Accelerating Reference Frames · Guide · glidepathscience.org
GuideAccelerating Reference Frames
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round, in the order a textbook would take them. Part 1 prepares the Intro track and takes about fifteen minutes to read; Part 2 prepares the Advanced track and takes about twenty. Each section ends with a worked example that is not one of the sim's own problems.
Part 1 · The essentials
What a scale reads
Ride an elevator and you feel heavier as it starts up and lighter as it slows near the top, although the Earth pulls on you exactly as before. What changes is how hard the floor pushes, and every instrument a rider carries, from a bathroom scale to the inner ear, measures a push like that, never gravity itself.
A bathroom scale is a spring in a box. It compresses until it pushes back up on your feet just hard enough, and the dial reports that push: the normal force N, the scale's force on you. Two forces act on you as you stand there, your weight W = mg, the Earth's pull, and N. At rest they balance:
$N - mg = 0$ so $N = mg$
Here m is your mass in kilograms and g = 9.81 m/s² is the acceleration of free fall near the Earth's surface, so N is in newtons. What a scale reads in any situation is your apparent weight; it equals your true weight, mg, only when nothing is accelerating. A scale neither weighs gravity nor measures mass: a dial marked in kilograms just divides its push by 9.81 m/s², which is right only at rest on the Earth.
Now set the elevator moving.
Heavier and lighter: N = m(g + a)
In an elevator the rider moves with the floor, and the scale shows the force that makes it happen. Only two forces act on the rider, W down and N up. Take up as positive and let a be the elevator's vertical acceleration, which is also the rider's. Then ΣF = ma reads
$N - mg = ma$ $N = m(g + a)$
The velocity does not appear: the reading depends on the acceleration alone.
| The elevator is… | a points | The scale reads |
|---|---|---|
| speeding up going up, or slowing going down | up | more than mg |
| speeding up going down, or slowing going up | down | less than mg |
| moving at a steady speed | (none) | exactly mg |
In units of your weight at rest, the "g-force" of everyday speech, the reading is
$N/mg = 1 + a/g$
The usual mistake is to go by the direction of travel: "it is going up, so I am heavier." Near the top of a trip up, the elevator is still rising but slowing, so a points down and the scale reads less than mg. Ask which way the velocity is changing. With up positive, that a is negative, and the equation handles the sign.
What happens if the elevator's acceleration reaches −g?
Free fall: weightless, not gravity-free
Riders on a drop tower float above their seats, and astronauts float in orbit. Most people say that gravity has gone. It has not, and the scale equation shows why. If the floor falls freely, its acceleration is −g:
$a = -g$ $N = m(g + a) = 0$
The rider and the scale fall together, each speeded up by gravity alone, so the scale never has to push. The rider is weightless, which means unsupported, while the Earth pulls at full strength; all of its pull goes into speeding the rider up. In the sim's drop tower a ball let go beside the rider stays beside the rider: everything inside falls at the same rate.
The wrong explanations, that there is no gravity in free fall or that the speed cancels it, fail at once: gravity is what makes the gondola fall, and the scale reads zero from the instant of release, at every speed. Braking at 3g on the way down means a = +3g, so the scale reads m(g + 3g) = 4mg.
So far every description has been from the ground. The rider, at rest in the cabin, sees things differently.
The view from inside: a fictitious force
Inside the vehicle, Newton's laws need one more force. In the elevator's frame the rider never moves, so the forces should add to zero, yet N and W do not balance while it speeds up. To keep ΣF = 0 for a body at rest in an accelerating frame, we add the fictitious force
$F_{fict} = -ma$
where a is the frame's acceleration, measured from the ground, and m is the body's mass. While the elevator speeds up going up, it points down. The sim draws it dashed. Nothing exerts it: it has no agent, so it has no third-law partner, and it comes from the choice of frame, not from an interaction. Both descriptions are right. From the ground, $N - mg = ma$; from inside, $N - mg - ma = 0$: the same equation, rearranged.
One mistake is to treat −ma as a real push, as if the floor pressed harder "because of" it; the floor's push is N, already in the diagram. The other is to leave −ma out of the inside view and conclude that N = mg, which the scale contradicts.
Now turn the elevator on its side.
Sideways: the plumb bob tilts
Hang a plumb bob of mass m in a car that speeds up at a steady a. From the ground, the line's pull T does two jobs: its vertical part holds the bob up, and its horizontal part speeds the bob up with the car. With θ the line's angle from vertical,
$T\cos\theta = mg$ $T\sin\theta = ma$ $\tan\theta = a/g$
From inside, gravity and the fictitious −ma add to an effective gravity g − a, and the bob hangs along it. Either way the line tilts by θ: toward the back while the car speeds up, toward the front while it brakes. The bob points opposite the acceleration, whatever the velocity. Turned around, $a = g\tan\theta$, the bob is an accelerometer: its angle gives the size of the acceleration, and the side it swings to gives the direction, opposite the swing.
The trap is braking. The car still moves forward, so many people expect the bob to swing back, but the acceleration points backward and the bob swings forward, the way a passenger who is not holding on lurches. Smaller slips are to use the cosine, or the sine, in place of the tangent (the sine comes close only for small tilts), and to measure θ from the horizontal.
A bathroom scale on the car's level floor still reads mg: it feels only the floor's push straight up, while friction on the shoes pushes forward. A turn at a steady speed is an acceleration too, toward the inside, so in a bus turning left the bob hangs to the right. An airliner in a coordinated turn banks so that effective gravity points straight into the cabin floor: a bob hangs square to it, a scale reads 1.15 g in a 30° bank, and with the shades down the turn feels like level flight.
Not everything in the car hangs back like the bob.
Balloons and coffee: everything lines up with effective gravity
In a car that speeds up, a helium balloon tied to the floor leans forward and coffee rises at the back of its cup. Both obey the bob's rule. In a vehicle with a steady acceleration, effective gravity g − a plays the part gravity plays at rest: "down" is along it, and "level" is perpendicular to it. A liquid's surface lies level, so where effective gravity tilts back, the surface tilts too, rising at the back of the cup with the bob's slope, tan θ = a/g.
A helium balloon floats "up," which here means opposite effective gravity. The air in a closed car is pushed forward along with the car, so its pressure grows toward the back, as it grows toward the ground. Buoyancy pushes from high pressure to low, up and forward, so the balloon, lighter than the air it displaces, goes the way the car accelerates.
The mistakes are to think liquids are always level and that everything swings back. Level means perpendicular to effective gravity, and what moves the balloon is the air around it, not its own tiny inertia; with the windows open, the air is not carried along with the car, and the lean is lost.
A crate on the floor has nothing to hang from. Friction has to do the job.
A crate on the floor: friction supplies ma
Static friction keeps cargo riding along, up to a limit. A crate at rest on the floor of a vehicle that speeds up at a must speed up too, and on a level floor only friction can push it horizontally. Static friction supplies whatever is needed, up to $\mu_s N$, where $\mu_s$ is the coefficient of static friction and N = mg:
$f_s = ma \le \mu_s mg$ $a_{\mathrm{max}} = \mu_s g$
The mass cancels: a heavy crate needs more friction and gets more, so heavy and light crates slip at the same acceleration. Past $a_{\mathrm{max}}$ the crate slides, and kinetic friction $\mu_k mg$ takes over, with the smaller coefficient $\mu_k$. From the ground, the sliding crate still speeds up forward, at $\mu_k g$, only less quickly than the floor; relative to the floor it moves backward until the back wall stops it.
Three mistakes are common: using $\mu_k$ for the slip limit (it applies only once the crate slides), multiplying by the mass, which gives a force rather than an acceleration, and saying the crate "moves backward." Relative to the road it never does.
Part 2 asks where the fictitious force comes from, and follows it into rotating frames.
Part 2 · Going further
Where fictitious forces come from
The fictitious force comes from measuring positions from a moving origin, and it is exact for any frame that translates without turning. Let the vehicle's origin be at R(t), measured from the ground, and a body at r′ in the vehicle's coordinates, whose axes stay parallel to the ground's. Then the body's position from the ground is r = R + r′. Differentiating twice, Newton's second law, ma = ΣF in the ground's frame, becomes
$r = R + r'$ $a = A + a'$ $ma' = \Sigma F - mA$
where A is the vehicle's acceleration and a′ the body's acceleration as seen inside. In the vehicle's coordinates, the second law gains one term, −mA: the fictitious force. It is proportional to the body's mass, exactly as gravity is, so it gives every body the same extra acceleration, −A, whatever it is made of. That shared property is the root of effective gravity, and of the equivalence principle.
The term is no approximation, and it is not the engine's or the air's push: what rides inside is pushed only by what touches it.
The next step is to combine −mA with gravity.
Effective gravity is a vector
In a vehicle with a steady acceleration, everything inside behaves as if gravity had changed both its size and its direction. Per unit mass, gravity g and the fictitious −A add to effective gravity:
$g_{eff} = g - A$ $|g_{eff}| = \sqrt{g^2 + A^2}$ $\tan\theta = A/g$
The second and third forms hold when A is horizontal, and θ is the tilt of $g_{eff}$ from vertical. The total push on a rider, the floor's and the friction's together, grows by the factor $|g_{eff}|/g$ and slants with "down," though a bathroom scale shows only its upward part, mg. A ball let go from rest relative to the vehicle falls in a straight line along $g_{eff}$. It takes $t = \sqrt{2h/g}$ to fall a height h, and in that time it drifts back by ½At², or
$\text{drift} = hA/g$
The common mistakes are to add g and A as numbers, though they are perpendicular, and to expect the drift to depend on the van's speed: cruising, the ball lands right below where it was let go.
Effective gravity reaches even the air, which is why the balloon leans the other way.
Buoyancy follows effective gravity
The balloon's lean shows that effective gravity acts on fluids exactly as gravity does. Air at rest in the car accelerates with it, and only a difference in pressure, higher behind each parcel of air than in front, can push it forward. In a fluid at rest in effective gravity, the pressure grows along $g_{eff}$:
$\nabla p = \rho\,g_{eff}$
where ρ is the fluid's density and ∇p points the way the pressure grows fastest: toward the floor, as always, and toward the back of a car that speeds up. Buoyancy is the net push of that pressure on a body of volume V:
$B = -V\nabla p = -\rho V g_{eff}$
It points opposite effective gravity, up and forward. A balloon whose average density is less than the air's gets more buoyancy than its own effective weight, and the difference, $(\rho V - m)|g_{eff}|$, pulls its string tight along $-g_{eff}$, opposite the bob. The balloon's own inertia would make it lag back, and its string only holds it down; the effect needs the windows closed, so that the car's air is carried along.
The scale, the bob and the balloon all respond to effective gravity. So does every accelerometer.
What an accelerometer reads
No accelerometer, in a phone, a rocket or the inner ear, can measure the acceleration alone, or gravity alone. An accelerometer is a small test mass held by a spring or by electric forces, and it reports the push needed to hold the mass, per unit mass. That push plus gravity gives the mass its acceleration, so the reading is the specific force,
$a - g$
the acceleration minus gravity, as vectors. On a table a = 0, and the reading is −g: 9.81 m/s² upward, the table's push. In free fall a = g, and the reading is zero. A bathroom scale is a one-axis accelerometer: its reading divided by your mass is the vertical part of a − g. The "g" that pilots and astronauts quote is the size of a − g in units of g, which on a vertical launch, with a upward, is $(a + g)/g$.
The tempting mistake is to read the g a crew feels as the rocket's acceleration. On a vertical launch the crew feels a + g, so the rocket speeds up at one g less than they feel.
If a scale reads a − g, a log of the scale is a log of the acceleration, and integrating it tells where the elevator went.
From a scale log to a height
The Ride Log tab integrates measured acceleration twice to find position, as inertial navigation does. Each reading N of a rider of mass m gives a = N/m − g. The velocity is the running total of a over time, and the height the running total of the velocity. With readings Δt apart, the trapezoid rule adds the average of each pair:
$v_{i+1} = v_i + \tfrac{1}{2}(a_i + a_{i+1})\,\Delta t$ $y_{i+1} = y_i + \tfrac{1}{2}(v_i + v_{i+1})\,\Delta t$
For a ride in three steady stages, speeding up for $t_1$, cruising for $t_2$ and slowing for $t_1$, the top speed is $v = a t_1$ and the climb, the area under the velocity, is $h = \tfrac{1}{2}v t_1 + v t_2 + \tfrac{1}{2}v t_1 = v(t_1 + t_2)$. The weakness is that errors are integrated too. A constant error δa, from a scale that reads 1 % high for instance, grows into a velocity error δa·t and a height error
$\delta h = \tfrac{1}{2}\,\delta a\,t^2$
That is why the Ride Log first zeroes the scale: it takes the mean reading at rest, $N_0$, which holds both mg and the calibration error, and uses $a = (N - N_0)/m$. Random noise adds up more slowly, into the height's standard uncertainty. The usual slips are to skip the zeroing and to forget the ½.
Back to the plumb bob, and an acceleration that starts suddenly.
The swing about the new vertical
After a sudden start a bob swings past its new angle, and the overshoot shows that, inside the vehicle, it simply is a pendulum in a new gravity. Before the start, the bob hangs straight down, at rest. The moment a steady horizontal A begins, effective gravity tilts by $\theta_{eq}$, and the bob is at rest $\theta_{eq}$ away from its new equilibrium. A pendulum released from rest swings symmetrically about its equilibrium, so with no damping the bob swings an equal angle past it, to $2\theta_{eq}$ from vertical, then back:
$\theta_{eq} = \arctan(A/g)$ $\theta_{\mathrm{max}} = 2\theta_{eq}$
The result is exact, even for wide swings, because it uses only the symmetry of a pendulum in uniform gravity, here $g_{eff}$. The swing's period is a pendulum's in $|g_{eff}|$: about $2\pi\sqrt{L/|g_{eff}|}$ for small swings on a line of length L, longer for wide ones. (Periods are the subject of Simple Harmonic Motion & Pendulums.) A real bob has damping, so its swings shrink until it settles at $\theta_{eq}$. The mistake is to stop the swing at $\theta_{eq}$: the bob passes it moving at its fastest.
The drop tower of Part 1 is a frame in which effective gravity vanishes altogether. So is an orbit.
Orbit, free fall and the equivalence principle
Weightless astronauts and Einstein's equivalence principle are one observation, seen from two sides. Gravity weakens with distance from the Earth's center as 1/r², so at a height h above the surface
$g(h) = g_0\,(R/(R + h))^2$
with $g_0$ = 9.81 m/s² and R = 6371 km, the Earth's mean radius. At the Space Station's 400 km, gravity keeps most of its surface strength. The station and crew float because they fall around the Earth together, a freely falling frame where $g_{eff} = g - A = 0$. In such a frame, a body with no other force on it keeps its velocity: the frame is locally inertial. The usual slips are to use h in place of R + h, and to forget the square.
Turned around, a sealed room accelerating at 1 g through empty space reproduces the Earth: the floor pushes with mg, a bob hangs straight, a dropped ball meets the floor in $\sqrt{2h/g}$. Both work because gravity's pull is proportional to a body's inertial mass, the m in F = ma, as −mA is. That inertial and gravitational mass are equal has been tested to about one part in $10^{15}$, by the MICROSCOPE satellite. The equivalence principle makes it a law: in a small enough room, no experiment tells uniform gravity from acceleration. It predicted that gravity bends light, since a beam crossing an accelerating room strikes the far wall a little low.
"In a small enough room" is the catch.
Tides: where equivalence fails
Over larger distances, gravity's changes in size and direction, the tides, cannot be removed by any choice of frame. In a freely falling room near a planet, gravity at the top is slightly weaker than at the bottom. Across a height d the difference is
$\Delta g = g_0\left[1 - \left(\frac{R}{R + d}\right)^2\right]$ $\Delta g \approx 2g d/R$
the second form holding because d is tiny beside R. Two balls released at rest, one above the other, drift apart at Δg, by ½Δg t² after a time t. Sideways, the pulls toward the planet's center converge, so two balls side by side drift together at about half that rate, g d/R. Stretched one way and squeezed the other is the pattern of the ocean tides that the Moon and the Sun raise on the Earth. The mistakes are to leave out the ½, and to think a falling room matches an inertial one at any size.
The last three sections turn the room into a spinning disk.
Rotating frames: the centrifugal force
In a frame turning at a steady rate ω, in radians per second, Newton's second law gains two fictitious terms (the spin is steady, so there is no Euler force):
$ma' = \Sigma F + m\omega^2 r_\perp - 2m\,\omega\times v'$
The first, the centrifugal force $m\omega^2 r$, points straight out from the axis, with r the distance from it, and acts on everything in the frame, moving or not. A coin at rest on a turntable shows it. From the ground the coin moves in a circle, which needs $m\omega^2 r$ toward the center, and only static friction can supply it. In the turntable's frame the coin is at rest, and the same friction balances the outward centrifugal force. Either way, it slips when $m\omega^2 r$ reaches $\mu_s mg$:
$\mu_s mg = m\omega^2 r$ $\omega = \sqrt{\mu_s g/r}$
The mass cancels, and coins nearer the rim slip first. The usual mistake is to treat the centrifugal force as real in the ground's frame. From the ground there is no outward force: once the coin slips, friction is too weak to keep turning its path, and it spirals out toward the rim. A second slip is the units: 1 rpm = 2π/60 rad/s.
The centrifugal force acts on everything in the frame; the Coriolis force acts only on what moves.
The Coriolis force
The second term in the rotating frame's law is the Coriolis force:
$-2m\omega\times v'$ $a_{Cor} = 2\omega v'$
Here v′ is the body's velocity in the rotating frame, and the vector ω points along the axis, up for a counterclockwise spin seen from above. The second form is its size per unit mass, for v′ in the plane of rotation. It is perpendicular to v′, so it turns the path without changing the speed: to the right of v′ for a counterclockwise spin. A puck launched outward from the center of a spinning air table goes straight from the ground, since nothing pushes it sideways, while the disk turns under it, so on the disk its path curves to the right. There the Coriolis force, always at right angles to v′, does the bending; the centrifugal force speeds the puck outward and, once the path has turned, bends it back a little. Integrating both reproduces the ground's straight line exactly, as the Rotating Frame tab checks.
On the Earth, Ω = 7.292 × 10⁻⁵ rad/s, and for horizontal motion only the vertical part of Ω counts, so the deflection is $2\Omega v\sin\lambda$ at latitude λ, zero at the equator. It is one part of the balance that steers winds, with pressure differences and friction. The mistakes are to curve the path the way the disk turns (the puck falls behind the disk), to drop the 2, and to think the curve is real from the ground.
Finally, the Earth's own spin.
The spinning Earth
A plumb line defines "vertical" for every builder, yet on the spinning Earth it does not point at the center, and gravity varies. At latitude λ a point on the surface is R cos λ from the axis, so the centrifugal term per unit mass is $\omega^2 R\cos\lambda$, pointing straight out from the axis. Its part along "up" reduces effective gravity, and its part along the ground, toward the equator, tilts the plumb line by an angle δ from the line to the center:
$\Delta g \approx \omega^2 R\,(\cos\lambda)^2$ $\tan\delta = \frac{\omega^2 R\cos\lambda\sin\lambda}{g_0 - \omega^2 R\,(\cos\lambda)^2}$
Here $g_0 = GM/R^2$ is gravity alone. With ω = 7.292 × 10⁻⁵ rad/s (one turn in 23.93 h) and R = 6378 km, the loss is a fraction of a percent of g, largest at the equator, where all of the centrifugal term points up. The tilt is zero at the equator and the poles and largest near 45°, a small fraction of a degree. The real Earth has been flattened by this same spin, which roughly doubles the tilt and adds about half again to the difference in g between the equator (9.780 m/s²) and the poles (9.832 m/s²). A common slip is to use cos λ once instead of twice for the "up" part. Rotation also gives itself away inside a closed room, as uniform acceleration cannot: the plane of a long pendulum's swing turns slowly against the floor (Foucault's pendulum), once a day at a pole and more slowly toward the equator.
Reference
Common mistakes
- Thinking a scale weighs gravity. It reads its push, $N = m(g + a)$.
- Going by the velocity. Scales, bobs and balloons follow the acceleration.
- Reading weightless as gravity-free. It means unsupported.
- Giving −ma an agent or a partner. It has neither.
- Expecting the balloon to swing back. It floats opposite effective gravity.
- Using $\mu_k$ for the slip limit. $a_{\mathrm{max}} = \mu_s g$, whatever the mass.
- Adding g and A as numbers. They are perpendicular.
- Reading the g a crew feels as the acceleration. They feel a + g.
- Stopping a sudden swing at $\theta_{eq}$. It reaches $2\theta_{eq}$.
- Curving a puck the way the disk turns. It curves the other way.
Equation sheet
| Part 1 | |
|---|---|
| scale, a up + | $N = m(g + a)$; in g, $N/mg = 1 + a/g$ |
| free fall | $a = -g$, so $N = 0$ |
| fictitious | $F_{fict} = -ma$ |
| bob, opposite a | $\tan\theta = a/g$; $a = g\tan\theta$ |
| crate holds while | $a \le \mu_s g$; $a_{\mathrm{max}} = \mu_s g$ |
| gravity | g = 9.81 m/s² at the Earth's surface |
| Part 2 | |
|---|---|
| effective gravity | $g_{eff} = g - A$; $|g_{eff}| = \sqrt{g^2 + A^2}$ |
| accelerometer | $a - g$ |
| ball, opposite A | $\text{drift} = hA/g$ |
| scale log | $h = v(t_1 + t_2)$; $\delta h = \tfrac{1}{2}\,\delta a\,t^2$ |
| sudden A, undamped | $\theta_{\mathrm{max}} = 2\theta_{eq}$ |
| altitude | $g(h) = g_0\,(R/(R + h))^2$, R = 6371 km |
| tides | $\Delta g \approx 2g d/R$ |
| rotating frame | centrifugal $m\omega^2 r$; Coriolis $-2m\omega\times v'$, $a_{Cor} = 2\omega v'$ |
| coin slips | $\omega = \sqrt{\mu_s g/r}$ |
| spinning Earth | $\Delta g \approx \omega^2 R\,(\cos\lambda)^2$ |
Symbols: m mass · g, $g_0$ gravity (9.81 m/s²) · a, A the vehicle's acceleration · N a scale's push · θ angle from vertical · $\mu_s$, $\mu_k$ friction coefficients · ρ, V density and volume · h height · R the Earth's radius (6371 km mean, 6378 km at the equator) · d a room's size · ω, Ω spin rate (rad/s) · r distance from the axis · v′ velocity in the rotating frame · λ latitude · δ a plumb line's tilt.
Further reading
OpenStax, University Physics Volume 1, chapter 6 (Applications of Newton's Laws), for apparent weight in an elevator and for noninertial frames and the Coriolis force; D. Kleppner and R. Kolenkow, An Introduction to Mechanics, 2nd ed., chapter 9, and J. R. Taylor, Classical Mechanics, chapter 9, for noninertial frames at the Advanced level. The Field Manual lists the sources of the sim's data.
Accelerating Reference Frames v1.6 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org