This action might not be possible to undo. Are you sure you want to continue?

# Revue d’histoire des math´ematiques

,

1 (1995), p. 3–38.

NATURPHILOSOPHIE AND ITS ROLE IN

RIEMANN’S MATHEMATICS

Umberto BOTTAZZINI and Rossana TAZZIOLI (*)

ABSTRACT. — This paper sets out to examine some of Riemann’s papers and notes

left by him, in the light of the “philosophical” standpoint expounded in his writings on

Naturphilosophie. There is some evidence that many of Riemann’s works, including

his Habilitationsvortrag of 1854 on the foundations of geometry, may have sprung

from his attempts to ﬁnd a uniﬁed explanation for natural phenomena, on the basis

of his model of the ether.

Keywords: ether theory, complex function theory, Riemannian diﬀerential geometry.

R

´

ESUM

´

E. — LE R

ˆ

OLE DE LA NATURPHILOSOPHIE DANS LES TRAVAUX

MATH

´

EMATIQUES DE RIEMANN. Dans cet article, nous proposons une lecture de

certains m´emoires et notes de Riemann `a la lumi`ere du point de vue philosophique

qu’il a d´evelopp´e dans ses ´ecrits sur la Naturphilosophie. Il apparaˆıt que l’origine de

nombreux travaux de Riemann, y compris l’Habilitationsvortrag de 1854 sur les fonde-

ments de la g´eom´etrie, peut ˆetre trouv´ee dans sa tentative d’explication unitaire des

ph´enom`enes naturels sur la base de son mod`ele de l’´ether.

INTRODUCTION

Riemann’s writings on Naturphilosophie

1

can be regarded as the result

of his attempt to ﬁnd a uniﬁed, mathematical explanation of various

physical phenomena such as gravitation, electricity, magnetism and light.

They also allow us to include some of his better known papers — such as

his Inauguraldissertation [1851], his Habilitationsvortrag [1854b] and other

papers on physical subjects as well — in a wide-ranging research program.

1

Heinrich Weber gathered these with others manuscripts of Riemann on philosoph-

ical subjects, such as psychology, metaphysics and gnosiology, and published them

in Riemann’s collected works under the title Fragmente philosophischen Inhalts [Rie-

mann 1876a].

(*) Texte re¸cu le 21 mars 1994, r´evis´e le 17 octobre 1994.

Umberto BOTTAZZINI, Rossana TAZZIOLI, Universit`a di Palermo, Dipartimento di

matematica, via Archiraﬁ 34, 90123 Palermo (Italia).

E.-mail: bottazzini@ipamat.math.unipa.it.

C SOCI

´

ET

´

E MATH

´

EMATIQUE DE FRANCE, 1995

4 U. BOTTAZZINI AND R. TAZZIOLI

As Klein once said, Riemann’s work was characterized by his continual

attempt to put “in mathematical form a uniﬁed formulation of the laws

which lie at the basis of all natural phenomena” [1894, p. 484]. Klein did

not hesitate to claim that “the origins of Riemann’s pure mathematical

developments” lay in this research which, in Riemann’s words, was at a

certain stage his own “main work”.

Searching for a mathematical description of the known physical phe-

nomena, Riemann thought of space as pervaded by substance (Stoﬀ

2

),

and in a section of his Fragmente he considered the state of a single parti-

cle of substance and analysed locally the space around it [Riemann 1853].

This passage from “local” to “global” constitutes the basic method

used by Riemann in some of his most important works in geometry as

well as in analysis and physics. In analytical terms this corresponds to

the analytical continuation of a complex function. This is “a well known

theorem” [Riemann 1857b, p. 88] which is at the basis of the “new method”

he set up in his thesis. This “method” [Riemann 1851, p. 37–39] could be

applied to Abelian functions, as he did in [1857b], and also “in its essential

lines” to “every function which satisﬁes a linear diﬀerential equation with

algebraic coeﬃcients” [1857a, p. 67]. Accordingly, in this paper he studied

the transcendental functions deﬁned by the hypergeometric diﬀerential

equation “almost without calculations” [Werke, p. 85] and “and in their

totality” on the complex sphere.

The same point of view inspired his Habilitationsvortrag where he

deﬁned metrics on manifolds by using the linear element ds. In particular,

Riemann stated that “questions about the immeasurably large are idle

questions for the explanation of Nature. But the situation is quite diﬀerent

with questions about the immeasurably small ” [Riemann 1854b/1979,

p. 151]. As Riemann explained in the introduction to the ﬁrst course

he gave in G¨ottingen as a Privatdozent, the laws for all space could

be deduced by integrating partial diﬀerential equations expressing some

“elementary” principles valid for inﬁnitely small portions of space.

3

2

Instead of this, in his later lectures on gravitation, electricity and magnetism Riemann

preferred to use the term ether.

3

“Wahre Elementargesetze k¨onnen nur im Unendlichkleinen, nur f¨ ur Raum —

und Zeitpunkte stattﬁnden. Solche Gesetze aber werden im Allgemeinen partielle

Diﬀerentialgleichungen sein, und die Ableitung der Gesetze f¨ ur ausgedehnte K¨orper

und Zeitr¨aume aus ihnen erfordert die Integration derselben. Es sind also Methode

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 5

Such a research method had already been announced by Riemann him-

self in a rather cryptic way in 1850. When lecturing at the P¨adagogische

Seminar he noticed that it was possible to formulate a mathematical the-

ory by moving from elementary principles toward general laws valid in

all of a given continuous space without distinguishing between gravity,

electricity, magnetism and equilibrium of heat.

4

As Klein pointed out, the method of studying functions on the basis

of their behaviour in the inﬁnitely small had a physical counterpart in

the concept of a line of force. Moreover, Klein suggested a kind of dualism

between Riemann’s mathematical thought and Faraday’s concept of action

by contact, writing that: “If I may dare to proceed with so forceful the

analogy, then I shall say that Riemann in the ﬁeld of mathematics and

Faraday in the ﬁeld of physics are parallel ” [Klein 1894, p. 484].

Supporting Klein’s point of view, in Raum Zeit Materie Weyl stated

that the passage from Euclidean to Riemannian geometry “is founded

in principle on the same idea as that which led from physics based

on action at a distance to physics based on inﬁnitely near action”

[1919a/1922, p. 91]. In fact, according to Weyl:

“The principle of gaining knowledge of the external world from the

behaviour of its inﬁnitesimal parts is the mainspring of the theory of

knowledge in inﬁnitesimal physics as in Riemann’s geometry, and, indeed,

the mainspring of all the eminent work of Riemann, in particular, that

dealing with the theory of complex functions” [1919a/1922, p. 92].

1. ON THE SOURCE OF RIEMANN’S ANALYTICAL WORK

Riemann introduced his ideas on complex function theory in his 1851

paper which concluded his studies at G¨ottingen. Riemann’s starting point

n¨othig, durch welche man aus den Gesetzen im Unendlichkleinen diese Gesetze im

Endlichen ableitet, und zwar in aller Strenge ableitet, ohne sich Vernachl¨assigungen zu

erlauben. Denn nur dann kann man sie an der Erfahrung pr¨ ufen” [Riemann 1869, p. 4].

4

“So z.B. l¨asst sich eine vollkommen in sich abgeschlossene mathematische Theorie

zusammenstellen, welche von den f¨ ur die einzelnen Punkte geltenden Elementar-

gesetzen bis zu den Vorg¨angen in dem uns wirklich gegebenen continuirlich erf¨ ullten

Raume fortschreitet, ohne zu scheiden, ob es sich um die Schwerkraft, oder die

Electricit¨at, oder den Magnetismus, oder das Gleichgewicht der W¨arme handelt”

(in [Dedekind 1876, p. 545]).

6 U. BOTTAZZINI AND R. TAZZIOLI

was given by the equations

(1.1)

∂u

∂x

=

∂v

∂y

,

∂u

∂y

= −

∂v

∂x

,

which have to be satisﬁed by the function w = u + iv of a variable

z = x + iy. From (1.1) he deduced the equations ∆u = 0, ∆v = 0 which

are the basis for investigating the properties of the functions u and v

[Riemann 1851, p. 7].

As Prym was to write to Klein after Riemann’s death,

5

since his student

days Riemann had attributed great importance to equations (1.1) for the

continuation of a function from one complex domain to another. According

to him, equations (1.1) explain why correct results can be obtained even

when working with divergent series, as Euler repeatedly did.

It is a well known fact that Riemann’s complex function theory is deeply

connected with potential theory in two dimensions — a theory he was

well acquainted with. Indeed, as a student Riemann had followed Weber’s

lectures in 1849 and the following year he participated in the physics

seminar jointly founded and led by Gauss and Weber. Gauss himself had

developed the theory of the Laplace equation in a paper of 1839. He had

determined the potential function in diﬀerent cases and, in particular, he

had studied the problem of the distribution of masses or electric charges

on a closed surface S, assuming the potential to be constant on S.

From a mathematical point of view, this reduced the problem to

minimizing the following integral

J =

V

| gradu|

2

dv.

5

“Nach einer Mittheilung, die mir Riemann in Fr¨ uhjahre 1865 w¨ahrend meines

Pisaner Aufenthalts machte, ist derselbe zu einer Theorie der Functionen einer

ver¨andlichen complexen Gr¨osse durch die Beobachtung gekommen, dass Beziehungen

zwischen Functionen, die durch Entwicklung der betreﬀenden Functionen in Reihen

erhalten worden, bestehen bleiben, auch wenn man ¨ uber die Convergenzgebiete der

darstellenden Reihen hinausging, und dass man in vielen F¨allen richtige Resultate

erh¨alt, wenn man, wie Euler z.B. es wiederholt getan, mit divergenten Reihen operiert.

Er frug sich dann, was denn eigentlich die Function aus dem einen Gebiete in das

andere fortgesetzt, und gelangte zu der Einsicht, dass dies die partielle Diﬀerential-

gleichung thue. Dirichlet, mit dem er den Gegenstand besprach, stimmte dieser Ansicht

vollst¨andig bei; es f¨allt also diese Idee wohl noch in die Studienjahre Riemanns, vor

die Auﬀassung seiner Inauguraldissertation”. This letter from February 6, 1882 is kept

in Klein’s Nachlass [11, 383].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 7

Since J > 0, “a [homogeneous] distribution must necessarily exist, so that

the integral J has a minimum”, Gauss wrote [1839, p. 233]. This argument

was used systematically by Dirichlet in his lectures on the forces which

are inversely proportional to the square of distance.

Dirichlet [1876, p. 127] faced the problem of proving that a function

with continuous ﬁrst partial derivatives on a given bounded domain, which

satisﬁes the Laplace equation within the domain and has given values on

the boundary, always exist. Dirichlet’s existence proof of the solution of

the “Dirichlet problem” was based on the fact that the minimum for the

integral J existed (“Dirichlet principle”).

According to Riemann, potential theory as developed by Gauss and

Dirichlet was well suited to a particular geometrical object, the “Riemann

surface”, he had introduced in order to study multi-valued functions

such as algebraic functions and their integrals. Riemann required that

the surface associated to a function be composed of as many sheets as

were the branches of the function, connected in such a way to preserve

continuity and to yield a single-valued function on the surface. In this

way, he attained an abstract conception of the space of complex variables

by means of a geometrical formulation which his contemporaries were

to ﬁnd very hard to understand. Referring to a conversation he had with

Prym in 1874, Klein reported that Prym “told me that Riemann’s surfaces

originally are not necessarily many-sheeted surfaces over the plane, but

that, on the contrary, complex functions of positions can be studied on

arbitrarily given curved surfaces in exactly the same way as on the surfaces

over the plane” [Klein 1882/1893, p. x].

Riemann made the surface simply connected with suitable transver-

sal cuts (Querschnitte) and analysed the behaviour of the function in

the neighbourhood of the singularities — poles and branch points. Then,

thanks to the Dirichlet principle, Riemann stated and proved a fundamen-

tal existence theorem for a function with given singularities and boundary

conditions [1851, p. 34–35]. This is the global theorem which, in Riemann’s

words, opens the way to the the study of complex functions independently

of their analytical expressions [Riemann 1851, p. 35].

Many of the ideas of this paper, as Klein ﬁrst emphasized, were inspired

by physical topics. As Riemann told Betti (see [Bottazzini 1985, p. 559]),

the idea of transversal cut on a surface struck him after a long discussion

8 U. BOTTAZZINI AND R. TAZZIOLI

with Gauss on a mathematical-physical problem. In Brill and Noether’s

opinion, the origin of the concepts of Riemann surface and transversal

cut could be found in an unpublished note [Riemann 1876c] on a problem

of electrostatic or thermal equilibrium on the surface of a cylinder with

transversal cuts.

6

Apparently Riemann sought to evaluate the distribution of the static

electricity on the surface of a cylinder, due to constant forces along the

directrices of the cylinder. Laplace had already observed that, when the

interacting masses are placed on an inﬁnite cylinder and the forces are

constant along straight lines parallel to the directrices of the cylinder, the

evaluation of the potential of the bodies can be seen as a plane problem. To

this end, it is suﬃcient to replace every directrix by its intersection point

with a plane orthogonal to the directrix. Thus, the diﬀerential equation

involved reduces to the Laplace equation ∆u = 0 in two variables and

the sought for potential function

¸

m/r reduces to the “logarithmic”

potential

¸

mlog(r).

Accordingly, Riemann stated that solving this problem was equivalent

to ﬁnding a function u having given boundary values and satisfying the

diﬀerential equation ∆u = 0 within a surface S, which he supposed to be

plane, simply connected, simple sheeted and bounded by n arbitrarily

given curves [Riemann 1876c, p. 440]. In turn, this problem could be

reduced to the easier one of determining a function ζ = ξ + iη of the

complex variable z = x + iy which is ﬁnite and continuous within S and

takes real values on the boundary, where it becomes inﬁnite of the ﬁrst

order in just one point for each curve of the boundary.

Under the condition that ζ goes from −∞ to +∞ when any of the

boundary curves is travelled in a positive sense, “one can easily” show

that ζ takes every real value exactly once on each curve of the boundary

whereas within S it takes every complex value n times (with Imζ > 0).

Thus, one gets a conformal map of S onto a n-sheeted surface T over

the upper half plane, whose boundary lines on the n sheets of the surface

coincide with the real axis. As within T there are (2n −2) branch points

[Riemann 1857b, p. 113], the problem reduces to the determination of

6

As Weber remarked, apart from some hints by Riemann, this note reduces to sheets

with calculations. “Wir sind geneigt, die[se] Note als eine der fr¨ uhesten Arbeiten

Riemann’s, oder doch ihren Gedankengang als den Ausgangspunkt f¨ ur Riemann’s

Arbeiten ¨ uber Functionentheorie zu bezeichnen” [Brill and Noether 1894, p. 259].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 9

a function of ζ branched like T, such that its real part u is continuous

within T and takes arbitrarily given values on the n lines of the boundary.

By means of both the Green function and the Riemann-Schwarz

principle of symmetry, Riemann showed how the problem could be solved

by following the same procedure he used in his paper on Abelian functions

[Riemann 1857b, pp. 113, 119] and determining the Green function by

means of an Abelian integral of the third kind. Next he discussed the

particular case in which the boundary of S is given by n circles. As Brill

and Noether [1894, p. 258] remarked, in this way Riemann established a

close connection between the theory of Abelian integrals, the conformal

mapping problem and the fundamental existence theorem Riemann stated

in his 1851 paper.

On the contrary, according to Klein the motivation for Abelian function

theory, which was to be developed by Riemann [1857b], lay in Riemann’s

researches in conductors and galvanic currents. In Klein’s opinion, Rie-

mann’s fundamental existence theorem of a harmonic function u could be

obtained from the following “thought experiment ” (Gedankenexperiment):

let a n-sheeted, closed Riemann surface over the complex plane be a uni-

form conductor with the two poles of a galvanic battery at the points A

1

and A

2

; on these assumptions a current is created, whose potential u is

deﬁned and single-valued on the surface, it satisﬁes the equation ∆u = 0,

and in A

1

and A

2

becomes inﬁnite when r

1

and r

2

go to inﬁnity as log r

1

and −log r

2

respectively [Klein 1926, p. 260–261]. Klein took this as the

starting-point of his presentation of Riemann’s theory of algebraic func-

tions and their integrals, claiming that:

“I have no doubt that he [Riemann] started from precisely those physical

problems, and then, in order to give what was physically evident the

support of mathematical reasoning, he afterwards substituted Dirichlet’s

Principle” [Klein 1882/1893, p. x].

After referring to “the conditions under which Riemann worked in

G¨ottingen” as well as to his writings on Naturphilosophie, Klein con-

cluded that “anyone who clearly understands” this “will, I think, share

my opinion”. In Klein’s view, Riemann’s “general problem” (allgemeine

Fragestellung) was the following: “to study the streamings in the ﬁrst

place and thence to work out the theory of certain analytical functions”

[Klein 1882/1893, p. 22]. Admittedly, Klein added that his presentation

10 U. BOTTAZZINI AND R. TAZZIOLI

“by no means include[d] the whole of what [Riemann] intended in the

theory of functions” and recognized that, even in the case of algebraic

functions and their integrals, his point of view was “necessarily very sub-

jective”.

In spite of Klein’s hope, his interpretation of the origins of Riemann’s

theory met strong criticism. In his review of Klein’s booklet, M. Noether

[1882] openly questioned Klein’s historical reconstruction. This was also

rejected by Prym

7

and Betti did the same in a letter to Klein from

March 22, 1882. Answering a precise question asked by Klein, he stated

that:

“Riemann ne m’a jamais dit avoir d´evelopp´e la th´eorie des courants

stationnaires dans un ﬂuide incompressible, dans un espace quelconque

`a trois dimensions et `a courbure quelconque, quoiqu’il se soit entretenu

avec moi plusieurs fois sur les travaux de Mr. Helmholtz” [Klein Nach-

lass 8, 86].

8

Some days later Klein addressed Bianchi [Opere 11, pp. 116–117], a

former student of Betti, who had spent a period in M¨ unchen with Klein.

In his reply Bianchi reported that, according to Betti’s record, Riemann

never made any connection between electric currents and closed surfaces

“free in space”. Instead, “[Betti] can ensure you with all certainty that

once Riemann told him that he [Riemann] had been led to his analytical

theory and his way of thinking [Anschauungsweise] after dealing with

questions and problems of physics” [Klein Nachlass 8, 100].

7

Contrary to Klein’s opinion, in the letter mentioned above (see footnote 5) Prym

stated that: “Ich halte es daher auch f¨ ur sehr wahrscheinlich, dass ¨ahnliche Ideen, wie

Sie sie entwickeln, von Riemann verfolgt worden sind, aber erst nachdem die Theorie

der Abelschen Functionen vollendet war. Sie dagegen scheinen der Ansicht zu sein,

dass Riemann von dem allgemeinen, auf beliebig beschaﬀene Fl¨achen bez¨ uglichen Falle

zu dem Falle der Abelschen Functionen als einem speciellen Hinabgestiegen sei. Ich

meine, der umgekehrte Weg, den ja auch Sie, wenn auch nicht in Ihrer Abhandlung,

so doch in Ihren Studien eingeschlagen haben, sein der nat¨ urliche und entspreche am

meisten den Gesetzen der geschichtlichen Entwicklung.”

8

French revised [Eds.].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 11

2. RIEMANN’S NEUE MATHEMATISCHEPRINCIPIEN DER

NATURPHILOSOPHIE

After the completion of his thesis, Riemann himself gave a hint of his

research projects by writing in an undated note that his “main work”

involved “a new interpretation of the known laws of nature — whereby

the use of experimental data concerning the interaction between heat, light,

magnetism, and electricity would make possible an investigation of their

interrelationship. I was led to this primarily through the study of the works

of Newton, Euler and, on the other side, Herbart ”.

9

As for the latter, it is worth noting here that Riemann could agree

“almost completely” with the results of Herbart’s early research while

he rejected his philosophy “at an essential point ” which involved Natur-

philosophie. In particular, Herbart’s psychology

10

inspired both Riemann’s

model of the substance (or ether) and his principles of Naturphilosophie.

Herbart had deﬁned the “psychic act ” (or representation) as an act

of self-preservation with which the ego opposed the perturbations coming

from the external world. He imagined a continuous ﬂow of representations

going from the ego to the conscious and back and studied the connections

between diﬀerent representations in mechanical terms as compositions

of forces.

Riemann’s more coherent attempt to give a systematic presentation

of his ideas on the propagation of physical phenomena like gravitation

9

Riemann wrote in fact: “Meine Hauptarbeit betriﬀt eine neue Auﬀassung der bekan-

nten Naturgesetze - Ausdruck derselben mittelst anderer Grundbegriﬀe — wodurch

die Benutzung der experimentellen Data ¨ uber die Wechselwirkung zwischen W¨arme,

Licht, Magnetismus und Electricit¨at zur Erforschung ihres Zusammenhangs m¨oglich

wurde. Ich wurde dazu haupts¨achlich durch das Studium der Werke Newton’s, Euler’s

und — anderseits — Herbart’s gef¨ uhrt” [1876a, p. 507].

10

“Was [Herbart] betriﬀt — Riemann wrote — so konnte ich mich den fr¨ uhesten

Untersuchungen Herbart’s, deren Resultate in seinen Promotions — und Habilitation-

sthesen (vom 22. u. 23. Oktober 1802) ausgesprochen sind, fast v¨ollig anschliessen,

musste aber von dem sp¨ateren Gange seiner Speculation in einem wesentlichen Punkte

abweichen, wodurch eine Verschiedenheit in Bezug auf seine Naturphilosophie und

diejenigen S¨atze der Psychologie, welche deren Verbindung mit der Naturphilosophie

betreﬀen, bedingt ist” [1876a, pp. 507–508]. Riemann summarized his philosophical

views by saying: “Der Verfasser ist Herbartianer in Psychologie und Erkenntnis-

stheorie (Methodologie und Eidologie), Herbart’s Naturphilosophie und den darauf

bez¨ uglichen metaphysischen Disciplinen (Ontologie und Synechologie) kann er meis-

tens nicht sich anschliessen” [1876a, p. 508].

12 U. BOTTAZZINI AND R. TAZZIOLI

and light was made in March 1853 in a paper which, echoing Newton’s

Principia, he did not hesitate to entitle Neue mathematische Principien

der Naturphilosophie (New mathematical principles of natural philoso-

phy). Indeed, Riemann himself regarded his writings on Naturphilosophie

as fundamental and intended to publish them, as he wrote to his brother

Wilhelm in December 1853 (see [Dedekind 1876, p. 547]).

In his paper Riemann began by claiming that the basis of the general

laws of motion for ponderable bodies, which are posed at the beginning of

Newton’s Principia, lies in the internal state of the bodies [1853, p. 528].

Led by the analogy with Herbart’s psychology Riemann made the hypoth-

esis that the universe (Weltraum) was ﬁlled with a substance (Stoﬀ ) ﬂow-

ing continually through atoms and there disappearing from the material

world (K¨orperwelt ) [1853, p. 529].

11

On the basis of this rather obscure idea Riemann tried to build a

mathematical model of the space surrounding two interacting particles

of substance. He introduced a cartesian coordinate system and considered

a single particle of substance as concentrated at the point O(x

1

, x

2

, x

3

) at

the time t and at the point O

(x

1

, x

2

, x

3

) at the time t

, where x

1

, x

2

, x

3

are functions of x

1

, x

2

, x

3

. Then, “according to a well known theorem”,

the two homogeneous diﬀerential forms

ds

2

= dx

2

1

+dx

2

2

+dx

2

3

and ds

2

= dx

2

1

+ dx

2

2

+ dx

2

3

could be expressed in the following way

ds

2

= G

2

1

ds

2

1

+G

2

2

ds

2

2

+G

2

3

ds

2

3

,

ds

2

= ds

2

1

+ ds

2

2

+ ds

2

3

,

where ds

1

, ds

2

, ds

3

was an appropriate new basis [Riemann 1853, p. 530].

12

Riemann called the quantities G

1

−1, G

2

−1, G

3

−1 the main dilatations

(Hauptdilatationen) of the particle at O and denoted them by λ

1

, λ

2

, λ

3

.

11

In a footnote Riemann added that: “In jedes ponderable Atom tritt in jedem

Augenblick eine bestimmte, der Gravitationskraft proportionale Stoﬀmenge ein und

verschwindet dort. Es ist die Consequenz der auf Herbart’schem Boden stehenden

Psychologie, dass nicht der Seele, sondern jeder einzelnen in uns gebildeten Vorstel-

lung Substantialit¨at zukomme” [1853, p. 529].

12

The conditions according to which the equation ds

2

= ds

2

is satisﬁed are the Lam´e

equations. They were ﬁrst published by Lam´e [1859, pp. 99, 101].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 13

Riemann’s result can be interpreted in terms of the classical theory of

elasticity. Supposing, as Riemann did, that the ether is an elastic, homo-

geneous, isotropic medium, then one can consider an elastic deformation

changing P(x

1

, x

2

, x

3

) and Q(x

1

+dx

1

, x

2

+dx

2

, x

3

+dx

3

) to P

(x

1

, x

2

, x

3

)

and Q

(x

1

+ dx

1

, x

2

+ dx

2

, x

3

+ dx

3

) respectively. Under these assump-

tions, it is possible to compare the distance ds between P and Q with the

corresponding distance ds

**after the deformation.
**

13

If x

i

= x

i

+ u

i

(where u

i

is the displacement of P due to the

deformation) one has

dx

i

= dx

i

+ du

i

= dx

i

+

3

¸

k=1

∂u

i

∂x

k

dx

k

.

Then

ds

2

= ds

2

+

3

¸

k,=1

∂u

∂x

k

dx

k

dx

+

3

¸

k,,i=1

∂u

∂x

i

∂u

∂x

k

dx

i

dx

k

.

Now it is possible to calculate the variation δ(ds

2

) = ds

2

− ds

2

(2.1) δ(ds

2

) =

3

¸

k,i=1

e

ik

dx

i

dx

k

,

where

e

ik

=

∂u

i

∂x

k

+

∂u

k

∂x

i

+

3

¸

=1

∂u

∂x

i

∂u

∂x

k

·

This quantity deﬁnes the deformation and coincides with the strain

tensor in elasticity theory. If the deformation is supposed inﬁnitely small,

the strain tensor becomes e

ik

=

∂u

i

∂x

k

+

∂u

k

∂x

i

. Since it is symmetric, it is

possible to ﬁnd a basis such that e

ik

= 0 if i = k.

Riemann compared ds

2

and ds

2

by a direct calculation on the same

basis ds

1

, ds

2

, ds

3

, with respect to which both ds

2

and ds

2

were orthog-

onal. He found that

(2.2) δ(ds

2

) = ds

2

− ds

2

= (G

2

1

−1)ds

2

1

+ (G

2

2

−1)ds

2

2

+ (G

2

3

−1)ds

2

3

.

13

See for instance Brillouin [1938, ch. X].

14 U. BOTTAZZINI AND R. TAZZIOLI

In this case for the components of the strain tensor one has

e

11

= G

2

1

−1, e

22

= G

2

2

−1, e

33

= G

2

3

−1, e

ik

= 0 if i = k.

Since the linear extensions after the deformation with respect to the axes

x

1

, x

2

, x

3

are deﬁned as ∆

i

=

√

1 +e

ii

−1, for i = 1, 2, 3, then

∆

1

= G

1

−1, ∆

2

= G

2

−1, ∆

3

= G

3

−1.

They coincide with Riemann’s Hauptdilatationen λ

1

, λ

2

, λ

3

, which there-

fore can be studied in the conceptual framework given by the theory of

elasticity.

Riemann then supposed that the variation δ(ds

2

) produced a force able

to modify the particle in such a way that the particle itself, opposing this

deformation, would propagate the physical forces through the space.

14

In

order to describe the reaction of the particle, Riemann introduced the

hypothesis that the dilatations were inﬁnitely small, and so the produced

forces were linear functions of λ

1

, λ

2

, λ

3

. By assuming the homogeneity of

the substance, the moment of force (Kraftmoment ) was given by

1

2

δ

a(λ

1

+λ

2

+λ

3

)

2

+b(λ

2

1

+λ

2

2

+λ

2

3

)

**where a, b were constants.
**

Riemann regarded this moment as a resultant of the forces which tend

to lengthen or shorten the line elements ending at O [1853, p. 531]. He

formulated the following “law of eﬀects” (Wirkungsgesetz): if dV and dV

represent the volumes of the particle at time t and t

respectively, then

the force arising from two diﬀerent states (Stoﬀzust¨ande) of the substance

in diﬀerent times t, t

**can be described in the following way
**

(2.3) a

dV − dV

dV

+b

ds − ds

ds

·

Moreover, since the two addenda in (2.3) are independent of each other,

the two actions produce diﬀerent eﬀects, whose development follows

diﬀerent laws. One could observe that if the Hauptdilatationen λ

1

, λ

2

, λ

3

,

of the particle of ether were equal to zero, the distance between two

inﬁnitely close points of ether remains constant after the deformation

14

See footnote 15 below.

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 15

at issue. In this case no physical phenomenon could occur in the space

surrounding the particle, and the Kraftmoment given by (2.3) vanishes.

Riemann calculated the variation of ds during the time dt and found

that the quantity δ ds/dt was equal to

t

−∞

dV

− dV

dV

ψ(t −t

) δt

+

t

−∞

ds

− ds

ds

ϕ(t −t

) δt

.

But how can we choose the functions ψ and ϕ in order that gravity,

heat and light propagate through space? This was the question Riemann

left open [1853, p. 532]. In this connection he limited himself to state that

“the eﬀects of ponderable matter on ponderable matter were attractive and

repulsive forces inversely proportional to the square of the distance” or

“light and radiant heat ”. They could be explained by assuming that every

particle of the homogeneous substance ﬁlling space has a direct eﬀect only

on its neighbourhood.

15

The mathematical law (2.3) according to which

this happens, can be divided into:

“1) the resistance with which a particle opposes a change of its volume,

and

2) the resistance with which a physical line element opposes a change

of length.

Gravity and electric attraction and repulsion are founded on the ﬁrst

part, light, heat propagation, electrodynamic and magnetic attraction and

repulsion on the second part ”.

16

15

“Es kann also die Wirkung der allgemeinen Gravitation auf ein ponderables Atom

durch den Druck des raumerf¨ ullenden Stoﬀes in der unmittelbaren Umgebung desselben

ausgedr¨ uckt und von demselben abh¨angig gedacht werden. Aus unserer Hypothese folgt

nothwendig, dass der raumerf¨ ullende Stoﬀ die Schwingungen fortpﬂanzen muss, welche

wir als Licht und W¨arme wahrnehmen” [Riemann 1853, p. 529].

16

“Beide Classen von Erscheinungen lassen sich erkl¨aren, wenn man annimt, dass

den ganzen unendlichen Raum ein gleichartiger Stoﬀ erf¨ullt, und jedes Stoﬀtheilchen

unmittelbar nur auf seine Umgebung einwirkt.

Das mathematische Gesetz, nach welchem dies geschieht, kann zerf¨allt gedacht wer-

den

1) in den Widerstand, mit welchem ein Stoﬀtheilchen einer Volum¨anderung, und

2) in den Widerstand, mit welchem ein physisches Linienelement einer L¨angen-

¨anderung widerstrebt.

Auf dem ersten Theil beruht die Gravitation und die electrostatische Anziehung und

Abstossung, auf dem zweiten die Fortpﬂanzung des Lichts und der W¨arme und die elec-

trodynamische oder magnetische Anziehung und Abstossung” [Riemann 1853, p. 532].

16 U. BOTTAZZINI AND R. TAZZIOLI

In Gravitation und Licht which constitutes the last section of his Frag-

mente on Naturphilosophie [1876a, pp. 532–538], Riemann presented a

mathematically more sophisticated attempt at a uniﬁed explanation of

both gravitation and light. Facing the problem of explaining Newton’s

theory of gravitation, Riemann supposed that a determinate cause (bes-

timmte Ursache) existed at every point of space. In virtue of this cause

the point was able to interact with other ponderable bodies by means of

a force, which was inversely proportional to the square of their distance.

He assumed that this cause should be sought in “the form of motion

of a substance spread continuously through the entire inﬁnite space. . . .

This substance can therefore be conceived as a physical space whose points

move in geometrical space”.

17

By means of this assumption, it was pos-

sible to explain the propagation of phenomena through space and then

light and gravitation had to be explained by means of the motion of this

substance.

18

The further development he drew from this hypothesis divided in two

parts:

(a) to ﬁnd the mathematical laws of motion of this substance which

can be used for explaining the phenomena;

(b) to clarify the causes (Ursachen) by means of which one could

explain these motions.

In commenting upon the second point, which was metaphysical in

character, Riemann observed that, contrary to Newton’s own opinion, his

law of attraction had not been thought to need any further explanation

for long time. Riemann was pleased to add in a footnote the celebrated

passage of Newton’s third letter to Bentley where Newton stated that

believing in action at a distance “without the mediation of anything else”

was “so great an absurdity” [1876a, p. 534].

In this connection, it is worth noting that in the General Scholium

17

“Die nach Gr¨osse und Richtung bestimmte Ursache . . . suche ich in der Bewe-

gungsform eines durch den ganzen unendlichen Raum stetig verbreiteten Stoﬀ. . . .

Dieser Stoﬀ kann also vorgestellt werden als ein physischer Raum, dessen Punkte sich

in dem geometrischen bewegen” [Riemann 1876a, p. 533].

18

“Nach dieser Annahme m¨ ussen alle von ponderablen K¨orpern durch den leeren

Raum auf ponderable K¨orper ausge¨ ubte Wirkungen durch diesen Stoﬀ fortgepﬂanzt

werden. . . . Diese beiden Erscheinungen, Gravitation und Lichtbewegung durch den

leeren Raum, aber sind die einzigen, welche bloss aus Bewegungen dieses Stoﬀes erkl¨art

werden m¨ ussen” [Riemann 1876a, p. 533].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 17

at the end of the third edition of the Principia, Newton himself had

spoken about “a certain most subtle spirit which pervades and lies hid

in all gross bodies”, through which one could explain both the attraction

of the “particles of bodies” and the action of the electric bodies and

light [1726/1934, p. 547]. The hypothesis about ether had also been

advanced by Euler, “in his magniﬁcent attempt ” — as A. Speiser deﬁned

it [1927, p. 106] — to draw up a uniﬁed theory of gravitation, light,

electricity and magnetism. Euler’s related papers, together with his Lettres

`a une princesse d’Allemagne, are included in the 1838 Brussels edition of

his works. Both these papers and the Lettres are likely to have been the

source of inspiration to which Riemann referred in his undated manuscript

note.

19

From the mathematical point of view, Riemann supposed that the real

motion of the substance v was composed of two motions u and w, that

is v = u + w, which were responsible for the propagation of gravitation

and light respectively. Then, he proposed to ﬁnd the “laws of motion of

substance in empty space” [1876a, p. 537]. To this end he identiﬁed the

forces with the motion caused by them and observed that the gravitational

force u, deduced from the potential V by means of the relation u = gradV ,

had to satisfy both the conditions of being a closed form and the equation

∆V = −4πρ as well [Riemann 1876a, pp. 534–535].

The diﬀerential equations which characterized the propagation of light

were a continuity equation

div w = 0

and a wave equation for transwerse oscillations in velocity w

∂

2

w

∂x

2

1

+

∂

2

w

∂x

2

2

+

∂

2

w

∂x

2

3

=

1

c

2

∂

2

w

∂t

2

,

where c was the speed of the light.

As N. Wise has remarked, “combining the two motions for gravity

and light produced a well-behaved velocity function, which conﬁrmed the

possibility of uniting the two processes” [1981, p. 290]. As a consequence

of the conditions valid for u and w, Riemann found for the motion of the

19

See above footnote 9.

18 U. BOTTAZZINI AND R. TAZZIOLI

substance in empty space, v, the following equations

div v = 0, (2.4)

1

c

2

∂

2

∂t

2

−

∂

2

∂x

2

1

+

∂

2

∂x

2

2

+

∂

2

∂x

2

3

∂v

2

∂x

3

−

∂v

3

∂x

2

= 0,

1

c

2

∂

2

∂t

2

−

∂

2

∂x

2

1

+

∂

2

∂x

2

2

+

∂

2

∂x

2

3

∂v

3

∂x

1

−

∂v

1

∂x

3

= 0,

1

c

2

∂

2

∂t

2

−

∂

2

∂x

2

1

+

∂

2

∂x

2

2

+

∂

2

∂x

2

3

∂v

1

∂x

2

−

∂v

2

∂x

1

= 0.

(2.5)

According to Riemann, these equations show how the motion of a

single particle of the substance depends only on the motions of the

particles around it.

20

Equation (2.4) proves that the density is not changed

during the motion of the substance while the condition expressed by the

equations (2.5) coincides with the condition that

1

c

2

∂

2

∂t

2

−

∂

2

∂x

2

1

+

∂

2

∂x

2

2

+

∂

2

∂x

2

3

(v

1

dx

1

+v

2

dx

2

+v

3

dx

3

)

is an exact diﬀerential dW.

Starting from a set of relations which were valid in the inﬁnitely small,

Riemann succeeded in formulating a theory valid in ﬁnite regions. In fact,

the solution of the diﬀerential system (2.4, 2.5) determined the motion of

the ether propagating light and gravity through the space. According to

his student Schering, by means of his model of the ether Riemann hoped

to eliminate from the laws of interaction those “speciﬁcations which refer

to action at a distance” as they “always depend on the properties of the

surrounding space” [1866/1991, p. 838].

3. PHYSICAL RESEARCHES

In 1854 during his Easter holidays, while preparing his Habilitationsvor-

trag on the principles of geometry, Riemann met the experimental physi-

cist Kohlrausch, who told him about an “unexplored phenomenon” con-

cerning the electric residuum in Leyden jars. Riemann intended to explain

20

“Diese Gleichungen zeigen, dass die Bewegung eines Stoﬀpunktes nur abh¨angt von

den Bewegungen in den angrenzenden Raum- und Zeittheilen, und ihre (vollst¨andigen)

Ursachen in den Einwirkungen der Umgebung gesucht werden k¨onnen” [Riemann

1876a, p. 537].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 19

Kohlrausch’s experiment, as he wrote to his brother, by means of his

research on the connections between electricity, gravity, light and mag-

netism. In such a way, Riemann hoped that his “main work could achieved

a favourable reception”.

21

According to Kohlrausch’s experiment, in a Leyden jar which had been

charged, then discharged and left isolated for some time, a residual charge

appeared. Kohlrausch [1854, pp. 61–62] divided the total charge in the jar

into two parts: the free charge, L, which could be discharged, and the

charge r that appeared (at least in part) only after the exhaustion of the

free charge (r was called das Residuum or der R¨uckstand). The “hidden”

residuum was a quantity of r that could no longer be traced in the jar,

while the “reappearing” residuum was the portion of residuum which came

back again after the discharge. Kohlrausch observed the “reappearing”

residuum by means of the Sinuselektrometer, an ingenious instrument

invented and built by Kohlrausch himself [1853].

In September 1854, lecturing at a meeting of the German “natural

philosophers”, Riemann tried to give an explanation of Kohlrausch’s

experiment. To this end, he assigned to every ponderable body a certain

conductivity coeﬃcient β, observing that in nature neither complete

insulators nor perfect conductors exist.

22

If u is the potential and ρ the

charge density, he deduced for the components of the electromotive force

at the point (x, y, z) at time t the following expressions

(3.1)

−

∂u

∂x

−β

2

∂ρ

∂x

,

−

∂u

∂y

−β

2

∂ρ

∂y

,

−

∂u

∂z

−β

2

∂ρ

∂z

.

Quantity (3.1) has to be proportional to the components of the current

intensity vector (ξ, η, ζ), and therefore the last expression is equal to

21

“Kohlrausch hatte nun einige Zeit vorher sehr genaue Messungen ¨ uber eine bis

dahin unerforschte Erscheinung (den electrischen R¨ uckstand in der Leidener Flasche)

gemacht und ver¨oﬀentlicht und ich hatte durch meine allgemeinen Untersuchungen

¨uber den Zusammenhang zwischen Electricit¨at, Licht und Magnetismus die Erkl¨arung

davon gefunden. . . . Mir ist diese Sache deshalb wichtig, weil es das erste Mal ist,

wo ich meine Arbeiten auf eine vorher noch nicht bekannte Erscheinung anwenden

konnte, und ich hoﬀe, dass die Ver¨oﬀentlichung dieser Arbeit dazu beitragen wird,

meiner gr¨osseren Arbeit eine g¨ unstige Aufnahme zu verschaﬀen” (in [Dedekind 1876,

pp. 548–549]).

22

Laplace in his M´ecanique c´eleste had already associated to each medium a grav-

itational coeﬃcient α, and had deduced the attraction law between the two masses

m

1

and m

2

: G(m

1

m

2

/r

2

) exp(−αr), where G is the constant of universal gravitation

and r the distance between m

1

and m

2

[Œuvres 5, pp. 445–452].

20 U. BOTTAZZINI AND R. TAZZIOLI

(αξ, αη, αζ), where α is a constant depending on the nature of the body

at issue. By means of both the principle of conservation of charge

∂ρ

∂t

+

∂ξ

∂x

+

∂η

∂y

+

∂ζ

∂z

= 0,

and the formula

∂

2

u

∂x

2

+

∂

2

u

∂y

2

+

∂

2

u

∂z

2

= −ρ,

Riemann deduced

(3.2) α

∂ρ

∂t

+ρ −β

2

∂

2

ρ

∂x

2

+

∂

2

ρ

∂y

2

+

∂

2

ρ

∂z

2

= 0.

He then compared “the consequences of the law in certain particu-

lar cases with experience” [Riemann 1854c, p. 51] and found well known

expressions in the case when the bodies were metallic and perfect conduc-

tors. The law (3.2) was also in accordance with the experimental data of

Kohlrausch about the free charge and the “reappearing” residuum.

In the last part of this paper, Riemann presented the two theories on

the nature of electric force, both discussed at that time — the unitary

conception and the dualistic one. According to the dualistic conception,

two opposite electricities existed in a body, a positive electricity and a

negative one. Riemann rejected this theory since, on Weber’s advice,

he had submitted it to calculations “without obtaining any satisfactory

results” [Riemann 1854c, p. 54]. Then Riemann chose Franklin’s unitary

theory that attaches a certain quantity of electricity to every body and

deﬁnes a positive or a negative electric state of a body as an excess or a

defect of this electricity respectively.

Riemann intended to publish an extended version of his results in

Annalen der Physik und Chemie but then, as Dedekind [1876, p. 550]

and Weber (in [Riemann 1876d, p. 367]) have reported, he changed his

mind because of the substantial changes requested by Kohlrausch, the

editor of the review.

23

Eventually his original manuscript was published by

Weber [Riemann 1876d] and it includes a section crossed out by Riemann

23

Instead of it, Riemann published in Kohlrausch’s Annalen his 1855 paper on Nobili’s

rings.

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 21

himself.

24

In this section, he developed a physical explanation of the

electromotive force and of electric propagation through a body on the

basis of the model of ether exposed in the Fragmente. Thus, instead

of Franklin’s unitary theory of electricity, Riemann assumed that the

electric current was caused by a reaction the body opposed to its own

electric state. This reaction was proportional to the charge density ρ

multiplied by the coeﬃcient β

2

, and it decreased or increased the electric

density according as the body contained positive or negative electricity.

Therefore, the transmission of electricity could not be instantaneous but,

Riemann continued, the electricity moves “against ponderable bodies” with

a speed which equals the electromotive force deriving from the potential u.

Moreover, this law of motion must be changed in order to show its

connection with heat and magnetism.

25

In a note dated July 1835 Gauss had already suggested a new theory of

electrodynamics. According to Gauss, two elements of electricity attract

and repulse each other with a force depending on their moving state.

26

He

gave for the force between two electric particles ε and ε

placed in (x, y, z)

and in (x

, y

, z

**) respectively, the following formula
**

(3.3) F =

εε

r

2

1 +

1

c

2

v

2

−

3

2

dr

dt

2

,

where r is the distance from ε to ε

, c is a constant velocity,

27

and v

is the velocity of ε(x, y, z) with respect to ε

(x

, y

, z

). As Maxwell

24

“Dieser ganze Artikel — Weber observed in a footnote — ist im Manuscript

durchgestrichen, wahrscheinlich nur aus dem Grunde, weil der Verfasser durch die

Eigenth¨ umlichkeit der hier vorgetragenen Auﬀassung, welche auf das Innigste mit

seinen naturphilosophischen Principien zusammenh¨angt, bei den Physikern damals

Anstoss zu erregen bef¨ urchtete” [Riemann 1876d, p. 371].

25

“Die Electricit¨at bewegt sich gegen die ponderabeln K¨orper mit einer Geschwindigkeit,

welche in jedem Augenblicke der aus diesen Ursachen hervorgehenden electromoto-

rischen Kraft gleich ist [the cause is the potential u which satisﬁes the equa-

tion ∆u = −ρ].

Uebrigens m¨ ussen diese Bewegungsgesetze der Electricit¨at, wenn deren Verh¨altniss

zu W¨arme und Magnetismus in Rechnung gezogen werden soll, vorbemerktermassen

selbst noch abge¨andert und umgeformt werden, und dann wird eine ver¨anderte Auf-

fassung dieser Erscheinungen n¨othig” [Riemann, 1876d, p. 371].

26

“Zwei Elemente von Electricit¨at in gegenseitiger Bewegung ziehen einander an oder

stossen einander ab nicht eben so als wenn sie in gegenseitiger Ruhe sind” [Gauss 1835,

p. 616].

27

Within the limit of observation, this is the velocity of light.

22 U. BOTTAZZINI AND R. TAZZIOLI

[1873, p. 484] showed, the formula (3.3) is inconsistent with the principle

of the conservation of energy.

In 1845, in a letter to Weber, Gauss [1845, p. 629] supposed that elec-

tricity propagated from one point to another not instantaneously, but in

time, as in the case of light. Gauss never published his ideas about elec-

trodynamics during his lifetime, and W. Weber’s theory of electrodynam-

ics, published in his celebrated Elektrodynamische Maassbestimmungen

(1846), was the ﬁrst result of this kind known to the scientiﬁc world.

According to Weber, the force between the two particles ε and ε

was

given by

(3.4) F =

εε

r

2

1 +

1

c

2

r

d

2

r

dt

2

−

1

2

dr

dt

2

.

The expressions (3.3) and (3.4) lead to the same result for mechanical

force between two electric currents, and this result coincides with that of

Amp`ere’s. Maxwell remarked that the formula (3.4) satisﬁed the principle

of the conservation of energy on assuming for the potential energy

(3.5) P =

εε

r

2

1 −

1

2c

2

dr

dt

2

.

But, as he pointed out, “an indeﬁnite amount of work cannot be generated

by a particle moving in a periodic manner under the action of the force

assumed by Weber” [Maxwell 1873, p. 484]. So Weber’s formula must be

rejected too.

As a follower of Gauss and Weber, Riemann himself tried to formulate

a new theory of electrodynamics in his 1858 paper Ein Beitrag zur

Elektrodynamik. He supposed that electric phenomena travel with the

velocity of light and that the diﬀerential equations for electric force are the

same of those valid for light and heat propagation [Riemann 1858, p. 288].

Riemann considered the electrodynamic system of two conductors C

and C

**moving one with respect to the other, and the galvanic currents
**

running through them. He studied the interaction of two particles ε and ε

,

the ﬁrst in C and the latter in C

, with coordinates (x, y, z) and (x

, y

, z

)

respectively. “With admirable directness — Rosenfeld has remarked — he

[wrote] down the generalised Poisson equation [for the potential ] involving

the operator now called ‘D’Alembertian’ and the expression for its solution

in the form of retarded potential ” [1956, p. 139].

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 23

Riemann indeed gave the following equation for the potential func-

tion U

(3.6)

∂

2

U

∂t

2

−α

2

∂

2

U

∂x

2

+

∂

2

U

∂y

2

+

∂

2

U

∂z

2

+ 4α

2

πρ = 0,

where ρ was the charge density in the point (x, y, z) and α the velocity

of propagation of electricity. According to (3.6), the potential U travels

with a velocity α and it reaches a point distant r from (x, y, z) after the

time r/α. As Reiﬀ and Sommerfeld [1902, p. 46] wrote, the interest in this

work is based on this result which makes Riemann appear as a “precursor”

of Maxwell.

Riemann supposed that the current was due to the positive and

negative electricities running through the wire, and that the two sums

¸

εf(x, y, z),

¸

ε

f(x

, y

, z

**) extended to all the charges were negligible
**

with respect to the same sums extended only to the positive or negative

charges. On these assumptions, Riemann found that the potential function

at the point (x, y, z) due to the point (x

, y

, z

) was f(t − r/α)/r, if the

charge −f(t) was placed in the point in question at time t, and r was the

distance between ε and ε

**. Moreover, if the coordinates of ε at the time t
**

were (x

t

, y

t

, z

t

), and those of ε

at time t

were (x

t

, y

t

, z

t

), he set

F(t, t

) =

(x

t

−x

t

)

2

+ (y

t

−y

t

)

2

+ (z

t

−z

t

)

2

−

1

2

,

and deduced the potential expressing the total eﬀect of the conductor C

on the conductor C

**from the time 0 to the time t
**

(3.7) P = −

t

0

¸¸

εε

F

t −

r

α

, τ

dτ.

Riemann tried to obtain the electrodynamic potential derived from the

Weber’s law from (3.7). But, as Clausius [1868] was to point out, Riemann

committed a mistake in the permutation of two integrations. According

to Weber (see [Riemann 1858, p. 293]), the mistake was discovered by

Riemann himself and this was the reason that convinced him to withdraw

the paper.

In Betti’s opinion however Riemann did not publish this paper because

it was in contrast with what he had stated in [Riemann 1854c, p. 54].

28

28

“Questo concetto della corrente elettrica, tutto ideale — Betti observed — `e poco in

armonia con ci`o che si conosce di essa, e pare che Riemann non ne fosse soddisfatto”

[1868, p. 242].

24 U. BOTTAZZINI AND R. TAZZIOLI

Betti himself suggested a new theory of electrodynamics. He supposed the

closed circuits to consist of polarized elements which acted on one another

as little magnets. According to Betti, the polarized elements interacted

not instantaneously but only after a time proportional to the distance

between the elements.

29

Maxwell was aware of Riemann’s and Betti’s electrodynamic researches,

and even he could not agree with “these eminent men”. In their theories,

indeed, the action travelled in a manner similar to that of light but,

Maxwell asked:

“If something is transmitted from one particle to another at a distance,

what is its condition after it has left the one particle and before it has

reached the other?” [1873/1954, II, p. 493].

In his 1858 paper Riemann did not mention any medium which prop-

agated the electric phenomena. He tried however to describe the ether

surrounding two interacting electric particles in his lectures on electric-

ity, gravity and magnetism, held in 1861 and published by Hattendorﬀ

in 1876. Here, Riemann tackled the electrodynamic problem of two con-

ductors moving one with respect to the other by using the calculus of

variations. He deduced from the principle of conservation of energy the

“extended theorem of Lagrange”

(3.8) δ

t

0

(T −D +S)dt = 0,

where T is the kinetic energy, S the potential depending only on the

coordinates (electrostatic potential) and D the potential depending on

the velocities too (electrodynamic potential) [Riemann, 1876e, p. 316].

If ε and ε

were two electric particles in C and in C

at the points

(x, y, z) and (x

, y

, z

**) respectively, S and D were expressed by the
**

following equations

D =

εε

r

u

2

2c

2

,

S = −

εε

r

,

where r was the distance between ε and ε

**, u the velocity of one particle
**

with respect to the other, and c the velocity of light.

29

Clausius also criticized some parts of Betti’s calculations.

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 25

By solving the Euler-Lagrange equations deduced from (3.8), Riemann

obtained the following expression for the force acting on ε and ε

(3.9)

F

x

=

εε

r

2

1 +

u

2

2c

2

∂r

∂x

+

εε

c

2

d

dt

x −x

r

,

F

y

=

εε

r

2

1 +

u

2

2c

2

∂r

∂y

+

εε

c

2

d

dt

y −y

r

,

F

z

=

εε

r

2

1 +

u

2

2c

2

∂r

∂z

+

εε

c

2

d

dt

z −z

r

.

If D is Weber’s electrodynamic potential, the interaction between ε

and ε

**deduced from (3.8) is given by Weber’s law (3.4) (see [Reiﬀ and
**

Sommerfeld 1902, pp. 48–49]).

By considering the eﬀect of all the particles of the conductor C

on the

particle ε, Riemann set S = εV and found for the potential function V

the relation

30

(3.10)

∂V

∂t

= div u =

∂u

1

∂x

+

∂u

2

∂y

+

∂u

3

∂z

,

where u

1

=

¸

ε

ε

r

dx

dt

,

u

2

=

¸

ε

ε

r

dy

dt

,

u

3

=

¸

ε

ε

r

dz

dt

·

According to Riemann, “from this diﬀerential equation it would be

possible to derive the meaning of the functions V , u

1

, u

2

, u

3

. One

could assume that electricity is propagated by an ether. By means of

equation [(3.10)] V may be regarded as the density and u

1

, u

2

, u

3

as

[the components of ] the intensity of the ﬂux of this ether”.

31

By this

identiﬁcation, the principle of conservation of charge held for the ether too.

Riemann explained his model of ether in his 1853 paper on the princi-

ples of Naturphilosophie. However, he was not able to incorporate electro-

static and electromagnetic eﬀects into it. In his 1861 lectures he assumed

that the ether satisﬁed physical properties which guaranteed electrical

propagation. As N. Wise has remarked: “a stable gradient in density would

30

For a discussion of this point see Wise [1981, p. 290].

31

“Auf Grund dieser Diﬀerentialgleichung k¨onnte man ¨ uber die Bedeutung der

Functionen V, u

1

, u

2

, u

3

eine Annahme machen. Man kann annehmen, die elektrische

Wirkung werde durch einen Aether vermittelt. Verm¨oge der Gleichung [(3.10)] liessen

sich dann V als die Dichtigkeit, u

1

, u

2

, u

3

als die Stromintensit¨aten dieses Aethers

ansehen” [Riemann 1876e, p. 330].

26 U. BOTTAZZINI AND R. TAZZIOLI

then correspond to electrostatic force, whereas a time rate of change in

density, and associated ﬂuxes, would correspond to electrodynamic eﬀects”

[1981, p. 290]. Therefore, the hypothesis (in [Riemann 1858]) that the

same diﬀerential equation held for light and electric propagation, made the

ether responsible for the transmission of both the phenomena. Referring

to his 1858 paper Riemann wrote to his sister that, notwithstanding the

fact that Gauss had formulated another theory of the connection between

electricity and light, “I am sure that my theory is right and that in a

couple of years it will be generally recognized as such”.

32

4. ON THE FOUNDATIONOF GEOMETRY

The attempt undertaken by Riemann in the Fragmente at formulating

a theory which could explain the transmission of phenomena is in our

opinion among the basic motivations for the study of manifolds presented

by Riemann in his 1854 Habilitationsvortrag.

As Dedekind [1876, p. 547] reported, even though Riemann should

have thought about this subject for long time, the question concerning

the principles of geometry that Gauss choose as the subject of Riemann’s

Probevorlesung was certainly the least prepared of those Riemann had

proposed, as he wrote to his brother on December 28, 1853. In this letter

Riemann stated that after the completion of his Habilitationsschrift [Rie-

mann 1854a] he once more took up his “other research” on Naturphiloso-

phie, which was almost ready for printing.

33

Some months later he con-

fessed to his brother to having been “so fully immersed” in his research on

Naturphilosophie that he was not able to rid himself of it when the subject

32

“Ich bin aber v¨ollig ¨uberzeugt, dass die meinige die richtige ist und in ein paar

Jahren allgemein als solche anerkannt werden wird” (in [Dedekind 1876, p. 553]).

33

“Meine andere Untersuchung ¨ uber den Zusammenhang zwischen Electricit¨at, Gal-

vanismus, Licht und Schwere hatte ich gleich nach Beendigung meiner Habilitation-

sschrift wieder aufgenommen und bin mit ihr so weit gekommen, dass ich sie in

dieser Form unbedenklich ver¨oﬀentlichen kann. Es ist mir dabei aber zugleich immer

gewisser geworden, dass Gauss seit mehreren Jahren auch daran arbeitet, und einigen

Freunden, u. A. Weber, die Sache unter dem Siegel der Verschwiegenheit mitgetheilt

hat, — Dir kann ich dies wohl schreiben, ohne dass es mir als Anmaassung ausgelegt

wird — ich hoﬀe, dass es nun f¨ur mich noch nicht zu sp¨at ist und es anerkannt werden

wird, dass ich die Sachen vollkommen selbst¨andig gefunden habe” (in [Dedekind 1876,

p. 547]).

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 27

of his lecture (Probevorlesung) was proposed to him (see [Dedekind 1876,

pp. 547–548]). H. Weyl quoted this passage from Riemann’s letter for sup-

porting his claim that Riemann’s research “on the connection between

light, electricity, magnetism and gravity” was not “objectively in any rela-

tion” to the content of his lecture [1919b/1991, p. 741].

Contrary to Weyl’s claim, in our opinion these letters suggest that

Riemann’s 1854 lecture was deeply connected with his research in mathe-

matical physics. In fact, one can argue that in this lecture Riemann tried

to generalize the ideas stated in his 1853 paper on Naturphilosophie to

n-dimensional manifolds, extending the “local” investigation of particles

of ether to the “global” analysis of space. In this connection the reference

Riemann made to Gauss and Herbart is also worth noting [1854b, p. 273].

From this point of view one can explain the seeming anomalies with

respect to contemporary research in foundations of geometry and in non-

Euclidean geometry, often mentioned by historians of mathematics. To be

sure, in his lecture Riemann only made a cryptic allusion in passing to

the possibility of the elliptic geometry. Nor did he refer to the work of

Bolyai and Lobachevsky, even though a paper of the latter had appeared

in 1837 in Crelle’s Journal and his Geometrische Untersuchungen had

been published in Berlin in 1840. However one can argue that Riemann

probably knew Lobachevsky’s geometry, at least in the version that

appeared in that Journal. In fact, the relevant volume of Crelle’s Journal

included both Lobachevsky’s paper and Dirichlet’s 1837 paper on the

representation of “arbitrary” functions in series of Legendre polynomials.

Thinking of Riemann’s paper [1854a], it seems very likely that he had

this volume in his hands when he worked at his Habilitationsschrift.

Had Riemann limited himself to reading the ﬁrst page of Lobachevsky’s

paper, he could have discovered that: “rien n’autorise, si ce ne sont les

observations directes, de supposer dans un triangle rectiligne la somme des

angles ´egale `a deux angles droits, et que la g´eom´etrie n’en peut pas moins

exister, sinon dans la nature, au moins dans l’analyse, lorsqu’on admet

l’hypoth`ese de la somme des angles moindre que la demi-circonf´erence du

cercle” [Lobachevsky 1837, p. 295].

After summarizing his main results concerning the geometry of a non-

Euclidean rectilinear triangle, Lobachevsky concluded that:

“L’hypoth`ese de la somme des angles d’un triangle moindre que deux

28 U. BOTTAZZINI AND R. TAZZIOLI

angles droits ne peut avoir d’application que dans l’analyse, puisque les

mesures directes ne nous montrent pas dans la somme des angles d’un

triangle la moindre d´eviation de deux angles droits. J’ai prouv´e ailleurs, en

m’appuyant sur quelques observations astronomiques, que dans un triangle

dont les cˆot´es sont de la mˆeme grandeur `a peu pr`es que la distance de

la terre au soleil, la somme des angles ne peut jamais diﬀ´erer de deux

angles droits d’une quantit´e qui puisse surpasser 0

, 0003 en secondes

sexag´esimales. Or cette diﬀ´erence doit ˆetre d’autant moindre que les cˆot´es

d’un triangle sont plus petits” [1837, pp. 302–303].

In addition to the results of the geodetical observation that Gauss had

reported at the end of his 1827 paper, this could have given Riemann fur-

ther suggestions for the remarks on the “empirical certainty” of geometry

which can be read in the concluding lines of the introduction to [Riemann

1854b]. According to Riemann, Euclidean geometry was not a “necessity”

but merely an “empirical certainty” and the facts on which it was based

were only “hypotheses”, no matter how high their probability within the

limits of observation [Riemann 1854b, p. 273]. As he wrote in a manuscript,

“the word hypothesis now has a slightly diﬀerent meaning than in Newton.

Today by hypothesis we tend to mean everything which is mentally added

to phenomena” [Riemann 1876a, p. 525].

As Gauss had done for the surfaces in his 1827 paper, Riemann too

attributed a crucial importance to the deﬁnition of the linear element ds

as starting point for the study of manifolds. With this in mind, we can

appreciate better the suggestions he could have drawn from the following

passage one can read in Lobachevsky’s paper:

“La g´eom´etrie imaginaire est con¸cue sur un plan plus g´en´eral que la

g´eom´etrie usit´ee qui n’en est qu’un cas particulier, et qui en d´erive dans

la supposition des lignes extrˆemement petites; de sorte que cette derni`ere

g´eom´etrie n’est sous ce rapport qu’une g´eom´etrie diﬀ´erentielle.

Les valeurs des ´el´ements diﬀ´erentiels des lignes courbes, des surfaces,

et du volume des corps sont les mˆemes dans la g´eom´etrie imaginaire et

dans la g´eom´etrie usit´ee” [1837, p. 302].

Whatever the case may have been, apparently Riemann had no real

interest in the problem of the foundations of geometry as such, in research

concerning the axioms of geometry (and in particular that of parallels).

“However interesting it may be to consider the possibility of this approach

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 29

to geometry, actually realising it would prove utterly sterile since in this

way we could never ﬁnd new theorems”, he wrote in a note kept in his

Nachlass.

34

In our opinion, Riemann’s Habilitationsvortrag can be interpreted as a

moment of both naturphilosophische and mathematical construction. In

an attempt to overcome “the shortcomings of the concepts” and to favour

progress “in recognizing the connection of things” [Riemann 1854b, p. 286],

he supplied a generalization to n-dimensional manifolds of what he had

elaborated concerning 3-dimensional space and the laws of propagation of

the phenomena as well.

Riemann deﬁned a metric in the whole space by associating to the

n-dimensional manifold V

n

the fundamental form

Φ =

n

¸

i,j=1

g

ij

dx

i

dx

j

(g

ij

positive deﬁnite matrix).

In this way, it was always possible to compare distances between diﬀerent

points and to study deformations of an elastic ﬂuid ﬁlling the whole

universe. Riemann tried to calculate how much V

n

, with the fundamental

form Φ, diﬀered from the Euclidean n-dimensional space E

n

. More

generally, he proposed to ﬁnd necessary and suﬃcient conditions such

that the forms Φ and Φ

=

n

¸

i,j=1

g

ij

dx

i

dx

j

could transform into each

other. This was equivalent to determine n new coordinates such that the

diﬀerential system

(4.1)

n

¸

i,j=1

g

ij

dx

i

dx

j

=

n

¸

i,j=1

g

ij

dx

i

dx

j

was satisﬁed. System (4.1) was the generalization to a n-dimensional

manifold of the expression

dx

2

1

+ dx

2

2

+ dx

2

3

= dx

2

1

+ dx

2

2

+ dx

2

3

,

which had been introduced in his 1853 paper on Naturphilosophie (see § 2

above). If the solution of (4.1) existed, it involved

1

2

n(n+1) arbitrary con-

stants. As n equations were satisﬁed by the new coordinates so

1

2

n(n −1)

34

“Wenn es aber auch interessant ist, die M¨oglichkeit dieser Behandlungsweise der

Geometrie einzusehen so w¨urde doch die Ausf¨ uhrung derselben ¨ausserst unfruchtbar

sein, denn wir w¨urden dadurch keine neuen S¨atze ﬁnden k¨onnen” (in [Scholz 1982,

p. 229]).

30 U. BOTTAZZINI AND R. TAZZIOLI

“functions of position” had to be deduced “by the nature of the manifold

to be represented” [Riemann 1854b/1979, p. 143]. In order to character-

ize these functions, Riemann introduced a system of normal coordinates,

which we denote x

1

, . . . , x

n

, and aﬃrmed that if ds

2

0

=

n

¸

i=1

(dx

i

)

2

then

there existed numbers c

ij,k

such that Φ became

Φ = ds

2

0

+

¸

ij,k

c

ij,k

(x

k

dx

i

−x

i

dx

k

)(x

dx

j

−x

j

dx

).

If Φ = ds

2

0

the manifold was called plane.

In 1861, Riemann once more considered this problem, when trying to

answer a question proposed by the Paris Academy on heat conduction in

homogeneous solid bodies. According to Weber, Riemann did not win this

prize since he did not completely explain the ways according to which he

found his results (see [Riemann 1861, p. 391]).

Riemann sought to answer the following general question: “quales esse

debeant proprietates corporis motum caloris determinantes et distributio

caloris, ut detur systema linearum quae semper isothermae maneant”.

In the second part of the paper he studied the particular case of a

homogeneous body. Riemann stated that the problem proposed by the

Academy was equivalent to ﬁnding necessary and suﬃcient conditions

according to which

n

¸

i,j=1

g

ij

dx

i

dx

j

=

n

¸

i,j=1

a

ij

dx

i

dx

j

where a

ij

are given constants. Since both the forms were positive deﬁnite,

the problem was to determine a new coordinate system in which Φ could

be expressed as the Euclidean form

n

¸

i=1

(dx

i

)

2

. To this end, Riemann

introduced the quantity (ij, k) which corresponded — but for the inessen-

tial factor 2 — to the curvature tensor

(4.2) R

ijk

=

1

2

∂

2

g

ik

∂x

j

∂x

+

∂

2

g

j

∂x

i

∂x

k

−

∂

2

g

i

∂x

j

∂x

k

−

∂

2

g

jk

∂x

i

∂x

+

n

¸

α,β=1

g

αβ

[j;α][ik;β] −[i;α][jk;β]

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 31

where

[ij;k] =

1

2

∂g

ik

∂x

j

+

∂g

jk

∂x

i

−

∂g

ij

∂x

k

**is the Christoﬀel symbol of ﬁrst kind. He found that the necessary
**

condition such that the system

(4.3)

n

¸

i,j=1

g

ij

dx

i

dx

j

=

n

¸

i=1

(dx

i

)

2

is satisﬁed, is given by R

ijk

= 0.

If we suppose, as Riemann had done in his 1853 paper, that an

ethereal medium ﬁlls all the universe, then we can associate to space the

fundamental forms

Φ =

n

¸

i,j=1

δ

ij

dx

i

dx

j

and Φ

=

n

¸

i,j=1

g

ij

dx

i

dx

j

at time t and t

respectively.

Now, let us consider an elastic deformation changing P(x

1

, x

2

, . . . , x

n

)

and Q(x

1

+ dx

1

, x

2

+ dx

2

, . . . , x

n

+ dx

n

) in P

(x

1

, x

2

, . . . , x

n

) and

Q

(x

1

+ dx

1

, x

2

+ dx

2

, . . . , x

n

+ dx

n

), and the distance ds between P

and Q in the distance ds

between P

and Q

. On these assumptions

ds

2

=

n

¸

i,j=1

δ

ij

(P)dx

i

dx

j

=

n

¸

i=1

(dx

i

)

2

,

ds

2

=

n

¸

i,j=1

g

ij

(P

)dx

i

dx

j

.

Let a

i

= ∂x

/∂x

i

, then

ds

2

=

n

¸

i,j=1

g

ij

(P

)a

i

a

k

j

dx

i

dx

j

=

n

¸

i,j=1

h

ij

(P

)dx

i

dx

j

.

Since the system of coordinates is the same for ds

2

and ds

2

, we can

evaluate the virtual displacement

(4.4) δ(ds

2

) = ds

2

− ds

2

=

n

¸

i,j=1

e

ij

(P

)dx

i

dx

j

,

32 U. BOTTAZZINI AND R. TAZZIOLI

where e

ij

= h

ij

− δ

ij

. The variation (4.4) is the generalization to any

curvilinear n-coordinate system of the expression (2.2) given by Riemann

[1853, p. 530] in cartesian, orthogonal 3-coordinates.

By extending the same procedure to any particle of ether one obtains

the quantities e

ij

(the strain tensor in elasticity theory) deﬁned on

the whole space. Moreover, according to Riemann’s assumption that an

ether particle propagates physical phenomena reacting to the deformation

described by e

ij

, one can suppose that the transmission of forces modiﬁes

the fundamental form Φ. From this point of view (4.4) can be interpreted

as a deep connection between physical phenomena and the variation of the

fundamental form. No variation occurs if e

ij

= 0: in this case no physical

force is propagated through space. The results obtained by Riemann in

his 1861 paper show that, if a solution of the system

(4.5)

n

¸

i,j=1

δ

ij

dx

i

dx

j

=

n

¸

i,j=1

g

ij

dx

i

dx

j

exists, then the curvature tensor R

ijk

is equal to zero. On the contrary,

if the strain tensor is diﬀerent from zero, then the fundamental form has

been modiﬁed and the curvature tensor does not vanish. Thus force and

curvature appear to be deeply connected, space being the responsible for

the transmission of phenomena by means of a variation of its fundamental

form. In other words, a force is always coupled to a change in the curvature

of space. This suggests a physical model of the space independent of

assumptions about the existence of the ether.

K. Pearson, the editor of Cliﬀord’s book [1885], emphasized this point

asking the question “whether physicists might not ﬁnd it simpler to assume

that space is capable of a varying curvature, and of a resistance to that

variation, than to suppose the existence of a subtle medium pervading

an invariable homaloidal [Euclidean] space” (in [Cliﬀord 1885, p. 203]).

According to Cliﬀord, “this variation of the curvature of space is what

really happens in that phenomenon which we call the motion of matter,

whether ponderable or ethereal ” and “in the physical world nothing else

takes place but this variation, subject (possibly) to the law of continuity”

[Cliﬀord 1876, p. 22].

Having this in mind and thinking of the research program on Natur-

philosophie which Riemann was pursuing at that time, we can try to

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 33

understand the cryptic and prophetic conclusion he drew at the end of

the lecture:

“Now it seems that the empirical notions on which the metric determi-

nations of space are based, the concept of a solid body and that of a light

ray, lose their validity in the inﬁnitely small; it is therefore quite deﬁnitely

conceivable that the metric relations of space in the inﬁnitely small do not

conform to the hypotheses of geometry” [Riemann 1854b/1979, p. 152].

Thus, “the question of the validity of the hypotheses of geometry in the

inﬁnitely small ” was linked to the determination of “metric relations of

space”. This was a question which, in Riemann’s view, might “still be

ranked as a part of the study of space”.

In this connection Riemann contrasted discrete manifolds with contin-

uous manifolds: “Either the reality underlying space” is given by a discrete

manifold which has the principle of its metrical relations in itself or “the

basis for the metric relations must be sought outside it [space], in binding

forces acting upon it ”. What has one to understand by this “reality”? Did

this involve Riemann’s speculations on ether which, as he wrote in the

manuscript Gravitation und Licht [Werke, p. 532 sq.], could be conceived

as “a substance spread continuously through the entire inﬁnite space”,

“as a physical space whose points move in geometrical space”? Riemann

left open the questions he raised concerning the nature of space. According

to his concluding remarks, an answer “can be found only by starting from

that conception of phenomena which has hitherto been approved by expe-

rience, for which Newton laid the foundation, and gradually modifying it

under the compulsion of facts which cannot be explained by it” [Riemann

1854b/1979, pp. 152–153]. “This leads us away into the domain of another

science, the realm of physics”, he concluded.

According to Weyl [1919a/1922, p. 97], the “full purport” of Riemann’s

concluding remarks “was not grasped by his contemporaries” with the

exception of “a solitary echo” in Cliﬀord’s writings. In Raum Zeit Materie

Weyl interpreted Riemann’s words in terms of Einstein’s theory of rela-

tivity: “Only now that Einstein has removed the scales from our eyes by

the magic light of his theory of gravitation do we see what these words

actually mean”. In this respect we cannot do better than referring to his

book (see in particular [Weyl 1919a/1922, pp. 96–102]). From a historical

point of view, however, Riemann’s statements seem to be better explained

34 U. BOTTAZZINI AND R. TAZZIOLI

in the light of his own attempts “to unify nature on the basis of a geo-

metrically conceived system of continuous dynamic processes in ether”;

“the ﬁrst attempt at a mathematically founded uniﬁed ﬁeld theory, much

in the spirit of Einstein’s later attempts”, as N. Wise [1981, p. 289] deﬁned

Riemann’s research on Naturphilosophie.

Acknowledgments

During the past years, sections of this paper have been presented at various

meetings. In particular, we would like to thank the colleagues attending the Work-

shop on the history of mathematics held at the Mittag-Leﬄer Institute (Djur-

sholm, 1991), the Annual G¨ottingen Workshop (1992) and the Oberwolfach meeting

(1994) for their useful comments. We are also grateful to the anonymous referees for

their helpful remarks and to Jeremy Gray for checking the English. We also wish

to thank the Nieders¨achsischen Staat-und Universit¨atsbibliothek G¨ottingen, Hand-

schriftenabteilung, for the kind permission of publishing excerpts from the letters kept

in Klein’s Nachlass.

BIBLIOGRAPHY

BETTI (E.)

[Opere] Opere matematiche (Accademia dei Lincei, ed.), 2 vols., Milano: Hoepli,

1903–1913.

[1868] Sopra la elettrodinamica, Opere 2, pp. 242–245.

BIANCHI (L.)

[Opere] Opere (Unione Matematica Italiana, ed.), 11 vols., Roma: Cremonese,

1952–1959.

BOTTAZZINI (U.)

[1985] Funzioni complesse e variet`a multidimensionali: globale e locale nella

matematica di B. Riemann, Scienza e ﬁlosoﬁa: Saggi in onore di L. Gey-

monat (C. Mangione, ed.), Milano: Garzanti, 1985, pp. 554–573.

[1986] The higher calculus: a history of real and complex analysis from Euler to

Weierstrass, New York: Springer, 1986.

[1994] Geometry and “Metaphysics of space” in Gauss and Riemann, Romanticism

in science (S. Poggi and M. Bossi, eds.), Dordrecht: Reidel, 1994, pp. 15–29.

BOTTAZZINI (U.) and TAZZIOLI (R.)

[1991] Etere, spazio, geometria: riﬂessioni sui frammenti ﬁlosoﬁci di Riemann,

Conoscenza e matematica (L. Magnani, ed.), Milano: Marcos y Marcos,

1991, pp. 383–409.

BRILL (A.) and NOETHER (M.)

[1894] Die Entwicklung der Theorie der algebraischen Funktionen in ¨alterer und

neuerer Zeit, Jahresbericht der Deutschen Mathematiker-Vereinigung, 3

(1894), pp. 107-565.

BRILLOUIN (L.)

[1938] Les Tenseurs en m´ecanique et en ´elasticit´e, Paris: Masson, 1938.

CLAUSIUS (R.)

[1868] Von Gauss angeregte neue Auﬀassung der elektrodynamischen Erscheinun-

gen, Annalen der Physik und Chemie, 135 (1868), p. 606.

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 35

CLIFFORD (W.K.)

[Papers] Mathematical papers (R. Tucker, ed.), London: Macmillan, 1882.

[1876] On the space-theory of matter, Proceedings of the Cambridge philosophical

Society, 2 (1876), pp. 157–58 (read Feb. 21, 1870); Papers, pp. 21–22.

[1885] The common sense of the exact sciences (K. Pearson, ed.), London: Macmil-

lan, 1885.

DEDEKIND (R.)

[1876] Bernhard Riemann’s Lebenslauf, in [Riemann Werke, pp. 541–558].

DIRICHLET (P.G. Lejeune-)

[Werke] Werke (L. Kronecker and L. Fuchs, eds.), 2 vols., Berlin: Mayer & M¨ uller,

1889–1897.

[1837] Sur les s´eries dont le terme g´en´eral d´epend de deux angles, et qui servent `a

exprimer des fonctions arbitraires entre des limites donn´ees, Journal f¨ ur

die reine und angewandte Mathematik, 17 (1837), pp. 35–56; Werke 1,

pp. 273–306.

[1876] Vorlesungen ¨ uber die im umgekehrten Verh¨altniss des Quadrats der Entfer-

nung wirkenden Kr¨afte (F. Grube, ed.), Leipzig: Teubner, 1876.

FARADAY (M.)

[1851] On the possible relation of gravity to electricity, Experimental researches in

electricity, t. III, London, 1885, pp. 161–168.

[1857] On the conservation of force, Experimental researches in chemistry and

physics, London, 1859, pp. 443–463.

FARWELL (R.) and KNEE (Ch.)

[1990] The missing link: Riemann’s ‘Commentatio’, diﬀerential geometry and

tensor analysis, Historia mathematica, 17 (1990), pp. 223–255.

GAUSS (C.F.)

[Werke]Werke (K¨oniglische Gesellschaft der Wissenschaften zu G¨ottingen, ed.),

12 vols., Leipzig-Berlin, 1863–1933.

[1813] Theorem ¨ uber die Anziehung, Werke 11, p. 71.

[1827] Disquisitiones generales circa superﬁcies curvas, Commentationes societatis

regiae scientiarum Gottingensis recentiores, 4 (1828); Werke 4, pp. 217–258.

French transl. 1855, repr. Blanchard, 1967. English transl. 1902, repr. in

Ast´erisque 62 (1979).

[1835] Grundgesetz f¨ ur alle Wechselwirkungen galvanischer Str¨ome, Werke 5,

pp. 616–620.

[1839] Allgemeine Lehrs¨atze in Beziehung auf die im verkehrten Verh¨altnisse wirk-

enden Anziehungs-und Abstossungs Kr¨afte, Resultate aus der Beobachtun-

gen des magnetischen Vereins im Jahre 1839 (K.F. Gauss and W. Weber,

eds.), Leipzig, 1840; Werke 5, pp. 194–242.

[1845] Brief an Weber, Werke 5, pp. 627–629.

KLEIN (F.)

[Ges. math. Abh.] Gesammelte mathematische Abhandlungen (Fricke and others,

eds.), 3 vols., Berlin: Springer, 1921–1923.

[1882]

¨

Uber Riemanns Theorie der algebraischen Funktionen und ihrer Integrale,

Leipzig: Teubner, 1882; Ges. math. Abh. 3, pp. 499–573. Engl. transl. 1893;

repr., Dover, 1963.

[1894] Riemann und seine Bedeutung f¨ ur die Entwicklung der modernen Math-

ematik, Jber. d. Dt. Math.-Verein., 4 (1894–1895), pp. 71-87; Ges. math.

Abh. 3, pp. 482–497.

36 U. BOTTAZZINI AND R. TAZZIOLI

[1926] Vorlesungen ¨ uber die Entwicklung der Mathematik im 19. Jahrhundert

(R. Courant and O. Neugebauer eds.), t. I, Berlin: Springer, 1926. Engl.

transl., Brooklin: Math. Sci. Press, 1979.

[Nachlass] Nieders¨achsischen Staat-und Universit¨atsbibliothek G¨ottingen, Hand-

schriftenabteilung, Prinzenstr. 1, D-3400 G¨ottingen.

KOHLRAUSCH (R.)

[1853] Das Sinuselektrometer, Ann. Phys. Chem., 88 (1853), p. 497.

[1854] Theorie des elektrischen R¨ uckstandes in der Leidener Flasche, Ibid., 91

(1854), pp. 56–82.

LAM

´

E (G.)

[1859] Le¸cons sur les coordonn´ees curvilignes et leurs applications, Paris: Mallet-

Bachelier, 1859.

LANDAU (L.D.) and LIFSHITZ (E.M.)

[1951] The classical theory of ﬁelds, Oxford: Pergamon Press, 1951, repr. 1979.

[1959] Th´eorie de l’´elasticit´e, Oxford: Pergamon Press, 1959.

LAPLACE (P.S.)

[Œuvres] Œuvres compl`etes, 14 vols., Paris: Gauthier-Villars, 1878–1912.

LOBACHEVSKY (N.)

[1837] G´eom´etrie imaginaire, J. reine angew. Math., 17 (1837), pp. 295–320.

MAXWELL (J.C.)

[1856] On Faraday’s lines of force, Transactions of the Cambridge philosophical

Society, 10 (1856), p. 27.

[1873] A treatise on electricity and magnetism, 2 vols., Oxford: Clarendon Press,

1873; 3rd ed. 1891; repr., New York: Dover, 1954.

NEWTON (I.)

[1726] Philosophiae Naturalis Principia Mathematica, 3rd ed., London 1726;

Engl. transl. by A. Motte revised by F. Cajori, Berkeley and Los Ange-

les: University of California Press, 1934; 5th ed., 1962.

NOETHER (M.)

[1882] Zu F. Klein’s Schrift ‘Ueber Riemann’s Theorie der algebraischen Functio-

nen’, Zeitschrift f¨ ur die Mathematik und Physik 27 (1882), pp. 201–206.

PORTNOY (E.)

[1982] Riemann’s contribution to diﬀerential geometry, Hist. math., 9 (1982),

pp. 1–18.

REIFF (R.) and SOMMERFELD (A.)

[1902] Elektrizit¨at und Optik, Encyklop¨adie der mathematischen Wissenschaften,

Physik, V, 2, Leipzig: Teubner, pp. 3–62.

RIEMANN (B.)

[Werke] Gesammelte mathematische Werke und wissenschaftlicher Nachlass

(H. Weber, ed.), Leipzig: Teubner, 1876; 2nd ed. 1892; repr. New York:

Springer, 1991. French edition, Paris 1898, repr. Blanchard, 1968. (All

references refer to the 1991 edition.)

[1851] Grundlagen f¨ ur eine allgemeine Theorie der Funktionen einer ver¨anderlichen

complexen Gr¨osse, Inauguraldissertation, G¨ottingen, 1851; Werke, pp. 3–48.

[1853] Neue mathematische Principien der Naturphilosophie, Werke, pp. 528–532.

[1854a]

¨

Uber die Darstellbarkeit einer Function durch eine trigonometrische Reihe,

Abhandlungen der K¨oniglichen Gesellschaft der Wissenschaften zu G¨ottin-

gen, 13 (1868); Werke, pp. 227–271.

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 37

[1854b]

¨

Uber die Hypothesen, welche der Geometrie zu Grunde liegen, Ibid.,

13 (1868); Werke, pp. 272–287. English transl. in [Spivak 1970/1979,

pp. 135–153].

[1854c]

¨

Uber die Gesetze der Vertheilung von Spannungselectricit¨at in ponder-

ablen K¨orpern, wenn diese nicht als vollkommene Leiter oder Nichtleiter,

sondern als dem Enthalten von Spannungselectricit¨at mit endlicher Kraft

widerstrebend betrachtet werden, Amtlicher Bericht ¨ uber die 31. Versamm-

lung Deutscher Naturforscher und Aerzte zu G¨ottingen im September 1854

(1854); Werke, pp. 49–54.

[1855] Zur Theorie der Nobili’schen Farbenringe, Ann. Phys. Chem., 95 (1855);

Werke, pp. 55–66.

[1857a] Beitr¨age zur Theorie der durch die Gauss’sche Reihe F(α, β, γ, x) darstell-

baren Functionen, Abh. K. Ges. Wiss. G¨ottingen, 7 (1857); Werke, pp. 67–83.

[1857b] Theorie der Abelschen Funktionen, J. reine angew. Math., 54 (1857); Werke,

pp. 88–144.

[1858] Ein Beitrag zur Elektrodynamic, Ann. Phys. Chem., 131 (1858); Werke,

pp. 288-293.

[1861] Commentatio mathematica, qua respondere tentatur quaestioni ab Ill

ma

Academia Parisiensi propositæ: ‘ Trouver quel doit ˆetre l’´etat caloriﬁque

d’un corps solide homog`ene ind´eﬁni pour qu’un syst`eme de courbes isother-

mes, `a un instant donn´e, restent isothermes apr`es un temps quelconque, de

telle sorte que la temp´erature d’un point puisse s’exprimer en fonction du

temps et de deux autres variables ind´ependantes’, Werke, pp. 391–404.

[1869] Partielle Diﬀerentialgleichungen und deren Anwendung auf physikalische

Fragen (K. Hattendorﬀ, ed.), Braunschweig: F. Vieweg, 1869.

[1876a] Fragmente philosophischen Inhalts, Werke, pp. 507–538.

[1876b]

¨

Uber das Potential eines Ringes, Werke, pp. 431–436.

[1876c] Gleichgewicht der Electricit¨at auf Cylindern mit kreisf¨ormigem Querschnitt

und parallelen Axen, Werke, pp. 440–444.

[1876d] Neue Theorie des R¨ uckstandes in electrischen Bindungsapparaten, Werke,

pp. 367–378.

[1876e] Schwere, Elektricit¨at, Magnetismus. Nach den Vorlesungen von B. Rie-

mann (K. Hattendorﬀ, ed.), Hannover: C. R¨ umpler, 1876.

[1902] Nachtr¨age (M. Noether and W. Wirtinger, eds.), 1902; Werke, viii + 116 p.

(separate numeration following p. 558).

ROSENFELD (L.)

[Papers] Selected papers of L´eon Rosenfeld (R.S. Cohen and J.J. Stachel, eds.),

Boston Studies in the Philosophy of Science 21, Dordrecht/Boston: Reidel

Publishing Company, 1979.

[1956] The velocity of light and the evolution of electrodynamics, Suppl. Nuovo

Cimento 4 (1956), pp. 1630–1669; Papers, pp. 134–177.

SCHAFFNER (K.F.)

[1972] Nineteenth-century aether theories, Oxford: Pergamon Press, 1972.

SCHERING (E.)

[Werke] Gesammelte mathematische Werke (R. Haussner and K. Schering, eds.),

2 vols., Berlin: Mayer and M¨ uller, 1902–1909.

[1866] Zum Ged¨achtniss an B. Riemann, Werke 2, pp. 367–383; repr. in [Riemann,

Werke 1991, pp. 828–847].

38 U. BOTTAZZINI AND R. TAZZIOLI

SCHOLZ (E.)

[1982] Riemanns fr¨ uhe Notizen zur Mannigfaltigkeitsbegriﬀ und zu den Grundlagen

der Geometrie, Archive for history of exact sciences, 27 (1982), pp. 213–232.

SPEISER (A.)

[1927] Naturphilosophische Untersuchungen von Euler und Riemann, J. reine

angew. Math., 157 (1927), pp. 105–114.

SPIVAK (M.)

[1970] A comprehensive introduction to diﬀerential geometry, vol. II, 1st ed. 1970;

2nd ed., Houston: Publish or Perish, 1979.

WEBER (W.)

[1846] Elektrodynamische Maassbestimmungen, Abhandlungen bei Begr¨ undung

der K¨oniglischen S¨achsichen Gesellschaft der Wissenschaften am Tage des

200 j¨ahrigen Geburtstages Leibnizens (1846), pp. 209–378.

WEYL (H.)

[1913] Die Idee der Riemannschen Fl¨ache, Leipzig: Teubner, 1913; repr. 1964.

[1919a] Raum Zeit Materie, Berlin: Springer, 1919. Engl. transl. 1922; repr. Dover,

1952. French transl. 1922, repr. Blanchard 1958.

[1919b] Bernhard Riemann,

¨

Uber die Hypothesen, welche der Geometrie zu Grunde

liegen (H. Weyl ed.), Berlin, 1919. Weyl’s Vorwort and Erl¨auterungen are

reprinted in Riemann [Werke, 1991, pp. 740–768].

[1923] Mathematische Analyse des Raumproblems, Berlin: Springer, 1923.

WHITTAKER (E.)

[1951] A history of the theories of aether and electricity, vol. I, New York: Harper,

1951; repr. Dover 1991.

WISE (N.M.)

[1981] German concepts of force, energy, and the electromagnetic ether: 1845–1880,

Conception of ether. Studies in the history of ether theory, 1740–1900

(G.N. Cantor and M.J. Hodge, eds.), Cambridge: Cambridge University

Press, 1981, pp. 269–307.

4

U. BOTTAZZINI AND R. TAZZIOLI

As Klein once said, Riemann’s work was characterized by his continual attempt to put “in mathematical form a uniﬁed formulation of the laws which lie at the basis of all natural phenomena” [1894, p. 484]. Klein did not hesitate to claim that “the origins of Riemann’s pure mathematical developments” lay in this research which, in Riemann’s words, was at a certain stage his own “main work ”. Searching for a mathematical description of the known physical phenomena, Riemann thought of space as pervaded by substance (Stoﬀ 2 ), and in a section of his Fragmente he considered the state of a single particle of substance and analysed locally the space around it [Riemann 1853]. This passage from “local” to “global” constitutes the basic method used by Riemann in some of his most important works in geometry as well as in analysis and physics. In analytical terms this corresponds to the analytical continuation of a complex function. This is “a well known theorem” [Riemann 1857b, p. 88] which is at the basis of the “new method ” he set up in his thesis. This “method ” [Riemann 1851, p. 37–39] could be applied to Abelian functions, as he did in [1857b], and also “in its essential lines” to “every function which satisﬁes a linear diﬀerential equation with algebraic coeﬃcients” [1857a, p. 67]. Accordingly, in this paper he studied the transcendental functions deﬁned by the hypergeometric diﬀerential equation “almost without calculations” [Werke, p. 85] and “and in their totality” on the complex sphere. The same point of view inspired his Habilitationsvortrag where he deﬁned metrics on manifolds by using the linear element ds. In particular, Riemann stated that “questions about the immeasurably large are idle questions for the explanation of Nature. But the situation is quite diﬀerent with questions about the immeasurably small ” [Riemann 1854b/1979, p. 151]. As Riemann explained in the introduction to the ﬁrst course he gave in G¨ttingen as a Privatdozent, the laws for all space could o be deduced by integrating partial diﬀerential equations expressing some “elementary” principles valid for inﬁnitely small portions of space. 3

2 Instead of this, in his later lectures on gravitation, electricity and magnetism Riemann preferred to use the term ether. 3 “Wahre Elementargesetze k¨nnen nur im Unendlichkleinen, nur f¨ r Raum — o u und Zeitpunkte stattﬁnden. Solche Gesetze aber werden im Allgemeinen partielle Diﬀerentialgleichungen sein, und die Ableitung der Gesetze f¨r ausgedehnte K¨rper u o und Zeitr¨ume aus ihnen erfordert die Integration derselben. Es sind also Methode a

NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS

5

Such a research method had already been announced by Riemann himself in a rather cryptic way in 1850. When lecturing at the P¨dagogische a Seminar he noticed that it was possible to formulate a mathematical theory by moving from elementary principles toward general laws valid in all of a given continuous space without distinguishing between gravity, electricity, magnetism and equilibrium of heat. 4 As Klein pointed out, the method of studying functions on the basis of their behaviour in the inﬁnitely small had a physical counterpart in the concept of a line of force. Moreover, Klein suggested a kind of dualism between Riemann’s mathematical thought and Faraday’s concept of action by contact, writing that: “If I may dare to proceed with so forceful the analogy, then I shall say that Riemann in the ﬁeld of mathematics and Faraday in the ﬁeld of physics are parallel ” [Klein 1894, p. 484]. Supporting Klein’s point of view, in Raum Zeit Materie Weyl stated that the passage from Euclidean to Riemannian geometry “is founded in principle on the same idea as that which led from physics based on action at a distance to physics based on inﬁnitely near action” [1919a/1922, p. 91]. In fact, according to Weyl: “The principle of gaining knowledge of the external world from the behaviour of its inﬁnitesimal parts is the mainspring of the theory of knowledge in inﬁnitesimal physics as in Riemann’s geometry, and, indeed, the mainspring of all the eminent work of Riemann, in particular, that dealing with the theory of complex functions” [1919a/1922, p. 92].

1. ON THE SOURCE OF RIEMANN’S ANALYTICAL WORK

Riemann introduced his ideas on complex function theory in his 1851 paper which concluded his studies at G¨ttingen. Riemann’s starting point o

n¨thig, durch welche man aus den Gesetzen im Unendlichkleinen diese Gesetze im o Endlichen ableitet, und zwar in aller Strenge ableitet, ohne sich Vernachl¨ssigungen zu a erlauben. Denn nur dann kann man sie an der Erfahrung pr¨fen” [Riemann 1869, p. 4]. u

4 “So z.B. l¨sst sich eine vollkommen in sich abgeschlossene mathematische Theorie a zusammenstellen, welche von den f¨r die einzelnen Punkte geltenden Elementaru gesetzen bis zu den Vorg¨ngen in dem uns wirklich gegebenen continuirlich erf¨llten a u Raume fortschreitet, ohne zu scheiden, ob es sich um die Schwerkraft, oder die Electricit¨t, oder den Magnetismus, oder das Gleichgewicht der W¨rme handelt” a a (in [Dedekind 1876, p. 545]).

Indeed. Dirichlet. BOTTAZZINI AND R. und dass man in vielen F¨llen richtige Resultate a erh¨lt. It is a well known fact that Riemann’s complex function theory is deeply connected with potential theory in two dimensions — a theory he was well acquainted with. this reduced the problem to minimizing the following integral J= V 5 “Nach einer Mittheilung. mit divergenten Reihen operiert. From (1. es f¨llt also diese Idee wohl noch in die Studienjahre Riemanns. dass dies die partielle Diﬀerentialgleichung thue. dass Beziehungen a o zwischen Functionen. as Euler repeatedly did.1) he deduced the equations ∆u = 0. According to him. die mir Riemann in Fr¨hjahre 1865 w¨hrend meines u a Pisaner Aufenthalts machte. auch wenn man uber die Convergenzgebiete der ¨ darstellenden Reihen hinausging. This letter from February 6. p. | grad u|2 dv.1) explain why correct results can be obtained even when working with divergent series. As Prym was to write to Klein after Riemann’s death. ∆v = 0 which are the basis for investigating the properties of the functions u and v [Riemann 1851. die durch Entwicklung der betreﬀenden Functionen in Reihen erhalten worden. mit dem er den Gegenstand besprach. a Er frug sich dann. 383]. in particular. From a mathematical point of view. wenn man.1) ∂u ∂v . as a student Riemann had followed Weber’s lectures in 1849 and the following year he participated in the physics seminar jointly founded and led by Gauss and Weber. was denn eigentlich die Function aus dem einen Gebiete in das andere fortgesetzt. stimmte dieser Ansicht vollst¨ndig bei. 5 since his student days Riemann had attributed great importance to equations (1. Gauss himself had developed the theory of the Laplace equation in a paper of 1839. TAZZIOLI was given by the equations (1. und gelangte zu der Einsicht. es wiederholt getan.6 U. ∂u =− ∂y ∂x which have to be satisﬁed by the function w = u + iv of a variable z = x + iy. = ∂x ∂y ∂v .1) for the continuation of a function from one complex domain to another.B. he had studied the problem of the distribution of masses or electric charges on a closed surface S. ist derselbe zu einer Theorie der Functionen einer ver¨ndlichen complexen Gr¨sse durch die Beobachtung gekommen. bestehen bleiben. equations (1. vor a a die Auﬀassung seiner Inauguraldissertation”. He had determined the potential function in diﬀerent cases and. . wie Euler z. assuming the potential to be constant on S. 7]. 1882 is kept in Klein’s Nachlass [11.

x]. he had introduced in order to study multi-valued functions such as algebraic functions and their integrals. “a [homogeneous] distribution must necessarily exist. 34–35]. As Riemann told Betti (see [Bottazzini 1985. Many of the ideas of this paper. This argument was used systematically by Dirichlet in his lectures on the forces which are inversely proportional to the square of distance. In this way. which satisﬁes the Laplace equation within the domain and has given values on the boundary. Riemann required that the surface associated to a function be composed of as many sheets as were the branches of the function. According to Riemann. This is the global theorem which. potential theory as developed by Gauss and Dirichlet was well suited to a particular geometrical object. 127] faced the problem of proving that a function with continuous ﬁrst partial derivatives on a given bounded domain. but that. 559]). Riemann made the surface simply connected with suitable transversal cuts (Querschnitte) and analysed the behaviour of the function in the neighbourhood of the singularities — poles and branch points. Riemann stated and proved a fundamental existence theorem for a function with given singularities and boundary conditions [1851. the “Riemann surface”. p. p. the idea of transversal cut on a surface struck him after a long discussion . on the contrary. complex functions of positions can be studied on arbitrarily given curved surfaces in exactly the same way as on the surfaces over the plane” [Klein 1882/1893. Then. Klein reported that Prym “told me that Riemann’s surfaces originally are not necessarily many-sheeted surfaces over the plane. Dirichlet [1876. always exist. in Riemann’s words. Dirichlet’s existence proof of the solution of the “Dirichlet problem” was based on the fact that the minimum for the integral J existed (“Dirichlet principle”). connected in such a way to preserve continuity and to yield a single-valued function on the surface. 233]. Referring to a conversation he had with Prym in 1874.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 7 Since J > 0. p. p. so that the integral J has a minimum”. p. were inspired by physical topics. thanks to the Dirichlet principle. he attained an abstract conception of the space of complex variables by means of a geometrical formulation which his contemporaries were to ﬁnd very hard to understand. Gauss wrote [1839. 35]. p. as Klein ﬁrst emphasized. opens the way to the the study of complex functions independently of their analytical expressions [Riemann 1851.

Riemann stated that solving this problem was equivalent to ﬁnding a function u having given boundary values and satisfying the diﬀerential equation ∆u = 0 within a surface S. oder doch ihren Gedankengang als den Ausgangspunkt f¨r Riemann’s u Arbeiten uber Functionentheorie zu bezeichnen” [Brill and Noether 1894. Thus. Thus. “one can easily” show that ζ takes every real value exactly once on each curve of the boundary whereas within S it takes every complex value n times (with Im ζ > 0). due to constant forces along the directrices of the cylinder. the problem reduces to the determination of 6 As Weber remarked. Under the condition that ζ goes from −∞ to +∞ when any of the boundary curves is travelled in a positive sense. 6 Apparently Riemann sought to evaluate the distribution of the static electricity on the surface of a cylinder. apart from some hints by Riemann.8 U. p. 259]. To this end. In Brill and Noether’s opinion. As within T there are (2n − 2) branch points [Riemann 1857b. one gets a conformal map of S onto a n-sheeted surface T over the upper half plane. p. Laplace had already observed that. which he supposed to be plane. simply connected. In turn. Accordingly. this problem could be reduced to the easier one of determining a function ζ = ξ + iη of the complex variable z = x + iy which is ﬁnite and continuous within S and takes real values on the boundary. ¨ . BOTTAZZINI AND R. it is suﬃcient to replace every directrix by its intersection point with a plane orthogonal to the directrix. when the interacting masses are placed on an inﬁnite cylinder and the forces are constant along straight lines parallel to the directrices of the cylinder. TAZZIOLI with Gauss on a mathematical-physical problem. die[se] Note als eine der fr¨hesten Arbeiten u Riemann’s. 113]. 440]. the origin of the concepts of Riemann surface and transversal cut could be found in an unpublished note [Riemann 1876c] on a problem of electrostatic or thermal equilibrium on the surface of a cylinder with transversal cuts. the evaluation of the potential of the bodies can be seen as a plane problem. the diﬀerential equation involved reduces to the Laplace equation ∆u = 0 in two variables and the sought for potential function m/r reduces to the “logarithmic” potential m log(r). whose boundary lines on the n sheets of the surface coincide with the real axis. this note reduces to sheets with calculations. p. simple sheeted and bounded by n arbitrarily given curves [Riemann 1876c. where it becomes inﬁnite of the ﬁrst order in just one point for each curve of the boundary. “Wir sind geneigt.

22]. As Brill and Noether [1894. closed Riemann surface over the complex plane be a uniform conductor with the two poles of a galvanic battery at the points A1 and A2 . Riemann’s “general problem” (allgemeine Fragestellung) was the following: “to study the streamings in the ﬁrst place and thence to work out the theory of certain analytical functions” [Klein 1882/1893. 113. Klein cono cluded that “anyone who clearly understands” this “will. By means of both the Green function and the Riemann-Schwarz principle of symmetry. In Klein’s view. which was to be developed by Riemann [1857b]. it satisﬁes the equation ∆u = 0. and then. in order to give what was physically evident the support of mathematical reasoning. 258] remarked. on these assumptions a current is created. 119] and determining the Green function by means of an Abelian integral of the third kind. p. p. and in A1 and A2 becomes inﬁnite when r1 and r2 go to inﬁnity as log r1 and − log r2 respectively [Klein 1926. pp. in this way Riemann established a close connection between the theory of Abelian integrals. I think. such that its real part u is continuous within T and takes arbitrarily given values on the n lines of the boundary. lay in Riemann’s researches in conductors and galvanic currents. Riemann’s fundamental existence theorem of a harmonic function u could be obtained from the following “thought experiment ” (Gedankenexperiment): let a n-sheeted. Klein added that his presentation . share my opinion”. p. Next he discussed the particular case in which the boundary of S is given by n circles. he afterwards substituted Dirichlet’s Principle” [Klein 1882/1893. 260–261]. the conformal mapping problem and the fundamental existence theorem Riemann stated in his 1851 paper. On the contrary. In Klein’s opinion. p. Riemann showed how the problem could be solved by following the same procedure he used in his paper on Abelian functions [Riemann 1857b. Klein took this as the starting-point of his presentation of Riemann’s theory of algebraic functions and their integrals. claiming that: “I have no doubt that he [Riemann] started from precisely those physical problems. After referring to “the conditions under which Riemann worked in G¨ttingen” as well as to his writings on Naturphilosophie. whose potential u is deﬁned and single-valued on the surface. x]. Admittedly.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 9 a function of ζ branched like T . according to Klein the motivation for Abelian function theory.

Helmholtz ” [Klein Nachlass 8. BOTTAZZINI AND R. Riemann never made any connection between electric currents and closed surfaces “free in space”. Ich meine. wenn auch nicht in Ihrer Abhandlung. 86]. u In his reply Bianchi reported that. 116–117]. dass ¨hnliche Ideen. In spite of Klein’s hope. aber erst nachdem die Theorie der Abelschen Functionen vollendet war. 1882. 100]. a former student of Betti. This was also rejected by Prym 7 and Betti did the same in a letter to Klein from March 22. who had spent a period in M¨ nchen with Klein. his interpretation of the origins of Riemann’s theory met strong criticism. der umgekehrte Weg. Sie dagegen scheinen der Ansicht zu sein. Instead.” 8 French revised [Eds. Answering a precise question asked by Klein. “[Betti] can ensure you with all certainty that once Riemann told him that he [Riemann] had been led to his analytical theory and his way of thinking [Anschauungsweise] after dealing with questions and problems of physics” [Klein Nachlass 8. auf beliebig beschaﬀene Fl¨chen bez¨glichen Falle a u zu dem Falle der Abelschen Functionen als einem speciellen Hinabgestiegen sei. pp. . 7 Contrary to Klein’s opinion. dass Riemann von dem allgemeinen. in the letter mentioned above (see footnote 5) Prym stated that: “Ich halte es daher auch f¨r sehr wahrscheinlich. even in the case of algebraic functions and their integrals. TAZZIOLI “by no means include[d ] the whole of what [Riemann] intended in the theory of functions” and recognized that. wie u a Sie sie entwickeln. Noether [1882] openly questioned Klein’s historical reconstruction. den ja auch Sie. he stated that: “Riemann ne m’a jamais dit avoir d´velopp´ la th´orie des courants e e e stationnaires dans un ﬂuide incompressible. M. his point of view was “necessarily very subjective”.10 U. sein der nat¨rliche und entspreche am u meisten den Gesetzen der geschichtlichen Entwicklung. according to Betti’s record. so doch in Ihren Studien eingeschlagen haben.]. dans un espace quelconque ` trois dimensions et ` courbure quelconque. In his review of Klein’s booklet. 8 Some days later Klein addressed Bianchi [Opere 11. von Riemann verfolgt worden sind. quoiqu’il se soit entretenu a a avec moi plusieurs fois sur les travaux de Mr.

deren Resultate in seinen Promotions — und Habilitationsthesen (vom 22. He imagined a continuous ﬂow of representations going from the ego to the conscious and back and studied the connections between diﬀerent representations in mechanical terms as compositions of forces. Euler and. Riemann’s more coherent attempt to give a systematic presentation of his ideas on the propagation of physical phenomena like gravitation Riemann wrote in fact: “Meine Hauptarbeit betriﬀt eine neue Auﬀassung der bekannten Naturgesetze . RIEMANN’S NEUE MATHEMATISCHE PRINCIPIEN DER NATURPHILOSOPHIE After the completion of his thesis. Euler’s a und — anderseits — Herbart’s gef¨hrt” [1876a . 507–508]. ¨ a Licht. on the other side. Ich wurde dazu haupts¨chlich durch das Studium der Werke Newton’s. light. wodurch eine Verschiedenheit in Bezug auf seine Naturphilosophie und diejenigen S¨tze der Psychologie. welche deren Verbindung mit der Naturphilosophie a betreﬀen.Ausdruck derselben mittelst anderer Grundbegriﬀe — wodurch die Benutzung der experimentellen Data uber die Wechselwirkung zwischen W¨rme. u 10 “Was [Herbart] betriﬀt — Riemann wrote — so konnte ich mich den fr¨hesten u Untersuchungen Herbart’s. I was led to this primarily through the study of the works of Newton. Herbart’s psychology 10 inspired both Riemann’s model of the substance (or ether) and his principles of Naturphilosophie. magnetism. 9 . Magnetismus und Electricit¨t zur Erforschung ihres Zusammenhangs m¨glich a o wurde. Herbart ”. 23. p. Riemann summarized his philosophical views by saying: “Der Verfasser ist Herbartianer in Psychologie und Erkenntnisstheorie (Methodologie und Eidologie). u. 508]. Herbart’s Naturphilosophie und den darauf bez¨glichen metaphysischen Disciplinen (Ontologie und Synechologie) kann er meisu tens nicht sich anschliessen” [1876a . Riemann himself gave a hint of his research projects by writing in an undated note that his “main work ” involved “a new interpretation of the known laws of nature — whereby the use of experimental data concerning the interaction between heat. Herbart had deﬁned the “psychic act ” (or representation) as an act of self-preservation with which the ego opposed the perturbations coming from the external world. 507]. it is worth noting here that Riemann could agree “almost completely” with the results of Herbart’s early research while he rejected his philosophy “at an essential point ” which involved Naturphilosophie. fast v¨llig anschliessen. o musste aber von dem sp¨teren Gange seiner Speculation in einem wesentlichen Punkte a abweichen. 9 As for the latter.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 11 2. In particular. Oktober 1802) ausgesprochen sind. and electricity would make possible an investigation of their interrelationship. p. bedingt ist” [1876a . pp.

lies in the internal state of the bodies [1853. 11 o On the basis of this rather obscure idea Riemann tried to build a mathematical model of the space surrounding two interacting particles of substance. the two homogeneous diﬀerential forms ds2 = dx2 + dx2 + dx2 1 2 3 and ds 2 = dx12 + dx22 + dx32 could be expressed in the following way ds 2 = G2 ds2 + G2 ds2 + G2 ds2 . x3 ) at the time t . x2 . p. 1 1 2 2 3 3 ds2 = ds2 + ds2 + ds2 . p. Then. TAZZIOLI and light was made in March 1853 in a paper which. p. as he wrote to his brother Wilhelm in December 1853 (see [Dedekind 1876. λ2 . x2 . 529]. sondern jeder einzelnen in uns gebildeten Vorstellung Substantialit¨t zukomme” [1853. he did not hesitate to entitle Neue mathematische Principien der Naturphilosophie (New mathematical principles of natural philosophy). In his paper Riemann began by claiming that the basis of the general laws of motion for ponderable bodies. p. which are posed at the beginning of Newton’s Principia. BOTTAZZINI AND R. x2 . x3 are functions of x1 . λ3 . He introduced a cartesian coordinate system and considered a single particle of substance as concentrated at the point O(x1 . Es ist die Consequenz der auf Herbart’schem Boden stehenden Psychologie. 547]). pp. 1 2 3 where ds1 . x3 . 530]. a 12 The conditions according to which the equation ds 2 = ds2 is satisﬁed are the Lam´ e equations. ds3 was an appropriate new basis [Riemann 1853. Riemann himself regarded his writings on Naturphilosophie as fundamental and intended to publish them. where x1 . In a footnote Riemann added that: “In jedes ponderable Atom tritt in jedem Augenblick eine bestimmte. G2 − 1. Indeed. dass nicht der Seele. Led by the analogy with Herbart’s psychology Riemann made the hypothesis that the universe (Weltraum) was ﬁlled with a substance (Stoﬀ ) ﬂowing continually through atoms and there disappearing from the material world (K¨rperwelt ) [1853. x3 ) at the time t and at the point O (x1 . “according to a well known theorem”. e 11 . ds2 . G3 − 1 the main dilatations (Hauptdilatationen) of the particle at O and denoted them by λ1 . x2 . 101]. 12 Riemann called the quantities G1 − 1. 529]. der Gravitationskraft proportionale Stoﬀmenge ein und verschwindet dort.12 U. p. They were ﬁrst published by Lam´ [1859. 528]. echoing Newton’s Principia. 99.

13 If xi = xi + ui (where ui is the displacement of P due to the deformation) one has 3 dxi = dxi + dui = dxi + k=1 ∂ui dxk . x3 + dx3 ) respectively. 1 1 2 2 3 3 13 See for instance Brillouin [1938. eik = ∂ui ∂uk + + ∂xk ∂xi 3 =1 ∂u ∂u · ∂xi ∂xk This quantity deﬁnes the deformation and coincides with the strain tensor in elasticity theory. Riemann compared ds 2 and ds2 by a direct calculation on the same basis ds1 . . with respect to which both ds 2 and ds2 were orthogonal. Under these assumptions. ∂xk Then 3 ds 2 = ds2 + k. x3 ) and Q(x1 + dx1 . If the deformation is supposed inﬁnitely small. X]. ch. ds3 .i=1 ∂u ∂u dxi dxk . it is ∂xk ∂xi possible to ﬁnd a basis such that eik = 0 if i = k. Supposing. ds2 . as Riemann did. that the ether is an elastic.i=1 eik dxi dxk . He found that (2. Since it is symmetric. ∂ui ∂uk the strain tensor becomes eik = + .NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 13 Riemann’s result can be interpreted in terms of the classical theory of elasticity. x2 . isotropic medium. x2 .2) δ(ds2 ) = ds 2 − ds2 = (G2 − 1) ds2 + (G2 − 1) ds2 + (G2 − 1) ds2 . homogeneous. x3 ) and Q (x1 + dx1 . x3 + dx3 ) to P (x1 . it is possible to compare the distance ds between P and Q with the corresponding distance ds after the deformation. x2 + dx2 . then one can consider an elastic deformation changing P (x1 . .1) where δ(ds2 ) = k. ∂xi ∂xk Now it is possible to calculate the variation δ(ds2 ) = ds 2 − ds2 3 (2. x2 + dx2 . =1 ∂u dxk dx + ∂xk 3 k.

x2 . 2. Since the linear extensions after the deformation with respect to the axes √ x1 . λ3 .14 U. ∆ 3 = G3 − 1. λ3 . Riemann then supposed that the variation δ(ds2 ) produced a force able to modify the particle in such a way that the particle itself. would propagate the physical forces through the space. One could observe that if the Hauptdilatationen λ1 . λ2 . the distance between two inﬁnitely close points of ether remains constant after the deformation 14 See footnote 15 below. which therefore can be studied in the conceptual framework given by the theory of elasticity. 3 eik = 0 if i = k. λ2 . By assuming the homogeneity of the substance. the moment of force (Kraftmoment ) was given by 1 δ a(λ + λ + λ )2 + b(λ2 + λ2 + λ2 ) 1 2 3 1 2 3 2 where a. He formulated the following “law of eﬀects” (Wirkungsgesetz): if dV and dV represent the volumes of the particle at time t and t respectively. whose development follows diﬀerent laws. Riemann introduced the hypothesis that the dilatations were inﬁnitely small. p. λ3 . the two actions produce diﬀerent eﬀects. b were constants. 1 e22 = G2 − 1. opposing this deformation. then ∆ 1 = G1 − 1. x3 are deﬁned as ∆ i = 1 + eii − 1. 2 e33 = G2 − 1. Riemann regarded this moment as a resultant of the forces which tend to lengthen or shorten the line elements ending at O [1853. then the force arising from two diﬀerent states (Stoﬀzust¨nde) of the substance a in diﬀerent times t. . since the two addenda in (2.3) are independent of each other. ∆ 2 = G2 − 1. and so the produced forces were linear functions of λ1 . λ2 . BOTTAZZINI AND R. of the particle of ether were equal to zero. They coincide with Riemann’s Hauptdilatationen λ1 . TAZZIOLI In this case for the components of the strain tensor one has e11 = G2 − 1. 3. 531]. 14 In order to describe the reaction of the particle. t can be described in the following way (2.3) a ds − ds dV − dV +b · dV ds Moreover. for i = 1.

light. They could be explained by assuming that every particle of the homogeneous substance ﬁlling space has a direct eﬀect only on its neighbourhood. a 16 “Beide Classen von Erscheinungen lassen sich erkl¨ren. Das mathematische Gesetz. und a 2) in den Widerstand. 532]. 529]. p. welche u wir als Licht und W¨rme wahrnehmen” [Riemann 1853. 16 “Es kann also die Wirkung der allgemeinen Gravitation auf ein ponderables Atom durch den Druck des raumerf¨llenden Stoﬀes in der unmittelbaren Umgebung desselben u ausgedr¨ckt und von demselben abh¨ngig gedacht werden. dass der raumerf¨llende Stoﬀ die Schwingungen fortpﬂanzen muss. heat and light propagate through space? This was the question Riemann left open [1853. mit welchem ein Stoﬀtheilchen einer Volum¨nderung. ds But how can we choose the functions ψ and ϕ in order that gravity. auf dem zweiten die Fortpﬂanzung des Lichts und der W¨rme und die eleca trodynamische oder magnetische Anziehung und Abstossung” [Riemann 1853. Aus unserer Hypothese folgt u a nothwendig. heat propagation. 15 The mathematical law (2. can be divided into: “1) the resistance with which a particle opposes a change of its volume. kann zerf¨llt gedacht wera den 1) in den Widerstand. In this connection he limited himself to state that “the eﬀects of ponderable matter on ponderable matter were attractive and repulsive forces inversely proportional to the square of the distance” or “light and radiant heat ”. and the Kraftmoment given by (2. electrodynamic and magnetic attraction and repulsion on the second part ”. a Auf dem ersten Theil beruht die Gravitation und die electrostatische Anziehung und Abstossung. 15 . p. wenn man annimt. 532]. Riemann calculated the variation of ds during the time dt and found that the quantity δ ds/dt was equal to t −∞ dV − dV ψ(t − t ) δt + dV t −∞ ds − ds ϕ(t − t ) δt . and 2) the resistance with which a physical line element opposes a change of length.3) vanishes.3) according to which this happens. und jedes Stoﬀtheilchen u unmittelbar nur auf seine Umgebung einwirkt. nach welchem dies geschieht. p. Gravity and electric attraction and repulsion are founded on the ﬁrst part. mit welchem ein physisches Linienelement einer L¨ngena ¨nderung widerstrebt. In this case no physical phenomenon could occur in the space surrounding the particle. dass a den ganzen unendlichen Raum ein gleichartiger Stoﬀ erf¨llt.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 15 at issue.

In virtue of this cause the point was able to interact with other ponderable bodies by means of a force. . He assumed that this cause should be sought in “the form of motion of a substance spread continuously through the entire inﬁnite space. 17 By means of this assumption. 18 “Nach dieser Annahme m¨ssen alle von ponderablen K¨rpern durch den leeren u o Raum auf ponderable K¨rper ausge¨bte Wirkungen durch diesen Stoﬀ fortgepﬂanzt o u werden. Diese beiden Erscheinungen. . which was inversely proportional to the square of their distance. suche ich in der Beweo gungsform eines durch den ganzen unendlichen Raum stetig verbreiteten Stoﬀ. 533]. Riemann observed that. TAZZIOLI In Gravitation und Licht which constitutes the last section of his Fragmente on Naturphilosophie [1876a. BOTTAZZINI AND R. it was possible to explain the propagation of phenomena through space and then light and gravitation had to be explained by means of the motion of this substance. contrary to Newton’s own opinion. 534]. Riemann supposed that a determinate cause (bestimmte Ursache) existed at every point of space. welche bloss aus Bewegungen dieses Stoﬀes erkl¨rt a werden m¨ssen” [Riemann 1876a . . In this connection.16 U. Gravitation und Lichtbewegung durch den leeren Raum. . . aber sind die einzigen. 18 The further development he drew from this hypothesis divided in two parts: (a) to ﬁnd the mathematical laws of motion of this substance which can be used for explaining the phenomena. . (b) to clarify the causes (Ursachen) by means of which one could explain these motions. Riemann presented a mathematically more sophisticated attempt at a uniﬁed explanation of both gravitation and light. Dieser Stoﬀ kann also vorgestellt werden als ein physischer Raum. 533]. Riemann was pleased to add in a footnote the celebrated passage of Newton’s third letter to Bentley where Newton stated that believing in action at a distance “without the mediation of anything else” was “so great an absurdity” [1876a. This substance can therefore be conceived as a physical space whose points move in geometrical space”. dessen Punkte sich in dem geometrischen bewegen” [Riemann 1876a . . p. . p. p. pp. . . . 532–538]. his law of attraction had not been thought to need any further explanation for long time. u 17 . In commenting upon the second point. it is worth noting that in the General Scholium “Die nach Gr¨sse und Richtung bestimmte Ursache . which was metaphysical in character. Facing the problem of explaining Newton’s theory of gravitation.

which were responsible for the propagation of gravitation and light respectively. The hypothesis about ether had also been advanced by Euler. are included in the 1838 Brussels edition of a his works. Riemann found for the motion of the 19 See above footnote 9. As N. 2 + ∂x2 + ∂x2 = c2 ∂t2 ∂x1 2 3 where c was the speed of the light. Wise has remarked. Then. To this end he identiﬁed the forces with the motion caused by them and observed that the gravitational force u. Both these papers and the Lettres are likely to have been the source of inspiration to which Riemann referred in his undated manuscript note. 19 From the mathematical point of view. electricity and magnetism. light. Newton himself had spoken about “a certain most subtle spirit which pervades and lies hid in all gross bodies”. p. . Euler’s related papers. 547]. “combining the two motions for gravity and light produced a well-behaved velocity function. which conﬁrmed the possibility of uniting the two processes” [1981. deduced from the potential V by means of the relation u = grad V .NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 17 at the end of the third edition of the Principia. together with his Lettres ` une princesse d’Allemagne. The diﬀerential equations which characterized the propagation of light were a continuity equation div w = 0 and a wave equation for transwerse oscillations in velocity w ∂2w ∂2w ∂2w 1 ∂2w . Speiser deﬁned it [1927. As a consequence of the conditions valid for u and w. Riemann supposed that the real motion of the substance v was composed of two motions u and w. “in his magniﬁcent attempt ” — as A. through which one could explain both the attraction of the “particles of bodies” and the action of the electric bodies and light [1726/1934. he proposed to ﬁnd the “laws of motion of substance in empty space” [1876a. p. 537]. p. had to satisfy both the conditions of being a closed form and the equation ∆V = −4πρ as well [Riemann 1876a. p. 106] — to draw up a uniﬁed theory of gravitation. that is v = u + w. 290]. pp. 534–535].

these equations show how the motion of a single particle of the substance depends only on the motions of the particles around it. According to Riemann. ∂2 ∂2 ∂2 2 + ∂x2 + ∂x2 ∂x1 2 3 ∂2 ∂2 ∂2 + + ∂x2 ∂x2 ∂x2 1 2 3 ∂2 ∂2 ∂2 2 + ∂x2 + ∂x2 ∂x1 2 3 ∂v2 ∂v3 − ∂x3 ∂x2 ∂v3 ∂v1 − ∂x1 ∂x3 ∂v2 ∂v1 − ∂x2 ∂x1 = 0.5) determined the motion of the ether propagating light and gravity through the space.4) 1 2 c 1 2 c 1 c2 ∂2 − ∂t2 ∂2 − ∂t2 ∂2 − ∂t2 div v = 0.5) = 0. 3. BOTTAZZINI AND R. p. 537]. who told him about an “unexplored phenomenon” concerning the electric residuum in Leyden jars. 2.4. TAZZIOLI substance in empty space. Riemann met the experimental physicist Kohlrausch. the solution of the diﬀerential system (2. the following equations (2. while preparing his Habilitationsvortrag on the principles of geometry.5) coincides with the condition that 1 ∂2 − c2 ∂t2 ∂2 ∂2 ∂2 2 + ∂x2 + ∂x2 ∂x1 2 3 (v1 dx1 + v2 dx2 + v3 dx3 ) is an exact diﬀerential dW . Riemann intended to explain 20 “Diese Gleichungen zeigen. Starting from a set of relations which were valid in the inﬁnitely small. (2. = 0. PHYSICAL RESEARCHES In 1854 during his Easter holidays. . v. p. dass die Bewegung eines Stoﬀpunktes nur abh¨ngt von a den Bewegungen in den angrenzenden Raum. 838]. und ihre (vollst¨ndigen) a Ursachen in den Einwirkungen der Umgebung gesucht werden k¨nnen” [Riemann o 1876a . In fact.18 U.und Zeittheilen. by means of his model of the ether Riemann hoped to eliminate from the laws of interaction those “speciﬁcations which refer to action at a distance” as they “always depend on the properties of the surrounding space” [1866/1991. 20 Equation (2.4) proves that the density is not changed during the motion of the substance while the condition expressed by the equations (2. Riemann succeeded in formulating a theory valid in ﬁnite regions. According to his student Schering.

an ingenious instrument invented and built by Kohlrausch himself [1853]. gravity. . . a residual charge appeared. To this end. by means of his research on the connections between electricity.1) has to be proportional to the components of the current intensity vector (ξ. Kohlrausch observed the “reappearing” residuum by means of the Sinuselektrometer. ζ). 22 If u is the potential and ρ the charge density. η. 548–549]). observing that in nature neither complete insulators nor perfect conductors exist. wo ich meine Arbeiten auf eine vorher noch nicht bekannte Erscheinung anwenden konnte. o u pp.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 19 Kohlrausch’s experiment. y. which could be discharged. 22 Laplace in his M´canique c´leste had already associated to each medium a grave e itational coeﬃcient α. The “hidden” u residuum was a quantity of r that could no longer be traced in the jar. Licht und Magnetismus die Erkl¨rung a a davon gefunden. he assigned to every ponderable body a certain conductivity coeﬃcient β. as he wrote to his brother. . ∂u ∂ρ . where G is the constant of universal gravitation and r the distance between m1 and m2 [Œuvres 5. Riemann hoped that his “main work could achieved a favourable reception”. and the charge r that appeared (at least in part) only after the exhaustion of the free charge (r was called das Residuum or der R¨ ckstand). Riemann tried to give an explanation of Kohlrausch’s experiment.1) − ∂ρ . pp. dass die Ver¨ﬀentlichung dieser Arbeit dazu beitragen wird. pp. und ich hoﬀe. lecturing at a meeting of the German “natural philosophers”. 21 . 445–452]. he deduced for the components of the electromotive force at the point (x. Mir ist diese Sache deshalb wichtig. Kohlrausch [1854. − − ∂x ∂x ∂y ∂y ∂z ∂z Quantity (3. while the “reappearing” residuum was the portion of residuum which came back again after the discharge. 61–62] divided the total charge in the jar into two parts: the free charge. and therefore the last expression is equal to “Kohlrausch hatte nun einige Zeit vorher sehr genaue Messungen uber eine bis ¨ dahin unerforschte Erscheinung (den electrischen R¨ckstand in der Leidener Flasche) u gemacht und ver¨ﬀentlicht und ich hatte durch meine allgemeinen Untersuchungen o u ¨ber den Zusammenhang zwischen Electricit¨t. L. In September 1854. light and magnetism. ∂u ∂ρ ∂u − β2 − β2 − β2 . z) at time t the following expressions (3. o meiner gr¨sseren Arbeit eine g¨nstige Aufnahme zu verschaﬀen” (in [Dedekind 1876. then discharged and left isolated for some time. 21 According to Kohlrausch’s experiment. weil es das erste Mal ist. In such a way. in a Leyden jar which had been charged. and had deduced the attraction law between the two masses m1 and m2 : G(m1 m2 /r 2 ) exp(−αr).

BOTTAZZINI AND R. Then Riemann chose Franklin’s unitary theory that attaches a certain quantity of electricity to every body and deﬁnes a positive or a negative electric state of a body as an excess or a defect of this electricity respectively. p. on Weber’s advice. He then compared “the consequences of the law in certain particular cases with experience” [Riemann 1854c. p. Riemann intended to publish an extended version of his results in Annalen der Physik und Chemie but then. 51] and found well known expressions in the case when the bodies were metallic and perfect conductors. Riemann presented the two theories on the nature of electric force. Riemann rejected this theory since. the editor of the review.2) α ∂ρ ∂2ρ ∂2ρ ∂2ρ + ρ − β2 + + 2 2 ∂t ∂x ∂y ∂z 2 = 0. he had submitted it to calculations “without obtaining any satisfactory results” [Riemann 1854c. αζ). αη. The law (3. 367]) have reported. According to the dualistic conception. where α is a constant depending on the nature of the body at issue. he changed his mind because of the substantial changes requested by Kohlrausch. p. Riemann published in Kohlrausch’s Annalen his 1855 paper on Nobili’s rings. a positive electricity and a negative one. . 23 Eventually his original manuscript was published by Weber [Riemann 1876d] and it includes a section crossed out by Riemann 23 Instead of it. both discussed at that time — the unitary conception and the dualistic one. 54]. In the last part of this paper. two opposite electricities existed in a body. TAZZIOLI (αξ. By means of both the principle of conservation of charge ∂ρ ∂ξ ∂η ∂ζ + + + = 0.20 U. 2 2 ∂x ∂y ∂z 2 Riemann deduced (3. as Dedekind [1876. 550] and Weber (in [Riemann 1876d. p. ∂t ∂x ∂y ∂z and the formula ∂2u ∂2u ∂2u + + = −ρ.2) was also in accordance with the experimental data of Kohlrausch about the free charge and the “reappearing” residuum.

As Maxwell 24 “Dieser ganze Artikel — Weber observed in a footnote — ist im Manuscript durchgestrichen. 371]. y. bei den Physikern damals a Anstoss zu erregen bef¨rchtete” [Riemann 1876d . p. According to Gauss. vorbemerktermassen a selbst noch abge¨ndert und umgeformt werden. Uebrigens m¨ssen diese Bewegungsgesetze der Electricit¨t. the following formula (3. Therefore. two elements of electricity attract and repulse each other with a force depending on their moving state. und dann wird eine ver¨nderte Aufa a fassung dieser Erscheinungen n¨thig” [Riemann. a o welche in jedem Augenblicke der aus diesen Ursachen hervorgehenden electromotorischen Kraft gleich ist [the cause is the potential u which satisﬁes the equation ∆u = −ρ]. welche auf das Innigste mit u seinen naturphilosophischen Principien zusammenh¨ngt. 26 He gave for the force between two electric particles ε and ε placed in (x. the electricity moves “against ponderable bodies” with a speed which equals the electromotive force deriving from the potential u. 25 In a note dated July 1835 Gauss had already suggested a new theory of electrodynamics. 27 and v is the velocity of ε(x. this is the velocity of light. z ) respectively. p.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 21 himself. and it decreased or increased the electric density according as the body contained positive or negative electricity. 27 Within the limit of observation. this law of motion must be changed in order to show its connection with heat and magnetism. weil der Verfasser durch die Eigenth¨mlichkeit der hier vorgetragenen Auﬀassung. the transmission of electricity could not be instantaneous but. Riemann continued. p. 371]. wahrscheinlich nur aus dem Grunde. y . Riemann assumed that the electric current was caused by a reaction the body opposed to its own electric state. wenn deren Verh¨ltniss u a a zu W¨rme und Magnetismus in Rechnung gezogen werden soll. y. Moreover. instead of Franklin’s unitary theory of electricity. Thus. z) and in (x . y . z ). . 1876d . This reaction was proportional to the charge density ρ multiplied by the coeﬃcient β 2 . where r is the distance from ε to ε . 616]. c is a constant velocity. z) with respect to ε (x . he developed a physical explanation of the electromotive force and of electric propagation through a body on the basis of the model of ether exposed in the Fragmente. 24 In this section. o 26 “Zwei Elemente von Electricit¨t in gegenseitiger Bewegung ziehen einander an oder a stossen einander ab nicht eben so als wenn sie in gegenseitiger Ruhe sind” [Gauss 1835.3) F = 1 εε 3 1 + 2 v2 − r2 c 2 dr dt 2 . u 25 “Die Electricit¨t bewegt sich gegen die ponderabeln K¨rper mit einer Geschwindigkeit.

288]. z) and (x .4) F = 1 d2 r 1 εε 1+ 2 r 2 − 2 r c dt 2 dr dt 2 . z ) respectively. y . p. p. 484] showed. He studied the interaction of two particles ε and ε . and the galvanic currents running through them. in a letter to Weber. But. Gauss [1845. 484]. Riemann himself tried to formulate a new theory of electrodynamics in his 1858 paper Ein Beitrag zur Elektrodynamik. Gauss never published his ideas about electrodynamics during his lifetime. Riemann considered the electrodynamic system of two conductors C and C moving one with respect to the other. y. as in the case of light. In 1845. TAZZIOLI [1873. and this result coincides with that of Amp`re’s. BOTTAZZINI AND R.4) satisﬁed the principle e of the conservation of energy on assuming for the potential energy (3.3) is inconsistent with the principle of the conservation of energy. 629] supposed that electricity propagated from one point to another not instantaneously. the ﬁrst in C and the latter in C . p. 139]. p. and W. but in time. “an indeﬁnite amount of work cannot be generated by a particle moving in a periodic manner under the action of the force assumed by Weber ” [Maxwell 1873. published in his celebrated Elektrodynamische Maassbestimmungen (1846). He supposed that electric phenomena travel with the velocity of light and that the diﬀerential equations for electric force are the same of those valid for light and heat propagation [Riemann 1858. “With admirable directness — Rosenfeld has remarked — he [wrote] down the generalised Poisson equation [for the potential ] involving the operator now called ‘D’Alembertian’ and the expression for its solution in the form of retarded potential ” [1956. with coordinates (x. According to Weber. as he pointed out.3) and (3.4) lead to the same result for mechanical force between two electric currents. p.5) P = εε 1 1− 2 r2 2c dr dt 2 . As a follower of Gauss and Weber. the force between the two particles ε and ε was given by (3. So Weber’s formula must be rejected too. was the ﬁrst result of this kind known to the scientiﬁc world. Weber’s theory of electrodynamics. .22 U. Maxwell remarked that the formula (3. the formula (3. The expressions (3.

p. tutto ideale — Betti observed — ` poco in e armonia con ci` che si conosce di essa. Riemann found that the potential function at the point (x. and those of ε at time t were (xt . As Reiﬀ and Sommerfeld [1902. the mistake was discovered by Riemann himself and this was the reason that convinced him to withdraw the paper. 242].7). and that the two sums εf (x. z ) extended to all the charges were negligible with respect to the same sums extended only to the positive or negative charges. α Riemann tried to obtain the electrodynamic potential derived from the Weber’s law from (3. and r was the distance between ε and ε . zt ). In Betti’s opinion however Riemann did not publish this paper because it was in contrast with what he had stated in [Riemann 1854c. yt . z) and α the velocity of propagation of electricity. ε f (x . zt ). and deduced the potential expressing the total eﬀect of the conductor C on the conductor C from the time 0 to the time t t (3. Riemann committed a mistake in the permutation of two integrations. the potential U travels with a velocity α and it reaches a point distant r from (x. where ρ was the charge density in the point (x. According to (3. 54]. p. yt .NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 23 Riemann indeed gave the following equation for the potential function U (3. y. 293]). . On these assumptions. he set F (t.6) ∂2U ∂2U ∂2U ∂2U − α2 + + ∂t2 ∂x2 ∂y 2 ∂z 2 + 4α2 πρ = 0. z) due to the point (x . 46] wrote. y.7) P =− 0 εε F t − r . p. But. τ dτ. the interest in this work is based on this result which makes Riemann appear as a “precursor” of Maxwell. y. z ) was f (t − r/α)/r. e pare che Riemann non ne fosse soddisfatto” o [1868. if the charge −f (t) was placed in the point in question at time t. z). 28 28 “Questo concetto della corrente elettrica. y. as Clausius [1868] was to point out. y . z) after the time r/α. Moreover. t ) = (xt − xt )2 + (yt − yt )2 + (zt − zt )2 −1 2 . y . if the coordinates of ε at the time t were (xt . According to Weber (see [Riemann 1858.6). p. Riemann supposed that the current was due to the positive and negative electricities running through the wire.

S and D were expressed by the following equations D= εε u2 . In their theories. p. He supposed the closed circuits to consist of polarized elements which acted on one another as little magnets. z) and (x . z ) respectively. what is its condition after it has left the one particle and before it has reached the other? ” [1873/1954. Riemann tackled the electrodynamic problem of two conductors moving one with respect to the other by using the calculus of variations. u the velocity of one particle with respect to the other. the action travelled in a manner similar to that of light but. 316]. If ε and ε were two electric particles in C and in C at the points (x. indeed. He deduced from the principle of conservation of energy the “extended theorem of Lagrange” t (3. 1876e. y . where T is the kinetic energy. 29 Clausius also criticized some parts of Betti’s calculations. In his 1858 paper Riemann did not mention any medium which propagated the electric phenomena. gravity and magnetism. TAZZIOLI Betti himself suggested a new theory of electrodynamics. Here. BOTTAZZINI AND R. and c the velocity of light. He tried however to describe the ether surrounding two interacting electric particles in his lectures on electricity. r 2c2 S=− εε . r where r was the distance between ε and ε . . 493]. 29 Maxwell was aware of Riemann’s and Betti’s electrodynamic researches.24 U. According to Betti. and even he could not agree with “these eminent men”. II.8) δ 0 (T − D + S) dt = 0. the polarized elements interacted not instantaneously but only after a time proportional to the distance between the elements. held in 1861 and published by Hattendorﬀ in 1876. p. Maxwell asked: “If something is transmitted from one particle to another at a distance. S the potential depending only on the coordinates (electrostatic potential) and D the potential depending on the velocities too (electrodynamic potential) [Riemann. y.

“from this diﬀerential equation it would be possible to derive the meaning of the functions V . u2 . u3 as [the components of ] the intensity of the ﬂux of this ether ”. u2 . Riemann obtained the following expression for the force acting on ε and ε 2 F = εε 1 + u x 2 r 2c2 u2 εε Fy = 2 1 + 2 r 2c 2 F = εε 1 + u z r2 2c2 ∂r εε d + 2 ∂x c dt ∂r εε d + 2 ∂y c dt ∂r εε d + 2 ∂z c dt x−x r y−y r z−z r . “Auf Grund dieser Diﬀerentialgleichung k¨nnte man uber die Bedeutung der o ¨ Functionen V. u3 . u2 = r dt ε dy . As N. .8). p. p. the principle of conservation of charge held for the ether too. However. Man kann annehmen. . u3 = r dt ε dz · r dt where u1 = ε ε ε According to Riemann.4) (see [Reiﬀ and Sommerfeld 1902. u2 . By considering the eﬀect of all the particles of the conductor C on the particle ε. pp. u1 . . 48–49]). 330]. u1 .NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 25 By solving the Euler-Lagrange equations deduced from (3. u3 eine Annahme machen. the interaction between ε and ε deduced from (3. u3 als die Stromintensit¨ten dieses Aethers ansehen” [Riemann 1876e . In his 1861 lectures he assumed that the ether satisﬁed physical properties which guaranteed electrical propagation. u2 . ∂V = div u = + + ∂t ∂x ∂y ∂z ε dx . he was not able to incorporate electrostatic and electromagnetic eﬀects into it.8) is given by Weber’s law (3. u1 . 31 By this identiﬁcation. die elektrische Wirkung werde durch einen Aether vermittelt. One could assume that electricity is propagated by an ether. Riemann explained his model of ether in his 1853 paper on the principles of Naturphilosophie. Verm¨ge der Gleichung [(3.10)] V may be regarded as the density and u1 . Riemann set S = εV and found for the potential function V the relation 30 (3. Wise has remarked: “a stable gradient in density would 30 31 For a discussion of this point see Wise [1981. (3. By means of equation [(3.10) ∂u1 ∂u2 ∂u3 .9) If D is Weber’s electrodynamic potential.10)] liessen o a sich dann V als die Dichtigkeit. 290].

Licht und Schwere hatte ich gleich nach Beendigung meiner Habilitationsschrift wieder aufgenommen und bin mit ihr so weit gekommen. ohne dass es mir als Anmaassung ausgelegt wird — ich hoﬀe. 547] reported. notwithstanding the fact that Gauss had formulated another theory of the connection between electricity and light. 33 Some months later he confessed to his brother to having been “so fully immersed” in his research on Naturphilosophie that he was not able to rid himself of it when the subject 32 “Ich bin aber v¨llig ¨ berzeugt. Gal¨ a vanismus. the question concerning the principles of geometry that Gauss choose as the subject of Riemann’s Probevorlesung was certainly the least prepared of those Riemann had proposed. the hypothesis (in [Riemann 1858]) that the same diﬀerential equation held for light and electric propagation. p. “I am sure that my theory is right and that in a couple of years it will be generally recognized as such”. In this letter Riemann stated that after the completion of his Habilitationsschrift [Riemann 1854a] he once more took up his “other research” on Naturphilosophie. 553]). dass Gauss seit mehreren Jahren auch daran arbeitet. p. even though Riemann should have thought about this subject for long time. die Sache unter dem Siegel der Verschwiegenheit mitgetheilt hat. BOTTAZZINI AND R.26 U. 547]). 1853. Referring to his 1858 paper Riemann wrote to his sister that. dass ich die Sachen vollkommen selbst¨ndig gefunden habe” (in [Dedekind 1876. 33 “Meine andere Untersuchung uber den Zusammenhang zwischen Electricit¨t. TAZZIOLI then correspond to electrostatic force. dass es nun f¨r mich noch nicht zu sp¨t ist und es anerkannt werden u a wird. dass ich sie in dieser Form unbedenklich ver¨ﬀentlichen kann. Weber. und einigen Freunden. . As Dedekind [1876. made the ether responsible for the transmission of both the phenomena. whereas a time rate of change in density. would correspond to electrodynamic eﬀects” [1981. 32 4. Therefore. 290]. dass die meinige die richtige ist und in ein paar o u Jahren allgemein als solche anerkannt werden wird” (in [Dedekind 1876. a p. A. ON THE FOUNDATION OF GEOMETRY The attempt undertaken by Riemann in the Fragmente at formulating a theory which could explain the transmission of phenomena is in our opinion among the basic motivations for the study of manifolds presented by Riemann in his 1854 Habilitationsvortrag. and associated ﬂuxes. p. Es ist mir dabei aber zugleich immer o gewisser geworden. u. which was almost ready for printing. — Dir kann ich dies wohl schreiben. as he wrote to his brother on December 28.

one can argue that in this lecture Riemann tried to generalize the ideas stated in his 1853 paper on Naturphilosophie to n-dimensional manifolds. From this point of view one can explain the seeming anomalies with respect to contemporary research in foundations of geometry and in nonEuclidean geometry. Nor did he refer to the work of Bolyai and Lobachevsky. even though a paper of the latter had appeared in 1837 in Crelle’s Journal and his Geometrische Untersuchungen had been published in Berlin in 1840. lorsqu’on admet l’hypoth`se de la somme des angles moindre que la demi-circonf´rence du e e cercle” [Lobachevsky 1837. 741]. de supposer dans un triangle rectiligne la somme des angles ´gale ` deux angles droits. Weyl quoted this passage from Riemann’s letter for supporting his claim that Riemann’s research “on the connection between light. often mentioned by historians of mathematics. In fact. However one can argue that Riemann probably knew Lobachevsky’s geometry. it seems very likely that he had this volume in his hands when he worked at his Habilitationsschrift. in his lecture Riemann only made a cryptic allusion in passing to the possibility of the elliptic geometry. electricity. extending the “local” investigation of particles of ether to the “global” analysis of space. au moins dans l’analyse. in our opinion these letters suggest that Riemann’s 1854 lecture was deeply connected with his research in mathematical physics. magnetism and gravity” was not “objectively in any relation” to the content of his lecture [1919b/1991. In this connection the reference Riemann made to Gauss and Herbart is also worth noting [1854b. pp. sinon dans la nature. Lobachevsky concluded that: “L’hypoth`se de la somme des angles d’un triangle moindre que deux e . Contrary to Weyl’s claim. After summarizing his main results concerning the geometry of a nonEuclidean rectilinear triangle. at least in the version that appeared in that Journal. Thinking of Riemann’s paper [1854a]. si ce ne sont les observations directes. 273]. p. In fact.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 27 of his lecture (Probevorlesung) was proposed to him (see [Dedekind 1876. To be sure. 547–548]). H. 295]. the relevant volume of Crelle’s Journal included both Lobachevsky’s paper and Dirichlet’s 1837 paper on the representation of “arbitrary” functions in series of Legendre polynomials. Had Riemann limited himself to reading the ﬁrst page of Lobachevsky’s paper. p. he could have discovered that: “rien n’autorise. p. et que la g´om´trie n’en peut pas moins e a e e exister.

In addition to the results of the geodetical observation that Gauss had reported at the end of his 1827 paper.28 U. With this in mind. des surfaces. TAZZIOLI angles droits ne peut avoir d’application que dans l’analyse. “However interesting it may be to consider the possibility of this approach . que dans un triangle dont les cˆt´s sont de la mˆme grandeur ` peu pr`s que la distance de oe e a e la terre au soleil. pp. la somme des angles ne peut jamais diﬀ´rer de deux e angles droits d’une quantit´ qui puisse surpasser 0 . As he wrote in a manuscript. apparently Riemann had no real interest in the problem of the foundations of geometry as such. As Gauss had done for the surfaces in his 1827 paper. According to Riemann. BOTTAZZINI AND R. J’ai prouv´ ailleurs. in research concerning the axioms of geometry (and in particular that of parallels). Or cette diﬀ´rence doit ˆtre d’autant moindre que les cˆt´s e e e oe d’un triangle sont plus petits” [1837. this could have given Riemann further suggestions for the remarks on the “empirical certainty” of geometry which can be read in the concluding lines of the introduction to [Riemann 1854b]. e e e Whatever the case may have been. en e e m’appuyant sur quelques observations astronomiques. e e e e e Les valeurs des ´l´ments diﬀ´rentiels des lignes courbes. 302–303]. 525]. de sorte que cette derni`re e e g´om´trie n’est sous ce rapport qu’une g´om´trie diﬀ´rentielle. we can appreciate better the suggestions he could have drawn from the following passage one can read in Lobachevsky’s paper: “La g´om´trie imaginaire est con¸ue sur un plan plus g´n´ral que la e e c e e g´om´trie usit´e qui n’en est qu’un cas particulier. Riemann too attributed a crucial importance to the deﬁnition of the linear element ds as starting point for the study of manifolds. Today by hypothesis we tend to mean everything which is mentally added to phenomena” [Riemann 1876a. no matter how high their probability within the limits of observation [Riemann 1854b. p. “the word hypothesis now has a slightly diﬀerent meaning than in Newton. 302]. et qui en d´rive dans e e e e la supposition des lignes extrˆmement petites. ee e et du volume des corps sont les mˆmes dans la g´om´trie imaginaire et e e e dans la g´om´trie usit´e” [1837. 0003 en secondes e sexag´simales. p. Euclidean geometry was not a “necessity” but merely an “empirical certainty” and the facts on which it was based were only “hypotheses”. 273]. p. puisque les mesures directes ne nous montrent pas dans la somme des angles d’un triangle la moindre d´viation de deux angles droits.

j=1 gij dxi dxj = i. 1 2 3 which had been introduced in his 1853 paper on Naturphilosophie (see § 2 above).NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 29 to geometry. In this way. Riemann’s Habilitationsvortrag can be interpreted as a moment of both naturphilosophische and mathematical construction. u a o p.1) was the generalization to a n-dimensional manifold of the expression dx2 + dx2 + dx2 = dx12 + dx22 + dx32 . he proposed to ﬁnd necessary and suﬃcient conditions such n that the forms Φ and Φ = i. p. More generally. Riemann tried to calculate how much V n . denn wir w¨rden dadurch keine neuen S¨tze ﬁnden k¨nnen” (in [Scholz 1982. it involved 1 n(n+1) arbitrary con2 stants. actually realising it would prove utterly sterile since in this way we could never ﬁnd new theorems”. As n equations were satisﬁed by the new coordinates so 1 n(n − 1) 2 34 “Wenn es aber auch interessant ist.1) i. with the fundamental form Φ.1) existed. This was equivalent to determine n new coordinates such that the diﬀerential system n n (4.j=1 gij dx i dx j was satisﬁed.j=1 gij dx i dx j could transform into each other. In an attempt to overcome “the shortcomings of the concepts” and to favour progress “in recognizing the connection of things” [Riemann 1854b. he supplied a generalization to n-dimensional manifolds of what he had elaborated concerning 3-dimensional space and the laws of propagation of the phenomena as well.j=1 gij dxi dxj (gij positive deﬁnite matrix). System (4. it was always possible to compare distances between diﬀerent points and to study deformations of an elastic ﬂuid ﬁlling the whole universe. . If the solution of (4. 34 In our opinion. diﬀered from the Euclidean n-dimensional space E n . 286]. Riemann deﬁned a metric in the whole space by associating to the n-dimensional manifold V n the fundamental form n Φ= i. he wrote in a note kept in his Nachlass. 229]). die M¨glichkeit dieser Behandlungsweise der o Geometrie einzusehen so w¨rde doch die Ausf¨hrung derselben ¨usserst unfruchtbar u u a sein.

30 U. Riemann sought to answer the following general question: “quales esse debeant proprietates corporis motum caloris determinantes et distributio caloris. p. the problem was to determine a new coordinate system in which Φ could n be expressed as the Euclidean form i=1 (dx i )2 .k (xk dxi − xi dxk )(x dxj − xj dx ). xn . k ) which corresponded — but for the inessential factor 2 — to the curvature tensor (4.k n (dxi )2 then i=1 cij. ut detur systema linearum quae semper isothermae maneant”. and aﬃrmed that if ds2 = 0 there existed numbers cij.β] − [i . 143]. Riemann introduced the quantity (ij.α][jk. 0 In 1861. when trying to answer a question proposed by the Paris Academy on heat conduction in homogeneous solid bodies. Riemann did not win this prize since he did not completely explain the ways according to which he found his results (see [Riemann 1861. In the second part of the paper he studied the particular case of a homogeneous body. BOTTAZZINI AND R. 391]).β=1 g αβ [j .β] . .j=1 i. .j=1 aij dx i dx j where aij are given constants. In order to characterize these functions. Riemann introduced a system of normal coordinates. Since both the forms were positive deﬁnite. If Φ = ds2 the manifold was called plane.k such that Φ became Φ = ds2 + 0 ij. TAZZIOLI “functions of position” had to be deduced “by the nature of the manifold to be represented ” [Riemann 1854b/1979. To this end. p. which we denote x1 . Riemann stated that the problem proposed by the Academy was equivalent to ﬁnding necessary and suﬃcient conditions according to which n n gij dxi dxj = i. .α][ik. . Riemann once more considered this problem.2) Rij k = 1 2 ∂ 2 gik ∂ 2 gj ∂ 2 gi ∂ 2 gjk + − − j ∂x i ∂xk j ∂xk ∂x ∂x ∂x ∂xi ∂x n + α. According to Weber.

.j=1 gij dx i dx j at time t and t respectively.j=1 n δij (P ) dxi dxj = i=1 (dxi )2 . . . . Since the system of coordinates is the same for ds2 and ds 2 . ds 2 = i. is given by Rij k = 0. then we can associate to space the fundamental forms n n Φ= i.j=1 gij (P ) dx i dx j . On these assumptions n n ds2 = i. . x2 + dx2 . If we suppose. and the distance ds between P and Q in the distance ds between P and Q .j=1 hij (P ) dxi dxj . x2 . Now. xn + dxn ). . x2 + dx2 . Let ai = ∂x /∂xi . . let us consider an elastic deformation changing P (x1 . that an ethereal medium ﬁlls all the universe. . He found that the necessary condition such that the system n n (4. xn ) and Q (x1 + dx1 .3) i. . we can evaluate the virtual displacement n (4.j=1 gij (P )ai ak dxi dxj = j i. as Riemann had done in his 1853 paper.4) δ(ds2 ) = ds 2 − ds2 = i.j=1 eij (P ) dxi dxj . . xn ) and Q(x1 + dx1 .j=1 δij dxi dxj and Φ = i.k] = 1 2 ∂gjk ∂gij ∂gik + − ∂xj ∂xi ∂xk is the Christoﬀel symbol of ﬁrst kind.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 31 where [ij. . . . . . . then n n ds 2 = i. .j=1 gij dxi dxj = i=1 (dx i )2 is satisﬁed. xn + dxn ) in P (x1 . x2 .

we can try to .2) given by Riemann [1853. According to Cliﬀord. The variation (4. From this point of view (4. if the strain tensor is diﬀerent from zero.4) can be interpreted as a deep connection between physical phenomena and the variation of the fundamental form.32 U. K. p. subject (possibly) to the law of continuity” [Cliﬀord 1876. BOTTAZZINI AND R. Moreover. a force is always coupled to a change in the curvature of space. space being the responsible for the transmission of phenomena by means of a variation of its fundamental form. 203]). “this variation of the curvature of space is what really happens in that phenomenon which we call the motion of matter. orthogonal 3-coordinates. emphasized this point asking the question “whether physicists might not ﬁnd it simpler to assume that space is capable of a varying curvature. then the fundamental form has been modiﬁed and the curvature tensor does not vanish. Thus force and curvature appear to be deeply connected. and of a resistance to that variation. if a solution of the system n n (4. whether ponderable or ethereal ” and “in the physical world nothing else takes place but this variation. By extending the same procedure to any particle of ether one obtains the quantities eij (the strain tensor in elasticity theory) deﬁned on the whole space.4) is the generalization to any curvilinear n-coordinate system of the expression (2. the editor of Cliﬀord’s book [1885]. p.5) i. On the contrary. 22]. one can suppose that the transmission of forces modiﬁes the fundamental form Φ. according to Riemann’s assumption that an ether particle propagates physical phenomena reacting to the deformation described by eij . Having this in mind and thinking of the research program on Naturphilosophie which Riemann was pursuing at that time. p. then the curvature tensor Rij k is equal to zero. TAZZIOLI where eij = hij − δij . This suggests a physical model of the space independent of assumptions about the existence of the ether. Pearson. than to suppose the existence of a subtle medium pervading an invariable homaloidal [Euclidean] space” (in [Cliﬀord 1885. No variation occurs if eij = 0: in this case no physical force is propagated through space.j=1 gij dx i dx j exists. The results obtained by Riemann in his 1861 paper show that. 530] in cartesian.j=1 δij dxi dxj = i. In other words.

Riemann’s statements seem to be better explained . In this connection Riemann contrasted discrete manifolds with continuous manifolds: “Either the reality underlying space” is given by a discrete manifold which has the principle of its metrical relations in itself or “the basis for the metric relations must be sought outside it [space]. 152–153]. the realm of physics”. “the question of the validity of the hypotheses of geometry in the inﬁnitely small ” was linked to the determination of “metric relations of space”. however. 96–102]). 532 sq. the “full purport” of Riemann’s concluding remarks “was not grasped by his contemporaries” with the exception of “a solitary echo” in Cliﬀord’s writings. for which Newton laid the foundation. pp. 97]. in binding forces acting upon it ”. In Raum Zeit Materie Weyl interpreted Riemann’s words in terms of Einstein’s theory of relativity: “Only now that Einstein has removed the scales from our eyes by the magic light of his theory of gravitation do we see what these words actually mean”. What has one to understand by this “reality”? Did this involve Riemann’s speculations on ether which. According to his concluding remarks. he concluded. p. 152]. From a historical point of view. In this respect we cannot do better than referring to his book (see in particular [Weyl 1919a/1922. and gradually modifying it under the compulsion of facts which cannot be explained by it” [Riemann 1854b/1979. lose their validity in the inﬁnitely small. This was a question which. “This leads us away into the domain of another science. According to Weyl [1919a/1922. an answer “can be found only by starting from that conception of phenomena which has hitherto been approved by experience.]. p. in Riemann’s view. as he wrote in the manuscript Gravitation und Licht [Werke. it is therefore quite deﬁnitely conceivable that the metric relations of space in the inﬁnitely small do not conform to the hypotheses of geometry” [Riemann 1854b/1979. the concept of a solid body and that of a light ray. pp.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 33 understand the cryptic and prophetic conclusion he drew at the end of the lecture: “Now it seems that the empirical notions on which the metric determinations of space are based. might “still be ranked as a part of the study of space”. “as a physical space whose points move in geometrical space”? Riemann left open the questions he raised concerning the nature of space. Thus. p. could be conceived as “a substance spread continuously through the entire inﬁnite space”.

p. for the kind permission of publishing excerpts from the letters kept in Klein’s Nachlass. [1986] The higher calculus: a history of real and complex analysis from Euler to Weierstrass. BOTTAZZINI (U. pp. pp. p. 1991). pp. we would like to thank the colleagues attending the Workshop on the history of mathematics held at the Mittag-Leﬄer Institute (Djursholm. Paris: Masson. 383–409. Magnani. pp. Romanticism in science (S.) and NOETHER (M. 3 (1894). ed. Dordrecht: Reidel. We are also grateful to the anonymous referees for their helpful remarks and to Jeremy Gray for checking the English. TAZZIOLI in the light of his own attempts “to unify nature on the basis of a geometrically conceived system of continuous dynamic processes in ether ”. 1991. 1986. 242–245. 1903–1913. Milano: Hoepli.). 554–573.. BIBLIOGRAPHY BETTI (E. BOTTAZZINI (U. “the ﬁrst attempt at a mathematically founded uniﬁed ﬁeld theory.) [Opere] Opere (Unione Matematica Italiana. 1985. Handa a o schriftenabteilung. ed. Milano: Garzanti.. 1994. Conoscenza e matematica (L.) [1985] Funzioni complesse e variet` multidimensionali: globale e locale nella a matematica di B. as N.). 107-565. Annalen der Physik und Chemie. spazio. ed. 289] deﬁned Riemann’s research on Naturphilosophie. Poggi and M. Wise [1981. much in the spirit of Einstein’s later attempts”. 2 vols. [1994] Geometry and “Metaphysics of space” in Gauss and Riemann.) [1894] Die Entwicklung der Theorie der algebraischen Funktionen in ¨lterer und a neuerer Zeit. Geymonat (C. pp. Milano: Marcos y Marcos. Acknowledgments During the past years. 135 (1868). sections of this paper have been presented at various meetings. New York: Springer. Opere 2.). Bossi. BRILLOUIN (L. BRILL (A.) [1991] Etere. In particular. Roma: Cremonese. geometria: riﬂessioni sui frammenti ﬁlosoﬁci di Riemann. Jahresbericht der Deutschen Mathematiker-Vereinigung. We also wish to thank the Nieders¨chsischen Staat-und Universit¨tsbibliothek G¨ttingen. Riemann. 606. 15–29. Mangione.) [1938] Les Tenseurs en m´canique et en ´lasticit´.) and TAZZIOLI (R. e e e CLAUSIUS (R. the Annual G¨ttingen Workshop (1992) and the Oberwolfach meeting o (1994) for their useful comments. BOTTAZZINI AND R.). eds. 11 vols.34 U. Scienza e ﬁlosoﬁa: Saggi in onore di L. 1952–1959. BIANCHI (L. . [1868] Sopra la elettrodinamica.). ed.) [Opere] Opere matematiche (Accademia dei Lincei. 1938.) [1868] Von Gauss angeregte neue Auﬀassung der elektrodynamischen Erscheinungen.

4 (1828). ed. repr. diﬀerential geometry and tensor analysis. 223–255. Ges. Leipzig: Teubner. Weber. Commentationes societatis regiae scientiarum Gottingensis recentiores. Engl. Tucker. Lejeune-) [Werke] Werke (L. eds. 157–58 (read Feb. . Werke 5. 1902. Werke 11. 17 (1990). 1863–1933. 21. pp. 3. 1963. ¨ [1827] Disquisitiones generales circa superﬁcies curvas. 1870). Werke 5. eds. pp. 4 (1894–1895). Experimental researches in electricity. et qui servent ` e e e e a exprimer des fonctions arbitraires entre des limites donn´es. Werke 5. pp.. Berlin: Mayer & M¨ ller.).). Resultate aus der Beobachtuna gen des magnetischen Vereins im Jahre 1839 (K. [1857] On the conservation of force. Journal f¨r e u die reine und angewandte Mathematik. 35–56. pp. pp. pp.). Leipzig. a FARADAY (M. transl. 1882. 1882.. pp. Proceedings of the Cambridge philosophical Society. pp.). math. Berlin: Springer. DEDEKIND (R. math. 1840. 17 (1837). London: Macmillan. 1921–1923. 194–242. London: Macmillan. 2 (1876). pp. 1876. III. repr. London. Experimental researches in chemistry and physics. 273–306.. 616–620. 482–497. Grube. Abh.K.) [Papers] Mathematical papers (R. 443–463. u o pp.) [Werke]Werke (K¨niglische Gesellschaft der Wissenschaften zu G¨ttingen. u 1889–1897.NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 35 CLIFFORD (W. [1837] Sur les s´ries dont le terme g´n´ral d´pend de deux angles. ¨ [1882] Uber Riemanns Theorie der algebraischen Funktionen und ihrer Integrale. p.) and KNEE (Ch. FARWELL (R. [1876] Vorlesungen uber die im umgekehrten Verh¨ltniss des Quadrats der Entfer¨ a nung wirkenden Kr¨fte (F. 3.). KLEIN (F. d.] Gesammelte mathematische Abhandlungen (Fricke and others. 541–558]. Werke 1. [1876] On the space-theory of matter. Werke 4. ed. 499–573.) [1876] Bernhard Riemann’s Lebenslauf. 217–258. Gauss and W. 1967. DIRICHLET (P. 71-87. French transl. Historia mathematica. [1894] Riemann und seine Bedeutung f¨ r die Entwicklung der modernen Mathu ematik. 1885. London.) [Ges. Papers. ed. 1859. pp. Abh. t. 1893. pp. [1885] The common sense of the exact sciences (K. pp. [1813] Theorem uber die Anziehung. o o 12 vols. math. 3 vols. Fuchs. repr. pp. Kronecker and L. pp. 161–168. Pearson. eds. 21–22. Leipzig: Teubner. Leipzig-Berlin. in Ast´risque 62 (1979).) [1990] The missing link: Riemann’s ‘Commentatio’. 71.F. Blanchard. [1845] Brief an Weber.).) [1851] On the possible relation of gravity to electricity.F. [1839] Allgemeine Lehrs¨tze in Beziehung auf die im verkehrten Verh¨ltnisse wirka a enden Anziehungs-und Abstossungs Kr¨fte. Ges. 627–629.). in [Riemann Werke. English transl. GAUSS (C. 1885..G. Math. Dover. 1855. e [1835] Grundgesetz f¨ r alle Wechselwirkungen galvanischer Str¨me. Abh. Dt. 2 vols. ed.. Jber.-Verein.

Motte revised by F. 1859. 3rd ed. 2. 1892. 91 u (1854). G¨ttingen.) [1726] Philosophiae Naturalis Principia Mathematica. ´ LAME (G. [1854] Theorie des elektrischen R¨ ckstandes in der Leidener Flasche. pp. Press. e e MAXWELL (J. Neugebauer eds.) [1851] Grundlagen f¨ r eine allgemeine Theorie der Funktionen einer ver¨nderlichen u a complexen Gr¨sse. Sci. repr. Courant and O.) [1853] Das Sinuselektrometer. pp. 1934. p.). NEWTON (I.. repr. Leipzig: Teubner. Math. TAZZIOLI [1926] Vorlesungen uber die Entwicklung der Mathematik im 19. 1979.M. e LOBACHEVSKY (N. Jahrhundert ¨ (R. [1959] Th´orie de l’´lasticit´. transl. I. repr.) [Œuvres] Œuvres compl`tes.) [1859] Le¸ons sur les coordonn´es curvilignes et leurs applications. 1878–1912. Werke. ed. pp. Engl. a a Physik. Ann. Leipzig: Teubner. 1891. 3rd ed. Paris 1898. o o [1853] Neue mathematische Principien der Naturphilosophie.. 2nd ed. Oxford: Clarendon Press.) [1882] Zu F. Werke.) [1837] G´om´trie imaginaire. 1954. 528–532. Brooklin: Math. 3–62..36 U. Weber. pp. pp. Encyklop¨die der mathematischen Wissenschaften. Prinzenstr. 227–271. Oxford: Pergamon Press.) [1951] The classical theory of ﬁelds. 1. t. (All references refer to the 1991 edition. REIFF (R. NOETHER (M. Paris: Gauthier-Villars. 27. [Nachlass] Nieders¨chsischen Staat-und Universit¨tsbibliothek G¨ttingen. 497. Paris: Malletc e Bachelier. RIEMANN (B.) and LIFSHITZ (E..) and SOMMERFELD (A. 1959. transl.) [1982] Riemann’s contribution to diﬀerential geometry. Inauguraldissertation. LANDAU (L. 13 (1868).D. Werke.). Transactions of the Cambridge philosophical Society. New York: Dover. 88 (1853). u PORTNOY (E.. BOTTAZZINI AND R. 14 vols. 9 (1982). Chem. 1968.C. [1873] A treatise on electricity and magnetism. London 1726. Ibid. p. Phys. Oxford: Pergamon Press.S. V. 1–18. Hist. 1991..) [1856] On Faraday’s lines of force. Berlin: Springer. 1979. 3–48. Engl. French edition. o KOHLRAUSCH (R. pp. repr. 1926. ¨ [1854a ] Uber die Darstellbarkeit einer Function durch eine trigonometrische Reihe. pp. 56–82... 201–206. 1873. Handa a o schriftenabteilung. 1851. . 5th ed. 2 vols. 1876. by A. Berkeley and Los Angeles: University of California Press. 1951. pp. 295–320. D-3400 G¨ttingen. reine angew. Blanchard. 17 (1837). e e e LAPLACE (P.. math. Abhandlungen der K¨niglichen Gesellschaft der Wissenschaften zu G¨ttino o gen. Zeitschrift f¨r die Mathematik und Physik 27 (1882). Cajori. J. New York: Springer.. 1962.) [Werke] Gesammelte mathematische Werke und wissenschaftlicher Nachlass (H. Klein’s Schrift ‘Ueber Riemann’s Theorie der algebraischen Functionen’.) [1902] Elektrizit¨t und Optik. 10 (1856).

a [1857a ] Beitr¨ge zur Theorie der durch die Gauss’sche Reihe F (α. Werke.. [1855] Zur Theorie der Nobili’schen Farbenringe..NATURPHILOSOPHIE AND ITS ROLE IN RIEMANN’S MATHEMATICS 37 ¨ [1854b ] Uber die Hypothesen. restent isothermes apr`s un temps quelconque. pp. Ges.). pp. Dordrecht/Boston: Reidel Publishing Company.S. 431–436. pp. Ann. . ROSENFELD (L. ed. 54 (1857). Werke. e [1869] Partielle Diﬀerentialgleichungen und deren Anwendung auf physikalische Fragen (K. 828–847]. 1972. ¨ [1876b ] Uber das Potential eines Ringes.. Chem. pp. Riemann. pp. Noether and W. SCHAFFNER (K. J. K. u [1876d ] Neue Theorie des R¨ ckstandes in electrischen Bindungsapparaten. u [1902] Nachtr¨ge (M. 440–444. 1979. ` un instant donn´. Amtlicher Bericht uber die 31. Hattendorﬀ. 367–383. 13 (1868). Haussner and K. Vieweg. Papers. 367–378. eds. pp. in [Spivak 1970/1979. G¨ttingen. pp.. Werke. Magnetismus.) [Papers] Selected papers of L´on Rosenfeld (R. 558). Schering. 135–153].). 507–538. Berlin: Mayer and M¨ ller. Ann.). o sondern als dem Enthalten von Spannungselectricit¨t mit endlicher Kraft a widerstrebend betrachtet werden. repr. de a e e telle sorte que la temp´rature d’un point puisse s’exprimer en fonction du e temps et de deux autres variables ind´pendantes’. Cohen and J. Hattendorﬀ. [1876c ] Gleichgewicht der Electricit¨t auf Cylindern mit kreisf¨rmigem Querschnitt a o und parallelen Axen.). 95 (1855). Riea mann (K. 1902. Hannover: C. Chem. eds. pp. Versamm¨ lung Deutscher Naturforscher und Aerzte zu G¨ttingen im September 1854 o (1854). pp. Wirtinger. Werke. R¨ mpler. pp. a Werke 1991. 272–287. x) darstellbaren Functionen. [1858] Ein Beitrag zur Elektrodynamic. Werke. a (separate numeration following p. Stachel. o [1857b ] Theorie der Abelschen Funktionen. Werke. Werke 2. 2 vols..F. u [1866] Zum Ged¨chtniss an B. β. SCHERING (E. pp. Phys. qua respondere tentatur quaestioni ab Illma Academia Parisiensi propositæ: ‘ Trouver quel doit ˆtre l’´tat caloriﬁque e e d’un corps solide homog`ne ind´ﬁni pour qu’un syst`me de courbes isothere e e mes. wenn diese nicht als vollkommene Leiter oder Nichtleiter. Werke. eds. 55–66. Nach den Vorlesungen von B. Nuovo Cimento 4 (1956). pp. 131 (1858). 1876. e Boston Studies in the Philosophy of Science 21. Phys. 134–177. 88–144. Math. Werke. pp. 67–83. Werke. Abh. 1902–1909. 1630–1669. reine angew. 1869. welche der Geometrie zu Grunde liegen. 7 (1857). Werke.J. ¨ a [1854c ] Uber die Gesetze der Vertheilung von Spannungselectricit¨t in ponderablen K¨rpern. English transl.) [Werke] Gesammelte mathematische Werke (R. γ. [1876a ] Fragmente philosophischen Inhalts. Elektricit¨t. pp. Wiss. viii + 116 p. Ibid. 391–404. 288-293. in [Riemann. pp. [1956] The velocity of light and the evolution of electrodynamics. Werke.). Suppl. Werke. [1876e ] Schwere.) [1972] Nineteenth-century aether theories. [1861] Commentatio mathematica. pp. Braunschweig: F. 49–54. ed. Oxford: Pergamon Press.

J.N..) [1927] Naturphilosophische Untersuchungen von Euler und Riemann. 1919. pp. welche der Geometrie zu Grunde liegen (H. vol. II. Math. Studies in the history of ether theory. 1964. Conception of ether.) [1913] Die Idee der Riemannschen Fl¨che. Berlin: Springer. 1913.) [1982] Riemanns fr¨ he Notizen zur Mannigfaltigkeitsbegriﬀ und zu den Grundlagen u der Geometrie. SPEISER (A. 213–232. 1923. Berlin: Springer. Berlin.). repr. WISE (N. 1981. repr. 1st ed. 209–378.) [1951] A history of the theories of aether and electricity. TAZZIOLI SCHOLZ (E.). 740–768]. 2nd ed. [1923] Mathematische Analyse des Raumproblems. 1922.. repr. energy.) [1970] A comprehensive introduction to diﬀerential geometry. 1952. 1951. 1970. Blanchard 1958. . vol. Archive for history of exact sciences. New York: Harper. Uber die Hypothesen.) [1846] Elektrodynamische Maassbestimmungen. ¨ [1919b ] Bernhard Riemann. repr. 1979. J. Dover. Weyl’s Vorwort and Erl¨uterungen are a reprinted in Riemann [Werke. Hodge. Dover 1991. 269–307. 27 (1982). pp.38 U. French transl. Cantor and M.M. pp. 1919. 1991. Engl. Cambridge: Cambridge University Press. SPIVAK (M. BOTTAZZINI AND R. eds. Abhandlungen bei Begr¨ndung u der K¨niglischen S¨chsichen Gesellschaft der Wissenschaften am Tage des o a 200 j¨hrigen Geburtstages Leibnizens (1846). a WEYL (H. WHITTAKER (E. I. 1740–1900 (G. a [1919a ] Raum Zeit Materie. WEBER (W.) [1981] German concepts of force. and the electromagnetic ether: 1845–1880. 157 (1927). 1922. Leipzig: Teubner. transl. Weyl ed. Houston: Publish or Perish. pp. 105–114. reine angew. pp.