 m, and the crossing was finished by smoke at metres per second — slow enough to stretch the timing onto a scale a camera can resolve. Even that human-sized chamber, roughly 0.4 m from nozzle to wall, was only barely in the readable regime. A benchtop tube puts its wall inside or just past the jet's own penetration length, where the exhaust arrives coherently within a millisecond, both accounts predict contact before visible motion, and the onset difference between them sits far inside one frame. At tube scale the discriminator is a load cell or nothing.

The delayed reaction, and what would actually decide it

The discriminator the book proposes on p. 80 is timing: “a more delayed reaction from the syringe as we increase the length of the chamber” would mean the vehicle is relying on the surrounding gases. p. 20 applies the same reading to The Action Lab's vacuum-chamber launch, taking the vehicle to start moving when the exhaust reaches the far wall rather than when it leaves the nozzle.

That video exists, so we measured it rather than arguing about it. The delay is real and is conceded — and so is the ordering the book expects: the exhaust reaches the wall about two seconds of playback before the syringe moves. The ordering carries no information, though: a push too small to see and a push not yet arrived look identical on video, so the coarse ordering is compatible with either account — and in a small chamber it is outright forced, by the arithmetic below. What the footage does rule out is the wall as the cause: the gap between contact and motion is 25 to 220 times too slow for a pressure wave, the emission history is featureless so returning gas has nothing to copy, and the acceleration answers the propellant flare at the nozzle within one or two frames. The frames, the arithmetic and the caveats are on their own page: The Action Lab Footage, Frame by Frame. What follows is the p. 80 rig's own regime — because a tube behaves very differently from a room, and nobody has published frames of that rig at all.

Start with what makes the timing argument feel compelling: the gas moves visibly, immediately and fast, while the vehicle appears to do nothing. That asymmetry is the third law being displayed, not violated. Momentum is conserved, so the ratio of speeds is the inverse ratio of masses — 73 mg of air against a 50 g carriage is 685 to 1. Exhaust leaving at 500 m/s leaves the vehicle at 0.73 m/s, and it takes the whole burn to get there. Equal and opposite refers to momentum, not to speed.

Now put a clock on it. At 0.05 N on 50 g the carriage needs about 45 milliseconds to travel a millimetre — the first displacement that reads as motion on video — and friction, ignored here, would only lengthen that. In those 45 ms an exhaust front at 500 m/s covers 22 metres. Run it the other way and the vehicle has moved half a micron when the cloud reaches a wall at 0.5 m, 8 microns at 2 m, and 0.2 mm at 10 m — still under the visible threshold.

Which is not to say the timing test is worthless. It is a good test, run wrong. The two models disagree sharply about the onset of motion: standard physics says onset is set by thrust and friction and is therefore independent of chamber length — the same 45 ms at half a metre and at two — while the book's mechanism gates it on the gas arriving somewhere, so it should scale with length. On p. 82 the gas compresses at the wall and pushes back on the vehicle, which is a round trip: 2 ms against 8 ms across that same pair, at the same 500 m/s used above. Double the tube, double the delay. That is a clean, pre-registered, falsifiable difference, and it does not need a 22-metre chamber.

What it needs is the right instrument, and that is not a camera. A camera measures displacement, which is the double integral of a force still building — and in a tube, even at 2,000 fps, the vehicle has moved a fraction of a pixel by the time the gas lands. (In a chamber large enough that the crossing takes visible time, as the companion page shows, a camera can resolve the interval — but that is a property of the chamber, not of the camera.) A force transducer registers the onset promptly: a piezo or strain-gauge load cell between vehicle and mount, kilohertz bandwidth, responding to thrust rather than to the motion thrust eventually produces. Put a second on the target wall and both events land on one clock. Run that at two clearly different wall distances with chamber pressure logged, and it decides the question in an afternoon.

A last point in fairness to the design. A wall can only affect the vehicle by sending gas back to it — nothing happens at the moment of the far-off impact itself, since the vehicle cannot know about it. Both of the competing near-wall terms above are perturbations on a motion that began at the nozzle, and neither is a delayed start. Total displacement cannot cleanly separate the models. The onset clock can — and in a tube it takes a force trace to read it, because the two predictions sit inside a millisecond of each other.

11 · Where we're genuinely willing to wait and see

In fairness: a new, independently run vacuum-chamber test isn't a bad idea just because the physics is already well established elsewhere — replication is a fine instinct, and Shape Debate's own page correctly names the single most common way an amateur version of this test goes wrong. What actually determines whether the Kampf test lands as evidence either way is whether the chamber and pump-down are large enough, relative to the thrust and burn duration used, that the chamber pressure stays low throughout the firing rather than rising as exhaust accumulates — exactly the sizing problem §9's amateur and NASA facilities were both built to avoid. That's a checkable, falsifiable detail once the results are up: if thrust holds steady while the chamber stays at a documented low pressure throughout the burn, that confirms standard physics; if thrust collapses while the chamber pressure is simultaneously rising, that shows a chamber-sizing problem, not a Newton's-third-law problem. And if the claimed evidence is the timing rather than the magnitude, the chamber has to be large enough that the exhaust does not cross it in a frame — otherwise the delayed reaction the rig is built to look for is guaranteed by the mass ratio and means nothing.

The claim is 106 years old, and the paper that printed it has retracted it

None of this is new, and the history is worth having because it is unusually clean.

In 1919 Robert Goddard published A Method of Reaching Extreme Altitudes, having already — in 1916 — fired rockets inside evacuated chambers, precisely to test whether they needed air to push against. They did not; they worked better without it, for the pressure-term reason set out in §5.

On 13 January 1920 the New York Times ran an editorial saying that Goddard “does not know the relation of action to reaction, and of the need to have something better than a vacuum against which to react.” That is Claim #2, in a national newspaper, word for word, a century early.

The honest bottom line

The claim's own physical intuition — “thrust needs something to push off” — correctly describes propellers and jet engines, which really do stop working without air. It just doesn't describe a system throwing its own mass overboard, because Newton's third law was never about air in the first place — it's about mass affecting mass, and exhaust gas has mass like anything else. That's true before a single rocket number gets involved, it's true in real Falcon 9 telemetry you can independently verify from the public webcast without trusting anyone's claimed specs, and it's true at every scale anyone has bothered to check — a $15 garage load cell, a combat-verified missile intercept, and a sixty-year-old NASA vacuum chamber all trace the same curve. NASA's numbers are just that curve drawn out to a bigger engine, not a separate claim resting on trusting NASA.

Sources & further reading

• Warped, Rockets in a Vacuum Chamber — Newton’s Third Law of Motion Visualized — YouTube, uploaded 17 February 2018, 15:44. The independent replication discussed in §9a: a purpose-built chamber, solid propellant supplied by an amateur rocketry association, three failed ignitions, then two successful burns with visible thrust under vacuum. Our account of it rests on a condensed summary of the automatic captions rather than on the audio, so nothing from it is quoted here inside quotation marks; the creator’s remark about the conspiracy theorists being partly right on ignition is paraphrased for that reason and should be checked before it is ever quoted.
• R. H. Goddard, “A Method of Reaching Extreme Altitudes,” Smithsonian Miscellaneous Collections 71(2), 1919 — including the evacuated-chamber firings of 1916. The New York Times editorial of 13 January 1920, and its correction of 17 July 1969.
• Sutton & Biblarz, Rocket Propulsion Elements, ch. 2–3 — the thrust equation, pressure thrust, and sea-level against vacuum specific impulse.
• Feynman, Lectures on Physics I-39 — kinetic theory, and why gas pressure on a wall is momentum flux.
• Vacuum and altitude test cells outside NASA: DLR Lampoldshausen P4.1 (Vinci upper-stage engine, ESA/ArianeGroup); USAF Arnold Engineering Development Complex J-4 and J-6 cells; JAXA Kakuda. Thrust in these is measured at the engine mount with cell pressure logged, upstream of any diffuser.
• U.S. Standard Atmosphere 1976 — 1.05 Pa at 80 km, 0.032 Pa at 100 km.
• Shape Debate, Claim #2 (shapedebate.com), and the referenced pending vacuum-chamber test by Alex Kampf, “Propulsion Tester.”
• NASA Glenn Research Center, Thrust Equation.
• NASA, Electric Propulsion and Power Laboratory, Glenn Research Center — Vacuum Facility 5/6 specifications and the NEXT ion thruster program; endurance-test duration per NASA, NEXT Provides Lasting Propulsion and High Speeds for Deep Space Missions.
• ThrustCurve.org — independent, community-maintained database of certified amateur and commercial motor thrust curves, no NASA affiliation.
• Richard Nakka, DIY strain-gauge load cell for thrust measurement — amateur rocketry reference site.
• CSIS Missile Defense Project — non-governmental catalogue of public missile and interceptor specifications; Arrow-2 altitude spec, plus event-specific sources for the Arrow-3 and THAAD intercepts cited in §9: Arrow 3, The War Zone on THAAD's first combat use, and Army Recognition's THAAD specifications.
• NASASpaceflight.com forum — independent enthusiasts hand-logging Falcon 9 webcast velocity/altitude telemetry.
• Flight Club — independent, community-built rocket trajectory simulator and visualizer, no SpaceX affiliation.
• U.S. Standard Atmosphere, 1976 — density/pressure model used for both physics charts on this page.
• Falcon 9 Block 5 public performance parameters (engine count and thrust, propellant mass, approximate engine cutoff time) per SpaceX's own Falcon Payload User's Guide and widely reproduced in spaceflight references.