THE HYPERBOLIC THEORY

OF SPECIAL RELATIVITY



J.F. BARRETT

--------------------------




















2 2 2





J.F. BARRETT




THE HYPERBOLIC THEORY
OF SPECIAL RELATIVITY

--------------------------

A reinterpretation of the Special Theory
in hyperbolic space.





'The principle of relativity corresponds to the hypothesis
that the kinematic space is a space of constant negative
curvature, the space of Lobachevski and Bolyai The
value of the radius of curvature is the speed of light.'

Borel 1913













First issued October 2006, Southampton, UK 2006 J.F.Barrett
Revised, corrected and reissued, November 2010






3 3 3
Preface

This book can be considered as the outcome of my early interest in the theory of
relativity when I felt uneasy at its presentation, particularly in its use of an imaginary
fourth dimensional time coordinate. There seemed to be something basically wrong so
that, many years later, when I saw by chance the work of Variak expressing the
theory in terms of Bolyai-Lobachevski geometry (or hyperbolic geometry), it came
as a revelation. Being convinced that this was without doubt the correct approach, I
started to work on it on my retirement publishing preliminary results at the
conferences Physical Interpretations of Relativity Theory (PIRT) held biannually in
London. The present book collects together ideas described there expanded with
additional material. It is intended to give an introductory systematic account of this
theory intended for the reader well acquainted with the standard theory of special
relativity.

Most of the mathematics in this book is elementary and known. The novelty lies in
arrangement of the material and showing inter-relationships. But there are also new
formulations and much use is made of historical aspects which the author believes
essential for a correct perspective. It is hoped that the book will demonstrate that, by
keeping close to the historical development, advances can be made without going into
sophisticated ideas even in such a well established field as the Special Theory of
Relativity.

I would like to record my gratitude to my family for their patience during the
preparation of the book.


Autumn 2006, 2010
Evia, Greece & Southampton, UK




Contact address:

Institute of Sound and Vibration,
University of Southampton,
Southampton UK

jfb@soton.ac.uk

4 4 4


Contents


Chapter 1

Comments on standard theory 6

Chapter 2

Product of Lorentz translations 18

Chapter 3

The spherical theory of Sommerfeld 26

Chapter 4

The hyperbolic theory 34

Chapter 5

Relative velocity 43

Chapter 6

Applications to optics 53

Chapter 7

Application to dynamics 60

Chapter 8

Differential Minkowski space and light propagation 72

General Bibliography 79

Appendix 1: Some historical notes 86

Appendix 2: Some mathematical notes 97


5 5 5


On the Hyperbolic Interpretation of Special Relativity

The Special Theory of Relativity, which received its initial formulation by Poincar
and Einstein in 1905, gained general acceptance in 1908 about the same time as
Minkowski's interpretation in terms of the 4 dimensional world Soon after, in the
years 1910-1914, the Yugoslav mathematician Vladimir Variak showed that this
theory finds a natural interpretation in hyperbolic (or Bolyai-Lobachevski) geometry,
an idea also put forward in less detail by a few other writers about the same time,
notably Robb (1910, etc.) and Borel (1913). Despite its apparently fundamental
nature, this hyperbolic interpretation remains little known and has not yet found its
way into standard texts on relativity theory, even after nearly a century. This lack of
interest has historical roots since hyperbolic geometry had from early times gained a
reputation as an imaginary geometry of interest only in pure mathematics and its
possible application to physical science therefore was, to most scientists, not seriously
considered.

The hyperbolic theory as put forward by Variak in 1910 arose in connexion with
the velocity composition law of Einstein. Sommerfeld in 1909 had shown how, using
Minkowski's ideas, this law may be reinterpreted in an intuitively clear way in terms
of spherical rotations in space time. But his interpretation relied essentially on
Minkowski's imaginary complex coordinate ict and was not without its difficulties.
Variak reinterpreted Sommerfelds theory in hyperbolic space and so avoided the
need for complex representation. His basic result is that the relativistic law of
combination of velocities can be interpreted as the triangle of velocities in hyperbolic
space and so the kinematic space of Special Relativity is hyperbolic. This view leads
to the redefinition of velocity as a corresponding hyperbolic velocity more appropriate
to relativity. It is needed for the correct definition of relative velocity which is
fundamental to the theory.

Although Variaks theory attracted some interest when it was first proposed, it
soon became overshadowed by the appearance of the General Theory of Relativity
which was of course also an interpretation using non-Euclidean geometry though in
its Riemannian form. Afterwards the exposition of Special Relativity continued to
follow the lines laid down by Einstein and Minkowski, the hyperbolic theory only
being mentioned rarely. It was however used in cosmology by Milne (1934) and
Fock (1955). Then, in the period just after the Second World War, ideas from
hyperbolic geometry were found to be of use when discussing collisions by Special
Relativity in atomic physics.

Recently the hyperbolic theory appears to be set for a revival as shown by, for
example, the historical review of Walter (1999) and the numerous publications of
Ungar. The theory however has yet to become generally known and accepted by the
majority of physicists.






6 6 6


CHAPTER 1 Comments on Standard Theory

1. The Lorentz Translation

The principle of relativity for mechanical phenomena dating from Galileo (1632)
applies to any frame of reference moving in uniform motion relative to an inertial
system The theory of Special Relativity arose from the realization that light
propagation must satisfy a similar principle of relativity. This led to the fundamental
contributions of Lorentz, Larmor, Poincar and Einstein establishing invariance of
Maxwell's equations of light propagation under the Lorentz transformation

The standard form of the Lorentz transformation and its inverse for translation along
the x-axis is

t' = (t - v x/c
2
) t = (t' + v x'/c
2
)
x' = (x - v t) x = (x' + v t')
y' = y y = y'
z' = z z = z'
(1)
Here is the nondimensional constant (Lorentz factor)

= 1/(1-v
2
/c
2
) (2)

The choice of this initial multiplying factor means that the transformation and its
inverse have similar form with only the sign change of v.

These equations relate two Cartesian frames of reference here called S, S', with S'
moving along the x-axis with velocity v relative to S (figure).
Fig: Frames S, S'
'
The frame S may be considered to be the base frame (the rest frame or observer
frame) and some phenomenon in the moving frame S' is to be referred back to S
which means using the inverse transformation. Sometimes however the frame S' is
taken to be that of a moving observer.
The standard assumption is that both the frames are inertial frames of reference and
that they move with constant velocity relative to one another along the x axis shown.


7 7 7


Associated with the frames of reference S, S' are the two local times t, t' The
standard convention, introduced by Lorentz and used by Poincar and Einstein, is that
these times are both set at zero at the initial instant when it is also assumed that the
two Cartesian frames coincide. This convention makes the equations homogenous.

Time t will here be taken as first variable in view of its special importance relative
to the space variables x,y,z. It is conveniently considered in the multiplied form ct,
the transformation then taking the homogeneous form

ct' = (ct x) ct = (ct' + x')
x' = (x ct) x = (x' + ct')
y' = y y = y'
z' = z z = z'
(3)
is here the nondimensional velocity v/c

The above transformation for translation along the x-axis was originally named by
Poincar (1905, 1906) a pure Lorentz transformation and by Minkowski (1909) a
special Lorentz transformation. More recently the term 'boost' has been introduced.
All these names are mathematically non descriptive. In this book the name used will
be Lorentz translation this being the natural analogue of a Euclidean translation in the
context of relativistic motion.

* Historical note: The Lorentz translation, as used by Lorentz himself in a slightly different
form and notation, included an arbitrary velocity dependent multiplier which was
subsequently set equal to unity by Poincar and Einstein for reasons of symmetry and of
establishing the group property. The equations then took their now-familiar form. Lorentz's
form of the transformation, which is slightly more general than the standard one, includes, for
example, the transformation given by Voigt (1887) under which the wave equation remains
invariant. The Lorentz multiplied form appears to have some importance in the theory (see
Chapter 8).


2. The Differential Form of the Lorentz Translation

The differential form of the Lorentz translation and its inverse is

dt = (dt' + v dx'/c
2
) dt' = (dt - v dx /c
2
)
dx = (dx' + v dt') dx' = (dx - v dt)
dy = dy' dy' = dy
dz = dz' dz' = dz
(1)
Or in homogeneous form,

c dt = (cdt' + dx') c dt' = (cdt dx)
dx = (dx' + cdt') dx' = (dx cdt)
dy = dy' dy' = dy
dz = dz' dz' = dz
(2)

8 8 8

This form has several advantages. It avoids the initial assumptions that the velocity
is uniform and that the origins coincide at time zero, assumptions which can be made
if and when necessary. If the velocity v is constant the equations can be integrated to
give the inhomogeneous form of the Lorentz translation.

From the equations the following results can immediately be deduced:

(a) Lorentz contraction: If two fixed points in the S' frame distance dx' apart are
viewed simultaneously in the S frame then dt = 0 and from the inverse of the second
equation of (1) follow the equations

dx' = dx (3)

dx = (1-v
2
/c
2
) dx' (4)

showing that the observed distance dx is contracted by the root factor.

(b) Time dilation: If two events in the S' frame are observed at a fixed value x' then
dx'=0 and from the first equation of (1) follows

dt = dt' = dt' /(1-v
2
/c
2
) (5)

So the time interval is dilated by the factor gamma, the Lorentz factor. Also

dt' = (1-v
2
/c
2
) dt = d (6)

This gives the time interval in frame S ' from the point of view of S. As indicated, it
is normally denoted by d in the notation of Minkowski.

3. Velocity Composition

A basic result, first clearly stated by Einstein (1905), is the composition rule for
finding the magnitude of the resultant of inclined velocities. Einstein assumed
that a point P has a uniform motion in frame S' which is Lorentz transformed to a
uniform motion in frame S from which is found the relation between the velocity
components u
x
, u
y
, u
z
in the S frame and the corresponding components u'
x
, u'
y
,
u'
z
in the S' frame:

u
x
= u'
x
+ v u'
x
= u
x
v
{1 + v u'
x
/c
2
} (1 - v u
x
/c
2
)

u
y
= ____u'
y
____ u'
y
= u
y
.
{1 + v u'
x
/c
2
} (1 - v u
x
/c
2
)

u
z
= ____u'
z
____ u'
z
= u
z
.
{1 + v u'
x
/c
2
} (1 v u
x
/c
2
)
(1)




9 9 9
If the differential form of the Lorentz transformation is used then these formulae
follow directly by division:

dx/dt = (dx'/dt'+v)/{1+(v/c
2
) dx'/dt'} dx'/dt' = (dx/dt - v)/{1 - (v/c
2
) dx/dt}
dy/dt = (dy'/dt')/{1+(v/c
2
) dx'/dt'} dy'/dt' = (dy/dt) /{1 - (v/c
2
) dx/dt}
dz/dt = (dz'/dt')/{1+(v/c
2
) dx'/dt'} dz'/dt' = (dz/dt) /{1 - (v/c
2
) dx/dt}
(2)
Here the initial assumption of uniform motion is avoided. This method shows
also that the composition equations are also valid if the Lorentz multiplier is used

* Einsteins composition rule: Denoting the magnitudes of the velocity of the
point P in S and S' by

u = (u
x
2
+ u
y
2
+ u
z
2
), u' = (u'
x
2
+ u'
y
2
+ u'
z
2
) (3)

and by the angle between v and u', the magnitude squared of u is found from
the above equations as

( )
2 2 2
2
2
2
v u ' 2vu ' cos (vu '/ c.sin )
u
1 (vu '/ c ) cos
+ +
=
+
(4)
or
{ }
2 2 2
2
v u ' 2vu ' cos (vu '/ c.sin )
u
1 (vu'/ c ) cos
+ +
=
+
(5)

This is Einstein's composition rule.

* Reduction to two-dimensional form: If it is assumed, as is always possible, that
the axes for x and y are taken in the plane of the two velocities u' and v the
equation for the z direction becomes redundant and the equations take the most
used form

u
x
= u'
x
+ v u'
x
= u
x
v
{1 + v u'
x
/c
2
} (1 v u
x
/c
2
)

u
y
= (1-v
2
/c
2
) u'
y
u'
y
= (1-v
2
/c
2
) u
y

{1 + v u'
x
/c
2
} (1 v u
x
/c
2
) (6)

where the last equation has been written using the square root to emphasis the
characteristic reduction of the transverse component of velocity by this factor. In the
one-dimensional case suffices can be omitted giving

u = u' + v u' = u v
(1 + v u'/c
2
) (1 v u /c
2
)
(7)





10 10 10
4. The Lorentz Group

The Lorentz group was at first defined by Poincar (1905) for one dimensional
motion and then (1906) as the group generated by the Lorentz translations in the x, y,
and z directions. These generate also the group of spatial rotations which
consequently is a subgroup of the Lorentz group. The group contains only
homogeneous transformations in x, y, z, t and is what would now be called the
restricted homogeneous Lorentz group. It only contains transformations retaining the
sense of direction of t, i.e.it is orthochronos in current terminology. As defined by
Poincar, it may include optionally scalar multiplication (dilations) but following
customary practice, this scalar will (for the present) be taken unity.

Any transformation of this homogeneous group satisfies an equation

x'
2
+y'
2
+z'
2
-c
2
t'
2
= x
2
+y
2
+z
2
-c
2
t
2
(1)

This is because the generating transformations do and so it leaves invariant the
quadratic form

x
2
+y
2
+z
2
c
2
t
2
(2)

The corresponding bilinear form is also left invariant.

The homogeneous Lorentz group is nowadays frequently defined as the group of
linear transformations leaving invariant the quadratic form (2). In this case the group
may optionally be extended to include also time reversals and space reversals.

Physically equation (1) implies that the sphere

x
2
+y
2
+z
2
= (c t)
2
(3)

describing the outward expansion of a light wave starting from the origin at time t
zero, becomes a similar sphere under Lorentz transformation. Any linear trans-
formation having this property must be a homogeneous Lorentz transformation in the
extended sense of Poincar. This property leads to an algebraic derivation which has
become standard for establishing the Lorentz transformation equations. Direct
consideration of Maxwells equations is avoided this way. Apart from the linear
transformations, nonlinear transformations also exist leaving invariant the quadratic
form (2) (see the last chapter)

If the Lorentz transformation is regarded as relating differential increments cdt, dx,
dy, dz then the group will include also nonhomogeneous transformations and will
coincide with what is nowadays, rather unhistorically, called the Poincar group.
Under such transformations

dx'
2
+dy'
2
+dz'
2
- c
2
dt'
2
= dx
2
+dy
2
+dz
2
-c
2
d t
2
(4)

which, interpreted physically, means that an infinitesimal sphere is transformed into
itself by a Lorentz transformation. Such an infinitesimal sphere may be regarded as a
Huyghens wavelet from which the finite wave (3) is generated. This approach
analyses the physical situation at a more fundamental level (see Chapter 8).

11 11 11
The group concept played an important part in the development of special relativity,
some writers relating it to the Erlanger Programme of Klein which classifies
geometries according to their invariance group. Thus Sommerfeld claimed that from
the invariance Maxwells equations under the Lorentz group, the whole of special
relativity theory could be deduced.

* References:
1) Poincar, Comptes Rendus Paris 1905, Rendiconti Palermo1906. What is nowadays called
the Poincar group occurred first in Minkowskis Space-Time lecture.
2) Sommerfeld, Lectures on Theoretical Physics, vol.3.


5. Pseudo-Euclidean Space

Towards the end of his 1906 paper Poincar introduced the pseudo-Euclidean space
with distance-squared function

x
2
+ y
2
+ z
2
(ct)
2
(1)

Using ict as fourth coordinate he took the distance squared function as

x
2
+ y
2
+ z
2
+ (ict)
2
(2)

which has the same form as Euclidean space of four dimensions. This analogy allows
Lorentz transformations to be visualised intuitively as rotations of a four dimensional
sphere

x
2
+ y
2
+ z
2
+ (ict)
2
= const. (3)

Poincar used this device to show that the only invariants under Lorentz trans-
formation are the quadratic form (1) and the corresponding bilinear form. His
representation subsequently played an important part in Special Relativity and the
geometrical difficulties associated with it tended to be ignored owing to its usefulness
in physics. One person who noted the difficulties was Robb who in his 1936 book
showed that some of the most basic geometrical ideas fail to hold in this space. As he
said in his introduction:

'This negative sign makes an enormous difference in the subject and renders
invalid a great part of what holds in ordinary Euclidean geometry'

* Complex Minkowski space: The general use of pseudo-Euclidean space came about
owing to the work of Minkowski who introduced the systematic notation

x
1
= x, x
2
= y, x
3
= z, x
4
=

ict (4)

creating the 4 dimensional world vector (x
1
, x
2
, x
3
, x
4
) with space and time on
equal terms, the 4 dimensional sphere then taking on the symmetric form

x
1
2
+ x
2
2
+ x
3
2
+ x
4
2
= const. (5)


12 12 12
In his 1908 paper On the fundamental equations of electromagnetic processes in
moving media Minkowski used this representation to give an impressive analysis of
Maxwells equations, putting them in a form which made clear their complete four
dimensional symmetry with respect to x
1
, x
2
, x
3
, x
4
. He used this symmetry to prove
their invariance with respect to Lorentz transformations by a very simple argument:
since they are clearly invariant under rotations of the space variables x
1
, x
2
, x
3
, they
must, by symmetry, also be invariant under rotations involving the time variable x
4
,
which typically would take the form

x
1
' = x
1
cos + x
4
sin
x
4
' = - x
1
sin + x
4
cos (6)

When is a purely imaginary angle Minkowski defined as i where

tanh = v/c (7)

Then equations (8) become the usual Lorentz equations. In his paper Minkowski also
introduced the well known 4- and 6-vectors of the electromagnetic quantities and in
the last part of the paper, he extended the four dimensional representation to
dynamics.

The methods and powerful analysis introduced by Minkowski exerted considerable
influence on the subsequent development of special relativity. Sommerfeld in
particular, followed and developed his line of thinking showing how the four
dimensional representation can be used for vector analysis and the equations of
mathematical physics.

However, as we hope will be seen from the alternative view of this book, the
Minkowski formulation has also had a negative influence Firstly in the promotion
and use of pseudo-Euclidean space which, while being very useful for certain
calculations, is unsatisfactory as a basis for theory. Secondly it has led to the view of
the space of special relativity as flat and essentially Euclidean which overlooks the
important relation with non-Euclidean space. Thirdly, putting time and space
variables philosophically on equal terms is a view which cannot be maintained and is
misleading, time being a distinguished variable as commonsense indicates.

Notes:
1) Minkowski used the Lorentz form of the Maxwell-Herz equations which can be written
concisely using modern suffix notation with summation convention as

f
ij
=
j
f*
ij
= 0 i, j = 1, 2, 3, 4
x
i
x
i

where the arrays f and f* are skew symmetric matrices of the electromagnetic and magnetic
vectors and the
j
give the four dimensional charge density vector.
2) Sommerfeld developed the Minkowski notation in his Ann. Phys 1909 paper, referred to in
the next chapter, in his papers in Ann. Phys 1910 on vector analysis and vector calculus, and
in his later book: Lectures on Mathematical Physics.
3) The view of Minkowski space as flat and essentially Euclidean of course also derives from
General Relativity owing to the vanishing of the Riemann-Christoffel tensor. This however
has no connexion with the non-Euclidean aspect referred to here as seen later.


13 13 13
6. Affine Minkowski Space

In his famous 1908 lecture Space and Time, Minkowski presented his vision of a
four dimensional space-time world without use of the complex representation but
instead representing the variables x, y, z, t geometrically in a space of 'world events'
by vectors (t,x,y,z). This representation is nowadays so familiar that it needs minimal
description here but some comments are necessary.

The structure of Minkowski space is determined by the group of affine trans-
formations of the variables t, x, y, z. Such transformations preserve parallelism but
make the coordinate axes oblique so distances are not preserved. For homogeneous
Lorentz transformations which are special affine transformations, Minkowski showed
that the obliqueness of the axes gives a geometrical explanation of the existence of the
Lorentz contraction.

The homogeneous Lorentz transformations conserve the family of hyperbolic surfaces

x
2
+ y
2
+ z
2
c
2
t
2
= const. (1)

which fill out the space and give it its characteristic structure. In the case when the
constant is zero, the surface becomes the two sided light cone

c
2
t
2
= x
2
+ y
2
+ z
2
(2)

which should perhaps properly be referred to as the Monge cone the properties of this
cone having previously been described in detail by Monge (1808, 1850). Events are
classified according to their relation with this cone as:

(a) space-like if c
2
t
2
< x
2
+y
2
+z
2
(b) null if c
2
t
2
= x
2
+y
2
+z
2
(c) time-like if c
2
t
2
> x
2
+y
2
+z
2
(3)

these relationships being all Lorentz invariant. This classification is usually illustrated
geometrically in reduced two dimensional form by the well known Minkowski
diagram. In this diagram the naming space-like is somewhat confusing since spatial
events in physics are mostly determined by time-like events, these being mutually
accessible by a signal of velocity less than that of light. Consequently time-like
events, not space-like events, are important in physical phenomena. Time-like events
all lie within the light cone, most evident geometrically in a 3-dimensional
representation.

* Scalar product: A scalar product of two event vectors (t, x, y, z), (t', x', y', z') may
be defined in either a space-like or time-like manner. Using the time-like manner, it is

ct ct' x x' y y' z z' (4)


It follows from the normal Cauchy inequality that the scalar product of two time-like
events is positive:

xx' + yy' + zz' {x
2
+ y
2
+ z
2
} {x '
2
+ y'
2
+ z '
2
} < ct ct' (5)

14 14 14
implying the strict inequality

ct ct' x x' y y' z z' > 0 (6)

From this follows the impossiblity of having two orthogonal time-like events with

ct ct' x x' y y' z z' = 0 (7)

If two events have such a property, one at least, must be space-like.

* Convexity: The convexity property of the cone of time-like events means that for
two time-like vectors (t,x,y,z), (t',x',y',z') the vector (t,x,y,z) + (t',x',y',z') where ,
> 0 is also time-like. This is deduced algebraically from

c
2
( t + t')
2
( x + x')
2


( y + y')
2
( z + z')
2


=
2
{c
2
t
2
(x
2
+y
2
+z
2
)} + {ct.ct' (xx'+yy'+zz')} +
2
{c
2
t'
2
(x '
2
+ y'
2
+ z'
2
)}

(8)
Here all terms on the right hand side are positive

THEOREM: (Reversed Cauchy inequality): Time-like vectors (t, x, y, z), (t', x', y', z')
satisfy

ct ct' xx' yy' zz' {(ct)
2
x
2
y
2
z
2
} {(c t')
2
x'
2
y'
2
z'
2
}
(9)
Equality holds only if the vectors are proportional.

Proof: Consider the quadratic in

f() =
2
{(ct)
2
x
2
y
2
z
2
} + 2 {ct ct' xx' yy' zz'}+ {(ct')
2
x'
2
y'
2
z'
2
}

= ( ct-ct')
2
(x-x')
2
(y-y')
2
( z-z')
2
(10)

As tends to plus or minus infinity, f() becomes positive while, when is t'/t f() is
negative or zero, being zero only if the vectors are proportional. So if the vectors are
nonproportional, f()=0 has two distinct real roots and the discriminant of the
quadratic f() is positive giving

(ct ct' xx' yy' zz')
2
{(ct)
2
x
2
y
2
z
2
}{(ct')
2
x'
2
y'
2
z'
2
} (11)

On taking positive square roots, inequality (9) follows from the positivity of the left
hand side. There is equality only if the vectors are proportional

----------------------------------------------------------------------------------------------------
References:
1) On the Monge cone see Kleins Die Entwicklung der Mathematik
2) This proof of the reversed Cauchy inequality is due to Aczl (1956). Another proof is given
later.



15 15 15

7. Ordering of Time-like Events

In the cone of time-like events a partial ordering may be defined where the relation
'after' is interpreted as meaning that a light signal can pass from the first event to the
second. Denoting the two events by (t
1
, x
1
, y
1
, z
1
), (t
2
, x
2
, y
2
, z
2
) the relation is

(t
2
, x
2
, y
2
, z
2
) (t
1
, x
1
, y
1
, z
1
) (1)

This has the meaning that

c(t
2
- t
1
) {(x
2
- x
1
)
2
+ (y
2
- y
1
)
2
+ (z
2
- z
1
)
2
} (2)

The inequality here is Lorentz invariant. This partial ordering breaks up the double
sided cone of time-like events into the future and past cones of events

(t, x,y,z) (0,0,0,0) (t,x,y,z) (0,0,0,0) (3)

* Trajectories: A physically realizable motion is represented by a trajectory or world-
line. This is a curve in the cone of time-like events where successive increments
satisfy

(c dt)
2
> (dx
2
+ dy
2
+ dz
2
) (4)

Then, with Minkowski, a variable may be defined having differential

d = {dt
2
- (dx
2
+ dy
2
+ dz
2
)/c
2
} (5)

By differentiation of the moving point (t, x, y, z) with respect to variable there are
derived Minkowskis velocity and acceleration four vectors

(dt/d, dx/d, dy/d, dz/d), (d
2
t/d
2
, d
2
x/d
2
, d
2
y/d
2
, d
2
z/d
2
) (6)

These are important for his four dimensional formulation of dynamics. The
components of the velocity vector satisfy the identity

c
2
(dt/d)
2
- (dx/d)
2
- (dy/d)
2
(dz/d)
2
= c
2
(7)

showing that the four-velocity is time-like. This identity shows that the 4 velocity
vector lies on a two sheeted hyperboloid and so is three dimensional. By
differentiation follows

c
2
(dt/d) (d
2
t/d
2
) (dx/d) (d
2
x/d
2
) (dy/d) (d
2
y/d
2
) (dz/d) (d
2
z/d
2
) = 0
(8)
i.e. the scalar product of velocity and acceleration four-vectors is zero. Since the
velocity vector has the time-like property the acceleration vector cannot also be time-
like and so has the space-like property (although, like the velocity four-vector, it is not
properly speaking an element in the Minkowski space).



16 16 16
* Reversed triangle inequality: A more precise definition of proper time follows from
the reversed triangle inequality which is an immediate deduction from the reversed
Cauchy inequality.

THEOREM: If (t, x, y, z), (t', x', y', z') are two time-like vectors.

{(ct)
2
-x
2
-y
2
-z
2
} + {(ct')
2
-x'
2
-y'
2
-z'
2
} {c
2
(t+t')
2
-(x+x')
2
-(y+y')
2
-(z+z')
2
} (9)

Proof: From the reversed Cauchy inequality follows

[{(ct)
2
-x
2
-y
2
-z
2
} + {(ct')
2
-x'
2
-y'
2
-z'
2
}]
2

= {(ct)
2
-x
2
-y
2
-z
2
} + 2{(ct)
2
-x
2
-y
2
-z
2
}{(ct')
2
-x'
2
-y'
2
-z'
2
} + {(ct')
2
-x'
2
-y'
2
-z'
2
}

< {(ct)
2
-x
2
-y
2
-z
2
} + 2{ct.ct'-xx'-yy'-zz'} + {(ct')
2
-x'
2
-y'
2
-z'
2
}

= c
2
(t+t')
2
- (x+x')
2
- (y+y')
2
(z+z')
2
(10)

Taking the positive square root gives the reversed triangle inequality which may be
written in terms of

T(t, x, y, z) = {t
2
-(x
2
+ y
2
+ z
2
)/c
2
} (11)

as

T(t, x, y, z) + T(t', x', y', z') T(t+t', x+x', y+y', z+z') (12)


* Definition of proper time along a trajectory: Another statement of the theorem is in
terms of intervals so that if e.g.

(t
1
, x
1
, y
1
, z
1
) (t
2
, x
2
, y
2
, z
2
) (t
3
, x
3
, y
3
, z
3
) (13)

then

T(t
3
t
2
, x
3
x
2
, y
3
y
2
, z
3
z
2
) + T(t
2
t
1
, x
2
x
1
, y
2
y
1
, z
2
z
1
)

T(t
3
t
1
, x
3
x
1
, y
3
y
1
, z
3
z
1
) (14)

This inequality immediately extends to a multiple division of a fixed interval by an
ordered set of time-space values. If these values lie on a world line trajectory then it
will be possible to take a minimum (or more precisely an infimum) as the number of
points on the trajectory tends to infinity so giving a rigorous definition of the proper
time for traversing the trajectory.








17 17 17

Notes:
1) The partial ordering of time-like events was observed by Robb (1913) and Carathodory
(1923) both of whom made it the basis of an axiomatic approach to relativity. Later the idea
was mentioned by other writers: Birkhoff: Lattice Theory 1948, Andronov: Canad. J.
Math.1957 and Zeeman: J Math. Phys.1960 the last two showing that ordering on the forward
cone implies its Lorentz structure.
2) Bellman proved a generalization of the reversed triangle inequality for pth powers by a
quite different method. See On an inequality ..., Amer. Math. Monthly, 1956































18 18 18

CHAPTER 2 - Product of Lorentz Translations

1. The Standard Form of a Lorentz Translation.

The standard form for the Lorentz translation for a velocity v having components v
1
,
v
2
, v
3
will be taken to be


1 2 3
2
1 1 1 2 1 3
2
2 2 1 2 2 3
2
3 3 1 3 2 3
v / c v / c v / c
cdt ' cdt
v / c 1 ( 1)n ( 1)n n ( 1)n n
dx' dx
dy' dy v / c ( 1)n n 1 ( 1)n ( 1)n n
dz' dz
v / c ( 1)n n ( 1)n n 1 ( 1)n




+


=


+



+

(1)

where n
1
, n
2
, n
3
are components of the unit vector n in the direction of the velocity.
The matrix of coefficients is symmetric and its inverse is obtained by changing v to
v. The relation can be written more concisely using partitioned matrices as

T
T
cdt ' cdt / c
/ c I ( 1)

=

+

v
dr' dr v nn
(2)

Bold letters are used for 3x1 column vectors for emphasis. Using nondimensional
parameters (= v/c) and the transformation is


1 2 3
2
1 1 1 2 1 3
2
2 2 1 2 2 3
2
3 3 1 3 2 3
n n n cdt ' cdt
n 1 ( 1)n ( 1)n n ( 1)n n dx ' dx
n ( 1)n n 1 ( 1)n ( 1)n n dy' dy
n ( 1)n n ( 1)n n 1 ( 1)n dz' dz



+

=
+

+



(3)
Equivalently, using partitioned matrices,


T
T
cdt ' cdt
I ( 1)

=

+

n
dr' dr n nn

(4)
The characteristic operator here is the 3x3 matrix

I + (-1) nn
T
= (I - nn
T
) + nn
T
(5)


This can also be considered vectorially as a dyadic (Silberstein 1914). When this
matrix acts on any vector, the first term on the right forms the component of the
vector perpendicular to n while the second term stretches the vector by a factor in
the direction n. Thus if

denotes component perpendicular to n,

{(I - nn
T
) + nn
T
}dr = dr

+ n (n
T
dr) (6)

19 19 19

2. The Product of Lorentz Translations.

The explicit representation of the product of two Lorentz translations has been an
enduring problem of Special Relativity. Since a Lorentz translation has a symmetric
matrix it is clear that such a product will not also be a Lorentz translation. As is now
well known, it is a Lorentz translation either followed or preceded by a spatial
rotation. This important observation was apparently first made by Silberstein (1914)
So that the composition of L
1
followed by L
2
may be written

L
2
L
1
= R L
r
= L
l
R (1)

R is a spatial rotation matrix and L
r
and L
l
the corresponding right and left Lorentz
translations resulting from the composition. Here it is has been assumed that the
rotation matrices for right and left multiplication are the same and this fact will be
clear from the canonical form shown below.

Note first that relations (1) imply on taking transposes that

L
1
L
2
= L
r
R
-1
= R
-1
L
l
(2)

since the Lorentz translation matrices are symmetric and the transpose of a rotation
matrix gives its inverse.

* The canonical form: The relation between right and left Lorentz translations can be
more clearly demonstrated as follows. Using the right hand Lorentz translation, let us
write the product as

T
T
1 0
0 I ( 1)


+

n
n nn
(3)

Here is a 3x3 rotation matrix.







This product is

T
T
( 1)(


+

n
n n)n

(4)
It is now convenient to introduce the unit vector n'

n' = n (5)

so that turns n into n'. From this follows, since transposition of gives its inverse,

20 20 20
n =
-1
n' =
T
n' (6)

The matrix product can now be written in the more symmetric form

T
T
( 1)


+

n
n' n'n
(7)

It can now be transformed with the same rotation matrix into the left translation form

T
T
1 0
0 I ( 1)


+

n'
n' n'n'
(8)

The form (7) will be taken as the canonical form for the product of two Lorentz
transformations. It represents the most general element of the Lorentz group
as may be seen as follows. Given two such matrices, the first may be put in the form
(3) and the second in the form (8). Their product then involves the product of two
Lorentz translations which is again a canonical matrix. Such products generate the
whole Lorentz group.
Each element of the group is characterized by the angle of which is additive on
multiplication of the matrices.


3. The Explicit Product

The explicit product can be found by the method of Silberstein. In obvious notation
the product L
2
L
1
is

T T
2 2 2 2 1 1 1 1
T T
2 2 2 2 2 2 1 1 1 1 1 1
I ( 1) I ( 1)



+ +


n n
n n n n n n


T
T
1 0
0 I ( 1)

=

+

n
n nn

(1)


There is found after adjustments for sign, transposition, etc.

=
1

2
{1 +
1

2
n
2
T
n
1
}

n =
2
{( I + (
1
-1) n
1
n
1
T
)
2
n
2
+
1

1
n
1
}

n =
1
{( I + (
2
-1) n
2
n
2
T
)
1
n
1
+
2

2
n
2
}

+ (-1) (n)n
T

=
2

2

1
n
2
n
1
T
+ (I + (
2
-1) n
2
n
2
T
)(I + (
1
-1) n
1
n
1
T
)
(2)

From the 1st and 2nd of these equations follows

21 21 21
n

= (I + (
1
-1) n
1
n
1
T
)
2
n
2
+
1

1
.n
1

1
{1 +
1

2
n
1
T
n
2
}


= {
1
-1
(I - n
1
n
1
T
) + n
1
n
1
T
}
2
n
2
+
1
.n
1
{1 +
1

2
n
1
T
n
2
}



= (1 -
1
2
) (I - n
1
n
1
T
)
2
n
2
+ {n
1
T
n
2

2
+
1
}n
1

{1 +
1

2
n
1
T
n
2
} (3)

Here the numerator is resolved into components orthogonal and parallel to n
1
and the
equation can be written, using the angle say between n
1

and n
2
,

as

n = (1 -
1
2
)
2
sin

n
1

+ {
2
cos +
1
}n
1
{1 +
1

2
cos } (4)

where n
1

is the unit vector orthogonal to n
1
. In a similar way is found

n = n' = (1 -
2
2
)
1
sin

n
2

+ {
1
cos +
2
}n
2

{1 +
1

2
cos } (5)

* The Einstein composition formula: These expressions show the relation to velocity
composition and there follows immediately for the magnitude squared


2
= (1 -
1
2
)
2
2

sin
2
+ {
2
cos +
1
}
2

{1 +
1

2
cos }
2


=
1
2
+ 2
1

2
cos +
2
2
(
1

2
)
2

sin
2

{1 +
1

2
cos }
2


(6)

Or in terms of velocity,

v
2
= v
1
2
+ 2 v
1
v
2
cos + v
2
2
(v
1
v
2
/c)
2

sin
2

{1 + v
1
v
2
/c
2
.cos }
2


(7)


------------------------------------------------------------------------------------------------------
* Reference: The calculation here is its equivalent using matrices of Silberstein's 1914
calculation using vector and dyadic notation.

4. Composition of Orthogonal Motions

The composition formulae may be simplified by taking the plane of the two velocities
as the xy plane so that the z-axis then becomes redundant. The Lorentz matrices can
then be written conveniently as 3 x 3 matrices transforming only the variables ct, x, y
while the spatial rotation matrix can be written as the 2x2 matrix

=
cos sin
si n cos




(1)

is the rotation angle, which is the angle turns n through to give n'.

22 22 22
* Composition of orthogonal velocities: This is an important special case. The
velocities may be taken along the x and y axes and then introducing as before
1
,
2
,
, for the ratios v
1
/c, v
2
/c, v/c the Lorentz matrices will be

L
1
=
1 1 1
1 1 1
0
0
0 0 1








(2)
L
2
=
2 2 2
2 2 2
0
0 1 0
0







(3)
The product is

L
2
.L
1
=
1 2 1 2 1 2 2
1 1 1
1 2 2 1 2 1 2 2
0








(4)
There follow the equations

=
1

2
n
T
= [
1

1
,
2

2
]
(n)
T
= [
1

1
,
1

2
]
(5)
Here the value of is given by


2
=
1
2
+
2
2
/
1
2
=
1
2
/
2
2
+
2
2
=
1
2
+
2
2
-
1
2

2
2
(6)

The unit vectors n, n' for the two velocity compositions are

n
T
= [
1
,
2
/
1
]
-1
= [
1
,
2
(1 -
1
2
)]
-1
n'
T
= [
1
/
2
,
2
]
-1
= [
1
(1 -
2
2
),
2
]
-1
(7)




The resulting right hand Lorentz translation can be constructed using n and the
product L
2
.L
1
written as


1 2 1
2
1 1 1 2 1
2
2 1 2 1 1 2 1
1 0 0 ( / )
0 cos si n 1 ( 1)( / ) ( 1)( / )( / )
0 si n cos ( / ) ( 1)( / )( / ) 1 ( 1)( / )



+

+



(8)

Similarly the left hand Lorentz translation can be constructed using n' and the product
written
23 23 23


1 2 2
2
1 2 1 2 1 2 2
2
2 2 1 2 2
( / ) 1 0 0
( / ) 1 ( 1)( / ) ( 1)( / )( / ) 0 cos si n
( 1)( / )( / ) 1 ( 1)( / ) 0 si n cos



+

+



(9)
The corresponding representation of L
1
L
2
is found by transposition which
interchanges the Lorentz matrices and replaces by - .


5. Rotation Angle for Orthogonal Motions

The rotation angle may be determined by forming scalar and vector products of n
and n' which give cos and sin . In the case of composition of orthogonal motions
there is found in this way, using values of n and n' found in the previous section,

cos = ((1 -
2
2
)

1
2
+ (1 -
1
2
)

2
2
)/
2
= (
2
-1

1
2
+
1
-1

2
2
)/
2

sin =
1

2
(1 - (1 -
1
2
)(1 -
2
2
))/
2
=
1

2
(1 -
1
-1

2
-1
)/
2

tan =
1

2
(
1

2
- 1)/ (
1

1
2
+
2

2
2
)
(1)
These may be expressed in various ways by algebraic transformation e.g.

cos = (
1
+
2
)(1 -
1

2
)/
2


= (
1
+
2
)/ (1 +
1

2
)
sin =
1

2
/ (1 +
1
-1

2
-1
) =
1

2
/ (1+
1

2
)
tan =
1

2
/ (
1
+
2
)
(2)
Corresponding formulae may be given in terms of the velocities v
1
, v
2,
e.g.

cos = v
1
2
(1- v
2
2
/c
2
) + v
2
2
(1- v
1
2
/c
2
)
v
1
2
+ v
2
2
(v
1
v
2
/c)
2

(3)
sin = v
1
v
2
{1 (1- v
1
2
/c
2
) (1 v
2
2
/c
2
)}
v
1
2
+ v
2
2
(v
1
v
2
/c)
2

= v
1
v
2

c
2
{1 + (1- v
1
2
/c
2
)(1 v
2
2
/c
2
) (4)

tan = v
1
v
2
{1 (1- v
1
2
/c
2
)(1 v
2
2
/c
2
)}
v
1
2
(1- v
2
2
/c
2
) + v
2
2
(1- v
1
2
/c
2
) (5)

* Half-angle formulae: The half-angle formulae for rotation angle take a simpler
form. They appear to have some importance in the theory. They may be found from:

cos
2
/2 = 1 + cos = (
1
+ 1) (
2
+ 1)
2 2 (1 +
1

2
)
sin
2
/2 = 1 - cos = (
1
- 1) (
2
- 1)
2 2 (1 +
1

2
)

tan
2
/2 = sin
2
/2 = (
1
- 1) (
2
- 1)
cos
2
/2 (
1
+ 1) (
2
+ 1)
(6)
24 24 24
From the last follows a formula due to Liebmann (quoted by Variak 1912)

cot /2 = /(
1
+ 1) (
2
+ 1)
(
1
- 1) (
2
1)
(7)

6. The Thomas Precession

The Thomas precession is a well known rotational effect occurring whenever
acceleration is in a different direction to velocity so that the velocities v, v + v at
successive instants t, t + t do not have the same direction. The combination of the
Lorentz matrices for these velocities results in an infinitesimal infinitesimal rotation
. One method of calculating this is by the formulae which have been derived for
combination of orthogonal velocities.

If vector velocities v and v are inclined at an angle , the increment v has
components parallel and orthogonal to v of v cos , v sin . Being infinitesimal,
their effects may be superimposed, and to the first order of small quantities, the
resulting infinitesimal rotation arises only from the orthogonal component. So the
rotation angle may be found from the previous formula

sin = v
1
v
2

c
2
{1 + (1- v
1
2
/c
2
) (1 v
2
2
/c
2
)}
(1)
On setting

v
1
= v, v
2
= v sin = t sin , = (2)

the following approximations will hold to first order:

(1 - v
1
/c) (1 - v/c) =
-1
,
(1 - v
2
/c) 1 (3)


The angular velocity relative to the observer is

d = v sin . = {1 - (1- v
2
/c
2
)} v sin
dt c
2
{1 + (1- v
2
/c
2
)} v
2
(4)

Relative to the moving point it is

d = {1 - (1- v
2
/c
2
)} v sin = { - 1} v sin
d v
2
(1- v
2
/c
2
) v
2
(5)

This can be written as the vector equation


d = ( - 1) v x (6)
d v

The rotation is about an axis perpendicular to v and and in the direction of v x .

25 25 25
* Approximation for v<<c: If the velocity v is small compared with that of light,

sin = v
1
v
2
v
1
v
2
= v v sin
c
2
{1 + (1- v
1
2
/c
2
) (1 v
2
2
/c
2
)} 2 c
2
2 c
2

(7)
resulting in

d = v sin
dt 2 c
2
(8)

d = v x
d 2 c
2
(9)




-------------------------------------------------------------------------------------------------------
Note: The Thomas rotation was encountered in connexion with the accelerated motion of an
electron in an electric field where the formula for spin of Goudschmidt and Uhlenbeck was
corrected by Thomas (1926, 1927). Thomas first paper gave the approximate formula, the
more accurate formula being given in his second paper. It is interesting that Borel had
previously deduced in 1913 on purely mathematical grounds that such an effect would occur.
Subsequent to Thomas work the spatial rotation associated with the composition of two
general Lorentz translations also became known as the Thomas rotation even though the
Thomas precession is only a special case. When, at a later period, the same phenomenon
found application in quantum mechanics and particle physics, the name 'Wigner rotation' also
came into use for the general rotation.


----------------------------------------------------------------------------------------------------------
References: Thomas: Nature 1926 p.514; Phil. Mag. 7 1927 1-23; Borel: C.R. Acad. Sci.
Paris 1913; Wigner: Rev. Mod. Phys 1957; Borels observation is described by Scott-Walker
1999.

26 26 26
CHAPTER 3 Sommerfelds Spherical Theory


1. The Lorentz Transformation as Rotation.

In his fundamental 1909 paper On the composition of velocities', Sommerfeld aimed
to show how the space-time view of Minkowski, then recently introduced, could be of
real use to physicists. His spherical interpretation of Einsteins composition formula
led on directly to the hyperbolic theory.

As previously described, Minkowski had, In his investigation of Maxwells equations,
somewhat incidentally, represented the Lorentz transformation as a Euclidean rotation

x' = x cos + ict sin
ict' = - x sin + ict cos (1)

The purely imaginary angle is defined by

tan = i v/c (2)

implying

cos = 1 = sin = i (v/c) = i (3)
(1 - v
2
/c
2
) (1 - v
2
/c
2
)

Sommerfeld pointed out that this representation simplifies the composition rule for
rectilinear motion since the result of two successive Lorentz transformations of angles

1
,
2
is the Lorentz transformation with angle
1
+
2
given by

iv/c = tan (
1
+
2
) = tan
1
+ tan
2
= iv
1
/c + iv
2
/c
1 - tan
1
tan
2
1 iv
1
/c iv
2
/c (4)

There follows the composition rule

v = v
1
+ v
2
(5)
1 + v
1
v
2
/c
2


This new method of deriving the composition rule soon came into general use and
was, for example, used by Einstein in his 1921 Princeton lectures replacing his earlier
method. It is now well known. Not so well known however, is the generalization of
this idea to the non-rectilinear case introduced by Sommerfeld in his 1909 paper and
subsequent writings.




----------------------------------------------------------------------------------------------
Reference: Sommerfeld A: ber die Zusammensetzung der Geschwindigkeiten
Phys.Z. 1909 826-829.


27 27 27
2. Non-commutativity of Velocity Addition

Einsteins 1905 derivation of the composition rule gave the magnitude of the
resultant of two velocities but had said nothing about its direction. Sommerfeld
attempted to clarify this situation by combining two orthogonal velocities

v
1
= (v
1
, 0), v
2
= (0, v
2
) (1)

This can be done in two possible ways - v
1
followed by v
2
and v
2
followed by v
1.

He found, in agreement with Einstein's calculation (cf chapter 1), for the resultant
velocities corresponding to these two ways the values

(v
1
, v
2
(1 v
1
2
/c
2
)), (v
1
(1 v
2
2
/c
2
), v
2
) (2)

These have the same magnitude squared

v
2
= v
1
2
+ v
2
2
(v
1
v
2
/c)
2
(3)

which can be written in either of two ways corresponding to the two resultants:

v
2
= v
1
2
+ v
2
2
(1 - v
1
2
/c) = v
1
2
(1 - v
2
2
/c) + v
2
2
(4)

From these are found Pythagoras formulae corresponding to the figure below. The
resultant velocities are represented by lines AC and C'A' of equal length.

Fig: Non-commutativity of velocity composition (Cartesian form)

Multiplication of the transverse velocity components by contraction factors results in
failure of the rectangular figure to close giving rise to what became known as non-
commutativity of velocity addition. The angle between the resultants, which was
already determined in the last chapter, is easily also found from this diagram by taking
scalar and vector products of the vectors (2).
As seen in the last chapter, composition of the Lorentz translations results in a
rotation through the angle . The curious consequence of this is apparently to make
an interchange of directions so that v
1
followed by v
2
results, not in vector AC, but in
C'A' - this vector turned through angle . It is similar for v
2
followed by v
1
.



28 28 28
3. The Spherical Representation

Sommerfeld explained noncommutativity of velocity addition as arising from addition
of displacements on a sphere determined by the Minkowski angles.

In the case of addition of orthogonal velocities, the relation between the velocities v
1
,
v
2
and their resultant v may be written

(1 - v
2
/c
2
) = (1 - v
1
2
/c
2
) (1 - v
2
2
/c
2
) (1)

This gives

1 = 1 1
(1 - v
2
/c
2
) (1 - v
1
2
/c
2
) (1 v
2
2
/c
2
) (2)

By using Minkowski angles ,
1,

2,
corresponding to v, v
1
, v
2
, the equation becomes

cos = cos
1
cos
2
(3)

which is the Pythagoras formula for the right angled spherical triangle the Minkowski
angles project on the surface of a sphere. This spherical form is equivalent to the
Cartesian form since the algebra is reversible. The Minkowski angles are all purely
imaginary as can be seen from (2) since all factors exceed unity

As in the Cartesian case the question of non-commutativity arises and there are two
triangles according to whether
1
or
2
is first. Sommerfeld represented the two
spherical triangles diagrammatically in the plane the sides of these triangles being
proportional to Minkowski angles. The figure below shows these triangles, denoted
ABC, A'B'C' the right angle being at vertices B, B'.

Fig: Non-commutativity of velocity composition (spherical form)

Unlike the Cartesian case, the triangles ABC, A'B'C' are now congruent (although
reversed) and overlap instead of being separated. But they still do not add to form a
rectangular figure because of spherical geometry which thus gives a natural
explanation of the noncommutativity. From this example Sommerfeld concluded:

For the combination of velocities in relativity theory there is valid, not formulae
of the plane, but instead spherical trigonometry (with imaginary sides.)



29 29 29
4. Spherical Form of the Einstein Composition Formula

Extending these ideas, Sommerfeld derived the general composition law for inclined
velocities. The Minkowski angle of the resultant velocity v is given by vector
addition of the arcs on a sphere of the Minkowski angles
1
,
2
for v
1
, v
2
so that by the
spherical cosine rule:

cos = cos
1
cos
2
+ sin
1
sin
2
cos ( ) (4)

where is the angle of inclination of velocities v
1
, v
2.
This is

cos = cos
1
cos
2
sin
1
sin
2
cos (5)

Changing from Minkowski angles to velocities gives


1 = 1 (iv
1
/c) (iv
2
/c) cos
(1 v
2
/c
2
) (1 v
1
2
/c
2
)(1 v
2
2
/c
2
) (1 v
1
2
/c
2
) (1 v
2
2
/c
2
)
(6)
= (1 + v
1
v
2
/c
2
cos )
(1 v
1
2
/c
2
)(1 v
2
2
/c
2
)
(7)
Inversion and squaring results in

(1 - v
2
/c
2
) = (1 v
1
2
/c
2
) (1 v
2
2
/c
2
)
(1 + v
1
v
2
/c
2
cos )
2
(8)

Then solving for v gives Einsteins composition law

_________________________________
v = {v
1
2
+ v
2
2
+ 2 v
1
v
2
cos (v
1
v
2
/c sin )
2
}
1 + (v
1
v
2
/c
2
) cos (9)


The algebra here is reversible so that starting from the composition formula the
spherical cosine rule can be deduced.















30 30 30

5. Spherical Excess and Rotation Angle

As later described by Sommerfeld (1931), the spherical representation gives a
geometrical method for the determination of the rotation angle . Reconsider the
figure illustrated below.

Fig: Rotation angle (spherical form)

The rotation angle is the angle between AC and C'A' which is seen to be

= (A + B + C) (1)

is the spherical excess E of either triangle ABC or A'B'C' and equality (1) is not
dependent on the triangles being right angled. By using its interpretation as spherical
excess, the determination of the rotation angle is reduced to the purely trigonometrical
problem of finding the spherical excess of a triangle.

Sommerfeld used this method to find the rotation angle for velocities at right angles
and applied the result to the Thomas rotation. To get the spherical excess E, he used
the formula from spherical trigonometry

sin E = sin
1
sin
2

1 + cos
1
cos
2
(2)

Here
1
,
2
are the Minkowski angles corresponding to the two velocities. Then on
going back to velocities he found the formula already derived in chapter 2:

sin = v
1
v
2
1 v
1
v
2

c
2
{1 + (1 v
1
2
/c
2
) (1 v
2
2
/c
2
)} 2 c
2
(3)

From which easily follows, as before, the formula for Thomas rotation.







31 31 31

* Note: Sommerfeld did not give details of the derivation of the relation (2) dismissing it as
elementary. Though elementary, the proof is somewhat tricky. It needs the use of sine, cosine
and cotangent formulae in the forms

sin A = sin
1
/ sin , sin C = sin
2
/ sin
cos = cos
1
cos
2

cot A cot C = cos

Then, since B is /2,

sin E = sin (A + C /2)
= cos (A+C)
= cos A cos C + sin A sin C
= sin A sin C (1 cot A cot C)
= (sin
1
/ sin ) (sin
2
/ sin ) (1 cos )
= sin
1
sin
2
(1 cos ) / (1 cos
2
)
= sin
1
sin
2
/ (1 + cos )
= sin
1
sin
2
/ (1 + cos
1
cos
2
) (4)

Quod erat demonstradum!

6. Central Projection and the Contraction Factor

An interesting aspect of the Sommerfeld representation, which Sommerfeld himself
did not analyse in detail is its relation with central projection of the sphere on to a
tangential plane. Here is seen the possiblity, through non-Euclidean geometry, of
explaining the contraction so typical of relativistic formulae. At the same time there
is revealed a shortcoming of the spherical representation.

The relation between velocities and their Minkowski angles, which may be written

v = R tan (1)

with R equal to c/i ( = ic), suggests the geometrical representation shown in the first
figure below; it leads to the idea of projection of the spherical triangle ABC on to
another triangle A
1
B
1
C
1
in the tangential plane at one vertex (A in the second figure)


Fig. Geometric meaning of Fig. Central projection of a right-
Minkowski angle -angled spherical triangle.

32 32 32
This geometrical representation of course ignores the fact that both R and are
imaginary and R is even negative. Such a figure is only permissible in the spirit of the
Sommerfeld representation which ignores all problems associated with geometrical
representation of complex quantities in the attempt to provide an intuitive picture to
aid thinking. The ultimate justification for doing this is that the resulting relationships
become strictly valid in the hyperbolic version considered below.

In this representation the tangential sides A
1
B
1
, A
1
C
1,
B
1
C
1
are

R tan
1
= v
1

R tan

= v
(R sec
1
) tan
2
= (R tan
2
) sec
1
= v
2
(1 v
1
2
/c
2
) (2)


The spherical triangle and its projection are shown below together for comparison.

ig. The spherical triangle
Fig. The projected triangle


The Cartesian components v
x
, v
y
of the resultant velocity v are A
1
B
1
, B
1
C
1
given by

v
x
= R tan
1
= v
1
_______
v
y
= (R tan
2
) sec
1
= v
2
(1 v
1
2
/c
2
) (3)

This explains the existence of the contraction factor affecting transverse velocity.

While this calculation and its diagrammatic illustration are in principle correct, a
contradiction arises from the geometrical representation. The figure above shows the
third side OB
1
greater than OA
1
after multiplication by the factor sec
1
whereas,
because
1
is purely imaginary, sec
1
is less than unity and so OB
1
should be less
than OA
1
.
33 33 33
The contradiction can be avoided by adhering to the following derivation which
however loses the geometrical picture. From the basic equations of spherical
trigonometry applied to the right-angled spherical triangle ABC we get

tan
1
= tan cos A
sin
2
= sin sin A
cos = cos
1
cos
2
(4)

From the second and third of these equations follows

tan sin A = tan
2
sec
1
(5)

so that the Cartesian components are, using these equations,

v
x
= v cos A = R tan cos A = R tan
1
= v
1

v
y
= v sin A = R tan sin A = R tan
2
sec
1
= v
2
(1 v
1
2
/c
2
) (6)






















-------------------------------------------------------------------------------------------------------
References:
1) On Sommerfelds approach to relativity see especially his Lectures on Theoretical Physics.
vol.3, Engl. tr. 1952 (Academic Press) See also the book of Rosenfeld: Non-Euclidean
Geometry. 1988 (Springer)
2) Sommerfelds paper on Thomas rotation was 'Vereinfachte Ableitung des Thomasfactors'.
Convegno di Fisica Nucleare, Rome 1931 reprinted in: Atombau und Spektrallinien 1931;
Engl.tr. Atomic Structure and Spectral Lines,1952. He again described the method on pp 234-
235 of vol.3 of his Lectures on Theoretical Physics. Engl. tr. New York 1952 (Academic
Press) A review paper with valuable historical comment is Belloni & Reina: 'Sommerfeld's
way to the Thomas precession', Eur. J. Phys. 1986. This paper credits the initial idea to
Langevin.



34 34 34

CHAPTER 4 The Hyperbolic Theory

1. Rapidity

The basic quantity of the hyperbolic theory is the rapidity w defined in terms of the
velocity by

th w = v/c (1)

the principal value of the inverse hyperbolic tangent being used for the determination
of w from this equation. Corresponding to any value of v less in magnitude to c this
equation determines a value of w lying in the range - < w < . Transforming the
Minkowski rotational representation and its inverse

x' = x cos + ict sin x = x' cos - ict' sin
ict' = - x sin + ict cos ict = x' sin + ict' cos (2)

by setting to be iw and using the identities

cos = cos iw = ch w
sin = sin iw = i sh w (3)

there arises the symmetric transformation and its inverse

ct' = ct ch w - x sh w ct = ct' ch w + x' sh w
x' = - ct sh w + x ch w x = ct' sh w + x' ch w (4)

which is the representation in terms of rapidity. Note that from (1) follows

ch w = 1 sh w = (v/c) (5)
(1- v
2
/c
2
) (1- v
2
/c
2
)

* Additivity: The characteristic property of rapidity is its additivity for velocity
composition. This additivity is obvious from the relation with Minkowski's imaginary
Euclidean form but can also be seen directly from the identity

th (w
1
+w
2
) = th w
1
+ th w
2

1 + th w
1
th w
2
(6)

which immediately results in the composition law

v = v
1
+ v
2

1 + v
1
v
2
/c
2
(7)


for the velocity v corresponding to rapidity w
1
+ w
2


so that for the rapidities w
1
,w
2
,w
corresponding to v
1
,v
2
,v it is true that

w = w
1
+ w
2
(8)


35 35 35

* Matrix representation: In terms of rapidity the Lorentz translation matrix takes the
symmetric form

chw shw
L(w)
shw chw

=

(9) (9)

with inverse

chw shw
L( w)
shw chw

=


(10)

The multiplication law is expressed by the equations

1 2 1 2 2 2 1 1
1 2 1 2 2 2 1 1
ch(w w ) sh(w w ) chw shw chw shw
sh(w w ) ch(w w ) shw chw shw chw
+ +

=

+ +

(11)

L(w
1
+w
2
) = L (w
1
) L (w
2
) (12) (12)

This semigroup property makes possible the exponential representation

L(w) = exp Kw = I + Kw + (Kw)
2
+ (Kw)
3
+ . (13). (13)
2! 3!

where K is the infinitesimal generator

0 1
K
1 0

=

(14)





----------------------------------------------------------------------------------------------------
* Historical note: Minkowski (1908) defined a quantity satisfying (in modern notation)

th = v/c

Here is the rapidity but Minkowski only used it to define his imaginary angle as equal to
i and then made no further use of it. Variak (1910 etc) used the notation w and described its
properties. Then it was independently defined by Robb in 1911 who gave it the name rapidity.
It has since been quite commonly used for representation of the Lorentz transformation even
though the Lorentz transformation loses the intuitive appeal of the imaginary rotational form
when written in this way. Notably, Whittaker, in his well known historical book (1910 etc),
used the rapidity consistently. Various notations have been used instead of w including the
original Minkowski . It has been commonly regarded as an arbitrary mathematical parameter
without any special physical significance



36 36 36
2. The Hyperbolic Triangle Law for Velocity Addition.

Since the early days of the theory of hyperbolic geometry it has been realized that
the trigonometric formulae can be obtained from those of spherical trigonometry by
using imaginary angles or equivalently by using a sphere of imaginary radius. It is
therefore a natural step to reinterpret the Sommerfeld theory in terms of hyperbolic
geometry. Since the Sommerfeld theory is based on the use of imaginary angles and
radii, the change to the hyperbolic form goes from imaginary to real form. This
change was made by Variak in 1910 soon after the appearance of Sommerfelds
1909 paper. Then followed independent work by a few others, e.g. Robb (1911) and
Borel (l9l3, 1914), also arriving at a hyperbolic representation.
The change from the spherical to the hyperbolic form was mathematically a small
step but it was of considerable importance in setting the theory on its correct path:
establishing the physical reality of hyperbolic geometry and its central significance for
the accurate statement of the Principle of Relativity.
Substituting variables iw for angles in Sommerfelds cosine formula leads to an
equation which can be written as the cosine rule in hyperbolic space:

ch w = ch w
1
ch w
2
- sh w
1
sh w
2
cos ( ) (1)

This equation is illustrated in the diagram below by the triangle ABC with slightly
inwardly curving sides to suggest the negative curvature.

Fig. A triangle of hyperbolic rapidities.

The direct derivation of formula (1) starting from Einsteins composition rule is:

v
2
= {v
1
2
+ v
2
2
+ 2 v
1
v
2
cos - (v
1
v
2
/c sin )
2
}
1 + (v
1
.v
2
/c
2
) cos (2)

(1 - v
2
/c
2
) = (1 - v
2
/c
2
) (1 - v
2
/c
2
)
(1 + v
1
v
2
/c
2
cos )
2
(3)

1 = 1 1 + (v
1
/c) (v
2
/c) cos
(1- v
2
/c
2
) (1- v
2
/c
2
) (1- v
2
/c
2
) (1- v
2
/c
2
) (1- v
2
/c
2
) (4)

Formula (1) results on substitution of hyperbolic functions.
37 37 37

* Historical note:
1) The transition from spherical to hyperbolic trigonometry by the use of complex quantities
is due to Taurinus (1826) and antedates much other work on hyperbolic geometry. See J.
Gray: Ideas of Space, 1969 Oxford Univ. Press. The fact used in Variak's transformation of
the Sommerfeld form is that a pseudospherical sphere of imaginary radius is identical with a
hyperboloid of revolution.
2) See the references to Variak, Robb and Borel in the general bibliography. Vladimir
Variak (1865-1942) was professor of mathematics at Zagreb University. His biography was
given by Kurepa (1965). Further contributions were made by Lewis & Tolman (1909), Ogura
(1913), LeRoux(1922), Milne(1934), Karapetoff (1944), Patria (1956), Smorodinski
(1963,1964). An extensive review has been given by Scott-Walker (1996, 1999)


3. Cartesian Projection from Hyperbolic Space

The analogue of central projection of a sphere on to a tangential plane exists in
hyperbolic geometry, the equivalent of the tangential plane at a point being a
Euclidean plane called the limiting plane. As a result, many trigonometric formulae
for hyperbolic space can be derived by projection in a similar way to the spherical
case though the procedure is not intuitive.
The equation corresponding to the imaginary R tan is

v = c th w (1)

is regarded as showing velocity as a Euclidean projection of the rapidity w om
hyperbolic space of negative radius of curvature c.
Considering again the composition of two velocities v
1
,v
2
at right angles, where a
hyperbolic triangle ABC with a right angle at B is projected on to a plane triangle
A
1
B
1
C
1
tangential at A=A
1
. Transcribing the equations of the spherical case gives the
to the sides a
1
, b
1
, c
1
of the Euclidean triangle A
1
B
1
C
1
the values

a
1
= c th w
2
sech w
1

b
1
= c th w
c
1
= c th w
1
(2)

The hyperbolic triangle ABC and its Cartesian projection A
1
B
1
C
1
are as illustrated
below
Fig: The right-angled hyperbolic triangle.

38 38 38

Fig: The projected triangle
(In terms of rapidities)

The sides of the projected triangle may alternatively be found using trigonometric
formulae for a hyperbolic triangle given in the mathematical appendix.

th w
1
= th w cos A
sh w
2
= sh w sin A
ch w = ch w
1
ch w
2
(3)

From the second and third equation follows

th w sin A = th w
2
sech w
1
(4)

The Cartesian velocity components v
x
, v
y
represented by sides c
1
, a
1
are then

v
x
= c th w cos A = c th w
1
= v
1
________
v
y
= c th w sin A = c th w
2
sech w
1
= v
2
{1- (v
1
/c)} (5)

as before. Here is seen the contraction factor in the transverse velocity. See the figure
below.


Fig: The projected triangle.
(In terms of velocities)




39 39 39
4. Hyperbolic Deficiency and Rotation Angle

The interpretation of noncommutativity in hyperbolic geometry corresponding to that
of Sommerfeld for the spherical case is shown in the figure below.


Fig: Non-commutativity (hyperbolic case)

Here the congruent triangles ABC, A'B'C' no longer overlap resulting in a difference
in sign between this and the spherical case. As indicated by the figure, the rotation
angle is the hyperbolic deficiency D of the triangle ABC

= - (A+B+C) = D (1)

* Calculation of rotation angle by trigonometry: The equality of rotation angle
with deficiency D reduces determination of rotation angle to a trigonometrical
problem of finding the deficiency of a triangle.

EXAMPLE 1: Two velocities at right angles.
The analogue of Sommerfelds formula proved in the last chapter is

sin D = sh w
1
sh w
2

1 + ch w
1
ch w
2
(2)

and it may be proved in a similar way merely transposing from spherical to
hyperbolic form, replacing E by D and
1
,
2
by iw
1
, iw
2
.

* Hyperbolic form of Liebmanns half-angle formula: From Sommerfeld's formula
may easily be deduced the Liebmann formula already proved in chapter 2.

sin D = (ch w
1
2
1) (ch w
2
2
1)
(1 + ch w
1
ch w
2
) (3)

cos D = (1 sin
2
D) = (ch w
1
+ ch w
2
)
(1 + ch w
1
ch w
2
) (4)

cos D + 1 = (ch w
1
+ 1)(ch w
2
+ 1)
(1 + ch w
1
ch w
2
) (5)
40 40 40

From this follows the required result:

cot D/2 = cos D + 1 = (ch w
1
+ 1)(ch w
2
+ 1) = (
1
+ 1)(
2
+ 1)
sin D (ch w
1
- 1)(ch w
2
- 1) (
1
1)(
2
- 1)
(6)
Now for any rapidity w

+ 1 = ch w + 1 = 2 ch
2
(w/2) = coth
2
(w/2)
- 1 ch w - 1 2 sh
2
(w/2) (7)

So the result may also be written in the hyperbolic form (Variak 1912).

cot D/2 = coth (w
1
/2) coth (w
2
/2) (8)

EXAMPLE 2 Two velocities inclined at an angle.

The formula for rotation angle is

cot /2 = C + cos (9)
sin

being angle of inclination of velocities, and C the constant

C = (
1
+ 1)(
2
+ 1) = coth (w
1
/2) coth (w
2
/2)
(
1
1)(
2
- 1) (10)

For orthogonal velocities, cos = 0, sin = 1 and the Liebmann formula.follows

The derivation is immediate from transcribing a formula of Lagrange for spherical
excess (see the mathematical appendix). The hyperbolic form of this formula is

cot (D/2) = ch (w
1
/2) ch (w
2
/2) + sh (w
1
/2) sh (w
2
/2) cos
sh (w
1
/2) sh (w
2
/2) sin (11)

On division of numerator and denominator by the sinh half angles, there results
formula (9). It may be transformed to the following symmetric form (see
mathematical appendix)

cot D/2 = (1 - ch
2
w
1
- ch
2
w
2
- ch
2
w
3
+ 2 ch w
1
ch w
2
ch w
3
)
1 + ch w
1
+ ch w
2
+ ch w
3
(12)

From there it is it transforms to the symmetrical formulae often quoted in the
particle physics literature:

sin D = 1 + ch w
1
+ ch w
2
+ ch w
3

2 (ch w
1
/2) (ch w
2
/2) (ch w
3
/2) (13)

cos D = (1 - ch
2
w
1
- ch
2
w
2
- ch
2
w
3
+ 2 ch w
1
ch w
2
ch w
3
)
2 4 (ch w
1
/2) (ch w
2
/2) (ch w
3
/2) (14)


41 41 41
* Note: The value of given by (10) was first given, using spinors, by van Wyk (1984)
and, apparently using Silberstein's method, by Ben-Menahem (1985). Subsequently many
papers have been devoted to its derivation by various methods. The present one by
hyperbolic trigonometry was given by the writer (PIRT 2000) and has apparently not
elsewhere appeared in the literature although Smorodinski (1962 etc.) used related
formulae from which it could easily have been deduced. The symmetrical formulae (14),
(15) were used in the physics literature from 1962. See Wick (1962), Smorodinski (1963)
etc. On the derivation see Hestenes Space-time Algebra 1966


5. Hyperbolic Velocity

Instead of the nondimensional rapidity w it is more natural in physical applications
to use the corresponding dimensional quantity

V = c w = c th
[-1]
(v/c) (1)

This was used by Variak who regarded it as the true velocity from which the usual
velocity v is found as a Euclidean projection. This is here accepted as a correct
view although for the sake of conforming with customary usage as regards the word
'velocity' the term hyperbolic velocity will be used to denote velocity as defined by
equation (1). Since v and V have the same physical dimensions the relation
between them can be shown as below where the scales of v and V are the same.

Fig: The relation between velocity and hyperbolic velocity.

* Properties of hyperbolic velocity

(a) Like rapidity, it can take any value from - to + , the hyperbolic velocity of
light being infinite. When v c correspondingly V
(b) At low velocities (v << c) hyperbolic velocity V approximates v or more
precisely,
.
V = v{ 1 + 1/3 (v/c)
2
+ 1/5 (v/c)
4
+ ... } (2)

For rectilinear motion, hyperbolic velocities combine by the same rules of addition
as do the proportional rapidities. So if the composition of velocities v
1
and v
2
gives
velocity v then corresponding hyperbolic velocities are added:

42 42 42
V = V
1
+ V
2
(3)


* The space of hyperbolic velocities: Hyperbolic velocity vectors V are defined by
their magnitude V and direction. The space of such vectors forms a hyperbolic
space of radius of curvature c and defines the kinematic space in Special Relativity.
This provides the constant c with a natural meaning and leads to a very satisfactory
view of the principle of relativity which was well expressed by Borel (1913)

'The principle of relativity corresponds to the hypothesis that the kinematic space
is a space of constant negative curvature, the space of Lobachevski and Bolyai
The value of the radius of curvature is the speed of light.'

The kinematic space approximates the classical velocity space locally for velocities
small compared with the speed of light.

The addition of hyperbolic velocities comes from rewriting the formula for the
combination of rapidities. The cosine rule giving the magnitude V for the
combination of V
1
and V
2
inclined at angle becomes

ch V/c = ch V
1
/c ch V
2
/c + sh V
1
/c sh V
2
/c cos (4)


Use of V instead of rapidity w simplifies diagrams, e.g.the Sommerfeld diagram in
hyperbolic space becomes a figure such as that shown below




-----------------------------------------------------------------------------------------------------
Note:
The addition law assumed in the present book follows the original idea of Variak (1912).
Addition of vectors in hyperbolic space by the method of parallel transport involves a
rotational effect (See e.g. Richtermeyer: Hyperbolic Geometry, Springer 1992). This is due to
the curvature of the space and occurs in relativity as the Thomson rotation. It leads to a result
similar to the gyrovector as defined by Ungar except that it applies to hyperbolic vectors and
not Cartesian vectors as does the gyrovector.
43 43 43


CHAPTER 5 Relative Velocity


1. Relative Velocities in Rectilinear Motion

All velocities are relative in Einsteins form of the principle of relativity. But most
frequently, expositions of the theory make only indirect reference to relative velocities
which are defined tacitly through the velocity composition law. This law is usually
thought of as in terms of group addition of velocities but is more properly interpreted
as combining two relative velocities to form a new relative velocity. Group addition is
most appropriate when all velocities are referred to the same origin.

The usual one dimensional situation (frames S and S') is as shown. Origin O' moves
away from origin O and the motion of a point P to the two frames is related Suppose
that, relative to origin O, the points O' and P move with velocities v
1,
v
2
and, relative
to O', P has velocity u.(see fig.)

|..u.|

Fig. Moving frames and O O' P
relative motions o------------o------------o-------------
v
1
v
2


Transferring the origin from O to the new origin O' gives, by the composition rule,

v
2
= v
1
+ u .
1 + v
1
u / c
2
(1)

On solving for u,

u = v
2
- v
1
.
1 - v
2
v
1
/ c
2
(2)

This formula gives the relative velocity of two points (here O' and P) moving with
velocities v
1
and v
2
relative to an origin (here O).

It might at first appear that since v
1
and v
2
are dependent on the origin O, then u
must also be dependent on this origin but it can be seen from the meaning that it is
not. This fact can also be verified algebraically by substituting in (2) for v
1
and v
2
the
expressions

v
1
+ u' , v
2
+ u' .
1 + v
1
u' / c
2
1 + v
2
u' / c
2
(3)

u' being a further arbitrary relative velocity, when it is found that the expression (2)
remains unchanged. In this respect formula (2) differs from the composition formula
(1) which is dependent on origin and for this reason formula (2) is preferable as a
starting point to formula (1).
44 44 44
* Re-derivation of the composition rule: With formula (2) for relative velocities as a
starting point it is possible to deduce the rule for composition of relative velocities in
a more convenient way. Consider the situation in the figure where 3 points P
1
, P
2
, P
3

are in motion relative to an origin O.


O P
1
P
2
P
3
Fig: Composition of relative o----------o-----------o---------o---------
rectilinear velocities v
1
v
2
v
3



The relative velocities u
2/1
of P
2
to P
1
and u
3/2
of P
3
to P
2
, are

u
2/1
= v
2
- v
1
u
3/2
= v
3
v
1 - v
2
v
1
/c
2
1 - v
3
v
2
/c
2
(4)

The relative velocity u
3/1
of P
3
relative to P
1
is

u
3/1
= v
3
- v
1
1 - v
3
v
1
/c
2
(5)

After some calculation, the composition rule follows purely algebraically as

u
3/1
= u
3/2
+ u
2/1
1 + u
3/2
u
2/1
/c
2
(6)

Reference: Prokhovnik: The Logic of Special Relativity, 1967

* Use of hyperbolic velocities: these relations simplify and become more transparent
by the use of rapidities or hyperbolic velocities. Suppose, as before, there are three
moving points P
1
, P
2
, P
3
referred to an origin O. Then

w
2/1
= w
2
- w
1
U
2/1
= V
2
- V
1
(7)

w
3/2
= w
3
- w
2
U
3/2
= V
3
- V
2
(8)

from which by addition,

w
3/2
+ w
2/1
= w
3
- w
1
= w
3/1
U
3/2
+ U
2/1
= V
3
- V
1
= U
3/1
(9)

* Galilean and Lorentzian translational invariance: The use of rapidity and
hyperbolic velocity makes clearer the distinction between two forms of translation
invariance. The two forms are

v v + u (Galilean) (10)

V V + U (Lorentzian) (11)




45 45 45


2. Definition of Relative Velocity in Three Dimensions

Assume two points P
1
, P
2
move with velocities v
1
, v
2
relative to an origin O (fig. 1).

Fig.1: Relative motion the space diagram

As suggested by the rectilinear case, the natural definition of the relative velocity of
these points would be that found by writing the composition rule for the difference of
velocities, i.e. for - v
1
combined with v
2
. This would lead to

v
2/1
= {v
2
cos - v
1
}n
1
+ (1 - v
1
2
/c
2
) v
2
sin

n
1

{1 v
1
v
2
/c
2
cos } (1)

Here is the angle between the two velocities The formula may be rewritten as

v
2/1
= n
1
n
1
T
(v
2
v
1
) + (1 v
1
2
/c
2
) (I n
1
n
1
T
) (v
2
v
1
}
(1 v
1
T
v
2
/c
2
) (2)

where n
1
n
1
T
projects in the direction of v
1
and (I - n
1
n
1
T
) projects perpendicular to v
1
.

The relation of this formula with the classical expression v
2
v
1
is illustrated below
showing v
2
v
1
resolved into components parallel and transverse to v
1
. The
transverse component is reduced by the root factor.

Fig: Relative velocity components
46 46 46
* Fock's derivation: This expression (2) was derived by Fock (1955) by transfering
the origin from O to P
1
so considering P
1
to be at rest. As in the classical case, the
velocity of P
2
with respect to this new origin is then defined as the velocity of P
2

relative to P
1
.

* Detail: Fock assumed uniform motions and wrote the Lorentz transformation for the the
move to the new origin at P
1
with coordinates r ', t '. This transformation has the differential
form

dt' =
1
(dt dr.v /c)
dr' = dr v
1
dt + (
1
1) n
1
(n
1
.dr v
1
dt) (3)

where n
1
is a unit vector in the direction of v
1
and factor
1
refers to this velocity:

By division Focks formula for the relative velocity v
2/1
is found as

v
2/1
= dr' = v
2
- v
1
+ (
1
- 1) n
1
.{(n
1
.v
2
)-v
1
} (4)
dt'
1
(1 - v
1
.v
2
/ c
2
)

By rearrangement this may be written in the form

v
2/1
= n
1
n
1
T
(

v
2
v
1
) + (1 v
1
2
/c
2
) (I n
1
n
1
T
) (v
2
v
1
}
(1 v
1
T
v
2
/c
2
) (5)

which is the same as formula (2).

By the same definition the reverse relative velocity, found by interchange of suffixes,
would be

v
1/2
= {v
1
cos v
2
}n
2
+ (1 v
2
2
/c
2
) v
1
sin

n
2

{1 v
1
v
2
/c
2
cos } (6)

= n
2
n
2
T
(

v
1
- v
2
) + (1 v
2
2
/c
2
) (I n
2
n
2
T
) (v
1
v
2
}
(1 - v
1
T
v
2
/c
2
) (7)

What is surprising is that the two relative velocities are not negatives of one another
as might have been expected. They do however have the same magnitude found by
the difference form of Einsteins composition rule:

v
2/1
2
=

v
1/2
2
= (v
1
2
2 v
1
v
2
cos + v
2
2
) (v
1
v
2
/c)
2
sin
2

{1 v
1
v
2
/c
2
cos }
2
(8)

The two relative velocities consequently differ only in direction. This rotational
effect is taken into account in the matrix definition given in the next section.

----------------------------------------------------------------------------------------------
References: Fock: The Theory of Space, Time and Gravitation, Moscow 1955, Engl.tr.
Oxford 1959. See also Mller: The Theory of Relativity 1955.






47 47 47
3. Matrix Representation of Relative Velocity

Relative velocity may be defined by a similar idea to that used previously for the one
dimensional case where all motions are referred to an observer O. For two moving
points P
1
, P
2
the relations between coordinate changes referred to observer O are, with
use of appropriate suffices

[ ] [ ]
0 0 1 2
1 2
0 0 1 2
cdt cdt cdt cdt
( ) , ( )

= =


v v
dr dr dr dr

(1)
the matrices here being Lorentz translations. From these follows

[ ][ ]
1 2 1
2 1
2 1
cdt cdt
( ) ( )

=


v v
dr dr
(2)

Although matrices L(v
2
), L (v
1
) are defined relative to O, their product L(v
2
)L (v
1
)
-1

is independent of O. For supposing that the relation between coordinate changes
between the two different observers O, O' is

[ ]
0 0
0
0 0
cdt cdt '
( )
'

=


v
dr dr
(3)

the matrices L (v
2
), L (v
1
) will transform by

L(v
2
) L (v
2
) (v
0
), L (v
1
) L (v
1
) (v
0
) (4)

leaving the product L (v
2
) L (v
1
)
-1
unchanged. The relative velocity matrix is
consequently well defined independently of observer as

2/1
= L(v
2
) L(v
1
)
-1
(5)

from which follows for the inverse

1/2
= L(v
1
) L(v
2
)
-1
= (
2/1
)

-1
(6)

Note that a relative velocity of two moving points is represented by a Lorentz
translation when and only when the origin is taken at one of the moving points. This
was in fact what Fock had done which gave the incorrect impression that a relative
velocity in general can be so represented.










48 48 48
4. Re-derivation of Focks Expression for Relative Velocity

The product L (v
2
)L (v
1
)
-1
when written as L (v
2
)L (-v
1
) may be evaluated as a product
of Lorentz translations and written in the form R L(v) where R is a spatial rotation and
(v) a Lorentz translation. More explicitly it will be

T
T
1 0 / c
0 / c I ( 1)


+

v
v nn
(1)

where is a 3x3 spatial rotation matrix. On forming the product there is found

=
1

2
{1 - v
1
T
v
2
/c
2
}
v =
2
{(I + (
1
-1) n
1
n
1
T
) v
2
-
1
v
1
}
( v) =
1
{(I + (
2
-1) n
2
n
2
T
)(-v
1
)+
2
v
2
} (2)

From the first and second of these equations is found

v = {I + (
1
-1) n
1
n
1
T
}v
2
-
1
v
1

1
{1 v
1
T
v
2
/c
2
} (3)

Slight rearrangement gives Focks expression

v = v
2
- v
1
+ (
1
- 1)n
1
{(n
1
T
v
2
)-v
1
} (4)

1
(1 - v
1
T
v
2
/c
2
)

so identifying v as v
2/1
By further rearrangement this may be written as

v
2/1
= n
1
n
1
T
(

v
2
- v
1
) + (1 v
1
2
/c
2
) (I - n
1
n
1
T
) (v
2
- v
1
}
(1 - v
1
T
v
2
/c
2
) (5)

This resolves the difference of velocities into longitudinal and transverse
components. Interchange of suffixes gives the reverse relative velocity:

v
1/2
= n
2
n
2
T
(

v
1
- v
2
) + (1 v
2
2
/c
2
) (I - n
2
n
2
T
) (v
1
- v
2
}
(1 - v
1
T
v
2
/c
2
) (6)

From above it is seen that

v
2/1
=

v
1/ 2
(7)

showing the relation between the two relative velocities which are not negatives of
one another. The matrix (1) may be written symmetrically in canonical form as

T
T
( 1)



+


2/1
1/2 1/2 2/1
n
n n n
(8)




49 49 49
5. Composition of Relative Velocity Matrices

Let there be three points P
1
, P
2
, P
3
moving with velocities v
1
,

v
2
,

v
3
relative to the
observer O. The relative velocity matrices are with L() denoting Lorentz translation,

2/1
= L(v
2
) L(v
1
)
-1

3/2
= L(v
3
) L(v
2
)
-1

3/1
= L(v
3
) L(v
1
)
-1
(1)

from which follows the composition rule:

3/1
=
3/2

2/1
(2)

which shows the transitivity of matrix multiplication for connected relative
velocities. This equation can obviously be extended to any number of consecutive
stages.

* Moving frames of reference: In the standard situation of systems S and S' with S'
moving with velocity v relative to S, consider the motion of a point P moving with
velocities u, u' relative to S and S'. As seen by an observer at rest at the origin O of
the S frame, the velocity of P relative to the origin O' in the S' frame is

(u') = L(u) L(v)
-1

(3)

where L(.) denotes a Lorentz translation matrix. Now the velocity u will be
defined by the equation

L(u') L(v) = R L(u)

(4)

R being here a rotation matrix So the observed relative velocity of P is

(u') = R
-1
L(u')

(5)

The transformation law then takes the form

L(u) = {R
-1
L(u')}L(v) = (u') L(v)

(6)

which implies that it is not possible to find L(u) by just multiplying Lorentz
translation matrices L(v) and L(u') as was done, for example, in the 1909
Sommerfeld paper discussed in chapter 3 While valid mathematically, this product
would have no physical meaning. The multiplication must be done as explained
here in which case the resultant has no rotation and addition takes place in the usual
way for vectors.



----------------------------------------------------------------------------------------------------
* Remark: The writer believes that in this way a resolution can be found of the difficulty in
multiplying Lorentz translations which has been much discussed in the literature following
the work of Mocanu and Ungar (see bibliography).

50 50 50

6. Relative Velocity and Hyperbolic Geometry

Fock (1955) deserves the credit for realizing the importance of hyperbolic space for
the correct representation of relative velocity in Special Relativity. He also
introduced the appropriate representation of this space - that of Beltrami-Klein.

Since the magnitude of the relative velocity is given by the Einstein composition
formula for the difference of velocities, the corresponding calculation for velocity
addition given previously is easily modified to give the relation

1 - v
2
= (1 - v
1
2
/c
2
)(1 - v
2
2
/ c
2
)
c
2
(1 - v
1
v
2
cos / c
2
)
2
(1)

between the velocities v
1
, v
2
, the angle between them, and the relative velocity v.
Taking square roots and inverting results in

1 = c
2
v
1
v
2


cos
(1 - v
2
/c
2
) (c
2
v
1
2
)(c
2
v
2
2
) (c
2
v
1
2
)(c
2
v
2
2
) (2)

which, using corresponding rapidities w, w
1
, w
2
, gives

ch w = ch w
1
ch w
2
sh w
1
sh w
2
cos (3)

or equally in terms of hyperbolic velocities V, V
1
, V
2

ch V/ c = ch V
1
/c ch V
2
/c sh V
1
/c sh V
2
/c cos (4)

The relationships between V, V
1
, V
2
may consequently be represented by the sides of
a triangle in hyperbolic space as shown.

Fig: Hyperbolic triangle of relative rapidities.

This diagram may be interpreted as showing vector subtraction of hyperbolic
velocities to form a relative hyperbolic velocity, as justified by the matrix form

L (v
2
) = (v
2/1
) L (v
1
) (5)

where the matrix (v
2/1
) includes a correcting rotation so that there is no resulting
turning angle in the right hand side.

51 51 51
7. The Beltrami Representation

In Special Relativity, relative to a chosen origin and Cartesian frame, admissible
velocities (v
x
, v
y
, v
z
) are restricted by the inequality

v = v
x
+ v
y
+ v
z
< c (1)

The velocity space is consequently represented by vectors drawn from the origin to
points inside this sphere which may be given by spherical coordinates:

v
x
= v n
1
= v sin cos
v
y
= v n
2
= v sin sin
v
z
= v n
3
= v cos (2)

where n
1
, n
2
, n
3
are direction cosines of the velocity direction and , spherical
coordinates. On using rapidity w as radial parameter instead of v, these velocity
components become

v
x
= c th w n
1
= c th w sin cos
v
y
= c th w n
2
= c th w sin sin
v
z
= c th w n
3
= c th w cos (3)

The rapidity w takes values from zero to infinity so that in terms of w, , the space
becomes infinite in extent. This is the Beltrami representation.of hyperbolic space in
its normal parametrization The Riemannian metric in this space, due to Beltrami, can
be introduced by considering the magnitude squared of the differential relative
velocity of points having velocities v and v + dv. This magnitude squared is given by
the Einstein composition rule for the difference which in this case results in the polar
coordinate expression

dv + 1 v [d + sin d]
(1 v/c) (1 v/c)
(4)

On change to rapidity, this leads to a Riemannian metric

c dw + c sh w [d + sin d] (5)

or, in terms of the corresponding hyperbolic velocity V it is,

dV + c sh V/c [d + sin d] (6)

which is the standard form of the Riemannian metric of a hyperbolic space.

Equivalently the metric squared element can be written
`
dV + [r d + r sin d] (7)

This expression shows clearly the deviation of the space from Euclidean form when
expressed in spherical coordinates.


52 52 52
The (nondimensional) metric in this space is given by the magnitude of the relative
rapidities w
2/1
, w
2/1
satisfying the equation which follows from the last section


c ch w
2/1
= c ch w
2/1
= c - v
1
.v
2

(c - v
1
.v
1
)(c - v
2
.v
2
) (8)

The equivalence of (12) with the composition rule for the relative velocity follows
from the calculation in the previous section.

Beltrami space has the convenience that velocities are represented by straight lines.
This may be illustrated as below in the case of three moving points B, C, D with
hyperbolic velocities V
1
, V
2
, V
3
relative to a point A. The relative velocities are
found by completing triangles as shown giving the triangular composition of two
relative hyperbolic velocities V
2/1
and V
3/2
to give V
3/1


Fig: Composition of relative hyperbolic velocities.

This figure corresponds to the matrix composition rule for relative velocities given
above.








-----------------------------------------------------------------------------------------------------
* Notes: Further details on the Beltrami representation are given in the mathematical
appendix. It was Milne (1934) who first used this representation in special relativity for his
model of the expanding universe (a use made more explicit by Whitrow 1939) Application of
the Beltrami representation to velocity space in special relativity is due to Fock (1955)






53 53 53
CHAPTER 6 Applications to Optics


1. Aberration and Oblique Doppler Effect

Here, starting from the original idea of Bradley on stellar aberration, the observer is
commonly thought of as moving while the source of light is considered to be
stationary. So frame S will represent the stationary source frame for emitted light and
frame S the moving frame of the observer.

* The principle of invariance of the phase: If the incoming wave has direction
cosines (l, m, n) relative to the source frame S, the phase relative to S is

= 2 f {t - (lx +my + nz)/c} (1)

f being frequency in Herz. Similarly if the wave has direction cosines (l', m', n')
relative to the observer frame S' the phase relative to S' is

= 2 f ' {t' - (l'x' +m'y' + n'z')/c} (2)

The basis of the calculation is that these two expressions must be the same. This is
the 'Principle of invariance of the phase' used by Lorentz (1886,1895) with classical
transformation formulae from which he found approximately correct formulae for
both Doppler effect and aberration. Einstein (1905) used the same method with the
corrected transformation and so found the relativistic formulae which have since
become very much used. Variak rephrased them in terms of rapidity

If the incoming wave is inclined at an angle to the direction of motion of the
observer relative to the source along the x-axis then the phase of the incoming wave
relative to the two frames S can be written in the simpler form

= 2 f {t - (x cos + y sin )/c}
= 2 f '{t' - (x' cos ' + y' sin ')/c} (3)

so that the principle of invariance gives

f (ct - x cos - y sin ) = f ' (ct' - x' cos ' - y' sin ') (4)

(a) True values in terms of observed values

From invariance of the phase combined with

ct' = ct ch w - x sh w
x' = - ct sh w + x ch w
y' = y (5)


There follows



54 54 54

f = f ' (1 + v /c.cos ')
(1-v/c) (6)

cos = cos ' + v/c
1 + v/c.cos ' (7)

sin = (1-v/c) sin '
1 + v/c.cos ' (8)

(b) Observed values in terms of true values

Similarly, starting from

ct = ct' ch w + x' sh w
x = ct' sh w + x' ch w
y = y' (9)

there is found through substitution

f ' = f (ch w - sh w cos ) = ch w f ( 1 - th w cos )
f ' cos ' = f (ch w cos - sh w ) = ch w f ( cos - th w )
f ' sin ' = f sin (10)

This gives the same formulae with change in sign for v

f ' = f (1 - v/c cos ) (11)
(1-v/c)

cos ' = cos - v/c
1 - v/c cos (12)

sin ' = (1-v/c) sin
1 - v/c cos (13)

* Transverse Doppler Effect: The transverse Doppler Effect occurs when the light
source is observed at right angles to the direction of motion. This requires solving the
equations for f ' when ' is /2. Using (7) to find cos in terms of cos ' and
substituting in (11) there is found observed frequency f ' in terms of observed angle '

f ' = f (1-v/c) .
(1 + v /c.cos ') (14)

When ' is /2 this gives the formula for the transverse Doppler effect:as

f ' = f (1-v/c) (15)

----------------------------------------------------------------------------------------------------
* References:
1) Lorentz's work on aberration and the Doppler Effect is well described by Miller (1998)
2) For history and interesting comments on the transverse Doppler Effect see the books of
Stevenson & Kilminster (1958), and Prokhovnik (1969)
55 55 55

2. The Radial Doppler Formula

When the angles , ' are zero so that the observation is in the direction of motion
the frequency formulae becomes

1 v / c
f ' f
1 v / c

=
+
(1)

This is easily found directly from the simplified form taken by the phase as

= 2 f (t - x /c) (2)

on making a Lorentz translation along the x-axis to new coordinate x', t' (Einstein
1907) Since frequency f and wave length are related by

f = c (3)

the formula may equally be written in terms of wave-length as

1 v / c
'
1 v/ c
+
=

(4)

* Transitivity of relativistic Doppler shift: The important transitivity property of the
relativistic formula is related to its dependence only on relative velocities. Suppose
that on the path of the wave P
0
, P
1
, P
2
are three collinear points having velocities v
0
,
v
1
, v
2
and relative velocities v' (= v
1/ 0
), v" (= v
2/ 1
), v (= v
2/ 0
) as in the figure below.

|....v...|
|v'........|...v"...|

Fig: Doppler Effect with ---o----------------o---------------------o----
collinear moving points P
0
P
1
P
2

v
0
v
1
v
2


Then v can be obtained from v', v" by the composition formula which can be put into
the form

1 v / c 1 v'/ c 1 v''/ c
1 v/ c 1 v'/ c 1 v''/ c
+ + +
=

(5)

Taking square roots this implies the consistency of the formulae (7) with (8) below.

1 0 2 1
1 v'/ c 1 v''/ c
,
1 v'/ c 1 v''/ c
+ +
= =

(6)


2 0
1 v / c
1 v/ c
+
=

(7)

56 56 56
* A modified redshift formula: The classical linear Doppler formula as used in
astronomy for nebula recession (Hubble 1936) arises from binomial theorem
approximation to the relativistic formula (5):

' = (1 + v/c) (8)

This can be written as

z = v/c (9)

where z is the redshift

z = ( ' - )/ (10)

This linearized Doppler formula does not have the transitivity property, implying for
example, that in astronomy, all redshifts are Earth centred. However a modified form
of the relativistic formula (5) may be found of similar form to the linearized Doppler
formula which does have the desired transitivity property. This is found by taking
logarithms of equation (5) and using the equation relating v to hyperbolic velocity V

[ 1]
1 v / c
ln th (v/ c)
1 v/ c

+
=

= V/c (11)

Then this may be written analogously to the classical linearized Doppler formula as

Z = V/c (12)

where Z is the modified redshift

Z = ln ( '/ ) (13)

This form of the Doppler shift law relates the quantities Z and V which are both
additive over subintervals as illustrated in the fig. below.


|....V...|
|V' .|.....V"..|

P
0
P
1
P
2

Fig. The additive relation between ----o--- ------------o-----------------------o--
hyperbolic velocity and
logarithmic Doppler shift < ------Z'----- >< ---------Z"------- >
< ------------ Z = Z' + Z" ---------- >


In this way the transitive property is restored to the linear Doppler formula.





57 57 57

* Logarithmic Doppler shift: The quantity Z which has been defined here provides an
alternative and improved measure of Doppler shift. It may be called 'logarithmic
measure of Doppler shift' or for short 'logarithmic Doppler shift. It is a strictly
increasing function of the wave-length ratio '/ reducing to normal redshift for low
relative velocities since, when ' differs little from ,

Z = ln {1 + (' - )/ } (' - )/ = / = z (14)

The following properties of logarithmic Doppler shift Z contrast favourably with
those of normal Doppler redshift z for which they fail to hold exactly:

(a) Change from wave-length to frequency or vice versa merely involves change of
sign

ln ( '/ ) = - ln (f '/ f) (15)

(b) Z has antisymmetry between emitter and receiver undergoing a simple sign
change when these are reversed:

ln ( '/ ) = - ln ( / '), ln (f '/ f) = - ln (f / f ') (16)

(c) Z is additive (transitive) as expressed by the equations

ln ( '/ ) + ln ( "/ ') = ln ( "/ ), ln (f '/f ) + ln (f "/ f ') = ln (f "/ f )

(17)








-------------------------------------------------------------------------------------------------------
* Notes:
1) The transitivity property of the radial Doppler formula was observed by Prokhovnik (1969)
and Jnossy (1971).
2) The logarithmic Doppler shift measure can also be used in General Relativity. Using the
known formulae for cosmological redshift and time-varying Hubble parameter H(t):


2


= R(t
2
) , H(t) = 1 dR(t)

1
R(t
1
) R dt

there is found the generalized logarithmic redshift law

Z = ln (
2
/
1
) = ln {R(t
2
)/ R(t
1
)} = ln R(t
2
) ln R(t
1
) = H(t) dt

Further details are given in the writers 1992 PIRT paper.



58 58 58

3. The Hyperbolic Interpretation of Relativistic Aberration

The wave theory interpretation of aberration loses contact with the simple picture of
triangular velocity addition of the classical Bradley theory. This picture can be
partially restored by using the velocity composition formulae. Consider an incoming
photon moving with velocity components relative to the source frame S of

u
x
= - c cos , u
y
= - c sin (1)

Using the velocity composition formula the corresponding components relative to the
observer frame S ' are

u
x
' = - (c cos + v) (2)
(1 + v/c. cos )

u
y
' = - c sin (1 - v
2
/c
2
) (3)
(1 + v/c. cos )
Here

u
x
'
2

+ u
y
'
2
= c
2
(4)

so that it is permissible to put

u
x
' = - c cos ', u
y
' = - c sin ' (5)

when substitution leads directly to the aberration formulae. There is however a
difficulty. Since both (u
x
, u
y
) and (u
x
', u
y
') have magnitude c it is not possible to form
a Euclidean triangle of velocities. This difficulty is overcome in the non-Euclidean
representation.

The aberration formula resulting from the substitution (5) in (2) may be written

c.cos ' = c cos + v
1 + cos . v/c (6)

and compounds the forward component of the light velocity c cos with v by the
composition rule for rectilinear velocities. So introducing rapidity components w, w ',
W by

cos = th w, cos ' = th w '
,
v = c th W (7)

(6) becomes

c cos ' = c th w + c th W = c th (w + W)
1 + th w. th W (8)






59 59 59

The simple addition

w ' = w + W (9)

of the x-components together with the fact that the y component of the light rapidity
remains infinite makes it possible to reconstruct the triangle of rapidities as shown in
the figure.
Fig: The triangle of rapidities in aberration.

Here the two sides representing velocity of light have become infinite, their
components in the direction of motion remaining finite. The light paths are both
Lobachevski parallels to the vertical line representing the transverse component of
light velocity. Angles ', are Lobachevski angles of parallelism as shown by the
equations (5). In this way the triangle addition of Bradleys theory is restored.








-----------------------------------------------------------------------------------------------
Note:
The use of Lobachevski parallels in light propagation was mentioned by Silberstein in his
1914 book but it did not appear in later literature.
60 60 60
CHAPTER 7 Applications to Dynamics


1. Hyperbolic Acceleration

Hyperbolic acceleration will be here defined as rate of change dV/d of hyperbolic
velocity V with respect to proper time , the time observed relative to the moving
body. In one dimension it coincides with the known expression for rectilinear
acceleration given e.g. by Rindler 1991:

= 1 dv (1)
(1-(v/c)
2
)
3/2
dt
= 1 1 dv = 1 d {th
[-1]
v/c} = dV
(1- v
2
/c
2
)
1/2
1 - v
2
/c
2
dt (1- v
2
/c
2
)
1/2
dt d (2)

Relative to the moving body v is zero and then from (2) the hyperbolic acceleration is
represented by the normal derivative expression. Hyperbolic acceleration dV/d has a
more familiar form than the expression (1) with its multiplying factor. It may also be
defined more intuitively as follows.

Consider the situation in the figure below representing an accelerating point P.

v v +v
0 |----------- P o ---------o P' --------------------------
x x +x x -axis

Fig: On the definition of rectilinear acceleration

In any time increment t, the increase in velocity v of an accelerating body does not
occur at a fixed value of x relative to an origin O but takes place over a certain
interval x to x+x. The increase of velocity relative the origin O should consequently
be calculated by the relative velocity formula as

(v+v)-v = v .
1 - (v+v).v/c
2
1 - (v+v).v/c
2
(3)

which gives, as t 0 and v 0, a first order increment of

v = c { th
[-1]
v/c } = V (4)
1 v
2
/c
2


Taking account of the time change from the origin to the frame of the moving particle,
the acceleration is found as the hyperbolic acceleration as defined above.

The three dimensional hyperbolic acceleration can be defined using Fock's definition
of relative velocity.





61 61 61
2. Motion under Constant Acceleration

The relativistic motion of a particle under constant acceleration was analysed in detail
by Born (1909) who called it 'hyperbolic motion' since the equation of the trajectory
in the x, t plane is a hyperbola instead of the parabola of classical physics. To find
the trajectory it is necessary to integrate the equation

2 2 3/ 2
dv
dt
(1 v / c )
=

(1)

for constant acceleration . This gives, if v = 0 when t=0,

2 2
v
t
(1 v / c )
=

(2)

On solving for v and again integrating, there is found

0
2
0 0
t t
t
x x vdt dt
(1 ( t / c) )

= =
+
(3)

which gives

2 2 2
0 0
x x (c / ) (1 ( t / c) ) 1 x t / 2 ...

= + + = +


(4)

from which the trajectory is found as the hyperbola

(x - x
0
+ b) - c t = b (5)

where b is the constant c/ .

* The Minkowski interpretation: The current standard interpretation of the motion
uses the ideas of Minkowski. The hyperbola is centralized by choosing x
0
equal to b
in which case.

x - c t = b (6)

and given the complex representation

x + (ict) = b (7)

In polar coordinates using an imaginary angle this is

x = b cos
ict = b sin (8)



62 62 62

Since b is constant, differentiation gives:

dx = - b sin d
ic dt = b cos d (9)

There follows

d = (dt - (dx /c)) = bi /c. d (10)

Consequently increases uniformly with giving the picture of constant angular
speed circular motion which Sommerfeld called 'cyclic motion'. It has a constant
(purely imaginary) central acceleration in the x, ict plane caused by the applied force.

* The Lobachevski interpretation: Here it is not necessary to make the same rather
artificial choice of initial condition but instead it is convenient to parametrize the
trajectory as

x = x
0
- b + b ch u, ct = b sh u (11)

from which

dx / dt = c th u (12)

so identifying u with rapidity. Further,
__________
d = {dx
2
(cdt)
2
} = b du (13)

again leading to the value of hyperbolic acceleration of

dV/d =c du/d = c
2
/ b = (14)

By integration is found an analogue of Galileos law for velocity increase as used in
elementary mechanics:

V = V
0
+ (15)







----------------------------------------------------------------------------------------------------
* References:
On hyperbolic motion see e.g. Pauli: Theory of Relativity, Prokhovnik: Theory of Relativity.
1967. Original references are: Minkowski Raum und Zeit 1908 (See p.84 in 'The Principle
of Relativity' translation which is annotated on p.96 with Sommerfeld's 1923 notes) Born
'Die Theorie des starren Elektrons ...' Ann. Phys. 1909, Sommerfeld Atombau und
Spektrallinien.1919 etc.320-.


63 63 63
3. Newtons Second Law for Rectilinear Motion

In 1906 Planck showed that Newton's second law of motion for rectilinear motion
can be written in the relativistic form

F = dp (1)
dt

F is applied force and p is momentum given by

p = mv = mv {1 + 1/6 (v/c)
2
+ ...} (2)
(1 v
2
/c
2
)

m being what is usually called the rest mass. Plancks equation may easily be
transformed to involve (hyperbolic) acceleration:

dp = m d v = m 1 dv = m (3)
dt dt (1-v
2
/c
2
) (1-v
2
/c
2
)
3/2
dt

This equation may also be written

dp = m dV (4)
dt d

Then Newton's law takes the familiar form

F = m = m dV (5)
d
In words,




(6)

This relation may be derived even more directly from the representation

p = mc sh V/c (7)

Differentiation gives:

F = dp = m ch V/c dV = m dV
dt dt d (8)

For constant force the acceleration is also constant and the motion is hyperbolic.



--------------------------------------------------------------------------------------------------
* References: See Planck (1906a) and (1906b). The form of Newtons Law given above was
proposed by the writer at the 2000 PIRT conference. It avoids the concept of velocity
dependent mass, velocity dependence being in the acceleration not the mass.
Force = Rest mass x (hyperbolic) acceleration
64 64 64
4. Remarks on Newton's Law of Motion in Three Dimensions

In his 1905 paper Einstein wrote the relativistic 3 dimensional Newtonian equations
of motion of a particle (an electron) in an electric field There were assumed inertial
systems, here denoted S and S', with S for the observer and S' for the particle. S'
moves with uniform velocity v relative to S along the x-axis and both S and S' are
coincident with the particle at t = 0 in position and S' also in velocity. The basic
assumption made is that Newton's law in its classical form applies in the frame S'
instantaneously coincident with the electron so that relative to the electron:

m d
2
x'/ dt '
2
= F'
x
= e E'
x

m d
2
y'/ dt '
2
= F'
y
= e E'
y

m d
2
z'/ dt '
2
= F'
z
= e E'
z


(1)

The accelerations in S' and in S are related by:

d
2
x'/ dt '
2
=
3
d2x/ dt
2
d
2
y'/ dt '
2
=
2
d
2
y/ dt
2
d
2
z'/ dt '
2
=
2
d
2
z/ dt
2
(2)

The electric fields are related by

E '
x
= E
x
E '
y
= (E
y
v/c H
z
)
E '
x
= (E
z
+ v/c H
z
) (3)

So from (1) it follows that to the observer in S the equations of motion are:

m
3
d
2
x/ dt
2
= F
x

m
2
d
2
y/ dt
2
= F
y
m
2
d
2
z/ dt
2
= F
z
(4)

where on the right hand side are the components of the Lorentz force:

F
x
= e E
x
F
y
= e (E
y
v/c H
z
)
F
z
= e (E
z
+ v/c H
z
) (5)

On the appearance of the factor in the lateral force components see Mller (1955)

* Plancks form of Newtons Law: Planck (1906) showed that for rectilinear motion,
Newton's law may be written relativistically in terms of rate of change of momentum.
Einstein (1907) rewrote the 3 dimensional equations in Planck's form and rederived
the equations of his 1905 paper. The 3 dimensional Planck-Newton equations are

[F
x
, F
y,
F
z
] = | dp
x
dp
y
dp
z
|
| dt dt dt | (6)





65 65 65

The components of vector momentum are

[p
x
, p
y
, p
z
] = | mv
x
, mv
y
, mv
z
. |
| (1 v
2
/c
2
) (1 v
2
/c
2
) (1 v
2
/c
2
) | (7)

Differentiating the right hand side of (6) gives the components as

d m v
x
= m (1-v
y
2
/c
2
-v
z
2
/c
2
) dv
x
+ m v
x
v
y
/c
2
dv
y
+ m v
x
v
z
/c
2
dv
z

dt (1-(v/c)
2
) (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt

d m v
y
= m v
y
v
x
/c
2
dv
x
+ m (1-v
z
2
/c
2
-v
x
2
/c
2
) dv
y
+ m v
y
v
z
/c
2
dv
z

dt (1-(v/c)
2
) (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt

d m v
z
= m v
z
v
x
/c
2
dv
x
+ m v
z
v
y
/c
2
dv
x
+ m (1-v
x
2
/c
2
-v
y
2
/c
2
) dv
z

dt (1-(v/c)
2
) (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt (1-(v/c)
2
)
3/2
dt

(8)
Applied to a particle with velocity components: v
x
= v, v
y
= 0, v
z
= 0 these give

F
x
= m 1 dv
x
= m
3
dv
x
(1 v
2
/c
2
)
3/2
dt dt
F
y
= m 1 dv
y
= m dv
y
(1 v
2
/c
2
) dt dt
F
z
= m 1 dv
z
= m dv
z
(1 v
2
/c
2
) dt dt (9)

On restoring the factor in the second and third equations there result equations
agreeing with those above

-------------------------------------------------------------------------------------------------------
* Note on variable mass:
Einstein (1905) made no mention of the interpretation of the left hand side of the equations
(4) as components of relativistic acceleration but interpreted these equations in terms of
velocity dependent mass. Dividing by he wrote the equations as

m
3
d
2
x/dt
2
= F
x

m d
2
y/dt
2
= F
y

m d
2
z/dt
2
= F
z


Then he interpreted second derivatives as components of acceleration in the classical way and
the multiplying factors m
3
and m as giving velocity dependent longitudinal and lateral
masses. In doing this Einstein was using the interpretation of Lorentz then current. The
notion of the variable mass of an electron had been introduced by Lorentz in 1895 and
modified by Abraham (1904) and Lorentz (1904c) to longitudinal and lateral masses.

------------------------------------------------------------------------------------------------------------
* References
Lorentz 1995 Versuch ; Abraham-Fppl 1904 'Lehrbuch der Elektrizitt ..' Lorentz
1904c. Planck 1906, Einstein Ann. Phys. 1905, Jahrb. Phys. 1907 ; Sommerfeld: 'Atombau
... ', 1933; Toulmin R.C. 'Note on the determination ', Phil. Mag. 21 1911 296-301. See
also the historical appendix,

66 66 66
5. Three-dimensional Particle Dynamics in Vector Form

The vector form of Plancks equation giving Newtons Law of Motion is

F = dp
dt (1)

F is applied force and p is the momentum vector

p = 1 (mv
x
, mv
y
, mv
z
)
(1 v
2
/c
2
) (2)

Following Einstein 1905, Planck's equation is usually discussed using the concept of
velocity dependent mass. Here we follow a method suggested by electron optics (see
e.g. Born & Wolf). The momentum is represented as

p = p n (3)

n is a unit vector in the direction of v and p denotes the scalar value:

p = mv .
(1 v
2
/c
2
) (4)

On differentiating the product Plancks equation becomes

F = dp n + p dn
dt dt (5)

The first term on the right, the tangential component, is in the direction of n while the
second term, the normal component, is perpendicular to n.

(a) The tangential component: This is

F.n = dp = m dv
dt {1 - v
2
/c
2
)
3/2
dt (6)

On the right, there is the same expression as in the rectilinear case. As a result is
deduced:





(7)








Tangential force =
Rest mass x tangential (hyperbolic) acceleration
67 67 67
(b) The normal component: this becomes more explicit by introducing differential arc
length ds when we can put

dn = dn ds = 1 n

v
dt ds dt (8)

Here is radius of curvature, and n

is the unit normal to the curve. Consequently the

resolution of the force into tangential and normal components is

F = dp.n + p v n

(9)
dt

The central force is

p v n

= mv
2
n

(10)


* Work and energy: Work is done solely by the tangential component of force and the
rate of doing work is

F.v = F
x
.v
x
+F
y
.v
y
+F
z
.v
z
= m v dv = dE (11)
(1 v
2
/c
2
) dt dt

It is seen to be equal to the rate of change of kinetic energy E

E = m c
2
.
(1 v
2
/c
2
) (12)

If a potential function V exists in the particular coordinate system used, then

F.dx = - dV (13)

and integration of (11) gives the energy equation

V + E = const. (14)


6. Calculation of Central Force in Uniform Circular Motion

In two dimensional circular motion referred to the centre

p = mc sh (V/c) (-sin , cos ) (1)

being the polar angle. If angular velocity is uniform,

dp = mc sh (V/c) (-cos , - sin ) d
dt dt (2)

This becomes, on transforming the time on the right hand side,



68 68 68
dp = mc th (V/c) (-cos ,-sin ) d = m v (-cos ,-sin ) d
dt d d (3)

Resembling the usual expression for the centrifugal force i.e. mv d/d along the unit
normal

Fig.1: The space diagram Fig.2: The velocity diagram
(Hodograph)

The same calculation may be done by using the velocity diagram (hodograph) which
in this case is a circle in the hyperbolic plane (fig.2). By the hyperbolic geometry of
the circle the increment V is related to incremental angle by

V = c sh (V/c) (4)

Then the central force can be calculated as mass times vector momentum change:

m dV = m d c sh (V/c) = m d c th (V/c) = m d v
dt dt. d d (5)

7. The Energy-Momentum Four-Vector

The energy-momentum four vector for a single particle of mass m and velocity v is
defined here as

[E/c, p
x
, p
y
, p
z
]
T
= | mc m v
x
mv
y
mv
z
|
| (1 - v
2
/c
2
) (1 - v
2
/c
2
) (1 - v
2
/c
2
) (1 - v
2
/c
2
) | (1)

Being a multiple of [cdt, dx, dy, dz]
T
it transforms the same way. Expressing the
components in terms of the unit vector n in the direction of the velocity there is found

E/c = mc ch w = m ch V/c
p
x
= mc sh w n
1
= m sh V/c n
1
= mc sh V/c sin cos

p
y
= mc sh w n
2
= m sh V/c n
2
= mc sh V/c sin sin

p
z
= mc sh w n
3
= m sh V/c n
3
= mc sh V/c cos (2)



69 69 69
The components satisfy identically the equation defining a hyperbolic surface
invariant under Lorentz transformation:

(E/c)
2
- (p
x
2
+p
y
2
+p
z
2
) = (mc)
2
(3)

In consequence the dimensionality is effectively three not four. On taking differentials
it follows that identically

(E/c) (dE/c) - (p
x
dp
x
+p
y
dp
y
+p
z
dp
z
) = 0 (4)

So that the differential vector

[dE/c, dp
x
, dp
y
, dp
z
]
T
(5)

is orthogonal to the energy-momentum vector and consequently space-like. The
spherical parametrization of the surface (4) leads to the surface element squared of

dp
x
2
+ dp
y
2
+ dp
z
2
d (E/c)
2
= (mc)
2
{dw
2
+ sh
2
w (d
2
+ sin
2
d
2
)}

= (m)
2
{dV
2
+ c
2
sh
2
(V/c) (d
2
+ sin
2
d
2
)} (6)

- the standard Riemannian form for a hyperbolic space. The expression on the right
corresponds to orthogonal incremental momentum components of m dV radially and
of mc sh V/c d transversally.


8. Energy-Momentum of a System of Particles

The energy - momentum vector of a system of particles is defined by summation over
all particles. The energy and momentum may be written

E/c = mc ch w
p = mc n sh w (1)

Superimposition of equations results in the same transformation equations as for a
single particle. Energy and momentum are connected by the identity:

(E/c)
2
- p.p = (Mc)
2
(2)

With the help of suffices, M is seen to have the value

M
2
= ( m
i
ch w
i
)
2
- ( m
i
n
i
sh w
i
) . ( m
j
n
j
sh w
j
)
= m
i
m
j
{ch w
i
ch w
j
- n
i
.n
j
sh w
i
sh w
j
}
= m
i
m
j
ch w
j / i
(3)

Here M can be thought of as the total mass-energy of the particles. Associated with M
a rapidity W can be defined by the equations

M ch W = m ch W
M n sh W = m n sh W (4)

70 70 70

Division gives

n th W = ( m
i
n
i
sh w
i
) / ( m
i
ch w
i
) (5)

So that

th
2
W = ( m
i
mj n
i
.n
j
sh w
i
sh w
j
)( m
i
ch w
i
)
2
(6)

The rapidity W here is similar to the classical velocity of the centre of mass.


9. Percussion

The equations for percussion were established by Lewis & Tolman 1908 in the
simple case of rectilinear collision. If the two particles have masses m
1,
m
2
and
velocities v
1
, v
2
the basic equations are

m
1
+ m
2
= m
1
+ m
2
.
(1-v
1
*
2
/c
2
) (1- v
2
*
2
/c
2
) (1-v
1
2
/c
2
) (1- v
2
2
/c
2
)

m
1
v
1
* + m
2
v
2
* = m
1
v
1
+ m
2
v
2
.
(1- v
1
*
2
/c
2
) (1- v
2
*
2
/c
2
) (1-v
1
2
/c
2
) (1-v
2
2
/c
2
) (1)

The asterisk denotes value after collision. The first equation is for the conservation of
mass equivalent of energy and the second equation is for conservation of momentum.

These equations must hold independently of the frame of reference so if we take these
equations to refer to inertial system S then they must also hold in a system S ' moving
uniformly relatively to it. This property is most conveniently established using the
hyperbolic representation when the conservation equations (1) become

m
1
ch w
1
* + m
2
ch w
2
* = m
1
ch w
1
+ m
2
ch w
2

m
1
sh w
1
* + m
2
sh w
2
* = m
1
sh w
1
+ m
2
sh w
2
(2)

In a system S ' moving with rapidity W relatively to S the equations will be

m
1
ch (w
1
* - W) + m
2
ch (w
2
* - W) = m
1
ch (w
1
- W) + m
2
ch (w
2
- W)
m
1
sh (w
1
* - W) + m
2
sh (w
2
* - W) = m
1
sh (w
1
- W)+ m
2
sh (w
2
- W) (3)

On expanding the hyperbolic functions these equations are seen to be equivalent to
the previous ones in (2). Further it can be seen that the validity of either of the
equations (3) with respect to arbitrary translations W (i.e. the relativity principle)
implies the validity of the other. It is convenient to write the relationships using the
two-dimensional energy-momentum vector relating mass-energy E and momentum p:

[E/c p ]
T
= [mc ch w mc shw ]
T
(4)





71 71 71

Moving from S to S ' expressed by

w = W + w' (5)

The energy-momentum vector obeys the usual Lorentz translation law, in matrix form

E / c chw shw E'/ c E'/ c chw shw E / c
p shw chw p' p' shw chw p

= =

(6)

By right matrix multiplication of the Lorentz translation w the equations

(E
1
*/c) + (E
2
*/c) = (E
1
/c) + (E
2
/c)
p
1
* + p
2
* = p
1
+ p
2
(7)

then immediately transform to the corresponding form in system S '.

* Three-dimensional percussion: equations for conservation of mass-energy in normal
form are:

m
1
+ m
2
= m
1
+ m
2

(1-v
1
* ) (1-v
2
* ) (1-v
1
) (1-v
2
) (8)

For conservation of momentum components they are:

m
1
v
x1
* + m
2
v
x2
* = m
1
v
x1
+ m

v
x2

(1- v
1
*
2
/c
2
) (1- v
2
*
2
/c
2
) (1-v
1
2
/c
2
) (1-v
2
2
/c
2
)

m
1
v
y1
* + m
2
v
y2
* = m
1
v
y1
+ m
2
v
y2
.
(1- v
1
*
2
/c
2
) (1- v
2
*
2
/c
2
) (1-v
1
2
/c
2
) (1-v
2
2
/c
2
)

m
1
v
z1
* + m
2
v
z2
* = m
1
v
z1
+ m
2
v
z2

(1- v
1
*
2
/c
2
) (1- v
2
*
2
/c
2
) (1-v
1
2
/c
2
) (1-v
2
2
/c
2
) (9)

These can be written concisely as

(E
1
*/c) + (E
2
*/c) = (E
1
/c) + (E
2
/c)
p
1
* n
1
* + p
2
* n
2
* = p
1
n
1
+ p
2
n
2
(10)

Here p is scalar momentum and n is the unit vector in the direction of the velocity










72 72 72
CHAPTER 9 Differential Minkowski Space and
Light Propagation


1. Differential Minkowski Space

The use of the differential form of the Lorentz transformation leads naturally to
the concept of the differential Minkowski space of four-vectors (dt, dx, dy, dz)
where all the ideas of normal Minkowski space apply to these differential vectors.
Thus differential vectors are classified as time-like or space-like according to
whether or not they lie in the differential Monge cone

c
2
dt
2
= dx
2
+dy
2
+dz
2
(1)

Only time-like vectors have significance for physical motion because the
requirement

(dx/dt)
2
+ (dy/dt)
2
+ (dz/dt)
2

< c
2
(2)

that the velocity should not exceed c, implies that

dx
2
+dy
2
+dz
2
< c
2
dt
2
(3)

So the differential vector (dt, dx, dy, dz) lies within the Monge cone. Usually it is
the forward light-cone which is of most interest characterized by this last condition
with dt > 0.

The Monge cone and the system of associated hyperbolic surfaces remain invariant
relative to differential changes brought about by any homogeneous Lorentz trans-
formation.

(cdt', dx', dy', dz')

= L (cdt, dx, dy, dz) (4)

The use of the differential form here allows the variables t', x', y', z' to be related to
t, x, y, z by nonhomogeneous as well as homogeneous Lorentz transformations.
This possibility is exactly in accord with Minkowskis original 1908 concept of the
absolute world as consisting of world vectors (t, x, y, z) invariant under the group
of non-homogeneous Lorentz transformations. However the form of space he
defined, i.e. the usual Minkowski space, is only invariant for homogeneous
transformations corresponding to the chosen fixed origin. Differential Minkowski
space gives more flexiblity permitting invariance also with respect to translations.
It is consequently more in accord with Minkowski's conception of the absolute
world.







73 73 73

2. The Cayley-Klein Metric in Differential Minkowski Space

In his Theory of Relativity which first appeared in 1921, Pauli briefly observed that
Variaks results could be derived from the Cayley-Klein theory of projective
geometry but neither he nor others followed up this idea which will be described
here in further detail

The space of differential vectors written in either of the two forms

(dt, dx, dy, dz), (c dt, dx, dy, dz) (1)

gives rise to a projective space with the differential Monge cone

c
2
dt
2
dx
2
dy
2
dz
2
= 0 (2)

as absolute. Since this locus is a real conic, the resulting projective space is
hyperbolic. Vectors in this space lie within the Monge cone representing physically
feasible motions satisfying the condition

dx
2
+ dy
2
+ dz
2

< c
2
d t
2
(3)

The Cayley-Klein projective distance between two differential vectors having
suffices 1, 2 is

ch
[-1]
(c
2
dt
1
dt
2
-dx
1
dx
2
-dy
1
dy
2
-dz
1
dz
2
) .
(c
2
dt
1
2
-dx
1
2
-dy
1
2
-dz
1
2
) (c
2
dt
2
2
-dx
2
2
-dy
2
2
-dz
2
2
) (4)

which is more conveniently written using vectors for the space part as


ch
[-1]
c
2
dt
1
dt
2
dr
1
.dr
2
.
(c
2
dt
1
2
dr
1
.dr
1
) (c
2
dt
2
2
dr
2
.dr
2
) (5)

On dividing through by dt
1
dt
2
and denoting velocities dr/ dt by v and we see that
the distance so defined is just the relative rapidity w of the two velocities v
1
, v
2

given by

ch w = c
2
v
1
.v
2
.
(c
2
v
1
.v
1
) (c
2
v
2
.v
2
) (6)

so identifying the hyperbolic space as the Beltrami space of these velocity vectors.

In this space the Beltrami parametric representation

dx/dt = v
x
= c th w n
1
= c th w sin cos
dy/dt = v
y
= c th w n
2
= c th w sin sin
dz/dt = v
z
= c th w n
3
= c th w cos (7)



74 74 74

can be viewed in a corresponding homogeneous form arising from

(c dt, dx, dy, dz) = const. (ch w, sh w n
1
, sh w n
2
, sh w n
3
) (8)

It implies

c
2
dt
2
- dx
2
- dy
2
- dz
2
= (const)
2
(9)

identifying the constant multiplier as c d. Thus

c dt = c d ch w = c d ch w
dx = c d sh w n
1
= c d sh w sin cos
dy = c d sh w n
2
= c d sh w sin sin
dz = c d sh w n
3
= c d sh w cos (10)

The parametrization (10) may also be written as giving the components of the
Minkowski four-velocity

V
0
= c dt/d = c ch w = c ch w
V
1
= dx/d = c sh w n
1
= c sh w sin cos
V
2
= dy/d = c sh w n
2
= c sh w sin sin
V
3
= dz/d = c sh w n
3
= c sh w cos (11)

Note that the space-time four-dimensionality arises from the use of homogeneous
coordinates for three-dimensional motion.

* Historical comment: In view of its importance for the history of the present subject the
remarks of Pauli are here quoted in full. In a footnote (The Principle of Relativity, p.74 in
Engl. trans.) he says:

Variak establishes a formal connexion between the Lorentz transformation, as
well as the relativistic formulae for the Doppler effect, aberration of light, and
reflection in a moving mirror, with the Bolyai-Lobachevski geometry This
connexion with the Bolyai-Lobachevski geometry can be briefly described in the
following way (this had not been noticed by Variak): if one interprets dx
1
, dx
2
,
dx
3
, dx
4
as homogeneous coordinates in a three dimensional projective space,
then the invariance of the equation dx
1
2
+dx
2
2
+dx
3
2
- dx
4
2
= 0 amounts to
introducing a Cayley system of measurement based on a real conic section. The
rest follows from the well known arguments of Klein (Math.Ann. 4 1871 12)

Pauli only allows a formal connexion and not a real connexion! As regards the
reference to Klein, it is remarkable that Klein himself, the authority on both projective
non-Euclidean geometry and relativity, had managed to miss the precise relation between
the two, even in his 1910 paper: ber die geometrische Grundlagen der Lorentz-gruppe
devoted to exactly this question. The explanantion appears to be that, as shown both in
Klein's paper and in Pauli's comment above, at that time it was customary to think in terms
of the pseudo-Euclidean space-like metric which does not lend itself to the Cayley metric.
The only place where the writer has found Klein using the appropriate time-like form is in
a footnote of his book Die Entwicklung . (p.131, vol.II) where he quoted the formula for
Cayley-Klein distance for special relativity though using the Minkowski imaginary angle
The reversed Cauchy inequality, the justification for this formula, was not mentioned.


75 75 75

3. The Light Cone Condition and Coordinate Invariance.

Suppose that a transformation of frames of reference is made resulting in the relation

[c dt', dx', dy', dz' ]
T
= [c dt, dx, dy, dz ]
T
(1)

being a general Lorentz transformation. From this follows

c
2
dt'
2
- dx'
2
- dy'
2
- dz'
2
= (c
2
dt
2
- dx
2
- dy
2
- dz
2
) (2)

So the Monge cones in the original and transformed spaces and their interiors
correspond to each other (causality condition). More generally the same conclusion
follows if there is a relation of the the type

c
2
dt'
2
- dx'
2
- dy'
2
- dz'
2
=
2
(c
2
dt
2
- dx
2
- dy
2
dz
2
) (3)

where is any non-zero scalar multiplier. Such a relation may come about from the
scalar multiplied form of the Lorentz transformation (e.g. the Voigt transformation) or
from a nonlinear transformation of coordinates and time. Condition (3), fundamental
to the theory; will be referred to as the light-cone condition. The following simple
fact is basic.

PROPOSITION: The light-cone condition implies absolute invariance of the Cayley-
Klein metric.

Proof: Any linear combination of two time-like differential vectors:

(c dt, dx, dy, dz) = (c dt
1
, dx
1
, dy
1
,

dz
1
) +

(c dt
2,
dx
2
, dy
2
, dz
2
) (4)

with , > 0 lies on the segment joining the two points with suffices 1 and 2 and so
from the convexity of the Monge cone is also time-like. Under linear transformation
of differentials it transforms to the vector

(c dt', dx', dy', dz') = (c dt'
1
, dx'
1
, dy'
1
, dz'
1
) +

(c dt'
2
, dx'
2
, dy'
2
, dz'
2
) (5)

This similarly lies within the Monge cone in the transform space and is time-like.
Applying the light-cone condition equating coefficients of
2
, and
2
we get

c
2
dt'
1
2
dx'
1
2
dy'
1
2
dz'
1
2
=
2
(c
2
dt
1
2
dx
1
2
dy
1
2
dz
1
2
)
c
2
dt'
1
dt'
2
dx'
1
dx'
2
dy'
1
dy'
2
dz'
1
dz'
2
=
2
(c
2
dt
1
dt
2
dx
1
dx
2
dy
1
dy
2
dz
1
dz
2
)
c
2
dt'
2
2
dx'
2
2
dy'
2
2
dz'
2
2
=
2
(c
2
dt
2
2
dx
2
2
dy
2
2
dz
2
2
)
(6)

From which follow invariance of the ratio giving Cayley-Klein metric and rapidity.
The hyperbolic spaces in the original and transformed light-cones are correspondingly
mapped on to one another isometrically. Rapidity and hyperbolic velocity are seen to
be independent of the coordinate system under any differentiable coordinate
transformation. Rapidity is an absolute invariant of the transformation and defines the
hyperbolic geometry in Beltrami representation as described previously.


76 76 76
4. Normal Light Propagation

In 1924 Carathodory published a general derivation of the light cone condition
starting from an initial set of axioms relating to light propagation. The background to
the paper is that in 1908, 1910 Bateman and Cunningham had discovered that the
equation of the wave surface as also Maxwell's equations were invariant under certain
nonlinear transformations related to inversion which they called 'spherical wave
transformations'. The papers of Cunningham and Bateman thus raised the question of
the exact nature of the group of transformations under which Maxwell's equations
remain invariant and the relation of this group with Special Relativity. This problem,
ascribed to Pauli by Carathodory, was investigated in the paper.

Out of the initial axiomatic discussion Carathodory established the existence of
differentiable equations relating two systems of locally Euclidean coordinates and
time of the form:

x' = X(x,y,z,t), y' = Y(x,y,z,t), z' = Z(x,y,z,t), t' = T(x,y,z,t) (1)

From these follow linear relations between corresponding differentials

[ dx', dy', dz', dt' ]
T
= [ J ] [dx, dy, dz, dt ]
T
(2)

[ J ] is the Jacobian matrix. Using these linear relations we may express as a quadratic
in terms of dx, dy, dz, dt the quantity

dx'
2
+ dy'
2
+ dz'
2
- c
2
dt'
2
(3)

For normal light propagation to be preserved by the transformation there must be
satisfied a condition:

dx'
2
+ dy'
2
+ dz'
2
- c
2
dt'
2
= (x, y, z, t) (dx
2
+ dy
2
+ dz
2
- c
2
dt
2
) (4)

- a similar condition to that previous found by Bateman in connexion with Maxwell's
equations. It resembles the previous light cone condition but the multiplier (which
should be positive) is a function of position and time instead of v. The condition is
fundamental to Carathodorys theory.

* Huyghens wavelets: The condition (4) can be given a physical interpretation for it
implies that the infinitesimal spherical wave (a Huyghens wavelet)

dx
2
+ dy
2
+ dz
2
= c
2
dt
2
(5)

transforms into a similar wavelet. These wavelets generate the wavefront which
propagates according to equation of the wave-equation

(V/x)
2
+ (V/y)
2
+ (V/z)
2
- 1/c
2
(V/t)
2
= 0 (6)

The light cone condition implies that this equation is also invariant under the
transformation considered. This is because the dual equation of the Monge cone,
giving the condition that a hyperplane with homogeneous coordinates (n
0
, n
1
, n
2
, n
3
) is
tangential to the cone, is

77 77 77
n
1
2
+ n
2
2
+ n
3
2
- (n
0
/c)
2
= 0 (7)

So the wave equation for V represents the condition that the vector

[V/x, V/y, V/z, V/t] (8)

is tangential to the Monge cone. On transforming to new coordinates, there will be a
condition dual to the light-cone condition:

(n
1
'

2
+ n
2
'

2
+ n
3
'

2
(n
0
'

/c)
2
) = (x, y, z, t)
1
(n
1
2
+ n
2
2
+ n
3
2
(n
0
/c)
2
) (9)

From this it immediately follows that the wave equation is satisfied in the new
coordinates.

Carathodory proved the invariance of the wavefront equation using the theory of
characteristics for a partial differential equation and showed that any piecewise linear
light path becomes transformed into another piecewise linear light path.

------------------------------------------------------------------------------------
References: See Carathodory: On the axiomatics of special relativity (German) Preuss.
Akad Wiss. 1923 reproduced in his collected works. There is also a shorter version in German
in his collected works which is the translation of an encyclopedia article published in Greek.
The papers of Bateman and Cunningham are listed in the bibliography.


5. Conformal Transformation in Four Dimensions

With l = ict, l' = ict' the light-cone condition can be written as

dx'
2
+ dy'
2
+ dz'
2
+ dl'
2
= (x, y, z, t) (dx
2
+ dy
2
+ dz
2
+ dl
2
) (1)

This is the condition for a conformal transformation in 4 dimensions. One such
conformal transformation is the Lorentz transformation. Another is the inversion:

x' = x / (x
2
+y
2
+z
2
+l
2
)
y' = y / (x
2
+y
2
+z
2
+l
2
)
z' = z / (x
2
+y
2
+z
2
+l
2
)
l' = l / (x
2
+y
2
+z
2
+l
2
) (2)

When translated back to x, y, z, t variables this coincides with one of the spherical
wave transformations of Bateman & Cunningham. The importance of inversion is
seen from the following result:

THEOREM (Liouville 1847): Every conformal transformation in a space of more than
2 dimensions is representable by a sequence of similarity transformations and
inversions.

The similarity transformations are those generated by orthogonal transformations and
translations. A simplified proof of the theorem was given by Carathodory in his 1923
paper.


78 78 78
Concerning the application of the spherical wave transformations to relativity, an
important example is the following which arose in Batemans 1910 paper

EXAMPLE Consider the infinitesimal transformation

t' = t (1 + f x)
x' = x (1 + f x) + f (-x
2
+ c
2
t
2
- y
2
- z
2
)
y' = y (1 + f x)
z' = z (1 + f x) (3)

f is here considered small so that only terms of the first order in f need be retained.
The conformal condition is satisfied with

(x,y,z,t) = 1 + 2 f x (4)

On substituting for t from the first equation into the second we find

x' = x + f (x
2
+ t'
2
y
2
z
2
) (5)

With x, y, z fixed, the point with coordinate x' moves with constant acceleration f. As
Bateman observed, the relation between t and t' agrees with that given by Einstein
(1907) in his attempt to extend the Special Theory to an accelerated system.

This is an example of a non-Galilean transformation which nevertheless satisfies the
condition of normal light propagation. Carathodory considered that such
transformations should be excluded from the invariance group of Special Relativity
since he was understanding in his paper the principle of Special Relativity to include
the principle of mechanical relativity, i.e. that all phenomena take place in an inertial
frame. As a result the conclusion of Carathodory is that the invariance group of
Special Relativity is the non-homogeneous Lorentz group with scalar multiplication

However, since those nonlinear transformations obey the light cone condition, it
follows from above that the Cayley-Klein metric defining an invariant rapidity can be
introduced leading to the Beltrami interpretation together with its consequences e.g.
all those which follow from the composition of velocities. The writers opinion is that
in view of this the status of the nonlinear transformations in the special theory must
remain an open and important question.









-----------------------------------------------------------------------------------------------
Reference: For further comments see the writer's paper at the congress 'Constantin
Carathodory ' Orestiada, Greece, Sept. 2000.

79 79 79
GENERAL BIBLIOGRAPHY

References

Aczl Ya: Some general methods in the theory of functional equations in one variable,
Usp. Mat. Nauk 11 1956 3-68.

Arzelis H: La cinmatique relativiste. Paris 1955 (Gauthier-Villars)
Engl.tr Relativistic Kinematics. Oxford 1966 (Pergamon)

Barrett J.F: Logarithmic redshift measure and a relativistic Hubble law.
PIRT Conf., Imperial Coll. London, Sept. 1992; Proceedings, vol.2, pp.1-8

Barrett J.F: Hyperbolic geometry representation of velocity space in special relativity
and Cayley-Klein metric. PIRT Conf. London, Sept. 1996; Proceedings
(Supplementary papers) pp. 28-31

Barrett J.F: Special relativity and hyperbolic geometry
PIRT Conf., London, Sept. 1998; Proceedings (Late papers), pp.18-31, ISBN: 1 873
757 530.

Barrett J.F: Relativity by conformal transformation and hyperbolic geometry.
PIRT Conf. London, Sept. 2000; Proceedings (Late papers): Univ. Liverpool (PD
Publications), ISBN: 1-873 694-05-9, pp.16-25,
http://www.physicsfoundations.org/PIRT_VII/texts/barrett.doc

Barrett J.F: On Carathodory's approach to relativity and its relation to hyperbolic
geometry. Congress 'Constantin Carathodory in his .. origins.' Orestiada, Greece,
Sept. 2000. Proceedings: Palm Harbor FL 2001 (Hadronic Press) pp.81-90, ISBN 1-
57485-053-9

Barrett J.F: Aberration and the luminosity-distance formula.
PIRT Conf., London, Sept. 2000; Proceedings (Late papers)
Univ. Liverpool (PD Publications) ISBN: 1-873-694-05-9, pp.26-32.

Barrett J.F: Hyperbolic geometry in special relativity and its relation to the cosmology
of Milne. pp. 27-33 in vol.1 of : M.Wegener & M.C. Duffy (eds.) 'Recent Advances in
Relativity Theory', Palm Harbour, FL 2001, (Hadronic Press) ISBN 1-57485-047-4.

Barrett J.F: Review of problems of dynamics in the hyperbolic theory of special
relativity. PIRT Conf., London, Sept 2002, pp.17-30 in the Proceedings, Univ.
Liverpool (PD Publications) ISBN 1 873 694 07 5.
http://www.physicsfoundations.org/PIRT_VIII/texts/Barrett.doc

Barrett J.F: Relative motion and hyperbolic geometry in special relativity.
PIRT Conf. London, Sept. 2004 pp. 14-24 in Proceedings II, Univ. Liverpool (PD
Publications) ISBN: 1 873 694 09 1,
http://www.physicsfoundations.org/PIRT_IX/texts/barrett.doc



80 80 80
Barrett J.F: Using relative velocities and hyperbolic geometry in special relativity
PIRT Conf. Budapest, Sept 2007,
http://www.phil-inst.hu/~szekely/PIRT_Budapest/ft/Barrett_ft.pdf

Bateman H: The transformation of the electrodynamical equations
Proc. LMS 8 1910 223-264.

Bellman R: On an inequality concerning an indefinite form.
Amer. Math. Monthly, 63 1956 108-109.

Ben-Menahem A. Wigner's rotation revisited.
Amer. J. Phys. 53 1985 62-65.

Belloni L.& Reina C: Sommerfeld's way to the Thomas precession.
Eur. J. Phys. 7 1986 55-61.

Birkhoff G.D. & Langer R.E: Relativity and Modern Physics
Cambridge, Mass. 1927 (Harvard Univ. Press)

Blumenthal O. (ed) Das Relativittsprinzip Leipzig 19l3, 1923 (Teubner), Engl. tr.
(W.Perrett & G.B. Jeffrey) The Principle of Relativity, London 1923 (Methuen); rpr.
New York 1952 (Dover)

Bondi H. Assumption and Myth in Physical Theory
Cambridge 1967 (Univ. Press)

Borel E. La thorie de la relativit et la cinmatique.
C.R. Acad. Sci. Paris, 156 1913 215-218.

Borel E. La cinmatique dans la thorie de la relativit.
C.R. Acad. Sci. Paris, 157 1913 703-705.

Borel E. Introduction geomtrique a quelques thories physiques.
Paris 1914 (Gauthier-Villars)

Born M. & Wolf E Principles of Optics.
Oxford 1959, 1965 (Pergamon)

Carathodory C.C. Zur Axiomatik der Speziellen Relativittstheorie.
Sitzb. Preuss. Acad. Wiss. 1924 12-27; Ges Math Schr. II 403-413.

Carathodory C.C. 'Xoroxronos' Megali Elliniki Encyclopedia, Athens 1934
(Eleftheroudakis); reprint in E. Spandagou ed. K.Karatheodori Athens 2000 (Aithra);
German transl.by Stephanos Carathodory: Der moderne Raum-Zeit-Begriff; Ges.
Math. Schr. 403-413.

Compte E: Relativity before relativity.
Proc. PIRT Conf. Sept 1998.

Cunningham E: The principle of relativity in electrodynamics.
Proc. LMS 8 1910 77-98.

81 81 81
Einstein A. Zur Elektrodynamik bewegte Krper.
Ann. Physik 17 1905 891- 921; Coll. Papers II 275-306.
Engl.tr. 'On the electrodynamics of moving bodies' in 'The Principle of Relativity'
Miller 1998.and Coll.Papers (Engl. version trans. by Anna Beck)

Einstein A. ber die Relativittsprinzip und die aus demselben gezogenen
Folgerungen. Jahrb. Radioaktivitat 4 1907 411-462; Coll. Papers II, 433-484.

Einstein A. ber die spezielle und die allgemeine Relativittstheorie.
Braunschweig 1917 (Vieweg) Engl.tr 'Relativity, the Special and the General Theory',
New York 1920 (Holt)

Einstein A. The Meaning of Relativity (1921 Princeton Lectures.);
Princeton 1921; London 1922 (Methuen)

Einstein A: The Collected Papers of Albert Einstein, Vol. 2. The Swiss Years:
Writings, 1900-1909, J. Stachel, ed. (Princeton: Princeton University Press, 1989).
Engl.tr. (Anna Beck) Princeton Univ.Press 1989 ISBN 0-691-08549-8

Fock V.A: Teoriya prostranstva, vremeni i tyagoteniya. Moscow 1955 (Gosizdat
fysmatlit.); Engl.tr 'The Theory of Space, Time and Gravitation', Oxford 1959
(Pergamon)

Goldstein H: Classical Mechanics
Reading Mass. etc 1980 (Addison-Wesley)

Herglotz G: ber die Mechanik der deformierbaren Krpers vom Standpunkte der
Relativittstheorie, Ann. Phys. 36 1911 493-533

Hestenes D Space-time Algebra
New York 1966 (Gordon & Breach)

Jnossy L: Theory of Relativity based on Physical Reality.
Budapest 1971 (Akademiai Kiado)

Karapetoff V: Special Theory of Relativity in hyperbolic functions.
Rev. Mod. Phys. 16 1944 33-52.

Klein F: ber die sogenannte nichteuklidische Geometrie.
Math Ann. 4 1871 573-625; Ges Math Abh I 254-; revised version: ibid. 6 1873 112-
145.

Klein F. Nicht-Euklidische Geometrie
Rosemann (ed.) Berlin 1923, 1928 (Springer); rpr New York n.d. (Chelsea)
[From lectures given in 1892]

Klein F. ber die geometrische Grundlagen der Lorentz-gruppe.
Jahrb. dtsch. Math. Verein. 19 1910 533-552; Phys. Z. 12 1911 17-27.



82 82 82
Klein F. Vorlesungen ber die Entwicklung der Mathematik im neunzehnten
Jahrhundert. Berlin 1926 (Springer) 2 vols; rpr New York 1956 (Chelsea) In 1 vol

Kurepa D: First centenary of the birth of mathematician Vladimir Variak.
Informations scientifiques, Univ. Belgrade 1965, 61-67. (In Serbo-Croat)

Le Roux J: Relativit restreinte et gometrie des systmes ondulatoires.
J de Math. 1 1922 205-253 esp 232-246; Paris 1922 (Gauthier-Villars).

Lewis G.N: A revision of the fundamental laws of matter and energy.
Phil.Mag. 16 1908 705-717.

Lewis G.N. & Tolman R.C: The principle of relativity and non-Newtonian mechanics.
Phil.Mag. 18 1909 510-523.

Minkowski H. Das Relativittsprinzip. Lecture to Math. Gesellschaft. Gttingen
Nov 1907. Published posthumously in Ann. Phys. 47 1915 927-938; Jahrb. dtsch.
Math. Verein. 24 1915 372-.

Minkowski H Die Grundgleichungen fr die elektromagnetischen Vorgnge in
bewegten Krpern. Gtt Nachr 1908 53-111, Math Ann 68 1910 472-525; Ges Abh
II 352-404

Minkowski H: Raum und Zeit (Address at 80th Assembly of German Natural
Scientists and Physicists), Cologne, Sept 1908; Phys. Z. 10 1909 104-111; Jahrb.
dtsch Math Verein 18 1909 75-88; Ges Abh II 431-446; Engl.tr. 'Space and Time'
reprint. in The Principle of Relativity.

Minkowski H Zwei Abhandlungen ber die Grundgleichungen der Elektrodynamik,
Max Born (ed.) Leipzig 1910 (Teubner) (rpr. from Math. Ann. 68 1910)

Minkowski H Gesammelte Abhandlungen (ed. A.Speiser, H.Weyl, D.Hilbert) 2 vols,
Leipzig-Berlin 1911 (Teubner); rpr. New York 1967 (Chelsea) 2 vols in 1

Milne A.E. Relativity, Gravitation and World Structure
Oxford 1935 (Clarendon Press)

Mocanu C. On the relativistic velocity composition paradox and the Thomas rotation
Foundations of Physics Letters, 5(5), 1992

Mller C. The Theory of Relativity,
New York 1952, 1955, 1972 (Oxford Univ. Press)

Ogura K. On the Lorentz transformation with some geometrical interpretations
Sci. Rep. Tohuku Imp. Univ., 2 1913 95-116.

Pathria R.K. Resultant momentum and kinetic energy in relativistic mechanics
Amer J Phys. 24 1956 47-48; 411-413.




83 83 83

Pathria R.K. Relativity
Dehli 1963, 1974 (Hindustan Pub. Corp.)

Pauli W. Relativittstheorie, Enz. math. Wiss. V.2(4) 19, Leipzig 1921 (Teubner);
Engl. tr. 'Theory of Relativity' Oxford 1958 (Pergamon)

Planck. M. Das Prinzip der Relativitt und die Grundgleichungen der Mechanik.
Verh. dtsch. Phys. Ges 8 1906 136-141.

Planck M Zur Dynamik bewegter Systeme.
Sitzb. Akad. Wiss. Berlin 13 1907 542-570; Ann. Physik 26 1908 1-34.

Poincar H. Sur la dynamique de l'electron.
C.R. Acad. Sci. Paris 140 1905 1504-1508, Oeuvres IX 489-493.

Poincar H. Sur la dynamique de l'electron.
Rend. Circ. mat. Palermo 21 1906 129-176, Oeuvres IX 494-550.
Engl. tr. (Keswani G.H. & Kilminster C.W.) Brit. J. Phil. Sci. 34 1983 343-.

Poincar H. Oeuvres de Henri Poincar, IX (ed. G.Darboux)
Paris 1916-1956 (Gauthier-Villars)

Prokhovnik S.J: The Logic of Special Relativity. Cambridge 1967 (Univ. Press)

Rindler W: Introduction to Special Relativity.
Oxford 1989 1991 (Univ. Press)

Robb A.A: A Theory of Space and Time
Cambridge 1910 (Heffers)

Robb A.A: Optical Geometry of Motion, a New View of the Theory of Relativity
Cambridge 1911, (Heffers)

Robb A.A: The Geometry of Time and Space.
Cambridge 1936 (Univ. Press)

Rosenfeld B.A: Istoriya neevlidvoi Geometri, Moscow 1976, (Nauka)
Engl. trans. A History of Non-Euclidean Geometry New York etc. 1988 (Springer)

Scott-Walter: The non-Euclidean style of Minkowskian relativity
pp. 91-127 The Symbolic Universe, J Gray (ed.), Oxford 1999 (Univ. Press)

Silberstein L: The Theory of Relativity.
London 1914, 1924 (MacMillan)

Smorodinski Ya A: Parallel transfer of spin in Lobachevski space
Soviet Physics 16(6) Jun 1963 1566-1570. Russian in JETP (USSR) 43 Dec 1962
2217-2223:



84 84 84
Smorodinski Ya A: Kinematika stolknovenii v geometriscom izlozhenia
(Kinematics of collisions presented geometrically), Proc. Conf. 'Questions of the
physics of elementary particles', Yerevan 1963 (Izdat. A.N. Arm. SSR) 242-271.

Smorodinski Ya A: The theory of spiral amplitudes.
Soviet Physics 18(2) Feb 1964 415-418. Russian in JETP 1963 604-609

Sommerfeld A. ber die Zusammensetzung der Geschwindigkeiten in der
Relativtheorie, Phys. Z. 10 1909 826-829; Ges. Schr. II 185-188.

Sommerfeld A. Zur Relativittstheorie, Ann. Phys. 1910,
I Vierdimensionale Vectoralgebra. 32 749-776
II Vierdimensionale Vektoranalyse. 33 649-689

Sommerfeld A. Vereinfachte Ableitung der Thomasfactors.
Convegno di Fisica Nucleare, Rome 1931; Proc 1932 (Atti Reale Accad.) 137-141

Sommerfeld A. Atombau und Spektrallinien, Braunschweig 1919, 1921, 1923
(Vieweg); Engl.trans: 'Atomic Structure and Spectral Lines' London 1923
(Methuen), rpr 1952 (Academic Press).

Sommerfeld A. Vorlesungen ber theoretischen Physik. vol.3 Elektrodynamik.
Engl. tr. Lectures on Theoretical Physics, vol.3 New York 1946, 1952 (Academic
Press)

Sommerfeld A. Gesammelte Schriften. 4 vols
Braunschweig 1948 (Vieweg)

Stachel J.J: History of Relativity Theory.
L.M. Brown et al. (eds.) Twentieth Century Physics, New York 1995 (Amer. Inst.
Phys) I 245-356.

Stevenson G. & Kilminster: Special Relativity for Physicists.
London 1958 (Longman-Green)

Stillwell J: Sources of Hyperbolic Geometry.
1996 Amer Math Soc & Lond Math Soc

Thomas L.H. The motion of a spinning electron
Nature 1926 514

Thomas L.H: The kinematics of an electron with an axis
Phil. Mag. 7 1927 1-23.

Tolman R.C. Note on the determination from the Principle of Relativity of the fifth
fundamental equation of the Maxwell theory. Phil. Mag. 21 1911 296-301.

Tolman R.C: Non-Newtonian mechanics, the direction of force and acceleration.
Phil. Mag. 22 1911 458-463.



85 85 85

Tolman R.C: Non-Newtonian mechanics, the mass of a moving body.
Phil. Mag. 23 1912 375-380.

Ungar A.A: Thomas rotation and the parametrization of the Lorentz group.
Found. Physics Lett 1 1988 57-89.

Ungar A.A: www.ndsu.edu/pubweb/~ungar/publications.html
(Complete list of publications)

Ungar A.A: Beyond Einsteins Addition Law and its Gyroscopic Thomas Precession.
The Theory of Gyrogroups and Gyrovector Spaces, Dordrecht 2001 (Kluwer Acad.);
Reviewed by Scott Walker in Found. Phys. 33 2002 327-330.

Variak V. Anwendung der Lobatschefkijschen Geometrie in der Relativtheorie.
Phys. Z. 11 1910 93-96.

Variak V. Die Relativtheorie und die Lobatschefskijsche Geometrie.
ibid. 287-293.

Variak V. ber die nichteuklidische Interpretation der Relativtheorie.
Jahrb. dtsch. math. Verein 21 1912 103-127.

Variak V. Darstellung der Relativittstheorie im dreidimensionalen
Lobatschefskijschen Rume, Zagreb 1924

Wegener M. & Duffy M.C.(eds.) Recent Advances in Relativity Theory, 2 vols
Palm Harbour FL 2001 (Hadronic Press) ISBN 1-57485-047-4, 1-57485-050-4

Weyl H. Raum, Zeit, Materie. Berlin 1918 (Springer)
Engl.tr. Space, Time, Matter, from 4th ed., London 1922 (Methuen); from 5
th
German
ed. New York 1952 (Dover)

Whittaker E.T. A History of the Theories of the Aether & Electricity
Edinburgh, (vol 1), London 1910 (Nelson), revised and enlarged edition 1951-3:
(Nelson) vol.I: The Classical Theories 1951, vol. II: The Modern Theories 1953;
reprint New York 1953 (Harper); New York 1973 (Humanities Press)

Wick G.C. Angular momentum states for three relativistic particles,
Annals Phys. 18 1962 65-80.

Wigner E: Relativistic invariance and quantum phenomena.
Rev. Mod.Phys. 29 1957 255-268.

Whitrow G.J: On the Lobatschewskian trigonometry of a static substratum.
Quart J Math 10 1939 313-

Woodhouse N.M.J. Special Relativity
London etc. 1949, 2003 (Springer) ISBN 1-85233-426-6

Wyk C. B.van Rotation associated with the product of two Lorentz transformations.
Amer. J. Phys. 92, 1984, 853-854.
86 86 86
APPENDIX 1 Some Historical Notes


1. Galilean Relativity and Newtonian Mechanics

The origin of the principle of relativity is nowadays usually credited to Galileo.
Following the publication in 1632 of his great book 'Dialogues on the two World
Systems - Ptolemaic and Copernican', Galileo was put on trial by the Inquisition and
found guilty of teaching that the Earth moves and is not the centre of the World.
Aristotle had made a clear distinction between rest and motion Therefore it was
natural to think if the Earth moved, either through rotation or by moving round the
Sun, then this motion would be noticed. A falling body, for example, would not go
straight down to Earth but fall askew. In the book the dialogue for the second day
discusses at length the common observation that a person on a ship moving uniformly
on a calm sea can be unaware of motion relative to the sea and can have the
impression of being at rest. Further that, for example, a body dropped from the top of
the mast falls straight down to the foot of the mast. This may be regarded as the first
clear statement of the principle of relativity. It was used by Galileo in his later book
'Dialogues on two new Sciences' to show that the path of a projectile is a parabola.

A similar principle of relativity for uniform motion was used by Huyghens (1656) to
establish the law governing the impact of colliding bodies. Knowing the law of
collision for direct impact gives the law for other impacts on referring the motion to
the moving centre of gravity of the two colliding bodies.

This principle of relativity was fundamental for the new mechanics which was then
replacing the teaching of Aristotle. Aristotle, with his distinction between rest and
motion, had actually stated that if left alone a body at rest will remain at rest and that a
body in motion will remain in the same motion and his statements were apparently
known to Newton. However in Newton's first law of motion of 1687- the law of
inertia - uniform motion is seen as equivalent to being at rest although this is not
apparent in the customary imprecise statement of the law which reads 'state of rest, or
of uniform motion in a straight line' where the comma appears to separate distinct
possibilities. As pointed out by Koyr, the more accurate translation of Newton's
original Latin statement would read:

'A body perseveres in its state, at rest or in uniform motion in a straight line,
unless compelled by an impressed force to change its state. '

Here the dynamical state is understood to be 'at rest or in uniform motion in a straight
line'.

Newton defined also absolute space and time. As illustrated by the well-known
rotating bucket experiment, he considered absolute space necessary to explain
centrifugal force in rotating systems. So Newtonian absolute space came to replace
Aristotelian 'rest. Newton offered no discussion regarding absolute time. His use of
absolute space, subsequently criticised, was defended by Euler (1748) as a necessary
foundation for the analytical mechanics he initiated so successfully. Maxwell (1877)
subsequently stated that while Newtonian mechanics accepts that all motions must be
defined relative to some origin and so are in this sense relative, the application of the
87 87 87
second law of motion needs the concept of absolute motion so that in defining
acceleration the motion of the origin does not have to be taken into account.

Notes: Principal historical references are in the bibliography at the end of this appendix.
1) Galileo's dialogues of 1632 were withdrawn after publication, but were published in
English translation by Salusbury in 1661, the year Newton went to Cambridge, and were
generally known at that time. (cf Herivel: Background to Newtons Principia, Oxford 1965)
Salusbury also translated Galileo's 'Two new sciences' published in 1638 in Leyden which had
more direct interest for mechanics with detailed derivations.
2) Huyghens' argument is described in Dugas: History of Mechanics.
3) Aristotles statements appeared in Physica and de Caelo. They were subsequently used by
dAlembert in his Encyclopedie (cf. Compte PIRT Conf. 1998)
4) On Newton's statement of his laws see Koyr Newtonian Studies, London 1965 (Chapman
& Hall) Chap. III, Appendix A The customary inaccurate statement of Newtons law is there
traced to Mottes 1727 translation of Principia. A manuscript predating Principia stating
clearly the equivalence of rest and uniform motion is quoted by Woodhouse: Special
Relativity, Springer 2003.
5) On Euler's support for Newtonian absolute space and time in reply to Berkeley and
Boscovich see Vasiliev. Space, Time, Motion London 1924 (Chatto & Windus)


2. Optics and the Ether

In 1827 Bradley discovered stellar aberration which causes an apparent displacement
of a star in the direction of the Earths motion so that its observed position completes
an ellipse round its true position in the course of a year.

Using Newtons corpuscular theory of light, Bradley explained aberration by vector
addition of the light velocity c with the forward velocity v of the observer O on the
Earth so that the star S is observed at angle shown in the figure.
Fig: Aberration velocity triangle

The sine rule then gives approximately for the small aberration angle

= (v/c) sin

From this the velocity of light c may be determined.



88 88 88
Bradley found the speed of light to be the same for all stars examined and close to the
value found by Rmer in1769 from eclipses of Jupiters satellites.Since in this
derivation the angle of aberration depends on the speed of light, it might be expected
to change if the light was passing through a material of refractive index greater than
one which would reduce its speed correspondingly. The test was made by Arago in
1810 with a negative result. Being unable to find a simple explanation he asked
Fresnel whether an explanation could be found using the wave theory of light which
Young (1904) had shown could explain normal aberration. In reply Fresnel proposed
his theory that the ether, at rest, is partially dragged by the Earth and as a result the
velocity of the light in a medium moving with velocity v is given by the formula

c' = c/N + kv

Here N is the index of refraction and k is the dragging coefficient having the value.

k = 1 1/N
2


This formula was later experimentally investigated by Fizeau (1851) and found to be
remarkably accurate. The result, strange at the time, subsequently found confirmation in
the theory of relativity since the formula approximates addition of the velocities c/N and
v by the composition rule when v/c is small (von Laue 1907, cf. Miller 1998)

* The Doppler Effect: Another optical phenomenon which came to be important is the
Doppler Effect which is by its nature a wave phenomenon. It was discussed in several
papers from 1842 by Doppler who derived his formula and suggested it might be
applied to the spectra of double stars. This was confirmed experimentally by Huggins
and Miller in 1868

* Note: Boscovich, already in 1766, had proposed using a telescope filled with water but this
was not tried out until about a century later by Airy with a negative result.


3. Maxwells Equations and the Lorentz Theory

As remarked above, Maxwell, in his 1877 book Matter and Motion stated that the
concept of absolute motion is necessary for the foundations of Newtonian mechanics
and he had assumed ether absolutely at rest when he established his electromagnetic
theory of light (1873). The difference between systems at rest and in motion was
emphasized in the Herz form of Maxwells equations which came into general use
after 1890.

* The work of Lorentz: Stokes in 1845 proposed a theory of aberration which assumed
that the Earth dragged the ether in such a way that the relative velocity between them
became zero on the surface. This theory led to results at variance with Fresnels
hypothesis. Lorentz's earlier work aimed at reconciling Maxwells equations with
Fresnels hypothesis. In connexion with this he developed a theory in 1886 covering
both aberration and Doppler effect using what he called the principle of invariance of
the phase which states that the phase

= 2f (t - (lx + my + nz))


89 89 89
remains unaltered with change from rest to motion. Here f denotes frequency and
(l, m, n) the direction cosines of the wave-normal.
Later, with the appearance of the Herz form of Maxwell's equations and the
discovery of the electron (1891), Lorentz developed his theory of electrons to study
electrical phenomena in moving media. This theory explained electrical phenomena in
terms of a free flow of electrons in the ether, the ether exerting a force on the
electrons due to electric and magnetic fields, a force which became known as the
Lorentz force.

* Evolution of the Lorentz transformation: The history of the Lorentz transformation
began with Voigt (1887) who, when investigating the Doppler Effect, showed that the
three dimensional wave equation is unaltered by the change of variables

x' = x - vt
y' = y (1-v
2
/c
2
)
z' = z (1-v
2
/c
2
)
t' = t - (v/c
2
).x

It differs only by a scalar multiple from the Lorentz transformation now used. The
contribution of Voigt was however isolated and not followed up.

Lorentzs derivation of the transformation was in answer to the question:.given that
Maxwell's equations are valid in a system S at rest relative to the ether, what
equations are followed in a system S' moving relative to S ? Lorentz first made the
transformation from coordinates in S to moving (relative) coordinates coinciding
with S' by

x
r
= x vt, y
r
= y, z
r
= z, t
r
= t

He then found that the further transformation is necessary to ensure that Maxwells
equations keep their form in S' given by

x' = l x
r
, y' = l y
r
, z' = l z
r
, t' = lt
r
/ l vx
r
/c
2

Here the variable t' is a fictitious local time, l is a positive constant multiplier, and
is the constant, nowadays usually denoted by and called the Lorentz factor, which
was set at the approximate value

= 1 + v/ 2c

With this theory it was found that, to the first order in v/c, light behaves in the moving
system S' exactly as it does in the stationary system S. (Further details are given in the
book of Miller (1998) from which this account is taken.)

* The contraction hypothesis: The famous Michelson-Morley experiments of 1887,
1888 established that there is not any second order effect which enables a moving
system to be distinguished from one that is at rest. The difficulty of explaining this
fact resulted in the contraction hypothesis proposed independently by FitzGerald
(1888) and Lorentz (1895). According to this hypothesis, motion of the electrons
relative to the ether causes a length contraction in the direction of motion by the factor
(1-v/c), the electrons then becoming ellipsoidal. This hypothesis led on to a
revision of the value of to its exact value
90 90 90

= 1/(1-v/c)

And to a revised theory by Lorentz (1904)

* The standard Lorentz transformation: By substitution from Lorentzs two sets of
equations follows

x' = l (x - vt)
y' = l y
z' = l z
t' = l (t - (v/c)x)

Poincar (1905) called these equations the Lorentz transformation observing that for
the value l = 1 they combine and have an inverse of the same form so giving rise to a
one dimensional group, the Lorentz group, which he then in 1906 extended to three
dimensions. The scalar-multiplied form of this transformation includes that of Voigt
as Lorentz (1909) later acknowledged saying that Voigt's paper had escaped him all
those years and adding:

'The idea of the transformation might therefore have been borrowed from Voigt,
and the proof that it does not alter the form of the equations for the free ether is
contained in his paper'

From a slightly different point of view, the same transformation had also been used by
Larmor in his 1900 book Aether and Matter


4. The Principle of Relativity

After the negative result of the Michelson-Morley experiments of 1887 and 1888
Poincar, who had taken a close interest in the work of Lorentz commenting critically
on underlying assumptions, was led to speculate that no difference would be found
between Maxwell-Hertz equations for fixed and moving systems even for higher order
approximations since electrical systems, like mechanical, depend only on relative
motions. This view received further verification from the revised theory of Lorentz
(1904) which accounted not only for second order effects and which demonstrated the
invariance of Maxwells equations under the Lorentz transformation. In a lecture at
St.Louis in 1904 Poincar proposed several principles applying to all physical
phenomena one of which was:

'The principle of relativity, according to which the laws of physical
phenomena must be the same for a stationary observer as for an observer
carried along in a uniform motion of translation so that we have not and
can not have any means of discerning whether or not we are carried along
in such a motion'




91 91 91
In his 1905 paper which appeared shortly before that of Einstein, Poincar
questioned the existence of absolute motion for linear motions although he had in his
previous writings been careful to point out the exceptional nature of rotating systems.

Einsteins interest in the problem of relativity dates from his student days in Zrich
where he had studied Herz-Maxwell theory in the form given to it by Lorentz in 1895
and then appears to have been influenced by Poincars writings. His initial ideas
formed in collaboration with his wife Mileva Mari appear to have been based on the
theory of relative motions. Thus the introduction to his 1905 paper says

'..Maxwell's electrodynamics - as usually understood at the present time
when applied to moving bodies, leads to asymmetries which do not appear
to be inherent in the phenomena. Take, for example, the reciprocal
electrodynamic action of a magnet and a conductor. The observable
phenomenon here depends only on the relative motion of the conductor and
the magnet, whereas the customary view draws a sharp distinction between
the two cases in which either the one or the other of these bodies is in
motion.'

However the theory of the 1905 paper was not based on relative motions but
followed on from the Lorentz theory. Einsteins version of the Principle of Relativity
denied absolutes both for space and, most remarkably, for time. All inertial systems in
relative uniform motion were considered equivalent for the description of physical
laws while each observer keeps his own time measured by personal clock. The status
of rotating systems was left undefined and in fact, ignored. Einstein's paper 1905 was
followed by a restatement with modifications in the 1907 paper in 'Jahrbuch der
Physik'. In another series of papers Einstein followed up his growing interest in the
relation between mass and energy.

General recognition of the approach of the 1905 paper depended on experimental
verification by measurements on fast moving electrons. After initial unclear results
and an unsuccessful attempt to verify the theory using the transverse Doppler effect,
final experimental proof was reported by Bcherer at the same time as Minkowski's
famous 1908 lecture and established a general acceptance of the relativity principle in
the form Einstein had given it.

At the time, Einstein's contribution was regarded as a variation on the Lorentz theory
although Lorentz himself later acknowledged the immense simplification Einstein's
relativity hypothesis presented:

If I had to write the last chapter now, I should certainly have given a more
prominent place to Einstein's theory of relativity by which the theory of
electromagnetic phenomena in moving systems gains a simplicity that I
had not been able to attain. The chief cause of my failure was my clinging
to the idea that the variable t only can be considered as the true time and
that my local time t' must be regard as no more than an auxiliary
mathematical quantity. In Einstein's theory, on the contrary, t' plays the
same part as t; if we want to describe phenomena in terms of x',y',z',t' we
must work with these variables exactly as we could do with x,y,z,t

Lorentz: 1915 note added to 'Theory of Electrons' (1906) p.321.)

92 92 92
Notes etc:
1) On Lorentz's work relating to Maxwell's equations see Miller A I: Einstein's Special
Theory of Relativity. New York 1998 (Springer)
2) Detailed references for Poincars contribution to the founding of the theory are given in
Whittakers book. The above 1904 quotation is from Halsteads English translation of
Poincars St Louis Lecture of 1904 as reprinted in Poincars Value of Science.
3) Detailed biographic comments for Einstein including his indebtedness to others and use of
texts such as those of Fppl and Drude may be found in his Collected Works, vol. II. Many of
the comments are reproduced in Stachels Einsteins Miraculous Year
4) The early work of Einstein is recorded only by two letters of 1901 referring to our work on
relative motions to his wife Mileva Mari who appears to have made (at least) an initial
contribution to the theory of relativity and its philosophy. Being from Croatia she came from
the same region as Variak and may well have been influenced by ideas originating from
Boscovich. Variak made an extensive study of Boscovichs works and wrote many scholarly
articles on them. He brought to notice Boscovichs article of 1755 which contained many
clear and radical ideas regarding the relativity of space, time and motion (quoted from
Silberstein 1914, p. 38)
5) The experimental evidence from electron measurements supporting Einsteins theory was
given by Bcherer Die experimentalle Besttigung der Lorentz-Einsteinsche Theorie.Phys.
Z. 9 1908 Also in Ann. Phys. 1909.


5. Bibliography on the History of the Theory of Relativity

These are in chronological order up to the appearance of the Poincar-Einstein theory.

1632 Galilei G: Dialogo di Galileo Galilei Linceo doue nei congresso di quattro
giornate se dicorre sopra i due massimi sistemi del mondo tolemaico e copernicano.
Florence 1632; Withdrawn shortly after publication but republished Strasbourg 1637
in Latin; Engl. tr: 1665; G de Santillana (ed.) 'Dialogue on the Great World Systems',
Chicago 1953 (Univ. Press); S. Drake (tr.) 'Dialogues concerning the two chief World
Systems - Ptolemaic and Copernican', Berkeley Calif. 1967 (Univ. Calif. Press)

1656 Huyghens C De motu corporum ex percussione. [1656] (posth 1703)
Oeuvres Compltes XVI 3-168; German transl. ber die Bewegung der Krper durch
den Stoss Ostwald's Klassiker 138 1-34. F. Hausdorf (ed.) Described in Dugas.

1687 Newton I: Philosophiae Naturalis Principia Mathematica.
London 1687 (Samuel Smith); 2nd ed. 1713; 3rd ed. 1726; Engl. tr. by Motte
Cajori: The Mathematical Principles of Natural Philosophy, Berkeley 1934 (Univ.
Calif. Press)

1728 Bradley J: Letter to Dr Halley giving an account of a new discovered motion of
the Fix'd Stars. Phil. Trans 35 1727/28 637- 661

1755 Bokovi R.J. (Boskovich): De moto absoluto, an possit a relativo distingui,
Rome 1755. Reprinted by Variak (ed.) South Slavic acad. sci. 190 1912.

1804 T.Young: Experiments and Calculations relative to physical optics.
Proc. Roy. Soc. 94 1804 1-16; Works I, 188- [Aberration explainable by wave theory]



93 93 93
1811 D.F-J.Arago Mmoire sur un modification remarkable qu'eprouvent les rayons
lumineux dans leur passage a traverse certains corps diaphanes d'optique.
Paris Mm. de l'Inst. 1811, 93-134; C.R. Acad. Sci Paris, 8 1829 326-; 36 1853 380-
[Arago showed experimentally that in all respects light behaves as though it were
coming from the direction of aberration.]

1818 Fresnel A.J. Sur l'influence du mouvement terrestre dans quelques
phenomnes d'optique. Ann. Chimie Phys. 9 1818 57- ; Oeuvres compltes l II 627-.
[Reply of Fresnel to Arago giving the Fresnel drag formula]

1841 Doppler C. Ueber das farbige Licht der Doppelsterne und einiger anderer
Gestirne des Himmels. Bohm. Ges. Abh. 2 1841-42 467-490; Abhandlungen, Leipzig
1907 (Engelmann), Ostwald's Klassiker No. 161 pp 1-24.

1845 Stokes G.G. On the aberration of light
Phil. Mag. 27 1845 9-15; 29 1846 62-63; Math & Phys Papers I 134-.
[Stokes theory on ether drag by the Earth.]

1846 Doppler C: Bemerkungen zu meiner Theorie des farbigen Lichtes der
Dopplesterne Pogg. Ann. 68 1846 1-34.

1846 Stokes G G: On Fresnel's theory of the aberration of light
Phil Mag 28 1846 76-81 Math & Phys Papers I.

1847 Doppler C. Ueber den Einfluss der Bewegung des Fortpflanzungsmittels auf
die Erscheinungen der Aether-, Luft- und Wasserwellen. Bhm. Ges. Abh. 5 1847
293-306.

1848 Stokes G.G. On the constitution of the luminiferous ether.
Phil.Mag. 32 1848 343-349; Math.& Phys. Papers I.

1851 Fizeau H. Sur les hypothses relatives .. l'aether lumineux, et sur une
exprience qui parait dmontrer que le mouvement des corps change le vitesse avec
laquelle la lumire se propage dans leur intrieur. C.R. Acad. Sci. Paris 33 1851 349-
355.

1868 Huggins W & Miller W.A: Notes on the lines in the spectra of some of the
fixed stars, Proc Roy Soc.18 1868 444-445
[Experimental verification of Doppler shift for stellar motion]

1873 Maxwell J C: Treatise on Electricity and Magnetism, 2 vols
Oxford 1873 (Univ. Press); rpr abridged 3rd ed. New York 1954 (Dover)

1877 Maxwell J C: Matter and Motion
Cambridge 1877 (Univ. Press) rpr New York 1952 (Dover)

1878 Maxwell J.C: Ether
Encyclopedia Britannica 1878: Scientific Papers 763-775.

1881 Michelson A.A:. The relative motion of the Earth and the luminiferous ether
Amer. J. Sci. 22 120-129. [Michelson's first experiment]

94 94 94
1886 Lorentz H.A. Sur l'influence de la movement de la terre sur les phenomnes
lumineux. Arch. nerl. Sci.21 1887 103-176. (Previously in Dutch in Versl Kon Akad
Wet 2 1886 297-372); Coll. Papers IV 153-214. [Derivation of optical Doppler Effect
and stellar aberration]

1887 Michelson A.A. & Morley E.W: On the relative motion of the Earth and the
luminiferous aether, Amer. J. Sci. 34 333-345 1887; Phil. Mag. 24 1887

1887 Voigt M. Ueber das Doppler'sche Princip.
Gtt Nachr 1887 rpr in Phys. Z. 1915

1892 Herz H: Electric Waves
Leipzig 1892 (Teubner); London 1893 (MacMillan); rpr New York 1962 (Dover)

1889 FitzGerald G F: The ether and the earth's atmosphere
Science 1889 [The Lorentz contraction proposed in a nontechnical way]

1892 Lorentz HA La thorie electromagnetique de Maxwell et son application aux
corps mouvants Arch nerl Sci 26 1892 363-; Leiden 1892 (Brill); Coll Papers II
164-343 [Lorentz-contraction]

1892 Lorentz H.A. De relatieve beweging van de Aarde en den Aether.
Zitt. Verslag. Akad. Wet. Am'dam 1 Nov. 1892, 74-79; Coll.Papers IV 219-223.

1892 Lodge O.J. On the present state of our knowledge of the connexion between
either and matter, Nature June 1892 p.164 [Refers to FitzGerald contraction]

1893 Lodge O J: A study on aberration
Phil. Trans. 184 1893 727-804 [Refers to FitzGerald contraction on p.749]

1894 Fppl A: Einfhrung in die Maxwell'sche Theorie der Elektrizitt.
Leipzig 1894 (Teubner); republished 1904 as vol I of 'Theorie der Elektrizitt' by
Abraham. [Influential textbook based on Lorentz theory]

1895 Lorentz H.A. Versuch einer Theorie der electrischen und optischen
Erscheinungen in bewegten Krpern. Leiden 1895 (Brill): Coll. Papers 5 1-137.
[Chap. V: Second derivation of optical Doppler Effect and stellar aberration using
principle of corresponding states. Local time defined in paras.89-92, p.122 refers to
FitzGerald, acknowledging his independent proposal of the contraction hypothesis.]

1899 Lorentz H.A. Thorie simplifie des phnomnes electriques et optiques dans
les corps en mouvement. Proc. Acad. Sci. Amsterdam .1 1899 427-442. Coll. Papers
IV 139-155 Previously in Dutch in Verslag. kon. Akad Wet 7 1899 507-522
[Lorentz transformation exact for square-root]

1900 Larmor J Aether and Matter
Cambridge 1900 (Univ. Press) (Previously Cambridge 1898, Adams Prize Essay)
[Wrote Lorentz transformation correct to second order terms and showed the
Fitzgerald-Lorentz contraction as well as the slowing of clocks to be consequences.
See pp.167-177 in the book, especially 173]


95 95 95

1900 Drude P.K.L. Lehrbuch der Optik 1900
Tr. as Textbook of Optics London 1902 [Limitation of Doppler formula emphasized]

1900 Poincar H. La thorie de Lorentz et le principe de reaction.
Arch. nerl. Sci. 1900; Oeuvres IX 464-488.

1901 Poincar H. Electricit et optique.
Paris (Sorbonne lectures to 1900)

1904 Lorentz H.A: Electromagnetic phenomena in a system moving with any velocity
smaller than that of light. Proc Acad Sci Amsterdam 6 1904 809- 834; Coll Papers V
172-197; rpr in 'The Principle of Relativity'

1904 Poincar H: Lecture to Congress of Arts and Sciences, St. Louis, USA Sept
Bull. Sci. Math. 28 1904 302-; Engl. tr. (G.B. Halstead) Monist Jan 1905.
[Stated The principle of relativity]

1904 Lorentz H.A. Maxwells Elektromagnetische Theorie.
Enz. Math. Wiss. V(2) 13 Leipzig 1904 (Teubner), 63-144.

1904 Lorentz H.A. Weiterbildung der Maxwell'schen Theorie: Electronentheorie.
Enz. Math. Wiss. V(2) 14 Leipzig 1904 (Teubner), 145-280.

1905 Poincar H. Sur la dynamique de l'electron.
C.R. Acad. Sci. Paris 140 1905 1504-1508, Oeuvres IX 489-493.

1905 Einstein A. Zur Elektrodynamik bewegte Krper. Ann Physik 17 1905 891-
921; Engl.tr 'On the electrodynamics of moving bodies' in 'The Principle of Relativity'
Coll. Works II 275- and 306 with extensive notes.

1906 Poincar H. Sur la dynamique de l'electron.
Rend. Circ. mat. Palermo 21 1906 129-176, Oeuvres IX 494-550.
Engl. tr. (Keswani G.H. & Kilminster C.W.) Brit. J. Phil. Sci. 34 1983 343-.

1907 Einstein A. ber die Relativittsprinzip und die aus demselben gezogenen
Folgerungen. Jahrb. Radioaktivitat 4 1907 411-462; Coll. Papers II 433-484.

1908 Bcherer A.H. Messungen an Becquerel strahlen. Die experimentalle
Besttigung der Lorentz-Einsteinsche Theorie. Phys. Z. 9 1908 755-762.
Also in Ann. Phys. 1909

1909 Lorentz H.A. The Theory of Electrons (Columbia University Lectures 1906)
Leipzig 1909 (Teubner); London 1909, 1915 (Nutt); rpr. New York 1952 (Dover)
[cf. Chapter 5: Optical phenomena in moving bodies.]







96 96 96

6. Secondary Literature and Collected Works

Doppler C. Abhandlungen
Leipzig 1907 (Wilhelm Engelmann), Ostwald's Klassiker No. 161.

Dugas R. Histoire de la mcanique.
Neuchatel 1950 (Griffon); Engl. tr. London 1957 (Routledge & Paul); 1988 (Dover)

Einstein A. The Collected Papers of Albert Einstein, ed J. Stachel et al.
(Princeton Univ. Press 1989) [Relativity papers 1900-1909 are in vol. II]

Lorentz H.A. Collected Papers, 9 vols,
The Hague 1935-39 (Nijhoff)

Miller A.I. Albert Einstein's Special Theory of Relativity
New York 1998 (Springer)

Poincar H. Oeuvres de Henri Poincar (ed. G.Darboux)
Paris 1916-1956 (Gauthier-Villars) [Relativity papers are in vol. IX]

Stachel J. Einstein's Miraculous Year
Princeton 1998 (Univ.Press)





























97 97 97

APPENDIX 2 - Mathematical Notes


1. Spherical Trigonometry

A spherical triangle is formed by three great-circle arcs on the surface of a sphere.
The vertices and the angles at these vertices are usually denoted by A, B, C and the
lengths of the sides by a, b, c. The sides may alternatively be characterized by the
angles , , they subtend at the centre of the sphere which are related to a, b, c by

= a/R, = b/R, = c/R

R is here the radius. The principal formulae are then, relative to vertex A:

(a) The cosine formulae:

cos = cos cos + sin sin cos A etc.
cos a/R = cos b/R cos c/R + sin b/R sin c/R cos A etc.

(b) The sine formula:

sin A/sin = sin B/sin = sin C/sin

(c) The polar cosine formulae: With any point on a sphere is associated the
equatorial great circle having this point as a pole. An orientation of this great
circle may be determined by the right handed corkscrew rule. Then the three
vertices of any triangle have three associated great circles forming what is called
the polar triangle. The relation between a triangle and its polar triangle is
reciprocal. The angles , , , A, B, C of either triangle correspond to -A, -B,
-C, -, -, - of the other. The cosine rule applied to the polar triangle gives
polar cosine formulae

cos A = - cos B cos C + sin B sin C cos etc.

* Remark: The sine rule may be deduced from the cosine formula by writing

sin
2
A = 1 - cos
2
A = (sin sin )
2
(cos - cos cos )
2

(sin sin )
2


The denominator on the right which will be denoted by f
2
(f > 0) can be brought to a
symmetrical form:

(sin sin )
2
(cos - cos cos )
2
= (1 - cos
2
) (1 - cos
2
) - cos
2
+ 2 cos cos cos - cos
2
cos
2

= 1 cos
2
cos
2
cos
2
+ 2 cos cos cos

The sine rule then follows from the symmetry of formula.

sin A = f .
sin sin sin sin

98 98 98


Fig. A right-angled spherical triangle

* Right-angled triangles:

The formulae for a right angled triangle are the following (with B as right angle)

(a) The cosine rule (Pythagoras theorem) (d) The adjacent side rule

cos = cos cos tan = tan A sin
tan = tan C sin

(b) The sine rule (e) The cotangent rule:

sin = sin A sin cos = cot A cot C
sin = sin C sin

(c) The tangent rule: (f) The adjacent angle rule:

tan = cos A tan cos A = sin C cos
tan = cos C tan cos C = sin A cos

Some may be deduced geometrically by projecting the triangle on to a tangent
plane as described in the text while others are simple deductions from these.
These 10 formulae cover all possible combinations of the unknowns A, B, a, b, c.
They may all be deduced diagramatically from Napiers Rule. For further details
on these matters see Todhunter & Leathems book.






---------------------------------------------------------------------------------------------
* Reference: Todhunter & Leathem: Spherical Trigonometry London (1919). An
excellent text written for schools The above formulae are as given in this book
transcribed for the case when B is a right angle.



99 99 99
* Spherical excess: From the polar cosine formula may be deduced that the sum
of the angles of a spherical triangle is greater than .

cos A = - cos B cos C + sin B sin C cos a/R
< - cos B cos C + sin B sin C
= - cos (B+C)
= cos (-B-C)
from which

A < B C, i.e. A+B+C <

The spherical excess is the positive difference

E = (A+B+C) -

It gives the area of a spherical triangle by Girauds formula

area = E R
2


* Lagranges formula for excess: numerous formulae have been given for finding
the excess. The most appropriate for this book is that described by Lagrange
(1799) which is

cot (E/2) = cos /2 cos /2 + sin /2 sin /2 cos A = cot /2 cot /2 + cos A
sin /2 sin /2 sin A sin A

Lagranges derivation was the following

cot (E/2) = - tan (A + B + C)/2 = - tan A/2 + tan (B + C)
1 tan A/2 tan (B + C)/2
Now use is made of the identity

tan (B + C)/2 = cos ( )/2
cot A/2 cos ( + )/2

which is one of Napiers analogies ( SeeTodhunter & Leathem) Substituting for tan
(B + C)/2, leads to the required expression

cot (E/2) = - tan A/2 cos ( + )/2 + cot A/2 cos ( - )/2
sin /2 sin /2 sin A

= cos /2 cos /2 + sin /2 sin /2 cos A
sin /2 sin /2 sin A

Lagrange showed that this may be converted to a symmetric form as shown in
the analogous hyperbolic case below.

Reference: Lagranges paper: Solution de quelques problmes ... may be read in his
Collected Works: Oeuvres VII 331-359 (see especially 339)
100 100 100
2. Hyperbolic Trigonometry

We pass from the formulae for a sphere of constant Gaussian curvature R
2
to
corresponding formulae for a two-dimensional hyperbolic space of constant
Gaussian curvature - R
2
by changing R into iR (Taurinus' theorem). This is
equivalent to changing angles from , , to /i, /i, /i i.e. to -i, -i, -i. Then
reinterpreting sine, cosine of imaginary angles as sinh, cosh (abbreviated sh, ch)
there are found the corresponding hyperbolic formulae:

(a) The hyperbolic cosine formula:

ch = ch ch - sh sh cos A etc.
ch(a/R) = ch(b/R) ch(c/R) - sh(b/R).sh(c/R).cos A etc.

(b) The hyperbolic sine formulae:

sin A/ sh = sin B/ sh = sin C/ sh

(c) The hyperbolic polar cosine formula:

cos A = - cos B cos C + sin B sin C ch etc.


* Remark: The hyperbolic sine rule may be deduced in the same way as in the spherical case
from the hyperbolic cosine formula by writing

sin
2
A = 1 - cos
2
A = 1 - (ch ch - ch )
2
(sh sh )
2

From this the sine rule follows from the symmetrical right hand side of

sin A = (1 - ch
2
- ch
2
- ch
2
+ 2 ch ch ch )
sh sh sh sh

* Infinite triangles: In hyperbolic geometry it is possible to have infinite triangles
with one or more of the vertex angles zero. One such triangle is shown below
formed by the perpendicular and an asymptotic parallel to a straight line BA from
a point C not on this line.
Fig: The parallel angle (a) of Lobachevsky



101 101 101
The angle B is the parallel angle of Lobachevsky denoted by him by (a)
showing its dependence on the length a of the perpendicular CB. The right-
angled triangle ABC has one zero angle and two infinite sides. On putting A = 0,
B = (a), C = /2 , b = , c = in the above polar cosine formula for cos A
follows,

1 = sin (a) ch

From this follows the formulae of Lobachevsky for the parallel angle:

sin (a) = 1/ ch (a/ R)
cos (a) = th (a/ R)
tan (a) = 1/ sh(a/R)

* References on hyperbolic geometry: The literature is considerable. Standard
texts are those of Bonola, Sommerville while the books of Coxeter, Rosenfeld
and Shirokov are to be recommended. Bonola includes a translation of Bolyais
work. Not much of the original work of Lobachevski is easily available in
translation Apart from a small introductory volume originally published in
German, there is available one article in French, Gometrie imaginaire and an
English translation of a later work Pangeometry both listed at the end of this
appendix.

* Right-angled hyperbolic triangles:

With B as right angle the formulae corresponding to the spherical case are as
follows

(a) The hyperbolic cosine rule (d) The hyperbolic adjacent side rule

ch = ch ch th = tan A sh
th = tan C sh

(b) The hyperbolic sine rule (e) The hyperbolic cotangent rule

sh = sin A sh ch = cot A cot C
sh = sin C sh

(c) The hyperbolic tangent rule: (f) The hyperbolic adjacent angle rule

th = cos A th cos A = sin C ch
th = cos C th cos C = sin A ch

As with spherical formulae, these 10 formulae cover all possible combinations of
the unknowns A, B, a, b, c and may be set out by a modified form of Napiers
Rule. Certain of these formulae may be deduced by projection on to a tangential
Euclidean plane (the limiting plane) as in the spherical case.
---------------------------------------------------------------------------------------------------------
References:
1) For the hyperbolic Napier Rule, due to Engel, see Sommerville: Non-Euclidean
Geometry
2) The method of projection on to the limiting plane is described in Shirokov: A sketch
of the fundamentals of Lobachevskian geometry, (1964)
102 102 102
* Hyperbolic defect: From the polar cosine formula it may be shown that the
sum of the angles of a hyperbolic triangle is less than . Following a similar
method to the spherical case we have

cos A = - cos B cos C + sin B sin C ch a/R
> - cos B cos C + sin B sin C
= - cos (B+C)
= cos (-B-C)

From which follows

A < -B-C i.e. A+B+C <

The hyperbolic defect is defined as the positive difference

D = - (A+B+C)

It gives the area of a hyperbolic triangle by a formula due to Gauss (cf the
spherical case)

area = D R
2


* The Hyperbolic Form of Lagrange's Formula for Excess (Area):

Lagrange's formula becomes

cot (D/2) = ch /2 ch /2 + sh /2 sh /2 cos A
sh /2 sh /2 sin A

LEMMA: In any hyperbolic triangle,

tan (B + C) = ch (-)/2 cot A
2 ch (+)/2 2

This is the first of Napier's analogies for a hyperbolic triangle. Its proof is similar to
the spherical case. From it the Lagrange formula follows easily:

cot (D/2) = tan (A+B+C)/2 = tan A/2 + tan(B+C)/2
1 + tan A/2 tan(B+C)/2
By the lemma the right hand side is

rhs = tan(A/2) ch(+)/2 - cot(A/2) ch(-)/2
ch(+)/2 - ch (-)/2

= ch /2 ch /2 + sh /2 sh /2 cos A
sh /2 sh /2 sin A

-------------------------------------------------------------------------------------------------
Reference: See Sommerville (1914) for Gauss simple proof of his formula for area of a
hyperbolic triangle.


103 103 103
* The symmetric form of the Lagrange formula: The formula may be further
transformed by multiplying both denominator and numerator by 4 ch /2 ch /2:

rhs = 4 (ch /2 ch /2) + 4 ch /2 ch /2 sh /2 sh /2 cos A
4 ch /2 ch /2 sh /2 sh /2 sin A

The numerator is then, converting to whole angles,

(ch + 1)(ch + 1) + sh sh cos A
= (ch -1) (ch - 1) + ch - ch ch
= 1 + ch + ch + ch

while the denominator is, by the hyperbolic sine rule,

sh sh sin A = (1 - ch
2
- ch
2
- ch
2
+ 2 ch ch ch )

resulting in the symmetric formula

cot (D/2) = 1 + ch + ch + ch .
(1 - ch
2
- ch
2
- ch
2
+ 2 ch ch ch )

Here cos (D/2) and sin (D/2) will be proportional to the numerator and denominator
respectively. The constant of proportionality is easily identified by squaring and
adding giving

cos (D/2) = 1 + ch + ch + ch
4 (ch /2)(ch /2)(ch /2)

sin (D/2) = (1 - ch
2
- ch
2
- ch
2
+ 2 ch ch ch )
4 (ch /2)(ch /2)(ch /2)

104 104 104
3. Cayley-Klein Projective Metric

Cayley (1869) showed how the concept of distance could be introduced into
projective geometry relative to an absolute conic (or simply 'absolute') having
equation

a(x, x) = 0

Here a(x, y) is a bilinear form in the vectors x, y of a projective space.

a(x, y) = a
ij
x
i.
y
j

i, j = 1

Projective distance is defined relative to the absolute by the formula

= cos
[-1]
a(x,y) .
{a(x,x) a(y,y)}

In three dimensions setting

a(x, x) = x
1
2
+ x
2
2
+ x
3
2


makes the absolute an imaginary locus and the Cayley definition gives the angle
between the vectors x and y or, if these lie on a sphere with centre at the origin, the
arc length on the surface of the sphere.

Klein (1871) considerably developed Cayley's idea by a method based on cross-
ratios leading to a unified treatment by projective geometry of Euclidean, spherical
and Lobachevskian geometries according to the nature of the absolute conic. He
introduced the terms parabolic, elliptic or hyperbolic respectively to denote the three
basic geometries.

In the hyperbolic case the quadratic a(x, x) is semi-definite so the absolute becomes a
real ellipse and in view of the reversed Cauchy inequality the Cayley formula must be
written as

= ch
[-1]
a(x,y) .
{a(x,x) a(y,y)}

Ordinary Euclidean geometrical ideas apply in the interior of this ellipse but straight
lines are considered to meet only if the meeting point lies in its interior. So lines do
not necessarily meet and are then considered parallel. Lines meeting on the ellipse are
defined as asymptotically parallel, the bounding conic thus representing infinity.
With these conventions it is found that all Euclidean axioms are satisfied except those
relating to parallels






105 105 105

* Relation to the Riemannian form: It is easy to show an equivalent formula for is

sh = {a(x,y)- a(x,x).a(y,y)}
{a(x,x) a(y,y)}

On putting y = x + dx there is found for the infinitesimal metric element squared:

ds = a(x,dx) a(x,x) a(dx,dx) = a(x,dx) a(dx,dx)
a(x,x) a(x,x) a(x,x)

It relates Cayley-Klein and Riemannian metrics. This formula shows that tangential
elements to the surface a(x, x) = const (> 0) satisfying the constraint that a(x,dx) = 0
have metric element squared

ds = a(dx, dx)/ const.

Note the unexpected minus sign.

* References: The original reference to Cayley's work is the last part of his 'Sixth Memoir on
Quantics': Phil. Trans. 1859. Kleins' work was published in his elegantly written: Nicht
Euklidische Geometrie reprinted by Chelsea 1927. See also Veblen & Young': Projective
Geometry (Ginn) for a fine exposition.


4 The Beltrami-Klein Representation:

Plane hyperbolic geometry is represented by the geometry of line segments within a
circle as in fig.1. Two line segments are considered to be asymptotically parallel if
they meet on the circle as in fig.2.

Fig.1 Intersecting and Fig.2: Asymptotic parallels
non-intersecting lines.

* The Cayley-Klein metric: If the equation is centred at the origin and has radius R, its
equation is

x
2
+ y
2
= R
2


which may be put into homogeneous form by writing X/Z, Y/Z for x and y so that

X
2
+ Y
2
R
2
Z
2
= 0

106 106 106
With this as absolute, the Cayley-Klein distance is,

ch = (R
2
Z
1
Z
2
X
1
X
2
Y
1
Y
2
) .
R {(R
2
Z
1
2
X
1
2
Y
1
2
)(R
2
Z
2
2
X
2
2
Y
2
2
)}

This in nonhomogeneous form is

ch = (R
2
- x
1
x
2
- y
1
y
2
) .
R {(R
2
- x
1
2
y
1
2
)(R
2
- x
2
2
y
2
2
)}

From the last equation is found

sh = R{(x
1
- x
2
)
2
+ (y
1
- y
2
)
2
- (x
1
y
2
- x
2
y
1
)}
R (R
2
- x
1
2
- y
1
2
)(R
2
- x
2
2
y
2
2
)}

th = R{(x
1
- x
2
)
2
+ (y
1
- y
2
)
2
- (x
1
y
2
- x
2
y
1
)}
R (R
2
- x
1
x
2
- y
1
y
2
)

* Polar formulae: Using the formula for th (/R), the distance from the origin (0,0)
to a general point (x,y) at radial distance r is found from

th = r
R R

giving the parametric Beltrami form for the nonhomogeneous case as

x = r cos = R th (/R) cos
y = r sin = R th (/R) sin

and for the homogeneous case,

X = R sh (/R) cos
Y = R sh (/R) sin
Z = ch (/R)

The coordinates X, Y, Z (usually termed Weierstrass coordinates), satisfy identically

-X
2
- Y
2
+ R
2
Z
2
= R
2


* The Riemannian form: The infinitesimal metric squared element (non-dimensional)
is found from the formula for sh(/R) or th (/R) as

R
2
{(R
2
-y
2
) dx
2
+ 2xy dxdy + (R
2
-x
2
) dy
2
}
(R
2
-x
2
-y
2
)
2

R
2
{(R
2
-x
2
-y
2
) dx
2
+ (R
2
-x
2
-y
2
) dy
2
+ (xdy-ydx)
2
}
(R
2
-x
2
-y
2
)
2


R
2
(dx
2
+ dy
2
) + (xdy ydx)
2
(R
2
-x
2
-y
2
) (R
2
-x
2
-y
2
)
2


107 107 107

Changing to polar coordinates gives the metric element squared as

dr
2
+ r
2
d
2

1 - r
2
/R
2
(1 r
2
/R
2
)
2


The further change to Beltrami coordinates gives it as

d
2
+ R
2
sh
2
(/R) d
2

* The 3-dimensional case: The region within the circle now becomes the region

x
2
+y
2
+z
2
< R
2


The radial distance from the origin (0, 0, 0) to point r = (x, y, z) is found as before as

th (/R) = r/R

It gives the parametric form of the Beltrami representation in spherical coordinates:

x = R th (/R) sin cos
y = R th (/R) sin sin
z = R th (/R) cos

The corresponding homogeneous form using Weierstrass coordinates (X, Y, Z, T) is

T = ch (/R)
X = R sh (/R) sin cos
Y = R sh (/R) sin sin
Z = R sh (/R) cos

These parametrize the hyperboloid:

X
2
Y
2
Z
2
+ R
2
T
2
= R
2

* The Cayley metric: Using r to denote vectors (x, y, z), equivalent forms for the
nonhomogeneous Cayley metric giving distance from r
1
to r
2
are

ch = R
2
r
1
.r
2
.
(R
2
r
1
.r
1
)(R
2
r
2
.r
2
)

sh = {R
2
(r
2
r
1
).(r
2
r
1
) [(r
1
.r
1
) (r
2
.r
2
) (r
1
.r
2
)
2
]}
(R
2
r
1
.r
1
)(R
2
r
2
.r
2
)

= {R
2
(r
2
- r
1
).(r
2
- r
1
) (r
1
x r
2
).(r
1
x r
2
)}
(R
2
r
1
.r
1
)(R
2
r
2
.r
2
)

The last step used the Lagrange identity

(r
1
.r
1
)(r
2
.r
2
) (r
1
.r
2
)
2
= (r
1
x r
2
).(r
1
x r
2
)

The formula for th may be found from these by division.
108 108 108



* Riemannian metric: Using the formula for sh or th and putting r
1
= r and
r
2
= r + dr, we find for the nondimensional metric element squared:

R
2
{(R
2
r.r) (dr.dr) + (r.dr)
2
}
(R
2
- r.r)
2
.

Inside the bracket, using spherical coordinates,

r.r = r
2

dr.dr = (dr)
2
+ r
2
(d
2
+ sin
2
d
2
)
r.dr = xdx+ydy+zdz = d(r
2
) = r dr

and so

(R
2
r.r)(dr.dr) + (r.dr) = R
2
(dr)
2
+ (R
2
r
2
) r
2
(d
2
+ sin
2
d
2
)

Now substitution results in

R
2


{dr
2
+ r
2
(d
2
+ sin
2
d
2
)} + R
2


(r.dr)
2

(R
2
-r
2
) (R
2
-r
2
)
2


= (dr/R)
2
+ 1 (r/R)
2
{d
2
+ sin
2
d
2
}
(1r
2
/R
2
)
2
(1r
2
/R
2
)

The further transformation to Beltrami coordinates gives the dimensional metric
element squared in the standard Riemannian form.

d
2
+ R
2
sh
2
(/R) {d
2
+ sin
2
d
2
}














-----------------------------------------------------------------------------------------------------
* Note: Beltramis original papers were published in the same year as the posthumously
published dissertation of Riemann in 1868. They were the first to consider the application of
Riemann's ideas to spaces of negative curvature. The Beltrami space was soon afterwards
treated by an alternative projective approach by Klein (1871) and became better known under
his name.

109 109 109


5. References for Mathematics Appendix

Beltrami E: Saggio di Interpretazione delle Geometria non-Euclidea.
Giorn di Mat.8 1868 284-312; Opere Mat. I 374-405; Engl.tr.in Stillwell 1996

Beltrami E: Teoria fondamentale degli spazii di curvatura costante.
Annali di Mat. Pura ed applicata, 11(2) 1868-69 2323-255; Engl.tr. In Stillwell 1996

Bonola R: La geometria non-euclidea, Bologna 1906
Engl.tr. Non-Euclidean Geometry, Chicago 1912 (Open Court); rpr New York 1955

Cayley A: Sixth memoir upon Quantics.
Phil. Trans. 44 1859 61-90; Coll. Works II 561-592.

Coxeter H.S.M: Non-Euclidean Geometry
Washington 1942 (Math Assoc America)

Klein F. Nicht-Euklidische Geometrie
Rosemann (ed.), Berlin 1923, 1928 (Springer); rpr. New York n.d. (Chelsea)

Lagrange J.L: Solution de quelques problmes relatifs aux Triangles Sphriques
J. Ecole Poly VI (II) 1799; Oeuvres VII 331-359 (see especially 339)

Lobachevski N.I. Gometrie imaginaire.
J. reine ang. Math. (Crelle) 17 1837 295-320.

Lobachevski N.I. Pangeometriya Sci. Pub. Univ. Kazan 1855;
Engl. tr. Smith D.E. Source Book in Mathematics, vol. II 360-374.

Lobachevski N.I. Polnoe Sobranie Sochineni (ed. Kagan et al), 4 vols
(Complete Collected Works) Moscow-Leningrad 1946-48

Rosenfeld B.A: Istoriya neevlidvoi Geometri, Moscow 1976, (Nauka)
Engl.tr: A History of Non-Euclidean Geometry, New York etc. 1988 (Springer)

Sommerville D. M. Y: The Elements of Non-Euclidean Geometry.
London 1914 (Bell); rpr New York 1958 (Dover)

Shirokov P. A: A Sketch of the Fundamentals of Lobachevskian Geometry.
Engl tr. from Russian, Groningen 1964 (Noordhoff)

Stillwell J: Sources of Hyperbolic Geometry.
AMS & LMS 1996

Todhunter I & Leathem J. G: Spherical Trigonometry
London 1901 etc. (MacMillan)

Veblen O &Young J.W: Projective Geometry.
Boston etc. 1910, 1913 (Ginn)
----------------------------------------------------------------------------------------------------