The Velocity Addition Formula

Velocity Addition Formula

Updated: November 15, 2026

On the Electrodynamics of Moving Bodies

Einstein-EDoMB-1905-Section §5

https://www.fourmilab.ch/etexts/einstein/specrel/www/

Einstein: Relativity, The Special and the General Theory, 1920, Chapter VI page 19, Chapter XIII page 45

https://www.google.com/books/edition/Relativity/n8QKAAAAIAAJ?hl=en&gbpv=1

Both sources show the velocity-addition-formula as it is today.

Shown: V = ( v + w ) / ( 1 + vw/c² )

Chapter VI uses upper case W on the left-hand-side.

#### Left-hand-side output

As in most of SRT, we use right-to-left equation logic ( an assignment operator so-called ) rather than left-to-right equality, an identity.

Assignment: 5 = 2 + 3

Identity: 2 + 3 = 5

Arrow: Bring in an arrow ← for right-to-left assignment:

Left ← Right

5 ← 2 + 3

#### Classical versus SRT

SRT = spezielle Relativitätstheorie, Special Relativity

Asterisk:  Use plain-V for classical mechanics and V*-asterisk for SRT.

Classical addition:  V ← ( v + w )

SRT addition: V* ← ( v + w ) / ( 1 + vw/c² )

#### Lorentz Transforms

So far we have not shown the Lorentz Transforms. So far, primed versus unprimed not shown.

Let’s not derive the v-a-f per Lorentz Transforms.  Instead, ask what the classical and SRT versions of the completed v-a-f are telling us. 

####  Obviously correct

The SRT version is obviously correct about something. Take the special case of w = c.  The reduction factor ( denominator ) is 1 + v/c, a dimensionless factor greater than 1, and V* = c.  Speed-v cannot be added to speed-c such that the end product is V = v + c.

####  Add classical speeds and distances

Let’s use RSGT-1920-Chapter-VI for nomenclature: an embankment, a carriage and a man.  Change that to pavement, a carriage and a projectile.

The classical set-up has a common meter stick on both the pavement and the carriage.  Likewise, simple proper time on both the pavement and the carriage.

Add classical speeds:  V ← v + w

The carriage moves at speed-v on the pavement.  The projectile moves at speed-w on the carriage.  The projectile moves at speed-V on the pavement.  A perfectly good equation.

Use proper time-t to rewrite as addition of distances:

Add classical distances: Vt ← vt + wt

A perfectly good equation.  Distance-wt of the projectile on the carriage is renumbered as distance-Vt on the pavement.  That distance is the sum of carriage motion on the pavement and projectile motion on the carriage.  A perfectly good equation.

####  Subtract classical speeds and distances

Subtract classical speeds:  w ← V – v

As before, the carriage moves at speed-v on the pavement.  Likewise, the renumbered projectile moves at speed-V on the pavement.  Subtraction renders speed-w of the projectile on the carriage.

Change that to subtraction of distances:

Subtract classical distances: wt ← Vt – vt

Distance of the projectile on the pavement minus distance of the carriage on the pavement is distance of the projectile on the carriage.

####  Bring in primed-unprimed time-t

Convert to subtraction and addition of distances with primed and unprimed time-t.  For convenience here, unprimed time-t is on the pavement, and primed time-t′ is on the carriage.  How our textbooks make this allocation is very ambiguous business.

Subtract classical distances: wt′ ← Vt – vt

Restated:  x′ ← x – vt

Add classical distances: Vt ← vt + wt′

Restated:  x ← vt + x′

Classical subtraction is two unprimed distances on the pavement cranking out primed distance on carriage.

Classical addition is unprimed-pavement plus primed-carriage cranking out unprimed pavement.  It works in terms of value, since in classical mode primed-unprimed are just names not qualities.

The SRT version of addition won’t crank out a correct value, since primed-unprimed are mixed qualities: a short and a long meter stick.  Can’t be added.

#### Location and segment

Subtract:  x′ ← x – vt

segment ← location – location

Location of carriage-front minus location of carriage-rear cranks out carriage size.

Add:  x ← vt + x′

location ← location + segment

Location of carriage-rear plus carriage size cranks out location of carriage-front. 

#### Bring in the Lorentz Transforms

The spatial LT shown below doing addition ( plus-sign ) papers over the deficit by writing both right-hand-side terms as primed ( x′ + vt′ ), both on the carriage when in fact term-vt′ is on the pavement.  Different meter sticks which cannot be added.

Section §4, doing length contraction and time dilation, uses the subtractive ( minus-sign ) spatial, two pavement distances, unprimed, on the right-hand-side.

Section §4: primed ← unprimed

Section §5, quote: “Solve for x.”

Section §5, not shown: unprimed ← primed

An LT derivation of the v-a-f is shown neither in EDoMB-1905 nor in RSGT-1920.  Modern textbooks and web pages show a derivation rooted in the plus-sign LT’s, and in fact the plus-sign temporal very cleverly delivers the denominator reduction-factor.

1920-Chapter-XIII deals with the right-hand-side primed-unprimed selection indirectly.   Hard to figure.

See also: Wiki, velocity addition formula.  Right-hand-side inputs x′ and t′ are explicitly primed.  Neither plus-sign nor minus-sign LT’s are announced though the algebra is pretty clearly rooted in the plus-sign pair.

#### Both versions of the LT’s

Subtract aka minus-sign:

x′ = gamma • ( x – vt )

t′ = gamma • ( t – vx / c² )

Add aka plus-sign:

x = gamma • ( x′ + vt′ )

t = gamma • ( t′ + vx′ / c² )

#### Compare LT sets

Look at the minus-sign set-of-two and the plus-sign set-of-two.

Minus-sign algebra:  the right-hand-side is input, unprimed and subtraction.

Plus-sign algebra:  the right-hand-side is input, primed and addition.

Minus-sign physics:  the right-hand-side is stationary and the pavement.

Plus-sign physics:  Is the right-hand-side stationary or moving?  Is it the carriage or the pavement?

#### Derive the v-a-f from the plus-sign pair

The general idea is simple.  Speed is distance divided by time.  In the plus-sign LT pair, change distance-x′ to distance-wt′:

x = gamma • ( wt′ + vt′ )

t = gamma • ( t′ + vwt′ / c² )

Divide: V* = x / t

Gamma divides out.  Time-t′ divides out.

numerator = w + v

denominator = 1 + vw / c²

V* = ( v + w ) / ( 1 + vw/c² )

#### Special case: w = c

Take the special case already noted, where input speed-w is w = c.  That simplifies denominator reduction factor to:

rf = ( 1 + v/c )

Numerator is:

( v + c )

Shown in Section §5:

V* = ( v + c ) / ( 1 + v/c )

V* = c

The “man” on the “carriage” of Chapter VI is now the special case of a “photon.”  As noted, the special case of ( w = c ) is correct in some sense.  Nothing can have speed V = c + v. Left-hand-side output photon on the pavement must have speed V* = c not speed V = c + v.

#### Special case: v = c

Postulate the special case of v = c, not w = c.

Numerator: ( c + w )

Denominator: ( 1 + w/c )

Quotient: V* = ( c + w ) / ( 1 + w/c )

V* = c

The v-a-f as such cannot be an LT quotient because the LT’s contain gamma and, with v = c, gamma is infinite.

1 ← infinity ÷ infinity

An exercise in futility.

#### Special case: w = 0

V* = ( v + w ) / ( 1 + vw/c² )

w = 0

V* = ( v + 0 ) / ( 1 + 0 )

V* = v

Let’s look at the two LT’s, spacial and temporal, which in quotient render speed-V*.  Write these LT’s out for the special case of w = 0

x = gamma • ( wt′ + vt′ )

t = gamma • ( t′ + vwt′ / c² )

w = 0

x = gamma • vt′

t = gamma • t′

V* =  v  = ( gamma • vt′ ) / ( gamma • t′ )

When the projectile on the carriage has speed w = 0, pavement speed is V = v + 0 = v and the v-a-f does not crank out pavement speed-V* less than V, less than v.  Makes sense. 

But separately, before dividing out in the quotient, each of distance and time doing carriage motion on the ground are gamma effects, non-Newtonian.   Or is gamma doing nothing because it divides out?  Do we really need distance-vt′ and time-t′ doctored up by gamma?  Problem revisited in the last ## section.

#### Problem: ( v + c )

Section §4 does time dilation using the minus-sign temporal LT.   What you might call the dilation equation is shown and it contains gamma. Derivation of the time equation from first principles occurs in Section §3.  A long equation is shown which contains the expression:

[ t + x′ / ( c – v ) + x′ / ( c + v ) ]

Time is distance divided by speed.  Both fractions are fixed distance-x′ divided by constant speeds, quotients being time-of-flight.  A total of three times are added.  Obscure, but in terms of algebraic value, it works.

Problem: denominator time in x′ / ( c + v ) is a speed greater than speed-c.

Worst case scenario.

#### Problem: Double-file

Go back to the special case of carriage photon-w of  speed w = c.  Photon-w is reconstructed as pavement photon-V or photon-V*.  Let’s not compare pavement photon-V-V* with carriage photon-w.  Instead, bring in new pavement photon-P of speed-c and then, on the pavement, compare photon-V-V* with photon-P.

Use classical mechanics to compare pavement photon V = v + c and pavement photon P = c. Obviously, pavement photon-V is a front-runner.

Use simple eyeball estimation to view pavement photon V* = c and pavement photon P = c.  Are these photons double-file or is photon-V* a front-runner, as was photon-V ?  Algebra V* = c renders these photons double-file.

However, if photon-V* is a front-runner ( compared to photon-P ) it can still have speed-c if its speed is determined by a succession of stationary clocks rendering time-of-flight as time-t rather than time-t′:

t = t′ • ( 1 + v / c )

Photon-P uses a succession of stationary clocks rendering proper time-t′.

A barrel of snaggletoothed fish heads.

#### Three clocks

Let’s start out with classical setups:

No time dilation – length contraction.
No primed – unprimed.
No ether drag.

The minus-sign temporal of Section §4:
 
A carriage of fixed length-L travels at speed-v.  Its rear end leaves pavement-zero at time-zero.  At its moment of departure, the carriage rear fires off a round trip photon, on-carriage speed-c, the photon going out to the front of the carriage and back, total distance of 2L.  Total time is t = 2L/c.  Classical time-t on-carriage is also time-t on the pavement and the carriage rear has moved to point x = vt. It looks like an equation of motion.  It isn’t.  Shown in Section §4, it is an equation of selection.  Output-x is location of the stationary milepost corresponding to the return of the photon on the carriage.

Two clocks are involved: the stationary pavement clock doing x = vt and a clock of speed-v on the pavement ( speed-zero on the carriage ) doing t = 2L/c.

SRT doctors up the carriage-zero clock doing round-trip photon return time t = 2L/c.  Why?  Bring in ether-drag and then denominator speed in the equation is less than speed-c.  Speed-c must be restored.

The plus-sign temporal of Section §5:

Same initial setup as the minus-sign temporal:  a carriage of fixed length-L travels at speed-v, its rear-end leaving pavement-zero at time-zero.  At moment of departure, the carriage-rear fires off a one-way photon, not the round-trip boustrophedon of Section §4.  The photon has one-way speed-c on the carriage.

We now have Eeyore’s hopeless quandary.  The event of interest is arrival of the one-way photon at carriage front at time t = L/c.  Over the course of duration-t, the carriage front moves from pavement location-L to location L + vt, and a stationary pavement clock at that location reports front arrival as time-t, and carriage speed as v = ( L + vt – L ) / t.

Problem: the photon on the carriage, which has arrival time t = L/c at carriage, is a tortfeasor if it has arrival time-t at pavement location L + vt, which is pavement speed c* = ( L + vt ) / t and c* > c.  Not allowed.

Three clocks at one place and one moment are associated with the tortfeasor’s pavement arrival: a clock at carriage front and two pavement clocks at pavement terminal L + vt.  In classical mode only one terminal clock is necessary, its time-t doing both carriage speed-v and the photon of extra distance and extra speed-c*.  In SRT mode, when the tortfeasor’s high speed-c* must be reduced to speed-c, one of terminal clocks must do extra time.  Two clocks of different face times at the pavement terminal.  Total of three clocks.

Guess what?  The Lorentz Transforms can’t do three clocks.  LT algebra is not cut out for the job.

#### deleted

Comments