The Lorentz Factor Gamma

A proton in the Large Hadron Collider carries $7\ \mathrm{TeV}$ of energy. Its Lorentz factor is $\gamma = 7460.5$ — and that single number says everything about it: its clock runs $7460$ times slow in the lab, the $26.7\ \mathrm{km}$ ring is $3.6\ \mathrm{m}$ long in its own frame, and it trails light by exactly $2.7\ \mathrm{m/s}$ . Meanwhile a passenger jet at $250\ \mathrm{m/s}$ has $\gamma - 1 = 3.5\times10^{-13}$ : thirteen zeros before anything happens at all. That gulf is the subject of this article. $\gamma = 1/\sqrt{1-\beta^2}$ is not a formula to memorise but a function to understand — its curve, its series, its limits, its derivative, its inverse. Where does relativity begin? Why is the everyday world so flat and the last one percent of $\beta$ so violent? Why does the Newtonian $\tfrac12 mv^2$ hide inside $(\gamma - 1)mc^2$ , and why does the innocent $d\gamma/d\beta = \gamma^3\beta$ mean that force and acceleration part company near $c$ ? The light-clock argument that first produces $\gamma$ belongs to The Light Clock and Time Dilation; the effects it scales — time dilation, length contraction — belong to Time Dilation and Length Contraction. What this article owns is $\gamma$ as a function: how fast is fast, with real velocities on a real plot, every number verified.

Key takeaways

The Lorentz factor is $\gamma = 1/\sqrt{1-\beta^2}$ with $\beta = v/c$ : dimensionless, always $\ge 1$ , and the single number that sizes every kinematic relativistic effect. It scales time dilation ( $\Delta t = \gamma\,\Delta\tau$ ) and length contraction ( $L = L_0/\gamma$ ), each derived in the article that owns it.
The curve is brutally non-linear. $\gamma(0.1) = 1.005$ ; $\gamma(0.6) = 1.25$ ; $\gamma(0.9) = 2.294$ ; $\gamma(0.99) = 7.089$ ; $\gamma(0.995) = 10.0125$ ; $\gamma(0.999) = 22.37$ . Everything interesting lives in the last one percent of $\beta$ .
The series $\gamma = 1 + \tfrac12\beta^2 + \tfrac38\beta^4 + \cdots$ is where Newton hides: $(\gamma-1)mc^2 \to \tfrac12 mv^2$ as $\beta \to 0$ . Truncated to the parabola $1 + \tfrac12\beta^2$ it is 1.0 % low by $\beta = 0.4$ ; keeping the $\tfrac38\beta^4$ term holds it to 0.14 % there, and to 1.7 % by $\beta = 0.6$ .
Structural facts you will use constantly: $\gamma\beta = \sqrt{\gamma^2-1}$ · $\beta(\gamma) = \sqrt{1 - 1/\gamma^2}$ · $d\gamma/d\beta = \gamma^3\beta$ .
Key idea: $\gamma$ is a dimensionless dial that reads 1 for everything you have ever touched and blows up without bound as $\beta \to 1$ . It is not "how much time slows down" for a single observer — it is the coefficient relating two coordinate systems, and both frames find the same $\gamma$ for the other.

The core insight

$\gamma = 1/\sqrt{1-\beta^2}$ is a pure number, never less than one, measuring how far a frame's motion has pushed it into relativity. It is flat almost everywhere — a jet, a satellite, a rifle bullet all sit at $\gamma - 1 < 10^{-9}$ — and then it detonates: from $\beta = 0.99$ to $\beta = 0.999$ , $\gamma$ triples. That asymmetry is why Newtonian physics survived two centuries, and why $c$ is unreachable for anything with mass.

Where this lives today

Every particle accelerator is a $\gamma$ -machine, and its designers think in $\gamma$ rather than $\beta$ precisely because $\beta$ stops being informative above $0.99$ . The LHC's $7\ \mathrm{TeV}$ protons sit at $\gamma = 7460.5$ ; the electrons of a $6\ \mathrm{GeV}$ synchrotron light source at $\gamma \approx 11\,700$ , and their radiation is beamed into a forward cone of width $\sim 1/\gamma$ — the entire reason such machines make useful X-ray beams. Detector physicists quote $\gamma c\tau$ , the mean flight distance before decay: at $\gamma = 10$ a muon covers $6.6\ \mathrm{km}$ instead of $659\ \mathrm{m}$ . At the other end, $\gamma - 1$ is the sort of correction that decides whether a technology works. GPS clocks lose $7.2\ \mu\mathrm{s/day}$ to the $\gamma - 1 = 8.35\times10^{-11}$ computed above, and are offset in frequency before launch to compensate. Optical lattice clocks now reach a stability of $10^{-18}$ , which resolves $\gamma - 1$ for a speed of about $0.4\ \mathrm{m/s}$ — walking pace. The flat left edge of the $\gamma$ curve, invisible for three centuries, is now routine metrology.

The physical picture

Think of $\gamma$ as a dial with a very peculiar scale. Turn it from a standstill up through a rifle bullet, an airliner, the Space Station, a GPS satellite — and the needle does not move. Not "barely moves": at $250\ \mathrm{m/s}$ the reading is $1.00000000000035$ . Every speed a human being has ever experienced lies where $\gamma$ is one to twelve decimal places, which is exactly why nobody noticed relativity until they went looking for it with light. Then the last stretch. At $\beta = 0.9$ the dial reads $2.29$ ; at $0.99$ , $7.09$ ; at $0.999$ , $22.37$ . Each factor-of-ten step toward $c$ roughly triples $\gamma$ , and no finite turn of the dial reaches the end of the scale. The blow-up is not a numerical accident: $1/\sqrt{1-\beta^2}$ has a genuine singularity at $\beta = 1$ , and that is why a massive particle cannot be pushed to the speed of light with any finite energy.

Interactive Simulation

The Lorentz Factor γ — drag β and watch γ = 1/√(1−β²) diverge near c, with the time dilation and length contraction it implies.

Figure

PUNCHLINE — γ against β, with real velocities on the plot. Everything human sits on the flat left edge: airliner 250 m/s (γ − 1 = 3.5×10⁻¹³), GPS satellite 3.874 km/s (γ − 1 = 8.35×10⁻¹¹), ISS 7.66 km/s (γ − 1 = 3.3×10⁻¹⁰). The Frisch–Smith muon at β = 0.995 reaches γ = 10.01. The LHC's 7 TeV proton has γ = 7460.5 — 620 times off the top of this plot.

Because the curve is so flat at the left, physics gets a free approximation there — the binomial series $\gamma \approx 1 + \tfrac12\beta^2$ — and that parabola is where Newton's kinetic energy is hiding. But an approximation is only useful if you know where it dies.

Figure

The low-β story. Exact γ (blue) against the two-term parabola 1 + β²/2 (grey) and the three-term series 1 + β²/2 + 3β⁴/8 (green). At β = 0.4 the parabola already reads 1.0800 against the exact 1.0911 — 1.0 % low; the β⁴ term repairs it to 1.0896, 0.14 % low. By β = 0.6 the parabola is 5.6 % low and the series is worthless.

The core equations

Two inertial frames in standard configuration; $v$ is their relative speed, $\beta \equiv v/c$ with $0 \le \beta < 1$ , and $c = 299\,792\,458\ \mathrm{m/s}$ exactly.

\gamma \;=\; \frac{1}{\sqrt{1-\beta^{2}}}\,, \qquad\quad \beta \;=\; \frac{v}{c} \qquad (1)

where: $\gamma$ — the Lorentz factor, dimensionless; $v$ — the relative speed of the two frames. Physical reading: $1 - \beta^2 \le 1$ , so its reciprocal square root is $\ge 1$ , with equality only at $\beta = 0$ . $\gamma$ belongs to a pair of frames, not to an object: a proton has no $\gamma$ until you say which frame you measure it from. The pair $(\gamma, \gamma\beta)$ satisfies a hyperbolic identity, and it lets you invert the definition.

\gamma^{2} - (\gamma\beta)^{2} = 1 \qquad\Longrightarrow\qquad \gamma\beta = \sqrt{\gamma^{2}-1}\,, \qquad \beta(\gamma) = \sqrt{1 - \frac{1}{\gamma^{2}}} \qquad (2)

where: $\gamma\beta$ — the combination that appears in relativistic momentum. Physical reading: $(\gamma, \gamma\beta)$ lies on a unit hyperbola, which is why boosts can be parametrised by a hyperbolic angle (the rapidity $\theta = \operatorname{artanh}\beta$ , developed in Rapidity and the Hyperbolic Form of Boosts). The inverse $\beta(\gamma)$ is the one you need in practice: accelerators quote energies, energy gives $\gamma$ , and $\gamma$ gives $\beta$ . Expand for small $\beta$ and the Newtonian world falls out of the first two terms.

\gamma \;=\; 1 + \tfrac{1}{2}\beta^{2} + \tfrac{3}{8}\beta^{4} + \tfrac{5}{16}\beta^{6} + \cdots \qquad (3)

where: the coefficients are those of the binomial series $(1-x)^{-1/2}$ at $x = \beta^2$ . Physical reading: every term is positive, so any truncation underestimates $\gamma$ . The $\tfrac12\beta^2$ term is Newton — multiply by $mc^2$ and $(\gamma - 1)mc^2 \to \tfrac12 mv^2$ — and the $\tfrac38\beta^4$ term is the first relativistic correction to kinetic energy. Finally the derivative, which is the reason relativistic dynamics is not just Newton with a heavier mass.

\frac{d\gamma}{d\beta} \;=\; \gamma^{3}\beta \qquad (4)

where: $\gamma^3$ — the cube, not the first power. Physical reading: at $\beta = 0.6$ the slope is $1.17$ ; at $\beta = 0.99$ , $352.7$ ; at $\beta = 0.999$ , $11\,178$ . The same $\gamma^3$ appears when you differentiate the momentum $\gamma m v$ along the direction of motion, which is why a force applied along $\mathbf{v}$ produces an acceleration smaller by $\gamma^3$ than Newton predicts. The dynamics belongs to the relativistic-momentum and relativistic-energy articles; here we note only where the cube comes from.

Where the results come from

Derivation AWhere γ comes from, and where Newton hides inside it

The goal is to get $\gamma$ honestly and then to mine the function for its physics. The transverse light clock is the shortest honest route, and it belongs to The Light Clock and Time Dilation — so we take one line from it and move on. The real work here is the series: expanding $\gamma$ for small $\beta$ shows why two centuries of physics never saw relativity, and produces the Newtonian kinetic energy as the leading term of a relativistic quantity. That is not a coincidence but a consistency requirement — any correct relativistic formula must reduce to its Newtonian ancestor as $\beta \to 0$ , and the $\gamma$ -series is the machine that performs the reduction.

Step 1. Take the light clock. A pulse bounces between mirrors a proper distance $L$ apart, transverse to the motion, so a tick at rest takes $\Delta t_0 = 2L/c$ . In a frame where the clock moves at $v$ , the pulse follows a zig-zag whose half-leg is the hypotenuse of a right triangle with legs $L$ and $v\Delta t/2$ ; since the pulse still travels at $c$ , $(c\Delta t/2)^2 = L^2 + (v\Delta t/2)^2$ . Solving, $\Delta t = 2L/\sqrt{c^2 - v^2} = \gamma\,\Delta t_0$ . Why we stop here: this is the whole of another article's business. What we want is not the tick but the coefficient, and the coefficient is $\gamma = 1/\sqrt{1-\beta^2}$ .

Step 2. Expand. Write $\gamma = (1 - \beta^2)^{-1/2}$ and use the binomial series $(1-x)^{-1/2} = 1 + \tfrac12 x + \tfrac38 x^2 + \tfrac{5}{16}x^3 + \cdots$ , valid for $|x| < 1$ , with $x = \beta^2$ . This gives equation (3): $\gamma = 1 + \tfrac12\beta^2 + \tfrac38\beta^4 + \tfrac{5}{16}\beta^6 + \cdots$ . Why the series is in $\beta^2$ and not $\beta$ : $\gamma$ is an even function of $\beta$ — it cannot care about the direction of travel — so every odd power must vanish, and it does.

Step 3. Test it. At $\beta = 0.1$ the exact value is $1.00503782$ and the three-term series gives $1.00503750$ — agreement to seven figures. At $\beta = 0.4$ the two-term parabola gives $1.0800$ against the exact $1.0911$ , an error of $1.0\,\%$ ; the three-term series gives $1.0896$ , an error of $0.14\,\%$ . At $\beta = 0.6$ the parabola is $5.6\,\%$ low. Why $\beta \approx 0.4$ is the practical death of the series: below it the error is smaller than most measurements, above it the neglected $\beta^4$ term is larger than the effect you are trying to compute.

Step 4. Collect the payoff. The relativistic kinetic energy is $(\gamma - 1)mc^2$ . Substituting the series, $(\gamma - 1)mc^2 = \left(\tfrac12\beta^2 + \tfrac38\beta^4 + \cdots\right)mc^2 = \tfrac12 m v^2 + \tfrac38\frac{mv^4}{c^2} + \cdots$ — whose leading term is exactly Newton's $\tfrac12 mv^2$ . Why this matters more than it looks: it is the correspondence principle made arithmetic. Newtonian kinetic energy is not an approximation someone chose; it is the first term of a relativistic expansion, and the next term says precisely how wrong it is. The dynamics belongs to the relativistic-energy and relativistic-momentum articles; the limit is $\gamma$ 's doing. *These results check out symbolically and numerically: the light-clock relation solves to $\Delta t = \gamma\,(2L/c)$ ; the binomial expansion of $(1-\beta^2)^{-1/2}$ to sixth order gives $1 + \beta^2/2 + 3\beta^4/8 + 5\beta^6/16$ with all odd-power coefficients zero; and $(\gamma-1)mc^2/(\tfrac12 mv^2) \to 1$ as $\beta \to 0$ . The exact-vs-series table confirms: $1.00503782$ vs $1.00503750$ at $\beta = 0.1$ ; $1.0911$ / $1.0800$ / $1.0896$ at $\beta = 0.4$ (errors $1.0\,\%$ and $0.14\,\%$ ); and $5.6\,\%$ low at $\beta = 0.6$ .*

Go deeper · Derivation B · 2 minThe structure of γ — six facts that do all the work

Derivation BThe structure of γ — six facts that do all the work

The goal is to stop treating $\gamma$ as a lookup and start treating it as a function with known properties. Six facts cover essentially every use you will make of it: it is bounded below by one; it is even; $\gamma\beta = \sqrt{\gamma^2-1}$ ; its derivative carries a cube; it inverts to $\beta(\gamma) = \sqrt{1 - 1/\gamma^2}$ ; and it diverges at $\beta = 1$ . Each is one line of algebra, and each is a physical statement in disguise.

Step 1. $\gamma \ge 1$ , with equality if and only if $\beta = 0$ . For $0 \le \beta < 1$ we have $0 < 1 - \beta^2 \le 1$ , so $0 < \sqrt{1-\beta^2} \le 1$ , so its reciprocal is $\ge 1$ ; equality needs $\beta^2 = 0$ . Why it matters: no frame ever measures another's clock as running fast. Dilation has a sign, and $\gamma \ge 1$ is that sign.

Step 2. $\gamma(-\beta) = \gamma(\beta)$ : $\gamma$ depends on $\beta^2$ only. Why it matters: the factor cannot distinguish approach from recession. Everything that does — the Doppler shift, for instance — must therefore involve something other than $\gamma$ alone.

Step 3. $\gamma^2 - (\gamma\beta)^2 = \dfrac{1 - \beta^2}{1-\beta^2} = 1$ , hence $\gamma\beta = \sqrt{\gamma^2 - 1}$ (equation 2). At $\beta = 0.995$ , $\gamma\beta = 9.9625$ and $\sqrt{\gamma^2-1} = 9.9625$ . Why it matters: $(\gamma, \gamma\beta)$ is a point on a unit hyperbola, exactly as $(\cos\theta, \sin\theta)$ is a point on a unit circle. That is the seed of the hyperbolic-rotation picture of boosts.

Step 4. Differentiate: $\dfrac{d\gamma}{d\beta} = -\tfrac12 (1-\beta^2)^{-3/2}\cdot(-2\beta) = \beta\,(1-\beta^2)^{-3/2} = \gamma^{3}\beta$ (equation 4). Numerically: $1.171875$ at $\beta = 0.6$ , $352.7$ at $\beta = 0.99$ , $11\,178$ at $\beta = 0.999$ . Why the cube is relativistic dynamics in one symbol: differentiate the momentum $p = \gamma m v$ along the motion and the same $\gamma^3$ appears, so $F = \gamma^3 m\,a$ for a longitudinal push. Force and acceleration stop being proportional, and no redefinition of "mass" repairs it in general.

Step 5. Invert. From $\gamma^2 = 1/(1-\beta^2)$ , solve for $\beta$ : $\beta(\gamma) = \sqrt{1 - 1/\gamma^{2}}$ . Check: $\gamma = 2 \Rightarrow \beta = \sqrt{3}/2 = 0.8660$ ; $\gamma = 10.0125 \Rightarrow \beta = 0.995$ exactly. Why this is the form you actually use: an accelerator hands you an energy, energy hands you $\gamma = E/mc^2$ , and this equation hands you the speed. For large $\gamma$ it is numerically dangerous — see worked example 1 for the stable rearrangement.

Step 6. Diverge. As $\beta \to 1^-$ , write $1 - \beta^2 = (1-\beta)(1+\beta) \approx 2(1-\beta)$ , so $\gamma \approx 1/\sqrt{2(1-\beta)} \to \infty$ . At $\beta = 0.995$ this shortcut gives $10.00$ against the exact $10.0125$ — $0.13\,\%$ out. Why $c$ is unreachable: $\gamma$ has a genuine singularity at $\beta = 1$ , so any quantity carrying a factor of $\gamma$ — the energy $\gamma mc^2$ above all — must diverge there. No finite push gets a massive body to $c$ ; the barrier is in the function itself. *Each of these structural facts can be confirmed directly: $\gamma^2 - (\gamma\beta)^2 = 1$ holds identically; differentiating $\gamma$ returns exactly $\gamma^3\beta$ ; inverting $\gamma(\beta)$ gives $\sqrt{1 - 1/\gamma^2}$ ; $\gamma$ is even in $\beta$ ; and $\lim_{\beta\to 1^-}\gamma = \infty$ . Numerically: $\gamma\beta = \sqrt{\gamma^2-1} = 9.9625$ at $\beta = 0.995$ ; the slopes are $1.171875$ , $352.7$ , and $11\,178$ at $\beta = 0.6$ , $0.99$ , and $0.999$ respectively; $\beta(2) = 0.8660$ and $\beta(10.0125) = 0.995$ ; and the asymptotic formula gives $10.00$ against the exact $10.0125$ .*

Limiting cases

Case	Result	Meaning
$\beta = 0$	$\gamma = 1$ exactly	no relative motion; the two frames agree about everything
$\beta \ll 1$	$\gamma \approx 1 + \tfrac12\beta^2$	the Newtonian world. $(\gamma-1)mc^2 \to \tfrac12 mv^2$ ; at $\beta = 10^{-4}$ the correction is $5\times10^{-9}$
$\beta = 0.4$	exact $1.0911$ , parabola $1.0800$	the series is $1.0\,\%$ low — the practical end of the low-speed expansion
$\beta = 0.995$	$\gamma = 10.0125$	the Frisch–Smith muon: one lifetime becomes ten in the lab frame
$\beta \to 1^-$	$\gamma \approx 1/\sqrt{2(1-\beta)} \to \infty$	at $\beta = 0.995$ this gives $10.00$ vs the exact $10.0125$ ; the divergence is why $c$ is unreachable
$\beta = 1$ (a photon)	$\gamma$ undefined	light has no rest frame; $\gamma$ simply does not apply to it

Common trap“γ is how much the moving clock slows down — it's what the observer sees.”

Two errors, tangled together. First, $\gamma$ is not a property of an object or of one observer; it is a coefficient between two coordinate systems, fixed by their relative speed. It is therefore perfectly symmetric: if you find my clock slow by $\gamma = 1.25$ , I find your clock slow by the very same $1.25$ . That looks like a contradiction and is not — we disagree about which events are simultaneous, and the bookkeeping is done in Relativity of Simultaneity and Time Dilation. Second, $\gamma$ is not what you see. Seeing means catching photons, and photons from the near and far ends of a fast object left at different times, so what a camera records is not the contracted shape at all: Terrell and Penrose showed in 1959 that a passing cube appears rotated rather than squashed. $\gamma$ tells you what a frame measures — rulers laid down, clocks read simultaneously in that frame — not what a lens receives.

Worked examples

1 · The LHC proton: γ = 7460 and 2.7 metres per second

A proton in the LHC is accelerated to a total energy of $E = 7\ \mathrm{TeV}$ . The proton's rest energy is $mc^2 = 0.938272\ \mathrm{GeV}$ . Find $\gamma$ , then find how far below $c$ the proton actually travels.

Step 1. Get $\gamma$ from the energy. Total energy is $E = \gamma mc^2$ , so $\gamma = E/(mc^2) = 7000/0.938272 = \boxed{7460.5}$ . This proton's clock runs $7460$ times slower than the laboratory's, and the $26.7\ \mathrm{km}$ ring is $3.6\ \mathrm{m}$ long in its own frame.

Step 2. Now find $\beta$ — carefully. Equation (2) gives $\beta = \sqrt{1 - 1/\gamma^2}$ , but $1/\gamma^2 = 1.80\times10^{-8}$ , so a naive subtraction throws away every digit that matters. Rearrange instead: $1 - \beta = \dfrac{1 - \beta^2}{1 + \beta} = \dfrac{1}{\gamma^{2}(1+\beta)}$ , and since $\beta \approx 1$ this is $\approx 1/(2\gamma^2)$ .

Step 3. Evaluate: $1 - \beta = \dfrac{1}{2\gamma^{2}} = \dfrac{1}{2(7460.5)^{2}} = \boxed{8.98\times10^{-9}}$ .

Step 4. Convert to a speed deficit. The proton falls short of $c$ by $c\,(1-\beta) = (2.998\times10^{8})(8.98\times10^{-9}) = \boxed{2.7\ \mathrm{m/s}}$ — walking pace. Seven tera-electronvolts buys a particle slower than light by less than a jogger's speed. That is the $\gamma$ curve's blow-up seen from the other side: near $\beta = 1$ , enormous energy buys almost no extra speed.

2 · GPS: 7.2 microseconds a day, from γ − 1 alone

A GPS satellite orbits at $v = 3.874\ \mathrm{km/s}$ . By how much does its clock lose against a ground clock in one day, from special-relativistic time dilation alone?

Step 1. Get $\beta$ : $\beta = 3874/299\,792\,458 = \boxed{1.292\times10^{-5}}$ . Deeply non-relativistic — so use the series, equation (3).

Step 2. Get $\gamma - 1$ . The parabola is overwhelmingly good here: $\gamma - 1 \approx \tfrac12\beta^2 = \tfrac12(1.292\times10^{-5})^2 = \boxed{8.35\times10^{-11}}$ (the exact value agrees to better than one part in $10^{9}$ ; the $\beta^4$ term is of order $10^{-20}$ ).

Step 3. Accumulate over a day. The satellite's proper time runs slow relative to coordinate time by the fraction $\gamma - 1$ , so in $86\,400\ \mathrm{s}$ the loss is $(\gamma - 1) \times 86\,400 = 8.35\times10^{-11} \times 86\,400\ \mathrm{s} = \boxed{7.2\ \mu\mathrm{s/day}}$ .

Step 4. Put it in context. Light travels $2.2\ \mathrm{km}$ in $7.2\ \mu\mathrm{s}$ , so an uncorrected satellite clock would corrupt a range measurement by kilometres within a day. The full GPS budget is not ours: the weaker gravity at orbit runs the clock fast by $+45.8\ \mu\mathrm{s/day}$ , and the net correction applied is $+38.6\ \mu\mathrm{s/day}$ . What this article supplies is the $7.2$ — and it comes from nothing but $\gamma - 1 = \tfrac12\beta^2$ .

Physics of the portrait

The portrait is a formula, and it arrived in physics with the wrong explanation attached: $\sqrt{1-\beta^2}$ was first written down not as a fact about time but as a desperate repair to the ether — and it was right, which is one of the strangest episodes in the subject.

Origins · 2 minLorentz factor and how this was discovered

Lorentz factor

1889–1908 · from an ad-hoc rescue to the geometry of spacetime

The factor begins as a patch. Michelson and Morley had found no ether wind, and in 1889 George FitzGerald suggested in a short note to Science that a moving body contracts along its motion; in 1892, independently, Hendrik Lorentz proposed the same. Lorentz–FitzGerald Contraction Hypothesis proves the striking thing about that guess: $\sqrt{1-\beta^2}$ is the unique factor that nulls the fringe shift — set the two arm times equal and you get a linear equation in the arm length, with exactly one root. No rival factor works. Lorentz then spent a decade seeking a reason, introducing his "local time" in the 1895 Versuch and reaching the full transformation by 1904, the factor sitting in his equations as a calculational device attached to a stationary ether. In 1905 Einstein derived the same factor from two postulates, with no ether and no contraction hypothesis — the patch had become a theorem. In 1908 Minkowski supplied the last word: writing $\gamma = \cosh\theta$ , $\gamma\beta = \sinh\theta$ makes a boost a hyperbolic rotation, and $\gamma$ stops being a correction factor and becomes geometry. The formula never changed. Only the reason for it did.

History

1889

— FitzGerald, in a brief note to Science, proposes that bodies moving through the ether change length by an amount depending on the square of their speed divided by that of light.

1892

— Lorentz, unaware of FitzGerald's note, proposes an almost identical contraction to reconcile the Michelson–Morley null result with a stationary ether.

1895

— Lorentz's Versuch introduces "local time" and the factor as part of a scheme for moving bodies, still tied to the ether.

1904

— Lorentz publishes the complete transformation, with the factor now appearing in both the space and the time equations.

1905

— Einstein derives the same factor from the two postulates alone, with no ether and no contraction hypothesis: it becomes a theorem rather than a repair.

1908

— Minkowski's Raum und Zeit recasts a boost as a hyperbolic rotation, with γ = cosh θ and γβ = sinh θ.

1959

— Terrell and Penrose show that a rapidly moving object does not photograph as contracted: light-travel time makes it look rotated instead.

Going further

Texts. Taylor and Wheeler, Spacetime Physics, 2nd ed., ch. 3–4 · Rindler, Introduction to Special Relativity, 2nd ed., §2 · Morin, Introduction to Classical Mechanics, ch. 11 · Terrell, "Invisibility of the Lorentz Contraction", Phys. Rev. 116 (1959) 1041. The key symbolic results are: the light-clock solve $\Delta t = \gamma\,(2L/c)$ ; the binomial series $1 + \beta^2/2 + 3\beta^4/8 + 5\beta^6/16 + \cdots$ with all odd coefficients zero; $(\gamma-1)mc^2 \to \tfrac12 mv^2$ ; the identity $\gamma^2 - (\gamma\beta)^2 = 1$ ; $d\gamma/d\beta = \gamma^3\beta$ ; the inverse $\beta(\gamma) = \sqrt{1-1/\gamma^2}$ ; and $\gamma \to \infty$ as $\beta \to 1^-$ . Key numerical values: $\gamma$ at $\beta = 0.1,\,0.4,\,0.6,\,0.8,\,0.9,\,0.99,\,0.995,\,0.999$ ; the series table ( $1.00503782$ vs $1.00503750$ ; $1.0911/1.0800/1.0896$ , errors $1.0\,\%$ and $0.14\,\%$ ; $5.6\,\%$ at $\beta = 0.6$ ); $\gamma - 1$ for airliner, GPS and ISS; the LHC proton ( $\gamma = 7460.5$ , $1-\beta = 8.98\times10^{-9}$ , $2.7\ \mathrm{m/s}$ ); GPS ( $7.2\ \mu\mathrm{s/day}$ ); the slopes $1.171875$ , $352.7$ , $11\,178$ ; $\gamma\beta = 9.9625$ ; $\beta(2) = 0.8660$ ; the asymptotic $10.00$ vs $10.0125$ ; and $\gamma c\tau = 6.6\ \mathrm{km}$ against $659\ \mathrm{m}$ . The uniqueness of $\sqrt{1-\beta^2}$ as the fringe-nulling factor is established in Lorentz–FitzGerald Contraction Hypothesis. Constant: $c = 299\,792\,458\ \mathrm{m/s}$ (exact).

Builds on. The Two Postulates of Special Relativity · The Light Clock and Time Dilation · Lorentz–FitzGerald Contraction Hypothesis. Leads to. The Lorentz Transformation Equations · Time Dilation · Length Contraction · Proper Time.

Why can't γ ever be less than 1? Surely a frame moving backwards should give something less.

The sign of $v$ never reaches $\gamma$ : only $\beta^2$ appears, so $\gamma(-\beta) = \gamma(\beta)$ . And for any $\beta$ in $[0,1)$ the quantity $1 - \beta^2$ lies in $(0, 1]$ , so its square root does too, and the reciprocal is at least $1$ . No arrangement of real speeds produces $\gamma < 1$ . Physically: dilation has a direction — moving clocks run slow, never fast, and no choice of which way to point the $x$ -axis reverses that. If you ever compute $\gamma < 1$ , you have used $v > c$ or dropped a sign inside the square root.

If each observer measures the other's γ as the same number, who is actually slowed?

Neither, and the question hides an assumption. $\gamma$ is a coefficient between two coordinate systems, and the relative speed is one number both frames agree on — so of course they compute the same $\gamma$ . The symmetry is real. What breaks the apparent paradox is that the two frames are not doing the same experiment: comparing one moving clock against your frame's clocks requires two of your clocks, at two places, and the other observer says those two are not synchronized. That offset accounts for the difference exactly. Time Dilation does the bookkeeping and Relativity of Simultaneity supplies the offset; here, $\gamma$ is symmetric by construction and cannot single anybody out.

Why is the derivative γ³β such a big deal? It looks like an ordinary calculus result.

Because the cube is what breaks $F = ma$ . Relativistic momentum is $p = \gamma m v$ ; push along the motion and working out $d(\gamma\beta)/d\beta$ brings in exactly the $\gamma^3$ of equation (4). A longitudinal force therefore produces an acceleration $\gamma^3$ times smaller than Newton predicts, while a transverse force is suppressed only by $\gamma$ — so acceleration is generally not parallel to force. That is why a single velocity-dependent "relativistic mass" fails: you would need two different masses depending on the direction of the push. The full treatment belongs to the relativistic-momentum and relativistic-energy articles; $\gamma$ 's job is to say where the cube comes from.

Does γ mean anything for a photon?

No, and it is worth being blunt. At $\beta = 1$ the expression $1/\sqrt{1-\beta^2}$ divides by zero: $\gamma$ is undefined, not infinite-but-usable. The physical statement underneath is that a photon has no rest frame — no inertial frame has light at rest, so there is no "clock carried with the photon" whose rate you could compare with yours. Formulas such as $E = \gamma mc^2$ are simply the wrong ones for massless particles; the correct relation is $E = pc$ , which contains no $\gamma$ at all. "Time stops for a photon" is a limit taken carelessly, not a fact.

A 7 TeV proton has γ = 7460. Does that mean it is 7460 times heavier?

That language was common a century ago and has been retired for good reason. True: the proton's total energy is $\gamma$ times its rest energy, and its momentum is $\gamma m v$ rather than $mv$ . Not true: that it acquires a scalar mass $7460\,m$ you can substitute into $F = ma$ — precisely because of the $\gamma^3$ above, the resistance to a push depends on whether the push is along or across the motion. Modern practice keeps $m$ as the invariant rest mass, the same in every frame, and puts all the velocity dependence into $\gamma$ where it is visible. The proton is not heavier; it is faster.

Where exactly does the low-speed series stop working, and what should I use instead?

As a rule of thumb, trust the two-term parabola $1 + \tfrac12\beta^2$ up to about $\beta = 0.1$ (relative error $4\times10^{-5}$ ) and the three-term version up to about $\beta = 0.3$ . By $\beta = 0.4$ the parabola is $1.0\,\%$ low and the three-term series $0.14\,\%$ low; by $\beta = 0.6$ they are $5.6\,\%$ and $1.7\,\%$ low. Above that, stop expanding and evaluate $1/\sqrt{1-\beta^2}$ directly — it costs one square root. Be careful at the opposite extreme: for $\gamma$ in the thousands, computing $\beta = \sqrt{1 - 1/\gamma^2}$ in floating point loses every significant digit. Use $1 - \beta \approx 1/(2\gamma^2)$ instead, as in worked example 1.