Inertial Frames of Reference

The operational definition is not a philosopher's nicety; it is an engineering specification. A ring-laser gyroscope in an airliner's navigation unit detects the aircraft's rotation with no outside reference at all, precisely because rotation — unlike velocity — is absolute and leaves a fingerprint inside a sealed box; every strapdown navigation system is a machine for measuring how far its own frame departs from inertial, integrating the departure twice to recover position. When astronomers need the best inertial frame available they build one: the International Celestial Reference Frame is fixed to hundreds of distant quasars, so remote that any residual acceleration is beneath detection — a quantitative descendant of Mach's "fixed stars". And physicists now manufacture free bodies to test the definition itself: LISA Pathfinder flew two gold-platinum cubes in free fall inside a spacecraft that steered itself around them, and they tracked inertial motion to femto-$g$ residuals.

Sit on a bench with an air-hockey puck and give it a shove. It runs dead straight at constant speed: your frame passes the test. Now do the same on a spinning carousel. The puck, once released, is untouched by any force — yet from your rotating seat you watch it curl away in a great arc, as though an invisible hand were shoving it sideways. Nothing is. In the ground frame it goes perfectly straight; it is you who are turning underneath it. The curve is the signature of a frame that has disqualified itself.

That is the whole diagnostic, stated as a procedure rather than a slogan. Shield a test body as well as you can, release it, and watch. Straight line at constant speed: the frame is inertial. Any swerve: the frame is not, and the swerve is a measurement of the frame's own acceleration. Ludwig Lange turned this into a definition in 1885 — throw a free particle from one point in three non-coplanar directions, and if all three paths are straight you have an inertial frame. The catch is that no real frame passes perfectly. A laboratory bench is bolted to a planet that spins, swings around the Sun, and is dragged around the Galaxy; each of those motions is an acceleration, so each leaves a residual a good-enough instrument will find. The honest question is never is my frame inertial? but inertial to what precision? — and that has a numerical answer.

The hierarchy tells the story. The dominant failure of a ground laboratory is the planet's spin, not its enormous orbital speed: what disqualifies a frame is acceleration, so the Earth's $29.8\ \mathrm{km/s}$ orbital velocity — undetectable in principle — contributes nothing, while its far smaller orbital acceleration contributes $0.0059\ \mathrm{m/s^2}$. Add everything up and a lab carries at most $0.040\ \mathrm{m/s^2}$, $0.41\%$ of $g$. That is why a bench experiment lasting seconds may treat the ground as inertial, while a Foucault pendulum swinging for hours cannot.

The definition comes first, and it is a statement about a frame, not a body: a frame is inertial when the first law is true in it. Integrating $\ddot{\mathbf r} = \mathbf 0$ twice gives the criterion in the form you check with a ruler and a stopwatch.

where: $\mathbf r_0, \mathbf v_0$ — the position and velocity at $t = 0$; nothing else appears. Physical reading: a free body traces a straight line at uniform rate, and that observation is the certificate. Lange's 1885 prescription makes it practical: launch free particles from one point in three non-coplanar directions and require all three tracks to be straight. Which other frames inherit the certificate? Slide the origin at constant velocity, turn the axes through a fixed angle, shift the clock's zero — the most general such transformation is the Galilean map.

where: $R$ — a fixed rotation matrix; $\mathbf V$ — a constant boost velocity; $\mathbf d_0, t_0$ — constant shifts of the space and time origins. Physical reading: differentiate twice and every constant dies, leaving $\mathbf a\,' = R\,\mathbf a$. Since $R$ is invertible, $\mathbf a = \mathbf 0 \iff \mathbf a\,' = \mathbf 0$: free bodies stay free and $S'$ is inertial too. The inertial frames form an infinite family with no member singled out as at rest — exactly the arena the relativity principle needs. Now break the condition: let the frame's origin accelerate at $\mathbf A$ instead of drifting at constant $\mathbf V$. Two derivatives no longer kill the extra term, and an observer who insists on Newtonian bookkeeping pays with a force that has no source.

where: $\mathbf A$ — the acceleration of the frame itself; $-m\mathbf A$ — the fictitious force, proportional to every mass and traceable to no second body (recall: Newton's First Law and Inertial Frames). Physical reading: put $\mathbf F_{\rm real} = \mathbf 0$ and a free body has $\mathbf a\,' = -\mathbf A \neq \mathbf 0$ — it swerves, untouched, and the size of the swerve hands you $\mathbf A$ exactly. This is the asymmetry: the boost velocity $\mathbf V$ has vanished from the physics; the frame acceleration $\mathbf A$ has not. A rotating frame matters most in practice, because every laboratory on Earth is one. The mechanics owner records the extra terms — centrifugal $-m\boldsymbol\Omega\times(\boldsymbol\Omega\times\mathbf r\,')$ and Coriolis $-2m\boldsymbol\Omega\times\mathbf v\,'$ — and for a body at rest at radius $R$ in a frame spinning at $\omega$ only the centrifugal one survives:

where: $\omega$ — the frame's angular velocity; $R$ — the distance from the rotation axis. Physical reading: this single number is the residual that disqualifies every Earth-fixed frame, and it is the reason "inertial" must always be quoted with a tolerance.

The last row eventually dismantles the others. A global inertial frame is a fiction in a universe with gravity: the best you can build is a freely falling box, valid over a region small enough that tidal differences stay below your resolution — and special relativity lives inside those boxes.

1 · How inertial is the ground beneath you?

A laboratory sits on the equator. Take $\omega = 7.292\times10^{-5}\ \mathrm{rad/s}$, $R = 6.378\times10^{6}\ \mathrm{m}$, $g = 9.81\ \mathrm{m/s^2}$. (a) Find the centrifugal residual. (b) Express it as a fraction of $g$. (c) Find the plumb-line tilt it produces at latitude $45^\circ$.

Step 1. The lab is carried around the rotation axis at radius $R$, so by equation (4) its fictitious acceleration is $a = \omega^2 R = (7.292\times10^{-5})^2 \times 6.378\times10^{6}$.

Step 2. Evaluate: $(7.292\times10^{-5})^2 = 5.317\times10^{-9}\ \mathrm{s^{-2}}$, so $a = \boxed{0.0339\ \mathrm{m/s^2}}$, directed outward from the axis.

Step 3. As a fraction of gravity: $0.0339 / 9.81 = 3.46\times10^{-3} = \boxed{0.35\%}$. The frame passes the first-law test only to about three parts in a thousand — fine for a bench experiment lasting seconds, hopeless for anything integrating over hours.

Step 4. Make the residual visible. At latitude $\phi$ the lab sits $R\cos\phi$ from the axis, so the centrifugal term is $\omega^2R\cos\phi$, whose component perpendicular to the local vertical is $\omega^2 R\cos\phi\sin\phi$ — and a plumb bob hangs off true by $\theta \approx \omega^2 R \sin\phi\cos\phi / g$.

Step 5. At $\phi = 45^\circ$: $\theta = 0.0339 \times 0.5 / 9.81 = 1.73\times10^{-3}\ \mathrm{rad} = \boxed{0.099^\circ}$, about $5.9$ arcminutes. Every plumb line at mid-latitude misses the centre of the Earth by a tenth of a degree — the non-inertiality of your frame, hanging on a string in front of you.

2 · The freely falling lift, and the size of the word "locally"

A lift's cable snaps. Inside, a passenger releases a ball. (a) What is the ball's acceleration in the lift frame? (b) How far does the lift fall in $3.0\ \mathrm{s}$? (c) Two balls are released $H = 2.5\ \mathrm{m}$ apart vertically — how far apart do they drift in that time?

Step 1. In the ground frame the ball is not free: gravity gives it $a = -g = -9.81\ \mathrm{m/s^2}$. But the lift frame accelerates too, with $A = -g$. Equation (3) then gives $a' = a - A = (-9.81) - (-9.81) = \boxed{0}$.

Step 2. So the released ball hangs motionless beside the passenger: it passes the first-law test. The falling lift is an inertial frame, even though gravity acts at full strength — the fictitious force $-mA$ has exactly cancelled the weight, and it cancels for every mass because both terms scale with $m$. (Why the cancellation is exact belongs to The Equivalence Principle; here we only measure how far it can be trusted.)

Step 3. Seen from the ground the whole box is plunging: in $t = 3.0\ \mathrm{s}$ it falls $\tfrac12 g t^2 = 0.5 \times 9.81 \times 9 = \boxed{44.1\ \mathrm{m}}$, with the ball floating beside the passenger the entire way.

Step 4. Now find the flaw. Gravity is not perfectly uniform — it weakens as $1/r^2$ — so a ball nearer the Earth falls slightly harder. Over a height $H$ the differential is $\Delta a = 2gH/R_\oplus = 2(9.81)(2.5)/6.378\times10^{6} = 7.69\times10^{-6}\ \mathrm{m/s^2}$.

Step 5. That residual drifts the two balls apart by $\tfrac12 \Delta a\, t^2 = 0.5 \times 7.69\times10^{-6} \times 9 = \boxed{35\ \mu\mathrm{m}}$ in $3.0\ \mathrm{s}$. That number is the size of the word "locally". The lift is inertial to about one part in $10^{6}$ of $g$ over its own height — good enough for any experiment coarser than $35\ \mu\mathrm{m}$, and not for one that is finer.

Newton knew the asymmetry this article turns on. Spin a bucket of water and the surface climbs the walls into a paraboloid — rotation announces itself from inside, with no reference to anything outside, while uniform velocity never does. Newton concluded that rotation must be absolute, measured against absolute space itself. Two centuries later Ernst Mach accepted the fact and refused the conclusion, asking the question that has never gone away: relative to what is the bucket rotating?

Texts. Rindler, Introduction to Special Relativity, 2nd ed., Ch. 1 · Taylor & Wheeler, Spacetime Physics, 2nd ed., Ch. 1 · Landau & Lifshitz, Mechanics, §3 · Mach, The Science of Mechanics (1883; English 1893), Ch. II.

Provenance. The definition and the closure argument follow Rindler Ch. 1 and Landau & Lifshitz §3; the operational criterion is Lange's (1885). Recall of the dynamical definition, the fictitious force $-m\mathbf A$, the centrifugal and Coriolis terms, and the $0.3\%$-of-$g$ failure of an Earth-fixed frame is by name to Newton's First Law and Inertial Frames (aq-mech-newtons-first-law-inertial-frames). A verifier should reproduce: $\mathbf a\,' = R\,\mathbf a$ under the general Galilean map; $\mathbf a\,' = \mathbf a - \mathbf A$ with $\partial\mathbf a\,'/\partial\mathbf A = -1$; $\omega^2 R = 0.0339\ \mathrm{m/s^2} = 0.35\%$ of $g$ and a $0.099^\circ$ plumb tilt at $45^\circ$ ($\omega = 7.292\times10^{-5}\ \mathrm{rad/s}$, $R = 6.378\times10^{6}\ \mathrm{m}$, $g = 9.81\ \mathrm{m/s^2}$); Earth's orbital $0.0059\ \mathrm{m/s^2}$ and the Sun's galactic $2\times10^{-10}\ \mathrm{m/s^2}$; and the free-fall lift ($a' = 0$, a $44.1\ \mathrm{m}$ drop in $3.0\ \mathrm{s}$, a $35\ \mu\mathrm{m}$ tidal drift over $H = 2.5\ \mathrm{m}$).

Builds on. Newton's Laws of Motion · Vectors and Coordinate Systems. Leads to. The Principle of Relativity · Galilean Relativity · Galilean Transformations · Absolute Space and Absolute Time · Einstein's Two Postulates.