Linearly accelerating frames

In 1907, working at the Bern patent office, Einstein was struck by what he later called "the happiest thought of my life": a person falling freely from a roof does not feel their own weight. Tools released mid-fall would float alongside; a dropped apple would hover. Gravity — the most obvious force in the universe — can be switched off completely, everywhere in a falling room at once, by nothing more than a change of reference frame. And what a falling frame can switch off, an accelerating frame can switch on: fire a rocket at $9.81\ \mathrm{m/s^2}$ in empty space and the crew walks its floor exactly as you walk on Earth. The stage for that idea is the simplest non-inertial frame there is: one whose origin accelerates with $\mathbf{A}(t)$ while its axes stay resolutely parallel to an inertial frame's — no rotation, ever. Where a rotating frame demands three separate correction forces, the linearly accelerating frame needs exactly one: the uniform fictitious force $-m\mathbf{A}$ , the same per kilogram at every point of the frame. Combined with real gravity it defines an effective gravity $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$ that tilts plumb lines in accelerating trains, retunes pendulum clocks, sloshes coffee to a new level surface, and reads out on elevator scales as apparent weight. This article builds that machinery from one differentiation, runs it through the classic laboratory of trains, elevators, and rockets, and follows it to the doorstep it famously opens: the equivalence principle, where the accelerating frame stops being a trick and becomes the blueprint for general relativity.

Key takeaways

A linearly accelerating frame translates with origin acceleration $\mathbf{A}(t)$ but never rotates; positions transform by simple vector addition and accelerations by $\mathbf{a}_{\mathrm{in}} = \mathbf{a}_{\mathrm{fr}} + \mathbf{A}$ .
Newton's law survives in the frame with exactly one correction: the uniform fictitious force $\mathbf{F}_{\mathrm{fic}} = -m\mathbf{A}$ — no centrifugal, Coriolis, or Euler terms, which all require rotation.
Folding $-m\mathbf{A}$ into gravity defines effective gravity $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$ : plumb lines, fluid surfaces, and pendula all obey ordinary statics with $\mathbf{g}$ replaced by $\mathbf{g}_{\mathrm{eff}}$ .
Horizontal acceleration tilts equilibrium by $\tan\theta = A/g$ and strengthens it to $g_{\mathrm{eff}} = \sqrt{g^2 + A^2}$ ; vertical acceleration rescales apparent weight to $N = m(g+A)$ .
Free fall is the perfect special case: $\mathbf{A} = \mathbf{g}$ makes $\mathbf{g}_{\mathrm{eff}} = 0$ — weightlessness is gravity transformed away, not gravity absent.
Because $-m\mathbf{A}$ is exactly proportional to mass, no interior mechanical experiment distinguishes uniform acceleration from uniform gravity: Einstein's equivalence principle, the on-ramp to general relativity.
Key idea: in a non-rotating frame whose origin accelerates with $\mathbf{A}(t)$ , Newton's law becomes $m\mathbf{a}_{\mathrm{fr}} = \mathbf{F}_{\mathrm{real}} - m\mathbf{A}$ — a single uniform fictitious force, identical in form to a uniform gravitational field, restores the inertial form of the law.

The core insight

In a non-rotating frame whose origin accelerates with $\mathbf{A}(t)$ , Newton's law becomes $m\mathbf{a}_{\mathrm{fr}}=\mathbf{F}_{\mathrm{real}}-m\mathbf{A}$ : a single uniform fictitious force $-m\mathbf{A}$ , identical in form to a uniform gravitational field, restores the inertial form of the law.

Where this lives today

The equivalence principle is now among the most precisely tested statements in physics, and the tests are direct descendants of this article's thought experiments. The MICROSCOPE satellite mission flew two concentric test masses of different composition in a drag-free orbit — a perpetual Einstein elevator — and reported in its 2022 final results that they fall identically to about one part in $10^{15}$ ; a violation at any level would have announced physics beyond general relativity. Successor concepts push toward $10^{-17}$ with cold-atom interferometers, which drop clouds of two different atomic species in the same vacuum tower and read the differential acceleration with matter waves — Galileo's Leaning Tower experiment, rebuilt at the quantum scale. The engineering side is everywhere you stop noticing it. Every smartphone accelerometer measures precisely $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$ — the specific force of this article — which is why your phone knows screen-up from screen-down and why it reads zero when dropped (a two-metre free fall is a genuine, if brief, Einstein elevator; some laptops use exactly that signature to park the hard drive before impact). Zero-g research aircraft fly $\mathbf{A} = \mathbf{g}$ parabolas twenty seconds at a time for training and microgravity experiments; drop towers buy quieter versions of the same frame. And every GPS fix you take silently honours the elevator's 1907 corollary: clocks in different effective gravity tick at different rates, and the constellation's atomic clocks are rate-corrected for it — the happiest thought, shipped as infrastructure.

The physical picture

Board a smoothly accelerating train and watch the interior physics reorganize itself. The strap-hangers lean; a plumb line settles at a fixed backward tilt; the surface of your coffee tips and stays tipped, perfectly flat at its new angle; a dropped phone falls not straight down but along the tilted "vertical." Nothing inside the carriage points at the old down anymore. The frame has voted, unanimously, for a new gravity: the vector sum of the old $\mathbf{g}$ and the backward fictitious field $-\mathbf{A}$ .

Figure

Inside an accelerating train — one fictitious force retilts the local vertical

Why exactly one fictitious force? A frame can misbehave in only two ways — its origin can accelerate, and its axes can turn. Pin the axes parallel to an inertial frame's and the second sin is gone: no basis vector ever changes, so time derivatives commute across frames with no $\boldsymbol{\omega}\times$ corrections, and the whole rotating-frame catalog — centrifugal, Coriolis, Euler — stays locked away (the general machinery lives in the non-inertial reference frames article, and the velocity-dependent member has its own Coriolis force deep-dive). What remains is the origin's acceleration, entering every ledger once, as the uniform field $-m\mathbf{A}$ . Uniform is the loaded word. The centrifugal force grows with distance from the axis; the Coriolis force reads your velocity; but $-m\mathbf{A}$ is the same vector per unit mass at the front of the train and the back, for the sprinting conductor and the parked luggage. A force field that is (i) uniform in space and (ii) exactly proportional to mass has precisely the résumé of a uniform gravitational field — and that is not a metaphor but an operational identity. Seal the windows and every experiment with masses, springs, and pendulums returns the same readings whether the carriage accelerates on a level track or hangs tilted in a stronger gravity.

Figure

Einstein's elevator — uniform acceleration and uniform gravity are interchangeable indoors

Einstein's move was to run the equivalence in both directions. Accelerating box mimics gravity; therefore falling box cancels it: put the frame in free fall, $\mathbf{A} = \mathbf{g}$ , and the fictitious force $-m\mathbf{g}$ erases every object's weight simultaneously. The astronauts "floating" on a space station are not beyond Earth's gravity — at 400 km altitude it is still almost 90% of surface strength — they are simply doing their physics in a frame that falls with them. Weightlessness is a coordinate choice. That so profound a fact follows from the humblest non-inertial frame is why this topic, and not the flashier rotating one, is the historical gateway to general relativity.

The mathematics of the accelerating frame

Only one derivative separates this frame from an inertial one. Let the frame's origin sit at $\mathbf{R}(t)$ in an inertial frame $S$ , with axes always parallel. A particle's positions in the two descriptions are related by pure vector addition, and because no basis vector ever turns, differentiating is completely safe — no operator identities, no cross products. Twice differentiating gives the only kinematic fact this article needs:

\mathbf{a}_{\mathrm{in}} = \mathbf{a}_{\mathrm{fr}} + \mathbf{A} \qquad (1)

where: $\mathbf{a}_{\mathrm{in}}$ — the particle's acceleration in the inertial frame; $\mathbf{a}_{\mathrm{fr}}$ — its acceleration measured in the accelerating frame; $\mathbf{A} = \ddot{\mathbf{R}}$ — the frame's own acceleration. Accelerations simply add, the moving frame contributing its own on top of whatever the particle does relative to it.

Derivation ANewton's law in a linearly accelerating frame

We are deriving the frame's complete equation of motion — the master result (2) with its single fictitious force — from nothing but the position bookkeeping and Newton's law in the inertial frame. The strategy is the standard two-differentiations route, but stripped of every rotational complication: with parallel axes the components of a vector are the same in both frames, so ordinary calculus applies with no operator identity. The point to watch is what does not appear — no term reads the particle's position or velocity, which is precisely why the resulting force is uniform and why this frame can impersonate uniform gravity.

Step 1. Write the position bookkeeping and differentiate twice:

\mathbf{r}_{\mathrm{in}} = \mathbf{R}(t) + \mathbf{r}_{\mathrm{fr}} \quad\Longrightarrow\quad \mathbf{a}_{\mathrm{in}} = \mathbf{A} + \mathbf{a}_{\mathrm{fr}}

Why the differentiation is legal term by term: the frame's axes stay parallel to the inertial ones, so $\mathbf{r}_{\mathrm{fr}}$ has identical components in both frames and its inertial second derivative is just $\mathbf{a}_{\mathrm{fr}}$ — the step that would fail, and spawn the whole cross-product catalog, if the axes rotated.

Step 2. Apply Newton's second law where it is valid — in the inertial frame: $m\mathbf{a}_{\mathrm{in}} = \mathbf{F}_{\mathrm{real}}$ , with $\mathbf{F}_{\mathrm{real}}$ the sum of genuine forces (gravity, normal, tension, friction). Substitute Step 1 and solve for the frame-measured acceleration:

m\,\mathbf{a}_{\mathrm{fr}} = \mathbf{F}_{\mathrm{real}} - m\mathbf{A} \qquad (2)

where: $-m\mathbf{A}$ — the translational fictitious force, the entire non-inertial correction. Why it deserves the name force: it enters the ledger exactly where forces enter, yet no body exerts it — it is the origin's acceleration transposed to the force side, and it would vanish from the books in any inertial description.

Step 3. Inspect its structure. It is independent of $\mathbf{r}_{\mathrm{fr}}$ and $\mathbf{v}_{\mathrm{fr}}$ — the same vector at every point, for every state of motion — and strictly proportional to $m$ , so the acceleration it imparts, $-\mathbf{A}$ , is universal. Why this is the whole ballgame: uniformity plus mass-proportionality is the exact specification of a uniform gravitational field, setting up the effective-gravity repackaging of the next derivation and, eventually, the equivalence principle. *Verified: matches Morin ch. 10 and Taylor ch. 9; setting $\mathbf{A}=0$ recovers bare Newton, and setting $\boldsymbol{\omega}\neq0$ instead reproduces the four-term catalog of the fictitious forces article, of which (2) is the pure-translation limit.*

The repackaging that makes the frame pleasant rather than merely tractable is to notice that gravity and the fictitious force are the same kind of animal — both proportional to mass — and merge them. Split the real forces into gravity and everything else, $\mathbf{F}_{\mathrm{real}} = m\mathbf{g} + \mathbf{F}_{\mathrm{other}}$ , and equation (2) factors:

m\,\mathbf{a}_{\mathrm{fr}} = m(\mathbf{g} - \mathbf{A}) + \mathbf{F}_{\mathrm{other}} \quad\Longrightarrow\quad \mathbf{g}_{\mathrm{eff}} \equiv \mathbf{g} - \mathbf{A} \qquad (3)

where: $\mathbf{g}_{\mathrm{eff}}$ — the effective gravity: the only "gravitational" field any experiment inside the frame can detect. Every statics and dynamics result you know holds verbatim with $\mathbf{g} \to \mathbf{g}_{\mathrm{eff}}$ : plumb lines align with it, fluid surfaces set perpendicular to it, projectiles arc along it, pendulums swing about it. One substitution buys the whole frame. Two standard configurations cover nearly every problem. For horizontal acceleration $A$ (the train), $\mathbf{g}_{\mathrm{eff}}$ is the diagonal of a right triangle with legs $g$ down and $A$ backward, so its direction and size are

\tan\theta = \frac{A}{g}, \qquad g_{\mathrm{eff}} = \sqrt{g^{2} + A^{2}} \qquad (4)

where: $\theta$ — the tilt of the new vertical from the old. Ordinary vehicles ( $A \lesssim 0.2g$ ) tilt equilibrium by at most a dozen degrees while strengthening it by at most 2% — the tilt is conspicuous, the strengthening subtle, since the Pythagorean sum grows only quadratically in $A/g$ . For vertical acceleration (the elevator), vectors collinear with gravity just add or subtract. A passenger standing on a scale is in frame equilibrium under the normal force $N$ , gravity, and the fictitious force, giving the apparent weight

N = m\,(g + A) \qquad (5)

where: $A > 0$ for upward acceleration (heavier: the floor must both hold you and accelerate you), $A < 0$ for downward (lighter), and $A = -g$ — free fall — gives $N = 0$ : weightlessness, with gravity still present and fully employed. The scale never read gravity; it always read the contact force, and the contact force answers to $\mathbf{g}_{\mathrm{eff}}$ .

Which formula answers which question

you want	use	notes
accelerations between frames	eq. (1)	plain addition — no rotation, no cross products
dynamics inside the frame	eq. (2)	one fictitious force $-m\mathbf{A}$ , uniform
any statics in the frame	eq. (3)	replace $\mathbf{g}$ by $\mathbf{g}_{\mathrm{eff}}$ everywhere
tilt of plumb line / fluid surface, horizontal $A$	$\tan\theta = A/g$ , eq. (4)	mass never appears
strength of the tilted gravity	$g_{\mathrm{eff}} = \sqrt{g^2+A^2}$ , eq. (4)	quadratic in $A/g$ — grows slowly
scale reading in an elevator	$N = m(g+A)$ , eq. (5)	signs: up positive; $A=-g$ → weightless
pendulum clock in the frame	$T = 2\pi\sqrt{\ell/g_{\mathrm{eff}}}$ , eq. (6)	swings about the tilted equilibrium

Go deeper · Derivation B · 2 minThe tilted pendulum in an accelerating train

Derivation BThe tilted pendulum in an accelerating train

The goal is the complete behaviour of a pendulum hanging in a train that accelerates horizontally at constant $A$ : its equilibrium angle, the strength of the gravity it experiences, and its small-oscillation period — the classic exam trio. The strategy is to work entirely in the train frame, where the bob feels three forces (gravity, string tension, and the fictitious $-m\mathbf{A}$ ), find the equilibrium by force balance, and then recognize that oscillations about it are an ordinary pendulum problem in the tilted effective field. Watch the mass: it must cancel from every observable, or something has gone wrong.

Step 1. Force balance at equilibrium. In the train frame the hanging bob is at rest, so tension $\mathbf{T}$ , gravity $m\mathbf{g}$ , and the backward fictitious force of magnitude $mA$ sum to zero. Resolving along horizontal and vertical:

T\sin\theta = mA, \qquad T\cos\theta = mg

Why the geometry: the string must lean backward so its tension acquires a horizontal component to cancel the fictitious force — leaning toward the rear of an accelerating train, the direction every strap-hanger already knows.

Step 2. Divide the two equations — tension and mass both cancel:

\tan\theta = \frac{A}{g}

Why the cancellation matters: every pendulum in the carriage, whatever its mass or length, tilts by the same angle; so does the coffee, the plumb line, and a helium balloon in reverse. The frame has a single well-defined new vertical.

Step 3. Read off the effective gravity's strength from the same triangle: $T = m\sqrt{g^{2}+A^{2}}$ , so the bob hangs in an effective field $g_{\mathrm{eff}} = \sqrt{g^{2}+A^{2}}$ pointing along the tilted string — equation (4) rederived by force balance.

Step 4. Small oscillations. Displace the bob slightly from the tilted equilibrium: in the train frame it is an ordinary pendulum in the uniform field $g_{\mathrm{eff}}$ , so

T_{\mathrm{period}} = 2\pi\sqrt{\frac{\ell}{g_{\mathrm{eff}}}} = 2\pi\sqrt{\frac{\ell}{\sqrt{g^{2}+A^{2}}}} \qquad (6)

where: $\ell$ — the pendulum length. Why the period shortens: acceleration can only add in quadrature to gravity, never subtract (for horizontal $A$ ), so an accelerating train's pendulum clock always runs slightly fast — a $0.2g$ acceleration quickens it by about 1%. *Verified: at $A = g$ the tilt is $45°$ and $g_{\mathrm{eff}} = \sqrt{2}\,g$ , matching Morin's worked case; $A \to 0$ recovers $\tan\theta = 0$ and the ordinary period $2\pi\sqrt{\ell/g}$ .*

Go deeper · Derivation C · 2 minApparent weight in an elevator and the free-fall limit

Derivation CApparent weight in an elevator and the free-fall limit

The goal is equation (5) — the elevator scale law $N = m(g+A)$ — followed to its most consequential special case, $A = -g$ , where the scale reads zero and Einstein's happiest thought begins. The strategy is force balance in the cabin frame on a passenger standing on a scale, keeping scrupulous track of signs. The step to watch is interpretive, not algebraic: what a scale measures is the contact force $N$ , never "gravity," and the entire drama of apparent weight lives in that distinction.

Step 1. Set up in the cabin frame, taking up as positive and the cabin accelerating upward at $A$ . The standing passenger is at rest in the frame, acted on by the normal force $N$ (up), gravity $mg$ (down), and the fictitious force $mA$ (down — opposite the frame's upward acceleration). Equilibrium in the frame demands

N - mg - mA = 0 \quad\Longrightarrow\quad N = m(g + A)

Why the fictitious force points down: the frame accelerates up, and $-m\mathbf{A}$ always opposes the frame's acceleration — the same rule that pushes you backward in a forward-accelerating car presses you floorward in a rising elevator.

Step 2. Run the sign through the service cases. Elevator accelerating upward at $A = 2\ \mathrm{m/s^2}$ : $N = m(9.81+2) = 1.20\,mg$ — a fifth heavier, the familiar start-of-ascent squeeze. Accelerating downward ( $A = -2$ ): $N = 0.80\,mg$ , the going-over-the-top lightness. Constant velocity, up or down: $N = mg$ exactly — steady motion is inertial, and no interior measurement can even detect it.

Step 3. Cut the cable: the cabin free-falls, $A = -g$ , and

N = m(g - g) = 0

Why everything floats, not just the passenger: the fictitious force $+mg$ upward cancels gravity $-mg$ for every object in the cabin simultaneously, because both scale with the same $m$ . The cabin frame has $\mathbf{g}_{\mathrm{eff}} = 0$ : releasing a phone leaves it hovering; a pendulum stops swinging; scales read zero for all occupants at once.

Step 4. Name what happened. Free fall did not remove gravity — it removed the frame's ability to detect it. A frame with $\mathbf{A} = \mathbf{g}$ is locally inertial: Newton's first law holds bare, and physics indoors is the physics of deep space. Why "locally" is load-bearing: real gravity varies from place to place, so a large enough falling frame sees residual tidal forces — the part no frame can cancel, and in general relativity the part that is real curvature. *Verified: numbers check as $N/mg = 1.20$ and $0.80$ for $A = \pm 2\ \mathrm{m/s^2}$ with $g = 9.81\ \mathrm{m/s^2}$ ; the $N = 0$ free-fall limit matches the orbital weightlessness of any spacecraft, where $A = \mathbf{g}$ continuously.*

Limiting cases

Case	Result	Meaning
$\mathbf{A} = 0$	fictitious force vanishes	the frame is inertial; bare $m\mathbf{a} = \mathbf{F}$ returns
constant $\mathbf{A}$	static, uniform $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$	equilibria tilt once and hold — trains, launched rockets at steady thrust
$\mathbf{A} = \mathbf{g}$ (free fall)	$\mathbf{g}_{\mathrm{eff}} = 0$	weightlessness with gravity present; the locally inertial frame of the equivalence principle
$A \ll g$ , horizontal	$\theta \approx A/g$ , $g_{\mathrm{eff}} \approx g$	small tilt, negligible strengthening — everyday vehicles
$A \gg g$	$\mathbf{g}_{\mathrm{eff}} \approx -\mathbf{A}$	launch and centrifuge-sled regime: the manufactured gravity swamps the real one
time-varying $\mathbf{A}(t)$	(2) holds instant by instant	$\mathbf{g}_{\mathrm{eff}}(t)$ wanders — the sloshing, jerky ride of a braking bus
$\boldsymbol{\omega} \neq 0$	beyond this article	rotation wakes centrifugal, Coriolis, and Euler terms — see the sibling articles

Common trapThe elevator accelerating upward increases gravity

It increases the scale reading — gravity itself never budges. Conflating the two costs points and understanding in equal measure. The gravitational force on the passenger is $m\mathbf{g}$ in every scenario, cable intact or cut; what changes with $\mathbf{A}$ is the contact force $N$ that the floor must supply, and $N$ is all a scale ever measures. Say "apparent weight" for $N = m(g+A)$ and reserve "gravity" for $m\mathbf{g}$ and every paradox dissolves — including "zero gravity" in orbit, which is actually near-full gravity read from a freely falling frame. Two adjacent traps: the fictitious force needs the frame's acceleration $\mathbf{A}$ , never the particle's own; and a linearly accelerating frame has no centrifugal or Coriolis forces whatsoever — those require rotation, and importing them here is the fastest way to fail this chapter's problems.

Worked examples

1 · Reading the commuter train (confidence builder)

A commuter train pulls out at a steady $A = 1.2\ \mathrm{m/s^2}$ . A pendant hangs from a passenger's phone cord, length $\ell = 0.50\ \mathrm{m}$ . Find the cord's equilibrium tilt, the effective gravity in the carriage, and the pendant's small-oscillation period during the acceleration.

Step 1. Tilt from equation (4): $\tan\theta = A/g = 1.2/9.81 = 0.122$ , so $\theta = \boxed{7.0°}$ toward the rear. Every hanging object in the carriage — cords, straps, the tea in the trolley — adopts the same $7.0°$ , because the mass cancels.

Step 2. Effective gravity: $g_{\mathrm{eff}} = \sqrt{g^{2} + A^{2}} = \sqrt{9.81^{2} + 1.2^{2}} = \sqrt{96.2 + 1.44} = \sqrt{97.7} = \boxed{9.88\ \mathrm{m/s^2}}$ — only 0.7% above $g$ : the direction change is conspicuous, the magnitude change nearly invisible, exactly the quadrature behaviour of (4).

Step 3. Period from equation (6): $T = 2\pi\sqrt{\ell/g_{\mathrm{eff}}} = 2\pi\sqrt{0.50/9.88} = 2\pi\sqrt{0.0506} = 2\pi\times0.225 = \boxed{1.41\ \mathrm{s}}$ , against $1.42\ \mathrm{s}$ when the train cruises — the pendant is a (terrible) accelerometer, running fast by half a percent.

Step 4. Sanity-check the direction in the inertial ledger: to accelerate the pendant forward at $1.2\ \mathrm{m/s^2}$ , the cord's tension needs a forward horizontal component, so the cord must lean back — same tilt, no fictitious force, ledgers agree.

2 · Launch, burnout, and the floating pen (exam level)

An $80\ \mathrm{kg}$ astronaut rides a booster accelerating vertically at $A = 3g$ . Find the seat force on her. At burnout (altitude $200\ \mathrm{km}$ , engines cut, drag negligible) she releases a pen. Find the seat force and the pen's motion in the cabin frame immediately after cutoff, given that gravity at that altitude is still $g(h) = g\,[R_\oplus/(R_\oplus + h)]^{2}$ with $R_\oplus = 6371\ \mathrm{km}$ .

Step 1. During the burn, equation (5) with upward $A = 3g$ : $N = m(g + 3g) = 4mg = 4\times80\times9.81 = \boxed{3.14\ \mathrm{kN}}$ — she "weighs" 320 kg. Crews call it pulling 4g; the seat, not gravity, supplies every newton of the excess.

Step 2. Gravity at burnout altitude: $g(200\ \mathrm{km}) = 9.81\times(6371/6571)^{2} = 9.81\times(0.9696)^{2} = 9.81\times0.940 = \boxed{9.22\ \mathrm{m/s^2}}$ — still 94% of surface gravity. Whatever happens next, gravity has not gone anywhere.

Step 3. Engines cut: the only force on the vehicle is gravity, so the cabin's acceleration is $\mathbf{A} = \mathbf{g}(h)$ — free fall (a ballistic arc counts, whatever the velocity). In the cabin frame the fictitious force $-m\mathbf{A} = -m\mathbf{g}(h)$ cancels gravity exactly for every object: $\mathbf{g}_{\mathrm{eff}} = 0$ . The seat force collapses from $3.14\ \mathrm{kN}$ to $\boxed{0}$ in the time the thrust tails off — the notorious instant "eyeballs-in" becomes floating.

Step 4. The released pen. In the cabin frame, with $\mathbf{g}_{\mathrm{eff}} = 0$ and no contact forces, the pen's equation of motion (2) reads $m\mathbf{a}_{\mathrm{fr}} = 0$ : it hovers, or drifts at whatever tiny velocity her hand gave it. In the inertial ledger, pen and cabin trace the same ballistic arc under the same $9.22\ \mathrm{m/s^2}$ , so their separation never changes — two descriptions, one floating pen.

Step 5. Read the moral like an examiner. "Weightless" here never meant "no gravity" — it meant $N = 0$ in a frame with $\mathbf{A} = \mathbf{g}$ . An orbiting station merely makes Step 3 permanent: it is a free-fall frame that keeps missing the ground. The answer weightlessness is a frame statement, not a gravity statement is the single most bankable sentence in this chapter.

Origins · 1 minAlbert Einstein and how this was discovered

Albert Einstein

1879–1955 · Bern patent office, Berlin, Princeton

Einstein spent 1907 writing a review of relativity while examining patents in Bern, and it was at his desk there that the falling-man thought arrived — "the happiest thought of my life," he called it decades later. The linearly accelerating frame was his laboratory of choice: no rotation, one fictitious force, and the clean statement that indoors it is indistinguishable from uniform gravity. He rode that equivalence hard. In 1907 it gave him gravitational time dilation; by 1911 it predicted that starlight grazing the Sun must bend (an accelerating elevator sees a light beam curve, so gravity must bend light too); by 1915 the full theory — general relativity — had recast gravity as spacetime geometry, with freely falling frames as the true heirs of Newton's inertial ones. The 1919 eclipse expeditions confirmed the doubled light-bending of the completed theory and made him the most famous scientist alive. The elevator never left the physics curriculum: it remains the shortest honest road from Newton to Einstein.

History

1632

— Galileo's ship argument establishes that uniform motion is undetectable from inside — the ancestor of every frame-equivalence claim.

1687

— Newton's Principia defines the inertial frames in which the laws of motion hold, leaving accelerated frames as the marked case.

1907

— Einstein's happiest thought: a freely falling observer feels no weight; uniform acceleration and uniform gravity are locally interchangeable.

1911

— From the accelerating-elevator argument Einstein predicts gravitational light bending and time dilation, testable consequences of the equivalence principle.

1915

— General relativity completes the program: gravity becomes geometry, and freely falling frames become the local inertial frames of physics.

Going further

Texts. Morin, Introduction to Classical Mechanics, ch. 10 (Accelerating frames of reference) · Goldstein, Classical Mechanics, 3rd ed., ch. 4 · Landau & Lifshitz, Mechanics, §39 · Taylor, Classical Mechanics, ch. 9 · Einstein, Relativity: The Special and the General Theory (the elevator chapters).

Provenance. Derivations follow Morin (Introduction to Classical Mechanics, Ch. 10), Taylor (Classical Mechanics, Ch. 9), Goldstein (Ch. 4), and Landau-Lifshitz (Mechanics §39). A verifier should recompute: (1) the acceleration transformation $\mathbf{a}_{\text{in}}=\mathbf{a}_{\text{fr}}+\mathbf{A}$ by twice differentiating $\mathbf{r}_{\text{in}}=\mathbf{R}+\mathbf{r}_{\text{fr}}$ with $\boldsymbol\omega=0$ ; (2) the pendulum results $\tan\theta=A/g$ and $T=2\pi\sqrt{\ell/\sqrt{g^2+A^2}}$ ; (3) the $A=g$ free-fall limit giving $g_{\text{eff}}=0$ ; and (4) the numeric check $g\tan30^\circ\approx 5.7\ \mathrm{m\,s^{-2}}$ using $g=9.81\ \mathrm{m\,s^{-2}}$ . Worked-example numbers on this page ( $\theta = 7.0°$ , $g_{\mathrm{eff}} = 9.88\ \mathrm{m/s^2}$ , $T = 1.41\ \mathrm{s}$ ; $N = 3.14\ \mathrm{kN}$ and $g(200\ \mathrm{km}) = 9.22\ \mathrm{m/s^2}$ ) are illustrative arithmetic recomputed from the boxed formulas.

Builds on: Non-inertial reference frames · Fictitious (inertial) forces. Leads to: Equivalence of uniform acceleration and gravity · Centrifugal force · Coriolis force · Time derivatives in rotating frames.

Why does the pendulum lean backward when the train accelerates forward — should it not be thrown forward?

Two ledgers, one answer. In the train frame, the fictitious force $-m\mathbf{A}$ points backward (opposite the frame's acceleration), so the bob settles rearward of the suspension point until the tension's horizontal component balances it: $\tan\theta = A/g$ . In the inertial frame there is no fictitious force at all — instead the bob must accelerate forward with the train, which requires a net forward force, which the string can only provide by leaning back. The "thrown forward" intuition belongs to braking ( $\mathbf{A}$ reversed), where both ledgers duly predict a forward lean. If you remember one rule: hanging objects lean opposite to $\mathbf{A}$ , fluid surfaces rise on the side $\mathbf{A}$ points away from.

Which way does a helium balloon lean in an accelerating car?

Forward — with the acceleration, opposite to every pendulum, and it is the best party trick in this chapter. The balloon does not respond to the fictitious force alone; it responds to buoyancy in air that itself feels $\mathbf{g}_{\mathrm{eff}}$ . The cabin air, denser than helium, piles up toward the rear (downhill along effective gravity), raising the pressure there; the balloon, like any buoyant object, moves against the effective-gravity gradient — "up" in the tilted field, which has a forward component. It is the same reason the balloon floats toward the ceiling in an unaccelerated car rather than falling. Watch one at a green light: the passengers lurch back, the balloon bows forward, and both are obeying $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$ exactly.

Is weightlessness in orbit really the same physics as a falling elevator?

Identical, with better plumbing. An orbiting station is in continuous free fall: gravity (about 90% of surface strength at 400 km) is the only force acting, so the station's $\mathbf{A} = \mathbf{g}$ at every instant, and in the station frame the fictitious force cancels gravity for every object aboard — $\mathbf{g}_{\mathrm{eff}} = 0$ , scales read zero, pens hover. The only difference from the severed elevator is horizontal speed: the station falls toward Earth and perpetually misses, because its 7.7 km/s sideways velocity carries it around the curve of the planet as fast as it drops. "Microgravity" rather than "zero gravity" is the honest word: tidal residuals of order $10^{-6}\,g$ survive, precisely the nonuniform part no frame can cancel.

If acceleration and gravity are indistinguishable, what eventually gives the game away?

Nonuniformity — the fine print in the word locally. A real gravitational field points toward a center and weakens with distance, so two test masses released far apart in a huge falling frame drift together (horizontally) and apart (vertically): tidal forces. No single acceleration $\mathbf{A}$ can cancel a field that varies from point to point; the residue is frame-proof. That is exactly the boundary where general relativity plants its flag: the cancellable part of gravity is coordinate bookkeeping, the uncancellable tidal part is spacetime curvature — the real, agent-independent content of the field. Practically, the elevator equivalence is superb for cabins and laboratories and fails for oceans: the tides in the harbor are the Moon's gravity gradient, the part of gravity no elevator could ever hide.

Does the equivalence trick work for light too, or only for mechanics?

Einstein's 1911 bet was that it works for everything — that is what promotes a mechanical curiosity to a principle. Fire a light beam across an upward-accelerating elevator: during the crossing time the cabin gains speed, so in cabin coordinates the beam's exit point is lower than its entry — the light path curves downward. If acceleration and gravity are indistinguishable, gravity must bend light by the same geometry. The 1919 eclipse measurements found starlight deflected by the Sun (the full general-relativistic value, twice the naive elevator estimate, the doubling coming from spatial curvature the simple argument cannot see). The same elevator argument yields gravitational redshift and time dilation, now verified in the lab to a few parts in $10^{19}$ with optical clocks a millimetre apart in height.

What does a phone accelerometer actually measure — acceleration or gravity?

Neither alone: it measures their difference, $\mathbf{g}_{\mathrm{eff}} = \mathbf{g} - \mathbf{A}$ , called specific force — precisely this article's effective gravity. A MEMS accelerometer is a microscopic mass on silicon springs; it reads the contact force per unit mass needed to hold that mass still in the phone's frame. Phone at rest: springs support against gravity, reading $g$ upward — which is why a "resting" accelerometer does not read zero. Phone in free fall: mass and case fall together, springs relax, reading zero — the drop-detection signature laptops have used to park hard drives. Phone accelerating in deep space at $g$ : reading indistinguishable from resting on Earth, the equivalence principle as a datasheet fact. Inertial navigation works by integrating specific force plus a gravity model — you cannot separate the two by measurement alone.

The train's acceleration changes from moment to moment — does the effective-gravity picture survive?

Yes, instant by instant: nothing in the derivation of $m\mathbf{a}_{\mathrm{fr}} = \mathbf{F}_{\mathrm{real}} - m\mathbf{A}$ required $\mathbf{A}$ constant, so $\mathbf{g}_{\mathrm{eff}}(t) = \mathbf{g} - \mathbf{A}(t)$ simply becomes time-dependent. What is lost is equilibrium: a plumb line no longer settles at a fixed tilt but chases the wandering vertical, overshooting and oscillating — the lurching of a bus in traffic is your vestibular system experiencing a rapidly varying $\mathbf{g}_{\mathrm{eff}}$ . The statics-with-tilted-gravity shortcuts of this article are exact for constant $\mathbf{A}$ and good approximations when $\mathbf{A}$ varies slowly compared to the system's response time (a short pendulum tracks a gentle brake; coffee does not track a pothole). For violently varying $\mathbf{A}$ , integrate (2) directly.

Why does this topic matter for learning rotating frames next?

Because it isolates one conceptual move — transposing frame acceleration into a force — with every geometric complication stripped away. Here the fictitious force is a single constant vector: no position dependence, no velocity dependence, no cross products, and you can see with complete clarity why it scales with mass, why it has no reaction partner, and how it merges with gravity into $\mathbf{g}_{\mathrm{eff}}$ . Rotating frames add nothing conceptually new — they just make the frame's acceleration depend on where you are (centrifugal, Euler) and how you move (Coriolis), which dresses the same transposition in cross products. Master the elevator and the carousel becomes bookkeeping; the taxonomy article and the Coriolis deep-dive are exactly that program.