Minkowski Papers PDF

Hermann Minkowski
Space and Time

Minkowski’s Papers on Relativity
Translated by Fritz Lewerto↵ and Vesselin Petkov

Edited by Vesselin Petkov
Hermann Minkowski
22 June 1864 – 12 January 1909
c 2012 by Minkowski Institute Press

All rights reserved. Published 2012
ISBN 978-0-9879871-4-3 (softcover)

ISBN 978-0-9879871-1-2 (ebook)
Minkowski Institute Press

Montreal, Quebec, Canada
minkowskiinstitute.org/mip/
For information on all Minkowski Institute Press publications visit our web-
site at minkowskiinstitute.org/mip/books/
May the hope be fulfilled, through this dissertation, that a
wider circle of people become motivated so that participants,
who immerse themselves in Minkowski’s ideas and the theory of
relativity, may each and all contribute their part to promote
and spread this theory in accordance with Minkowski’s bold
dream and that, hence, future generations of mankind will be
consciously aware that space and time recede completely to be-
come mere shadows and only the space-time-transformation still
stays alive.
Aachen, May 1910

Otto Blumenthal
From the Foreword to H. Minkowski, Zwei Abhandlungen über

die Grundgleichungen der Elektrodynamik (Teubner, Leipzig 1910)
i
Preface
This volume contains together for the first time Hermann Minkowski’s
three papers on relativity written by himself1 :
• The Relativity Principle 2 – lecture given at the meeting of the Göttin-

gen Mathematical Society on November 5, 1907.
• The Fundamental Equations for Electromagnetic Processes in Moving

Bodies 3 – lecture given at the meeting of the Göttingen Scientific So-
ciety on December 21, 1907.
• Space and Time 4 – lecture given at the 80th Meeting of the Natural
Scientists in Cologne on September 21, 1908.
The three papers were translated by Fritz Lewerto↵ and myself. Fritz
Lewerto↵ translated Das Relativitätsprinzip, which is the first English trans-
1
Almost immediately after Minkowski’s sudden and untimely departure M. Born (a
student of Minkowski) embarked on decoding the calculations Minkowski left and suc-
ceeded in assembling them in a fourth paper (which has never been translated into En-
glish and which will be included in a planned volume with Minkowski’s physics papers
to be published also by the Minkowski Institute Press): Eine Ableitung der Grundgle-
ichungen für die elektromagnetischen Vorgänge in bewegten Körpern vom Standpunkte
der Elektronentheorie (Aus dem Nachlaß von Hermann Minkowski bearbeitet von Max
Born. Mathematische Annalen 68 (1910) S. 526-551); reprinted in H. Minkowski, Zwei
Abhandlungen über die Grundgleichungen der Elektrodynamik, mit einem Einführungswort
von Otto Blumenthal (Teubner, Leipzig 1910) S. 58-82, and in Gesammelte Abhandlungen
von Hermann Minkowski, ed. by D. Hilbert, 2 vols. (Teubner, Leipzig 1911), vol. 2, pp.
405-430.
2
H. Minkowski, Das Relativitätsprinzip, Annalen der Physik 47 (1915) S. 927-938.
3
H. Minkowski, Die Grundgleichungen für die elektromagnetischen Vorgänge in be-
wegten Körpern, Nachrichten der K. Gesellschaft der Wissenschaften zu Göttingen.
Mathematisch-physikalische Klasse (1908) S. 53-111; reprinted in H. Minkowski, Zwei Ab-
handlungen über die Grundgleichungen der Elektrodynamik, mit einem Einführungswort
von Otto Blumenthal (Teubner, Leipzig 1910) S. 5-57, and in Gesammelte Abhandlungen
von Hermann Minkowski, ed. by D. Hilbert, 2 vols. (Teubner, Leipzig 1911), vol. 2, pp.
352-404.
4
H. Minkowski, Raum und Zeit, Physikalische Zeitschrift 10 (1909) S. 104-111; Jahres-
bericht der Deutschen Mathematiker-Vereinigung 18 (1909) S. 75-88; reprinted in Gesam-
melte Abhandlungen von Hermann Minkowski, ed. by D. Hilbert, 2 vols. (Teubner, Leipzig
1911), vol. 2, pp. 431-444, and in H.A. Lorentz, A. Einstein, H. Minkowski, Das Rela-
tivitätsprinzip (Teubner, Leipzig 1913) S. 56-68. This lecture also appeared as a separate
publication (booklet): H. Minkowski, Raum und Zeit (Teubner, Leipzig 1909).
ii
lation, and the Dedication – the last paragraph of Otto Blumenthal’s Fore-
word to H. Minkowski, Zwei Abhandlungen über die Grundgleichungen der
Elektrodynamik (Teubner, Leipzig 1910). I translated the other two papers.
My initial intention was to retranslate (by making corrections wherever
necessary) the only English translation of Die Grundgleichungen für die
elektromagnetischen Vorgänge in bewegten Körpern done in 1920 by Saha5 ,
but since I was anyway checking every single sentence and also typesetting
the paper in LATEX I ended up with a virtually new translation.
Raum und Zeit was translated anew. I would like to thank Fritz Lew-
erto↵ for his invaluable advice on the translation of three difficult passages
and for his patience – our discussions often lasted between one and two
hours.
Montreal Vesselin Petkov

July 2012
5
The Principle of Relativity: Original Papers by A. Einstein and H. Minkowski, Trans-
lated into English by M.N. Saha and S.N. Bose with a Historical Introduction by P.C. Ma-
halanobis. (The University of Calcutta, Calcutta 1920). Minkowski’s paper Die Grund-
gleichungen für die elektromagnetischen Vorgänge in bewegten Körpern is translated in
this book under the name “Principle of Relativity” and the paper’s Appendix Mechan-
ics and the Relativity Postulate had been extended by the inclusion of the translation of
Minkowski’s paper Raum und Zeit.
Contents
Introduction 1
The not-fully-appreciated Minkowski . . . . . . . . . . . . . . . . . 1
Minkowski and Einstein . . . . . . . . . . . . . . . . . . . . . . . . 4
Minkowski and Poincaré . . . . . . . . . . . . . . . . . . . . . . . . 19
Minkowski and gravitation . . . . . . . . . . . . . . . . . . . . . . . 23
Minkowski and the reality of spacetime . . . . . . . . . . . . . . . 30
The Relativity Principle 39
The Fundamental Equations for Electromagnetic Processes in

Moving Bodies 51
Space and Time 111
iii
iv CONTENTS
Introduction
The not-fully-appreciated Minkowski

The major reason for the publication of Minkowski’s papers on relativity is
to correct an injustice which has been ongoing – Minkowski’s contributions
to modern physics have not been fully and appropriately appreciated. The
very fact that so far his papers on relativity have not been published together
either in German or English (and even his Das Relativitätsprinzip has never
been translated into English) is an indication of that.
Since the first publication1 in April 1908 of Minkowski’s mathematical
formalism of what he regarded as a theory of an absolute four-dimensional
world there have been never stopping attempts to downplay his revolutionary
contributions to the modern spacetime physics. Here are several examples:
• Unfortunately, it was Einstein himself (with Jakob Laub) who ex-

pressed the first documented2 reservation towards Minkowski’s four-
dimensional physics. Einstein and Laub indicated in the first para-
graph of their first paper on Minkowski’s study Die Grundgleichungen
für die elektromagnetischen Vorgänge in bewegten Körpern that “In
view of the fact that this study makes rather great demands on the
reader in its mathematical aspects, we do not consider it superfluous to
derive here these important equations in an elementary way, which, is,
by the way, essentially in agreement with that of Minkowski”3 . Ein-
1
H. Minkowski, Die Grundgleichungen für die elektromagnetischen Vorgänge in be-
wegten Körpern, Nachrichten der K. Gesellschaft der Wissenschaften zu Göttingen.
Mathematisch-physikalische Klasse (1908) S. 53-111. This is the lecture Minkowski gave
at the meeting of the Göttingen Scientific Society on December 21, 1907.
2
A. Einstein, J. Laub, Über die elektromagnetischen Grundgleichungen für bewegte
Körper. Annalen der Physik 26 (1908) S. 532-540; Über die im elektromagnetischen
Felde auf ruhende Körper ausgeübten ponderomotorischen Kräfte. Annalen der Physik
26 (1908) S. 541-550.
3
The Collected Papers of Albert Einstein, Volume 2: The Swiss Years: Writings, 1900-
1
2 INTRODUCTION
stein called Minkowski’s approach “superfluous learnedness”4 (über-

flüssige Gelehrsamkeit). Also, Sommerfeld’s recollection of what Ein-
stein said on one occasion can provide further indication of his initial
attitude towards Minkowski’s development of the implications of the
equivalence of the times of observers in relative motion: “Since the
mathematicians have invaded the relativity theory, I do not under-
stand it myself any more”5 .
• Sommerfeld understood and accepted Einstein’s special relativity

thanks to Minkowski’s four-dimensional formulation. That is why it
is difficult to explain why he made changes to the original text of
Minkowski’s lecture Das Relativitätsprinzip given at the meeting of the
Göttingen Mathematical Society on November 5, 1907, which he pre-
pared for publication in 1915. Sommerfeld’s changes were favourable
to Einstein as Pyenson6 observed: “Sommerfeld was unable to resist
rewriting Minkowski’s judgement of Einstein’s formulation of the prin-
ciple of relativity. He introduced a clause inappropriately praising
Einstein for having used the Michelson experiment to demonstrate
that the concept of absolute space did not express a property of phe-
nomena. Sommerfeld also suppressed Minkowski’s conclusion, where
Einstein was portrayed as the clarifier, but by no means as the princi-
pal expositor, of the principle of relativity.” Giving credit to Einstein
for realizing the crucial role of the Michelson experiment is especially
unfortunate since Einstein himself stated the opposite: “In my own de-
velopment, Michelson’s result has not had a considerable influence. I
even do not remember if I knew of it at all when I wrote my first paper
on the subject (1905). The explanation is that I was, for general rea-
sons, firmly convinced that there does not exist absolute motion and
my problem was only how this could be reconciled with our knowledge
of electrodynamics. One can therefore understand why in my personal
1909 (Princeton University Press, Princeton 1989), p. 329.
4
A. Pais, Subtle Is the Lord: The Science and the Life of Albert Einstein (Oxford
University Press, Oxford 2005) p. 152.
5
A. Sommerfeld, To Albert Einstein’s Seventieth Birthday. In: Albert Einstein:
Philosopher-Scientist. P. A. Schilpp, ed., 3rd ed. (Open Court, Illinois 1969) pp. 99-
105, p. 102.
6
L. Pyenson, Hermann Minkowski and Einstein’s Special Theory of Relativity, Archive
for History of Exact Sciences 17 (1977) pp. 71-95, p. 82; see also L. Corry, Hermann
Minkowski and the Postulate of Relativity, Archive for History of Exact Sciences 51
(1997) p. 273-314, p. 276 and P. L. Galison, Minkowski’s Space-Time: From Visual
Thinking to the Absolute World, Historical Studies in the Physical Sciences, 10 (1979)
pp. 85-121, p. 93.
THE NOT-FULLY-APPRECIATED MINKOWSKI 3
struggle Michelson’s experiment played no role, or at least no deci-

sive role.”7 Minkowski’s view of the role of Einstein’s 1905 paper in
clarifying the physical meaning of the Lorentz transformations is ex-
pressed at the end of the first part of his 1908 paper The Fundamental
Equations for Electromagnetic Processes in Moving Bodies (see this
volume): “The paper of Einstein which has been cited in the Intro-
duction, has succeeded to some extent in presenting the nature of the
transformation from a physical standpoint.”
• Despite his initial negative reaction towards Minkowski’s four-dimensio-

nal physics Einstein relatively quickly realized that his revolutionary
theory of gravity would be impossible without the revolutionary con-
tributions of Minkowski. At the beginning of his 1916 paper on general
relativity Einstein wrote: “The generalization of the theory of relativ-
ity has been facilitated considerably by Minkowski, a mathematician
who was the first one to recognize the formal equivalence of space coor-
dinates and the time coordinate, and utilized this in the construction
of the theory.” This quote is hardly from the new 1997 translation.8
Quite strangely, the first page of the paper containing the recognition
of Minkowski’s work had been omitted in the first English translation.9
• Many physicists (including relativists) do not appear to have been fully

appreciating the depth of Minkowski’s four-dimensional physics and
his general explanation of relativistic phenomena – “The whole world
presents itself as resolved into such worldlines, and I want to say in
advance, that in my understanding the laws of physics can find their
most complete expression as interrelations between these worldlines”
(this volume). In 1960 Synge wrote: “It is to support Minkowski’s
way of looking at relativity that I find myself pursuing the hard path
of the missionary. When, in a relativistic discussion, I try to make
things clearer by a space-time diagram, the other participants look at
it with polite detachment and, after a pause of embarrassment as if
some childish indecency had been exhibited, resume the debate in their
own terms”10 . Now the situation does not appear to be that bad, but it
7
University Press, Oxford 2005) p. 172.
8
The Collected Papers of Albert Einstein, Volume 6: The Berlin Years: Writings,
1914-1917 (Princeton University Press, Princeton 1997), p. 146.
9
H. A. Lorentz et al., The Principle of Relativity, translated by W. Perrett and G. B.
Je↵ery (Methuen 1923; Dover repr., 1952).
10
J. L. Synge, Relativity: the general theory (North-Holland, Amsterdam 1960) p. IX.
4 INTRODUCTION
is not much better either – everyone can check how many kinematical
relativistic e↵ects are explained through spacetime diagrams in recent
textbooks on relativity. Given the fact that it is only Minkowski’s
four-dimensional physics that provides the correct explanations of the
relativistic e↵ects (see below and also the next section), it is difficult
to understand the reluctance and sometimes even resistance against
explaining the kinematical relativistic e↵ects as manifestations of the
four-dimensionality of the world as Minkowski advocated. A possible
but disturbing explanation may be an approach that appears to be
held by some physicists – that it is merely a matter of description
whether we will use Einstein’s or Minkowski’s versions of special rel-
ativity. I think such an approach is a sure recipe for a double failure
– in genuinely understanding physical phenomena and in making dis-
coveries in physics – because it is certainly not a matter of description
whether the world is three- or four-dimensional.
• There have been authors of books on general relativity, spacetime and

gravitation, including of recent (21st century) ones, who abundantly
use Minkowski’s four-dimensional mathematical formalism and space-
time concepts introduced by him, but in a whole book mention his
name just once, for example. I prefer not to give any references.
• What is also unfortunate is that some well-known physicists who write

papers and books for the general public virtually do not mention
Minkowski’s contributions and often omit even his name. As a re-
sult most who have read about spacetime appear to believe it was
introduced by Einstein.
• There have been claims by di↵erent authors that Minkowski did not
understand Einstein’s special relativity. The actual situation had been
just the opposite as will be shown in the next section.
Minkowski and Einstein

Let me make it clear right away – it is not my intention at all to try to
downplay Einstein’s contributions to special relativity. As stated at the
beginning of the Introduction the main purpose of this book is to correct
an injustice towards Minkowski, and an injustice cannot be corrected by
committing another injustice. I hope it would be fair to both Minkowski
and Einstein to shed some additional light (based on the historical facts we
MINKOWSKI AND EINSTEIN 5
know now) on what they knew and understood in the period 1905-1908. I
think the best approach in such situations is to imagine that they both were
alive and would read what is written about them.
Let me start with very brief information about Minkowski’s academic
background (Einstein’s background is well-known) and several facts.
In April 1883 the French Academy granted the Grand Prize in Mathe-
matics jointly to the eighteen year old Hermann Minkowski for his innovative
geometric approach to the theory of quadratic forms and to Henry Smith.
Thirteen years later, in 1896, Minkowski published his major work in math-
ematics The Geometry of Numbers.11
By 1905 Minkowski was already internationally recognized as an ex-
ceptional mathematical talent. At that time he became interested in the
electron theory and especially in an unresolved issue at the very core of
fundamental physics – at the turn of the nineteenth and twentieth century
Maxwell’s electrodynamics had been interpreted to show that light is an
electromagnetic wave, which propagates in a light carrying medium (the lu-
miniferous ether), but its existence was put into question since Michelson’s
interference experiments failed to detect the Earth’s motion in that medium.
Minkowski’s documented involvement with the electrodynamics of moving
bodies began in the summer of 1905 when he and his friend David Hilbert
co-directed a seminar in Göttingen on the electron theory. The paper of
Minkowski’s student – Einstein – on special relativity was not published
at that time; Annalen der Physik received the paper on June 30, 1905.
Poincaré’s longer paper “Sur la dynamique de l’électron” was not published
either; it appeared in 1906. Also, “Lorentz’s 1904 paper (with a form of the
transformations now bearing his name) was not on the syllabus.”12
Minkowski’s student Max Born, who attended the seminar in 1905, re-
called in 1959 what Minkowski had said during the seminar:13 “I remember
that Minkowski occasionally alluded to the fact that he was engaged with
the Lorentz transformations, and that he was on the track of new interre-
lationships.” Again Born wrote in his autobiography about what he had
heard from Minkowski after Minkowski’s lecture “Space and Time” given on
September 21, 1908:14 “He told me later that it came to him as a great shock
11
H. Minkowski, Geometrie der Zahlen (Teubner, Leipzig 1896).
12
S. Walter, Minkowski, Mathematicians, and the Mathematical Theory of Relativity,
in H. Goenner, J. Renn, J. Ritter, T. Sauer (eds.), The Expanding Worlds of General
Relativity, Einstein Studies, volume 7, (Birkhäuser, Basel 1999) pp. 45-86, p. 46.
13
Quoted from T. Damour, “What is missing from Minkowski’s “Raum und Zeit” lec-
ture”, Annalen der Physik 17 No. 9-10 (2008), pp. 619-630, p. 626.
14
M. Born, My Life: Recollections of a Nobel Laureate (Scribner, New York 1978) p.
6 INTRODUCTION
when Einstein published his paper in which the equivalence of the di↵erent
local times of observers moving relative to each other were pronounced; for
he had reached the same conclusions independently but did not publish
them because he wished first to work out the mathematical structure in all
its splendour. He never made a priority claim and always gave Einstein his
full share in the great discovery.”
These facts and especially the results of Minkowski’s publications are the
best proof that in the period 1905-1908 Minkowski had found a truly revolu-
tionary resolution of the difficult issues surrounding the electrodynamics of
moving bodies – that the relativity principle implies, as will be briefly sum-
marized below, that the Universe is a four-dimensional world with time as
the fourth dimension. Unfortunately, Minkowski had never indicated exactly
when he arrived at that discovery. In any case, it had been sufficiently long
before his December 1907 lecture The Fundamental Equations for Electro-
magnetic Processes in Moving Bodies when he presented the fully developed
mathematical formalism of the four-dimensional physics introduced by him
(a formalism that could not have been created in just several months).
So in the fall of 1907 Minkowski was the only one who had genuine
understanding of a number of difficult and unresolved at that time issues:
• The profound physical meaning of the relativity principle – that physi-

cal phenomena are the same for all inertial observers in relative motion.
As a mathematician it may have been easier for Minkowski (than for
Einstein) to postulate that the (real) time t of a stationary observer
and the abstract mathematical time t0 , which Lorentz introduced call-
ing it the local time of a moving observer, are equivalent and to ex-
plore the consequences of such a hypothesis. Unfortunately, we will
never know how Minkowski arrived at the idea that t and t0 should
be treated equally. What appears certain is that his path had been
di↵erent from Einstein’s. The mathematical way of thinking surely
had helped Minkowski to realize that if two observers in relative mo-
tion have di↵erent times they necessarily must have di↵erent spaces
as well (since space is perpendicular to time), which is impossible in a
three-dimensional world, but in a four-dimensional world with time as
the fourth dimension. Here is how Minkowski in his own words at his
lecture Space and Time explained how he had realized the profound
physical meaning of the relativity principle – that the world is four-
dimensional. In the case of two inertial reference frames in relative
131.
motion along their x-axes “one can call t0 time, but then must neces-
sarily, in connection with this, define space by the manifold of three
parameters x0 , y, z in which the laws of physics would then have exactly
the same expressions by means of x0 , y, z, t0 as by means of x, y, z, t.
Hereafter we would then have in the world no more the space, but an
infinite number of spaces analogously as there is an infinite number
of planes in three-dimensional space. Three-dimensional geometry be-
comes a chapter in four-dimensional physics” (this volume). Minkowski
suddenly found the answers to many questions in his four-dimensional
physics, e.g. the answer to the question of why the relativity principle
requires that physical phenomena be the same in all inertial reference
frames – this is so because every inertial observer describes the phe-
nomena in exactly the same way – in his own reference frame (i.e. in
terms of his own space and time) in which he is at rest. Also, the
answer to the question of the failure of Michelson’s experiments to
detect the motion of the Earth appears obvious – the Earth is at rest
with respect to its space and therefore not only Michelson’s but any
other experiments would confirm this state of rest. As every observer
always measures the velocity of light (and anything else) in his own
(rest) space and by using his own time, the velocity of light is the same
for all observers.
• Minkowski’s realization that the relativity principle implies many times

and spaces, which in turn implies that the world is four-dimensional,
naturally explained why there is no absolute motion (since there are
many spaces, not just one absolute space), and why there is a di↵erence
between inertial and accelerated motion (a body moving by inertia is
represented by a straight timelike worldline, whereas the worldline of
an accelerated body is curved). Minkowski found it necessary to stress
that “Especially the concept of acceleration acquires a sharply promi-
nent character” (this volume). This sharply prominent character of
the acceleration comes from the absolute geometric property of the
worldline of an accelerated body – the worldline of such a body is
curved (deformed); therefore the absoluteness of acceleration merely
reflects the absolute fact that the worldline an accelerating body is
curved (deformed) and does not imply an absolute space with respect
to which the body accelerates.
• Minkowski’s four-dimensional physics allowed him not only to explain

the physical meaning of length contraction, but to realize clearly that,
8 INTRODUCTION
exactly like the relativity principle, that e↵ect is also a manifestation

of the four-dimensionality of the world.
• In his four-dimensional physics Minkowski found that pairs of ordinary

mechanical quantities are in fact space and time components of four-
dimensional vectors and the ordinary electromagnetic quantities are
components of new types of four-dimensional structures.
Einstein won the race with his mathematics professor Minkowski (of
the existence of which neither of them suspected) and first published his
special relativity in 1905 in which he postulated the equivalence of t and
t0 . The realization of this equivalence took him many years and it came
as a result of the persistent analysis of his thought experiment of racing a
light beam. This thought experiment became a paradox for Einstein when
he studied Maxwell’s equations at the Polytechnic Institute in Zurich. In
Maxwell’s theory the velocity of light is a universal constant (c = (µ0 ✏0 ) 1/2 )
which meant for Einstein (due to his trust in “the truth of the Maxwell-
Lorentz equations in electrodynamics” and that they “should hold also in
the moving frame of reference.”15 ) that if he travelled almost at the speed
of light (relative, say, to Earth), a beam of light would still move away from
him at velocity c, which is in Einstein’s own words “in conflict with the
rule of addition of velocities we knew of well in mechanics”16 Later Einstein
acknowledged that “the germ of the special relativity theory was already
present in that paradox”17 and explained that his “solution was really for the
very concept of time, that is, that time is not absolutely defined but there
is an inseparable connection between time and the signal velocity. With
this connection, the foregoing extraordinary difficulty could be thoroughly
solved. Five weeks after my recognition of this, the present theory of special
relativity was completed.”18
Einstein’s realization that inertial observers in relative motion have dif-
ferent times had been accomplished through conceptual analyses à la Galileo.
The development of this powerful method had later helped Einstein to make
one of the greatest discoveries in the intellectual history of our civilization
– that gravitational phenomena are not caused by gravitational forces but
are a manifestation of the non-Euclidean geometry of spacetime. However,
in 1905 Einstein still did not understand fully all implications of his major
15
University Press, Oxford 2005) p. 139
16
A. Pais, Ibid.
17
A. Folsing, Albert Einstein: A Biography (Penguin Books, New York 1997) p. 166
18
A. Pais, Ibid.
discovery that t and t0 should be treated equally. As a result, at that time

and at least in the following several years Einstein did not have complete
understanding of the above list of issues which Minkowski clarified in 1907
and 1908. For example, unlike Minkowski Einstein had to postulate the rel-
ativity principle without being able to explain its physical meaning. He also
simply stated that the luminiferous ether was superfluous without any ex-
planation, that is, he merely postulated that absolute motion does not exist.
Einstein did not have the correct understanding of the physical meaning of
length contraction either since at that time he had not yet fully understood
and adopted Minkowski’s four-dimensional physics.
One of the indications that Einstein did not fully comprehend the im-
plications of the fact that observers in relative motion have di↵erent times
is the very name of his theory – the theory of relativity. Einstein believed
that all uniform motion is relative, whereas Minkowski demonstrated that
that relativity is a manifestation of (or implies) an absolute four-dimensional
world. What is even worse, is that Einstein insisted on relativity as the core
concept of his theories and called his revolutionary theory of gravitation
the general theory of relativity, which is a further indication of his slow ac-
ceptance of Minkowski’s four-dimensional physics. As Synge remarked19
Minkowski “protested against the use of the word ’relativity’ to describe a
theory based on an ’absolute’ (space-time), and, had he lived to see the gen-
eral theory of relativity, I believe he would have repeated his protest in even
stronger terms.”
It is well known that Einstein was “for general reasons, firmly convinced
that there does not exist absolute motion”20 and that Einstein regarded
all motion as relative mostly due to Mach. And indeed Einstein kept the
term “relativity” in his general theory because he believed that in that the-
ory acceleration should also be treated as relative. In his 1914 paper The
Formal Foundation of the General Theory of Relativity 21 Einstein repeated
and extended Mach’s argument for a relative acceleration. This fact alone
is sufficient to demonstrate that even in 1914 Einstein had not fully un-
derstood Minkowski’s spacetime physics.22 As indicated above Minkowski
19
J. L. Synge, Relativity: the general theory (North-Holland, Amsterdam 1960) p. IX.
20
A. Pais, loc. cit., p. 172.
21
The Collected Papers of Albert Einstein, Volume 6: The Berlin Years: Writings,
1914-1917 (Princeton University Press, Princeton 1997) p. 31.
22
However, later in his life Einstein seems to have fully realized the implications of
spacetime not only for physics but for our entire worldview as well (see last section).
Regarding Mach, Einstein wrote in 1954: “As a matter of fact, one should no longer speak
of Mach’s principle at all” (A. Pais, loc. cit., p. 288).
10 INTRODUCTION
particularly pointed out the prominent character of the concept of accelera-

tion since the acceleration’s absoluteness comes from the absolute fact that
the worldline of an accelerating body is curved (deformed). It is true that
Minkowski’s explanation of the absoluteness of acceleration was given for the
case of flat spacetime, whereas in 1914 Einstein was completing his theory of
general relativity. However, the situation regarding the absoluteness of ac-
celeration is exactly the same in the case of curved spacetime (i.e. in general
relativity) – a body moving by inertia is represented by a geodesic world-
line (which is the analog of a straight worldline in curved spacetime since
it is curved only due to the curvature of spacetime, but is not additionally
curved, i.e. it is not deformed), whereas an accelerating body is represented
by a deformed (non-geodesic) worldline. Therefore acceleration in both flat
and curved spacetime is absolute which demonstrate that Mach’s view of
relative acceleration is clearly wrong. Here is a concrete example to see why
this is so. Mach argued that one could not say anything about the state of
motion of a single particle in the Universe since he believed that one can
talk only about motion relative to another body. However, that situation
is crystal clear in Minkowski’s spacetime physics – the worldline of a single
particle in the Universe is either geodesic or deformed, which means that
the particle is either moving by inertia or accelerating.
Despite the difficulties Einstein had had with understanding and adopt-
ing Minkowski’s spacetime physics, the mastering of the method of con-
ceptual analyses involving thought experiments helped him draw all three-
dimensional implications of the equivalence of the times of observers in rela-
tive motion. For example, the thought experiments led Einstein to the rela-
tion between mass and energy E = mc2 which now bears his name although
it was discovered before him in the framework of the electron theory.23
In view of all these facts it is inexplicable how could anyone say that
Minkowski had not understood Einstein’s 1905 paper on special relativity.
I will give two examples which are even more inexplicable since they come
from the authors of two very informative and otherwise excellent papers.
In 1979 Galison24 wrote: “At this early time (1907) it is clear that
Minkowski did not understand the import of Einstein’s theory.” As we have
seen the actual situation had been just the opposite. Galison had in mind
23
When it was initially derived in the electron theory that expression contained the
famous factor of 4/3, which was later accounted for; see V. Petkov, Relativity and the
Nature of Spacetime, 2nd ed. (Springer, Heidelberg 2009) Chap. 9, particularly Sec. 9.3
and the references therein.
24
P. L. Galison, Minkowski’s Space-Time: From Visual Thinking to the Absolute World,
Historical Studies in the Physical Sciences, Vol. 10 (1979) pp. 85-121, p. 93.
Minkowski’s enthusiasm for arriving at an electromagnetic picture of the

world based on his world postulate and the electron theory as suggested by
the last paragraph of Space and Time (this volume): “The validity with-
out exception of the world postulate is, I would think, the true core of an
electromagnetic world view which, as Lorentz found it and Einstein further
unveiled it, lies downright and completely exposed before us as clear as day-
light.” First, not only in 1907 but also in 1908 (when Space and Time was
presented in Cologne) Minkowski had the same view; moreover his Cologne
lecture essentially explained in a non-technical language the main results of
his lecture given on December 21, 1907. And I do not see anything wrong
with Minkowski’s hope for a unified world picture; at that time the other
fundamental interactions were unknown, so it was perfectly natural to try
to find a unified picture of the world on the basis of what was known. Most
important, however is the following. If “Minkowski did not understand the
import of Einstein’s theory” because he was positively looking at the electron
theory, then by exactly the same argument Einstein did not understand the
import of his own theory. In January 1909 Einstein wrote25 “In conclusion, I
would also like to point to the importance of the recently published paper by
Ph. Frank, which, by taking into account the Lorentz contraction, restores
the agreement between Lorentz’s treatment, based on the electron theory,
and Minkowski’s treatment of the electrodynamics of moving bodies. The
advantage of the treatment based on the electron theory consists, on the
one hand, in providing a graphic interpretation of the field vectors and, on
the other hand, in dispensing with the arbitrary assumption that the deriva-
tives of the velocity of matter do not appear in the di↵erential equations.”
As seen from this quote, in 1909 Einstein viewed “Minkowski’s treatment
of the electrodynamics of moving bodies” as di↵erent from Lorentz’ treat-
ment “based on the electron theory” and pointed out the “advantage of the
treatment based on the electron theory.”
Now the prevailing view is that the electron theory was wrong. I am
afraid that that is rather a simplistic view. It is now clear what in the
electron theory was undoubtedly wrong – e.g. the electron is not a small
charged sphere. A completely wrong theory cannot make a number of cor-
rect predictions – e.g. the electron theory predicted that the electron mass
increases as the electron’s velocity increases before the theory of relativity,
yielding the correct velocity dependence, and that the relation between en-
25
A. Einstein, Comment on the paper of D. Mirimano↵ “On the Fundamental Equa-
tions. . . ” Annalen der Physik 28 (1909) pp. 885-888. In: The Collected Papers of Albert
Einstein, Volume 2: The Swiss Years: Writings, 1900-1909 (Princeton University Press,
Princeton 1989), p. 356.
12 INTRODUCTION
ergy and mass is E = mc2 . That is why it is maybe more appropriate to say
that today “the state of the classical electron theory reminds one of a house
under construction that was abandoned by its workmen upon receiving news
of an approaching plague. The plague in this case, of course, was quantum
theory. As a result, classical electron theory stands with many interesting
unsolved or partially solved problems.”26
Unfortunately, exactly a hundred years after Minkowski’s lecture Space
and Time Damour27 wrote: “First, I would like (after many others. . . )
to stress that Minkowski probably did not really comprehend the concep-
tual novelty of Einstein’s June 1905 paper on Special Relativity, and espe-
cially the results therein concerning time. Indeed, in his Cologne lecture
Minkowski says that, while Einstein “deposed [time] from its high seat”,
“neither Einstein nor Lorentz made any attack on the concept of space. . . ”
However, this was precisely one of the key new insights of Einstein, namely
the relativity of simultaneity!”
Now, thanks to Minkowski, we know that relativity of simultaneity does
imply many spaces since a space constitutes a class of simultaneous events –
two observers in relative motion have di↵erent classes of simultaneous events
and therefore di↵erent spaces and vice versa (as Minkowski discovered two
observers in relative motion have di↵erent spaces and therefore di↵erent
classes of simultaneous events). However, in 1905 Einstein was totally un-
aware of this. He had been occupied with the idea of time and how to
measure times and distances. Even a quick look at how Einstein arrived at
the idea of relativity of simultaneity in his 1905 paper shows that he did that
in an operational way – by analyzing the procedure of synchronizing distant
clocks through light signals; relativity of simultaneity follows immediately
from the fact that the velocity of light is c for all observers. That is why
Einstein himself had never claimed that he had realized that observers in
relative motion have di↵erent spaces. On the contrary, as indicated above
three years after his 1905 paper (in May 1908) he reacted negatively towards
the introduced by Minkowski absolute four-dimensional world and therefore
negatively towards the very idea of many spaces since it was the idea of
many spaces that led Minkowski to the absolute four-dimensional world. As
we saw above Minkowski’s geometrical approach helped him to realize first
that as observers in relative motion have di↵erent times they necessarily
must have di↵erent spaces as well, and then he had probably immediately
26
P. Pearle, Classical Electron Models. In: Electromagnetism: Paths to Research, ed.
by D. Teplitz (Plenum Press, New York 1982) pp. 211-295, p. 213.
27
T. Damour, “What is missing from Minkowski’s “Raum und Zeit” lecture”, Annalen
der Physik. 17, No. 9-10, (2008) pp. 619-630, p. 627.
seen that many spaces imply an absolute four-dimensional world.

As unfounded as the statement above (that Einstein had discovered that
observers in relative motion have di↵erent spaces), is another statement in
Damour’s article:
In addition, when Minkowski introduces the (geometrically moti-

vated) concept of proper time, he does not seem to fully grasp its
physical meaning. However, this is the second key new insight
brought in by Einstein concerning time, namely the fact (ex-
plicitly discussed by Einstein) that, when comparing a moving
clock to one remaining at rest (and marking the corresponding
‘rest’ coordinate time t), the moving clock will mark (upon being
reconvened with the sedentary clock) the time
Z p
⌧ = dt 1 v 2 /c2
i.e. Minkowski’s proper time. It seems that Minkowski was not

aware of this.
Minkowski was certainly aware of this expression without the integral

(there is no integral in Einstein’s paper as Damour admits but in a foot-
note) – on October 9, 1907 he wrote to Einstein to request a copy of his
1905 paper.28 Damour’s suggestion that Minkowski might have misread the
paper – “This is another example of a scientist misreading a paper which he
knew, however, to be central to his research topic!”29 – seems virtually im-
possible since “Minkowski had written to Einstein asking for a reprint of his
1905 paper, in order to study it in his joint seminar with Hilbert”30 (could
Minkowski have misread a key paper that had been studied at the seminar
he co-directed with Hilbert?).
What is most important, however, is that, like the above issue of many
spaces, Damour again seems to read more in Einstein’s 1905 paper. Ein-
stein had completed that paper only five weeks after he had realized the
equivalence of the times of observers in relative motion and had been still
struggling with its consequences. By contrast, Minkowski seems to have had
28
29
T. Damour, loc. cit., p. 627.
30
L. Corry, Hermann Minkowski and the Postulate of Relativity, Archive for History of
Exact Sciences 51 (1997) p. 273-314, p. 276.
14 INTRODUCTION
more than two years to explore those consequences – Minkowski appears ho

have realized independently the equivalence of the times of observers in rel-
ative motion almost certainly as late as the summer of 1905.31 The best
proof that Minkowski fully understood the physical meaning of proper time
(which is quite natural given that this concept was introduced by himself) is
the fact that the modern introduction and definition of proper time is iden-
tical to that of Minkowski. Only an in-depth and complete understanding
of the new concepts of space and time and their union made their introduc-
tion and definition so precise that they remained unchanged more than a
hundred years later. As this should be self-evident since it was Minkowski
who thoroughly developed these new concepts it is inexplicable why not
only did Damour make the above claim but found it necessary to repeat
it: “Minkowski did not fully grasp the physical meaning of what he was
doing.”32
Minkowski’s understanding of the physical meaning of time and space-
time had been so deep that with the introduction of proper time he essen-
tially demonstrated that an observer should use two times in his rest frame
– proper and coordinate times (⌧ and t) – which provided the correct physi-
cal treatment of time (i) in accelerated reference frames in special relativity,
and later (ii) in general relativity. Minkowski did not call the time t coordi-
nate time, but the presence of the two times in the same reference frame is
obvious from the way he defined proper time (this volume):
1p 2 2
d⌧ = c dt dx2 dy 2 dz 2 .
c
The expression c2 dt2 dx2 dy 2 dz 2 is the interval (the spacetime distance)
ds2 (in a reference frame) between the two infinitesimally close events on the
worldline of a particle; the length of the worldline between these events is
the proper time d⌧ . If the particle’s worldline is straight, which means that
the particle moves with constant velocity, in its inertial reference frames
proper and coordinate times coincide. However, if the particle accelerates,
31
There are two indications of that which cannot be merely ignored. First, Born’s
recollection quoted in the first section that Minkowski had been shocked when Einstein’s
paper appeared in 1905; there is no reason whatsoever to suspect that Born would invent
such a recollection (moreover, he had another recollection, as indicated also in the first
section, which supports it). Second, what is far more important, however, is the full-blown
four-dimensional formalism Minkowski reported on December 21, 1907 and the depth of his
understanding of the electrodynamics of moving bodies and the absolute four-dimensional
world; such a revolution in both physics and mathematics could not have been possible if
he had merely developed others’ ideas.
32
its worldline is curved and an observer in the particle’s accelerating frame

should use both proper and coordinate times.
If Damour had insisted on keeping in his paper the repeated unfortu-
nate expression “did not fully grasp the physical meaning of what he was
doing,” he should have used it for Einstein’s understanding of the physical
meaning of the time (in the case discussed by Damour) which Minkowski
later called proper time (but that would have been equally unfair since as
indicated above Einstein completed his 1905 paper only five weeks after his
profound insight that the times of observers in relative motion should be
treated equally). In the above calculation quoted by Damour, Einstein de-
termined the time of a clock in circular motion: “If there are two synchronous
clocks in A, and one of them is moved along a closed curve with constant
velocity33 until it has returned to A, which takes, say, t sec, then this clock
will lag on its arrival at A 12 t(v/V )2 sec behind the clock that has not been
moved.”34 Einstein arrived at this result by using the Lorentz transforma-
tion of the times of two inertial clocks in relative motion, which generally
deals with coordinate time. As coordinate and proper time coincide in iner-
tial reference frames (moving with constant velocity) no misunderstanding
is likely. But in an accelerating reference frame coordinate and proper time
do not coincide. When Einstein compared the times of the accelerating clock
(moving along the closed curve) and the stationary clock he used what was
later called the proper time of the accelerating clock without having any idea
that that time is a second time in the reference frame of the accelerating
clock, which is di↵erent from the coordinate time (Minkowski introduced
the concept of proper time more than two years later).
Damour further wrote35 that Minkowski “had (seemingly) not fully
grasped the striking result of Einstein that proper time along any polyg-
onal (or curved) time-like line between two points in spacetime is smaller
than the proper time along the straight line joining the two points. If he
had realized it clearly, he would have commented that this is just the oppo-
site of the usual triangular inequality.” First, the wording of “the striking
result of Einstein that proper time...” is inappropriate – it is well known and
indicated above that in 1905 Einstein could not have had any idea of what
33
Even in the new translation of Einstein’s 1905 paper the German word Geschwindigkeit
has been again erroneously translated in this sentence as velocity. Obviously, the velocity
of the clock along a closed curve is not constant; what is constant is the clock’s speed.
34
A. Einstein, On the electrodynamics of moving bodies, The Collected Papers of Albert
Einstein, Volume 2: The Swiss Years: Writings, 1900-1909 (Princeton University Press,
Princeton 1989), p. 153.
35
16 INTRODUCTION
proper time is. Second, as Minkowski defined proper time as a length along
a timelike worldline he knew perfectly what proper time is, and it is indeed
a valid question why he did not define the triangle inequality in spacetime
as well.
I think the most probable explanation is that since he had been com-
pletely occupied with developing the spacetime physics and its four-dimensional
mathematical formalism his first priority had been (as seen from his three
papers) the electrodynamics of moving bodies. The work on the kinemat-
ical consequences of the absolute four-dimensional world (e.g. the special
role of acceleration stressed by Minkowski) had been scheduled for later as
Minkowski clearly alluded to such a plan: “The whole world presents it-
self as resolved into such worldlines, and I want to say in advance, that in
my understanding the laws of physics can find their most complete expres-
sion as interrelations between these worldlines” (this volume). The triangle
inequality is clearly such an interrelation between worldlines.
To expect more from someone who had already done so much for such a
short period of time, and who would have indisputably done even more, if he
had not been taken away from us when he was at the peak of his intellectual
strength, is very unfair.
It is important to stress that after his initial hostile attitude towards
Minkowski’s spacetime physics Einstein gradually adopted it since it was
essential for his general relativity. In 1946 in his Autobiography Einstein
summarized Minkowski’s main contribution:36
Minkowski’s important contribution to the theory lies in the

following: Before Minkowski’s investigation it was necessary to
carry out a Lorentz-transformation on a law in order to test its
invariance under such transformations; he, on the other hand,
succeeded in introducing a formalism such that the mathemati-
cal form of the law itself guarantees its invariance under Lorentz-
transformations. By creating a four-dimensional tensor-calculus
he achieved the same thing for the four-dimensional space which
the ordinary vector-calculus achieves for the three spatial dimen-
sions. He also showed that the Lorentz-transformation (apart
from a di↵erent algebraic sign due to the special character of
time) is nothing but a rotation of the coordinate system in the
four-dimensional space.
36
A. Einstein, “Autobiographical notes.” In: Albert Einstein: Philosopher-Scientist.
Paul A. Schilpp, ed., 3rd ed. (Open Court, Illinois 1969) pp. 1-94, p. 59.
As seen from his estimation of Minkowski’s contribution Einstein did not

explicitly credit Minkowski for demonstrating that the relativity postulate
and length contraction imply an absolute four-dimensional world; we will
return to this point in the last section. On the other hand, Einstein credited
Minkowski for showing that the Lorentz transformations are rotations in
spacetime, whereas it was Poincaré who first published that result in 1906.37
Let me stress it one more time – Einstein’s achievements speak for them-
selves, so no one can downplay his contributions. I think Minkowski’s four-
dimensional (spacetime) physics and Einstein’s discovery that gravity is a
manifestation of the spacetime curvature will forever remain as the two
greatest intellectual achievements. The approaches of Minkowski and Ein-
stein are distinctly di↵erent, but they both proved to be so extraordinarily
productive that should become integral parts of the way of thinking of any
scientist who works on the front line of research in any field. Minkowski’s
and Einstein’s proven but not fully studied approaches form the core of a
research strategy that is being developed and will be employed at a new re-
search institute – Institute for Foundational Studies ‘Hermann Minkowski’
(http://minkowskiinstitute.org/).
In addition, I have a personal reason not to even think of downplaying
Einstein’s contributions. I have always admired him for the way he arrived
at his two theories – by employing and extending Galileo’s way of doing
physics through conceptual analyses and thought experiments. Moreover,
my own way of thinking about physical phenomena was consciously formed
by studying the methods of great physicists which led them to ground-
breaking discoveries, particularly those of Galileo and Einstein; much later
I discovered and started to appreciate thoroughly Minkowski’s approach to
physics.
Also, I fully share Einstein’s firm position that quantum mechanics does
not provide a complete description of the quantum world in a sense that it
does not contain a model of the quantum object itself. I believe a theory that
describes only the state of something, not the something itself, is intrinsically
incomplete. As now no one can seriously question the probabilistic nature
of quantum phenomena it appears easily tempting to state that Einstein’s
intuition that God does not play dice was wrong. I think such a temptation
will remain baseless until we understand what the quantum object is.
Leaving aside the issue of whether God would care about a human’s
opinion on how he should behave, just imagine the following (very prob-
37
H. Poincaré, Sur la dynamique de l’électron. Rendiconti del Circolo matematico Ren-
diconti del Circolo di Palermo 21 (1906) pp.129-176.
18 INTRODUCTION
able in my view) development in quantum physics, which may reveal an

unanticipated meaning of Einstein’s intuition. As Galileo’s and Einstein’s
conceptual analyses (which proved to be physics at its best) are now almost
explicitly regarded as old-fashioned (no leading physics journal would pub-
lish a paper containing a deep conceptual analysis of an open question), it
is not surprising that the so called quantum paradoxes remained unresolved
almost a century after the advent of quantum mechanics.
Despite Feynman’s desperate appeal to regard Nature as absurd38 the
history of science teaches us that all apparent paradoxes are caused by some
implicit assumptions. A consistent conceptual analysis of only one of those
quantum mechanical paradoxes – say, the famous double-slit experiment,
discussed by Feynman – almost immediately identifies an implicit assump-
tion39 – we have been taking for granted that quantum objects exist contin-
uously in time although there has been nothing either in the experimental
evidence or in the theory that compels us to do so. Just imagine – a fun-
damental continuity (continuous existence in time) at the heart of quantum
physics. And no wonder that such an implicit assumption leads to a paradox
– an electron, for example, which is always registered as a pointlike entity
and which exists continuously in time, is a classical particle (i.e. a world-
line in spacetime) that cannot go simultaneously through both slits in the
double-slit experiment to form an interference pattern.
However, if we abandon the implicit assumption and replace it explicitly
with its alternative – discontinuous existence in time – the paradox disap-
pears. Then an electron is, in the ordinary three-dimensional language, an
ensemble40 of constituents which appear-disappear ⇠ 1020 times per second
(the Compton frequency). Such a quantum object can pass simultaneously
through all slits at its disposal.
In Minkowski’s four-dimensional language (trying to extract more from
his treasure), an electron is not a worldline but a “disintegrated” worldline
whose worldpoints are scattered all over the spacetime region where the
electron wavefunction is di↵erent from zero. Such a model of the quantum
object and quantum phenomena in general provides a surprising insight
38
Feynman wrote: “The theory of quantum electrodynamics describes Nature as absurd
from the point of view of common sense. And it agrees fully with experiment. So I hope
you can accept Nature as She is – absurd.” R. P. Feynman, QED: The Strange Theory of
Light and Matter (Princeton University Press, Princeton 1985) p. 10.
39
V. Petkov, Relativity and the Nature of Spacetime, 2nd ed. (Springer, Heidelberg
2009) Chap. 10.
40
A. H. Anastassov, Self-Contained Phase-Space Formulation of Quantum Mechanics as
Statistics of Virtual Particles, Annuaire de l’Universite de Sofia “St. Kliment Ohridski”,
Faculte de Physique 81 (1993) pp. 135-163.
MINKOWSKI AND POINCARÉ 19
into the physical meaning of probabilistic phenomena in spacetime – an

electron is a probabilistic distribution of worldpoints which is forever given
in spacetime.
Had Minkowski lived longer he might have described such a spacetime
picture by the mystical expression “predetermined probabilistic phenomena.”
And, I guess, Einstein would be also satisfied – God would not play dice since
a probabilistic distribution in spacetime exists eternally there.
Minkowski and Poincaré

This section was the most difficult to write since I have not found any clue
of how Minkowski would have explained the obvious fact – that Poincaré
was not mentioned in his Cologne lecture Space and Time. Minkowski was
certainly aware of Poincaré’s paper Sur la dynamique de l’électron published
in 1906 (but received by Rendiconti del Circolo matematico Rendiconti del
Circolo di Palermo on July 23, 1905) since he quoted it in his previous lec-
tures given in November and December 1907. In his paper Poincaré first
published the important result that the Lorentz transformations had a ge-
ometric interpretation as rotations in what he seemed to have regarded as
an abstract four-dimensional space with time as the fourth dimension.41
Here are two attempts to explain Minkowski’s omission to mention
Poincaré’s paper in his Cologne lecture.
In the absence of any clear indication why Minkowski left Poincaré

out of his lecture, a speculation or two on his motivation may
be entertained. If Minkowski had chosen to include some men-
tion of Poincaré’s work, his own contribution may have appeared
derivative. Also, Poincaré’s modification of Lorentz’s theory of
electrons constituted yet another example of the cooperative role
played by the mathematician in the elaboration of physical the-
ory. Poincaré’s “more mathematical” study of Lorentz’s electron
theory demonstrated the mathematician’s dependence upon the
insights of the theoretical physicist, and as such, it did little to es-
tablish the independence of the physical and mathematical paths
to the Lorentz group. The metatheoretical goal of establishing
the essentially mathematical nature of the principle of relativ-
ity was no doubt more easily attained by neglecting Poincaré’s
41
H. Poincaré, Sur la dynamique de l’électron, Rendiconti del Circolo matematico Ren-
diconti del Circolo di Palermo 21 (1906) pp. 129-176.
20 INTRODUCTION
elaboration of this principle.42
My conjecture is that Minkowski, helped by his background read-

ing of some of the works of Lorentz and Poincaré (which, how-
ever, did not include their most recent contributions of 1904-
1905. . . ) had discovered by himself, in the summer of 1905 (with-
out knowing about the 1905 papers of Poincaré) the fact that
Lorentz transformations preserve the quadratic form c2 t2 +
x2 (+y 2 + z 2 ). If that reconstruction is correct, he must have
been all the more eager, when he later realized that he had been
preceded by Poincaré, to find reasons for downplaying Poincaré’s
work.43
I think one should also ask why in 1946 in his Autobiography44 (as quoted
in Section 2) Einstein wrote that Minkowski “showed that the Lorentz-
transformation [. . . ] is nothing but a rotation of the coordinate system in
the four-dimensional space.” It seems Einstein was either unaware in 1946
(which is highly unlikely) of the fact that it was Poincaré who first published
that result, or he knew (perhaps from Born) that Minkowski independently
had made the same discovery.
Another interesting fact is that not someone else but a famous French
physicist credited Minkowski for the discovery of spacetime. In 1924 Louis de
Broglie wrote in his doctoral thesis Recherches sur la théorie des quanta:45
“Minkowski showed first that one obtains a simple geometric representa-
tion of the relationships between space and time introduced by Einstein by
42
43
T. Damour, What is missing from Minkowski’s “Raum und Zeit” lecture, Annalen der
Physik. 17, No. 9-10, (2008) pp. 619-630, p. 626.
44
A. Einstein, “Autobiographical notes.” In: Albert Einstein: Philosopher-Scientist.
Paul A. Schilpp, ed., 3rd ed. (Open Court, Illinois 1969) pp. 1-94, p. 59.
45
“Minkowski a montré le premier qu’on obtenait une représentation géométrique sim-
ple des relations de l’espace et du temps introduites par Einstein en considérant une
multiplicité euclidienne à 4 dimensions dite Univers ou Espace-temps,” Louis de Broglie,
Recherches sur la théorie des quanta, Réédition du texte de 1924. (Masson, Paris 1963),
p. 27. Strangely, the word “appears” (which is clearly not in the original French text)
had been inserted into the sentence translated into English by Kracklauer: “Minkowski
appears to have been first to obtain a simple geometric representation of the relationships
introduced by Einstein between space and time consisting of a Euclidian 4-dimensional
space-time,” Louis-Victor de Broglie, On the Theory of Quanta, translated by A. F.
Kracklauer (2004); available at the website of Annales de la Fondation Louis de Broglie
(http://aflb.ensmp.fr/LDB-oeuvres/De_Broglie_Kracklauer.htm).
MINKOWSKI AND POINCARÉ 21
considering an Euclidean manifold of 4 dimensions called Universe or space-

time.” Another contemporary French physicist – Thibault Damour (quoted
above) – also thinks that “the replacement of the separate categories of space
and time with the new physical category of space-time is [. . . ] more properly
attributed to Hermann Minkowski and not to Poincaré.”46
Probably we will never learn why Minkowski did not quote Poincaré
in his lecture Space and Time in 1908. However, a similar question ap-
plies to Poincaré himself: “In the lecture Poincaré delivered in Göttingen on
the new mechanics in April 1909, he did not see fit to mention the names
of Minkowski and Einstein.”47 Poincaré could have used the fact that his
lecture was only around three months after Minkowski’s death to credit
Minkowski for fully developing the four-dimensional physics based on the
idea of spacetime which Poincaré first published.
I think the discovery of spacetime is a doubly sad story. First, unlike
Minkowski, Poincaré seems to have seen nothing revolutionary in the idea of
a mathematical four-dimensional space as Damour remarked48 – “although
the first discovery of the mathematical structure of the space-time of spe-
cial relativity is due to Poincaré’s great article of July 1905, Poincaré (in
contrast to Minkowski) had never believed that this structure could really
be important for physics. This appears clearly in the final passage that
Poincaré wrote on the question some months before his death”:
Everything happens as if time were a fourth dimension of space,

and as if four-dimensional space resulting from the combination
of ordinary space and of time could rotate not only around an
axis of ordinary space in such a way that time were not altered,
but around any axis whatever. . .
What shall be our position in view of these new conceptions?
Shall we be obliged to modify our conclusions? Certainly not;
we had adopted a convention because it seemed convenient and
we had said that nothing could constrain us to abandon it. Today
some physicists want to adopt a new convention. It is not that
they are constrained to do so; they consider this new convention
more convenient; that is all. And those who are not of this
46
T. Damour, Once Upon Einstein, Translated by E. Novak (A. K. Peters, Wellesley
2006) p. 49.
47
48
22 INTRODUCTION
opinion can legitimately retain the old one in order not to disturb
their old habits. I believe, just between us, that this is what they
shall do for a long time to come.49
Poincaré believed that our physical theories are only convenient descrip-
tions of the world and therefore it is really a matter of convenience and
our choice which theory we would use. As Damour stressed it, it was “the
sterility of Poincaré’s scientific philosophy: complete and utter “convention-
ality” [. . . ] which stopped him from taking seriously, and developing as a
physicist, the space-time structure which he was the first to discover.”50
What makes Poincaré’s failure to comprehend the profound physical
meaning of the relativity principle and the geometric interpretation of the
Lorentz transformations especially sad is that it is perhaps the most cruel
example in the history of physics of how an inadequate philosophical position
can prevent a scientist, even as great as Poincaré, from making a discovery.
However, this sad example can serve a noble purpose. Science students and
young scientists can study it and learn from it because scientists often think
that they do not need any philosophical position for their research:
Scientists sometimes deceive themselves into thinking that philo-

sophical ideas are only, at best, decorations or parasitic com-
mentaries on the hard, objective triumphs of science, and that
they themselves are immune to the confusions that philosophers
devote their lives to dissolving. But there is no such thing as
philosophy-free science; there is only science whose philosophical
baggage is taken on board without examination.51
49
H. Poincaré, Mathematics and Science: Last Essays (Dernières Pensées), Translated
by J.W. Bolduc (Dover, New York 1963) pp. 23-24. Poincaré even appeared to have
thought that the spacetime convention would not be advantageous: “It quite seems, indeed,
that it would be possible to translate our physics into the language of geometry of four
dimensions. Attempting such a translation would be giving oneself a great deal of trouble
for little profit, and I will content myself with mentioning Hertz’s mechanics, in which
something of the kind may be seen. Yet, it seems that the translation would always be
less simple than the text, and that it would never lose the appearance of a translation, for
the language of three dimensions seems the best suited to the description of our world,
even though that description may be made, in case of necessity, in another idiom.” H.
Poincaré, Science and Method, In: The Value of Science: Essential Writings of Henri
Poincaré (Modern Library, New York 2001) p. 438.
50
51
D. C. Dennett, Darwin’s Dangerous Idea: Evolution and the Meanings of Life (Simon
and Schuster, New York 1996) p. 21.
MINKOWSKI AND GRAVITATION 23
Second, it seems virtually certain that Minkowski independently arrived

at two important results – (i) the equivalence of the times of observers in rel-
ative motion and (ii) the fact that the Lorentz transformations preserve the
quadratic form c2 t2 x2 y 2 z 2 and can therefore be regarded geometrically
as rotation in a four-dimensional space with time as the fourth dimension.
But these results were first published by Einstein and Poincaré, respectively.
As indicated in Section 2 the best proof that Minkowski, helped by his ex-
traordinary geometrical imagination, had made these discoveries indepen-
dently of Einstein and Poincaré, is the introduced by him four-dimensional
(spacetime) physics with a fully developed mathematical formalism and his
deep understanding of the new worldview and its implications. Born’s rec-
ollections given in Section 2 only confirm what follows from a careful study
of Minkowski’s results.
Minkowski and gravitation

On January 12, 1909 only several months after his Cologne lecture Space
and Time at the age of 44 Minkowski tragically and untimely departed from
this strange world (as Einstein would call it later). We will never know how
physics would have developed had he lived longer.
What seems undeniable is that the discovery of the true cause of gravita-
tion – the non-Euclidean geometry of spacetime – would have been di↵erent
from what actually happened. On the one hand, Einstein’s way of thinking
based on conceptual analyses and thought experiments now seems to be the
only way powerful enough to decode the unimaginable nature of gravitation.
However, on the other hand, after Minkowski had written the three papers
on relativity included here, he (had he lived longer) and his friend David
Hilbert might have formed an unbeatable team in theoretical physics and
might have discovered general relativity (surely under another name) before
Einstein.
As there is no way to reconstruct what might have happened in the
period 1909-1915 I will outline here what steps had been logically available
to Minkowski on the basis of his results. Then I will briefly discuss whether
their implications would lead towards the modern theory of gravitation –
Einstein’s general relativity.
In 1907 (most probably in November) Einstein had already been well
ahead of Minkowski when he made a gigantic step towards the new theory
of gravity:52
52
Quoted from: A. Pais, Subtle Is the Lord: The Science and the Life of Albert Einstein
24 INTRODUCTION
In the first row of this photograph (probably taken around 1905) are Minkowski
(left) David Hilbert’s wife, Käthe, and David Hilbert. Source: D. E. Rowe, A Look
Back at Hermann Minkowski’s Cologne Lecture “Raum und Zeit,” The Mathemat-
ical Intelligencer, Volume 31, Number 2 (2009), pp. 27-39.
I was sitting in a chair in the patent office at Bern when all of a

sudden a thought occurred to me: “If a person falls freely he will
not feel his own weight.” I was startled. This simple thought
made a deep impression on me. It impelled me toward a theory
of gravitation.
Einstein had been so impressed by this insight that he called it the

“happiest thought” of his life.53 And indeed this is a crucial point – at that
time Einstein had been the only human who realized that no gravitational
force acted on a falling body. Then he struggled eight years to come up with
a theory – his general relativity – according to which gravity is not a force
but a manifestation of the curvature of spacetime.
Here I will stress particularly the core of general relativity which reflects
Einstein’s “happiest thought” – the geodesic hypothesis according to which
a falling particle is not subject to a gravitational force. In other words,
the geodesic hypothesis in general relativity assumes that the worldline of
a free particle is a timelike geodesic in spacetime. The geodesic hypothesis
(Oxford University Press, Oxford 2005) p. 179.
53
A. Pais, Ibid.
is regarded as “a natural generalization of Newton’s first law,”54 that is,

“a mere extension of Galileo’s law of inertia to curved spacetime.”55 This
means that in general relativity a particle, whose worldline is geodesic, is a
free particle which moves by inertia.
The geodesic hypothesis has been confirmed by the experimental fact
that particles falling towards the Earth’s surface o↵er no resistance to their
fall – a falling accelerometer, for example, reads zero resistance (i.e. zero ac-
celeration; the observed apparent acceleration of the accelerometer is caused
by the spacetime curvature caused by the Earth). The experimental fact
that particles do not resist their fall (i.e. their apparent acceleration) means
that they move by inertia and therefore no gravitational force is causing their
fall. It should be emphasized that a gravitational force would be required to
accelerate particles downwards only if the particles resisted their accelera-
tion, because only then a gravitational force would be needed to overcome
that resistance.
Let us now imagine how Minkowski would have approached the issue
of gravitation. By analogy with Maxwell’s electrodynamics he had already
modified Newton’s gravitational theory in order that the speed of gravity
be equal to that of light c (Poincaré also proposed such a modification in
his 1906 paper on the dynamics of the electron). Now, thanks to the ge-
nius of Einstein, we know that electromagnetism is fundamentally di↵erent
from gravitation – electromagnetic phenomena are caused by electromag-
netic forces, whereas gravitational phenomena are manifestation of the non-
Euclidean geometry of spacetime which means that there are no gravitational
forces in Nature.
The natural question is whether Minkowski would have found any reasons
to revise his modified version of Newton’s theory of gravity. Perhaps many
physicists would say ‘highly unlikely.’ And they might be right. But looking
at what Minkowski had achieved for so short a period of time, I think his
genius should never be underestimated (even because that would constitute
a contradiction in terms). Let us see what logical options Minkowski had
after his third lecture Space and Time.
Minkowski had been aware of two relevant facts – (i) the motion of par-
ticles with constant velocity cannot be detected experimentally since the
particles move non-resistantly, i.e. by inertia (in other words, an experi-
ment always detects the lack of resistance of an inertial particle, and in this
54
J. L. Synge, Relativity: the general theory. (Nord-Holand, Amsterdam 1960) p. 110.
55
W. Rindler, Relativity: Special, General, and Cosmological (Oxford University Press,
Oxford 2001) p. 178.
26 INTRODUCTION
sense inertial motion is absolute or frame-independent), and (ii) the acceler-

ated motion of a particle can be discovered experimentally since the particle
resists its acceleration (so accelerated motion is also absolute in this sense
and therefore frame-independent).
The accelerated motion had already been causing problems after the
publication of Einstein’s special relativity in 1905 since it appeared that the
experimental detection of accelerated motion provided experimental sup-
port for the absolute space – if a particle’s acceleration is absolute (since
it is measurable), then such an acceleration is with respect to the absolute
space, which contradicts both Einstein’s special relativity and particularly
Minkowski’s interpretation of the relativity principle according to which ob-
servers in relative motion have di↵erent times and spaces (whereas an abso-
lute space implies a single space).
However, Minkowski had not been concerned about such an apparent
contradiction at all. He provided rigorous criteria for inertial and acceler-
ated motion56 – a free particle, which moves by inertia, is a straight timelike
worldline in Minkowski spacetime, whereas the timelike worldline of an ac-
celerating particle is clearly di↵erent – it is curved (i.e. deformed ). That
is why Minkowski wrote at the beginning of Section III of Space and Time:
“Especially the concept of acceleration acquires a sharply prominent char-
acter.”
These criteria show that in spacetime the absoluteness of inertial (non-
resistant) and accelerated (resistant) motion become more understandable
– the straightness of a timelike worldline (representing inertial motion) and
the curvature or rather the deformation of a timelike worldline (representing
accelerated motion) are absolute (frame-independent) properties of world-
lines. These absolute properties of worldlines (straightness and deformation)
correspond to the absoluteness (frame-independence) of inertial and accel-
erated motion in terms of experimental detection – it is an experimental
fact that a particle moving by inertia o↵ers no resistance to its uniform mo-
tion, and it is an experimental fact that an accelerating particle resists its
acceleration.
Then, as indicated in Section 2, it becomes evident that absolute accel-
eration is a mere manifestation of the deformation of the worldline of an
accelerating particle and does not imply some absolute space with respect to
which the particle accelerates. Exactly in the same way, absolute inertial
56
In the beginning of Section II of his paper Space and Time (this volume) Minkowski
wrote: “a straight line inclined to the t-axis corresponds to a uniformly moving substantial
point, a somewhat curved worldline corresponds to a non-uniformly moving substantial
point.”
motion reflects the straightness of the worldline of an inertial particle and

does not imply some absolute space with respect to which the particle moves
with constant velocity.
Perhaps Minkowski knew all this well. What is more important, how-
ever, is that he certainly knew that an accelerating particle is represented
by a curved (deformed) worldline. Then he might have realized that inertia
– the resistance a particle o↵ers to its acceleration – could be regarded as
arising from a four-dimensional stress57 in the deformed worldline, or rather
worldtube, of an accelerating particle. Certainly, Minkowski would have
been enormously pleased with such a discovery because inertia would have
turned out to be another manifestation of the four-dimensionality of the
absolute world since only a real four-dimensional worldtube could resist its
deformation (by analogy with an ordinary deformed rod which resists its de-
formation). Of course, the question of whether or not Minkowski could have
noticed this surprising four-dimensional explanation of the origin of inertia
will forever remain unanswerable; but that explanation of inertia follows
logically from the fact that an accelerating particle is a deformed worldtube
and therefore would have been a legitimate logical option for Minkowski, es-
pecially given the fact that all his contributions to mathematics and physics
demonstrated his innovative ability to explore the deep logical structure of
what he studied.
We saw that Minkowski’s spacetime criteria for inertial and accelerated
motion spectacularly resolved the old (since Newton) question of the mean-
ing of absolute acceleration – the acceleration of a particle is absolute not
because it accelerates with respect to an absolute space, but because the
particleŠs worldline is curved (deformed) which is an absolute geometric
property. Then by asking the obvious question “What is the link between
the two absolute properties of an accelerating particle – the absolute geo-
metric property (the deformation of its worldline) and the absolute physical
property reflected in the fact that an accelerating particle resists its acceler-
ation?” we are led to the surprising insight about the origin of inertia – the
resistance a particle o↵ers to its acceleration is in fact the static resistance
in the deformed worldline of the accelerating particle.
To see even better the enormous potential of Minkowski’s criteria for
inertial and accelerated motion let us imagine two scenarios.
First, imagine that Minkowski or someone else who had had profound
understanding of Minkowski’s spacetime physics had read Galileo’s works.
57
2009) Chap. 9.
28 INTRODUCTION
That would have played the role of Einstein’s “happiest thought” because
Galileo came close to the conclusion that a falling body does not resist its
fall:58
But if you tie the hemp to the stone and allow them to fall freely
from some height, do you believe that the hemp will press down
upon the stone and thus accelerate its motion or do you think
the motion will be retarded by a partial upward pressure? One
always feels the pressure upon his shoulders when he prevents
the motion of a load resting upon him; but if one descends just
as rapidly as the load would fall how can it gravitate or press
upon him? Do you not see that this would be the same as trying
to strike a man with a lance when he is running away from you
with a speed which is equal to, or even greater, than that with
which you are following him? You must therefore conclude that,
during free and natural fall, the small stone does not press upon
the larger and consequently does not increase its weight as it
does when at rest.
Then the path to the idea that gravitational phenomena are manifesta-
tions of the curvature of spacetime would have been open – the experimen-
tal fact that a falling particle accelerates (which means that its worldtube
is curved), but o↵ers no resistance to its acceleration (which means that
its worldtube is not deformed) can be explained only if the worldtube of a
falling particle is both curved and not deformed, which is impossible in the flat
Minkowski spacetime where a curved worldtube is always deformed. Such a
worldtube can exist only in a non-Euclidean spacetime whose geodesics are
naturally curved due to the spacetime curvature, but are not deformed.
Second, imagine that after his Space and Time lecture Minkowski found a
very challenging mathematical problem and did not compete with Einstein
for the creation of the modern theory of gravitation. But when Einstein
linked gravitation with the geometry of spacetime Minkowski regretted his
change of research interests and started to study intensely general relativity
and its implications.
As a mathematician he would be appalled by what he saw as confusing
of physics and geometry:
• The new theory of gravitation demonstrates that gravitational physics

58
Galileo, Dialogues Concerning Two Sciences. In: S. Hawking (ed.), On The Shoulders
Of Giants, (Running Press, Philadelphia 2002) pp. 399-626, p. 447
is in fact geometry of curved spacetime; no general relativity of any-

thing can be found there.
• How could physicists say that in the framework of general relativity

itself gravitational phenomena are caused by gravitational interac-
tion? According to what general relativity itself tells us gravity is not
a physical interaction since by the geodesic hypothesis particles falling
towards a planet and planets orbiting the Sun all move by inertia and
inertia by its very nature presupposes no interaction. The mass of the
Sun, for example, curves spacetime no matter whether or not there
are other planets in its vicinity, and the planets move by inertia while
orbiting the Sun (the correct expression is: the planets’ worldlines are
geodesics which represent inertial motion).
• How could physicists talk about gravitational energy in the framework

of general relativity? There is no gravitational field and no gravita-
tional force; the gravitational field is at best a geometric not a physical
field, and as such it does not possess any energy. Moreover, the math-
ematical formalism of general relativity itself refuses to yield a proper
(tensorial) expression for gravitational energy and momentum.
I guess some physicists might be tempted to declare that such questions

are obvious nonsense. For instance, they might say that the decrease of the
orbital period of a binary pulsar system, notably the system PSR 1913+16
discovered by Hulse and Taylor in 1974, provided indirect experimental ev-
idence for the existence of gravitational energy that is carried away by the
gravitational waves emitted by the neutrons stars in the system.
It may sound heretical, but the assumption that the orbital motion of
the neutron stars in the PSR 1913+16 system loses energy by emission of
gravitational waves contradicts general relativity, particularly the geodesic
hypothesis and the experimental evidence which confirmed it. The reason
is that by the geodesic hypothesis the neutron stars, whose worldlines are
geodesics (the neutron stars in the PSR 1913+16 system had been modeled
by Taylor and Hulse “as a pair of orbiting point masses” which means that
they are exact geodesics) move by inertia without losing energy since the
very essence of inertial motion is motion without any loss of energy.
Therefore no energy is carried away by the gravitational waves emitted by
the binary pulsar system. For this reason the experimental fact of the decay
of the orbital motion of PSR 1913+16 (the shrinking of the stars’ orbits)
does not constitute evidence for the existence of gravitational energy. That
30 INTRODUCTION
fact may most probably be explained in terms of tidal friction as suggested

in 1976 as an alternative to the explanation given by Hulse and Taylor.
A detailed critical examination of the “confusing of physics and geome-
try” (as Minkowski might have called it) is part of an analysis of the nature of
inertia and gravitation by explicitly following Minkowski’s approach, which
is reported in V. Petkov, Inertia and Gravitation (Minkowski Institute Press,
Montreal 2012), to appear in 2012.
Minkowski and the reality of spacetime

Since 1908 there has been no consensus on the reality of the absolute four-
dimensional world no matter whether it is the flat Minkowski spacetime or a
curved spacetime since both spacetimes represent a four-dimensional world
with time wholly given as the fourth dimension. What makes this issue
truly unique in the history of science is that for over a hundred years not
only has it remained an unresolved one, but for some it has been even a
non-issue, whereas Minkowski had already provided the necessary evidence
for the reality of spacetime in 1907 and 1908. He had fully realized the
profound physical meaning of the relativity principle (reflecting the exist-
ing experimental evidence) – the impossibility to discover absolute motion
experimentally unequivocally implies that observers in relative motion have
di↵erent times and spaces, which in turn implies that what exists is an ab-
solute four-dimensional world.
Apparently Minkowski had realized the entire depth and grandness of the
new view of the absolute four-dimensional world imposed on us by the ex-
perimental evidence. A draft of his Cologne lecture Space and Time reveals
that he appears to have tried to tone down his excitement in the announce-
ment of the unseen revolution in our understanding of the world. As the
draft shows Minkowski’s initial intention had been to describe the impact of
the new world view in more detail – he had written that the essence of the
new views of space and time “is mightily revolutionary, to such an extent
that when they are completely accepted, as I expect they will be, it will be
disdained to still speak about the ways in which we have tried to understand
space and time.”59 In the final version of the lecture Minkowski had reduced
this sentence about the new views of space and time to just “Their tendency
is radical.”
59
See: P. L. Galison, Minkowski’s Space-Time: From Visual Thinking to the Absolute
World, Historical Studies in the Physical Sciences, 10 (1979) pp. 85-121, p. 98.
MINKOWSKI AND THE REALITY OF SPACETIME 31
Given this rather restrained (compared to the draft version) announce-

ment of the successful decoding of the physical meaning of the relativity
principle – that the world is four-dimensional – it is surprising that Damour
referred to that announcement as “the somewhat theatrical tone of Cologne’s
non-technical exposé.”60 The tone of the Cologne lecture could look theatri-
cal only to someone who does not see the major issue in it in the way
Minkowski saw it. This seems to be precisely the case since Damour ap-
parently regards Minkowski’s unification of space and time into an absolute
four-dimensional world as nothing more than a mathematical abstraction:61
Though Minkowski certainly went much farther than Poincaré

in taking seriously the 4-dimensional geometry as a new basis
for a physico-mathematical representation of reality, it does not
seem that he went, philosophically and existentially, as far as
really considering ‘the flow of time’ as an illusory shadow. By
contrast, let us recall that the old Einstein apparently did take
seriously, at the existential level, the idea that ‘time’ was an
illusory shadow, and that the essence of (experienced) reality
was timeless.
Minkowski’s paper does not contain anything that even resembles a hint
of what Damour wrote – that “it does not seem that he went, philosophically
and existentially, as far as really considering ‘the flow of time’ as an illusory
shadow.” On the contrary, the whole paper and even its “theatrical tone”
(in Damour’s own words) unambiguously demonstrates that Minkowski con-
sciously announced a major discovery about the world, not a discovery of
a mathematical abstraction (moreover Minkowski was fully aware that that
mathematical abstraction was already published by Poincaré two years be-
fore Minkowski’s Cologne lecture).
It is particularly disturbing when especially relativists do not regard
spacetime as representing a real four-dimensional world and still hold the
unscientific 62 view that time flows. Such an opinion of spacetime as nothing
60
T. Damour, “What is missing from Minkowski’s “Raum und Zeit” lecture”, Annalen
der Physik. 17, No. 9-10, (2008) pp. 619-630, p. 620.
61
62
This everyday view is unscientific since there is no scientific evidence whatsoever for
the sole existence of the present moment, which is the central element of the concept of
time flow (what is sufficient for the issue of the reality of spacetime is that there is no
physical evidence for the existence of time flow). If the flow of time were a feature of the
physical world (not of the image of the world in our mind), physics would have discovered
it by now.
32 INTRODUCTION
more than a mathematical space was openly defended by another physicist,

Mermin, in a recent article What’s bad about this habit in the May 2009 issue
of Physics Today where he argued that “It is a bad habit of physicists to take
their most successful abstractions to be real properties of our world.” 63 He
gave the issue of the reality of spacetime as an example – “spacetime is an
abstract four-dimensional mathematical continuum” – and pointed out that
it is “a bad habit to reify the spacetime continuum”. Mermin specifically
stressed that spacetime does not represent a real four-dimensional world:
“The device of spacetime has been so powerful that we often reify that
abstract bookkeeping structure, saying that we inhabit a world that is such
a four- (or, for some of us, ten-) dimensional continuum.”
I think the proper understanding of Minkowski’s spacetime physics (which
requires more e↵ort than learning its four-dimensional formalism) is crucial
not only for deep understanding of modern physics, but more importantly
such understanding is a necessary condition for making discoveries in the
twenty-first century physics.
The best proof that the experimental evidence against the existence of ab-
solute motion (reflected in the relativity postulate) implies that the Universe
is an absolute four-dimensional world is contained in Minkowski’s paper it-
self. As discussed in Section 2 Minkowski first realized the important hidden
message in the experimental fact that physical phenomena are the same in
all inertial reference frames (which Einstein merely stated in the relativity
postulate without explaining it) – physical phenomena are the same in all
inertial reference frames because every inertial observer has his own space
and time64 and therefore describes the phenomena in his reference frame (i.e.
in his own space and time) in which he is at rest. For example, the Earth
is at rest with respect to its space and therefore all experiments confirm
this state of rest. Due to his excellent geometrical imagination Minkowski
appears to have immediately realized that many spaces are possible in a
four-dimensional world. In this way he managed to decode the physical
meaning of the experimental fact that absolute motion cannot be discovered
– that fact implies that the Universe is an absolute four-dimensional world
in which space and time are inseparably amalgamated; only in such a world
one can talk about many spaces and many times. Minkowski noted that “I
think the word relativity postulate used for the requirement of invariance
under the group Gc is very feeble. Since the meaning of the postulate is
63
N. D. Mermin, What’s Bad About This Habit? Physics Today 2009, p. 8.
64
As we saw in Section 2 Minkowski showed that the equivalence of the times of observers
in relative motion (which is necessary to explain why absolute motion cannot be detected)
means that the observers have not only di↵erent times but di↵erent spaces as well.
The transparency which Minkowski used at his lecture in Cologne on September

21, 1908. It shows Fig. 1 in his paper (this volume). Source: Cover of The
Mathematical Intelligencer, Volume 31, Number 2 (2009).
that through the phenomena only the four-dimensional world in space and
time is given, but the projection in space and in time can still be made with
certain freedom, I want to give this affirmation rather the name the postulate
of the absolute world ” (this volume).
To see why Minkowski’s absolute four-dimensional world adequately rep-
resents the dimensionality of the real world, assume the opposite – that the
real world is three-dimensional and time really flows (as our everyday ex-
perience so convincingly appears to suggest). Then there would exist just
one space, which as such would be absolute (i.e. it would be the same
for all observers since only a single space would exist). This would imply
that absolute motion should exist and therefore there would be no relativity
principle.
Another example of why special relativity (as we now call the physics
of flat spacetime) would be impossible in a three-dimensional world is con-
tained in Minkowski’s four-dimensional explanation of the physical mean-
ing of length contraction, which is shown in the above figure (displaying
the transparency Minkowski used in 1908). Consider only the vertical (red)
strip which represents a body at rest with respect to an observer. The proper
length of the body is the cross section P P of the observer’s space, represented
by the horizontal (red) line, and the body’s strip. The relativistically con-
34 INTRODUCTION
tracted length of the body measured by an observer in relative motion with

respect to the body is the cross section P 0 P 0 of the moving observer’s space,
represented by the inclined (green) line, and the body’s strip (on the trans-
parency P 0 P 0 appears longer than P P because the two-dimensional pseudo-
Euclidean spacetime is represented on the two-dimensional Euclidean surface
of the page).
To see that no length contraction would be possible in a three-dimensional
world,65 assume that the world is indeed three-dimensional. This would
mean that all objects are also three-dimensional. Therefore the four-
dimensional vertical strip of the body would not represent anything real in
the world and would be merely an abstract geometrical construction. Then,
obviously, the cross sections P P and P 0 P 0 would coincide and there would be
no length contraction since the observers in relative motion would measure
the same three-dimensional body which has just one length P P ⌘ P 0 P 0 .
The impossibility of length contraction in a three-dimensional world also
follows even without looking at the spacetime diagram: it follows from the
definition of a three-dimensional body – all its parts which exist simultane-
ously at a given moment; when the two observers in relative motion measure
the length of the body, they measure two di↵erent three-dimensional bodies
since the observers have di↵erent sets of simultaneous events, i.e. di↵erent
sets of simultaneously existing parts of the body (which means two di↵er-
ent three-dimensional bodies). If the world and the physical bodies were
three-dimensional, then the observers in relative motion would measure the
same three-dimensional body (i.e. the same set of simultaneously existing
parts of the body), which means that (i) they would have a common set of
simultaneous events in contradiction with relativity (simultaneity would be
absolute), and (ii) they would measure the same length of the body, again
in contradiction with relativity.
The same line of reasoning demonstrates that no relativity of simultane-
ity, no time dilation, and no twin paradox e↵ect would be possible in a
three-dimensional world.66
As I gave examples of how some physicists do not fully appreciate the
depth of Minkowski’s discovery that the physical world is four-dimensional,
it will be fair to stress that there have been many physicists (I would like to
65
A visual representation of Minkowski’s explanation of length contraction is given
in V. Petkov, Spacetime and Reality: Facing the Ultimate Judge, Sect. 3 (http:
//philsci-archive.pitt.edu/9181/). I think this representation most convincingly
demonstrates that length contraction is impossible in a three-dimensional world.
66
2009) Chap. 5.
think the majority) who have demonstrated in written form their brilliant
understanding of what the dimensionality of the world is. Here are several
examples.
A. Einstein, Relativity: The Special and General Theory (Routledge,
London 2001) p. 152:
It appears therefore more natural to think of physical reality as a

four-dimensional existence, instead of, as hitherto, the evolution
of a three-dimensional existence.
A. S. Eddington, Space, Time and Gravitation: An Outline of the Gen-

eral Relativity Theory (Cambridge University Press, Cambridge 1920), p.
51:
In a perfectly determinate scheme the past and future may be

regarded as lying mapped out – as much available to present
exploration as the distant parts of space. Events do not happen;
they are just there, and we come across them.
A. S. Eddington, Space, Time and Gravitation: An Outline of the Gen-

eral Relativity Theory (Cambridge University Press, Cambridge 1920), p.
56:
However successful the theory of a four-dimensional world may

be, it is difficult to ignore a voice inside us which whispers: “At
the back of your mind, you know that a fourth dimension is all
nonsense.” I fancy that that voice must often have had a busy
time in the past history of physics. What nonsense to say that
this solid table on which I am writing is a collection of electrons
moving with prodigious speeds in empty spaces, which relatively
to electronic dimensions are as wide as the spaces between the
planets in the solar system! What nonsense to say that the thin
air is trying to crush my body with a load of 14 lbs to the square
inch! What nonsense that the star cluster which I see through
the telescope obviously there now, is a glimpse into a past age
50 000 years ago! Let us not be beguiled by this voice. It is
discredited.
Eddington made his most explicit comment on the reality of spacetime

when he discussed the fact (discovered by Minkowski) that not only do ob-
servers in relative motion have di↵erent times but they also have di↵erent
36 INTRODUCTION
spaces, which however are fictitious since according to the theory of rela-
tivity the world is not objectively divided into such spaces and times (A.S.
Eddington, The Relativity of Time, Nature 106 (1921) pp. 802–804, p.
803):
It was shown by Minkowski that all these fictitious spaces and

times can be united in a single continuum of four dimensions.
The question is often raised whether this four-dimensional space-
time is real, or merely a mathematical construction; perhaps it
is sufficient to reply that it can at any rate not be less real than
the fictitious space and time which it supplants.
H. Weyl, Philosophy of Mathematics and Natural Science (Princeton

University Press, Princeton 2009) p. 116:
The objective world simply is, it does not happen. Only to the
gaze of my consciousness, crawling upward along the life line of
my body, does a section of this world come to life as a fleeting
image in space which continuously changes in time.
H. Weyl, Mind and Nature: Selected Writings on Philosophy, Mathemat-

ics, and Physics (Princeton University Press, Princeton 2009) p. 135:
The objective world merely exists, it does not happen; as a whole

it has no history. Only before the eye of the consciousness climb-
ing up in the world line of my body, a section of this world “comes
to life” and moves past it as a spatial image engaged in temporal
transformation.
R. Geroch, General relativity from A to B (University of Chicago, Chicago

1978) pp. 20-21:
There is no dynamics within space-time itself: nothing ever

moves therein; nothing happens; nothing changes [. . . ] one does
not think of particles as ‘moving through’ space-time, or as ‘fol-
lowing along’ their world-lines. Rather, particles are just ‘in’
space-time, once and for all, and the world-line represents, all at
once, the complete life history of the particle.
In a real four-dimensional world there is no time flow since all moments

of time have equal existence as they all form the fourth dimension (which
like the other three dimensions is entirely given), whereas the very essence of
time flow is that only one moment of time exists which constantly changes.
But it is a well known fact that there does not exist any physical evidence
whatsoever that only the present moment exists. On the contrary, all rela-
tivistic experimental evidence confirms Minkowski’s view that all moments
of time have equal existence due to their belonging to the entirely given
time dimension. So the old Einstein was wise67 to take seriously the abso-
lute four-dimensional world and the idea that the flow of time was merely
“a stubbornly persistent illusion” as evident from his letter of condolences
to the widow of his longtime friend Besso:68
Now Besso has departed from this strange world a little ahead of
me. That means nothing. People like us, who believe in physics,
know that the distinction between past, present and future is
only a stubbornly persistent illusion.
Minkowski succeeded in demonstrating how the power of mathematical

thinking applied to unresolved physical problems can free us from such illu-
sions and can reveal the existence of a reality that is difficult to comprehend
at once. Galison masterfully summarized the essence of Minkowski’s discov-
ery by pointing out that in his lectures The Relativity Principle and Space
and Time “the idea is the same: beyond the divisions of time and space
which are imposed on our experience, there lies a higher reality, changeless,
and independent of observer.”69
I think there are still physicists and philosophers who have been e↵ec-
tively refusing to face the implications of a real four-dimensional world due
to the huge challenges they pose. But trying to squeeze Nature into our pre-
set and deceivingly comfortable views of the world should not be an option
for anyone in the 21st century.
Montreal Vesselin Petkov

July 2012
67
I think it is this context that is the right and fair one for using the word ‘old’ especially
if it refers to such a scientist and person as Einstein.
68
Quoted from: Michele Besso, From Wikipedia, the free encyclopedia (http://en.
wikipedia.org/wiki/Michele_Besso). Besso left this world on 15 March 1955; Einstein
followed him on 18 April 1955.
69
P. L. Galison, Minkowski’s Space-Time: From Visual Thinking to the Absolute World,
Historical Studies in the Physical Sciences, 10 (1979) pp. 85-121, p. 98.
38 INTRODUCTION
The Relativity Principle
With regard to the electromagnetic light theory, it appears that recently a

complete transformation that changes our thinking about space and time
wants to take place, and to become acquainted with such thinking must in
any case be of quite particular interest for a mathematician. Also, the math-
ematician is particularly well predisposed to absorb these new intuitions be-
cause this involves becoming acclimatized and to conceptualize things anew,
that is, it involves a process that the mathematician has practiced for the
longest time. Although the physicists must now partially and newly invent
these concepts whereby they must clear for themselves and with great e↵ort
a jungle like path through an unexplored territory, while quite nearby there
already exists an excellent highway of the mathematicians’ which comfort-
ably leads onwards. After all, the new attempts, if they in fact interpret
the phenomena correctly, would present almost the greatest triumph ever
that the application of mathematics has brought about as of today. What
is being dealt with here is, expressed as foreshortened as possible – I will
present a more explicit account later – that the world in space and time in a
certain sense is a four dimensional, non-Euclidean manifold. Apparently, as
to the fame of mathematicians and the astonishment of the rest of mankind,
the mathematicians within their pure fantasies opened up a huge territory,
although these ideal craftsmen never had any such intentions, which one day
would contain the most complete real existence.
The principle of relativity, to which I want to draw your attention today,
has been invented as a means to find an explanation why every experimental
attempt which would show that the earth moves relative to a luminiferous
aether must necessarily fail. Experiments, which rely on magnitudes of the
order of quotients which take the speed of the earth in the solar system
over the speed of light as a basis for observation, have shown, so far, that
it is impossible to determine the direction of the earth’s motion through
experiments that take place on the surface of the earth. This is so because
circumstances are such that a comparison of two clocks placed at a distance
39
40 THE RELATIVITY PRINCIPLE
from each other at two separate points must be made whereby signals must
necessarily travel forth and back again between these clocks. Moreover, A.
Michelson performed in 1881 an experiment (which in 1887, together with
Morley, was repeated on a larger scale) which took into account detection of
a second order magnitude in the above mentioned quotients, nevertheless,
the result turned out negative just the same. In order to explain this negative
result as well, H. A. Lorentz (1892) and independently Fitz Gerald (1893)
formulated the hypothesis that on account of the earth’s motion a quite
determinable contraction of matter occurs parallel to the earth’s motion.
From this highly peculiar sounding hypothesis finally evolved the postulate
of relativity in a form that particularly suited the mathematician’s way of
understanding. Credits for the general principle’s development are shared
by Einstein, Poincaré, and Planck. I will talk in more detail about their
work a little later.
Now I will finally get to the actual subject under discussion and, in order
to maintain clarity, I shall divide what follows into four subject headings,
namely: 1. Electricity, 2. Matter, 3. Dynamics, and 4. Gravitation.
1. First of all, what is being dealt with here explicitly is a purely math-
ematical relationship of a certain formal character of those di↵erential equa-
tions which Lorentz takes to be fundamental to his electron theory and,
moreover, that these equations regulate the functioning of the electromag-
netic field in pure aether in the same way as they apply to an infinite space
that is filled with electricity. These fundamental equations, although they
do not pertain to a right angled coordinate system of space of which they
stand quite independently, contain nevertheless a certain additional symme-
try which cannot be expressed through our familiar way of writing. Fore-
most, although this by the way did not happen with regard to the above
named authors, not even with Poincaré, I will render here an expression of
the above mentioned symmetry, whereby in fact, if I may say so, the form
of the equations becomes quite transparent. To start with let x, y, z desig-
nate fixed right angled coordinates in space, in the aether, and t designates
time. Furthermore, the follow up discussion involves the quadratic expres-
sion x2 + y 2 + z 2 c2 t2 whereby c denotes the propagation speed of light
in empty space. A unit of time may be chosen in such a way that c = 1,
in other words, as 1/3.1010 s whereby a unit of length becomes 1 cm. Now
one may write x1 , x2 , x3 , instead of x, y, z and furthermore x4 may replace
it. In that way, with regard to the following, x4 stands, of course, for a
purely imaginary magnitude. The above quadratic expression may now be
transformed into
x21 + x22 + x23 + x24
41
which means that from now on we will be dealing with constructions in a

four dimensional manifold of x1 , x2 , x3 , x4 . This also indicates that, at any
point in time, the entire electromagnetic state in space can now be described
through the behaviour of a single four dimensional vector. The components
of this vector can be expressed as 1 , 2 , 3 , 4 whereby again 1 , 2 , 3
are real and 4 is purely imaginary. In accord with the descriptions which
Abraham gives us in his electromagnetic theory of radiation, Ux , Uy , Uz are
the components of the electromagnetic vector potentials U, whereby 4 = i
and is the scalar electromagnetic potential. This vector ( ), on account of
its dependence upon x1 , x2 , x3 , x4 , must now fulfill the following condition:
@ 1 @ 2 @ 3 @ 4
+ + + =0 (1)
@x1 @x2 @x3 @x4
The di↵erential expression to the left may be called Div( ). In addition
a second four dimensional vector gains validity, namely (%) = %1 , %2 , %3 , %4 .
Therefore, %4 = i% whereby % indicates the density of electricity per unit
volume, and %1 , %2 , %3 are the components of the spatial vectors % b according
to Abraham, whereby b indicates the velocity of the convective motion of
electricity. Finally, the abbreviation ⇤ can be used with the di↵erential
expression
@2 @2 @2 @2
+ + +
@x21 @x22 @x23 @x24
which allows that we take the combinations
@ k @ j
@xj @xk
so that we may designate the latter by jk whereby, in general, jk = kj

and with the same subscript twice jj = 0. This now renders the following
di↵erential equations valid:
⇤ j = %j (j = 1, 2, 3, 4) (2)
and from the previously given equation (1) obviously follows Div(%) = 0,
which is the equation of continuity of electricity. Finally,
23 , 31 , 12 ; 14 , 24 , 34
can be identified with
Hx , H y , H z ; iEx , iEy , iEz ,

whereby Ex , Ey , Ez the electrical components and Hx , Hy , Hz the magnetic

components describe field strength. These certainly quite understandably
and easily arrived at formulas which are the total expression of the Lorentzian
foundational fundamental equations, contain naturally in a special sense, if
the vector (%) = 0 is applied, Maxwell’s equations for the electromagnetic
processes in pure aether. Now the usual somewhat artful looking expres-
sions for the interference with charges by ponderomotive forces of the field
become quite transparent. Let X, Y, Z be the components of this force which
should be calculated per unit volume (not at all per unit charge), and besides
these magnitudes one must also take a fourth magnitude into consideration,
namely, the work done by this force per second which gives us,
A = Xbx + Y by + Zbz ;
so that X, Y, Z, iA are the components of a four dimensional vector X1 , X2 ,

X3 , X4 , whereby in general we may write
Xj = %1 j1 + %2 j2 + %3 j3 + %4 j4
so that now, during the motion of electricity, the following expression will
always hold:
X1 dx1 + X2 dx2 + X3 dx3 + X4 dx4 = 0.
Within these equations, with reference to how I have transformed them in my
writing, lies a perfectly obvious fact which hereafter is tied to the relativity
principle. That is, if instead of x, y, z, t one establishes through a perfectly
acceptable linear transformation a new set of primed coordinates x0 , y 0 , z 0 , t0 ,
then the expression x21 + x22 + x23 + x24 remains, so to say, invariant; and if one
transforms the vector 1 , 2 , 3 , 4 in the same way as x1 , x2 , x3 , x4 were
transformed, then the entire system of the above equations also pertains to
the corresponding primed system which incorporates the same equations. In
that sense, without having to claim a new law which would lead us astray
to something that has not been previously contained, we may proclaim it
trivial that the fundamental equations of the electron theory allow for the
orthogonal transformations of the four dimensional space.
2. We will now turn to a further contemplation of matter. Here we must
deal, on the one hand, with electrodynamics and then, on the other hand,
with mechanics. To do this, we adopt a point of view under which we are not
as yet totally familiar with the prevailing physical laws. Perhaps one day a
reduction to a pure doctrine of electricity will become possible, but as for
now, particularly in light of Michelson’s experiment, it has been shown that,
43
as Einstein so succinctly expresses this, the concept of an absolute state of

rest entails no properties that correspond to phenomena. And this fact can
easily be explained if we assume that also the equations of the electrody-
namics of matter are in any case of such a kind, that they likewise remain
invariant under that group which Poincaré named after Lorentz. Here the
relativity principle takes up its place as a truly new physical law, that is, the
relativity principle advances a demand for equations of phenomena which
have yet to be established. To what conclusions this postulate leads will
now be discussed in detail. For the contemplation of matter we must first
of all focus on a new vector as a function of x, y, z, t, the visible velocity of
mater at any point. Let wx , wy , wz be the components of the velocity at one
position of matter, whereby w is the magnitude of the velocity, thus a vector
in space corresponds to real motion and not, however, to the case of rest. I
will now consider instead a four dimensional vector; I will take the important
result in advance, namely, that velocities of matter equal or greater than the
velocity of light proves itself to be an absurdity, that therefore w < 1 must
always hold true. Now I posit that
p
w x , wy , wz , i 1 w 2
= w 1 , w2 , w3 , w4 ,
so that w1 , w2 , w3 , w4 is always a point upon the surface
w12 + w22 + w32 + w42 = 1
Or, if you want, upon

t2 x2 y2 z 2 = 1, (3)
and represents at the same time the four dimensional vector from the ori-
gin to this point; and this corresponds also to the velocity of zero, that is,
something like a real vector represents a state of rest. The non-Euclidean
geometry, of which I already spoke in a preliminary sense, now evolves for
these velocity vectors. The transformations, of which I spoke just above,
become the real transformations of this four dimensional hyperboloid (3)
with regard to conjugated diameters, whereby it becomes perfectly clear
that one can begin with insert, as t-diameter, the light ray at any predeter-
mined point of this surface; here we are dealing with transformations under
which not only spatial coordinates are transformed, but, at the same time
and under a certain connection therewith, time is also transformed. In par-
ticular, as we have just seen, the velocity of any single point of matter can
therefore be transformed to a state of rest, and when this happened, what
remains thereafter by choosing the other three conjugated diameters of (3)

is nothing other than the freedom of a single orthogonal transformation of
the spatial coordinates. After this explanation it should at once be clear,
that the invariance of physical laws with regard to Lorentz’ group can be ex-
pressed accordingly, as I have already done above, namely, that the absolute
state of rest entails no properties of phenomena. Although I will talk about
mechanics only afterwards, I will insert here, for a clearer understanding
of the ideas presented, how Galileo’s law of inertia stands in relation to the
postulate of relativity. To this end, we will not at once designate the velocity
of light by 1, instead, we will stay for now with the general symbol c for the
velocity of light. In that case, with respect to a velocity w in the direction
of the x-axis, the transformation, through which this velocity is reduced to
a state of rest, becomes:
c(x wt) c2 t wx
x0 = p , y 0 = y, z 0 = z, t0 = p .
c2 w 2 c c2 w 2
Now we see clearly, with respect to a limit under which c = 1, that the
above equations can be expressed as:
x0 = x wt, y 0 = y, z 0 = z, t0 = t,
that is, new right angled coordinates are simply inserted with respect to an
axis system which, with respect to the initial right angled coordinates, moves
uniformly with the translational motion of the initial coordinate system.
Hereby, and in accordance with the law of inertia, the laws of mechanics
should remain unchanged in their expressions. Accordingly now, the law of
inertia signifies an invariance of mechanics for the transformations of the
expression x2 + y 2 + z 2 c2 t2 whereby c = 1, that is, the law of inertia
carries the same meaning as does the postulate of relativity for c = 1.
Now we shall take a closer look at how electrodynamics results founda-
tionally from the principle of relativity. Hereby Lorentz’ initial attempts are
to be seen as particularly necessary. Besides the velocity vector we have to
take into consideration two additional formations of matter; here I present
the results again in such a way that the invariance of the Lorentz group be-
comes evident. To start with, I introduce a four dimensional vector, namely,
( )= 1, 2, 3, 4,
and I shall call this vector the electrical current. Hereby 1 , 2 , 3 are to
be identified with the components ix , iy , iz of the electrical current, and
45
4 = i indicates, with being the charge density, the density of the actual
electricity of the Maxwell-Hertz theory. This vector satisfies the continuity
equation Div( ) = 0. In addition, I introduce another thing which for now,
if you don’t mind, I shall call a Traktor. The latter shall be a formation
consisting of six components which, in a four dimensional space, become
attached to the concept of the vector, namely,
p23 , p31 , p12 , p14 , p24 , p34 .
A Traktor carries, like wise, significance which does not depend on the
choice of a four dimensional coordinate system. That is to say that new
orthogonal coordinates for x1 , x2 , x3 , x4 , namely, y1 , y2 , y3 , y4 are inserted in
such a way that the latter quadruple of variables remains congruent with
the former quadruple, that is, through this linear substitution each set of
variables is transformable into the other and, moreover, at the same time
the values of pjk are to be substituted by this transformation in such a way
that they correspond with the coefficients of the bilinear expression
p23 (x2 y3 x3 y2 ) + p31 (x3 y1 x1 y3 ) + . . . + p34 (x3 y4 x4 y3 ).
I shall now describe the fundamental equations of the electrodynamics of

moving mediums: The following points have to be taken into consideration:
For every spacetime point x, y, z, t there is a corresponding vector potential
( ) = 1 , 2 , 3 , 4 of the electromagnetic field, further a velocity vector
(w) = w1 , w2 , w3 , w4 of matter conforms to this, as does the vector of the
electric current given by ( ) = 1 , 2 , 3 , 4 , and finally a Traktor, which I
will call a Polarisationstraktor, (p) = p23 , . . . p34 comes into play.
The vector ( ), although it is not grounded in a four dimensional space,
must satisfy the fundamental equation Div( ) = 0. From this vector arises,
as shown above, a Traktor 23 , . . . 34 , whereby now the first three compo-
nents indicate the magnetic induction and the following three the electrical
field strength. In addition we already have the di↵erential equation,
@p1j @p2j @p3j @p4j
+ + + = j +⇤ j ... [pkj = pjk ].
@x1 @x2 @x3 @x4
Finally the following facts are valid:
If one transforms the components in such a way that a determined posi-
tion x, y, z, t stays at rest, then p14 , p24 , p34 , multiplied by i, becomes itself
the dielectric polarization in proportion to the vector 14 , 24 , 34 , whereby
the proportionality factor decreases the dielectric constant of the medium
by 1; furthermore, with regard to objects that cannot be magnetized, the
vector p23 , p31 , p12 is in that case zero; finally, the vector 1 , 2 , 3 is also
proportional to the vector 14 , 24 , 34 , and the proportionality factor is the
electrical capacity of conductivity. Evidently, under these terms or stipula-
tions the invariance of the Lorentzian group rests ascertained.
3. Now we will turn to mechanics. According to what I have already said
about the relationship between the relativity principle and the law of inertia,
it should, for a start, be clearly understood that the prevailing fundamental
laws of mechanics count for nothing more than an approximation of reality
if, with regard to mechanics, the postulate of relativity is to be validated.
But such has to be again the case because otherwise the possibility that
the earth moves relative to the aether would have to be reinstated. Planck
clearly points to the necessity of dismissal of the law of inertia under the
following consideration. If one takes thermal radiation into account, then it
becomes impossible to separate the energy of a uniformly moving body from
any other energy of that body. In general, one cannot define the kinetic
energy of a body in a determined manner, that is, there always remains
a gap in the laws of mechanics which can only be filled by applying the
relativity principle. If we think of an empty space or cavity surrounded
by a very expansive and perfect conductor, then, on account of its huge
volume, the influence of the matter of the walls of this cavity can further be
considered negligible. Accordingly, if inside this cavity the thermal radiation
is designated by some temperature T , and if, thereby, a body moves with
a constant velocity w in any direction, then the volume of this cavity can
finally be designated by V . Hence, the theory that accounts for the energy
contained in radiation can be formulated as:
2
1 + w8
E= T 4 V,
(1 w2 )3
however, a suitable temperature unit must be chosen for this application.

Now, provided that one keeps in mind that w4 stands against 1, one cannot
determine by this mathematical expression a term for a state of rest and
a proportional term for w2 , that is, one cannot separate the kinetic energy
of uniform motion from the inherent energy of a body. Moreover, such a
separation, in the context of the energy of thermal radiation, is naturally
and generally in all cases impossible. For example, Planck calculates that,
for a contained space filled with a stationary mono-atomic gas at the melting
point temperature of platinum and under a pressure of 0,001 mm, the heat
that is added under constant pressure accounts for a fourth part of the
increase in radiation energy, that is, molecular motion accounts but for three
47
quarters of the radiation energy thus generated. Apparently then, the first
quarter of this energy cannot be ignored. In his recently published article
in the Berliner Akademie (18. June of this year), Planck tries to establish
a dynamic formulation on the basis of the relativity principle. As a further
goal plausible mathematical attempts should be worked out in such a way
that experimental tests can be performed and interpreted under the new
as well as past theories. Planck ’s concern is the dynamic behaviour of a
single point when it is subjected to a change of temperature. This shows
that if, at times, one turns away from cosmic events, then thermodynamic
influences make it possible to detect important phenomena in the first place.
Planck ’s thoughts encompass a body which on the whole only experiences
translational motions, that is, a body whose motions can be determined by
three velocity components, namely, wx , wy , wz , and whose state depends in
addition on its volume V and its temperature T . Hence, what Planck is
searching for are the laws of the dynamics of such a system. The above
mentioned body is, however, at the same time in a state of rest with regard
to a certain coordinate system; the question is, at what temperature is this
body in a state of rest? Changes in volume are given by this substitution of
the coordinate system. Now Planck plausibly explains that the entropy with
respect to both coordinate systems must remain the same; accordingly then,
with regard to a thermal radiation filled cavity, we can in this case apply the
known
p laws, with constant pressure taken into consideration, so that T in
T / 1 w2 undergoes a change with respect to the new coordinate system.
Planck accepts that this relationship holds generally true on account that
pressure and temperature in this case present equilibrium parameters. Still
missing are the laws under which this system is subjected by given influential
outside forces in terms of pressure and temperature, and therefore Planck
accepts Helmholtz ’ extension of the principle of least action which can be
expanded to also hold true for thermodynamic relationships. The change
between two states, at a constant energy level, should therefore be given by
the minimum of the integral
Z
W = Hdt,
whereby H, the so called kinetic potential, depends on the three velocity

components, yet this holds only in connection with w, as well as with the
volume V and the temperature T .
The result, which Planck arrives at via a detour by dragging thermo-
dynamic relations into place, could simply have been foreseen through the
equation
p
H dt = K dt2 dx2 dy 2 dz 2 ,
whereby the second factor of the Lorentzian group remains invariant, which
means, that the first factor stays invariant as well. The noteworthy conclu-
sion that now ties up with this is that the transferral motion of the body,
which changes with time and thus presents the external force, can be estab-
lished by the following equations:
@H w(E + pV )
G= =
@w c2
whereby
@H @H
E=w +T H
@w @T
gives us the total energy of the body; I wrote the above mathematical ex-
pression, which does not specify any particular unit of time, in order to add
the following remark. If now the velocity changes only in accord with a
transversal vector and the affiliated change in size of the motion is M times
the acceleration, then M denotes an increase in size which can be called the
mass of the body. Therefore:
G E + pV
M= = .
w c2
Accordingly: Considering that mass denotes the size of the body, what
matters here in particular is that the quantity of mass changes with any
addition of heat. That is, by an addition of heat under constant pressure the
increase in mass amounts exactly to the added heat divided by the square of
the speed of light. For example, if 18 g (1 12 mol) of oxy-hydrogen gas under
atmospheric pressure and at room temperature condenses to form liquid
water, then a decrease of mass of 3.2 10 6 mg should occur, a diminution
of matter so small, however, that we cannot observe it. It would naturally
stand to reason, nevertheless, that a further development of this theory will
consequently allow us to observe this phenomenon.
4. Finally I want to say a few words about gravitation. The important
question remains, how the law of gravitation pertains to the realm of the
principle of relativity. In his article submitted to the Rendiconti del Circolo
Matematico di Palermo in 1906, Poincaré addressed this question. Laplace
believed that he had proven that the propagation of gravitation must happen
momentarily or at a much higher speed than the speed of light. Yet, if one
were to ground gravitation in electromagnetism, then Laplace’s proof would
49
surely emerge as incomplete. In fact, Poincaré as well thinks that gravitation

propagates in no other way than at the speed of light. Apparently then, the
purely mathematical task consists in finding a law which corresponds to
the relativity principle and which, at the same time, must be applicable to
Newton’s laws, that is, if one can neglect the squares of the speed of stars
as well as the multiplied result of outward acceleration (i.e., proportional to
the square of the speed of light). Poincaré points out such a law by taking
into consideration the invariants of the Lorentzian group, however, this law
is but one amongst many other possible laws, and the relevant investigations
are in no way of a definitive character. Perhaps I will write in greater detail
about this at another time. I hope my above account has made one thing
clear, namely, that the current investigations should claim, or even demand,
the highest degree of interest of mathematicians.
(Received, June 15, 1915)
Footnote on the first page of the original German publication of The Rel-
ativity Principle in 1915 (H. Minkowski, Das Relativitätsprinzip, Annalen
der Physik 47 (1915) S. 927-938):
Exactly 10 years ago the Annalen der Physik published Einstein’s work
about the electrodynamics of moving bodies. With respect to the important
role of the relativity principle, based upon this work and played out in the
Annalen, Mr. Sommerfeld, in agreement with the editorial department of
the Annalen and from Minkowski’s posthumous papers, as the most success-
ful interpretation of the relativity principle, publicly presented the following
account. This lecture was given at the meeting of the Göttinger Mathema-
tischen Gesellschaft on November 5, 1907, that is, almost one year before
Minkowski’s lecture on space and time that took place in Cologne.
The Fundamental Equations
for Electromagnetic
Processes in Moving Bodies
Introduction
At present di↵erences of opinion on the basic equations of electrodynam-

ics for moving bodies are still prevailing. The approach of Hertz 1 (1890) has
to be abandoned because it has been found that it contradicts various ex-
perimental results.
In 1895 H. A. Lorentz 2 published his theory of optical and electrical
phenomena in moving bodies, which was based on an atomistic understand-
ing of electricity, and whose many successes seem to have justified the bold
hypotheses. Lorentz’ theory3 assumes some initial equations, which should
be valid at every point of “aether”; then by forming the average values over
“physically infinitely small” regions that already contain many “electrons,”
the equations for electromagnetic processes in moving bodies can be ob-
tained.
In particular, Lorentz’ theory gives an account of the non-existence of
a relative motion of the Earth with respect to the luminiferous aether; it
brings this fact in connection with a covariance of those initial equations
with certain simultaneous transformations of the space and time parameters,
which have received from H. Poincaré 4 the name Lorentz transformations.
1
Über die Gründgleichungen der Elektrodynamik für bewegte Körper. Wiedemanns
Annalen, Bd. 41, S. 369, 1890 (auch in: Gesammelte Werke, Bd. I, S. 256, Leipzig 1892).
2
Versuch einer Theorie der elektrischen und optischen Erscheinungen in bewegten
Körpern, Leiden 1895.
3
Vgl. Enzyklopädie der mathematischen Wissenschaften, V 2, Art. 14. Weiterbildung
der Maxwellschen Theorie. Elektronentheorie.
4
Rendiconti del Circolo Matematico di Palermo, T. XXI (1906), p. 129.
51
52 THE FUNDAMENTAL EQUATIONS
For those initial equations, the covariance under the Lorentz transformations
is a purely mathematical fact, which I will call the theorem of relativity; this
theorem is essentially based on the form of the di↵erential equation for the
propagation of waves with the velocity of light.
It is now possible without any hypothesis about the connection between
electricity and matter, to expect that this mathematically evident theorem
will have its consequences extended so far that they may hold even for those
laws of ponderable media which are yet unknown, and which may possess this
covariance under Lorentz transformations. This expresses therefore more a
confidence than already an existing understanding, and this confidence I will
call the postulate of relativity. This situation is approximately such, as if
one postulates the conservation of energy in cases where the common forms
of energy are still not recognized.
If afterwards the expected covariance is manifested as a specific relation
between observable quantities for moving bodies, this particular relation
may then be called the principle of relativity.
These distinctions seem to me useful and can characterize the current
state of the electrodynamics of moving bodies.
H. A. Lorentz found the relativity theorem and established the relativity
postulate as a hypothesis that electrons and matter experience contractions
as a result of their motion according to a certain law.
A. Einstein 5 expressed it most sharply so far that this postulate is not
an artificial hypothesis but is rather a new concept of time imposed upon
us by the phenomena.
However, the Principle of Relativity, in the sense indicated by me, has
not yet been formulated for the electrodynamics of moving bodies. As I for-
mulate this principle in the present paper I obtain the fundamental equations
for moving bodies in a form entirely and unambiguously determined by this
principle. We will see that none of the previously accepted forms for these
equations exactly comply with this principle.
One would particularly expect that the adopted by Lorentz fundamental
equations for moving bodies correspond to the relativity postulate. It turns
out however, that this is not the case for the general equations derived by
Lorentz for an arbitrary, also magnetized, body but this is the case approxi-
mately (neglecting the squares of the speeds of matter against the square of
the speed of light) for those equations which Lorentz deduced afterwards for
a non-magnetic body; but this later adjustment to the relativity postulate,
in view of the fact that the condition of non-magnetization has been for-
5
Annalen der Physik, Bd. 17, S. 891, 1905.
53
mulated in a way not corresponding to the relativity postulate, is achieved

by an accidental compensation of two violations to the relativity postulate.
However, this statement means no objection against the molecular hypoth-
esis of Lorentz, but it is only clear that the adoption of the contraction of
the electrons in motion in the Lorentz’ theory would have to be introduced
at an earlier point than this was done by Lorentz.
In an appendix I discuss the status of classical mechanics with respect
to the relativity postulate. An easy change of mechanics in accordance with
the requirements of the postulate of relativity would give for the observed
phenomena hardly noticeable di↵erences, but would lead to very surprising
results: Starting with the relativity postulate, there are sufficient means for
deducing henceforth the complete set of laws of mechanics from the law of
conservation of energy alone (and statements about the forms of energy).
§1.
Notations
Let a reference system of rectangular coordinates x, y, z, t be given in

space and time. The unit of time is selected in such a relation to the unit
of length that the velocity of light in empty space is 1.
Although I would prefer not to change the notations used by Lorentz it
seems important to me to use from the very beginning a di↵erent selection
of symbols and to establish their relationships. I will call the vector
electric force by E, the magnetic induction by M, the electric
induction by e and the magnetic force by m,
so that E, M, e, m are used instead of E, B, D, H used by Lorentz.
I will also use the complex variables in a manner that has not been
previously common practice in physical studies, i.e., insteadpof operating
with t, I will operate with it, where i is the imaginary unit 1. On the
other hand, very important circumstances become evident as I use a notation
with indices, namely, often instead of
x, y, z, it x1 , x 2 , x 3 , x 4
is used which will then establish a general use of the indices 1, 2, 3, 4. The
advantage of this method will be, as I explicitly emphasize here, that it only
provides a more evident expression of purely real relations; we can however
at any moment go to real equations if it is assumed that symbols with one

index 4 represent purely imaginary values, while those which have no index
4, or have it twice represent real values.
An individual system of values of x, y, z, t, i. e. of x1 , x2 , x3 , x4 will
be called a spacetime point.
Further let w denote the velocity vector of matter, " the dielectric con-
stant, µ, the magnetic permeability, the conductivity of matter, which all
are functions of x, y, z, t (or x1 , x2 , x3 , x4 ), while ⇢ denotes the density
of electricity in space, and s the vector of “electric current” which we will
introduce in §7 and §8.
First Part.
Consideration of aether as a limiting case
§2.
The basic equations for ether
By using the electron theory, Lorentz in his above mentioned essay traces
the laws of electrodynamics of ponderable bodies to still simpler laws. Let us
now adhere to these simpler laws, whereby we require that for the limiting
case " = 1, µ = 1, = 0, they should constitute the laws for ponderable
bodies. In this ideal limiting case " = 1, µ = 1, = 0, E = e, and M = m.
At every space time point x, y, z, t we shall have the equations
@e
curl m = %w, (I)
@t
div e = %, (II)
@m
curl e + = 0, (III)
@t
div m = 0. (IV)
p
I will now write x1 , x2 , x3 , x4 for x, y, z, it (i = 1) and
%1 , %2 , %3 , %4
55
for
%wx , %wy , %wz , i%
i.e. the components of the convection current %w and the multiplied by i
electric density. Further I will write
f23 , f31 , f12 , f14 , f24 , f34
for
mx , my , mz , iex , iey , iez
i.e., the components of m and ie along the three axes; finally for any two
indices h, k of the numbers 1, 2, 3, 4
fkh = fhk ,
therefore
f32 = f23 , f13 = f31 , f21 = f12 ,
f41 = f14 , f42 = f24 , f43 = f34 .
Then the three equations in (I), and the equation (II) multiplied by i
become:
@f12 @f13 @f14
+ + = %1 ,
@x2 @x3 @x4
@f21 @f23 @f24
+ + = %2 ,
@x1 @x3 @x4
@f31 @f32 @f34
+ + = %3 , (A)
@x1 @x2 @x4
@f41 @f42 @f43
+ + = %4 .
@x1 @x2 @x3
On the other hand, the three equations in (III) multiplied by i, and
the equation (IV) multiplied by 1, become
@f34 @f42 @f23
+ + = 0,
@x2 @x3 @x4
@f43 @f14 @f31
+ + = 0,
@x1 @x3 @x4
@f24 @f41 @f12
+ + = 0, (B)
@x1 @x2 @x4
@f32 @f13 @f21
+ + = 0.
@x1 @x2 @x3
One notices immediately in this way of writing the perfect symmetry of

the 1st as well as the 2nd system of equations with respect to permutation
of the indices 1, 2, 3, 4.
§3.
The Theorem of Relativity of Lorentz
It is well-known that writing the equations (I)–(IV) in vectorial notation

reveals an invariance (or rather a covariance) of the system of equations (A)
as well as of (B), when the coordinate system is rotated around the origin.
For example, if we take a rotation of the axes about the z axis at an angle
', keeping e, m, w fixed in space, and instead of x1 , x2 , x3 , x4 introduce
new variables x01 , x02 , x03 , x04 by
x01 = x1 cos ' + x2 sin ', x02 = x1 sin ' + x2 cos ',
x03 = x3 , x04 = x4 ,
and introduce new quantities %01 , %02 , %03 , %04 by
%01 = %1 cos ' + %2 sin ', %02 = %1 sin ' + %2 cos ',
%03 = %3 , %04 = %4 ,
0 , · · · , f 0 by
and also new quantities f12 34
0 0 0
f23 = f23 cos ' + f31 sin ', f31 = f23 sin ' + f31 cos ', f12 = f12 ,
0 0 0
f14 = f14 cos ' + f24 sin ', f24 = f14 sin ' + f24 cos ', f34 = f34 ,
0 0
fkh = fhk (h, k = 1, 2, 3, 4),
then out of the equations (A) would follow a corresponding system of primed
equations (A0 ) composed of the newly introduced primed quantities.
Now the theorem of relativity, which was discovered by Lorentz, follows
immediately without any calculation from the symmetry of the equations (A)
and (B) with respect to the indices 1, 2, 3, 4.
I will understand by i a purely imaginary quantity and will consider
the substitution
x01 = x1 , x02 = x2 , x03 = x3 cos i + x4 sin i , x04 = x3 sin i + x4 cos i .

(1)
57
By means of
e e 1 1+q
i tg i = = q, = log nat (2)
e +e 2 1 q
we have
1 iq
cos i = p , sin i = p ,
1 q2 1 q2
p
where 1 < q < 1 and 1 q 2 is taken with positive sign. By writing
x01 = x0 , x02 = y 0 , x03 = z 0 , x04 = it0 , (3)
the substitution (1) takes the form
z qt qz + t
x0 = x, y 0 = y, z 0 = p , t0 = p (4)
1 q 2 1 q2
with only real coefficients.
If now in the above equation describing a rotation about the z axis, we
replace 1, 2, 3, 4 throughout by 3, 4, 1, 2, and ' by i , and at the same time
introduce new quantities %01 , %02 , %03 , %04 through the substitution
%01 = %1 , %02 = %2 , %03 = %3 cos i + %4 sin i ,
%04 = %3 sin i + %4 cos i ,
and also new quantities 0 ,···
f12 0 introduced by
, f34
0 0 0
f41 = f41 cos i + f13 sin i , f13 = f41 sin i + f13 cos i , f34 = f34 ,
0 0 0
f32 = f32 cos i + f42 sin i , f42 = f32 sin i + f42 cos i , f12 = f12 ,
0 0
fkh = fhk (h, k = 1, 2, 3, 4),
then the systems of equations in (A) and (B) are transformed into equations
(A0 ), and (B0 ), and we see that the new equations can be obtained by simply
priming the old equation.
All these equations can be now rewritten in a purely real form and the
final result can be formulated in the following way:
If the real transformations (4) are used, and x0 , y 0 , z 0 , t0 are regarded as
a reference system for space and time, and we introduce6
! !
qw z + 1 w z q
%0 = % p , %0 w0 z 0 = % p , (5)
1 q2 1 q2
6
The equations (5) are here in a di↵erent order, whereas the equations (6) and (7) are
in the same sequence as the above equations, from which they follow.
%0 w0 x0 = %wx , %0 w0 y0 = %wy ,
further
ex qmy qex + my
e0 x0 = p , m0 y0 = p , e0 z 0 = ez (6)
1 q 2 1 q 2
and
mx + qey qmx + ey
m0 x0 = p , e0 y0 = p , m0 z 0 = mz , (7)
1 q 2 1 q2
then for the vectors w0 , e0 , m0 with components w0 x0 , w0 y0 , w0 z 0 ; e0 x0 , e0 y0 , e0 z 0 ;
m0 x0 , m0 y0 , m0 z 0 in the new coordinate system x0 , y 0 , z 0 , and also for the quan-
tity %0 the equations (I0 )–(IV0 ), which are analogous to (I)–(IV), are exactly
satisfied, whereas the system (I), (II) goes into (I0 ), (II0 ) and the system
(III), (IV) into (III0 ), (IV0 ).
We remark that here ex qmy , ey + qmx , ez , are the components of the
vector e + [b m], where b is a vector in the direction of the positive z axis,
and is of absolute value |b| = q, and [b m] is the vector product of b and
m. Analogously mx + qey , my qex , mz are the components of the vector
m [b e].
The equations (6) and (7), as they are in pairs with each other, can be
combined by another use of imaginary quantities in
e0 x0 + im0 x0 = (ex + imx ) cos i + (ey + imy ) sin i ,

0 0
e y0 + im y0 = (ex + imx ) sin i + (ey + imy ) cos i ,
0 0
e z 0 + im z 0 = ez + imz .
and we notice that if ' is any real angle, from the these relationships the
following combinations can be formed:
(e0 x0 + im0 x0 ) cos ' + (e0 y0 + im0 y0 ) sin '

= (ex + imx ) cos(' + i ) + (ey + imy ) sin(' + i ), (8)
(e0 x0 + im0 x0 ) sin ' + (e0 y0 + im0 y0 ) cos '

= (ex + imx ) sin(' + i ) + (ey + imy ) cos(' + i ). (9)
59
§4.
Special Lorentz transformations
The role which is played by the z axis in the transformation (4) can
easily be transferred to any other direction. Each of the systems of coordi-
nate axes x, y, z and x0 , z 0 , y 0 is subjected to the same rotation with respect
to itself. We come now to a more general proposition.
Let b be a vector in any direction with components bx , by , bz , and with
such a non-zero absolute value |b| = q which is smaller than 1. By b we
shall denote any vector which is perpendicular to b, and further denote the
components of a vector r in the direction of b and b as rb and rb , respectively.
Instead of x, y, z, t, new variables x0 , y 0 , z 0 , t0 will be introduced in the
following way. If for the sake of shortness, r is written for the vector with
the components x, y, z in the first system of reference, r0 for the same vector
with the components x0 , y 0 , z 0 in the second system of reference, then for the
direction of b we have
rb qt
r0 b = p , (10)
1 q2
and for each b in a direction perpendicular to b
r0 b = rb , (11)
and also
qrb + t
t0 = p . (12)
1 q2
The notations r0 b and r0 b should be understood in the sense that the
direction of b and every direction b perpendicular to b in the system x, y, z
a direction is always associated with the same direction cosines in the system
x0 , y 0 , z 0 .
A transformation which is represented by (10), (11), (12) with the con-
dition 0 < q < 1 I will call a special Lorentz transformation, and b will be
called the vector, the direction of b – the axis, and the magnitude of b – the
moment of this special Lorentz transformation.
If further %0 and the vectors w0 , e0 , m0 , in the system x0 , y 0 , z 0 are so
defined that,
%( qwb + 1)
%0 = p , (13)
1 q2
%wb %q
%0 w0 b = p , %0 w0 b = %wb (14)
1 q 2
and further7
(e + im i[w, e + im])b
(e0 + im0 )b = p ,
1 q2
(e0 + im0 )b = (e + im i[w, e + im])b , (15)
then it follows that the systems of equations (I), (II), and (III), (IV) every
time are transformed into the exactly corresponding systems with primes.
The solution of the equations (10), (11), (12) leads to
r0 b + qt0 qr0 b + t0
rb = p , rb = r0 b , t = p . (16)
1 q2 1 q2
We shall now make a very important observation about the vectors
w and w0 . We can again introduce the indices 1, 2, 3, 4, so that we write
x01 , x02 , x03 , x04 instead of x0 , y 0 , z 0 , it0 and %01 , %02 , %03 , %04 instead of %0 w0 x0 ,
%0 w0 y0 , %0 w0 z 0 , i%0 . Like the rotation about the z axis, the transformation
(4), and the more general transformations (10), (11), (12), are also linear
transformations with the determinant +1, with the help of which
x21 + x22 + x23 + x24 , i.e. x2 + y 2 + z 2 t2 (17)
is transformed into
2 2 2 2
x01 + x02 + x03 + x04 , i.e. x02 + y 02 + z 02 t02 .
On the basis of the equations (13), (14)
(%21 + %22 + %23 + %24 ) = %2 (1 w2x w2y w2z ) = %2 (1 w2 )
is transformed into %0 2 (1 w0 2 ) or in other words the square root of it

p
% 1 w2 , (18)
which is positive, is an invariant under the Lorentz transformations.

If we divide %1 , %2 , %3 , %4 by this magnitude, we obtain the 4 values
wx wy wz i
w1 = p , w2 = p , w3 = p , w4 = p ,
1 w 2 1 w 2 1 w 2 1 w2
between which the relationship
w12 + w22 + w32 + w42 = 1 (19)

7
The parentheses only combine the terms with the same index, whereas [w, e + im] is
the vector product of w and e + im.
61
exists. Obviously these 4 values are uniquely determined by the vector w

and vice versa – the quantities w1 , w2 , w3 , w4 , where w1 , w2 , w3 are real
and iw4 is real and positive, through the condition (19) uniquely determine
the vector w with absolute value < 1.
The meaning of w1 , w2 , w3 , w4 here is that they are the ratios of
dx1 , dx2 , dx3 , dx4 to
q p
(dx21 + dx22 + dx23 + dx24 ) = dt 1 w2 . (20)
where the di↵erentials are the displacements of matter occupying the space-
time point x1 , x2 , x3 , x4 to an adjacent spacetime point. Now the equations
(10), (11), (12) immediately apply to the matching di↵erentials dx, dy, dz, dt
and dx0 , dy 0 , dz 0 , dt0 and will therefore lead to
2 2 2 2
(dx21 + dx22 + dx23 + dx24 ) = (dx01 + dx02 + dx03 + dx04 ).
After performing the Lorentz transformation the velocity of matter in the

new system of reference for the same spacetime point x0 , y 0 , z 0 , t0 is the vector
0 dy 0 dz 0
w0 with the ratios dx
dt0 , dt0 , dt0 interpreted as components.
Now it is evident that the system of variables
x1 = w 1 , x2 = w 2 , x3 = w 3 , x4 = w 4
by virtue of the Lorentz transformation (10), (11), (12) transforms in the

new system of variables
x01 = w10 , x02 = w20 , x03 = w30 , x04 = w40 ,
which has exactly the same meaning for the velocity w0 after the trans-
formation as the former system of variables for the velocity w before the
transformation.
If in particular the vector b of the special Lorentz transformation is equal
to the velocity vector w of matter at the spacetime point x0 , y 0 , z 0 , t0 then it
follows from (10), (11), (12) that
w10 = 0, w20 = 0, w30 = 0, w40 = i.
Therefore, under these circumstances after the transformation at the cor-

responding spacetime point the velocity w0 = 0, i.e. it is transformed
p to
rest as we can express ourselves. We may call the invariant % 1 w2 the
rest-density of electricity.
§5.
Spacetime Vectors of the I st and II nd kind
By taking into account the main result of the special Lorentz transfor-
mation together with the fact that the system of equations (A) as well the
system of equations (B) is covariant with respect to a rotation of the refer-
ence system about the origin, we obtain the general theorem of relativity. To
make the facts easily comprehensible, it may be more convenient to define
a series of abbreviated terms for the purpose of expressing the ideas in a
concise form, while on the other hand I shall adhere to the practice of using
complex variables in order to make certain symmetries evident.
The linear homogeneous transformation
x1 = ↵11 x01 + ↵12 x02 + ↵13 x03 + ↵14 x04 ,

x2 = ↵21 x01 + ↵22 x02 + ↵23 x03 + ↵24 x04 ,
(21)
x3 = ↵31 x01 + ↵32 x02 + ↵33 x03 + ↵34 x04 ,
x4 = ↵41 x01 + ↵42 x02 + ↵43 x03 + ↵44 x04
of determinant +1, in which all coefficients without the index 4 occurring

once are real, while ↵14 , ↵24 , ↵34 and ↵41 , ↵42 , ↵43 , are purely imaginary, but
↵44 is real and > 0, and through which
2 2 2 2
x21 + x22 + x23 + x24 transforms into x01 + x02 + x03 + x04 ,
I will call a general Lorentz transformation.

If we put
x01 = x0 , x02 = y 0 , x03 = z 0 , x04 = it0 ,
then it immediately arises from here a homogeneous linear transformation
of x, y, z, t to x0 , y 0 , z 0 , t0 with real coefficients, whereby the expression
2 2 2 2
x2 y2 z 2 + t2 transforms into x0 y0 z 0 + t0 ,
and to every such system of variables x, y, z, t with a positive t, for which

this expression > 0, there always corresponds a positive t0 ; the latter is easily
seen from the continuity of the expression in x, y, z, t.
The last vertical column of the system of coefficients (21) has to satisfy
the condition
2 2 2 2
↵14 + ↵24 + ↵34 + ↵44 = 1. (22)
63
If ↵14 = 0, ↵24 = 0, ↵34 = 0, then ↵44 = 1 and the Lorentz transforma-

tion reduces to a simple rotation of the spatial coordinate system about the
origin.
If not all ↵14 , ↵24 , ↵34 are zero and we set
↵14 : ↵24 : ↵34 : ↵44 = bx : by : bz : i,
then the absolute quantity

q
q= b2x + b2y + b2z < 1.
follows from (22). On the other hand, for every set of coefficients ↵14 , ↵24 ,
↵34 , ↵44 satisfying the condition (22) with real values of bx , by , bz , we can
construct the special Lorentz transformation (6) with ↵14 , ↵24 , ↵34 , ↵44 as
the last vertical column, and then every Lorentz transformation with the
same last vertical column of coefficients is composed of the special Lorentz
transformation and a rotation of the spatial coordinate system about the
origin.
The totality of all Lorentz transformations forms a group.
Under a spacetime vector of the I-st kind shall be understood a system
of four quantities %1 , %2 , %3 , %4 with the requirement that for every Lorentz
transformation (21) it should be replaced by the system %01 , %02 , %03 , %04 which
follows from (21) for the variables x01 , x02 , x03 , x04 if for x1 , x2 , x3 , x4 the values
of %1 , %2 , %3 , %4 are taken.
Besides the time-space vector of the I-st kind x1 , x2 , x3 , x4 we shall also
make use of another spacetime vector of the I-st kind u1 , u2 , u3 , u4 , and will
consider the linear combination
f23 (x2 u3 x3 u2 ) + f31 (x3 u1 x1 u3 ) + f12 (x1 u2 x2 u1 )+
(23)
+f14 (x1 u4 x4 u1 ) + f24 (x2 u4 x4 u2 ) + f34 (x3 u4 x4 u3 )
with six coefficients f23 · · · f34 . We note that in vectorial notation this can
be constructed out of four vectors
x 1 , x2 , x3 ; u 1 , u2 , u 3 ; f23 , f31 , f12 ; f14 , f24 , f34
and the constants x4 and u4 , and, on the other hand as a form symmetric in
the indices 1, 2, 3, 4. If we subject x1 , x2 , x3 , x4 and u1 , u2 , u3 , u4 simultane-
ously to the Lorentz transformation (21), the combination (23) is changed
to
0
f23 (x02 u03 x03 u02 ) + f31
0
(x03 u01 x01 u03 ) + f12
0
(x01 u02 x02 u01 )+
0 (24)
+f14 (x01 u04 x04 u01 ) + f24
0
(x02 u04 x04 u02 ) + f34
0
(x03 u04 x04 u03 )
where the six coefficients f23 0 · · · f 0 depend solely on the six f · · · f

34 23 34 and
the sixteen coefficients ↵11 , ↵12 , · · · ↵44 .
We shall define a spacetime vector of the II-nd kind as a system of six
quantities f23 , f31 , f12 , f14 , f24 , f34 with the condition that when subjected to
a Lorentz transformation, it is changed to a new system f23 0 , f0 , f0 , f0 , f0 , f0
31 12 14 24 34
which satisfies the connection between (23) and (24).
Now I express the general theorem of relativity concerning the equations
(I) - (IV), which are the “fundamental equations for Aether,” as follows.
If x, y, z, it (space coordinates, and time t ⇥ i) is subjected to a Lorentz
transformation, and at the same time %wx , %wy , %wz , i% (convection cur-
rent, and charge density ⇥i) is transformed as a spacetime vector of the
I-st kind, while mx , my , mz , iex , iey , iez (magnetic force, and electric in-
duction ⇥ i) is transformed as a spacetime vector of the II-nd kind, then
the system of equations (I), (II) and the system of equations (III), (IV)
transform into exactly the same relationships between the respective newly
introduced variables.
These facts can be more concisely expressed in these words: the system
of equations (I), and (II) as well as the system of equations (III), (IV) are
covariant under any Lorentz transformation, where %w, i% should be trans-
formed as a spacetime vector of the I-st kind, whereas m, ie should be
treated as a vector of the II-nd kind. Or even more concisely:
%w, i% is a spacetime vector of the I-st kind, m, ie is a spacetime vector
of the II-nd kind.
I add a few comments here to clarify the idea of spacetime vector of
the II-nd kind. Invariants of such a vector m, ie for the group of Lorentz
transformations are obviously
m2 e2 = f23
2 2
+ f31 2
+ f12 2
+ f14 2
+ f24 2
+ f34 , (25)
me = i(f23 f14 + f31 f24 + f12 f34 ). (26)
A spacetime vector of the II-nd kind m, ie (where m and e are real
space vectors), may be called singular, when the scalar square (m ie)2 = 0,
i.e. m2 e2 = 0, and at the same time (me) = 0, i.e. the vectors m and
e are equal and perpendicular to each other. If such is the case, these two
properties for the spacetime vector of the II-nd kind remain unchanged for
for every Lorentz transformation.
If the spacetime vector of the II-nd kind m, ie is not singular, we rotate
the spacial coordinate system in such a way that the vector product [me] is
in the direction of the z axis, then mz = 0, ez = 0. Then
(mx iex )2 + (my iey )2 6= 0,
65
e +im
that is exy +imyx is di↵erent from ±i and therefore we can determine a complex
argument ' + i in such a way that
ey + imy
tg(' + i ) = .
ex + imx
Then taking into account the equation (9), we combine the transformation
(1) with a parameter and a subsequent rotation about the z axis at an
angle ', we perform a Lorentz transformation and at the end my = 0, ey = 0,
and therefore m and e will both be directed along the new x axis; through
the invariants m2 e2 and (me) one can fix the final values of these vectors,
and also whether they are of the same or of opposite directions, or whether
one of them is equal to zero.
§6.
Concept of Time
Due to the Lorentz transformations certain changes to the time param-

eter are allowed. Consequently, it is no longer permissible to speak of the
simultaneity of two events per se. The use of this term implies, rather, that
the freedom of choosing 6 parameters for defining a reference system for
space and time is already limited in some way to freedom of choosing only
3 parameters. Only because we are used to this constraint, we consider the
concept of simultaneity of two events as inherently existent8 . The actual
situation may be described in the following way.
Let a reference system x, y, z, t for spacetime points (events) be somehow
known. If a space point A (x0 , y0 , z0 ) at the time t0 = 0 is compared with
a space point P (x, y, z) at the time t and if the time di↵erence t t0 (let
t > t0 ) is smaller than the length AP i.e. smaller than the time required for
the propagation of light from A to P , and if q is the ration tAPt0 < 1, then
by a special Lorentz transformation, in which AP is taken as the axis, and
which has the moment q, we can introduce a new time parameter t0 which
(see equation (12) in §4) has the same value t0 = 0 for both spacetime points
A, t0 , and P, t; therefore the two events can be regarded as simultaneous.
Let us take at the same time t0 = 0, two di↵erent space-points A, B, or
three space-points (A, B, C) which do not lie on the same space-line, and
8
Just as beings which are confined within a narrow region surrounding a point on a
spherical surface may be led to think that a sphere is a geometric figure in which one
diameter is specifically distinguished from the others.
compare with them a space point P , which is outside the line AB, or the
plane ABC at another time t, and let the time di↵erence t t0 (t > t0 )
be smaller than the time which light requires for propagation from the line
AB, or the plane ABC to P . Let q be the quotient of the first to the
second time; then if a Lorentz transformation is carried out in which the
perpendicular from P on AB, or from P on the plane ABC is the axis, and
q is the moment, then all the three (or four) events A, t0 ; B, t0 ; (C, t0 ) and
P, t are simultaneous.
If four space-points, which do not lie on one plane are conceived to be at
the same time t0 , then it is no longer possible to make a change of the time
parameter by a Lorentz transformation without at the same time destroying
the character of the simultaneity of these four space points.
To the mathematician, accustomed on the one hand to the above con-
siderations in multidimensional manifolds, and on the other hand to the
concepts of the so-called non-Euclidean geometry, there can be no difficulty
in adopting this concept of time for the application of the Lorentz trans-
formation. The paper of Einstein which has been cited in the Introduction,
has succeeded to some extent in presenting the nature of the transformation
from a physical standpoint.
Second Part.
The electromagnetic phenomena
§7.
Fundamental Equations for Bodies at Rest
After these preparatory works, which have been first developed on ac-
count of the small amount of mathematics involved in the limiting case
" = 1, µ = 1, = 0, let us turn to the electromagnetic phenomena in matter.
We look for those relations which make it possible for us – under appropri-
ate boundary condition – to obtain the following quantities at every place
and time, and therefore at every spacetime point as functions of x, y, z, t:
the vector of the electric force E, the magnetic induction M, the electrical
induction e, the magnetic force m, the electrical space-density %, the vector
of the “electric current” s (whose relation hereafter to the conduction current
is known by the manner in which conductivity occurs in the process), and
lastly the vector w, the velocity of matter.
The relations in question can be divided into two classes,
67
firstly, those equations, which by using the vector w, the velocity of

matter, as a function of x, y, z, t, allow us to determine other quantities as
functions of x, y, z, t I will specifically call this first class of equations the
fundamental equations,
secondly, the expressions for the ponderomotive forces, which by using
the laws of mechanics give us additional information about the vector w as
a function of x, y, z, t.
For the case of a body at rest, i.e. when w(x, y, z, t) = 0, the theories
of Maxwell (Heaviside, Hertz) and Lorentz lead to the same fundamental
equations. These are
1) the di↵erential equations, which contain no constants related to mat-
ter:
@e
curl m = s, (I)
@t
div e = %, (II)
@M
curl E + = 0, (III)
@t
div M = 0. (IV)
2) additional relationships, which characterize the influence of existing
matter; for the most important case, to which we limit ourselves, i.e. for
isotopic bodies, they take the form
e = "E = 0, M = µm, s = E, (V)

where the dielectric constant ", the magnetic permeability µ, the conductiv-
ity of matter , are all regarded as known function of x, y, z and t. Here s
is the conduction current.
I will now return to these equations by a modified notation and an even
more hidden symmetry will emerge. I use as in the above text
x1 = x, x2 = y, x3 = z, x4 = it
and will write

s1 , s 2 , s 3 , s 4
instead of
sx , sy , sz , i%,
further
f23 , f31 , f12 , f14 , f24 , f34
for
mx , my , mz , iex , iey , iez ,
and also
F23 , F31 , F12 , F14 , F24 , F34
for
Mx , My , Mz , iEx , iEy , iEz ;
finally for all unequal pairs h, k of indices 1, 2, 3, 4 the relation
fkh = fhk , Fkh = Fhk ,
should be satisfied. (The letters f, F should remind us of the word field

(Feld), whereas s – the word current (Strom).)
Then the fundamental equations (I), (II) can be written as:
@f12 @f13 @f14
+ + = s1 ,
@x2 @x3 @x4
@f21 @f23 @f24
+ + = s2 ,
@x1 @x3 @x4
@f31 @f32 @f34
+ + = s3 , (A)
@x1 @x2 @x4
@f41 @f42 @f43
+ + = s4 ,
@x1 @x2 @x3
and the equations (III) and (IV) as
@F34 @F42 @F23
+ + = 0,
@x2 @x3 @x4
@F43 @F14 @F31
+ + = 0,
@x1 @x3 @x4
@F24 @F41 @F12
+ + = 0, (B)
@x1 @x2 @x4
@F32 @F13 @F21
+ + = 0.
@x1 @x2 @x3
69
§8.
The Fundamental Equations for Moving Bodies
We can now determine in an unambiguous way the fundamental equa-

tions for arbitrarily moving bodies through the following three axioms:
The first axiom should be:
When a detached region of matter is at rest at any moment, so in a
system x, y, z, t the vector w is zero – the surroundings of that region may
be understood to be in any motion – then for the spacetime point x, y, z, t the
same relations (A), (B), (V), which hold when all matter is at rest, should
also hold between %, and the vectors s, e, m, E, M and their derivatives
with respect to x, y, z, t.
The second axiom should be:
Every velocity of matter is < 1, i.e. smaller than the velocity of propa-
gation of light.
The third axiom should be:
The fundamental equations are of such kind that when x, y, z, it are sub-
jected to a Lorentz transformation and the the pairs m, ie and M, iE are
transformed as spacetime vectors of the II-nd kind, whereas s, i% is trans-
formed as a spacetime vector of the I-st kind, the equations are transformed
into essentially identical forms involving the transformed quantities.
I will express this third axiom shortly as:
m, ie and M, iE are spacetime vectors of the II-nd kind, whereas s, i%
is a spacetime vector of the I-st kind. I call this axiom the principle of
relativity.
In fact these three axioms unambiguously lead us from the previously
formulated fundamental equations for bodies at rest to the fundamental
equations for moving bodies.
Namely, according to the second axiom in each spacetime point the mag-
nitude of the velocity vector |w| < 1. As a result, we can always represent
the vector w by the following set of four quantities
wx wy
w1 = p , w2 = p ,
1 w2 1 w2
wz i
w3 = p , w4 = p ,
1 w2 1 w2
between which the relationship
w12 + w22 + w32 + w42 = 1 (27)

holds.
From the concluding remarks of §4 it is seen that under Lorentz trans-
formations this quadruple behaves as a spacetime vector of the I-st kind,
and we will to call it spacetime velocity vector.
Let us now have a specific point x, y, z of matter at a given time t. If
at this spacetime point w = 0, then, according to the first axiom, we have
for this point the equations (A), (B), (V) of §7. If w 6= 0 and since |w| < 0
then there exists according to (16) a special Lorentz transformation, whose
vector b is equal to the vector w(x, y, z, t) and this transformation leads us
to a new system of reference x0 , y 0 , z 0 , t0 . For the considered spacetime as we
saw in §4 new values arise
w10 = 0, w20 = 0, w30 = 0, w40 = i, (28)
so the new velocity vector w0 = 0 and the spacetime point is, as we put
it then, as if transformed to rest. Now according to the third axiom from
the fundamental equations for the spacetime point x, y, z, t one should ob-
tain the corresponding fundamental equations for the point x0 , y 0 , z 0 , t0 , ex-
pressed through the transformed quantities w0 , %0 , s0 , e0 , m0 , E0 , M0 and
their derivatives with respect to x0 , y 0 , z 0 , t0 . According to the first axiom
and since w0 = 0 these equations must be exactly equivalent to
1) the di↵erential equations (A0 ), (B0 ) which are obtained from the equa-
tions (A), (B) by simply priming the symbols in (A) and (B),
2) and the equations
e0 = "E0 , M0 = µm0 , s0 = E0 , (V0 )
where ", µ, are the dielectric constant, magnetic permeability, and conduc-
tivity for the system x0 , y 0 , z 0 , t0 corresponding to the properties of matter in
the spacetime point x, y, z, t.
Now we go through the inverse Lorentz transformation to the original
variables x, y, z, t, and the quantities w, %, s, e, m, E, M; then the equa-
tions, which we obtain from this transformation, will be the fundamental
equations sought by us for the moving bodies.
Now from §4, and §6, it can be seen that the system of equations (A), as
well as the system of equations (B) are covariant under the Lorentz trans-
formations, i.e. the equations, which we obtain backwards from (A0 ) (B0 ),
must be exactly of the same form as the equations (A) and (B), which hold
for bodies at rest. We have therefore as the first result:
The di↵erential equations expressing the fundamental equations of elec-
trodynamics for moving bodies, when written in % and the vectors s, e, m, E,
71
M are exactly of the same form as the equations for bodies at rest. The ve-
locity of matter does not enter in these equations. In vector form these
equations are therefore once again
@e
curl m = s, (I)
@t
div e = %, (II)
@M
curl E + = 0, (III)
@t
div M = 0. (IV)
The velocity of the matter is only referred to the additional conditions
that characterize the influence of matter through the special constants ", µ, .
Let us now transform these additional conditions (V’) into the original co-
ordinates x, y, z, and the original time t.
According to formula (15) in §4, the component of e0 in the direction of
the vector w is the same us that of e+[wm], the component of m0 is the same
as that of m [we], but for the perpendicular direction w the components of
e0 and m0 are the same as those of e+[wm] and m [we], multiplied by p1 1 w2 .
On the other hand, E0 and M0 stand to E + [wM], and M [wE] in the same
relation as e0 and m0 to e + [wm] and m [we]. Then, by distinguishing in the
relation e0 = "E0 three components – one parallel to the vector w and the
other two mutually perpendicular and perpendicular
p to w – in which case
the last two components are multiplied by 1 w2 , we have
e + [wm] = "(E + [wM]). (C)

Analogously, the relation M0 = µm0 leads to
M [wE] = µ(m [we]). (D)
Further it follows from the transformation equations (12), (10), (11) in

§4, when we replace q, rb , rb , t, r0 b , r0 b , t0 by |w|, sw , sw , %, s0 w , s0 w , %0 that
|w|sw + % sw |w|%
%0 = p , s0 w = p , s0 w = sw
1 w 2 1 w 2
so that from s0 = E0 now we have

sw |w|%
p = (E + [wM])w
1 w2
(E)
(E + [wM])w
sw = p
1 w2
According to the way in which the conductivity enters into these relations
it will be appropriate to call the vector s %w conduction current with
components sw %|w| in the direction of w, and sw in the direction of w;
this vector vanishes for = 0.
We remark that for " = 1, µ = 1 the equations e0 = E0 , m0 = M0 through
the inverse Lorentz transformation with the vector w are immediately
transformed to the equations e = E, m = M, and for = 0 the equation
s0 = 0 becomes s = %w, so that as a result the limiting case of the equations
obtained here with % = 1, µ = 1, = 0 turn out to be the “fundamental
equations of Aether” discussed in §2.
§9.
The fundamental equations in the theory of Lorentz
Now let us see to what extent the fundamental equations accepted by

Lorentz correspond to the relativity postulate as defined in §8. In the article
“Electron theory” (Enzykl. der math. Wiss., V2, Art. 14) Lorentz used
di↵erential equations for any possible, even magnetized bodies (see there
page 209, Eq. XXX’, formula (14) on page 78 of the same issue):
@D
curl (H [wE]) = J + + w div D curl [wD] (IIIa00 )
@t
div D = % (I00 )
@B
curl E = (IV00 )
@t
div B = 0. (IV00 )
Then for moving non-magnetized bodies, Lorentz puts (page 223, 3)
µ = 1, B = H and assumes the occurrence of the dielectric constant " and
conductivity according to
D E = (" 1)(E + [wE]), (Eq. XXXIV000 , p. 227)
J = (E + [wE]). (Eq. XXXIII00 , p. 223)
73
The Lorentz symbols E, B, D, D are replaced here by E, M, e, m whereas

Lorentz denotes the conduction current by J.
The last three of the di↵erential equations cited above now coincide
immediately with the equations (II), (III), (IV) here, but the first equation,
since = 0 for vanishing currents the vector J can be identified with the
current s w%, becomes
@D
curl (H [wE]) = s + curl [wD], (29)
@t
which does not coincide with (I) here. Therefore the general di↵erential
equations of Lorentz for any magnetized body do not correspond to the
relativity principle.
On the other hand, the form corresponding to the relativity principle
for the condition of non-magnetization should be taken from (D) in §8 with
µ = 1 not as B = H, as Lorentz did, but as
B [wE] (from M [wE] = m [we]), (30)
Now by putting H = B, the di↵erential equation (29) is transformed into the
same form as equation (I) here when m [we] = M [wE]. So it comes with
a compensation of two contradictions with the relativity principle that the
di↵erential equations of Lorentz for moving non-magnetized bodies finally
agree with the relativity principle.
If we make use of (30) for non-magnetized bodies, and put accordingly
H = B + [w, D E], then as a consequence of (C) in §8
(" 1)(E + [wB]) = D E + [w[w, D E]],
i.e. in the direction of w
(" 1)(E + [wB])w = (D E)w ,
and in a direction w perpendicular to w
(" 1)(E + [wB])w = (1 w2 )(D E)w ,
i.e. it coincides with Lorentz’ assumption, if we neglect terms of the order
of w2 which are small compared to 1.
Also to the same order of approximation, Lorentz’ form for J corresponds
to the conditions imposed by the relativity principle (comp. (E) in §8) that
the components p Jw and Jw should be equal to the components of (E + wB)
2 1
multiplied by 1 w and 1 w2 , respectively.
p
§10.
The fundamental equations of E. Cohn
E. Gohn9 adopted the following fundamental equations:
@E
curl (M + [wE]) = + w div E + J
@t (31)
@M
curl (E [wM]) = + w div M,
@t
J = E, E = "E [wM ]), M = µM + [wE], (32)
where where E, M are the electric and magnetic field intensities (forces),
E, M are the electric and magnetic polarization (induction). These equations
also permit the existence of true magnetism; if we do not take into account
this consideration, it should be taken that div M = 0.
An objection to this system of equations is that for " = 1, µ = 1 the
vectors force and induction do not coincide. If, however, in the equations
for the electric and magnetic force we use not E and M , but E [wM] and
M + [wE], and substitute E, M, E, M, div E with e, M, E + [wM], m [we], %,
then his di↵erential equations transform to our equations, and the conditions
(32) take the form
J = (E + [wM]),
e + [w, m [we]] = "(E + [wM]),
M [w, E + [wM]] = µ(m [we]),
and then the equations of Cohn obtain the form required by the relativity
principle, if terms of the order of w2 are neglected compared to 1.
It should be mentioned that the equations of Hertz (in the notation of
Cohn) di↵er from (31) with the additional conditions
E = "E, M = µM, J = E, (33)
9
Gött. Nachr. 1901, S. 74 (also in Ann. d. Phys. 7 (4), 1902, S. 29)
75
but it seems it is not possible to interpret the symbols there in terms of

observable quantities in order to arrive at an agreement with the relativity
principle even as an approximation of the order of w2 .
§11.
Typical representation of the fundamental equations
In the statement of the fundamental equations, our leading idea had

been that they should retain a covariance of form, when subjected to a
group of Lorentz transformations. Now we have to deal with ponderomotive
reactions and energy in the electromagnetic field and there should be no
doubt that the resolution of this question is in some way connected with the
simplest forms which can be given to the fundamental equations, satisfying
the conditions of covariance. In order to arrive at such forms, I shall first of
all put the fundamental equations in a typical form which brings out clearly
their covariance in case of a Lorentz transformation. Here I am using a
method of calculation, which enables us to deal in a simple manner with the
spacetime vectors of the I-st, and II-nd kind, and of which the rules, as far
as required are given below.
10 . A system of quantities
a11 , · · · , a1q
.. .. ,
. .
ap1 , . . . , apq
arranged in p horizontal rows, and q vertical columns is called a p ⇥ q
matrix 10 and will be denoted by the letter A.
If all the quantities ahk are multiplied by c, the resulting matrix with
quantities cahk will be denoted by cA.
If the roles of the horizontal rows and vertical columns be interchanged,
we obtain a q ⇥ p matrix, which will be known as the transposed matrix of
A, and will be denoted by A:
a11 , · · · , ap1
A= .. .. .
. .
a1q , . . . , apq
10
You might also think that instead of Cayley’s matrix calculus one could use Hamilton’s
quaternion calculus, but the latter seems to me to be too restrictive and cumbersome for
our purposes.
If we have a second matrix with the same p and q series as in A
b11 , · · · , a1q
B= .. .. ,
. .
bp1 , . . . , apq
then A + B shall denote the same p ⇥ q matrix whose members are ahk + bhk .
20 . If we have two matrices
a11 , · · · , a1q b11 , · · · , a1r

A= .. .. , B= .. .. ,
. . . .
ap1 , . . . , apq bq1 , . . . , aqr
where the number of horizontal rows of B, is equal to the number of vertical
columns of A, then by AB, the product of the matrices A and B, will be
denoted the matrix
c11 , · · · , c1r
C= .. .. ,
. .
cp1 , . . . , cqr
where each element is formed by combination of the horizontal rows of A
with the vertical columns of B according to the rule
✓ ◆
h = 1, 2, . . . , p
chk = ah1 b1k + ah2 b2k + . . . + ahk bqk
k = 1, 2, . . . , r
For such products, the associative law (AB)S = A(BS) holds, where S is a
third matrix whose horizontal rows are equal to the vertical columns of B
(or AB).
For the transposed matrix of C = BA we have C = BA.
30 . Here it will only be considered matrices with no more than four
horizontal rows and no more than four vertical rows.
A unit matrix (in matrix equations it will be for brevity given as 1) will
be defined as the following 4 ⇥ 4 matrix with the elements
e11 , e12 , e13 , e14 1, 0, 0, 0

e21 , e22 , e23 , e24 0, 1, 0, 0
= (34)
e31 , e32 , e33 , e34 0, 0, 1, 0
e41 , e42 , e43 , e44 0, 0, 0, 1
77
For products c · 1 with the participation of the unit matrix (as a special case
of the matrix product cA introduced in 10 ) for the sake of brevity we will
simply write c in matrix equations.
For a 4 ⇥ 4 A matrix, DetA will denote the determinant formed of the
4 ⇥ 4 elements of the matrix. If DetA 6= 0, then there is a reciprocal matrix
A 1 corresponding to A so that A 1 A = 1.
A matrix
0, f12 , f13 , f14

f21 , 0, f23 , f24
f= ,
f31 , f32 , 0, f34
f41 , f42 , f43 , 0
whose elements fulfil the relation fkh = fhk , is called an alternating matrix.
These relations say that the transposed matrix f = f . Then by f ⇤ will be
the dual, alternating matrix
0, f34 , f42 , f23

f43 , 0, f14 , f31
f⇤ = . (35)
f24 , f41 , 0, f12
f32 , f13 , f21 , 0
Then
f ⇤ f = f32 f14 + f13 f24 + f21 f34 (36)
is a 4⇥4 matrix in which all elements outside the main diagonal from top left
to bottom right are zero and that all elements in this diagonal correspond
to each other and are equal to the combination of the coefficients f given
by(36). The determinant of f turns out to be the square of this combination
1
and we want by Det 2 f to denote the expression
1
Det 2 f = f32 f14 + f13 f24 + f21 f34 (37)
40 . Any linear transformation
xk = ↵h1 x01 + ↵h2 x02 + ↵h3 x03 + ↵h4 x04 (h = 1, 2, 3, 4) (38)

can be represented by the 4 ⇥ 4 matrix with coefficients
↵11 , ↵12 , ↵13 , ↵14

↵21 , ↵22 , ↵23 , ↵24
A= .
↵31 , ↵32 , ↵33 , ↵34
↵41 , ↵42 , ↵43 , ↵44
which we will call transformation A. By the transformation A, the expression
x21 + x22 + x23 + x24
changes into the quadratic form
⌃ahk x0h x0k (h, k = 1, 2, 3, 4),
where
ahk = ↵1h ↵1k + ↵2h ↵2k + ↵3h ↵3k + ↵4h ↵4k ,
so that the 4 ⇥ 4 (symmetric) matrix formed by the coefficients of this form
turns out to be the product AA of the transposed matrix of A and the matrix
A itself. If the transformation results in the form
2 2 2 2
x01 + x02 + x03 + x04 ,
the product of A and A must be the matrix 1.
AA = 1. (39)
This relation is satisfied by A if its elements correspond to a Lorentz trans-

formation in the transformation (38). For the determinant of A it follows
out from (39) that (Det A)2 = 1, Det A = ±1. The condition (39) can be
written as
A 1 = A, (40)
i.e. the reciprocal matrix of A is equivalent to the transposed matrix of A.
For A representing a Lorentz transformation, we find that Det A = +1,
that the coefficients ↵14 , ↵24 , ↵34 , ↵41 , ↵42 , ↵43 are purely imaginary (or zero),
the other coefficients are real, and ↵43 > 0.
50 . Any spacetime vector of the I-st kind s = (s1 , s2 , s3 , s4 ) should be
represented by a 1 ⇥ 4 matrix which is formed by its four components:
s = |s1 , s2 , s3 , s4 |, (41)
then a Lorentz transformation A turns it into sA

79
A spacetime vector of the II-nd kind with components f23 , f31 , f12 , f14 ,
f24 , f34 will be represented by the alternating matrix
0, f12 , f13 , f14

f21 , 0, f23 , f24
f= , (42)
f31 , f32 , 0, f34
f41 , f42 , f43 , 0
and under a Lorentz transformation (see the rules (23), (24) in §5) f should
be replaced by Af A = A 1 f A. Using the expression (37), we have the
1 1 1
identity Det 2 (Af A) = Det A Det 2 f . Therefore Det 2 f is an invariant under
the Lorentz transformations (see also equation (26) in §5).
For the dual matrix f ⇤ we have by taking into account (36):
1 1
1 ⇤ 1 1 ⇤ 1
(A f A)(A f A) = A f f A = Det 2 f · A A = Det 2 f,
from where it is easy to see that the dual matrix f ⇤ is a spacetime vector
of the II-nd kind since it transforms exactly like the spacetime vector of the
II-nd kind f . Therefore f ⇤ with components f14 , f24 , f34 , f23 , f31 , f21 will be
called the dual spacetime vector of f .
60 . If w and s are two spacetime vectors of the I-st kind then by ws (as
well as by sw) we will be understand the combination of their components
w1 s1 + w2 s2 + w3 s3 + w4 s4 . (43)
Under a Lorentz transformation A this combination is invariant because of

(wA)(As) = ws. When ws = 0 w and s are normal to each other.
Two spacetime vectors of the I-st kind w, s form a 2 ⇥ 4 matrix
w1 , w2 , w3 , w4
.
s1 , s 2 , s 3 , s 4
It can be seen immediately that the system of six quantities
w2 s3 w 3 s2 , w 3 s1 w 1 s3 , w 1 s2 w 2 s1 , w 1 s4 w4 s1 ,
(44)
w2 s4 w 4 s2 , w 3 s4 w4 s3
behaves under the Lorentz transformations as a spacetime vector of the II-

nd type. This vector of the II-nd kind with the components (44) is denoted
1
by [w, s]. It is easily seen that Det 2 [w, s] = 0. The dual vector of [w, s] will
be written as [w, s]⇤ .
If w is a spacetime vector of the I-st kind, and f is a spacetime vector
of the II-st kind, then the product wf is a 1 ⇥ 4 matrix. Under a Lorentz
transformation A the quantity w goes into w0 = wA, f into f 0 = A 1 f A;
then w0 f 0 = wAA 1 f A = (wf )A, i.e. wf transforms as a spacetime vector
of the I-st kind.
If w is a spacetime vector of the I-st kind and f is a vector of the II-nd
kind, the important identity can be easily verified
[w, wf ] + [w, wf ⇤ ]⇤ = (ww)f. (45)
The sum of the two space time vectors of the II-nd kind on the left side should
be understood in the sense of the addition of two alternating matrices.
Concretely for w1 = 0, w2 = 0, w3 = 0, w4 = i we have
wf = |if41 , if42 , if43 , 0|; wf ⇤ = |if32 , if13 , if21 , 0|;

[w, wf ] = 0, 0, 0, f41 , f42 , f43 ; [w, wf ⇤ ] = 0, 0, 0, f32 , f13 , f21 ,
and in this particular case the relation(45) is satisfied, which is sufficient

to show that the identity (45) holds generally because this relation has a
covariant character under a Lorentz transformation, and is homogeneous in
w1 = 0, w2 = 0, w3 = 0, w4 = i.
After these preparations, we deal first with the equations (C,) (D,) (E)
by which the constants ", µ, are introduced.
Instead of the space vector w, the velocity of matter, we will introduce,
as in §8, the spacetime vector of the I-st kind w with its four components
wx wy wz i
w1 = p , w2 = p , w3 = p , w4 = p ,
1 w2 1 w2 1 w2 1 w2
where
(ww) = w12 + w22 + w32 + w42 = 1 (46)
and iw4 > 0.

81
Now F and f will be again understood as the occurring in the funda-

mental equations spacetime vectors of the II-nd kind, M, iE, and m, ie.
Through = wF we have another spacetime vector of the I-st kind
with components
1 = w2 F12 + w3 F13 + w4 F14

2 = w1 F21 + w3 F23 + w4 F24
3 = w1 F31 + w2 F32 + w4 F34
4 = w1 F41 + w2 F24 + w3 F34 .
The first three quantities 1, 2, 3 are the x-,y-,z-components of the space-
vector
E + [wM]
p , (47)
1 w2
and
i(wE)
4 =p . (48)
1 w2
Since the F is an alternating matrix, we obviously have
w = w1 1 + w2 2 + w3 3 + w4 4 = 0, (49)
which shows that the vector is normal to w; this relation can be also
written as
4 = i(wx 1 + wy 2 + wz 3 ). (50)
I will call the spacetime vector of the I-st kind the electric rest force.
Analogous relations between such as between w, F, E, M, w hold also
between wf, e, m, w, and, in particular, wf is normal to w. The relation
(C) can be now replaced with
wf = "wF. {C}
a formula which, although provides four equations for the relevant compo-
nents, but such that the fourth, due to (50), is a consequence of the first
three.
Let us now form the time-space vector I-st kind = iwf ⇤ whose com-
ponents are
1 = i( w2 f34 + w3 f42 + w4 f23 )

2 = i(w1 f43 + w3 f14 + w4 f31 )
3 = i(w1 f24 + w2 f41 + w4 f12 )
4 = i(w1 f32 + w2 f13 + w3 f21 ).
The first three 1, 2, 3, are the x-, y-, z-components of the space-
vector
m [we]
p , (51)
1 w2
and also
i(wm)
4 =p . (52)
1 w2
and between them there exists the relation
w = w1 1 + w2 2 + w3 3 + w4 4 = 0, (53)
which can be also written as
4 = i(wx 1 + wy 2 + wz 3 ). (54)
which shows that the vector is normal to w. I will call the spacetime
vector of the I-st kind the magnetic rest force.
Relations analogous to the ones between imf ⇤ , m, e, w hold also between
imF ⇤ , M, E, w, and the relation (D) can be replaced with
wF ⇤ = µwf ⇤ . {D}
The relations C and D can be used to express the field vectors F and f
in terms of and . We have
wF = , wF ⇤ = iµ , wf = " , wf ⇤ = i ,
83
and applying the rule (45) by taking into account (46) leads to
F = [w, ] + iµ[w, ]⇤ , (55)

⇤
f = "[w, ] + i[w, ] , (56)
i.e. explicitly
F12 = (w1 2 w2 1) + iµ(w3 4 w4 3 ), etc.

f12 = "(w1 2 w2 1) + i(w3 4 w4 3 ), etc.
We also find the spacetime vector of the II-nd kind [ , ] with the six
components
2 3 3 2, 3 1 1 3, 1 2 2 1,
1 4 4 1, 2 4 4 2, 3 4 4 3.
Then the corresponding spacetime vector of the I-st kind
w[ , ] = (w ] + (w )
vanishes identically due to (49) and (53). Let us now consider the spacetime
vector of the I-st kind
⌦ = iw[ , ]⇤ (57)
with the components
w2 , w2 , w3
⌦1 = i 2, 3, 4 , etc.
2, 3, 4
and by applying rule (45) it follows
[ , ] = i[w, ⌦]⇤ (58)
i.e. explicitly
1 2 2 1 = i(w3 ⌦4 w4 ⌦3 ), etc.
The vector ⌦ obviously satisfies the relation
(w, ⌦) = w1 ⌦1 + w2 ⌦2 + w3 ⌦3 + w4 ⌦4 = 0, (59)
which can be written as
⌦4 = i(wx ⌦1 + wy ⌦2 + wz ⌦3 ),
which means that ⌦ is normal to w. If w = 0 one has 4 = 0, 4 = 0, ⌦4 = 0
and
⌦1 = 2 3 3 2, ⌦2 = 3 1 1 3, ⌦3 = 1 2 2 1. (60)
I will call the spacetime vector of the I-st kind rest ray.
Regarding the relation (E), which introduces the conductivity , we see
first that
{w}sw + %
ws = (w1 s1 + w2 s2 + w3 s3 + w4 s4 ) = p = %0
1 w2
is the rest density of electricity (see §8 and the end of §4). Then I will call
the spacetime vector of the I-st kind
s + (ws)w, (61)
which due to ww = 1 is obviously normal to w, the rest current. Let us

regard the first three components of this vector as the x-,y-,z-components
of a space vector, so the component in the direction of w is:
|w|%0 sw |w|% Jw
sw p = 2
= ,
1 w 2 1 w 1 w2
and the component in any perpendicular to w direction w
sw = Jw ;
this space vector is analogous to the space vector J = s %w which we called
in §8 the conduction current.
Now, by comparison with = wF the relation (E) can be written in
the form:
85
s + (ws)w = wF. {E}

This formula contains four equations, but since on both sides there are
spacetime vectors of the I-st kind which are normal to w, the fourth is a
consequence of the first three.
Finally, we will transform the di↵erential equations (A) and (B) into a
typical form.
§12.
The di↵erential operator Lor
A 4 ⇥ 4 matrix
S11 , S12 , S13 , S14

S21 , S22 , S23 , S24
S= = |Shk |, (62)
S31 , S32 , S33 , S34
S41 , S42 , S43 , S44
which obeys the rule that under a Lorentz transformation A it is always

replaced by ASA, can be called a spacetime matrix of the II-nd kind. Such
a matrix is one in particular:
the alternating matrix f , which corresponds to the spacetime vector of
the II-nd kind f ,
the product f F of two such matrices f and F which under a transfor-
mation A is replaced by (A 1 f A)(A 1 F A) = A 1 f F A,
the 4⇥4 matrix with the elements Shk = wh ⌦k which are the products of
the components w1 , w2 , w3 , w4 and ⌦1 , ⌦2 , ⌦3 , ⌦4 of two spacetime vectors
of the I-st kind,
finally, a 4 ⇥ 4 matrix proportional to the unit matrix, i.e. a multiple
L of the unit matrix, whose elements in the main diagonal have the same
value L and the remaining elements are all zero.
Here we always deal with functions of spacetime points x, y, c, it and
it can be advantageous to use a 1 ⇥ 4 matrix formed from the di↵erential
symbols
@ @ @ @
, , , ,
@x @y @z i@t
or also
@ @ @ @
, , , . (63)
@x1 @x2 @x3 @x4
For this matrix, I will need the abbreviation lor.

If S as in (62) is a spacetime matrix of the II-nd kind, then by the rule
for product formation of matrices, under lor S we will understand the 1 ⇥ 4
matrix
|K1 , K2 , K3 , K4 | ,
with the expressions
@S1k @S2k @S3k @S4k

Kk = + + + (k = 1, 2, 3, 4). (64)
@x1 @x2 @x3 @x4
When a Lorentz transformation A introduces a new reference system

x01 , x02 , x03 , x04 for the spacetime point, so the analogous operator can be con-
structed
@ @ @ @
lor0 = , , , .
@x01 @x02 @x03 @x04
As S transforms into S 0 = ASA = |Shk
0 |, so lor0 S 0 should be a 1 ⇥ 4 matrix,
with the expressions

0
@S1k 0
@S2k 0
@S3k 0
@S4k
Kk0 = + + + (k = 1, 2, 3, 4).
@x01 @x02 @x03 @x04
Now for the di↵erentiation of any function of a spacetime point applies the
rule
@ @ @x1 @ @x2 @ @x3 @ @x4

0 = 0 + 0 + 0 + =
@xk @x1 @xk @x2 @xk @x3 @xk @x4 @x0k
@ @ @ @
= ↵1k + ↵2k + ↵3k + ↵4k ,
@x1 @x2 @x3 @x4
which in an easily understandable way symbolically is
lor0 = lor A
from which it follows immediately
87
lor0 S 0 = lor (A(A 1

SA)) = (lor S)A, (65)
i.e. if S is a spacetime matrix of the II-nd kind, then lor S transforms as a

spacetime vector of the I-st kind.
If L is a multiple of the unit matrix, then lor L is the matrix with elements
@L @L @L @L
, , , . (66)
@x1 @x2 @x3 @x4
If s = |s1 , s2 , s3 , s4 | is a spacetime vector of the I-st kind, then
@s1 @s2 @s3 @s4

lor s = + + + . (67)
@x1 @x2 @x3 @x4
If upon application of a Lorentz transformation A the signs lor0 , and s0 are

used in place of lor and s, it follows that
lor0 s0 = (lor A)(As) = lor s,

i.e. lor s is an invariant under the Lorentz transformations.
In all these relations the operator lor plays the role of a spacetime vector
of the I-st kind
If f is a spacetime vector of the II-nd kind, lor f is a spacetime vector of
the I-st kind with components
@f12 @f13 @f14

+ + ,
@x2 @x3 @x4
@f21 @f23 @f24
+ + ,
@x1 @x3 @x4
@f31 @f32 @f34
+ + ,
@x1 @x2 @x4
@f41 @f42 @f43
+ + .
@x1 @x2 @x3
After that the system of di↵erential equations (A) can be expressed in the
short form
lor f = s. {A}
In exactly the same way the system of di↵erential equations (B) can be
written as:
lor F ⇤ = 0. {B}
In regard to definition (67) the compounds formed from lor (lor f ) and
lor (lor F ⇤ ) vanish identically when f and F ⇤ are alternating matrices. After
this, the relationship between the components of the current s
@s1 @s2 @s3 @s4

+ + + =0 (68)
@x1 @x2 @x3 @x4
follows from {A} while the relation
lor (lor F ⇤ ) = 0 (69)
has the meaning that out of the four equations for the evolution of the field
vectors in {B} only three are independent.
I now summarize the results together:
Let w denote the spacetime vector of the I-st kind p1w w2 , p1 i w2 (w is
the velocity of matter), F – a spacetime vector of the II-nd kind M, E (M
– magnetic induction, E – electric force), f – a spacetime vector of the II-
nd kind m, e (m – magnetic force, e – electric induction), s – a spacetime
vector of the I-st kind s, i% (% – density of the electrical charge, s %w –
conductivity current), " – dielectric constant, µ – magnetic permeability,
– conductivity, then these are (with the explained symbols of matrices in
§10 and §11) the fundamental equations for the electromagnetic processes in
moving bodies
lor f = s {A}
lor F ⇤ = 0 {B}
wf = "wF {C}
wF ⇤ = µwf ⇤ {D}
s + (ws)w = wF. {E}

89
Here ww = 1 and the spacetime vectors of the I-st kind wF , wf , wF ⇤ ,

wf ⇤ , s + (ws)s are normal to w and finally for the system of equations for
{B} the relationship
lor (lor F ⇤ ) = 0
holds.
In view of the the last relation, we see that we have the necessary inde-
pendent equations in order to determine the processes in appropriate bound-
ary conditions, if the motion of matter, expressed by the vector w as a func-
tion of x, y, z, t is given.
§13.
The product of the field vectors f F
Finally, we ask for the laws determining the vector w as a function of

x, y, z, t. In the investigations relating to that, of interest are expressions
resulting from the multiplication of two alternating matrices
0, f12 , f13 , f14 0, F12 , F13 , F14

f21 , 0, f23 , f24 F21 , 0, F23 , F24
f= , F = .
f31 , f32 , 0, f34 F31 , F32 , 0, F34
f41 , f42 , f43 , 0 F41 , F42 , F43 , 0
The matrix product f F can be represented in the form
S11 L, S12 , S13 , S14

S21 , S22 L, S23 , S24
fF = , (70)
S31 , S32 , S33 L, S34
S41 , S42 , S43 , S44 L
such that
S11 + S22 + S33 + +S44 = 0. (71)
L is the symmetrical combination of all indices 1, 2, 3, 4
1
L = (f23 F23 + f31 F31 + f12 F12 + f14 F14 + f24 F24 + f34 F34 ), (72)
2
and then
1
S11 = (f23 F23 + f34 F34 + f42 F42 f12 F12 f13 F13 f14 F14 ),
2 (73)
S12 = f13 F32 + f14 F42 , etc.
As I express the conditions of reality, I want to write
S11 , S12 , S13 , S14 Xx , Yx , Zx , iTx

S21 , S22 , S23 , S24 Xy , Yy , Zy , iTy
S= = , (74)
S31 , S32 , S33 , S34 Xz , Yz , Zz , iTz
S41 , S42 , S43 , S44 iXt , iYt , iZt , Tt
where
1
Xx = (mx Mx my My mz Mz + ex Ex ey Ey ez Ez )
2
Xy = mx My + ey Ex , Yx = my Mx + ex Ey , etc.
Xt = ey Mz ez My , (75)
Ty = mz Ey my Ez , etc.
1
Tt = (mx Mx + my My + mz Mz + ex Ex + ey Ey + ez Ez )
2
and also
1
L = (mx Mx + my My + mz Mz ex Ex ey Ey ez Ez ) (76)
2
are all real. In the theories of bodies at rest the combinations Xx , Xy , Xz ,

Yx , Yy , Yz , Zx , Zy , Zz are known as “Maxwell’s Stresses”, the quantities Tx ,
Ty , Tz are known as the “Poynting’s vector,“ Tt as the “electromagnetic
energy density,” and L as the “Lagrangian function.”
On the other hand, we can take the dual matrices of f and F and multiply
them in reverse order
S11 L, S12 , S13 , S14

S21 , S22 L, S23 , S24
F ⇤f ⇤ = , (77)
S31 , S32 , S33 L, S34
S41 , S42 , S43 , S44 L
91
and we can hereafter set
fF = S L, F ⇤f ⇤ = S L (78)
where by L, we mean the matrix L·1 that is proportional to the unit matrix,
i.e. the matrix with elements
✓ ◆
ehh = 1, ehk = 0, h 6= k
|Lehk | .
h, k = 1, 2, 3, 4
As SL = LS we conclude further
F ⇤f ⇤f F = ( S L)(S L) = SS + L2 ,
1 1
from here, since f ⇤ f = Det 2 f , F ⇤ F = Det 2 F , we find the interesting
relation
1 1
SS = L2 Det 2 f Det 2 F, (79)
i.e. the product of the matrix S with itself is proportional to the unit matrix.
relations
Sh1 S1k + Sh2 S2k + Sh3 S3k + Sh4 S4k = 0 (80)
for unequal indices h, k from the series 1, 2, 3, 4 and
1 1
Sh1 S1h + Sh2 S2h + Sh3 S3h + Sh4 S4h = L2 Det 2 f Det 2 F (81)
for h = 1, 2, 3, 4.
Now if the quantities F and f in (72), (73) are expressed, by taking into
account (55), (56) and (57), through the electrical rest force , the magnetic
rest force and the rest ray ⌦, we arrive at the expressions:
1 1
L= " + µ , (82)
2 2
1 1
Shk = " ehk + µ ehk
2 2
+ "( wh wk ) + µ( wh wk ) (83)
h k h k
⌦h w k "µwh ⌦k (h, k = 1, 2, 3, 4),
where
2 2 2 2 2 2 2 2
= 1 + 2 + 3 + 4, = 1 + 2 + 3 + 4,
ehh = 1, ehk = 0 (h 6= k).

Namely, in any case, the right side of (82) as well as L is an invariant
under Lorentz transformations and the right side of (83) as well as Shk repre-
sent the elements of a 4 ⇥ 4 spacetime matrix of the II-nd kind. Considering
this, it is enough for the relations (82) and (83) to hold generally, if they are
verified only for the case w1 = 0, w2 = 0, w3 = 0, w4 = i. But in this case
w = 0 equations (82), (83) follow, on the one hand, from (47), (51), (60),
and, on the other hand, from e = "E, M = µm directly due to the equations
(75), (76).
The expression on the right in (81)
✓ ◆2
1
= (mM eE) + (em)(EM),
2
due to the relations (em) = " , (EM) = µ is always 0; the square
1
root of it, which is 0, will be in accordance with (79) denoted by Det 4 S.
For S, the transposed matrix of S, as f = f and F = F , it follows
from (78) that
Ff = S L, F ⇤f ⇤ = S L (84)
Then
S S = |Shk Skh |
is an alternating matrix and also means a spacetime vector of the II-nd kind.
From the expressions (83) we immediately have
S S= ("µ 1)[w, ⌦], (85)

93
from where it is also deduced see (57),(58))
w(S S)⇤ = 0, (86)

w(S S) = ("µ 1)⌦. (87)
When at a spacetime point the matter is at rest, i.e. w = 0, then (86)

reduced to the following equations
Zy = Yz , Xz = Z x , Yx = X y ,
and then one has from (83):
Tx = ⌦1 , Ty = ⌦2 , Tz = ⌦3 ,
Xt = "µ⌦1 , Yt = "µ⌦2 , Zt = "µ⌦3
Now by a suitable rotation of the spatial coordinate system x, y, z about the

origin, we can have
Zy = Yz = 0, Xz = Zx = 0, Yx = Xy = 0.
According to (71) one has
Xx + Yy + Zz + Tt = 0, (88)
and always T > 0 as seen from (83). In the special case when ⌦ also vanishes,
one obtains from (81)
1
Xx2 = Yy2 = Zz2 = Tt2 = (Det 4 S)2
1
and if Tt and one of the three magnitudes Xx , Yy , Zz are = +Det 4 S, the re-
1
maining two are = Det 4 S. If ⌦ does not vanish, we can take, for example,
⌦3 6= 0 and then according to (80) in particular
Tz Xt = 0, Tz Yt = 0, Zz Tz + Tz Tt = 0,
and therefore ⌦1 = 0, ⌦2 = 0, Zz = Tt . From (81) and in view of (88)
then it follows
1
Xx = Yy = ± Det 4 S,
q
1 1
Z z = Tt = Det 2 S + "µ⌦23 > Det 4 S.
Finally, of particular importance is the spacetime vector of the I-st kind
K = lor S, (89)
for which we now want to prove an important transformation.

According to (78) S = L + f F and it follows that
lor S = lor L + lor f F.

The symbol lor represents di↵erentiation, which in lor f F applies, on the
one hand, to the components of f , and, on the other hand, to the components
of F . Accordingly, lor f F is additively decomposed into a first and a second
part. The first part is obviously the matrix product (lor f ) F , where lor f
is interpreted by itself as a 4 ⇥ 4 matrix. The second term is that part from
lor f F , in which the di↵erentiations a↵ect only the components of F . Now
we see from (78)
fF = F ⇤f ⇤ 2L,
consequently the second part of lor f F is (lor F )⇤ f ⇤+ the part of 2 lor L,
in which the di↵erentiations apply only to the components of F . After this,
the result is
lor S = (lor f )F (lor F ⇤ )f ⇤ + N, (90)
where N is a vector with components
✓ ◆
1 @f23 @f31 @f12 @f14 @f24 @f34
Nh = F23 + F31 + F12 + F14 + F24 + F34
2 @xh @xh @xh @xh @xh @xh
✓ ◆
1 @F23 @F31 @F12 @F14 @F24 @F34
f23 + f31 + f12 + f14 + f24 + f34
2 @xh @xh @xh @xh @xh @xh
(h = 1, 2, 3, 4).
Using the fundamental equations {A} and {B} transforms (90) into the
fundamental relation
lor S = sF + N. (91)
95
In the limiting case " = 1, µ = 1, when f = F , N vanishes identically.

In general, due to (55), (56) and with regard to the expression (82) for L
and from (57) we arrive the following expressions of the components of N :
1 @" 1 @µ
Nh =
2 @xh 2 @xh
✓ ◆
@w1 @w2 @w3 @w4 (92)
+ ("µ 1) ⌦1 + ⌦2 + ⌦3 + ⌦4
@xh @xh @xh @xh
for h = 1, 2, 3, 4.
By using (59) and introducing the space vector W with ⌦1 , ⌦2 , ⌦3 as

its x-, y-, z-components, then the third component of (92) can be concisely
expressed as
✓ ◆
"µ 1 @w
p W , (93)
1 w2 @xh
where the bracket indicates the scalar product of two vectors therein.
§14.
The ponderomotive forces
In a more detailed form the equation K = lor S = sF +N is represented

by four equations:
@Xx @Xy @Xz @Xt

K1 = + + = %Ex + sy Mz sz My
@x @y @z @t
✓ ◆ (94)
1 @" 1 @µ "µ 1 @w
+p W ,
2 @x 2 @x 1 w2 @x
@Yx @Yy @Yz @Yt

K2 = + + = %Ey + sz Mx sx Mz
@x @y @z @t
✓ ◆ (95)
1 @" 1 @µ "µ 1 @w
+p W ,
2 @y 2 @y 1 w2 @y
@Zx @Zy @Zz @Zt

K3 = + + = %Ez + sx My sy Mx
@x @y @z @t
✓ ◆ (96)
1 @" 1 @µ "µ 1 @w
+p W ,
2 @z 2 @z 1 w2 @z
1 @Tx @Ty @Tz @Tt

K4 = = sx Ex + sy Ey + z Ez
i @x @y @z @t
✓ ◆ (97)
1 @" 1 @µ "µ 1 @w
+ + p W .
2 @t 2 @t 1 w2 @t
It is now my opinion that in the electromagnetic processes the pondero-

motive force which acts on the matter at a spacetime point x, y, z, t, calcu-
lated for a unit volume, has as its x-,y-,z-components, the first three com-
ponents of the spacetime vector
K + (wK)w, (98)
which is normal to the spacetime vector w, and the energy law is represented
by the fourth of the above equations.
Detailed reasons for this opinion will be reserved for a subsequent trea-
tise; here I just want to give some supporting remarks on the mechanics of
this opinion.
In the limiting case " = 1, µ = 1, = 0 we have N = 0, s = %w, from
where wK = 0 and our approach coincides with those commonly used in
the electron theory.
Appendix.
Mechanics and the relativity postulate.
Most likely it would be unsatisfactory if the new understanding of the

concept of time, which is characterized by the freedom of the Lorentz trans-
formations, is applied to only a branch of physics.
97
Now many authors say that classical mechanics is contrary to the rela-
tivity postulate, which is set here as the basis of electrodynamics.
In order to decide this let us concentrate on a special Lorentz trans-
formation as determined by the equations (10), (11), (12) with a non-zero
vector b from any direction and a magnitude q < 1. For the moment we will
not choose in advance any relation between the unit of length and the unit
of time and will write not t, t0 , q, but ct, ct0 , qc , where c is a positive constant
and q < c. The equations mentioned above are therefore transformed into
c(rb qt) qrb + c2 t

r0b = rb , r0b = p , t0 = p ,
c2 q 2 c c2 q 2
where as we recall r is the space vector with components x, y, x and r0 is the
space vector with components x0 , y 0 , z 0 .
If we keep v constant, in the limiting case c = 1 these equations become
r0b = rb , r0b = rb qt, t0 = t.

These new equations would now mean a transition from the current
spatial coordinate system x, y, z to a di↵erent spatial coordinate systems
x0 , y 0 , z 0 with parallel axes, whose origin moves in a straight line with con-
stant velocity with respect to the first, while the time parameter will remain
completely una↵ected.
Based on this observation one might say:
Classical mechanics postulates a covariance of physical laws for the group
of homogeneous linear transformations of the expression
x2 y2 z 2 + c 2 t2 , (1)
with the setting c = 1.

Now it would be absolutely confusing to find that in one branch of physics
we find a covariance of the laws for the transformation of expression (1) with
a finite value of c, whereas in another part – for c = 1.
The situation that Newtonian mechanics deals with this covariance only
for c = 1 and not with the case of c as the speed of light, needs no explana-
tion. Is it then possible to regard that traditional covariance for c = 1 only
as a first approximation, consistent with experience, of the exact covariance
of natural laws holding for a certain finite value of c?
I want to say that by reforming the mechanics such that instead of New-
ton’s postulate of relativity with c = 1 another one with a finite c holds,
even the axiomatic structure of mechanics appears to become considerably

complete.
The ratio of the time unit to length unit is normalized in agreement with
the relativity postulate with c = 1.
As I now want to introduce geometrical figures in the manifold of the
four variables x, y, z, t it may be convenient not to take y, z into account,
and to interpret x and t as any oblique coordinates in a plane.
A spacetime origin O(x, y, z, t = 0, 0, 0, 0) is held unchanged under
Lorentz transformations. The structure
x2 y2 z 2 + t2 = 1, t > 0, (2)
is a hyperboloidal shell, which contains the spacetime point A(x, y, z, t =

0, 0, 0, 1) and all points A0 (x0 , y 0 , z 0 , t0 ) which are obtained from the first
due to the Lorentz transformations.
The direction of a radius vector OA0 from 0 to a point A0 on the structure
(2) and the directions of tangents at A0 are normal to one another.
Let us trace a certain location of matter in its path at all times t. The
entirety of the spacetime points x, y, z, t of the locations corresponding to
di↵erent times t, I will call a spacetime line.
The task to determine the motion of matter should be understood in
this way: the direction of the spacetime line passing through each spacetime
point should be found.
A spacetime P (x, y, z, t) can be transformed to rest by introducing through
a Lorentz transformation a reference system x0 , y 0 , z 0 , t0 such that its t0 -axis
should be chosen from the allowed directions OA0 in such a way that it is
parallel to the spacetime line passing through P . The space t0 =const, which
goes through P, is normal to the spacetime line through P . The increase dt
of the time t at P corresponds to the increase
p p dx4
d⌧ = dt2 dx2 dy 2 dz 2 = dt 1 w2 = (3)
w4
of the parameter t0 introduced here11 . The value of the integral

Z Z q
d⌧ = (dx21 + dx22 + dx23 + dx24 )
11
The indices and the symbols w, w have the meaning adopted above (see §3 and §4).
99
taken along the spacetime line from any fixed initial point P 0 to a point P
is called proper time (Eigenzeit), corresponding to the location of matter at
the spacetime point P . (This is a generalization of the concept of local time
(Ortszeit) used by Lorentz in the case of uniform motion.)
Consider a spatially extended body R0 at a given time t0 , then the area
consisting of the spacetime lines passing through the spacetime point R0 , t0
is called a spacetime thread (Raum-Zeitfaden).
Let us have an analytical expression ⇥(x, y, z, t) such that ⇥(x, y, z, t) =
0 is taken from each spacetime line of the thread in one point, with
✓ ◆2 ✓ ◆2 ✓ ◆2 ✓ ◆2
@⇥ @⇥ @⇥ @⇥ @⇥
+ > 0, > 0,
@x @y @z @t @t
then the entirety Q of the points resulting from the intersection of the space
of extended body R0 and its spacetime thread will be called a cross section
(Querschnitt) of the thread. At any point P (x, y, z, t) of such across sec-
tion, we can introduce through a Lorentz transformation a reference system
x0 , y 0 , z 0 , t0 in which
@⇥ @⇥ @⇥ @⇥
= 0, = 0, = 0, > 0.
@x0 @y 0 @z 0 @t0
The direction of such uniquely determined t0 -axis will be called the up-
per
RRR normal of the cross section Q at the point P and the value dJ =
dx0 dy 0 dz 0 for a neighborhood of P on the cross section will be called
the volume element (Inhaltselement) of the cross section. In this sense,
R0 , t0 itself should be regarded as the cross section at t = t0 which is normal
to the t0 -axis of the thread and the (ordinary) volume of the body R0 should
be called the volume (Inhalt) of this cross section.
By letting the space R0 to converge toward a point we come to the
concept of an infinitely thin spacetime thread. In such a thread a spacetime
line will be regarded as the main line (Hauptlinie) and by proper time of the
thread we will understand the proper time along this main line; by a normal
cross section of the thread we will understand the space, which is normal to
the main line at a given point on it.
We now formulate the principle of conservation of mass.
To every space R at a time t belongs a positive quantity, the mass in R at
time t. When R converges to a point x, y, z, t the quotient of this mass and
the volume of R approaches a limiting value µ(x, y, z, t), the mass density
at the spacetime point x, y, z, t.
The principle of conservation of mass states that: For an infinitely thin

spacetime thread the product µ dJ of the mass density at a point x, y, z, t
on the thread (i.e., on the main line of the thread) and the volume dJ of
the cross section, which is normal to the t-axis, is constant along the entire
thread.
Now, the volume dJn of the normal cross section of the thread at point
x, y, z, t is
1 dt
dJn = p dJ = iw4 dJ = dJ (4)
1 w2 d⌧
and it should be expected for
µ p d⌧
⌫= =µ 1 w2 = µ , (5)
iw4 dt
to be defined as the rest mass density at the point x, y, z, t. Then the

principle of conservation of mass can be also formulated as follows:
For an infinitely thin spacetime thread the product of the rest mass den-
sity and the volume of the normal cross section at a point on the thread is
always constant along the entire thread.
In an arbitrary spacetime thread to a cross section Q0 a second cross
section Q1 can be mounted which with Q0 has points on the boundary of
the thread, but with only this in common, and for the spacetime lines within
the thread the values of t increase from Q0 to Q1 . The finite area located
between Q0 and Q1 will be called spacetime sickle (Raum-Zeit-Sichel ), where
Q0 is the lower boundary and Q1 – the upper boundary of the sickle.
Let us suppose that the thread is divided into many very thin spacetime
threads, so every entry of a thin thread in the lower boundary of the sickle
corresponds to an exit from the top, and each time for both cases the product
⌫ dJn , whose terms are defined by (4) and (5), has the same value.
R
Therefore the di↵erence between the two integrals ⌫ dJn , the first
extending over the upper, the second over the lower boundary of the sickle,
vanishes. This di↵erence can be found by using a known theorem of integral
calculus and is equal to the integral
ZZZZ
lor ⌫w dx dy dz dt,
which extends over the whole area of sickle, where (see (67) in §12)
101
@⌫w1 @⌫w2 @⌫w3 @⌫w4

lor ⌫w = + + + .
@x1 @x2 @x3 @x4
When the sickle shrunk to a spacetime point x, y, z, t we arrive at the di↵er-
ential equation
lor ⌫w = 0, (6)
i.e. at the continuity condition
@µwx @µwy @µwz @µ

+ + + = 0.
@x @y @z @t
We form also the integral which extends over the whole area of a space-
time sickle
ZZZZ
N= ⌫ dx dy dz dt. (7)
We cut the sickle into thin the spacetime threads and cut each of these
threads into small elements d⌧ of its proper time, which are however still
large compared to the linear dimensions of the normal cross section of the
thread. If we take the mass of such a thread to be ⌫ dJn = dm and ⌧ 0 and
⌧ 1 are the proper time of the thread at the lower and the upper boundary
of the sickle, respectively, the integral (7) can be interpreted as
ZZ Z
⌫ dJn d⌧ = (⌧ 1 ⌧ 0 ) dm.
over all the threads in the sickle.

Now I take the spacetime lines within a spacetime sickle as material
curves consisting of material points and think of them as being subjected to
a continuous change of position within the sickle in the following way. All the
curves are somehow fixed at the end points of the lower and upper boundary of
the sickle and the individual material points on them always move normal
to the curves. The whole process should be analytically represented by a
parameter ✓ and the value ✓ = 0 corresponds to the curves that represent
the actual spacetime lines within the sickle. Such a process is called a virtual
displacement in the sickle.
The point x, y, z, t in the sickle for ✓ = 0 changes its position to x +
x, y + y, z + z, t + t for the value ✓ of parameter; the latter quantities
are then functions of x, y, z, t, ✓. If we consider again an infinitely thin
spacetime thread, the volume dJn of its normal cross section at the initial
point x, y, z, t becomes the volume dJn + dJn of the normal cross section
at the corresponding point of the varied thread. When we take into account
the principle of conservation of mass and that the rest mass density at the
varied position is ⌫ + ⌫, where ⌫ is the real rest mass density at x, y, z, t,
we can write
(⌫ + ⌫)(dJn + dJn ) = ⌫ dJn = dm. (8)
On the basis of this result, the integral (7) taken over the area of the sickle,
varies due the virtual displacement as a some function N + N of ✓, and we
will call this function N + N of ✓ the mass e↵ect of the virtual displacement.
If we go by the notation with indices, we will have:
X @ xh ✓ ◆
@ xh k = 1, 2, 3, 4
d(xh + xh ) = dxh + dxk + d✓ . (9)
@xk @✓ h = 1, 2, 3, 4
k
Then on the basis of the comments already made above is immediately seen
that the value of N + N for the parameter ✓ will be
ZZZZ
d(⌧ + ⌧ )
N+ N= ⌫ dx dy dz dt (10)
d⌧
which is taken over the sickle, and where the magnitude d(⌧ + ⌧ ) follows
from
p
(dx1 + d x1 )2 (dx2 + d x2 )2 (dx3 + d x3 )2 (dx4 + d x4 )2 ,
by taking account of (9), and
dx1 = w1 d⌧, dx2 = w2 d⌧, dx3 = w3 d⌧, dx4 = w4 d⌧, d✓ = 0,
we have
v !2
u ✓ ◆
d(⌧ + ⌧ ) u X X @ xh k = 1, 2, 3, 4
=t wh + wk . (11)
d⌧ @xk h = 1, 2, 3, 4
k k
103
Now we want to subject the value of the derivative

✓ ◆
d(N + N )
(12)
d✓ (✓=0)
to a transformation. Since each xh as a function of the arguments x, y, z, y, ✓

vanishes in general for ✓ = 0, it is also generally @@xxkh = 0 for ✓ = 0. If we
now denote
✓ ◆
@ xh
= ⇠h (h = 1, 2, 3, 4), (13)
@xk ✓=0
then taking account of (10) and (11) the expression (12) becomes
ZZZZ X ✓ ◆
@⇠h @⇠h @⇠h @⇠h
⌫ wh w1 + w2 + w3 + w4 dx dy dz dt.
@x1 @x2 @x3 @x4
h
For the system x1 , x2 , x3 , x4 on the boundary of the sickle x1 , x2 , x3 , x4

should vanish for every value of ✓ and therefore also ⇠1 , ⇠2 , ⇠3 , ⇠4 are all zero.
After that, the last integral transforms by integration by parts
ZZZZ X ✓ ◆
@⌫wh w1 @⌫wh w2 @⌫wh w3 @⌫wh w4
⇠h + + + dx dy dz dt.
@x1 @x2 @x3 @x4
h
The expression in the brackets

X @⌫wk X @wh
= wh +⌫ wk .
@xk @xk
k k
The first sum here vanishes according to the continuity condition (6), the
second can be represented as
✓ ◆
@wh dx1 @wh dx2 @wh dx3 @wh dx4 dwh d dxh
+ + + = = ,
@x1 d⌧ @x2 d⌧ @x3 d⌧ @x4 d⌧ d⌧ d⌧ d⌧
d
where d⌧ indicates di↵erentiation in the direction of the spacetime line at a
given point. Finally, for the derivative(12) we have the expression
ZZZZ ✓ ◆
dw1 dw2 dw3 dw4
⌫ ⇠1 + ⇠2 + ⇠3 + ⇠4 dx dy dz dt. (14)
d⌧ d⌧ d⌧ d⌧
For a virtual displacement in the sickle we have set the requirement that
the material points should move normal to the curves produced from them,
which means that for ✓ = 0 the quantities ⇠h must satisfy the condition
w1 ⇠1 + w2 ⇠2 + w3 ⇠3 + w4 ⇠4 = 0. (15)
Consider now Maxwell’s stresses in the electrodynamics of bodies at rest

and, on the other hand, at our results in §12 and §13 then there is some
adjustment of the Hamilton principle for continuously extended elastic media
to the relativity postulate.
Let at each spacetime point (as in §13) a spacetime matrix of the II-nd
kind be known
S11 , S12 , S13 , S14 Xx , Yx , Zx , iTx

S21 , S22 , S23 , S24 Xy , Yy , Zy , iTy
S= = , (16)
S31 , S32 , S33 , S34 Xz , Yz , Zz , iTz
S41 , S42 , S43 , S44 iXt , iYt , iZt , Tt
where Xx , Yx , · · · , Zz , Tx , · · · , Xt , · · · , Tt are real quantities.

For a virtual displacement in a spacetime sickle with the previously used
terms the value of the integral
0 1
ZZZZ X
@ @(x k + x )
k A
W+ W = Shk dx dy dz dt, (17)
@xh
h,k
which extends over the area of the sickle, may be called the stress e↵ect in
the virtual displacement.
The sum occurring here, given in more detail and written with real quan-
tities, is
Xx + Y y + Z z + Tt
@ x @ x @ z
+Xx + Xy + · · · + Zz
@x @y @z
105
@ x @ t @ t
Xt · · · + Tx + · · · + Tt .
@t @x @t
We will now introduce the following minimum principle for mechanics:

For any bounded spacetime sickle the sum of the mass e↵ect and the
stress e↵ect should always be an extremum for each virtual displacement in
the sickle for the actually occurring configurations of the spacetime lines in
the sickle.
The meaning of this statement is that, in the previously explained sym-
bols, for every virtual displacement it should be
✓ ◆
d( N + w)
= 0. (18)
d✓ ✓=0
By using the methods of the calculus of variations the four di↵eren-

tial equations below follow immediately from the minimum principle under
consideration after taking into account the condition (15) and using the
transformation (14)
dwh
⌫ = Kh + {wh (h = 1, 2, 3, 4), (19)
d⌧
where
@S1h @S2h @S3h @S4h

Kh = + + + (20)
@x1 @x2 @x3 @x4
are the components of the spacetime vector of the I-st kind K = lor S, and
{ is a factor whose determination has to be made on the basis of ww = 1.
By multiplying each equation of (19) with wh , and subsequent summation
over h = l, 2, 3, 4 we find { = Kw and obviously K + (Kw)w is a spacetime
vector of the I-st kind, which is normal to w. If we write the components of
this vector
X, Y, Z, iT,
we now come to the following laws for the motion of matter :

d dx
⌫ = X,
d⌧ d⌧
d dy
⌫ = Y,
d⌧ d⌧ (21)
d dz
⌫ = Z,
d⌧ d⌧
d dt
⌫ = T.
d⌧ d⌧
We have
✓ ◆2 ✓ ◆2 ✓ ◆2 ✓ ◆2
dx dy dz dt
+ + = 1,
d⌧ d⌧ d⌧ d⌧
and
dx dy dz dt
X +Y +Z =T ,
d⌧ d⌧ d⌧ d⌧
and on the basis of these circumstances the fourth equation in (21) can be
viewed as a consequence of the first three.
From (21) we derive further the laws for the motion of a material point,
that is to say for the course of an infinitely thin spacetime thread.
Let x, y, z, t denote a point on the accepted main line in the thread. We
form the equations (21) for the points of the normal cross section of the
thread through x, y, z, t and integrate them, which is to multiply them by
the volume element of the cross section, over the whole space of the normal
cross section.
If the integrals on the right side of (21) are Rx , Ry , Rz , Rt and m is
the constant mass of the thread we can write
d dx
m = Rx ,
d⌧ d⌧
d dy
m = Ry ,
d⌧ d⌧ (22)
d dz
m = Rz ,
d⌧ d⌧
d dt
m = Rt .
d⌧ d⌧
107
Here again R with the components Rx , Ry , Rz , iRt is a spacetime vector

of the I-st kind, which is normal to the spacetime vector of the I-st kind w,
the velocity of the material point, with the components
dx dy dz dt
, , , i .
d⌧ d⌧ d⌧ d⌧
We will call this vector R the moving force of the material point.
If, however, instead of integrating the equations above over the normal
cross section of the thread, we integrate them over the cross section that
is normal to the t axis and passes through x, y, z, t, the results is (see (4))
that the equations (22), multiplied with d⌧dt , in particular the last equation
which becomes
✓ ◆
d dt d⌧ d⌧ d⌧
m = wx Rx + wy Ry + wz Rz .
dt d⌧ dt dt dt
Now the right side can be interpreted as the work done per unit time (Ar-
beitsleistung) on the material point. This equation itself is then the law of
energy (Energiesatz ) for the motion of the material point, and the expression
✓ ◆ ✓ ◆ ✓ ◆
dt 1 1 2 3 4
m 1 =m p 1 =m |w| + |w| + · · ·
d⌧ 1 w2 2 8
will be seen as the kinetic energy of the material point.

Since always dt > d⌧ the quotient dtd⌧d⌧ could be called the advance
(Vorgehen) of time in comparison with the proper time of the material point,
and then we can say: The kinetic energy of a material point is the product
of its mass and the advance of time in comparison with its proper time.
The quadruple of equations (22) again shows the required by the relativ-
ity postulate full symmetry in the x, y, z, it, while to the fourth equation, as
we have already encountered an analogous case in electrodynamics, a higher
physical evidence (Evidenz) should be attributed. Due to the requirement of
symmetry along the lines of the fourth equation we can now set up the triple
of the first three equations and in view of this fact, the following statement
is justified: When the relativity postulate is put at the head of mechanics the
complete laws of motion follow solely from the law of energy.
I would not omit to make even more plausible that it is not expected from
the phenomena of gravitation to produce an opposition to the acceptance of
the relativity postulate12 .
12
In a completely di↵erent way, from the one I follow here, H. Poincaré tried to adapt
If B ⇤ (x⇤ , y ⇤ , z ⇤ , t⇤ ) is a fixed spacetime point, the area of all those space-

time points B(x, y, z, t) for which
(x x⇤ )2 + (y y ⇤ )2 + (z z ⇤ )2 = (t t⇤ ) 2 , t t⇤ 0 (23)
is called the lightray structure (Strahlgebilde) of the spacetime point B ⇤ .

This structure intersects an arbitrary spacetime line always only in a
single spacetime point B, which, on the one hand, follows from the convexity
of the structure, and, on the other hand, from the circumstance that all
directions of a spacetime line are only directions from B ⇤ toward the concave
side of the structure. We will call B ⇤ the point light source (Lichtpunkt) of
B.
If in the condition (23) the point B(x, y, z, t) is regarded as fixed, whereas
point B ⇤ (x⇤ , y ⇤ , z ⇤ , z ⇤ ) is thought as variable, then the same relation defines
the structure consisting of all spacetime points B ⇤ , which are point light
sources of B, and analogously it turns out that on any spacetime line there
is always only a single point B ⇤ , which is the point light source of B.
There may now be a material point F of mass m, which in the presence
of another material point of mass m⇤ experiences a moving force according
to the following law. Imagine the spacetime threads of F and F ⇤ with the
main lines in them. Let us take an infinitely small element BC on the main
line of F , further on the main line of F ⇤ , B ⇤ is the point light source of
B and C ⇤ is the point light source of C; then OA0 is a radius vector of a
point on the hyperboloidal fundamental structure (2) parallel to B ⇤ C ⇤ , and
finally D⇤ is the intersection point of the line B ⇤ C ⇤ with the space which is
normal to it and passes through B. The moving force of the point mass F at
the spacetime point B may now be a spacetime vector of the I-st kind, which
is normal to BC, and which is additively composed of the vector
✓ ◆3
⇤ OA0
mm BD⇤ (24)
B ⇤ D⇤
in the direction of BD⇤ and of a suitable vector in the direction of B ⇤ C ⇤ .

0
Here BOA⇤ D⇤ is the ratio of the two respective parallel vectors.
It is clear that this construction has a covariant character with respect

to the Lorentz group.
Newton’s law of attraction to the relativity postulate (Rendiconti del Circolo Matematico
di Palermo, T. XXI (1906), p. 129).
109
We now ask how the spacetime thread of F behaves, if the material point
F⇤ is in a uniform translational motion, which means that the main line of
the thread of F ⇤ is a straight line. We move the spacetime origin 0 to it,
and by a Lorentz transformation we can regard this line as the t axis. Now
x, y, z, t represent the point B and ⌧ ⇤ is the proper time of the point B ⇤ ,
calculated from O. Our construction here leads to the equations
d2 x m⇤ x d2 y m⇤ y d2 z m⇤ z
= , = , = (25)
d⌧ 2 (t ⌧ ⇤ )3 d⌧ 2 (t ⌧ ⇤ )3 d⌧ 2 (t ⌧ ⇤ )3
and
d2 t m⇤ d(t ⌧ ⇤ )
= , (26)
d⌧ 2 (t ⌧ ⇤ )2 dt
where
x2 + y 2 + z 2 = (t ⌧ ⇤ )2 (27)
and
✓ ◆2 ✓ ◆2 ✓ ◆2 ✓ ◆2
dx dy dz dt
+ + = 1. (28)
d⌧ d⌧ d⌧ d⌧
The three equations (25) in view of (27) are exactly as the equations for
the motion of a material point according to Newton’s laws for a fixed center
of attraction, except that instead of the time t the proper time ⌧ of the
material point occurs. The fourth equation (26) then gives the relation
between proper time and time for the material point.
Now be the orbit of the spatial point x, y, z for di↵erent ⌧ is an ellipse
with a semi-major axis a, eccentricity e, and let E be the eccentric anomaly,
then T is the increase in proper time for a full revolution on the orbit, finally
nT = 2⇡ with a suitable initial value of ⌧ we have the Kepler equation
n⌧ = E e sin E (29)
Changing the unit of time and again denoting the velocity of light by c, it
follows from (28):
✓ ◆2
dt m⇤ 1 + e cos E
1= (30)
d⌧ ac2 1 e cos E
Neglecting terms proportional to c 4 compared to 1, it follows that

✓ ◆
1 m⇤ 1 + e cos E
n dt = n d⌧ 1 + ,
2 ac2 1 e cos E
which together with the use of (29) is
✓ ◆
1 m⇤ m⇤
nt + const = 1 + n⌧ + sin E. (31)
2 ac2 ac2
m ⇤
The factor ac 2 here is the square of the ratio of a some average orbital
velocity of F to the velocity of light. If m⇤ is the mass of the Sun and a is
half the major axis of the Earth’s, then this factor is 10 8 .
The law of attraction for masses according to the formulation just dis-
cussed and related to the relativity postulate would also imply a propagation
of gravitation with the velocity of light. Given the smallness of the periodic
term in (31), a decision should not be inferred from astronomical observa-
tions against such a law and the proposed modified mechanics and in favor
of the Newtonian law of attraction of the Newtonian mechanics.
Space and Time
Gentlemen! The views of space and time which I want to present to you arose
from the domain of experimental physics, and therein lies their strength.
Their tendency is radical. From now onwards space by itself and time by
itself will recede completely to become mere shadows and only a type of
union of the two will still stand independently on its own.
I.
I want to show first how to move from the currently adopted mechanics
through purely mathematical reasoning to modified ideas about space and
time. The equations of Newtonian mechanics show a twofold invariance.
First, their form is preserved when subjecting the specified spatial coordi-
nate system to any change of position; second, when it changes its state of
motion, namely when any uniform translation is impressed upon it; also,
the zero point of time plays no role. When one feels ready for the axioms of
mechanics, one is accustomed to regard the axioms of geometry as settled
and probably for this reason those two invariances are rarely mentioned in
the same breath. Each of them represents a certain group of transforma-
tions for the di↵erential equations of mechanics. The existence of the first
group can be seen as reflecting a fundamental characteristic of space. One
always tends to treat the second group with disdain in order to unburden
one’s mind that one can never determine from physical phenomena whether
space, which is assumed to be at rest, may not after all be in uniform trans-
lation. Thus these two groups lead completely separate lives side by side.
Their entirely heterogeneous character may have discouraged any intention
to compose them. But it is the composed complete group as a whole that
gives us to think.
We will attempt to visualize the situation graphically. Let x, y, z be
orthogonal coordinates for space and let t denote time. The objects of our
perception are always connected to places and times. No one has noticed a
111
112 SPACE AND TIME
place other than at a time and a time other than at a place. However I still
respect the dogma that space and time each have an independent meaning.
I will call a point in space at a given time, i.e. a system of values x, y, z, t
a worldpoint. The manifold of all possible systems of values x, y, z, t will be
called the world. With a hardy piece of chalk I can draw four world axes
on the blackboard. Even one drawn axis consists of nothing but vibrating
molecules and also makes the journey with the Earth in the Universe, which
already requires sufficient abstraction; the somewhat greater abstraction
associated with the number 4 does not hurt the mathematician. To never
let a yawning emptiness, let us imagine that everywhere and at any time
something perceivable exists. In order not to say matter or electricity I
will use the word substance for that thing. We focus our attention on the
substantial point existing at the worldpoint x, y, z, t and imagine that we
can recognize this substantial point at any other time. A time element dt
may correspond to the changes dx, dy, dz of the spatial coordinates of this
substantial point. We then get an image, so to say, of the eternal course of
life of the substantial point, a curve in the world, a worldline, whose points
can be clearly related to the parameter t from 1 to +1. The whole world
presents itself as resolved into such worldlines, and I want to say in advance,
that in my understanding the laws of physics can find their most complete
expression as interrelations between these worldlines.
Through the concepts of space and time the x, y, z-manifold t = 0 and
its two sides t > 0 and t < 0 fall apart. If for simplicity we hold the chosen
origin of space and time fixed, then the first mentioned group of mechanics
means that we can subject the x, y, z-axes at t = 0 to an arbitrary rotation
about the origin corresponding to the homogeneous linear transformations
of the expression
x2 + y 2 + z 2 .
The second group, however, indicates that, also without altering the expres-
sions of the laws of mechanics, we may replace
x, y, z, t by x ↵t, y t, z t, t,
where ↵, , are any constants. The time axis can then be given a com-
pletely arbitrary direction in the upper half of the world t > 0. What has
now the requirement of orthogonality in space to do with this complete
freedom of choice of the direction of the time axis upwards?
To establish the connection we take a positive parameter c and look at
the structure
113
c 2 t2 x2 y2 z 2 = 1.
t
t'
Q' Q'
B'
A A'
B
x' P' P'
1
c D
C'
1
O D' C x P P Q Q
Fig. 1
It consists of two sheets separated by t = 0 by analogy with a two-sheeted

hyperboloid. We consider the sheet in the region t > 0 and we will now
take those homogeneous linear transformations of x, y, z, t in four new vari-
ables x0 , y 0 , z 0 , t0 so that the expression of this sheet in the new variables has
the same form. Obviously, the rotations of space about the origin belong
to these transformations. A full understanding of the rest of those trans-
formations can be obtained by considering such among them for which y
and z remain unchanged. We draw (Fig. 1) the intersection of that sheet
with the plane of the x- and the t-axis, i.e. the upper branch of the hyper-
bola c2 t2 x2 = 1 with its asymptotes. Further we draw from the origin
O an arbitrary radius vector OA0 of this branch of the hyperbola; then we
add the tangent to the hyperbola at A0 to intersects the right asymptote at
B 0 ; from OA0 B 0 we complete the parallelogram OA0 B 0 C 0 ; finally, as we will
need it later, we extend B 0 C 0 so that it intersects the x-axis at D0 . If we
now regard OC 0 and OA0 as axes for new coordinates x0 , t0 , with the scale
units OC 0 = 1, OA0 = 1/c, then that branch of the hyperbola again obtains
the expression ct02 x02 = 1, t0 > 0, and the transition from x, y, z, t to
x0 , y 0 , z 0 , t0 is one of the transformations in question. These transformations
plus the arbitrary displacements of the origin of space and time constitute a
group of transformations which still depends on the parameter c and which
I will call Gc .
If we now increase c to infinity, so 1/c converges to zero, it is clear from
the figure that the branch of the hyperbola leans more and more towards the
x-axis, that the angle between the asymptotes becomes greater, and in the
limit that special transformation converts to one where the t0 -axis may be
in any upward direction and x0 approaches x ever more closely. By taking
114 SPACE AND TIME
this into account it becomes clear that the group Gc in the limit c = 1,
that is the group G1 , is exactly the complete group which is associated
with the Newtonian mechanics. In this situation, and since Gc is mathe-
matically more understandable than G1 , there could have probably been a
mathematician with a free imagination who could have come up with the
idea that at the end natural phenomena do not actually possess an invari-
ance with the group G1 , but rather with a group Gc with a certain finite
c, which is extremely great only in the ordinary units of measurement. Such
an insight would have been an extraordinary triumph for pure mathematics.
Now mathematics expressed only staircase wit here, but it has the satisfac-
tion that, due to its happy antecedents with their senses sharpened by their
free and penetrating imagination, it can grasp the profound consequences of
such remodelling of our view of nature.
I want to make it quite clear what the value of c will be with which
we will be finally dealing. c is the velocity of the propagation of light in
empty space. To speak neither of space nor of emptiness, we can identify
this magnitude with the ratio of the electromagnetic to the electrostatic unit
of the quantity of electricity.
The existence of the invariance of the laws of nature with respect to the
group Gc would now be stated as follows:
From the entirety of natural phenomena, through successively enhanced
approximations, it is possible to deduce more precisely a reference system
x, y, z, t, space and time, by means of which these phenomena can be then
represented according to certain laws. But this reference system is by no
means unambiguously determined by the phenomena. One can still change
the reference system according to the transformations of the above group
Gc arbitrarily without changing the expression of the laws of nature in the
process.
For example, according to the figure depicted above one can call t0 time,
but then must necessarily, in connection with this, define space by the mani-
fold of three parameters x0 , y, z in which the laws of physics would then have
exactly the same expressions by means of x0 , y, z, t0 as by means of x, y, z, t.
Hereafter we would then have in the world no more the space, but an infi-
nite number of spaces analogously as there is an infinite number of planes
in three-dimensional space. Three-dimensional geometry becomes a chapter
in four-dimensional physics. You see why I said at the beginning that space
and time will recede completely to become mere shadows and only a world
in itself will exist.
II.
115
Now the question is, what circumstances force us to the changed view
of space and time, does it actually never contradict the phenomena, and
finally, does it provide advantages for the description of the phenomena?
Before we discuss these questions, an important remark is necessary.
Having individualized space and time in some way, a straight worldline par-
allel to the t-axis corresponds to a stationary substantial point, a straight
line inclined to the t-axis corresponds to a uniformly moving substantial
point, a somewhat curved worldline corresponds to a non-uniformly moving
substantial point. If at any worldpoint x, y, z, t there is a worldline passing
through it and we find it parallel to any radius vector OA0 of the previously
mentioned hyperboloidal sheet, we may introduce OA0 as a new time axis,
and with the thus given new concepts of space and time, the substance at
the worldpoint in question appears to be at rest. We now want to introduce
this fundamental axiom:
With appropriate setting of space and time the substance existing at any
worldpoint can always be regarded as being at rest.
This axiom means that at every worldpoint1 the expression
c2 dt2 dx2 dy 2 dz 2
is always positive, which is equivalent to saying that any velocity v is al-

ways smaller than c. Then c would be an upper limit for all substantial
velocities and that is precisely the deeper meaning of the quantity c. In this
understanding the axiom is at first glance slightly displeasing. It should be
noted, however, that a modified mechanics, in which the square root of that
second order di↵erential expression enters, is now gaining ground, so that
cases with superluminal velocity will play only such a role as that of figures
with imaginary coordinates in geometry.
The impulse and true motivation for accepting the group Gc came from
noticing that the di↵erential equation for the propagation of light waves in
the empty space possesses that group Gc 2 . On the other hand, the concept of
a rigid body has meaning only in a mechanics with the group G1 . If one has
optics with Gc , and if, on the other hand, there were rigid bodies, it is easy
to see that one t-direction would be distinguished by the two hyperboloidal
sheets corresponding to Gc and G1 , and would have the further consequence
that one would be able, by using appropriate rigid optical instruments in
1
Editor’s note: Minkowski means at every worldpoint along the worldline of the sub-
stance.
2
An important application of this fact can already be found in W. Voigt, Göttinger
Nachrichten, 1887, S. 41.
116 SPACE AND TIME
the laboratory, to detect a change of phenomena at various orientations

with respect to the direction of the Earth’s motion. All e↵orts directed
towards this goal, especially a famous interference experiment of Michelson
had, however, a negative result. To obtain an explanation, H. A. Lorentz
made a hypothesis, whose success lies precisely in the invariance of optics
with respect to the group Gc . According to Lorentz every body moving at a
velocity v must experience a reduction in the direction of its motion namely
in the ratio
r
v2
1: 1 .
c2
This hypothesis sounds extremely fantastical. Because the contraction is not
to be thought of as a consequence of resistances in the ether, but merely as
a gift from above, as an accompanying circumstance of the fact of motion.
I now want to show on our figure that the Lorentzian hypothesis is com-
pletely equivalent to the new concept of space and time, which makes it
much easier to understand. If for simplicity we ignore y and z and think
of a world of one spatial dimension, then two strips, one upright parallel
to the t-axis and the other inclined to the t-axis (see Fig. 1), are images
for the progression in time of a body at rest and a body moving uniformly,
where each preserves a constant spatial dimension. OA0 is parallel to the
second strip, so we can introduce t0 as time and x0 as a space coordinate
and then it appears that the second body is at rest, whereas the first –
in uniform motion. We now assume that the first body has length l when
considered at rest, that is, the cross section P P of the first strip and the
x-axis is equal to l · OC, where OC is the measuring unit on the x-axis,
and, on the other hand, that the second body has the same length l when
regarded at rest; then the latter means that the cross-section of the second
strip measured parallel to the x0 -axis is Q0 Q0 = l · OC 0 . We have now in
these two bodies images of two equal Lorentz electrons, one stationary and
one uniformly moving. But if we go back to the original coordinates x, t, we
should take as the dimension of the second electron the cross section QQ of
its associated strip parallel to the x-axis. Now as Q0 Q0 = l ·OC 0 , it is obvious
that QQ = l · OD0 .qIf dx/dt for the second strip is = v, anqeasy calculation
2 2
gives OD0 = OC · 1 vc2 , therefore also P P : QQ = 1 : 1 vc2 . This is
the meaning of the Lorentzian hypothesis of the contraction of electrons in
motion. Regarding, on the other hand, the second electron as being at rest,
that is, adopting the reference system x0 , t0 , the length of the first electron
will be the cross section P 0 P 0 of its strip parallel to OC 0 , and we would find
117
the first electron shortened with respect to the second in exactly the same
proportion; from the figure we also see that
P 0 P 0 : Q0 Q0 = OD : OC 0 = OD0 : OC = QQ : P P.
Lorentz called t0 , which is a combination of x and t, local time of the
uniformly moving electron, and associated a physical construction with this
concept for a better understanding of the contraction hypothesis. However,
it is to the credit of A. Einstein3 who first realized clearly that the time of
one of the electrons is as good as that of the other, i.e. that t and t0 should
be treated equally. With this, time was deposed from its status as a concept
unambiguously determined by the phenomena. The concept of space was
shaken neither by Einstein nor by Lorentz, maybe because in the above-
mentioned special transformation, where the plane of x0 , t0 coincides with
the plane x, t, an interpretation is possible as if the x-axis of space preserved
its position. To step over the concept of space in such a way is an instance
of what can be achieved only due to the audacity of mathematical culture.
After this further step, which is indispensable for the true understanding of
the group Gc , I think the word relativity postulate used for the requirement
of invariance under the group Gc is very feeble. Since the meaning of the
postulate is that through the phenomena only the four-dimensional world in
space and time is given, but the projection in space and in time can still be
made with certain freedom, I want to give this affirmation rather the name
the postulate of the absolute world (or shortly the world postulate).
III.
Through the world postulate an identical treatment of the four identi-

fying quantities x, y, z, t becomes possible. I want to explain now how, as
a result of this, we gain more understanding of the forms under which the
laws of physics present themselves. Especially the concept of acceleration
acquires a sharply prominent character.
I will use a geometric way of expression, which presents itself immediately
when one implicitly ignores z in the triple x, y, z. An arbitrary worldpoint
O can be taken as the origin of space-time. The cone
c 2 t2 x2 y2 z2 = 0
with O as the apex (Fig. 2) consists of two parts, one with values t < 0, the
other one with values t > 0.
3
A. Einstein, Annalen der Physik 17 (1905), S. 891; Jahrbuch der Radioaktivität und
Elektronik 4 (1907), S. 411.
118 SPACE AND TIME
after O
r
cto
futur ve
int e lig
htco ke
ern ne el i
al tim or
spacelike vect
O
ola t c on
e
p erb ligh
hy past
before O
Fig. 2
The first, the past lightcone of O, consists, we can say, of all worldpoints
which “send light to O”, the second, the future lightcone of O, consists of all
worldpoints which “receive light from O”4 . The area bounded solely by the
past lightcone may be called before O, whereas the area bounded solely by
the future lightcone – after O. Situated after O is the already considered
hyperboloidal sheet
F = c 2 t2 x2 y2 z 2 = 1, t > 0
The area between the cones is filled with the one-sheeted hyperboloidal struc-
tures
F = x2 + y 2 + z 2 c 2 t2 = k 2
for all constant positive values of k 2 . Essential for us are the hyperbolas with
O as the center, located on the latter structures. The individual branches
of these hyperbolas may be briefly called internal hyperbolas with center O.
Such a hyperbola would be thought of as the worldline of a substantive point,
which represents its motion that increases asymptotically to the velocity of
light c for t = 1 and t = +1.
If we now call, by analogy with vectors in space, a directed line in the
manifold x, y, z, t a vector, we have to distinguish between the timelike vec-
tors with directions from O to the sheet +F = 1, t > 0, and the spacelike
vectors with directions from O to F = 1. The time axis can be parallel to
any vector of the first kind. Every worldpoint between the future lightcone
4
Editor’s and translator’s note: I decided to translate the words Vorkegel and Nachkegel
as past lightcone and future lightcone, respectively, for two reasons. First, this translation
reflects the essence of Minkowski’s idea – (i) all worldpoints on the past lightcone “send
light to O”, which means that they all can influence O and therefore lie in the past of O;
(ii) all worldpoints on the future lightcone “receive light from O”, which means that they
all can be influenced by O and therefore lie in the future of O. Second, the terms past
lightcone and future lightcone are now widely accepted in spacetime physics.
119
and the past lightcone of O can be regarded, by a choice of the reference sys-
tem, as simultaneous with O as well as earlier than O or later than O. Each
worldpoint within the past lightcone of O is necessarily always earlier than
O, each worldpoint within the future lightcone is necessarily always later
than O. The transition to the limit c = 1 would correspond to a complete
folding of the wedge-shaped section between the cones into the flat manifold
t = 0. In the figures this section is intentionally made with di↵erent widths.
We decompose any vector, such as that from O to x, y, z, t into four
components x, y, z, t. If the directions of two vectors are, respectively, that
of a radius vector OR from O to one of the surfaces ⌥F = 1, and that of
a tangent RS at the point R on the same surface, the vectors are called
normal to each other. Accordingly,
c2 tt1 xx1 yy1 zz1 = 0

is the condition for the vectors with components x, y, z, t and x1 , y1 , z1 , t1 to
be normal to each other.
The measuring units for the magnitudes of vectors in di↵erent directions
may be fixed by assigning to a spacelike vector from O to F = 1 always
the magnitude 1, and to a timelike vector from O to +F = 1, t > 0 always
the magnitude 1/c.
Let us now imagine a worldpoint P (x, y, z, t) through which the worldline
of a substantial point is passing, then the magnitude of the timelike vector
dx, dy, dz, dt along the line will be
1p 2 2
d⌧ = c dt dx2 dy 2 dz 2 .
c
R
The integral d⌧ = ⌧ of this magnitude, taken along the worldline from any
fixed starting point P0 to the variable end point P , we call the proper time
of the substantial point at P . On the worldline we consider x, y, z, t, i.e. the
components of the vector OP , as functions of the proper time ⌧ ; denote their
first derivatives with respect to ⌧ by ẋ, ẏ, ż, ṫ; their second derivatives with
respect to ⌧ by ẍ, ÿ, z̈, ẗ, and call the corresponding vectors, the derivative of
the vector OP with respect to ⌧ the velocity vector at P and the derivative
of the velocity vector with respect to ⌧ the acceleration vector at P . As
c2 ṫ2 ẋ2 ẏ 2 ż 2 = c2
it follows that
c2 ṫẗ ẋẍ ẏ ÿ ż z̈ = 0,
120 SPACE AND TIME
i.e. the velocity vector is the timelike vector of magnitude 1 in the direction
of the worldline at P , and the acceleration vector at P is normal to the
velocity vector at P , so it is certainly a spacelike vector.
hyperbola
ature
ne
curv
worldli
Fig. 3
Now there is, as is easily seen, a specific branch of the hyperbola, which
has three infinitely adjacent points in common with the worldline at P , and
whose asymptotes are generators of a past lightcone and a future lightcone
(see Fig. 3). This branch of the hyperbola will be called the curvature
hyperbola at P . If M is the center of this hyperbola, we have here an internal
hyperbola with center M . Let ⇢ be the magnitude of the vector M P , so we
recognize the acceleration vector at P as the vector in the direction M P of
magnitude c2 /⇢.
If ẍ, ÿ, z̈, ẗ are all zero, the curvature hyperbola reduces to the straight
line touching the worldline at P , and we should set ⇢ = 1.
IV.
To demonstrate that the adoption of the group Gc for the laws of physics
never leads to a contradiction, it is inevitable to undertake a revision of
all physics based on the assumption of this group. This revision has been
done successfully to some extent for questions of thermodynamics and heat
121
radiation5 , for the electromagnetic processes, and finally, with the retention
of the concept of mass, for mechanics.6
For the latter domain, the question that should be raised above all is:
When a force with the spatial components X, Y, Z acts at a worldpoint
P (x, y, z, t), where the velocity vector is ẋ, ẏ, ż, ṫ, as what force this force
should be interpreted for any change of the reference system? Now there
exist some proven approaches to the ponderomotive force in the electromag-
netic field in cases where the group Gc is undoubtedly permissible. These
approaches lead to the simple rule: When the reference system is changed,
the given force transforms into a force in the new space coordinates in such
a way that the corresponding vector with the components
ṫX, ṫY, ṫZ, ṫT

remains unchanged, and where
1 ẋ ẏ ż
T = 2
( X + Y + Z)
c ṫ ṫ ṫ
is the work done by the force at the worldpoint divided by c2 . This vector is
always normal to the velocity vector at P . Such a force vector, representing
a force at P , will be called a motive force vector at P .
Now let the worldline passing through P represent a substantial point
with constant mechanical mass m. The multiplied by m velocity vector
at P will be called the momentum vector at P , and the multiplied by m
acceleration vector at P will be called the force vector of the motion at P .
According to these definitions, the law of motion for a point mass with a
given force vector is:7
The force vector of the motion is equal to the motive force vector.
This assertion summarizes four equations for the components for the
four axes, wherein the fourth can be regarded as a consequence of the first
three because both vectors are from the start normal to the velocity vector.
According to the above meaning of T , the fourth equation is undoubtedly
5
M. Planck, “Zur Dynamik bewegter Systeme,” Sitzungsberichte der k. preußischen
Akademie der Wissenschaften zu Berlin, 1907, S. 542 (auch Annalen der Physik, Bd. 26,
1908, S. 1).
6
H. Minkowski, “Die Grundgleichungen für die elektromagnetischen Vorgänge in be-
wegten Körpern”, Nachrichten der k. Gesellschaft der Wissenschaft zu Göttingen,
mathematisch-physikalische Klasse, 1908, S. 53 und Mathematische Annalen, Bd. 68,
1910, S. 527
7
H. Minkowski, loc. cit., p. 107. Cf. also M. Planck, Verhandlungen der Physikalischen
Gesellschaft, Bd. 4, 1906, S. 136.
122 SPACE AND TIME
the law of conservation of energy. The kinetic energy of the point mass is
defined as the component of the momentum vector along the t-axis multiplied
by c2 . The expression for this is
dt mc2
mc2 =q ,
d⌧ 2
1 vc2
which is, the expression 12 mv 2 of Newtonian mechanics after the subtraction

of the additive constant term mc2 and neglecting magnitudes of the order
1/c2 . The dependence of the energy on the reference system is manifested
very clearly here. But since the t-axis can be placed in the direction of
each timelike vector, then, on the other hand, the law of conservation of en-
ergy, formed for every possible reference system, already contains the whole
system of the equations of motion. In the discussed limiting case c = 1,
this fact will retain its importance for the axiomatic structure of Newtonian
mechanics and in this sense has been already noticed by J. R. Schütz8
From the beginning we can determine the ratio of the units of length and
time in such
p a way that the natural limit of velocity becomes c = 1. If we
introduce 1t = s instead of t, then the quadratic di↵erential expression
d⌧ 2 = dx2 dy 2 dz 2 ds2
becomes completely symmetric in x, y, z, s and this symmetry is carried over
to any law that does not contradict the world postulate. Thus the essence of
this postulate can be expressed mathematically very concisely in the mystical
formula:
p
3 · 105 km = 1 seconds.
V.
The advantages resulting from the world postulate may most strikingly
be proved by indicating the e↵ects from an arbitrarily moving point charge
according to the Maxwell-Lorentz theory. Let us imagine the worldline of
such a pointlike electron with charge e, and take on it the proper time ⌧
from any initial point. To determine the field induced by the electron at any
worldpoint P1 we construct the past lightcone corresponding to P1 (Fig. 4).
It intersects the infinite worldline of the electron obviously at a single point P
8
J. R. Schütz, “Das Prinzip der absoluten Erhaltung der Energie”, Nachrichten der k.
Gesellschaft der Wissenschaften zu Göttingen, mathematisch-physikalische Klasse, 1897,
S. 110.
123
because the tangents to every point on the worldline are all timelike vectors.
At P we draw the tangent to the worldline and through P1 construct the
normal P1 Q to this tangent. Let the magnitude of P1 Q be r. According
to the definition of a past lightcone the magnitude of P Q should be r/c.
Now the vector of magnitude e/r in the direction P Q represents through its
components along the x-, y-, z-axes, the vector potential multiplied by c,
and through the component along the t-axis, the scalar potential of the field
produced by e at the worldpoint P1 . This is the essence of the elementary
laws formulated by A. Liénard and E. Wiechert.9
e1 e t
y
P1 Q
r
N
P
x
Fig. 4
Then it emerges in the description itself of the field caused by the electron
that the division of the field into electric and magnetic forces is a relative one
with respect to the specified time axis; most clearly the two forces considered
together can be described in some, though not complete, analogy with the
wrench in mechanics. I now want to describe the ponderomotive action of an
arbitrarily moving point charge exerted on another arbitrarily moving point
charge. Let us imagine that the worldline of a second pointlike electron of
charge e1 goes through the worldpoint P1 . We define P, Q, r as before, then
9
A. Liénard, “Champ électrique et magnétique produit par une charge concentré en un
point et animée d’un mouvement quelconque”, L’Éclairage électrique, T. 16, 1898, pp. 5,
53, 106; E. Wiechert, “Elektrodynamische Elementargesetze”, Archives Néerlandaiaes des
Sciences exactes et naturelles (2), T. 5, 1900, S. 549.
124 SPACE AND TIME
construct (Fig. 4) the center M of the curvature hyperbola at P , and finally

the normal M N from M to an imagined straight line from P parallel to
QP1 . We now fix a reference system with its origin at P in the following
way: the t-axis in the direction of P Q, the x-axis in the direction of QP1 ,
the y-axis in the direction of M N , and lastly the direction of the z-axis is
determined as being normal to the t-, x-, y-axes. Let the acceleration vector
at P be ẍ, ÿ, z̈, ẗ, the velocity vector at P1 be x˙1 , y˙1 , z˙1 , t˙1 . Now the motive
force vector exerted by the first arbitrarily moving electron e on the second
arbitrarily moving electron e1 at P1 will be
ẋ1
ee1 (t˙1 )K
c
where for the components Kx , Ky , Kz , Kt of the vector K three relations exist:
1 ÿ
cKt Kx = , Ky = 2 , Kz = 0
r2 c r
and fourthly this vector K is normal to the velocity vector at P1 , and this
circumstance alone makes it dependent on the latter velocity vector.
If we compare this assertion with the previous formulations10 of the same
elementary law of the ponderomotive action of moving point charges on one
another, we are compelled to admit that the relations considered here reveal
their inner being in full simplicity only in four dimensions, whereas on a
three dimensional space, forced upon us from the beginning, they cast only
a very tangled projection.
In mechanics reformed in accordance with the world postulate, the dis-
turbing disharmony between Newtonian mechanics and the modern electro-
dynamics disappears by itself. In addition, I want to touch on the status
of the Newtonian law of attraction with respect to this postulate. I will
consider two point masses m, m1 , represented by their worldlines, and that
m exerts a motive force vector on m1 exactly as in the case of electrons,
except that instead of ee1 +mm1 should be used. We can now specifically
consider the case when the acceleration vector of m is constantly zero, then
we may choose t in such a way that m is regarded as at rest, and assume that
only m1 move under the motive force vector which originatesqfrom m. If we
2
now modify this specified vector by adding the factor ṫ 1 = 1 vc2 , which
up to magnitudes of the order 1/c2 is equal to 1, it can be seen11 that for the
10
K. Schwazschild, Nachrichten der k. Gesellschaft der Wissenschaften zu Göttinger,
mathematisch-physikalische Klasse, 1903, S. 132; H. A. Lorentz, Enzyklopädie der math-
ematischen Wissenschaften, V, Art. 14, S. 199.
11
H. Minkowski, loc. cit., p. 110.
125
positions x1 , y1 , z1 of m1 and their progression in time, we arrive exactly at

Kepler’s laws, except that instead of the times t1 the proper times ⌧1 of m1
should be used. On the basis of this simple remark we can then see that the
proposed law of attraction associated with the new mechanics is no less well
suited to explain the astronomical observations than the Newtonian law of
attraction associated with the Newtonian mechanics.
The fundamental equations for the electromagnetic processes in ponder-
able bodies are entirely in accordance with the world postulate. Actually,
as I will show elsewhere, there is no need to abandon the derivation of these
equations which is based on ideas of the electron theory as taught by Lorentz.
The validity without exception of the world postulate is, I would think,
the true core of an electromagnetic world view which, as Lorentz found it
and Einstein further unveiled it, lies downright and completely exposed be-
fore us as clear as daylight. With the development of the mathematical
consequences of this postulate, sufficient findings of its experimental valid-
ity will be arrived at so that even those to whom it seems unsympathetic
or painful to abandon the prevailing views become reconciled through the
thought of a pre-stabilized harmony between mathematics and physics.

Minkowski Papers PDF

Uploaded by

Copyright:

Available Formats

Minkowski Papers PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Minkowski Papers PDF

Uploaded by

Copyright:

Available Formats

Hermann Minkowski

Space and Time

Translated by Fritz Lewerto↵ and Vesselin Petkov

c 2012 by Minkowski Institute Press

ISBN 978-0-9879871-4-3 (softcover)

Minkowski Institute Press

Aachen, May 1910

From the Foreword to H. Minkowski, Zwei Abhandlungen über

• The Relativity Principle 2 – lecture given at the meeting of the Göttin-

• The Fundamental Equations for Electromagnetic Processes in Moving

Montreal Vesselin Petkov

The Relativity Principle 39

The Fundamental Equations for Electromagnetic Processes in

Space and Time 111

The not-fully-appreciated Minkowski

• Unfortunately, it was Einstein himself (with Jakob Laub) who ex-

stein called Minkowski’s approach “superfluous learnedness”4 (über-

• Sommerfeld understood and accepted Einstein’s special relativity

struggle Michelson’s experiment played no role, or at least no deci-

• Despite his initial negative reaction towards Minkowski’s four-dimensio-

• Many physicists (including relativists) do not appear to have been fully

• There have been authors of books on general relativity, spacetime and

• What is also unfortunate is that some well-known physicists who write

Minkowski and Einstein

• The profound physical meaning of the relativity principle – that physi-

• Minkowski’s realization that the relativity principle implies many times

• Minkowski’s four-dimensional physics allowed him not only to explain

exactly like the relativity principle, that e↵ect is also a manifestation

• In his four-dimensional physics Minkowski found that pairs of ordinary

discovery that t and t0 should be treated equally. As a result, at that time

particularly pointed out the prominent character of the concept of accelera-

Minkowski’s enthusiasm for arriving at an electromagnetic picture of the

seen that many spaces imply an absolute four-dimensional world.

In addition, when Minkowski introduces the (geometrically moti-

i.e. Minkowski’s proper time. It seems that Minkowski was not

Minkowski was certainly aware of this expression without the integral

more than two years to explore those consequences – Minkowski appears ho

its worldline is curved and an observer in the particle’s accelerating frame

Minkowski’s important contribution to the theory lies in the

As seen from his estimation of Minkowski’s contribution Einstein did not

able in my view) development in quantum physics, which may reveal an

into the physical meaning of probabilistic phenomena in spacetime – an

Minkowski and Poincaré

In the absence of any clear indication why Minkowski left Poincaré

elaboration of this principle.42

My conjecture is that Minkowski, helped by his background read-

considering an Euclidean manifold of 4 dimensions called Universe or space-

Everything happens as if time were a fourth dimension of space,

Scientists sometimes deceive themselves into thinking that philo-

Second, it seems virtually certain that Minkowski independently arrived

Minkowski and gravitation

I was sitting in a chair in the patent office at Bern when all of a

Einstein had been so impressed by this insight that he called it the

is regarded as “a natural generalization of Newton’s first law,”54 that is,

sense inertial motion is absolute or frame-independent), and (ii) the acceler-

motion reflects the straightness of the worldline of an inertial particle and

• The new theory of gravitation demonstrates that gravitational physics

is in fact geometry of curved spacetime; no general relativity of any-

• How could physicists say that in the framework of general relativity