The Geometry of The Second Fundamental Form: Curvature Properties and Variational Aspects

KATHOLIEKE UNIVERSITEIT LEUVEN
Faculteit Wetenschappen
Departement Wiskunde
AFDELING MEETKUNDE
The Geometry of the

Second Fundamental Form:
Curvature Properties
and Variational Aspects
Steven Verpoort
Proefschrift voorgedragen tot het

Promotor: behalen van de graad van
Prof. dr. L. Verstraelen Doctor in de Wetenschappen
23 mei 2008

c 2008 Katholieke Universiteit Leuven — Faculteit Wetenschappen
Geel Huis, Kasteelpark Arenberg 11, 3001 Heverlee, België.
Alle rechten voorbehouden. Niets uit deze uitgave mag worden vermenigvuldigd en/of openbaar gemaakt worden door middel van druk,
fotokopie, mikrofilm, elektronisch of op welke andere wijze ook zonder voorafgaande schriftelijke toestemming van de uitgever.
All rights reserved. No part of the publication may be reproduced in any form by print, photoprint, microfilm or any other means without
written permission from the publisher.
The electronic version of this dissertation is publicly available and can be reached
by browsing the catalogue of the university library at
http://bib.kuleuven.be .
ISBN 978-90-8649-183-4
D/2008/10.705/29
KATHOLIEKE UNIVERSITEIT LEUVEN
Faculteit Wetenschappen
Departement Wiskunde
AFDELING MEETKUNDE
The Geometry of the

Second Fundamental Form:
Curvature Properties
and Variational Aspects
(De Meetkunde van de Tweede Fundamentaalvorm:

Krommingseigenschappen en Variationele Aspecten)
Steven Verpoort
Proefschrift voorgedragen tot het

Promotor: behalen van de graad van
Prof. dr. L. Verstraelen Doctor in de Wetenschappen
23 mei 2008
PREFACE.
Approach.
The broad subject of this dissertation can be described as “the geometry of

the second fundamental form.”
I have not restricted myself to a mere presentation of the hand-
ful new achievements which have been obtained, but rather did I endeav-
our to provide the reader with an intelligible essay on the subject. In pur-
suit of this goal I have complemented the results which were reached with
a choice of previously known theorems, along with a full proof in most
cases.
The last chapter, which contains an overview of the literature
on the geometry of the second fundamental form, is intended to give the
reader some insight in the origin, the context and the development of the
topic. The aim of the extensive bibliography has not only been to give ac-
count of the works which were frequently consulted during the preparation
of the manuscript, but primarily to collect most of the sources which are
relevant to the subject. For convenience, a notation index and an author
and subject index have been included at the very end. Ample examples
and illustrations have been inserted into the text in the hope of making the
matter more tangible to the reader.
In the author’s personal opinion it is also part of the duty of
the researcher to report on those undertakings which turned out to be only
partially successful; in consequence, also some partial results have been
i
ii PREFACE.
included. To this category the attempts towards a solution of the problems

which are stated in § 2.5.4 may be counted.
Acknowledgements.
This thesis contains the results which have been achieved during my term
as doctoral student at the section of geometry at K.U.Leuven, and as such
has been highly enriched by several mathematicians whom I had the op-
portunity to discuss with.
In the first place I wish to thank my colleague dr. S. Haesen
for having suggested the study of a variational problem associated to the
second fundamental form. Moreover, a substantial part of the results have
been obtained in real collaboration with him. For this and for the plenty of
motivating discussions I am very grateful.
I also wish to thank prof. dr. F. Dillen, who helpfully found
the time to share his opinion on certain parts of my work in between his
occupations as head of our section.
I gratefully acknowledge the help of my adviser prof. dr. L.
Verstraelen who has given me considerable freedom, hereby greatly con-
tributing towards the stimulating working atmosphere.
I have also had advantageous discussions with the present and
former members of our section—in particular dr. E. Boeckx, dr. J. Faste-
nakels, mr. D. Kowalczyk and dr. J. Van der Veken—as well as with ms. W.
Goemans.
It has always been a pleasure to discuss with other researchers
on several occasions. In particular, I wish to extend a word of thanks to
prof. dr. U. Abresch, who has sketched the idea of the second proof of theo-
rem 4.1; to prof. dr. J. Guven, with whom I have had instructive discussions
on infinitesimal deformations; to prof. dr. R.D. Kamien, for the pleasant
communication on the variations of the curvature of a surface; and to prof.
dr. A. Romero for various valuable remarks.
In addition, I would like to express my sincerest gratitude to
prof. dr. B.-Y. Chen and prof. dr. em. L. Vanhecke for several useful com-
ments, in particular on the method of power series expansions as applied
in § 5.7.
ii
PREFACE. iii
I cannot forget to mention prof. dr. em. U. Simon for several in-
structive remarks, and in particular for his comments on an earlier version
of the last chapter.
I am also indebted to the K.U.Leuven for having given me the
opportunity to study as doctoral bursary.
I wish to thank all staff members of Arenberg library (K.U.Leu-
ven) for having retrieved numerous journal articles.
My thanks also go to the members of the examination board:
prof. dr. F. Dillen, dr. S. Haesen, prof. dr. I. Van de Woestyne, prof. dr. L.
Verstraelen, prof. dr. W. Veys, and prof. dr. L. Vrancken.
In conclusion of this list I wish to expressly mention my parents
and my brother for their support and understanding.
STEVEN VERPOORT.
iii
iv PREFACE.
iv
TABLE OF CONTENTS.
Preface. i
Table of Contents. v
Summary. xi
Nederlandstalige Samenvatting (Dutch Summary). xv
1. Variational Formulae for Surfaces in the Three-Dimensional

Euclidean Space. 1
1.1. Deformations of Surfaces. . . . . . . . . . . . . . . . . 1
1.2. Applications to Integral Formulae and Unicity Results for

the Sphere. . . . . . . . . . . . . . . . . . . . . . . . . 12
2. The Geometry of the Second Fundamental Form of an Oval-

oid in the Three-Dimensional Euclidean Space. 19
2.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . 19
v
vi TABLE OF CONTENTS.
2.2. Four Geometrical Interpretations of the Second Funda-

mental Form. . . . . . . . . . . . . . . . . . . . . . . . 20
2.3. The Second Fundamental Form as a Metric and the Cor-

responding Intrinsic Geometry. . . . . . . . . . . . . . . 23
2.3.1. Introduction. . . . . . . . . . . . . . . . . . . . . 23
2.3.2. Some Technical Results. . . . . . . . . . . . . . . . 25
2.3.3. The Gaussian curvature of the Second Fundamental
Form. . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4. The Extrinsic Geometry of the Second Fundamental

Form. . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.1. The Mean Curvature of the Second Fundamental Form. 30

2.4.2. A Beltrami formula for HII . . . . . . . . . . . . . . . 34
2.4.3. The Unit Normal Variation of the Curvatures of the Sec-
ond Fundamental Form. . . . . . . . . . . . . . . . 36
2.5. Some Characterisations of Euclidean Spheres, in

Which the Curvatures of the Second Fundamental Form
are Involved. . . . . . . . . . . . . . . . . . . . . . . . 38
2.5.1. Some First Results.. . . . . . . . . . . . . . . . . . 38

2.5.2. A Jellett-Minkowski Integral Formula for the Area of the
Second Fundamental Form.. . . . . . . . . . . . . . 41
2.5.3. An Adaption of Christoffel’s Variational Characterisa-
tion of Euclidean Spheres. . . . . . . . . . . . . . . 42
2.5.4. Isoperimetric Inequalities for the Second Fundamental
Form. . . . . . . . . . . . . . . . . . . . . . . . . 44
2.5.5. Further Integral Formulae and Unicity Results.. . . . . 46
2.6. Calculation of the Curvatures for Some Surfaces of Revol-

ution. . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.7. Infinitesimal Deformations of Surfaces in E3 . . . . . . . . 57
vi
TABLE OF CONTENTS. vii
3. Variational Formulae for Abstract semi-Riemannian Mani-

folds. 61
3.1. Deformations of Metrics. . . . . . . . . . . . . . . . . . 62
3.2. Variational Problems for Abstract Ovaloids. . . . . . . . 66
4. The Variation of the Shape Operator of a Hypersurface in a

semi-Riemannian Manifold. 69
4.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . 69
4.2. Hypersurfaces in a semi-Riemannian Manifold. . . . . . 70
4.2.1. Tensors and Curvature.. . . . . . . . . . . . . . . . 70

4.2.2. Deformations of Hypersurfaces.. . . . . . . . . . . . 72
4.3. The Variation of the Shape Operator. . . . . . . . . . . . 72
4.4. Applications of the Formula for the Variation of the Shape

Operator. . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.4.1. The Variation of the Curvatures of a Surface in a Three-

Dimensional semi-Riemannian Manifold. . . . . . . . 81
4.4.2. Properties of Equidistant Surfaces in Space Forms. . . . 82
5. The Geometry of the Second Fundamental Form of a Hyper-

surface in a semi-Riemannian Manifold. 87
5.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . 87
5.2. Some Technical Results. . . . . . . . . . . . . . . . . . 89
5.2.1. The Tensor B and the Vector Field Z . . . . . . . . . . 90

5.2.2. The Differential Invariants of the Second Fundamental
Form. . . . . . . . . . . . . . . . . . . . . . . . . 90
5.2.3. The Difference Tensor L and some Related Tensors. . . 91
5.2.4. A Lemma on the Connections. . . . . . . . . . . . . 96
vii
viii TABLE OF CONTENTS.
5.3. The Intrinsic Geometry of the Second Fundamental

Form. . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.3.1. The Scalar Curvature of the Second Fundamental Form. 98

5.3.2. Applications for Surfaces in the Hyperbolic Space. . . . 102
5.3.3. Further Applications.. . . . . . . . . . . . . . . . . 116
5.4. The Extrinsic Geometry of the Second Fundamental

Form. . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.5. Some Characterisations of Extrinsic Hyperspheres, in

Which the Curvatures of the Second Fundamental Form
are Involved. . . . . . . . . . . . . . . . . . . . . . . . 122
5.5.1. Hypersurfaces in a Space Form. . . . . . . . . . . . . 122
5.5.2. Hypersurfaces in an Einstein Space. . . . . . . . . . . 124
5.5.3. Surfaces in a three-dimensional semi-Riemannian
Manifold. . . . . . . . . . . . . . . . . . . . . . . 126
5.5.4. Curves in a semi-Riemannian Surface.. . . . . . . . . 129
5.6. Calculation of the Curvatures for Some Special Classes of

Hypersurfaces. . . . . . . . . . . . . . . . . . . . . . . 133
5.6.1. Calculation of the Curvatures for Some Surfaces of
Revolution. . . . . . . . . . . . . . . . . . . . . . 133
5.6.2. Calculation of the Curvatures for Some Surfaces in the
Minkowski Space. . . . . . . . . . . . . . . . . . . 134
5.6.3. The Slices of a Generalised Robertson–Walker Space. . 137
5.7. Geodesic Hyperspheres in a Riemannian Manifold. . . . 138
5.7.1. On the Technique of Power Series Expansions for Global

Invariants of Geodesic Hyperspheres. . . . . . . . . . 138
5.7.2. The Mean Curvature of the Second Fundamental Form
of the Geodesic Hyperspheres. . . . . . . . . . . . . 144
5.7.3. The Area of the Geodesic Hyperspheres, as Measured
by Means of the Second Fundamental Form.. . . . . . 152
viii
TABLE OF CONTENTS. ix
6. A Literature Overview on the Geometry of the Second Fun-

damental Form and the Geometry of Ovaloids. 157
6.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . 157
6.2. Notation and Preliminaries. . . . . . . . . . . . . . . . 158
6.2.1. Surfaces in E3 .. . . . . . . . . . . . . . . . . . . . 158

6.2.2. Hypersurfaces in semi-Riemannian Manifolds. . . . . . 159
6.3. The Second Fundamental Form as a Metrical Tensor and

the Curvatures of the Second Fundamental Form. . . . . 159
6.4. The Geometry of the Second Fundamental Form and Rel-

ative Differential Geometry.. . . . . . . . . . . . . . . . 162
6.5. The Conformal Geometry of the Second Fundamental

Form. . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
6.6. II-minimal Surfaces. . . . . . . . . . . . . . . . . . . . . 167
6.7. Congruence Theorems. . . . . . . . . . . . . . . . . . . 168
6.7.1. Some Classical Congruence Theorems. . . . . . . . . 168

6.7.2. Congruence Theorems Involving the Second Fundamen-
tal Form. . . . . . . . . . . . . . . . . . . . . . . 171
6.7.3. Congruence Theorems Involving the Third Fundamen-
tal Form. . . . . . . . . . . . . . . . . . . . . . . 173
6.7.4. Congruence Theorems Involving the Shape Operator. . 174
6.7.5. The Weyl and Minkowski Problem. . . . . . . . . . . 175
6.8. Infinitesimal Deformations. . . . . . . . . . . . . . . . . 176
6.8.1. Infinitesimal Deformations Which Preserve the Second

Fundamental Form. . . . . . . . . . . . . . . . . . 177
6.8.2. Infinitesimal Deformations Which Preserve the Third
Fundamental Form. . . . . . . . . . . . . . . . . . 178
6.8.3. Infinitesimal Deformations Which Preserve the Shape
Operator. . . . . . . . . . . . . . . . . . . . . . . 178
ix
x TABLE OF CONTENTS.
6.9. On the Global Geometry of Ovaloids. . . . . . . . . . . . 179

6.9.1. Isoperimetric Inequalities and Related Integral Inequal-
ities and Equalities. . . . . . . . . . . . . . . . . . 179
6.9.2. Existence of Umbilics and Closed Geodesics.. . . . . . 181
6.10. Characterisations of Totally Umbilical Hypersurfaces. . . 182

6.10.1. Characterisations of Spheres in E3 . . . . . . . . . . . 182
6.10.2. Characterisations of Hyperspheres in Em+1 . . . . . . . 188
6.10.3. Characterisations of Extrinsic Spheres in Space Forms. . 190
6.10.4. Characterisations of Totally Umbilical Hypersurfaces of
semi-Riemannian Manifolds. . . . . . . . . . . . . . 191
6.11. Miscellaneous Related Topics. . . . . . . . . . . . . . . 192
6.11.1. Convexity. . . . . . . . . . . . . . . . . . . . . . . 192
6.11.2. Submanifolds in a Euclidean Space of Arbitrary Codi-
mension. . . . . . . . . . . . . . . . . . . . . . . 192
6.11.3. Non-smooth Convex Surfaces. . . . . . . . . . . . . 193
6.12. Ten Interesting Open Problems. . . . . . . . . . . . . . 194
Bibliography. 197
Notation Index. 215
Author and Subject Index. 221
x
SUMMARY.
In this dissertation, hypersurfaces with a non-degenerate second funda-

mental form are investigated. The second fundamental form of such a
hypersurface can be studied as an abstract semi-Riemannian metric in its
own right.
As such, the geometrical properties of the hypersurface can not

only be investigated with respect to the usual geometry (of the first funda-
mental form), but as well with respect to the geometry which is determined
by its second fundamental form.
The most well-studied example of hypersurfaces fulfilling the

above requirement are the compact surfaces in the three-dimensional
Euclidean space, which satisfy a certain convexity condition. These are
known as ovaloids. The geometry of an ovaloid as determined by its second
fundamental form deviates from its usual geometry in the sense that the
norm of a tangent vector should be multiplied with the square root of the
normal curvature in the corresponding direction.
First of all, the intrinsic geometry of the second fundamental

form, and in particular the curvature properties of this geometry, can be
studied. Sometimes simple requirements on this alternative geometry are
reflected in natural conditions on the shape of the hypersurface. Perhaps
the most illustrative result in this direction is a theorem due to R. Schneider
which characterises the spheres as the only ovaloids the second fundamen-
tal form of which has constant Gaussian curvature.
xi
xii SUMMARY.
But a second point of view can be adopted: it could be asked

how certain quantities which have been determined in the geometry of the
second fundamental form change in first-order approximation with respect
to small deformations of the hypersurface. It could be said that the ob-
jects which describe these first-order changes and are thus obtained by a
comparison of the geometry of the second fundamental form of the hyper-
surface with the corresponding geometry of neighbouring hypersurfaces,
belong to the “extrinsic geometry of the second fundamental form.”
By means of preparation for our investigations from the second

point of view, the first chapter contains a short overview on variational
formulae within the context of classical differential geometry. This enables
us to briefly state several well-known unicity results for the sphere, which
are interesting to compare with some of the subsequent results.
In the second chapter the second fundamental form geome-
try is studied for surfaces in the three-dimensional Euclidean space. Our
discussion of this geometry within the framework of the first approach is
mainly restricted to a description of known results. In regard of the sec-
ond approach, most attention will be paid to the variation of the II-area of
ovaloids, i.e., their area as measured by means of the second fundamental
form. In this context, three questions naturally arise:
— Are the spheres the only ovaloids of which the II-area is stationary with
respect to all deformations under which the enclosed volume
is preserved?
respect to all deformations under which the (classical) area is
preserved?
respect to all deformations under which the total mean curva-
ture is preserved?
xii
SUMMARY. xiii
As one of the new results which is contained in this chapter, theorem 2.20
on p. 43 could be mentioned. In this theorem a modified Jellett–Minkowski
integral formula is applied in order to provide an affirmative answer to the
last question.
In the two following shorter chapters some further variational
formulae are established.
Namely, in chapter 3 some classical variational formulae, which
hold in an abstract setting, are brought together. As a by-product, some
progress concerning the second problem can be given (p. 67).
In the fourth chapter the problem of how the shape operator
of a hypersurface is altered under deformations of the hypersurface is ad-
dressed. Although some minor applications are considered, this chapter is
mainly a technical preliminary which facilitates a calculation in the fifth
chapter.
In that chapter the geometry of the second fundamental form
is studied more generally for hypersurfaces in an arbitrary semi-Riemannian
manifold. Some of the theorems which will be uncovered in this chapter
extend well-known results for surfaces in the three-dimensional Euclidean
space.
Amongst others, an expression for the scalar curvature of the
second fundamental form is given in this setting.
In the same chapter the area functional associated to the sec-
ond fundamental form of hypersurfaces is studied more closely. The con-
cept of “mean curvature of the second fundamental form,” which was given
originally by E. Glässner for surfaces in the three-dimensional Euclidean
space, will be introduced in this general context.
It turns out that several relations between these curvatures
of the second fundamental form can be satisfied only by extrinsic hyper-
spheres. As such, some new characterisations of extrinsic hyperspheres are
obtained. For example, the extrinsic spheres are the only compact surfaces
in the three-dimensional hyperbolic space endowed with a positive-definite
second fundamental form of constant Gaussian curvature.
In the last section of that chapter, the technique of power se-
ries expansions will be used in order to study the II-areas of small geodesic
hyperspheres in a Riemannian manifold.
xiii
xiv SUMMARY.
In this section, the phenomenon that the low-dimensional

Euclidean spaces can be distinguished from non-flat spaces by means of
these II-areas, is discovered. More precisely, it will be shown that a Rie-
mannian manifold of dimension at most five, for which the area of every
geodesic hypersphere as determined by its second fundamental form coin-
cides with the corresponding area of a Euclidean hypersphere of the same
radius, is locally isometric to a Euclidean space.
In the last chapter an overview on the literature will be given.
The discussion on relative geometry includes modest progress concerning
the first-stated problem (p. 166).
xiv
NEDERLANDSTALIGE SAMENVATTING.
In dit proefschrift worden hyperoppervlakken met een niet-ontaarde twee-

de fundamentaalvorm onderzocht. Voor zo’n hyperoppervlakken kan de
tweede fundamentaalvorm op zichzelf als een abstracte semi-Riemannse
metriek geïnterpreteerd worden.
De meetkundige eigenschappen van zulke hyperoppervlakken
kunnen niet enkel tegenover de gewone meetkunde, dewelke bepaald
wordt door de eerste fundamentaalvorm, onderzocht worden; evenzeer
kan men de meetkunde van het hyperoppervlak tegenover zijn tweede fun-
damentaalvorm onderzoeken.
Het meest bekende voorbeeld van hyperoppervlakken die aan
de gestelde vereisten voldoen zijn de compacte oppervlakken in de driedi-
mensionale Euclidische ruimte die een convexiteitsvoorwaarde vervullen.
Deze worden ovaloïden genoemd. De meetkunde van een ovaloïde, zoals
bepaald door haar tweede fundamentaalvorm, wijkt van de gewone meet-
kunde af in de zin dat de lengte van een rakende vector vermenigvuldigd
dient te worden met de wortel van de normale kromming in de overeen-
komstige richting.
Vooreerst kunnen de intrinsieke meetkunde van de tweede fun-
damentaalvorm, en in het bijzonder de krommingseigenschappen van deze
meetkunde, bestudeerd worden. Soms worden eenvoudige eisen op de-
ze alternatieve meetkunde weerspiegeld in natuurlijke voorwaarden op de
vorm van het hyperoppervlak. Het resultaat in deze richting dat wellicht
het meest sprekend is werd door R. Schneider gevonden en karakteriseert
xv
xvi NEDERLANDSTALIGE SAMENVATTING.
de sferen als de enige ovaloïden waarvan de tweede fundamentaalvorm

constante Gausskromming heeft.
Daarnaast kan een tweede benaderingswijze aangeno-
men worden: men kan nagaan hoe bepaalde grootheden, die tegenover de
meetkunde van de tweede fundamentaalvorm bepaald werden, in eerste-
orde benadering veranderen tegenover kleine vervormingen van het hyper-
oppervlak. Men zou kunnen zeggen dat de grootheden die deze eerste-orde
veranderingen beschrijven en die dus verkregen werden door een vergelij-
king van de meetkunde van de tweede fundamentaalvorm van het hyper-
oppervlak met de overeenkomstige meetkunde van naburige hyperopper-
vlakken, tot de “extrinsieke meetkunde van de tweede fundamentaalvorm”
behoren.
In het eerste hoofdstuk wordt bij wijze van voorbereiding voor

onze navorsingen vanuit het tweede standpunt een kort overzicht gepre-
senteerd van formules uit de variatierekening in het kader van klassieke
differentiaalmeetkunde. Dit stelt ons in staat om enkele welbekende ka-
rakterisaties van sferen te bekomen, die vergeleken kunnen worden met
sommige ernavolgende resultaten.
In het tweede hoofdstuk wordt de door de tweede fundamen-
taalvorm bepaalde meetkunde bestudeerd voor oppervlakken in de driedi-
mensionale Euclidische ruimte. Onze discussie binnen het gedachtenkader
van de eerste benaderingswijze beperkt zich hoofdzakelijk tot een beschrij-
ving van reeds gekende resultaten. Betreffende de tweede benadering, zal
onze aandacht voornamelijk uitgaan naar de variatie van de II-oppervlakte
van ovaloïden. Met dit laatste wordt de oppervlakte van een ovaloïde, zo-
als opgemeten met behulp van haar tweede fundamentaalvorm, bedoeld.
De volgende drie vragen rijzen spontaan in deze context:
— Zijn de sferen de enige ovaloïden waarvan de II-oppervlakte stationair

blijft onder alle vervormingen die het omsloten volume behou-
den?
xvi
NEDERLANDSTALIGE SAMENVATTING. xvii

blijft onder alle vervormingen die de (klassieke) oppervlakte
behouden?

blijft onder alle vervormingen die de totale gemiddelde krom-
ming behouden?
Eén van de nieuwe resultaten die in dit hoofdstuk bewezen wordt, is stel-
ling 2.20 op blz. 43. In die stelling wordt een gewijzigde Jellett–Minkowski
integraalformule aangewend teneinde de laatste vraag positief te beant-
woorden.
In de twee volgende kortere hoofdstukken worden verdere va-
riatieformules aangetoond.
In hoofdstuk 3 zullen namelijk enkele klassieke variatieformu-
les, die in een abstracte context geldig zijn, samengebracht worden. Als
nevenproduct wordt vooruitgang aangaande het tweede probleem geboekt
(blz. 67).
In het vierde hoofdstuk wordt de vraag, hoe de vorm-operator
(“shape operator”) van een hyperoppervlak verandert onder deformaties
van het hyperoppervlak, behandeld. Hoewel enkele onmiddellijke toepas-
singen vermeld worden is dit hoofdstuk hoofdzakelijk een technische voor-
bereiding die een berekening in het vijfde hoofdstuk zal vergemakkelijken.
In dat hoofdstuk wordt de meetkunde van de tweede funda-
mentaalvorm van hyperoppervlakken meer algemeen in een willekeurige
semi-Riemannse ruimte bestudeerd. Enkele resultaten die in dit hoofdstuk
aan het licht gebracht worden, zijn uitbreidingen van welgekende resulta-
ten voor oppervlakken in de driedimensionale Euclidische ruimte.
Zo wordt ondermeer een formule voor de scalaire kromming
van de tweede fundamentaalvorm gegeven in dit kader.
In hetzelfde hoofdstuk wordt de oppervlaktefunctionaal die
aan de tweede fundamentaalvorm geassocieerd is, onderzocht. Het con-
cept “gemiddelde kromming van de tweede fundamentaalvorm,” dat oor-
spronkelijk door E. Glässner ingevoerd werd voor oppervlakken in de drie-
dimensionale Euclidische ruimte, zal in deze algemenere context geïntro-
duceerd worden.
xvii
xviii NEDERLANDSTALIGE SAMENVATTING.
Het blijkt dat aan verscheidene voorwaarden op deze krom-

mingen van de tweede fundamentaalvorm enkel door extrinsieke hypersfe-
ren voldaan kan worden. Op deze wijze worden enkele nieuwe kenmer-
kende eigenschappen voor de extrinsieke hypersferen bekomen. Zo zijn de
extrinsieke sferen de enige compacte oppervlakken in de driedimensionale
hyperbolische ruimte, dewelke een positief-definiete tweede fundamen-
taalvorm met constante Gausskromming hebben.
In het laatste deel van dat hoofdstuk zal de techniek van macht-
reeksontwikkelingen toegepast worden in de studie van de II-oppervlaktes
van kleine geodetische hypersferen in een Riemannse ruimte.
In dit deel wordt het fenomeen, dat de laagdimensionale Eu-
clidische ruimten aan de hand van deze II-oppervlaktes onderscheiden kun-
nen worden van de niet-platte ruimten, ontdekt. Meer precies wordt aan-
getoond dat een Riemannse ruimte, van dimensie ten hoogste vijf, waar-
voor de oppervlakte van elke geodetische hypersfeer zoals bepaald tegen-
over haar tweede fundamentaalvorm samenvalt met de overeenkomstige
oppervlakte van een Euclidische hypersfeer met eenzelfde straal, lokaal
isometrisch is met een Euclidische ruimte.
In het laatste hoofdstuk wordt een overzicht op de literatuur
gegeven. De discussie over relatieve differentiaalmeetkunde omvat enige
vooruitgang betreffende het eerstvernoemde probleem (blz. 166).
xviii
Chapter 1.
VARIATIONAL FORMULAE FOR

S URFACES IN THE T HREE -D IMENSIONAL
EUCLIDEAN SPACE.
The purpose of this chapter is to establish, by means of introduction, some

well-known theorems which are situated on the connection between the
calculus of variation and classical differential geometry. Of some of these
theorems, an adaption to the geometry of the second fundamental form
will be given in the following chapters.
1.1. Deformations of Surfaces.
Assume M ⊆ E3 is a compact surface with interior unit normal vector field

N . We consider a deformation
µ : ]−", "[ × M → E3 : (t, p) 7→ µ t (p) = µ(t, p)
of the surface. Here it is required that the function µ is smooth and that
µ0 (p) = p for all points p of the original surface M . It should be remarked
that the deforming mapping µ t , which sends the initial surface M onto
the deformed surface
µ t (M ) = µ t (p) | p ∈ M ,

1
2 1. VARIATIONAL FORMULAE FOR SURFACES IN E3 . 1.
l'4sd
(l'l)
rMr
Figure 1.1: The deformation vector is the initial velocity vector of the trajectory
which is described by a point moving under the deformation.
is a diffeomorphism for small values of |t|. The deformation vector field

Z ∈ X(M ) attaches to any point of the surface the vector of its initial motion
under the deformation:
∂

Z(p) = µ t (p) , (p ∈ M ) .
∂ t t=0
Thus, this vector is the tangent vector of the trajectory which is traced out
by the point p under the deformation (see figure 1.1). Here the collection
of all vector fields on M which take values in E3 has been denoted by X(M ).
However, should the vector field Z be tangent to the surface in any of its
points, then the deformation will be called a tangent deformation. The
concept of a normal deformation is defined analogously.
We will denote I(µ t ) for the first fundamental form of µ t (M ),
which is the restriction of the Euclidean scalar product 〈· , ·〉 to the tan-
gent spaces of the surface. A similar notation will be adopted for all other
tensors defined on µ t (M ). The unit normal vector field of µ t (M ) will be
2
1.1. DEFORMATIONS OF SURFACES. 1.1. 3
denoted by Nt .
If we are to investigate the way a tensor changes under the
deformation, the way tensors on µ t (M ) shall be compared with tensors on
Ö) by
M has to be laid down. Herefore we define I(µ t
Ö) : X(M ) × X(M ) → F(M )

I(µ t
: (V, W ) 7→ I(µ t ) dµ t (V ), dµ t (W ) = dµ t (V ), dµ t (W ) .

Here X(M ) and F(M ) are defined, respectively, as the set of all vector fields
on the surface M and the set of all functions mapping M in R. Then the
variation of the first fundamental form along the deformation µ is defined
as the tensor, sending a pair (V, W ) of vector fields to the function
∂

Ö)(V, W ) .
I(µ t
∂ t t=0
For all other tensors B which are defined on the set of all surfaces in E3 the
similar convention
×) = µ∗ B(µ )

B(µ t t t
will be made so as to obtain a tensor on M ; also here,

the variation of
∂
the tensor along the deformation µ is defined as ∂ t B(µ t ). Hence by
×
t=0
an application of the tangent mapping of the deforming mapping µ t the
variation of a tensor could be defined as an ordinary partial derivative in
the tangent spaces to M .
Assume to any surface M in E3 a tensor B on M is assigned. It
will be said that B is smooth if smooth functions Fıab······ exist such that for
any parametrisation x : R2 → E3 of any surface, the co-ordinate coefficients
Bıab······ = B(xi , x j , · · · , dxa , dx b , · · · )
of B can be expressed as
= Fıab······ (x, xu , x v , xuu , · · · ) . (1.1)
For example, for the co-ordinate coefficient II1 1 of the second fundamental
form, there holds
II1 1 = xuu , N

3
xu × x v

= xuu ,
kxu × x v k
and this obviously can be expressed as
= F1 1 (x, xu , x v , xuu )
where
4 the real-valued function F1 1 , which is defined on an open domain of
R3 , is given by
w×x

F1 1 : (v, w, x, y) 7→ y, .
kw × xk
It seems noteworthy to clearly state a fundamental and easy

result.
Theorem 1.1 ( R-Linearity of the Variation-Operator).

Assume to any surface in E3 , a tensor B on M is assigned in a smooth
way. Let three deformations µ, µ] and µ¶ be given and denote their re-
spective deformation vector fields by Z, Z ] and Z ¶ . If the deformation
vector fields are related by
Z = α Z] + β Z¶ (for real numbers α and β) ,
then the variations of B along these deformations are likewise related

by
∂ ∂ ∂

×) = α ] ¶
B(µ B(µ t ) + β B(µ t).
× ×
t
∂t
t=0 ∂t
t=0 ∂t
t=0
Proof of Theorem 1.1. This basically follows from the fact that the varia-
tion of a tensor was constructed as a partial derivative.
For a parametrisation x : R2 → M ⊆ E3 of a surface M , the
composition µ t ◦ x with the deforming mapping gives us a parametrisation
of the deformed surface µ t (M ). For these parametrisations of the surfaces,
the co-ordinates of the varied tensor are nothing but the partial derivatives
4
of the tensor co-ordinates:

a b ···
∂ ∂

a b ···
×)
B(µ = B(µ t ) ı  ··· .
t
∂ t t=0

ı  ···
∂ t t=0

(In the left-hand side, co-ordinate indices with respect to the parametrisa-
tion x of M appear. The co-ordinate indices after the partial derivative in
the right-hand side refer to the parametrisation µ t ◦x of the surface µ t (M ).)
By means of (1.1), this can be re-written as
= d(Fıab······ )(x,xu ,xv ,xuu ,··· ) Z, Zu , Z v , Zuu , · · · ,

∂2
where Zuu = ∂ u2
(Z ◦ x) and so forth. The result follows from the above
expression.
Since the variation of a tensor B is completely determined by

the deformation vector field Z, it is justified to denote this variation as δ Z B
(which will often be abbreviated to δB).
Remark. Many further properties of the δ-operator are obvious. For ex-
ample, the Leibniz rule δ (B ⊗ C) = (δB) ⊗ C + B ⊗ (δB) is satisfied.
Let us briefly recall that the shape operator of M is defined by
A : X(M ) → X(M ) : V 7→ −DV N ,
where D stands for the operation of partial derivation, i.e., the standard
connection of E3 ; that the eigenvalues of the shape operator are called
the principal curvatures λ1 and λ2 ∈ F(M ), whereas the corresponding
eigenvectors, the principal directions, are denoted by e1 and e2 ∈ X(M );
and that the mean curvature H and the Gaussian curvature K are defined
respectively as the arithmetic mean and the product of these principal cur-
vatures.
A first consequence of the above theorem 1.1 is that the vari-
ations of the mean curvature H and the Gaussian curvature K of a surface
along a tangent deformation can simply be expressed as partial derivatives
5
along the deformation vector field. More generally, the variation of any
tensor B along such a tangent deformation is given by the Lie derivative of
this tensor with respect to the deformation vector field:

δ Z H = Z [H] 
δ Z K = Z [K] (Z tangent to M ).
δ Z B = L Z (B)

Furthermore by the linearity of the δ-operation it will be sufficient to re-

strict attention to linear deformations in the study of first-order variational
problems. By a linear deformation will be understood a deformation in
which every point of the surface moves steadily in a fixed direction with
growing t:
µ : ]−", "[ × M → E3 : (t, p) 7→ µ t (p) = µ(t, p) = p + t Z(p) . (1.2)
A vector field W ∈ X(M ) is transferred to µ t (M ) according to
dµ t (W ) = W + tDW Z (as free vectors). (1.3)
It immediately follows that the variation satisfies

 ¬ ¶
 I(µ t ) (V, W ) = V + tDV Z , W + tDW Z ;
Ö
 ¬ ¶ ¬ ¶
δI (V, W ) = V, DW Z + DV Z, W .
In a E3 -neighbourhood of a point on M where the deformation

vector does not vanish, a vector field N exists which extends all unit normal
vector fields Nt of the surfaces µ t (M ) (see figure 1.2).
Another straightforward corollary of equation (1.3) is that for
every W ∈ X(M ) and every p ∈ M
∂

¬ ¶ ¬ ¶ ¬ ¶
0= dµ t W(p) , N(µ t (p)) = DW Z, N + W, D Z N (p)
∂ t t=0

and consequently (denoting δN for the vector field D Z N ∈ X(M ) ),

¬ ¶ ¬ ¶ ¬ ¶ ¬ ¶
δN = D Z N , e1 e1 + D Z N , e2 e2 = − De1 Z, N e1 − De2 Z, N e2 . (1.4)
6
1
{
\
Figure 1.2: The unit vector field N is normal to the surfaces µ t (M ) everywhere.
Lemma 1.2.
The
i
. \ linear
.'f
l '! t
t deformation (1.2) satisfies, for every W ∈ X(M ),
'r-J
t , t r t
δA (W ) = −DW (δN ) − DA(W ) Z .

(1.5)
Proof of Lemma 1.2. Let p denote an arbitrary point of the surface M .

Its image under the deforming mapping µ t will be denoted by q = µ t (p) ∈
µ t (M ). If wq is a vector ofTq E3 , the vector of T p E3 which is parallel to wq

will be denoted by wq y . Let Z † be an extension of Z to a vector field
p
which is defined on an open neighbourhood of M . The tangent mapping of
µt
dµ t : T p M → Tq µ t (M )
7
can be extended to the invertible linear map


dµ†t : T p E3 → Tq E3 : vp 7→ vp + tD vp Z † y .

q
The inverse of this map is given by1


† ←

dµ t 3 3
: Tq E → T p E : wq 7→ wq y − tD (wq )y Z † + O (t 2 )

p p
and clearly extends the inverse of the tangent mapping dµ t . Now choose
a curve α : R 7→ M : s 7→ α(s) which passes with velocity Wp through p for
parameter value 0.
 
∂

δA (Wp ) = − dµ← Ddµ t (Wp ) N 

∂ t t=0 t
 
∂

= − dµ← D ∂ N
∂ t t=0 t µ (α(s))
∂ s s=0 t
 
∂ ← ∂

†
= − dµ t  Nµ (α(s)) 
∂ t t=0 ∂ s t
s=0
This last derivative can be split in two parts. We obtain
δA (Wp )

   
 
∂ ∂ ∂
    
† ←
  
=− dµ N + N
   
α(s) µ (α(s))
 
∂ t t
∂ s s=0 ∂ s s=0

  t
 

t=0 
 y  y 
q p
= −DA(Wp ) Z − DWp (δN ) .
Let ∇ stand for the Levi-Civita connection of the surface. The Codazzi
equation expresses that ∇A is a symmetric tensor:

 e2 λ1 e1 = λ2 − λ1 ∇e1 e2 ;

(1.6)
e1 λ2 e2 = λ1 − λ2 ∇e2 e1 .

1
The inverse of a bijection f : A → B will be denoted as f ← : B → A .
8
The following expression for the variation of the shape opera-

tor can be obtained from (1.4–1.6):
e2 λ 1 ¬
¨ «
¬ ¶ ¶ ¬ ¶
δA (e1 ) = De1 De1 Z, N +

De2 Z, N − 2λ1 De1 Z, e1 e1
λ2 − λ 1
e2 λ 1 ¬
¨
¶ ¬ ¶
+ De1 Z, N + De1 De2 Z, N
λ1 − λ2
«
¬ ¶ ¬ ¶
− λ1 De2 Z, e1 + De1 Z, e2 e2 . (1.7)
The vector fields


(µ ) = d(µ← t ) e1 (µ t ) ∈ X(M ) ;

eØ
 1 t

←
2 (µ t ) = d(µ t ) e2 (µ t ) ∈ X(M ) ,

 eØ


can be expanded in the basis e1 , e2 :

(µ ) = e1 + tζ1 e2 + O (t 2 ) ;
eØ
 1 t

2 (µ t ) = tζ2 e1 + e2 + O (t ) ,
 eØ
 2
where ζ1 , ζ2 ∈ F(M ). If the variation is taken of both sides of the equation

A(µ t )e1 (µ t ) = λ1 (µ t )e1 (µ t ), there results
δA (e1 ) = δλ1 e1 + λ1 − λ2 ζ1 e2 .

Consequently, equation (1.7) can be written as
e2 λ 1 ¬

¬ ¶ ¶ ¬ ¶
δλ1 = De1 De1 Z, N + De2 Z, N − 2λ1 De1 Z, e1 ;
λ2 − λ1






e2 λ 1 ¬

¶ ¬ ¶
λ1 − λ2 ζ 1 = De1 Z, N + De1 De2 Z, N


λ1 − λ2



 ¬ ¶ ¬ ¶
−λ1 De2 Z, e1 + De1 Z, e2 .
9
t n
Let now vp and vp stand for the tangent and the normal part of a
vector vp ∈ T p M . If the deformation vector field is decomposed as Z =
f N + Z t , the first equation can be rewritten as
2
δλ1 = Hess f (e1 , e1 ) + f λ1 + Z t [λ1 ] .
This implies the following theorem.

(In this work the following sign convention on the Laplacian
will be adopted: the Laplacian of a function is defined as the trace of its
Hessian. Thus there holds ∆ f = f 00 for a function f : R 7→ R.)
Theorem 1.3 (Variation of the Gaussian and Mean Curvature).

The variation of the Gaussian and mean curvature are given by:

1

 δH = 2
∆ f + 2H 2
− K f + Z t [H] ;
δK = λ2 Hess f (e1 , e1 ) + λ1 Hess f (e2 , e2 ) + 2K H f + Z t [K] .


(1.8)
Ú) = η dΩ, where η (p) is the

It is also self-evident that dΩ(µ t t t
area of the parallellogram spanned by dµ t (e1 (p)) and dµ t (e2 (p)). A further
calculation which makes use of (1.3) yields

δdΩ = div Z t − 2H f dΩ . (1.9)
As a corollary we obtain two classical integral formulae.
Corollary 1.4 (Divergence Theorem).

Let a compact surface M ⊆ E3 and a vector field X ∈ X(M ) be given.
There holds Z
div X dΩ = 0 . (1.10)
Proof of Corollary 1.4. Let µ : ]−", "[ × M → M : (t, p) 7→ µ t (p) be the
10
flow of the vector field X on the surface M . Since any of these mappings
µ t (at least for small |t|) is a diffeomorphism of M onto M , equation (1.9)
gives
∂
Z
0= Area µ t (M ) = div X dΩ .

∂ t t=0
Corollary 1.5.
Let a compact surface M ⊆ E3 and a vector field X ∈ X(M ) be given.
There holds
Z
¬ ¶ ¬ ¶
λ2 ∇e1 X , e1 + λ1 ∇e2 X , e2 dΩ = 0 . (1.11)
R
Proof of Corollary 1.5. The formula expresses that 0 = div A(X ) dΩ.
Remark. Hence it can be concluded R from theorem 1.3 and corollary 1.5
that the total Gaussian curvature K dΩ does not alter under deforma-
tions. This fact was already noticed by Poisson prior to the discovery of the
Gauss–Bonnet theorem.
If the integral formulae (1.10) and (1.11) are applied to the

vector field grad f g (for f , g ∈ F(M )), and the Codazzi equation is taken

into account, there results

 Z Z

 f ∆g dΩ = g∆ f dΩ ;





 Z

 λ2 f Hess g (e1 , e1 ) + λ1 f Hess g (e2 , e2 ) dΩ



 Z
= λ2 g Hess f (e1 , e1 ) + λ1 g Hess f (e2 , e2 ) dΩ .


A consequence is the following theorem. For clarity we will first recall a

common definition.
11
Assume that to any surface M ⊆ E3 a real number F (M ) has

been associated in such a way that the composition F ◦ µ of F with any
deformation µ of any surface in E3 is smooth. Let M ⊆ E3 be a surface, and
µ a deformation thereof. The (first) variation of the functional F along
the deformation µ is denoted by and defined as
0
δF (M ) = F ◦ µ (0) .
(The more elaborate notation δ Z F (M ) will seldom be used.)
Theorem 1.6.
Let F : (u, v) 7→ F (u, v) be a function of two variables. Let a defor-
mation µ of a compact surface M in E3 withRdeformation vector field
Z = Z t + f N be given. The first variation of F (H, K) dΩ is given by
Z Z ¨
1
δ F (H, K) dΩ = ∆ Fu (H, K) + Fu (H, K) 2H 2 − K

f
2
+ λ2 Hess(Fv (H,K)) (e1 , e1 ) + λ1 Hess(Fv (H,K)) (e2 , e2 )
«
+ 2 H K F v (H, K) − 2 F (H, K) H dΩ .
1.2. Applications to Integral Formulae and Unicity

Results for the Sphere.
As a first application of theorem 1.6 we will provide three proofs of two

classical integral formulae. These proofs are not independent but should
rather be seen as “variations on a theme.” The position vector field P of
E3 assigns to every point p in E3 the free vector p (with respect to some
chosen origin). The scalar product
〈−N , P〉 : M → R
12
1.2. APPLICATIONS TO I NTEGRAL FORMULAE. . . 1.2. 13
which gives the signed distance of the tangent plane to the origin is called
the support function (see figure 1.3).
Theorem 1.7 (Integral Formulae of Jellett and Minkowski).

Every compact surface M ∈ E3 satisfies the following integral formu-
lae:  Z

 Area(M ) = H 〈−N , P〉 dΩ ;


Z Z (1.12)


HdΩ =

 K 〈−N , P〉 dΩ .
Historical Remark. It is interesting to notice that these integral formu-

lae of Minkowski, as they are often called, have been obtained already by
Jellett [Jellett 1853] (see also [Bonnet 1853]). The fact that the support
function, which is one of the main ingredients of these integral formulae,
has been employed already in Hamilton’s article [Hamilton 1833], suggests
that it might have been Hamilton’s influence which made the other mathe-
matician of Dublin find the formulae.
Proof of Theorem 1.7 (First Version). By deforming M in the direction

of the position vector field, we obtain the hypersurface
¦ ©
µ t (M ) = p + t P(p) | p ∈ M ,
which is a rescaling of M with a factor 1 + t. Such a homothety magnifies

areas with a factor (1 + t)2 and consequently,
δArea(M ) = 2 Area(M ) .
If, on the other hand, µ t (M ) is considered as a deformation of M with

deformation vector field P, theorem 1.6 gives us

Z
δArea(M ) = 2 H 〈−N , P〉 dΩ .
13
|r)
=
HJ
UI
+-
^
tt
I
/
I
r'
l-
otr
E
4.
V
a
I
I
I
a-
^
-
l
',
'l
JI
\
\
I
t'
:
F
Figure 1.3: The support function.

{
}-*
The first integral formula is a consequence of these two lastR equalities. The
second integral inequality similarly follows if the fact that H dΩ is homo-
geneous of the first order with respect to homothetic deformations is taken
into account.
Proof of Theorem 1.7 (Second Version). The integral formulae can be

obtained by a substitution of the tangent part of the position vector field
for the vector field X in corollaries 1.4 and 1.5. Here the small calculation
below can be useful (for W ∈ X(M )):
¬ ¶ ¬ ¶ ¬ ¶ ¬ ¶
∇W P t , W = DW P t , W = DW P, W − DW (〈P, N 〉 N ) , W
¬ ¶
= 〈W, W 〉 − 〈P, N 〉 DW N , W = 〈W, W 〉 + 〈P, N 〉 〈A(W ), W 〉 .
Proof of Theorem 1.7 (Third Version). An application of the divergence

theorem to the compact region M of E3 which is bounded by M gives
Z Z
〈−N , P〉 dΩ = div P dΩ = 3 Vol(M ) ,
M M
where quantities with a bar refer to the surrounding Euclidean geometry.

Here as well as in the sequel a slight inconsistency will enter in our nota-
14
tion, since Vol(M ) will actually stand for the volume of the set M which
is bounded by M . The first formula (1.12) is obtained by an investigation
of the behaviour of these quantities under the unit normal deformation,
which is the linear deformation determined by the unit normal vector field.
This deformation sends the original surface into equidistant surfaces. Now
the first formula expresses that
Z
δN 〈−N , P〉 dΩ = δN (3 Vol(M )) .
M
The second formula follows similarly from an application of the operator

δN to both sides of the first integral formula.
The three theorems below are an immediate corollary of the

integral formulae of Jellett and Minkowski. By an ovaloid we will under-
stand a compact surface in E3 satisfying K > 0. For a given ovaloid, the
origin can always be chosen in such a way that the support function is
strictly positive.
Corollary 1.8 (Christoffel).

H
The spheres are the only ovaloids in E3 for which the ratio K
is a
constant.
Corollary 1.9 (Jellett–Liebmann).

The spheres are the only ovaloids in E3 with constant mean curvature.
Corollary 1.10 (Liebmann–Hilbert).

The spheres are the only compact surfaces in E3 with constant Gauss-
ian curvature.
All three of these theorems characterise the sphere as the

unique solution of a certain
R variational problem. In this work, the integral
of the mean curvature H dΩ will be called the total mean curvature.
15
Corollary 1.11 (Variational Rewording of Corollary 1.8).

The spheres are the only ovaloids which are a critical point of the total
mean curvature functional with respect to deformations under which
the area is preserved.

The spheres are the only ovaloids which are a critical point of the area
functional with respect to deformations under which the enclosed vol-
ume is preserved.

The spheres are the only critical points of the total mean curvature
functional with respect to deformations under which the enclosed vol-
ume is preserved.
Remark. The Hilbert–Liebmann theorem also gives us that an isometry of

surfaces between an arbitrary ovaloid and a sphere is necessarily a congru-
ence. Thus it can be said that the sphere is determined by its first funda-
mental form. This fact is referred to as the “rigidity of the sphere.”
As another application of corollaries 1.4 and 1.5, we can give

two obvious unicity results for the sphere.
Corollary 1.14.
The spheres are the only ovaloids in E3 which
R are a critical point of
the total squared mean curvature functional H 2 dΩ.
Proof of Corollary 1.14. An ovaloid which is a critical point of this func-
16
tional satisfies the corresponding Euler-Lagrange equation

∆H = −2H H 2 − K
which can be derived from theorem 1.6. An application of the divergence

theorem gives that the integral of the non-negative function H H 2 − K
vanishes, and this is only possible if H 2 = K throughout on M .
ItRshould
be remarked
that the above functional is variationally
equivalent to H 2 − K dΩ. Since a point is umbilical if and only if the
local quadratic approximation of the surface is rotationally symmetric, this
functional can be seen as a measure for the global lack of local rotational
symmetry. The corollary gives us that the only totally umbilical surface is the
R p 2
only critical point of this “total umbilicity deficiency.” But H − K dΩ
can be ascribed to measure this deficiency as well, and the corresponding
result still holds.
Corollary 1.15.
The spheres are the only ovaloids in E3 which are a critical point of
R p 2
the functional H − K dΩ.
Proof of Corollary 1.15. The corresponding Euler-Lagrange equation is

now
p p
∆ H − K + 2 H − K H2 − K
= λ2 Hess pH (e1 , e1 ) + λ1 Hess pH (e2 , e2 )
K K
and the theorem follows from an application of corollaries 1.4–1.5.
Bibliographical Notes for Chapter 1. The formulae for δH and δK are already described
in, e.g., [Blaschke 1923] § 117 and [Voss 1956]. With respect to theorem 1.6, see also,
e.g., [Reilly 1973] and [Nitsche 1993] and [Rosso–Virga 1999]. See also, e.g., [Minkowski
1901, Minkowski 1903, Osserman 1990] for the integral formulae. A theorem much more
general than corollary 1.8 was proved in [Christoffel 1865] (see also [Liebmann 1899b]).
17
For corollaries 1.8–1.10 see [Jellett 1853, Liebmann 1899a, Hilbert 1901, Amur 1971,
Nakajima 1926, Chern 1945]. For textbooks on the present topic, see, e.g., [Blaschke 1923,
Klingenberg 1978] and [Montiel–Ros 2005].
18
Chapter 2.
T H E G E O M E T RY O F T H E S E C O N D
FUNDAMENTAL FORM OF AN OVALOID IN
T H E T H R E E -D I M E N S I O N A L E U C L I D E A N
SPACE.
2.1. Introduction.
The second fundamental form is a traditional object of study in the classical
differential geometry of surfaces in the Euclidean space E3 . Less traditional
is the study of this form from the metrical point of view: the second funda-
mental form can be regarded as a Riemannian metric on an ovaloid.
The corresponding geometry deviates from the usual geometry
in the sense that the length of a tangent vector should be multiplied with
the square root of the normal curvature in the corresponding direction. The
present chapter focusses on the geometry which arises in this way.
In the first § 2.2, we recall how the second fundamental form
is involved in several common descriptions of the shape of a surface in E3 .
In the subsequent part of this chapter, the second fundamental
form will be seen as an abstract metrical tensor on an ovaloid, and the
corresponding geometry is the actual object of study.
In analogy with the classical study of the geometry of surfaces,
a distinction can be made here between the “intrinsic geometry of the sec-
ond fundamental form,” which is determined by measurements of II-lengths
19
20 2. THE GEOMETRY OF THE SECOND FUNDAMENTAL FORM OF M ⊆ E3 . 2.
on the ovaloid only, and the “extrinsic geometry of the second fundamental
form,” which is constituted of all measurements in which the geometry of
the second fundamental form is compared with the corresponding geome-
try of nearby surfaces.
The chief invariant of the intrinsic geometry of the second fun-
damental form is its Gaussian curvature.
As for the extrinsic geometry of the second fundamental form,
we will mainly devote our attention to the comparison of the areas of neigh-
bouring surfaces, as measured by means of their second fundamental form.
2.2. Four Geometrical Interpretations of the Second

Fundamental Form.
Let M ⊆ E3 be a surface, and N a unit normal vector field on M . The

second fundamental form is the symmetric bilinear form defined by
II : X(M ) × X(M ) → F(M ) : (V, W ) 7→ 〈A(V ), W 〉 .
The second fundamental form of a surface in the Euclidean

space measures the change of the unit normal direction from point to point
on the surface. More precisely, suppose V and W are tangent vector fields
on a surface M ⊆ E3 . If γ : R → M is an integral curve of V , there holds
(see figure 2.1)
∂
® ¸
II V(γ(0)) , W(γ(0)) = −
N(γ(t)) , W(γ(0)) .
∂ t t=0
Hence this form records the twisting of the tangent plane along
different directions which are tangent to the surface.
According to Meusnier’s theorem, the second fundamental
form describes the normal component of curvature vectors of curves which
are situated entirely on the surface. Thus, as is indicated in figure 2.2, for
an arcwise parametrised curve γ : R → M ⊆ E3 , the normal component of
the curvature vector k = Dγ0 γ0 of this curve as seen in E3 is given by
n
k = II(γ0 , γ0 ) N .
20
2.2. FOUR GEOMETRICAL INTERPRETATIONS OF THE SECOND. . . 2.2. 21
Figure 2.1: The second fundamental form

normal vector field.
( il,=,
AIW,ht) = ' keeps trackAIl
of the twisting of the unit
ND ,,w)
,/t
-l
I
v
f
MsEg
Figure 2.2: The second fundamental form describes the normal components of
curvature vectors. N,1,
Another characteristic property of the second5=flot

fundamental
form makes use of the distance function ψ : E → R to the surface. (This
3
I
distance ψ(q) between cUUntUU.t
the point qectn
and rt.
the surface M should be counted
4-?
21
an(fr)
t4sE3
PlgE'
PlgE'
4 \
Tfll
Figure 2.3: The second fundamental form describes the deviation of the surface
from its tangent surface.
negative if the shortest connection lies on the opposite side of the tangent
plane from the unit normal.) Then, up to the sign, the restriction of the
Hessian of ψ to T p M × T p M is (for every p ∈ M ) the second fundamental
form of M at p. (A generalisation of this assertion is proved on page 77.)
This can also be stated in a slightly different way. For every
vector vp ∈ T p M (where the tangent plane T p M is seen as a plane in E3 ),
2
there holds ψ(svp ) = − s2 II(vp , vp ) + O (s3 ). This can be summarised by
saying that the deviation of the surface from its tangent plane is described
by the second fundamental form (see figure 2.3).
Finally, the second fundamental form describes the difference
between the first fundamental form of the surface and that of an equidis-
tant surface at a short distance t. Namely, the variation of the first fun-
MsE,
damental form under the unit normal deformation (see figure 2.4), which
arises if each point of the surface evolves along its unit normal vector, is
given by
δN I = −2 II.
22
,/ttl*4;
2.3. THE SECOND FUNDAMENTAL FORM AS A METRIC . . . 2.3. 23
MsE,
,/ttl*4;
Figure 2.4: The second fundamental form as the unit normal variation of the first
fundamental form.
(A generalisation of this assertion is proved on page 76.)
2.3. The Second Fundamental Form as a Metric and

the Corresponding Intrinsic Geometry.
2.3.1. Introduction.
We will now turn our attention to compact surfaces in the Euclidean space
with a positive-definite second fundamental form. Such surfaces will be
called ovaloids. This condition means that the Gaussian curvature is strictly
positive and that the unit normal vector is chosen so as to make both prin-
cipal curvatures positive. (It should be noted that we need to choose the
interior unit normal vector field on an ovaloid in order to make the second
fundamental form positive-definite.)
This condition implies that the surface is strictly convex in the
sense that it meets every of its tangent planes only in one point. (This is not
a characterisation, since a strictly convex surface may have planar points.
Consider for example the surface given by x 6 + y 6 + z 6 = 1.)
23
Since the second fundamental form of an ovaloid M is positive-

definite, it furnishes an abstract metrical tensor on M . The interplay be-
tween the geometry of this abstract metrical tensor and the shape of the
original ovaloid is one of the central themes in this booklet.
All geometrical objects of Riemannian geometry can be con-
structed on M either in the usual way, or with respect to the abstract Rie-
mannian metric II. To distinguish them, we will attach an upper or lower
index II to the objects which are constructed in the latter way.
Let us first ask how the geometry of a tangent space of a sur-
face is altered if lengths are measured by means of the second fundamental
form. The indicatrix of Dupin is now seen as the unit circle in this tangent
plane, and conjugate directions will be seen as orthogonal directions. The
principal directions e1 and e2 span a rectangle in the geometry of the sec-
ond fundamental form, since

 II(e1 , e1 ) = λ1 ;
II(e1 , e2 ) = 0 ;
II(e2 , e2 ) = λ2 .

p
The area of this rectangle is K and this implies that the area element
p
determined by the second fundamental form is given by dΩII = K dΩ.
Furthermore, V1 = p1 e1 and V2 = p1 e2 provide us with a II-
λ1 λ2
orthonormal frame field on M (at least, if umbilics are omitted). Similarly,
W1 = λ1 e1 and W2 = λ1 e2 form a III-orthonormal frame field on M . Here
1 2
the third fundamental form is defined by III(V, W ) = 〈A(V ) , A(W )〉.
The connections of the second and the third fundamental form
are given by   
←
 ∇V W = A ∇V (A(W )) ;
 III

(2.1)



1

∇IIV W = 2
∇V W + ∇IIIV W ,
where V and W are tangent vector fields on M . These facts were found by
Weingarten (see, e.g., [Bianchi 1894], § 85). A convenient way to establish
the above relation is to verify that the mappings sending the pair (V, W )
to the right-hand side of the above equations satisfy the five conditions
24
(e.g., as stated in [O’Neill 1983], pp. 59–61) which uniquely determine

the Levi-Civita connection of (M , II) resp. (M , III).
It then follows that

H
 ∇V Vi = K ∇ei ei ;
II
i
1
W Wi =
∇III

∇ e
K ei i
.
i
2.3.2. Some Technical Results.

In this §, we will collect some basic formulae which will be used in the
sequel. The following fact is obvious.
Lemma 2.1.
A function f ∈ F(M ) on an ovaloid M satisfies
A gradII f = grad f .

The difference tensor between the connection of the first and

the second fundamental form will be denoted by L:
L : X(M ) × X(M ) → X(M ) : (V, W ) 7→ ∇IIV W − ∇V W .
It is obvious that L is a symmetric tensor. Further, it follows from Wein-

garten’s equations (2.1) that
1
L(V, W ) = A← ∇V A (W ) .

2
This last expression establishes in turn the following symmetry property:
II(L(V, W ), Z) = II(L(V, Z), W )
for all V, W, Z ∈ X(M ). The following notation will be adopted for the
different traces of the tensor L:

 traceL = trace {V 7→ L(V, ·)} ∈ Λ(M ) ;
trII L = trII {(V, W ) 7→ L(V, W )} ∈ X(M ) ;
tr L = tr {(V, W ) 7→ L(V, W )} ∈ X(M ) .

25
Here Λ(M ) stands for the collection of all differential forms (of first order)
on the surface. The natural logarithm (with base e) will be denoted as log.
The trace of an operator X : X(M ) 7→ X(M ) will be denoted by trace X in
this work, whereas a metric contraction of a tensor Y will be denoted by
tr Y or trII Y (depending on which metric is involved in the contraction).
Lemma 2.2.
The traces of the difference tensor L are given by
1

traceL = d log K ;
2





 1
trII L = gradII log K ;


 2


trL = gradII H .

Proof of Lemma 2.2. The third expression follows from the fact that every
tangent vector field X on M satisfies
¬ ¶ ¬ ¶
2 II (tr L, X ) = ∇e1 A e1 , X + ∇e2 A e2 , X
¬ ¶ ¬ ¶
= e1 , ∇e1 A X + e2 , ∇e2 A X
= e1 , ∇X A e1 + e2 , ∇X A e2

= e1 , ∇X A e1 − A(∇X e1 ) + e2 , ∇X A e2 − A(∇X e2 )

= e1 , X [λ1 ]e1 + e2 , X [λ2 ]e2

= X [λ1 ] + X [λ2 ]
= 2 II(X , gradII H) .
The other expressions can be obtained similarly.
Lemma 2.3.
Every vector field X ∈ X(M ) satisfies
26
divII X = div X + (traceL)(X ) .
Proof of Lemma 2.3. This is a consequence of

¦ ©
divII X = trace V 7→ ∇IIV X
= trace V 7→ ∇V X + trace {V 7→ L(V, X )} .

The squared norm of the tensor L with respect to the second

fundamental form is defined as
 2
X
II(L, L) = II L(Vı , V) , Vk  .
ı k
In the following lemma, which is immediate, the Hessian of the surround-

ing Euclidean space is denoted with a bar.
Lemma 2.4.
Assume a function F : E3 → R is given. Let f stand for the restriction
of F to a surface M ⊆ E3 . On X(M ) × X(M ) there holds
Hess f = Hess F + N [F ]II .
In accordance with [O’Neill 1983] the following vector field

which is obtained by contraction of the covariant derivative of a symmetric
(1,1)-tensor B will be called a divergence of B:

div B = ∇e1 B e1 + ∇e2 B e2 .
It immediately follows that

¬ ¶ ¬ ¶
div (B(X )) = 〈div B, X 〉 + B(∇e1 X ), e1 + B(∇e2 X ), e2 .
for a vector field X ∈ X(M ). Similarly a contraction of the covariant deriva-

tive of a tensor h (where the contraction is taken on an original slot of h
and the new slot of ∇h) will be called a divergence of h.
27
Lemma 2.5.
The shape operator satisfies

 div A = 2 grad H ;
div A← = − gradII log K .


Lemma 2.6.
For a function f ∈ F(M ), defined on an ovaloid M ⊆ E3 , there holds
∆II f = trII Hess f −II(gradII f , trII L) .
Proof of Lemma 2.6. This is a standard calculation:

2 ¬
X ¶
trII Hess f = ∇Vi grad f , Vi
i=1
2 ¦
X ¬ ¶©
=

Vi grad f , Vi − grad f , ∇Vi Vi

i=1
2 ¦
X ©
= Vi II(gradII f , Vi ) − II(gradII f , ∇Vi Vi )

i=1
X2 n o
= II(∇IIV gradII f , Vi ) + II(gradII f , L(Vi , Vi ))
i
i=1
= ∆II f + II gradII f , trII L .

2.3.3. The Gaussian curvature of the Second Fundamental Form.
In the context of an ovaloid’s geometry as determined by its second funda-

mental form, a study of the curvature of this metric spontaneously catches
28
our attention. The intrinsic (Gaussian) curvature of the second funda-

mental form will be denoted by KII . This curvature can be expressed as
1 1
KII = H + II(L, L) − II(gradII K, gradII K) , (2.2)
2 8K 2
or alternatively
1 λ2 λ1

=H+ e1 [λ2 ] e1 log + e2 [λ1 ] e2 log ,
4K λ1 λ2
or yet
 2
K H H KX eı e 

=H− II gradII , gradII +  L , .

λı λ 

2 K K 2 ı

II
(See respectively [Schneider 1972] (or a further chapter, p. 101), [Cartan

1943] and [Baikoussis–Koufogiorgos 1987a].) Erard has also established
“formula (Q)”
2 ® 2 ¸
K H 1 H
II gradII , gradII H − grad , grad K (Q)
2 K 4 K
= 2H KII − H (H 2 − K) .

Let us conclude this section by establishing a characteristic property for the

sphere [Schneider 1972].
Theorem 2.7 (Schneider).

The spheres are the only ovaloids for which the curvature of the sec-
ond fundamental form is constant.
Proof of Theorem 2.7. Assume an ovaloid M ⊆ E3 satisfying KII = C ∈ R+

is given. Let K attain its maximal value at the point p ∈ M . It can be seen
from expression (2.2) that every q ∈ M satisfies
 
p II gradII K, gradII K p p
C = KII ( p) ¾ K(p) −  2
 = K(p) ¾ K(q) .
8K
(p)
29
p
Thus C ¾ K and hence
Z Z Z
p
4π = KII dΩII = C K dΩ ¾ K dΩ = 4π .
p
This is only possible if C = K which implies that M is a sphere.
Remark. This theorem, which characterises the sphere by the constancy

of a certain curvature, can be seen as an adaption of the Liebmann–Hilbert
theorem. In particular, the theorem implies that a diffeomorphism between
an ovaloid and a sphere, which preserves the second fundamental form, is
a congruence.
However, it has been pointed out by [Becker–Kühnel 1996]
that the situation is subtle. They have constructed examples of non-spher-
ical C 2 ovaloids M of revolution, such that (M , II) is C 1 -isometric to the
unit sphere, the Gaussian curvature K is well-defined, but gradII K is ill-
defined. Consequently, the expression (2.2) for the Gaussian curvature of
the second fundamental form is not valid.
Many more similar characterisations for the sphere have been

found (see the overview on the literature in the last chapter). Furthermore,
it is known that the second fundamental form is a flat Lorentzian metric on
a minimal surface.
2.4. The Extrinsic Geometry of the Second Funda-

mental Form.
2.4.1. The Mean Curvature of the Second Fundamental Form.

As was known already to Gauss, the Gaussian curvature is an intrinsic in-
variant of a surface in E3 . In extension, a Gaussian curvature can be asso-
ciated to every abstract metric, and it is in this way that the curvature of
the second fundamental form was determined in the previous section.
In contrast, the mean curvature of a surface in E3 belongs to
its extrinsic geometry, for it depends on both the first and the second fun-
damental form. In order to adapt this concept of mean curvature to the
30
2.4. THE EXTRINSIC GEOMETRY OF THE SECOND FUNDAMENTAL. . . 2.4. 31
geometry of the second fundamental form, we will go back to a variational

characterisation of the mean curvature: the variation of the area along a
deformation µ with deformation vector field Z = Z t + f N is given by
Z
δArea(M ) = −2 f H dΩ (2.3)
and this property is characteristic to the function H. Thus the mean curva-
ture of a surface in the Euclidean space can be described as a measure for
the rate of area growth under deformations of the surface.
In this section, the notion of mean curvature will be tailored to
the second fundamental form. Since we are studying surfaces the second
fundamental form of which is a Riemannian metric, areas can be measured
with respect to the second fundamental form as well, and we can associate
to any such surface M its area, as surveyed in the geometry of the second
fundamental form. This area is related to the classical area element dΩ by
Z
p
AreaII (M ) = K dΩ .
Now the “mean curvature of the second fundamental form” (HII ) will be in-
troduced as a measure for the rate at which this total II-area changes under
a deformation: Z
δAreaII (M ) = − f HII dΩII . (2.4)
This curvature HII , which resulted from a comparison of II-areas, and as

such belongs to the extrinsic geometry of the second fundamental form of
the surface, was introduced by E. Glässner.
Theorem 2.8.
Let M be an ovaloid in E3 . The derivative of the area functional of the
second fundamental form along a deformation µ with deformation
vector field Z = Z t + f N is given by
Z
1

δAreaII (M ) = − f H + ∆II (log K) dΩII .
4
31
Proof of Theorem 2.8. An application of theorem 1.6 yields

Z Z ¨
p
δ K dΩ = f λ2 Hess p1 (e1 , e1 )
2 K
«
p
+ λ1 Hess 1
p
(e , e ) − H
2 2 K dΩ .
2 K
Denote ϕ for the function ϕ : R+ 1

0 → R : x 7→ 2 log( 2x ). We obtain
2
X 1
trII Hess(log K) = Hessϕ( 1
) (ei , ei )
i=1
λi p
2 K
2
1 1
X
= ∇ei grad(ϕ( p )) , ei
λ
i=1 i 2 K
2
1 p 1
X
= ∇ei −4( K) grad( p ) , ei .
λ
i=1 i 2 K
Consequently, we have
2
p X 1
trII Hess(log K) = −4( K) Hess( p1 ) (ei , ei ) (2.5)
λ
i=1 i
2 K
1
+ II gradII (log K), gradII (log K) .

2
On the other hand, lemma 2.2 and lemma 2.6 give
1
∆II (log K) = trII Hess(log K) − II gradII (log K), gradII (log K) .

(2.6)
2
In consequence of (2.5) and (2.6), we obtain
 
−4
∆II (log K) = p λ2 Hess( p1 ) (e1 , e1 ) + λ1 Hess( p1 ) (e2 , e2 ) .
K 2 K 2 K
A substitution of the above equation in the first equation of this proof gives
the desired result.
32
Definition 2.9.
Let M be an ovaloid. The mean curvature of the second fundamen-
tal form HII is defined by
1
HII = H + ∆II (log K) .
4
If HII = 0, the surface will be called II-minimal.
Remark. There may be an inaccuracy in the terminology, since it is not

clear whether a critical point of AreaII is necessarily a minimum. Despite
what might be suggested by the name mean curvature of the second funda-
mental form, the quantity HII is not determined by the second fundamental
form II solely, but belongs to the extrinsic geometry of the second funda-
mental form.
Remark. The function HII is homogeneous with respect to rescalings, in

the following sense. Let an ovaloid Me be obtained by rescaling a given
ovaloid M ⊆ E with a factor φ. Let H
3 fII stand for the composition of the
mean curvature of the second fundamental form of M e with the rescaling
homothety which maps M onto M e . Then there holds
1
fII =
H HII .
φ
Remark. The concept of HII is an adaption of the function H to the geom-

etry of the second fundamental form in the sense that the two variational
formulae
 Z Z

 δ dΩ = −2 f H dΩ ;

 Z Z

δ dΩII = − f HII dΩII ,

33
are valid for deformations of the given ovaloid with deformation vector
field Z = f N . However, since the divergence proof was applied in the proof
of theorem 2.8, it is not true that δ dΩII and −HII f dΩII are identical.

Instead, the formulae


δ (dΩ) = −2 H f dΩ ;

1

δ dΩII =

trII Hess f −H f dΩII ,

2
are valid if the deformation vector field is Z = f N .
Remark. It is easy to recognise that there do not exist II-minimal ovaloids

in E3 . This follows already from the fact that the functional AreaII is strictly
positive but not scaling-invariant. Alternatively, it can be seen that HII (p)
is strictly positive at a point p of an ovaloid where K is minimal.
2.4.2. A Beltrami formula for HII .
A classical result, due to Beltrami, states that the mean curvature of a sur-
face M ⊆ E3 can be expressed by means of the Laplacian of the position
vector field P. This Laplacian of the position vector field is defined by
〈∆P, X 〉 = ∆〈P, X 〉 for every constant (i.e., absolutely parallel) vector field
X ∈ X(E3 ). Thus ∆ is the Laplacian of the first fundamental form of the
surface acting on the vector field P. We ask whether a similar result holds
for the Laplacian of the second fundamental form. Therefore, we define
the II-Laplacian of the position vector field similarly by the requirement
that 〈∆II P, X 〉 = ∆II 〈P, X 〉 for every constant vector field X ∈ X(E3 ).
Theorem 2.10 (Beltrami’s Formula).

A surface M ⊆ E3 satisfies
∆P = 2H N .
34
Proof of Theorem 2.10. Choose an arbitrary point p• ∈ M and a rectilinear

co-ordinate system (X , Y, Z) with p• as origin and the Z-axis along the unit
•
normal direction Np• . The symbol = will indicate that an equality is valid
at the point p• . The co-ordinate vector fields, which are constant, will be
denoted by u1 , u2 and u3 ∈ X(E3 ). In accordance with the notation adopted
in lemma 2.4, we will denote x, y and z for the restrictions of X , Y and Z
to the surface. An application of this lemma gives

•
 ∆P, u1 = ∆ P, u1 = ∆x = tr Hess x = tr HessX = 0 ;


• ¦ ©
∆P, u3 = ∆ P, u3 = ∆z = tr Hessz = tr Hess Z + II = 2H ,

and this establishes Beltrami’s equation.
Remark. A fourth proof of the first Jellett-Minkowski integral formula on

page 13 results by a mere substitution of Beltrami’s formula in
Z Z ( 3 )
X
0 = ∆ 〈P , P〉 dΩ = ∆

P , ui P , ui dΩ .
i=1
(The theorem below is already contained in [Haesen–Verpoort–

Verstraelen 0000].)
Theorem 2.11 (Adaption of Beltrami’s Formula to the Second Funda-

mental Form).
An ovaloid M ⊆ E3 satisfies
1
∆II P = − gradII log K + 2N . (2.7)
2
It is still possible to express HII in terms of the position vector field:
 
1 t
HII = 〈∆P, N 〉 − divII ∆II P  .
2
35
Proof of Theorem 2.11. We adopt the same notation as in the previous

proof. We calculate
∆II P, u1 = ∆II P, u1 = ∆II x

which can be rewritten with lemmata 2.6 and 2.2

1
= trII Hess x − II gradII log K, A← grad x

2
• 1
= − gradII log K, u1 .

2
•
Similarly, by making use of grad z = 0 and lemma 2.4, one finds
•
∆II P, u3 = ∆II P, u3 = ∆II z = trII Hessz = trII II = 2 .

2.4.3. The Unit Normal Variation of the Curvatures of the Sec-

ond Fundamental Form.
It seems natural to ask for an expression for the variations of the curvatures
which have been associated to the geometry of the second fundamental
form. The following theorem, which is thus a modification of theorem 1.3
on page 10, provides us with a partial answer. Some variational formulae
for abstract manifolds, which are collected in the next chapter, will already
be applied in the proof.
Theorem 2.12 (Unit Normal Variation of the Curvatures of the Second

Fundamental Form).
There holds:

1 1
δN HII = 2H 2 − K + ∆ log K + ∆II H
4 2


1 1


− II gradII H, gradII log K +


2
grad K, grad K
 4 8K


1 1 1


δN KII = H KII − ∆ log K + ∆II H −

2
grad K, grad K .
4 2 8K
(2.8)
36
Proof of Theorem 2.12. The unit normal variation of the second funda-
mental form is equal to δN II = −III. The variation tensor of the Levi-Civita
connection of the second fundamental form, which will be denoted by X II ,
satisfies the equation (cf. page 64)

4II X II (W, Z), V = −2 ∇IIZ III (W, V ) − 2 ∇IIW III (Z, V ) + 2 ∇IIV III (W, Z)

which can be rewritten if the Weingarten formula for the connection of the
second fundamental form is used:
= − ∇ Z III (W, V ) − ∇W III (Z, V ) + ∇V III (W, Z) .

It can easily be deduced from this expression that every vector field Z ∈
X(M ) satisfies
trII (V, W ) 7→ II(X II (W, Z), V ) = −II(Z, gradII H) .

(2.9)
In agreement with lemma 2.1 and theorems 3.1–3.2, a fixed function f and
a fixed tensor of type (0, 2) satisfy
(
(δ gradII ) f = grad f ;
(δ trII )B = tr B .
If f ∈ F(M ) and V, W ∈ X(M ) are fixed,

δ HessIIf (W, V ) = (δII) ∇IIW gradII f , V (2.10)

+II X II (W, gradII f ), V + II ∇IIW (δ gradII ) f , V ,

and the variation of the II-Laplacian of the function log K, which varies
under the deformation as well, can be expressed as

δ(∆II log K) = trII δ HessIIlog K + tr HessIIlog K +∆II δ log K .

The first term on the right-hand side can be reduced with help of equa-
tions (2.10) and (2.9); the second term can be rewritten by means of an
obvious adaption of lemma 2.6; and the last term can be simplified if the
expression δN K = 2 K H for the normal variation of the Gaussian curvature
37
is taken into account. A further calculation finally gives an expression for

δ(∆II log K) which, together with the expression for the normal variation
of the mean curvature, leads to the first expression for δN HII .
The second equation follows from the result on page 66.
2.5. Some Characterisations of Euclidean Spheres,

in Which the Curvatures of the Second Funda-
mental Form are Involved.
In this section, several characterisations of the sphere, in which the mean

and the Gaussian curvature of the second fundamental form are involved,
will be given. In particular, Christoffel’s result which describes the spheres
as the only solutions of a certain variational problem, will be adapted to
the second fundamental form.
2.5.1. Some First Results.
For the result below, see also [Stamou 1981].
Theorem 2.13 (Stamou).

An ovaloid M ⊆ E3 satisfies either HII = H + f (K) or HII = H f (K) for
an increasing function f : R → R, if and only if M is a sphere.
Proof of Theorem 2.13. Assume first that M is an ovaloid satisfying one

of the given conditions. Let p+ and p− be two points where K attains its
maximum and minimum, respectively. In the first case, we necessarily have

4 f (K) p = ∆II log(K) p ¶ 0 ¶ ∆II log(K) p
+ + −

= 4 f (K) p ¶ 4 f (K) p .

− +
This shows that K is constant, which is only possible if M is a sphere.
38
2.5. SOME CHARACTERISATIONS OF EUCLIDEAN SPHERES. . . 2.5. 39
In the second case,

1

H f (K) p = HII (p = H + ∆II log(K) ¶ H|(p+ )
+ +) 4 p+
and consequently, every p ∈ M satisfies

f (K) p ¶ f (K) p ¶ 1 .
+
But this means ∆II log(K) ¶ 0, and M is a sphere.

The converse direction of the theorem is straightforward.
The following theorem can be found in [Manhart 1989a], Satz

3.7 and [Stamou 2003], Korollar 1, (b) with α = 0.
Theorem 2.14 (Manhart–Stamou).

The spheres are the only ovaloids M ⊆ E3 which satisfy HII ¾ KII .
Proof of Theorem 2.14. It is easy to verify that HII = R1 = KII on a sphere

of radius R. Conversely, suppose that the inequality HII ¾ KII is satisfied. In
consequence of the equation
∆II K II(gradII K, gradII K)

HII = H + −
4K 4K 2
and expression (2.2) for KII , we have
0 ¶ HII − KII
∆II K II(gradII K, gradII K) 1
= − − II(L, L)
4K 8K 2 2
∆II K
¶ .
4K
But the achieved inequality 0 ¶ ∆II K implies that K is constant. This is only
possible if M is a sphere.
39
Theorem 2.15.
If M ⊆ E3 is an ovaloid, then there holds
Z Z
HII dΩ ¾ HdΩ
with equality if and only if M is a sphere.
Proof of Theorem 2.15. This is an immediate consequence of the relation

1
∆II (log K) = div gradII (log K) + II(gradII (log K), gradII (log K)) ,
2
which can easily be obtained from lemma 2.3.
The following theorem is obvious.
Theorem 2.16.
Let M ⊆ E3 be an ovaloid. Then the chain of inequalities
Z Z
4π ¶ HII dΩII ¶ H 2 dΩ
is valid, with equality anywhere in the chain if and only if M is a

Euclidean sphere.
An ovaloid in E3 of which it is known that any of the functions

p
KII − K or HII − H does not change sign, is a sphere [Koutroufiotis 1974,
Stamou 1981]. We mention a similar result by G. Stamou (see [Stamou
2003], Korollar 1, (c), with α = 12 ; the present proof makes also use of
ideas of theorem 3 of [Stamou 1981]).
Theorem 2.17 (Stamou).

An ovaloid M ⊆ E3 is a sphere if and only if the function HII − KII does
not change sign.
40
Proof of Theorem 2.17. If KII ¾ HII , it follows that

Z Z Z
p
K dΩII = K dΩ = 4π = KII dΩII
Z Z Z
p
¾ HII dΩII = H dΩII ¾ K dΩII .
Consequently, M is a sphere.
The case where HII ¾ KII is covered by theorem 2.14.
2.5.2. A Jellett-Minkowski Integral Formula for the Area of the

Second Fundamental Form.
It is possible to adapt a previous integral formula to the geometry of the

second fundamental form.
Theorem 2.18 (Adaption of the First Jellett-Minkowski Formula to the

Second Fundamental Form).
Let an ovaloid M ⊆ E3 be given. The following integral formula is
valid: Z
AreaII (M ) = HII 〈−N , P〉 dΩII . (2.11)
Proof of Theorem 2.18 (First Version). The integral equality can be ob-
tained by the rescaling principle which has already been applied in the first
version of the proof of theorem 1.7.
Proof of RTheorem 2.18 (Second Version). The integral formula expresses

that 0 = divII P t dΩII .
(See the literature overview on p. 166 for a bibliographical

comment on this integral formula.)
41
2.5.3. An Adaption of Christoffel’s Variational Characterisation

of Euclidean Spheres.
In this § 2.5.3, three new characteristic properties for the sphere will be
described (see also [Verpoort 0000a]).
As an immediate consequence of theorem 2.13, the sphere is
the only ovaloid in E3 satisfying HII = C H for some constant C. The fol-
lowing result, which follows from the above integral equality and is of a
similar nature.
Theorem 2.19.
p
Let M ⊆ E3 be an ovaloid which satisfies HII = C K for some constant
C. Then M is a sphere.
Proof of Theorem 2.19. To prove this theorem, we observe that the con-
stant C is of necessity greater or equal than unity:
Z Z Z
p
C K dΩ = C K dΩII = HII dΩII
Z Z Z
p
= H dΩII = H K dΩ ¾ K dΩ .
Remark. The fact that C ¾ 1 can also be deduced from theorem 2.15:
Z Z Z Z
p
C H dΩ ¾ C K dΩ = HII dΩ ¾ H dΩ .
Or yet, it follows from the fact that a point p where ∆II log K (p) ¾ 0

satisfies
p 1
C H(p) ¾ C K(p) = HII (p) = H(p) + ∆II log K (p) ¾ H(p) .

4
On the other hand, an application of formula (2.11) and of an earlier-
mentioned integral formula gives
Z Z Z
p
K dΩ = AreaII (M ) = HII 〈−N , P〉 dΩII = C K〈−N , P〉 dΩ
42
Z Z Z
p
= C H dΩ ¾ H dΩ ¾ K dΩ .
This is only possible if M is a sphere.
Under a deformation of an ovaloid M ⊆ E3 with deformation

vector field Z = Z t + f N , there holds
 Z Z

 δ H dΩ = − f K dΩ ;


Z
p


δ AreaII = −

 f HII K dΩ .
Hence, the previous theorem can be restated as follows.
Theorem 2.20 (Variational Rewording of Theorem 2.19).

3
R M ⊆ E is a critical point of the total mean curvature
If an ovaloid
functional H dΩ with respect to deformations under which the area
of the second fundamental form is preserved, then M is a sphere.
Moreover, the inequality

Z Z
p
H dΩ ¾ K dΩ = AreaII
shows that this is actually a minimum.

The last result is a modification of the theorem of Christoffel
which was mentioned in the first chapter. We shall give a proof of another
modification of Christoffel’s result, although the similarity is only on a for-
mal level. Under the additional assumption KII ¾ 0, the theorem below
becomes a consequence of theorem 1.d of [Stamou 1987].
Theorem 2.21.
The spheres are the only ovaloids M ⊆ E3 such that HII = C KII for
some C ∈ R.
43
Proof of Theorem 2.21. Let M be an ovaloid such that HII = C KII . It

should be noticed that the constant which occurs in the condition satisfies
C ¾ 1. This inequality follows from
Z Z
4πC = C KII dΩII = HII dΩII (2.12)
Z Z Z
p
= H dΩII = H K dΩ ¾ K dΩ = 4 π .
The following inequalities are valid at a point p• where K achieves its max-
imal value:
• 1
0 ¾ ∆II log K = HII − H = C KII − H
4
• C
= (C − 1) H + II(L, L) ¾ (C − 1) H .
2
This only possible if C = 1, such that equality occurs in (2.12). This finishes
the proof.
2.5.4. Isoperimetric Inequalities for the Second Fundamental

Form.
Manhart has established the II-isoperimetric inequality for ovaloids M ⊆ E3
(see [Manhart 1989b]).
Theorem 2.22 (The Isoperimetric Inequality for the Second Fundamen-

tal Form).
3
AreaII (M ) ¶ 48π2 (Vol(M )) ,
and equality characterises the spheres.
Proof of Theorem 2.22. Since both members of the inequality behave ho-
mogeneously of the third degree with respect to homotheties, it is sufficient
44
to establish the theorem for ovaloids enclosing the same volume as the unit
sphere. For such an ovaloid, consider the function
Z
1

τ
+
f : , 2 → R : τ 7→ K 2 dΩ .
2 M
Z
1 2 τ
This is a convex function, for f 00 (τ) = log K K 2 dΩ ¾ 0. According
4
to the equiaffine isoperimetric inequality, f ( 12 ) ¶ 4π with equality if and
only if M is an ellipsoid. Furthermore, f (2) = 4π. This gives already that
AreaII (M ) = f (1) ¶ 4π. Moreover, equality is only possible if f ≡ 4π,
which implies that log K = 0, i.e., M is the unit sphere.
Fairly obvious is the following related result.
Theorem 2.23.
2
AreaII (M ) ¶ 4π Area(M ) ,
and equality characterises the spheres.
Proof of Theorem 2.23. This follows from

Z ÈZ Z
p p
AreaII (M ) = K dΩ ¶ K dΩ dΩ = 4π Area(M ) ,
where equality occurs if and only if K is a constant.
There are some interesting questions of a variational char-

acter which are related with the previous theorems. First of all, the II-
isoperimetric problem has not been solved yet [Glässner 1974, Glässner–
Simon 1973]: “Are the spheres the only critical points of the II-area func-
tional under volume constraint?” Critical points of this variational problem
need to satisfy the relation
p
HII K = C (C ∈ R) (2.13)
45
but it is not known whether the speres are the only ovaloids satisfying this
equation. (Theorem 2.22 gives us that the spheres are the only absolute
maximum of the constrained functional.)
A similar question is: “Are the spheres the only critical points of
the II-area functional under area constraint?” Critical points of this varia-
tional problem need to satisfy the relation
p
HII K = C H (C ∈ R) . (2.14)
(The above theorem states that the spheres are the only absolute maximum
of the constrained functional.)
However, in both cases it can be concluded that an operator-
valued equation, which is stronger than the above equations, is satisfied.
This is discussed on p. 166 and p. 67 ff, respectively.
2.5.5. Further Integral Formulae and Unicity Results.

The aim of this section is to develop some further integral formulae, for
these often serve as a valuable tool for the purpose of establishing unique-
ness results for the sphere. In first instance, the rescaling principle which
yielded already theorems 1.7 and 2.18, can be applied to other functionals
as well (this has been described in, a.o., [Li 1983]). In the following, only
the homogeneous functional
Z
τ
K 2 dΩ (2.15)
will be studied. Here τ is a fixed real number.
Theorem 2.24.
Let M be an ovaloid in E3 . The variation of the functional (2.15)
along a deformation µ with deformation vector field Z = Z t + f N is
given by
τ(τ − 3) II(gradII K, gradII K)

Z Z
τ
δ K 2 dΩ = (τ − 2) f H +
8 K2
46
τ ∆II K

τ
+ K 2 dΩ . (2.16)
4 K
Proof of Theorem 2.24. An application of theorem 1.6 gives

τ τ
Z Z ¨
τ
δ K 2 dΩ = f λ2 Hess τ−2 (e1 , e1 ) + λ1 Hess τ−2 (e2 , e2 )
2 K 2 2 K 2

τ
+ (τ − 2)H K 2 dΩ .
It is easy to show that

(τ − 2)(τ − 4) τ−6 (τ − 2) τ−4
Hess τ−2
= K 2 dK ⊗dK + K 2 HessK .
K 2 4 2
When this is substituted in the first equation of this proof, there results
τ τ−2
Z Z
τ
δ K 2 dΩ = (τ − 2) f K 2 trII HessK
4
τ(τ − 4) τ II(gradII K, gradII K)

τ
+ K2 + H K 2 dΩ .
8 K2
Lemma 2.6 finishes the proof.
Remark (Generalised Fundamental Forms). Besides the second or the

third fundamental form, the following bilinear forms could be studied on
an ovaloid:
τ + 1 : X(M ) × X(M ) → F(M ) : (V, W ) 7→ 〈(Aτ ) V, W 〉 .

R τ
Here τ is an arbitrary real number. The functional K 2 dΩ is the area func-
tional of this fundamental form, and the quantity between curly brackets
in formula (2.16) could be called the mean curvature of this generalised
fundamental form (notation H τ+1 ). The formula then becomes
Z
δArea τ+1 = (τ − 2) f H τ+1 dΩ τ+1 .
47
In particular, for τ = 0 we obtain the formula (2.3) which char-

acterises the usual mean curvature H 1 = H; for τ = 1 we similarly recover
formula (2.4) in which the mean curvature of the second fundamental form
H 2 = HII occurs; and the variation of the area functional associated to the
third fundamental form vanishes identically since it measures the area of
the Gaussian image.
However, since these generalised fundamental forms do not
pertain to the subject of this dissertation,
R τ we will not pursue this theme.
The study of the first variation of K 2 dΩ is primarily motivated by the
search for integral equalities related to the second fundamental form.
Theorem 2.25.
Let an ovaloid M ⊆ E3 be given. There holds:
τ(τ − 3) II(gradII K, gradII K)

Z Z
τ
K 2 dΩ = 〈−N , P〉 H +
8 K2
τ ∆II K

τ
+ K 2 dΩ . (2.17)
4 K
Proof of Theorem 2.25. Assume first τ 6= 2. A first integral equality is

obtained if the fact that the functional (2.15) is homogeneous of degree
2 − τ, is exploited in a similar way as in theorem 2.18. The equation (2.17)
is obtained upon division by (τ − 2).
For τ = 2, the equation results by evaluating the limit of (2.17)
for τ approaching 2.
Thus the following formulae are valid:

Z Z
1 5 II(gradII K, gradII K)

(τ = −2) dΩ = 〈−N , P〉 H +
K 4 K2
1 ∆II K 1

− dΩ ;
2 K K
48
Z Z
1 1 II(gradII K, gradII K)

(τ = −1) p dΩ = 〈−N , P〉 H +
K 2 K2
1 ∆II K 1

− p dΩ ;
4 K K
Z Z
(τ = 0) dΩ = 〈−N , P〉 H dΩ ;
Z Z
(τ = 1) dΩII = 〈−N , P〉 HII dΩII ;
Z Z
1 II(gradII K, gradII K)

(τ = 2) K dΩ = 〈−N , P〉 H −
4 K2
1 ∆II K

+ K dΩ ;
2 K
3 ∆II K
Z Z
3 3
(τ = 3) K dΩ =
2 〈−N , P〉 H + K 2 dΩ ;
4 K
Z Z

2
(τ = 4) K dΩ = 〈−N , P〉 H +
2 K2
∆II K

+ K 2 dΩ .
K
Remark. The integral formula of theorem 2.25 can be rewritten as

Z Z ¨
τ
K 2 dΩ = 〈−N , P〉 (1 − τ)2 H + τHII
«
1 τ
+ τ(1 − τ)KII + τ(τ − 1) II(L, L) K 2 dΩ .
2
Theorem 2.26.
Let an ovaloid M ⊆ E3 be given. There holds:
49
Z Z

log K dΩ = 〈−N , P〉 H log K −
4 K2
1

+ ∆II log K dΩ .
2

∂
Proof of Theorem 2.26. This is obtained by taking ∂ τ τ=0
of both sides
of (2.17).
Theorem 2.27.
An ovaloid satisfies
p
4
1¶K and HII + KII ¶ 2 H (1 − log K)
if and only if it is the unit sphere.
Proof of Theorem 2.27. The requirement on HII + KII implies that
H log K + 2HII + 2KII ¶ 4H .
Combined with the integral equality of the previous theorem, this gives
Z
0¶ log K dΩ
Z
1

= 〈−N , P〉 H log K + 2(HII − H) + 2 KII − H − II(L, L) dΩ .
2
¶ 0.
It is maybe interesting to compare this theorem with the following charac-

terisation of the sphere, which can be proven easily by integrating the con-
dition with respect to the area element of the second fundamental form.
50
Theorem 2.28.
An ovaloid is a sphere if and only if
HII + KII ¾ 2H .
We will now present another integral formula, which can ei-

ther be obtained by exploiting the scaling invariance of the integral
Z Z Z
p
H K dΩ = HdΩII = HII dΩII
or from calculating the normal variation of both sides of the integral for-
mula for the II-area, which was previously obtained. The result can accord-
ingly be seen as an extension of the second integral formula of Jellett–
Minkowski to the geometry of the second fundamental form. Unfortu-
nately, the result is rather technical, and I was not able to concoct an appli-
cation of the formula to a unicity result for the sphere.
Theorem 2.29 (Adaption of the Second Jellett-Minkowski Formula to

the Second Fundamental Form).
Let M be an ovaloid in E3 . There holds
p p
∆ K
Z ¨ «
2
K
0 = 〈−N , P〉 H − K + p + trII Hess pH
dΩII (2.18)
2 K 2 K
Z
KII 1

= 〈−N , P〉 H(H − HII ) + (H 2 − K) + ∆ log K (2.19)
H 4
H 1

+ ∆II log H +

grad log H, grad log K dΩII .
2 4
R p
Proof of Theorem 2.29 (First Version). Since the integral H K dΩ
is invariant under rescalings, equation (2.18) immediately follows from
51
theorem 1.6. This is equivalent with the equation

Z ¨

2
1 grad K, grad K 1
0 = 〈−N , P〉 H − K + ∆ log K + 2
+ ∆II H
4 8K 2
«
H H II(gradII K, gradII K) II gradII H, gradII K
− ∆II K + − dΩII ,
4K 4 K2 4K
which can be rewritten as

Z
1 1

= 〈−N , P〉 2H 2 − K − H HII + ∆ log K + ∆II H
4 2

«
grad K, grad K II gradII H, gradII K
+ − dΩII .
8K 2 4K
The formula (Q) on page 29 can be written as

grad K, grad K II gradII H, gradII K
−
8K 2 4K

1 2
II gradII H, gradII H grad H, grad K
= (KII − H)(H − K) − + .
H 2H 4H K
When this is substituted in the last expression, the desired expression (2.19)
is obtained.
Proof of Theorem 2.29 (Second Version). The integral equality can also
be obtained by a evalating the effect of the operator δN on both sides of
equation (2.11) on page 41. In this way, the formula becomes an applica-
tion of the first formula of theorem 2.12.
Remark.
R I was not able to establish the result starting from the fact that
0 = divII AP t dΩII .
2.6. Calculation of the Curvatures for Some Surfaces

of Revolution.
Let f be a strictly positive function defined on an open interval, subject
to the condition f 00 < 0. For the surface which is obtained by revolution
52
2.6. CALCULATION OF THE CURVATURES FOR SOME SURFACES. . . 2.6. 53
5fr
f-
I
..r
I
I
I
^) 2,/l
I
I
I
I
I
-
I
'C\,,--l/
f-.s------
iJ
-
I
:LN
ss)
-\
o!
NI
Irl
-
I
F
-
.u
o$1'
-
-
\
F
--J
I
t ,,
I
/
Figure 2.5: Construction of the surface of revolution.
of the curve z = f (x) around the x-axis (as in figure 2.5), the following
expressions for the curvatures are valid with respect to the interior unit
normal vector field:
− f 00
K = ;
f (1 + f 02 )2
1 + f 02 − f f 00
H = 3
;
2 f (1 + f 02 ) 2
2 f 00 + f 00 f 02 − f f 0 f 000
KII = p
4 f f 00 1 + f 02
p
1 + f 02 00 02 0 000
f 00 (−1 + f 02 )
− f f + f f f + ;
4 f 2 f 002 3
2(1 + f 02 ) 2
p
1 + f 02

HII = 2 003
6 f f 003 − 2 f f 0 f 00 f 000
8f f
53
Figure 2.6: The ellipsoid of revolution of the last example (on page 55) together
with its generating function.

02 002 2 0002 2 00 0000
−f f + 3f f − 2f f f
f 0 f 000 f 00 − 2 f 00 f 02
+ p + 3
.
2 f 00 1 + f 02 2(1 + f 02 ) 2
(Here the left-hand side has to be evaluated at a point (x, y, z) of the sur-
face, while the value x has to be substituted in the right-hand side.)
Example. We calculate the p curvatures of the two-sheeted hyperboloid.

Herefore we define f (x) = x 2 − 1 for x > 1. There results
1 (−2x 2 + 3)x 2
K= ; KII = ;
(2x 2 − 1)2 3
(2x 2 − 1) 2
x2 −4x 4 + 3x 2 − 2
H= 3
; HII = 3
. z
(2x 2 − 1) 2 (2x 2 − 1) 2
p
Example. For f (x) = x we find the curvatures of a paraboloid,
4 2
K= ; KII = ;
(4x + 1)2 (4x + 1) 2
3
54
2.6. CALCULATION OF THE CURVATURES FOR SOME SURFACES. . . 2.6. 55
p
Figure 2.7: The functions K (thin line) and H (thick line) drawn along the
parameter x for the ellipsoid (2.20).
4x + 2 4x − 2
H= 3
; HII = 3
. z
(4x + 1) 2 (4x + 1) 2
Example. An investigation of the surface which is obtained by revolution

3
of the graph f (x) = x 4 (for x > 0) around the x-axis gives:
48 12 p1x
K= p ; KII = p ;
x(16 x + 9)2 3
(16 x + 9) 2
32 + 24 p1x 32
H= p 3
; HII = p 3
. z
(16 x + 9) 2 (16 x + 9) 2
Remark. It can be seen that the surfaces originating from the graph of
f (x) = x a (for some a ∈ ]0, 1[ ) satisfy H = HII + 2KII .
Example. The curvatures of the ellipsoid of revolution (see also figure 2.6)
55
Figure 2.8: The functions KII (thin line) and HII (thick line) drawn along the pa-
rameter x for the ellipsoid (2.20).
constructed from the function

p
f (x) = 2 1 − x2 (2.20)
are given by
1 9x 4 − 6x 2 + 5
K= ; KII = ;
(3x 2 + 1)2 3
4(3x 2 + 1) 2
3x 2 + 5 18x 4 + 39x 2 − 1
H= 3
; HII = 3
.
4(3x 2 + 1) 2 4(3x 2 + 1) 2
(See also figures 2.7–2.8.) Notice that the graphs of HII and KII intersect,
as they have to do by virtue of theorem 2.17. For every point p of the el-
lipsoid there has been placed a dot with co-ordinates (K(p), KII (p)) in the
figure 2.9, which may be called the (K, KII )–curvature diagram. Here it
should be remarked that according to a theorem of [Simon 1976] KII could
impossibly have been a decreasing function of K. This is indeed what we
56
2.7. INFINITESIMAL DEFORMATIONS OF SURFACES IN E3 . 2.7. 57
Figure 2.9: KII drawn against K for the ellipsoid (2.20).
observe. z
2.7. Infinitesimal Deformations of Surfaces in E3 .
In this section Weingarten’s equation from the theory of infinitesimal de-

formations will be linked with the geometry of the second fundamental
form. The field of infinitesimal deformations is devoted to the first-order
behaviour of deformations, which means that all formulae have to be read
modulo O (t 2 ). Accordingly, an infinitesimal deformation is completely de-
termined by the corresponding linearised deformation1 .
1
Although it cannot be said that the study of infinitesimal deformations coincides with
the study of linear deformations. This claim is substantiated already by contrasting the pres-
ence of non-trivial isometric infinitesimal deformations of a plane with the non-existence
of non-trivial isometric linear deformations of a plane.
57
In this section, we will allow the second fundamental form of

a surface in E3 to be either a Riemannian or a Lorentzian metric.
If for an infinitesimal deformation µ a one-parameter family of
congruences ν can be found such that
µ t = ν t + O (t 2 )
it will be said that µ is an infinitesimal congruence or a trivial infinitesi-

mal deformation.
An infinitesimal deformation is said to be isometric if the
lengths of curves on the surface are stationary under the deformation. This
condition, that lengths remain unaltered up to O (t 2 ), exactly means that
the variation of the first fundamental form vanishes. The necessary and
sufficient condition on a deformation vector
¬ field¶ Z ¬to determine
¶ an iso-
metric infinitesimal deformation is that DV Z, W + V, DW Z = 0 for all
tangent vector fields V and W . For any infinitesimal deformation µ which
is isometric, a unique rotation vector field Y ∈ X(M ) can be found which
satisfies
∂

dµ t (V ) = DV Z = Y × V (2.21)
∂ tt=0
for all V ∈ X(M ). This can be seen as the analogue of the Darboux vector
field which describes the instantaneous rotation of the Frenet frame along
a curve. The infinitesimal isometric deformation is trivial if and only if Y is
a constant vector field.
Now assume Z is the deformation vector field of an infinitesi-
mal isometric deformation of a surface M of non-vanishing Gaussian cur-
vature. Let Y = Y t + ωN stand for the corresponding rotation vector field.
If equation (2.21) is substituted in the equality
DV DW Z − DW DV Z = D[V,W ] Z
(which holds for all vector fields V, W ∈ X(M )), the following integrability
conditions on the rotation vector field Y are obtained:
¨
gradII (ω) = −Y t ;
div(Y t ) = 2 H ω .
58
2.7. INFINITESIMAL DEFORMATIONS OF SURFACES IN E3 . 2.7. 59
From this we deduce, with help of lemma 2.3,
1
∆II ω = div gradII ω + II gradII ω, gradII log |K|

2
1
= −2 H ω + II gradII ω, gradII log |K| .

(2.22)
2
This equation is equivalent with Weingarten’s characteristic equation (e.g.,
as given in [Bianchi 1922] Vol I, p. 9, eq. (I∗ )) for the function ω.
It is not clear why an equation appears in which the second
fundamental form plays such a decisive rôle. The above equation simplifies
considerably if M is either the unit sphere or a minimal surface, and a proof
of the infinitesimal rigidity of the sphere and a theorem of Lie results.
Theorem 2.30 (Liebmann).

Every infinitesimal isometric deformation of the unit sphere is an in-
finitesimal congruence.
Proof of Theorem 2.30. Equation (2.22) becomes ∆ω = −2ω, for a func-

tion ω defined on the unit sphere. It is well-known that every solution of
this differential equation is the restriction of a linear mapping to the unit
sphere which has been centred at the origin, and the rotation vector fields
which are obtained in this way are constant vector fields.
Theorem 2.31 (Lie).

Every simply connected domain of a minimal surface (subject to K <
0) admits a non-trivial isometric deformation satisfying δH = 0.
Proof of Theorem 2.31. For a minimal surface, equation (2.22) obviously

has ω = 1 as a solution. The rotation vector field of the infinitesimal
isometric deformation which is thus obtained, is the unit normal vector
field Y = N .
59
Since (2.22) is the integrability condition of equation (2.21),

a deformation vector field Z can be found with Y as rotation vector field.
Since Y is a non-constant vector field, the infinitesimal deformation cannot
be an infinitesimal congruence. Let e1 and e2 now stand for the principal di-
rections on M and assume that N = e1 × e2 is the chosen unit normal vector
field. The deformation vector field can be written as Z = α e1 + β e2 + γ N .
If the principal directions are substituted for the vector field V ∈ X(M )
in relation (2.21), six first-order differential equations in the functions α,
β and γ are obtained. If these are used in previous formulae, it can be
checked that δH = 0.
Remark. The above theorem is an obvious corollary of Lie’s result that a

simply connected domain of a minimal surface belongs to a one-parameter
family of isometric minimal surfaces.
Theorem 2.32.
A simply connected surface with nowhere vanishing Gaussian curva-
ture is minimal if and only if the unit normal vector field is the rotation
vector field of an infinitesimal isometric deformation.
Proof of Theorem 2.32. If the unit normal vector field is such a rotation
vector field, ω = 1 is a solution of (2.22), and this is obviously only possi-
ble if the surface is minimal.
Bibliographical Note for Chapter 2. See also [Efimow 1957] in relation with § 2.7.
60
Chapter 3.
VARIATIONAL FORMULAE FOR

ABSTRACT SEMI -R IEMANNIAN M ANIFOLDS .
Deformations of hypersurfaces and the corresponding variational formu-

lae are two main themes of this dissertation. In the previous chapters
it has been studied how certain geometric quantities change if a surface
is perturbed. In this case, the metric tensors of the deformed surfaces
were pulled back through the deforming mappings, which yielded a one-
parameter family of metrics on the initial surface. It is an immediate con-
sequence of Gauss’ theorema egregium that the variation of the Gaussian
curvature, which has been computed already in terms of the fundamen-
tal forms and the deformation vector field, is only dependent on the first
fundamental form along with its variation.
The aim of this chapter is to indicate how the formulae which

describe the variation of such geometric quantities with respect to a one-
parameter family of metrics can be recovered.
These formulae will enable us to study variational problems

associated to intrinsic Lagrangians, whereas in the last two chapters we
sought for critical points of functionals which are defined on a certain class
of hypersurfaces. With this alternative point of view some progress with
respect to a previously stated problem can be made.
61
62 3. VARIATIONAL FORMULAE FOR ABSTRACT. . . 3.
3.1. Deformations of Metrics.

Our interest is carried towards an abstract manifold the metrical tensor of
which is varied. As has been mentioned, the application we have in mind
Ö), or the path
is a study of either the path of first fundamental forms I(µ t
of second fundamental forms II(µ t ), with respect to a deformation µ of a
Ö
hypersurface.
Let us first explain the adopted notation for a semi-Riemannian

manifold (M , g). The set of all vector fields on M will be denoted as X(M ),
whereas the symbol F(M ) stands for the set of all real-valued functions
defined on M . The Riemann–Christoffel curvature tensor R of (M , g) is
given by
R(X , Y )Z = ∇[X ,Y ] Z − ∇X ∇Y Z + ∇Y ∇X Z , (3.1)
where X , Y, Z ∈ X(M ) and ∇ is the Levi-Civita connection. The Ricci (0, 2)-
tensor is the trace of the curvature tensor. For vector fields X , Y ∈ X(M ),
there holds
m
X
Ric(X , Y ) = g R(X , ei )Y, ei "i ,

i=1
where e1 , . . . , em is an orthonormal frame such that g(ei , ei ) = "i = ±1.

We will denote the Ricci operator by Rc : X(M ) → X(M ). This operator is

completely determined by the requirement that for all X , Y ∈ X(M ), there
holds
g (Rc(X ), Y ) = Ric(X , Y ) .
The scalar curvature of (M , g) will be denoted by S = tr Ric.
It will now be assumed that a deformation of the metric, i.e.,

a one-parameter family of semi-Riemannian metrical tensors
g (t) = g + t h + O (t 2 ) , (3.2)
62
3.1. DEFORMATIONS OF METRICS. 3.1. 63
is present on an abstract manifold M . Since we are only interested in first-

order variational problems, it is sufficient to determine the family of metrics
up to O (t 2 ). All geometric quantities constructed by means of the metric
g (t) will bear an upper index (t).
If by means of the metric g (t) a tensor B (t) has been con-
structed, which is for this reason dependent on the parameter t, the varia-
tion of the tensor B is defined as
∂

B (t) .
∂ t t=0
A first example is provided by the variation of the metric, which will be

denoted by h = ∂∂ t g (t) . Moreover, we define the symmetric operator σ

t=0
by the requirement
h(V, W ) = g(σ(V ), W )
for all V, W ∈ X(M ).
By similar methods as in the previous chapter it can be shown

∂
that ∂ t t=0
B (t) depends in an R-linear manner on the variation of the met-
ric h. This explains why the elaborate notation δh B for the variation of
the tensor B could be used; however, the shorthand notation δB will be
adopted almost exclusively.
A particular case of this last definition is the variation of the
gradient of a function. If f ∈ F(M ) is a function, there holds
∂

grad(t) f .

δ grad f =
∂ t t=0
The following theorems can be found immediately.
Theorem 3.1.
For any function f ∈ F(M ), there holds
δ grad f = −σ(grad f ) .
63
Theorem 3.2.
If B is a (0, 2)–tensor, there holds
δ tr B = −g(h, B) .
The variation tensor of the Levi-Civita connection is the ten-

sor X defined by
∂

(t)

X : X(M ) × X(M ) → X(M ) : (V, W ) 7→ X (V, W ) = ∇V W .
∂ t t=0
This is a symmetric tensor. A consideration of the variation of the Koszul

formula immediately gives the following theorem.
Theorem 3.3.
For all V, W, Z ∈ X(M ), there holds
 
1
g(X (V, W ), Z) =  ∇V h (W, Z) + ∇W h (V, Z) − ∇ Z h (V, W ) .

2
(3.3)
If both sides of equation (3.1) are varied, the expression of

the variation of the curvature tensor in terms of the variation tensor of the
connection results.
Theorem 3.4.
For all V, W, Z ∈ X(M ), there holds
δR(V, W )Z = ∇W X (V, Z) − ∇V X (W, Z) .

The covariant derivative ∇B of a tensor B is a tensor again.

The covariant derivative of this tensor will be denoted as ∇2 B. In particular,
for a function f , ∇2 f is the Hessian of f . We define the Laplacian of a
64
3.1. DEFORMATIONS OF METRICS. 3.1. 65
tensor B as the trace of ∇2 B over the first two slots:

n o
∆B = tr (V, W ) 7→ ∇2V,W B .
For a one-form ϕ, we denote the symmetric two-form which is obtained by

symmetrisation of ∇ϕ as d∗ ϕ:
 
1
d∗ ϕ : X(M ) × X(M ) → F(M ) : (V, W ) 7→  ∇V ϕ (W ) + ∇W ϕ (V ) .

2
In the following formula for the variation of the Ricci tensor,

the curvature tensor R acts as a derivation on the tensor h. The divergence
of a symmetric tensor has been defined on p. 27.
Theorem 3.5.
For all U, V ∈ X(M ), there holds
−1 1
δRic(U, V ) = (∆h)(U, V ) + (d∗ div h)(U, V ) −
Hess(tr h) (U, V )
2 2
1

− tr (P, Q) 7→ R(P, V ) · h (Q, U) + R(P, U) · h (Q, V ) .

2
Theorem 3.6.
The variation of the scalar curvature is given by
δS = −∆(tr h) + div (div h) − g(Ric, h) .
Remark. For a surface in E3 which is deformed along the deformation

vector field f N , the variation of the first fundamental form is given by
δI = −2 f II. If this is substituted in the above formula, the expression for
δK is recovered.
65
Theorem 3.7.
The variation of the volume element is given by
1
δdΩ = (trace σ) dΩ .
2
3.2. Variational Problems for Abstract Ovaloids.

In order to apply the foregoing results to variational problems for ovaloids
in E3 , an abstract ovaloid is defined as a compact two-dimensional Rie-
mannian manifold (M , g) such that M is diffeomorphic to a sphere and g
has strictly positive Gaussian curvature. The notation
EABS = { abstract ovaloids }
will be adopted.
Remark. Since the Weyl embedding problem has been solved by Weyl,
Lewy, et al (see the literature overview on p. 175), an abstract ovaloid is
actually not so abstract. More precisely, for every abstract ovaloid (MABS , g)
an ovaloid M ⊆ E3 can be found which is isometric with (MABS , g). More-
over, the congruence theorem of Cohn-Vossen (see the literature overview
on p. 169) gives us that M is unique up to a congruence of E3 .
If a deformation g (t) of the metric of an abstract ovaloid (M , g)

is given, then (M , g (t) ) is still an abstract ovaloid for sufficiently small |t|.
Thus variational problems can be posed on the class EABS . Because of the
previous remark, this is essentially not different from the study of varia-
tional problems on the class of “real” ovaloids.
We introduce the following functionals:
 Z

 F : EABS → R : (M , g) 7→ dΩ ;


Z
 F : E → R : (M , g) 7→ pK dΩ .



II ABS
66
3.2. VARIATIONAL PROBLEMS FOR ABSTRACT OVALOIDS. 3.2. 67
It has been asked already whether the spheres are the only critical points of
the II-area functional if an area constraint is imposed (p. 46). This question
will now be posed in the abstract setting.
If f : M → R is a real-valued function on an abstract ovaloid,
the Hessian operator of f , which is the operator which is metrically equiv-
alent with the Hessian of f , will be denoted by Hs f . It is straightforward
to show that

trace Hs f ◦ σ = div σ(grad f ) − div f div h + f div (div h) .

The following variational formulae can be inferred from the aforemen-

tioned preliminaries:
 Z
1
 δF = (trace σ) dΩ ;
2




Z (  )
p

1 1

 δFII = K −∆ p id + Hs p1  ◦ σ dΩ .

trace 

4 K K
By an application of the lemma which is stated on p. 168 of

[Blair 2000], it can be concluded that an ovaloid is a critical point of the
functional FII with respect to deformations satisfying δF = 0 if and only if
p 1

K −∆ p id + Hs p1 = C id (C ∈ R). (3.4)
K K
In particular, since this variational problem is equivalent with the corre-

sponding variational problem for “real” ovaloids, the above equation is
satisfied for ovaloids in E3 which are a critical point of the II-area func-
tional under area constraint. For instance, the fact that it does not matter
whether we substitute the operator which appears either in the left- or the
right-hand side of the above equation for Q in the expression
λ1 e1 , Q(e1 ) + λ2 e2 , Q(e2 )

p
can be formulated as (2.14), i.e., HII K = C H.
Although the above (operator-valued) equation contains more
information than the previous equation (2.14), I was not able to verify
67
whether the sphere is the only ovaloid satisfying (3.4). (In this respect,
a future application of the results of [Kühnel 1988] could provide further
information.)
Bibliographical Notes for Chapter 3. Equation (3.3), which is sometimes called the Pala-
tini equation, as well as related information, can be found in, e.g., [Yano 1949] eq. (2.58),
[Lichnerowicz 1961] eq. (17.5), [Besse 1987] § 1.K, [Kühnel 2002] lemma 8.5.
68
Chapter 4.
THE VARIATION OF THE SHAPE

OPERATOR OF A HYPERSURFACE IN A SEMI-
RIEMANNIAN MANIFOLD.
4.1. Introduction.
The formula which expresses the first-order change of the shape operator
of a hypersurface under a deformation is a key result for various applica-
tions of the calculus of variation within the framework of semi-Riemannian
geometry.
Two proofs of this well-known formula will be given. The first
proof is based on a comparison between the given deformation of the hy-
persurface with the unit normal deformation, which deforms the hyper-
surface in equidistant hypersurfaces. The second proof makes use of the
interpretation of the second fundamental form as the Hessian of the dis-
tance function.
This chapter will be concluded with two applications of the
established formula. First of all, the generalisation of theorem 1.3 for sur-
faces in a semi-Riemannian manifold of dimension three is written down.
Secondly, the generalisation of Bonnet’s theorem on equidistant surfaces
will be treated for surfaces in space forms.
But the main reason why a chapter has been devoted to the
derivation of this formula is its application in a variational problem asso-
69
70 4. THE VARIATION OF THE SHAPE OPERATOR OF A HYPERSURFACE. . . 4.
ciated to the geometry of the second fundamental form. This has been
included in the next chapter.
4.2. Hypersurfaces in a semi-Riemannian Manifold.

In this §, the notation for hypersurfaces in a semi-Riemannian manifold,
which will be adopted in the present and the following chapters, will be
explained.
4.2.1. Tensors and Curvature.

Assume (M , g) is a semi-Riemannian hypersurface in a semi-Riemannian
manifold (M , g), endowed with a unit normal vector field N . In this work,
all hypersurfaces will be assumed to be connected and embedded. Geometric
objects in (M , g) will be distinguished from their analogues in (M , g) with a
bar. For example, ∇ stands for the Levi-Civita connection of (M , g) whereas
∇ is the Levi-Civita connection of (M , g). The metric g of M , which is the
restriction of the metric of the surrounding space to the tangent spaces of
M , is also called the first fundamental form (which is denoted as I).
The set of all vector fields on M which take values in the tan-
gent bundle TM is denoted by X(M ). The orthogonal projections of T p M
to the tangent space T p M of M and its orthogonal complement T⊥ p M will
be denoted by
 t
 pT M → T p M : w p →
7 wp ,
n
T p M → T⊥

p M : w p 7→ wp .
The volume form of (M , g) is written as dΩ.

The shape operator A, the second fundamental form II and the
third fundamental form III of the hypersurface M are defined through the
formulae

A : X(M ) → X(M ) : V 7→ −∇V N ;
II : X(M ) × X(M ) → F(M ) : (V, W ) 7→ α g(A(V ), W ) ;
III : X(M ) × X(M ) → F(M ) : (V, W ) 7→ g(A(V ), A(W )) ,

70
4.2. HYPERSURFACES IN A SEMI -R IEMANNIAN MANIFOLD. 4.2. 71
where α = g(N , N ) = ±1. This sign convention concerning the second

fundamental
n form is motivated by the fact that in this case the relation
∇V W = II(V, W ) N holds true for V, W ∈ X(M ). Further, let m + 1 be
the dimension of (M , g).
The shape operator is a symmetric tensor on M . Since no plain
generalisation of the principal axes theorem holds for symmetric operators
on a vector space endowed with a non-definite scalar product, a basis of
principal directions need not exist.
The mean curvature H of the hypersurface M is defined as
m
α 1X
H= trace(A) = II(ek , ek )"k
m m k=1
with respect to a local orthonormal basis {e1 , e2 , . . . , em } of X(M ), where

"i = g(ei , ei ).
The Riemann-Christoffel curvature tensor R of the hypersur-
face M is related to the second fundamental form by means of the Gauss
equation
g(R(X , Y )Z, W ) = (4.1)

 
g(R(X , Y )Z, W ) + α II(X , Z)II(Y, W ) − II(X , W )II(Y, Z) ,
which is valid for all tangent vector fields X , Y, Z, W ∈ X(M ). As a conse-

quence hereof, we obtain
Ric(X , Y ) = Ric(X , Y )−α g(R(X ,N ) Y, N )+α m H II(X , Y )−α III(X , Y ) . (4.2)
If σ is a plane tangent to M , the sectional curvature of this plane as viewed

in (M , g) will be denoted as K(σ). The sectional curvature of this plane,
viewed in (M , g), will be denoted as K(σ).
The Codazzi equation of the hypersurface is
(∇X A)Y − (∇Y A)X = R(X , Y )N , (4.3)
for all X , Y ∈ X(M ).
71
4.2.2. Deformations of Hypersurfaces.
A mapping µ : ]−", "[ × M → M : (t, p) 7→ µ t (p) = µ(t, p) which satisfies
(i). for all p ∈ M , µ0 (p) = p ;
(ii). there exists a compact set M ⊆ M such that for all p ∈ M \ M and
all t ∈ ]−", "[ there holds µ t (p) = p ,
will be called a deformation of the hypersurface

M . The deformation
∂
vector field Z is defined by Z(p) = ∂ t µ t (p) ∈ T p M for all p ∈ M .
t=0
Let A(µ t ) stand for the shape operator of µ t (M ). This (1, 1)-
tensor on µ t (M ) can be seen as a tensor on M in the following way:
 
7 dµ←
×) : X(M ) → X(M ) : X → A(µ t ) dµ t (X )  .

A(µ t t
The variation of the shape operator is then defined as the partial derivative
∂

δA(X ) := ×)(X )
A(µ t (X ∈ X(M )) .
∂ t t=0
We will adopt a similar convention for all other tensors which are defined
on the class of all hypersurfaces of (M , g). It is clear that theorem 1.1 on
p. 4 remains valid.
4.3. The Variation of the Shape Operator.
The main purpose of this chapter is to establish the following expression

for the variation of the shape operator.
Theorem 4.1 (The Variation of the Shape Operator).

Let (M , g) be a semi-Riemannian hypersurface of a semi-Riemannian
72
4.3. THE VARIATION OF THE SHAPE OPERATOR. 4.3. 73
manifold (M , g). Under a deformation of M , with deformation vector

field Z = f N + Z t , there holds for every X , Y ∈ X(M ):
  

 δA(X ) = f R(N , X )N + A2 (X ) + α Hs f (X )



+ (L(Z t ) A)(X ) ;





  

δII(X , Y ) = α f  g(R(N , X )N , Y ) − III(X , Y ) + Hess f (X , Y )






+ (L(Z t ) II)(X , Y ) .
We need some further notation and lemmata before we can

prove the above theorem.
As can be inferred from the generalisation of theorem 1.1, we
only need to prove theorem 4.1 for some deformation with deformation
vector field f N (for arbitrary f ∈ F(M )). In consequence, it is not a restric-
tion to suppose that the deformation µ can be written as
µ : ]−", "[ × M → M : (t, p) 7→ µ t (p) = exp t f (p) Np .

We will also consider the unit normal deformation ξ which deforms M

into equidistant hypersurfaces ξs (M ):
ξ : ]−", "[ × M → M : (s, p) 7→ ξs (p) = exp s Np .

Let Nt be the unit normal vector field of µ t (M ). Let N be

the unit vector field (which is defined on a M -neighourhood of a point
n ∈ M subject to f (n) 6= 0) which coincides with Nt on µ t (M ) for every
t. Similarly the vector field U extends the unit normal vector fields Us of
ξs (M ) (see figure 4.1). It should be noted that there holds N = U = ∂s on
M.
The real-valued function t ∈ F(M ) can be defined by µ t (p) 7→
t. In a similar way the function s can be defined (this is the distance func-
tion to M ). For any φ ∈ F(M ), the real-valued function φ ∈ F(M ) is de-
73
Figure 4.1: The vector fields N and U.
termined by φ(ξs (m)) = φ(m). The functions s, t and f on M are clearly

related by s = t f .
If X ∈ X(M ), we define
µ
 
µ = dµ t (X p )
 X ∈ X(M )
Xµ
 t (p)
through the equations
ξ
X ξ ∈ X(M ) X ξ (p) = dξs (X p ) .
 
s
Lemma 4.2.
Let X ∈ X(M ). Then, for all p ∈ M and all t ∈ ]−", "[,
µ ξ
= t X p f Uµ t (p) + X µ (p) .

Xµ
t (p) t
Proof of Lemma 4.2. Let an arbitrary vector vp ∈ T p M be given. Assume

the real numbers s and t are related by t f (p) = s such that the points µ t (p)
74
and ξs (p) coincide. What we have to prove is that the relation
dµ t (vp ) = t vp[ f ] Uµ t (p) + dξs (vp )
holds. This is essentially a consequence of the linearity of tangent map-

pings.
Let γ : R → M : x 7→ γ(x) be a curve passing through the point
p with velocity vp for parameter value x = 0. Let ω be the curve in the
normal bundle, obtained by lifting γ through the vector field t f N :
ω : R → T⊥ M : x 7→ t f (γ(x)) Nγ(x) .
Let further Λ be the diffeomorphism
Λ : T⊥ M → R × M : λNq 7→ (λ, q)
with inverse
Γ : R × M → T⊥ M : (λ, q) 7→ λNq .
Further we define two more curves in T⊥ M by

¨
η : R → T⊥ M : x 7→ (t f (p) + x)Np ,
ν : R → T⊥ M : x 7→ t f (p)Nγ(x) .
Then we see that

 
∂

dµ t (vp ) = d exp  ((Γ ◦ Λ ◦ ω)(x))
∂ x x=0
 
∂

= d(exp ◦ Γ)  t f (γ(x)), γ(x) 

∂ x x=0
 
= d(exp ◦ Γ) (t vp [ f ], 0) + (0, vp )
   
= t vp [ f ] d exp dΓ(1, 0)(t f (p),p)  + d exp dΓ(0, vp )(t f (p),p) 
   
∂ ∂

= t vp [ f ] d exp  η(x) + d exp  ν(x)
∂ x x=0 ∂ x x=0
75
∂

= t vp [ f ] exp (t f (p) + x)Np
∂ x x=0
∂

+ exp(t f (p)N(γ(x)) )
∂ x x=0
= t vp [ f ]Uµ t (p) + dξs (vp ) .
Lemma 4.3.
If a vector field W ∈ X(M ) is everywhere tangent to the deformed
hypersurfaces µ t (M ), there holds for every p ∈ M and for every θ
  
∂

dµθ  d(µ← ) Wµ (p)  = f U, W .
∂ t t=θ
t t
µ (p) θ
This follows from the standard interpretation of the Lie bracket
∂

dΨ←
t (W ) = f U, W
∂ t t=0
where Ψ

t stands for the flow of the vector field f U. In particular, the Lie
bracket f U, X µ vanishes on M for all X ∈ X(M ).
Let us conclude these preparations by providing a proof of the
generalisation of two earlier-mentioned characteristic properties for the
Proposition 4.4 (The Second Fundamental Form as the Variation of

the First Fundamental Form).
The following expression for the variation of the first fundamental
form is valid:
δI = −2α f II + L Z t I .
Proof of proposition 4.4. It is sufficient to prove the theorem in case
76
Z = f N for some function f ∈ F(M ). By definition, the variation of the

first fundamental form is given by
∂

δI(V, W ) = g dµ t (V ), dµ t (W ) .

∂ t t=0
The above expression can be seen as the derivative of a scalar function

along the curve t 7→ µ t (p) (for any point p ∈ M ). This derivative can also
be calculated by means of the connection ∇ of (M , g). We obtain
= ∇ Z g(V µ , W µ )

= g(∇ Z V µ , W µ ) + g(V µ , ∇ Z W µ )
which by means of the previous lemma can be written as
= g(∇V µ Z, W µ ) + g(V µ , ∇W µ Z)
= g(∇V ( f N ), W ) + g(V, ∇W ( f N ))
= − f g(A(V ), W ) − f g(V, A(W )) .
The distance function to the hypersurface M is defined on an open neigh-

bourhood of M in M and defined by
ψ : M → R : exp(s Np ) 7→ s.
Proposition 4.5 (The Second Fundamental Form as the Hessian of the

Distance Function).
The distance function ψ with respect to the hypersurface M satisfies

 Hessψ = −II on T p M × T p M ;
Hessψ = 0 on T p M × T⊥

pM.
77
Proof of Proposition 4.5. It is clear that gradψ = α U throughout on M .

Assume vp , w p ∈ T p M and x p ∈ T p M are given. We see that

 Hessψ (vp , w p ) = g(∇ vp gradψ, w p ) = g(α∇ vp U, w p ) = −II(vp , w p ) ;
 Hess (x , U ) = g(∇ gradψ, U ) = αg(∇ U, U ) = 0 .
ψ p p xp p xp p
We are now ready to give two proofs of the theorem.
Proof of Theorem 4.1 (First Version). We remark that lemma 4.3 implies
that
∂

←
)(W ) = + f (p) A(Wp ) − Wp f Np (4.4)

d(µ µ (p) ∇ f N W (p)
∂t
t=0
t t
for W ∈ X(M ) tangent to the hypersurfaces µ t (M ) everywhere, and for

p ∈ M . Furthermore, since g(N , X µ ) = 0 on M , there holds for X ∈ X(M )
and p ∈ M ,
g(∇N N , X µ )(p) = − g(N , ∇N X µ )(p)

 
−α
= − g U, ∇U X ξ + t X f U  =

g(grad f , X )(p)

f (p)
(p)
From this it follows that

−α
∇N N = grad f (4.5)
f
on M and in consequence, on M holds
α α α
∇X ∇N N = X f grad f − ∇X grad f − II(X , grad f )N .

2
(4.6)
f f f
Now let S = −∇N denote the following (1, 1)-tensor on M :
S : X(M ) → X(M ) : Y 7→ S(Y ) = −∇Y N .
By making use of expression (4.4), it can be verified that

 
∂

µ

←
δA(X p ) = d(µ ) S X
µ t (p)

∂ s s=0 t
78
 
= ∇ f N (S(X µ )) + f (p) A2 (X p ) − (A(X p )) f Np

(p)

= −∇ f U ∇X µ N + ∇X µ ∇ f U N − ∇X µ ∇ f U N

2
+ f A (X ) − α II(X , grad f ) N 
(p)

= −R(X , f U)N − X f ∇N N − f ∇X ∇N N


+ f A2 (X ) − α II(X , grad f ) N  .
(p)
An application of equations (4.5–4.6) concludes the proof of the formula

for the variation of the shape operator. The formula for the variation of the
second fundamental form now follows from proposition 4.4.
Proof of Theorem 4.1 (Second Version). We let ψ t : M → R denote the

distance function to µ t (M ). It has been mentioned already that ψ t (q) is
strictly positive (resp. negative) for a point q ∈ M which is reached by a
geodesic emanating from µ t (M ) in the direction Nt (resp. −Nt ). There
holds α gradψ t = N on µ t (M ), and α gradψ0 = U throughout on M . In
the present proof use will be made of proposition 4.5.
First of all, we notice that the variation of the distance func-
tions is equal to minus the variation function:
∂

ψ t (q) = − f (q) (for q ∈ M .) (4.7)
∂ t t=0
To see this, we fix a point q = µ t 0 (p) ∈ M . (It will be assumed that t 0 > 0.)
Let γ be the geodesic joining p and q:
γ : 0, t 0 7→ M : t 7→ γ(t) = µ t (p) = exp t f (p)Np .

For every t, let β(t) be the end point of the shortest geodesic from q to
µ t (M ). Since β(t) ∈ µ t (M ), a point ζ(t) ∈ M can be chosen such that
79
Figure 4.2: Second proof of theorem 4.1.
µ t (ζ(t)) = β(t) (See figure 4.2). It follows that

∂

0
β (0) = µ(t, ζ(t)) = γ0 (0) + ζ0 (0) .
∂ t t=0
Let d denote the distance function of the ambient manifold (M , g). (Since
γ is not a null geodesic, the same holds true for the geodesics connecting q
to β(t) and β(t) to ζ(t), at least for small values of |t|. Consequently, this
function d is well-defined and smooth in the regions under consideration.)
It is a straightforward
consequence of the definition of β(0)
∂
that the quantity ∂ t d(ζ(t), q) vanishes, and consequently there holds
t=0
∂ ∂

ψ t (q) = d(β(t), q)
∂ t t=0 ∂ t t=0
∂ ∂

= d(γ(t), q) + d(ζ(t), q) = − f (q).
∂ t
t=0 ∂ t
t=0
Consequently, (4.7) is proven. Since ψ0 = s, there holds
Hessψ t = Hesss − tHess f + O (t 2 ) .
80
4.4. APPLICATIONS OF THE FORMULA. . . 4.4. 81
A (0, 2)-tensor field H along the geodesic γ will be defined by

H(t) = Hessψ t T .
γ(t) M ×Tγ(t) M
From this, the variation of the second fundamental form can be computed:
∂

µ µ
δII(X p , Yp ) = − H(t)(X µ (p) , Yµ (p) )
∂ t t=0 t t
 
= − ∇ f (p)Up H (X p , Yp ) − H(∇ f (p)Up X µ , Yp ) − H(X p , ∇ f (p)Up Y µ )
 
= − f (p) ∇Up Hesss  (X p , Yp ) + Hess f (X p , Yp )
−H(∇X p ( f U), Yp ) − H(X p , ∇Yp ( f U))

= − f (p) g ∇Up ∇X µ (grad s), Yp + f (p) g ∇(∇U X µ ) (grad s), Yp
p
+Hess f (X p , Yp ) − f (p) H(∇X p U, Yp ) − f (p) H(∇Yp U, X p ) .
Finally, this can be rewritten as the second equation of the theorem. The
formula for the variation of the shape operator follows by means of propo-
sition 4.4.
4.4. Applications of the Formula for the Variation of

the Shape Operator.
4.4.1. The Variation of the Curvatures of a Surface in a Three-

Dimensional semi-Riemannian Manifold.
Assume (M , g) is a Riemannian surface in a three-dimensional semi-Rie-
mannian manifold (M , g). We will now derive the formulae for the vari-
ation of the mean curvature H = α2 trA, the Gauss–Kronecker curvature
H2 = det A, the sectional curvature K of the tangent plane to M in M , and

the intrinsic (Gaussian) curvature K of M . By e1 , e2 will be denoted an
orthonormal basis of principal directions on the tangent spaces of M with
corresponding principal curvatures λ1 and λ2 (i.e., A(ei ) = λi ei ).
81
Theorem 4.6.
The variations of the curvatures of a Riemannian surface M in a three-
dimensional semi-Riemannian manifold (M , g) along a deformation
with variational vector field f N are given by
δH = f 12 Ric(e1 , e1 ) + 12 Ric(e2 , e2 ) + 12 ∆ f + f (2αH 2 − K) ;







  

 δH2 = 2 f α H H2 + λ2 α Hess f (e1 , e1 ) + f g(R(N ,e1 ) N , e1 )









  


 +λ1 α Hess f (e2 , e2 ) + f g(R(N ,e2 ) N , e2 ) ;







δK = f g( ∇N R (e1 , e2 )e1 , e2 ) + 2 Ric(grad f , N ) ;








δK = δK + α δH2 .
Proof of Theorem 4.6. The first two formulae follow immediately from
the expression for the variation of the shape operator. A short calculation
suffices to establish the third formula. The last formula follows from the
contracted Gauss equation H2 = α(K − K), but can be derived alternatively
from theorem 3.7.
4.4.2. Properties of Equidistant Surfaces in Space Forms.
We will now turn our attention to equidistant surfaces in semi-Riemannian

three-dimensional space forms. A classical theorem of Bonnet states that
there is an equidistant surface of a given surface in E3 with constant mean
curvature which has constant strictly positive Gaussian curvature, and con-
versely.
82
The purpose of this § is to illustrate how the formula for the

variation of the shape operator can be applied in the derivation of Bon-
net’s theorem for surfaces in space forms. In this way, the results which al-
ready have been described in [Klotz 1983] theorem 5, [Terng 1987] p. 103
and [Baikoussis–Defever–Koufogiorgos–Verstraelen 1996] for surfaces in
the three-dimensional Minkowski space E31 , the unit sphere, and Lorentzian
space forms, respectively, can be recovered.
We will restrict our attention to Riemannian surfaces in the
de Sitter space S31 . By a branched surface we will understand a surface
which may have singularities. The deforming mapping ξs between two
equidistant surfaces will be called a Bonnet transformation in this context.
Theorem 4.7.
Let M be a Riemannian surface in S31 with constant mean curvature
H satisfying H 2 > 1. If s = −arccotanh(H), the equidistant sur-
face ξs (M ) is a branched surface of constant mean curvature −H.
−arccotanh(H)
If s = 2
, the branched surface ξs (M ) has constant Gauss–
Kronecker curvature
1
p 2 .
−H + H 2 − 1
Conversely, assume M is a Riemannian surface in S31 withp constant

Gauss–Kronecker curvature H2 > 1. If s = arccotanh( H2 ), the
1+H
branched surface ξ±s (M ) has constant mean curvature ± p 2 .
2 H2
Proof of Theorem 4.7. Let a Riemannian surface M in S31 be given. It has

already been shown that there holds

 δI = 2II ;
δA = −id + A2 ; (4.8)
δII = I + III ,

under the unit normal deformation variation ξ.
83
A first consequence of these formulae (4.8) is that the principal

directions are preserved under the Bonnet transformation, in the sense that
dξs (e1 ) and dξs (e2 ) are the principal directions on ξs (M ). Thus, from (4.8)
the following two formulae can be inferred:
(
δλ1 = −1 + (λ1 )2 ;
δλ2 = −1 + (λ2 )2 .
This means that the differential equation
d 2
λ
Ø (ξ
i s ) = −1 + λØ (ξ
i s ) (i = 1, 2)
ds
is satisfied for s = 0. The above reasoning, applied to the surface ξs (M ) in-

stead of M , shows that the last-mentioned differential equation is generally
valid. There consequently holds
λi − (tanh s)
λ
Ø
i (ξs ) = (i = 1, 2) .
1 − (tanh s)λi
This implies

(tanh s) + (1 + (tanh s)2 )H + (tanh s)H2
 H(ξ
×
s ) = ;
1 + 2(tanh s)H + (tanh s)2 H2



(4.9)
(tanh s)2 + 2(tanh s)H + H2


2 (ξs ) =

HÙ .
1 + 2(tanh s)H + (tanh s)2 H2
and the theorem is now an easy consequence.
A space-like surface in S31 will be called of type (a), (b) or (c)

if the condition

 (a). A(1 − H2 ) = a(1 − H) for some A, a ∈ R (not both = 0) ;
(b). B(1 − H2 ) = b(1 + H) for some B, b ∈ R (not both = 0) ;
(c). C(1 + H2 ) = (not both = 0) ,

cH for some C, c ∈ R
is satisfied.
84
Theorem 4.8.
A branched surface, equidistant to a surface in S31 of type (a), (b) or
(c), is of the same type.
Proof of Theorem 4.8. As can be seen from equation (4.9), the Bon-
net transformations act on the (H, H2 )–curvature diagram D as a one-
parameter family of projective transformations, which means that straight
lines in D are mapped onto straight lines in D. The infinitesimal projec-
tive vector field on D which generates this action vanishes at exactly three
points. The conclusion is that the action of any Bonnet transform on a
straight line passing through such a zero is a straight line passing through
the same zero. See also figure 4.3.
Remark. If M ⊆ S13 is a Riemannian surface and s ∈ R, the surface ξs (M )

was called a branched surface since the Bonnet transformation
ξs : M → S31 : p 7→ exp(s Np )
is not necessarily an immersion. From the first equation of (4.8) together

with the expression for λØ
i (ξs ) a differential equation for g(dξs (ei ), dξs (ei ))
(as a function of s) is obtained, with the following solution:
2
λi tanh s − 1
g(dξs (ei ), dξs (ei )) = .
1 − (tanh s)2
Thus, for a given s, the collection of all points where ξs fails to be an

immersion is described by
λ1 = cotanh s or λ2 = cotanh s .
It follows that this set is “small” under “mild conditions.” For example, if
neither of both principal curvatures is constant, the set has measure zero.
85
Figure 4.3: The (H, H2 )–curvature diagram D for surfaces in the de Sitter space
is depicted above. Along the horizontal axis, the mean curvature H is measured;
along the vertical axis, the Gauss–Kronecker curvature H2 . (Only the domain H2 ¶
( H )2 , bounded by above by the black parabola, is of importance.) The Bonnet
transformations act on D as a one-parameter family of projective transformations
and the corresponding projective vector field has been displayed. The three zeroes
of this vector field have been indicated by black dots.
86
Chapter 5.
T H E G E O M E T RY O F T H E S E C O N D
FUNDAMENTAL FORM OF A HYPERSURFACE
IN A SEMI -R IEMANNIAN M ANIFOLD .
5.1. Introduction.
In this chapter, some of the results of the second chapter will be gener-
alised for hypersurfaces in an enveloping space which can be curved. Thus
we shall be concerned with semi-Riemannian hypersurfaces of a semi-Rie-
mannian manifold, of which the second fundamental form II is a semi-Rie-
mannian metrical tensor. In analogy with the second chapter, the geometry
of such a hypersurface can be explored with respect to either the first or
Some technical results, which should be compared with those
of § 2.3.2, are gathered in the initiating § 5.2. The four subsequent parts of
this chapter have been organised in a manner similar to the second chapter:
§ 5.3–§ 5.6 are devoted respectively to the intrinsic geometry of the second
fundamental form; the extrinsic geometry of the second fundamental form;
characterisations of extrinsic hyperspheres by means of the curvatures as-
sociated to the second fundamental form; and explicit examples. In the last
§ 5.7 the area functional of the second fundamental form is investigated for
sufficiently small geodesic hyperspheres in a Riemannian manifold.
87
88 5. THE GEOMETRY OF THE SECOND FUNDAMENTAL FORM OF A . . . 5.
|*4st'4
/l
/i
I
I
I
I
I
I
I
l
l
I
I
I
I
I
I
I
I
I
I
I
1r'4
Figure 5.1: The second fundamental form describes the distance function between
the hypersurface and its osculating geodesic umbrella.
It should be remarked that all four of the geometric interpreta-

tions of the second fundamental form can be adapted to this more general
situation. For example, the aforementioned relation between the distance
function ψ : M → R from M and the second fundamental form of M ⊆ M
cannot be literally taken over here since T p M is not a hyperplane in M any
more. However, the relation remains valid upon replacement of T p M by its
image under the exponential mapping of M at the point p.
Thus the second fundamental form of M at a point p ∈ M de-

scribes the distance function between the hypersurface M and the hyper-
88
5.2. SOME TECHNICAL RESULTS. 5.2. 89
surface of M which is spanned by all (M , g)–geodesics, emanating from the

point p in a direction tangent to M . This hypersurface is sometimes called
the “osculating geodesic umbrella” at the point p. (See also figure 5.1.)
5.2. Some Technical Results.
We repeat that all hypersurfaces are assumed to be connected and embedded.

Furthermore, a hypersurface will be said to be (semi-)Riemannian if the
first fundamental form is a (semi-)Riemannian metric.
We study a semi-Riemannian hypersurface (M , g) of (M , g).
The first fundamental form is denoted by g or, equivalently, I. All previous
notations, as introduced on pages 62 and 70, remain valid.
A hypersurface will be called totally umbilical if every point
is an umbilical point, i.e., there exists a function ρ ∈ F(M ), the normal
curvature function, for which the relation II = ρI is valid. A compact
and totally umbilical hypersurface of which the normal curvature function
is constant, will be called an extrinsic hypersphere. It is remarked that
an extrinsic hypersphere is not necessarily topologically equivalent to a
common hypersphere in the Euclidean space.
Throughout this chapter, (M , g) will stand for a semi-Riemann-
ian manifold of dimension m + 1. Further, M is a hypersurface of which
both the first and the second fundamental form are assumed to be semi-
Riemannian metrical tensors.
As earlier, a bar which is drawn over a symbol indicates that
the corresponding geometric object should be determined with respect to
the surrounding semi-Riemannian manifold (M , g), and a sub- or super-
script II signals that the object should be derived from the geometry of the
Let {e1 , . . . , em } denote a frame field on M , which is orthonor-
mal with respect to the first fundamental form I. Define "i (i = 1, . . . , m)
by "i = I(ei , ei ) = ±1.
Furthermore, let {V1 , . . . , Vm } be a frame field on M , which is
orthonormal with respect to the second fundamental form II. Define κi
(i = 1, . . . , m) by κi = II(Vi , Vi ) = ±1.
89
5.2.1. The Tensor B and the Vector Field Z .

We define a tensor B by
¨ «
t
←
B : X(M ) × X(M ) → X(M ) : (V, W ) 7→ A R(V,N )W .
This tensor vanishes if the surrounding space (M , g) is a space form. The

vector field Z in X(M ) is defined by
m
X
Z = trII B = B(Vi , Vi ) κi . (5.1)
i =1
If M has dimension three and A is diagonalisable, the expression

t
Rc N
Z =α .
det A
for the vector field Z is valid.
5.2.2. The Differential Invariants of the Second Fundamental

Form.
The following lemma can easily be deduced.
Lemma 5.1.
Every function f ∈ F(M ) satisfies
α A gradII f = grad f .

The following expression for the connection of the second fun-

damental form generalises Weingarten’s equation (2.1) on page 24.
Lemma 5.2 (The Connection of the Second Fundamental Form).

For all vector fields X , Y ∈ X(M ), there holds
n o
∇IIX Y = 21 ∇X Y + 21 A← ∇X (AY ) − 12 B(X , Y ) .
90
Proof of Lemma 5.2. The Koszul formula for the second fundamental form
reads (for X , Y, Z ∈ X(M ))

2 II ∇IIY X , Z = Y [II(X , Z)] + X [II(Z, Y )] − Z [II(Y, X )]
− II (Y, ¹X , Zº) + II (X , ¹Z, Y º) + II (Z, ¹Y, X º)
= α Y g(X , AZ) + α X g(Z, AY ) − α Z g(Y, AX )

− α g(AY, ∇X Z) + α g(AY, ∇ Z X )
+ α g(AX , ∇ Z Y ) − α g(AX , ∇Y Z)
+ α g(AZ, ∇Y X ) − α g(AZ, ∇X Y ) .
If some terms of the right-hand side are collected, this can be written as
§ ª
= α g(Y, (∇X A)Z) + g X , (∇Y A)Z − (∇ Z A)Y

+ 2 II(∇Y X , Z) .
An application of the Codazzi equation (4.3) and the symmetry of R finally

reduces this to
n o
= II 2∇Y X + A← ∇X A Y − B(X , Y ), Z .
The following expression results:

n o
2∇IIY X = 2∇Y X + A← ∇X A Y − B(X , Y ) . (5.2)
By making use of the fact ∇X Y − ∇Y X = ¹X , Y º = ∇IIX Y − ∇IIY X , the last

equation can be rewritten as
n o
2∇IIX Y = 2∇X Y + A← ∇X A Y − B(X , Y )
n o
= ∇X Y + A← ∇X (AY ) − B(X , Y ) .
5.2.3. The Difference Tensor L and some Related Tensors.

The difference tensor L between the two Levi-Civita connections ∇II and ∇
is defined as
L(V, W ) = ∇IIV W − ∇V W ,
91
Figure 5.2: Interpretation of the difference tensor in terms of parallel transport.
where V, W ∈ X(M ). It follows from the above formula (5.2) that
1 n o 1
L(V, W ) = A← (∇V A)W − B(V, W ) . (5.3)
2 2
Remark. The difference tensor L can be interpreted easily in terms of par-

allel transport. Assume p ∈ M and v, w ∈ T p M are given. Choose a curve
γ : R → M such that γ(0) = p and γ0 (0) = w. By v"• we will denote the
vector of Tγ(") M obtained by parallel translation of v along γ with respect
Æ
to ∇. By v" we will denote the vector of T p M which is obtained by par-
allel transport of the vector v"• back to p along γ with respect to ∇II (see
figure 5.2). It is not hard to show that
Æ
v" − v
L(v, w) = lim .
"→0 "
Of course, the tensor L is symmetric. According to the next

lemma, the associated cubic form is totally symmetric if (M , g) is a space
form.
Lemma 5.3.
The difference tensor L satisfies the following symmetry formula:
92
II(L(V, W ), Z) − II(L(V, Z), W ) = −αg(R(V,N ) W, Z) .
Proof of Lemma 5.3. This follows immediately from formula (5.3).
The different traces of L are defined in the same way as on

page 25 (although the lemmata, as given there, do not remain valid).
Lemma 5.4.
The traces of the difference tensor L are given by
1

traceL = d log |det A| ;
2





 1
trII L =

gradII log |det A| − Z ;

 2

 m § t ª
tr L = gradII H − A←

Rc N .


2
Proof of Lemma 5.4. With help of the expression (5.3), we deduce (for
X ∈ X(M ))
m
X
(traceL)(X ) = II(L(X , Vi ), Vi ) κi
i=1
m m
1X n o αX
= II A← (∇X A)Vi , Vi κi − g R(X , N )Vi , Vi κi .
2 i=1 2 i=1
The second sum vanishes by means of the symmetries of R. The first trace
can be written alternatively by means of Jacobi’s formula for the derivative
of the determinant.
1
= trace A← ◦ (∇X A)

2
93
1 X [det A]
= .
2 det A
The second formula follows from the above calculation and the symmetry
property of the difference tensor:
m
X
II trII L, X = II(L(Vi , Vi ), X ) κi
i=1
m §
X ª
= II(L(Vi , X ), Vi ) − αg R(Vi , N )Vi , X κi
i=1
= (trace L)(X ) − II(trII B, X )

1
= II gradII log |det A| , X − II(Z , X ) .

2
It similarly follows from expression (5.3) that
m m
αX αX
II (trL, X ) = g (∇ei A)ei , X "i − g R(ei , N )ei , X "i .
2 i=1 2 i=1
If the Codazzi equation is applied, this becomes:

m
α X
= trace(∇X A) − α g R(ei , X )ei , N
2 i=1
α
= X [trace A] − α Ric(X , N ) .
2
Lemma 5.5.
If the dimension of the hypersurface M is equal to m = 2, then
2 II(Z , Z ) = II(B, B) .
Proof of Lemma 5.5. An immediate consequense of the symmetries of the

curvature tensor R is the fact that
(
II B(V1 , V1 ) , V1 = 0 ;

II B(V2 , V2 ) , V2 = 0 ,

94
which means that B(V1 , V1 ) is proportional to V2 and vice versa. In particu-

lar,
II(Z , Z ) = II(trII B , trII B)

= II B(V1 , V1 ) κ1 + B(V2 , V2 ) κ2 , B(V1 , V1 ) κ1 + B(V2 , V2 ) κ2

B(V , V ) 2
X
= ı ı II
ı
X 2
= II B(Vı , Vı ), V)
ı
X 2
= g R(Vı , N )Vı , V .
ı
By making another time use of the symmetries of the curvature tensor of

M , we obtain
1 X 2
= g R(Vı , N )V, Vk
2 ı k
 2
1X
= II B(Vı , V), Vk 
2 ı k
1
= II(B, B) .
2
The following lemmata remain unaltered, as compared to the
second chapter.
Lemma 5.6.
Every vector field X ∈ X(M ) satisfies
divII X = div X + (traceL)(X ) .
Lemma 5.7.
For a function f ∈ F(M ) there holds
∆II f = trII Hess f −II(gradII f , trII L) .
95
Furthermore, we define the operator ` as Z ù L. This means:
` : X(M ) → X(M ) : Y 7→ L(Z , Y ) .
There holds
1
trace ` =

II gradII log |det A| , Z . (5.4)
2
5.2.4. A Lemma on the Connections.
In the sequel of this chapter we will rely on the following lemma (see also
[Hicks 1965, Thm. 7; Simon 1972] and [Gardner 1972a, Cor. 13]).
Lemma 5.8.
Let M be a compact hypersurface of a semi-Riemannian manifold
(M , g). Suppose that both the first and the second fundamental form
are positive-definite and that these metrical tensors induce the same
Levi-Civita connection. Furthermore, assume that (M , I) has either
strictly positive or strictly negative sectional curvature. Then M is an
extrinsic hypersphere.
Proof of Lemma 5.8 (First Version). As an immediate consequence of

∇ = ∇II , we see that
R(X , Y )Z = RII (X , Y )Z
holds for all X , Y, Z ∈ X(M ). Let p ∈ M be an arbitrary point and choose

an orthonormal basis {e1 (p), . . . , em (p)} of principal directions:
A(ei (p)) = λi (p) · ei (p) (i = 1, . . . , m) .
A smooth orthonormal frame field {e1 , . . . , em } which extends these vectors

on a neighbourhood of p in M can be chosen. For any choice of i 6= j ∈
{1, . . . , m}, there holds
!
II RII (ei , e j )ei , e j
KII (ei (p), e j (p)) =
II(ei , ei ) II(e j , e j )
(p)
96
!
α λ j I R(ei , e j )ei , e j
=
λi λ j
(p)
α
= K(ei (p), e j (p)) .
λi (p)
Since the above equation remains valid if the rôle of i and j is interchanged
and K(ei (p), e j (p)) 6= 0, it follows that M is totally umbilical. This means
that II = ρ I for a function ρ : M → R. Furthermore, for all X , Y, Z ∈ X(M ),

0 = ∇IIX II (Y, Z) = ∇X II (Y, Z) = X [ρ] I(Y, Z) .

Consequently, the normal curvature function ρ is constant.
Proof of Lemma 5.8 (Second Version). Assume that the conditions, as

stated in the lemma, are satisfied for a hypersurface which is not an ex-
trinsic hypersphere. Thus the open set of all non-umbilical points on M
is supposed to be non-empty. Hence a non-umbilical point p ∈ M can be
chosen, in a neighbourhood of which the principal directions {e1 , . . . , em }
are smooth. The fact that p is non-umbilical means, after a possible re-
numbering of indices, that

λ1 (p) = λ2 (p) = · · · = λk (p) ;
 λ1 (p) 6= λk+1 (p) ;


..


 .
λ1 (p) 6= λm (p) ,
for some k ∈ {1, . . . , m − 1} .

Since the first and the second fundamental form induce the
same Levi-Civita connection, there holds ∇A = 0. Thus for X ∈ X(M ) and
i, j ∈ {1, . . . , m}, there holds

0 = g (∇X A)ei , e j = (λi − λ j )g(∇X ei , e j ) − X [λi ]δi j ,
and in particular,
0 = g(∇X ei , e j ) (for 1 ¶ i ¶ k < k + 1 ¶ j ¶ m.) (5.5)
97
Now choose a curve γ on M satisfying γ(0) = p = γ(1), and

choose a vector vp ∈ span ek+1 (p), . . . , em (p) . The vector field, which is

obtained by parallel translation of this vector vp along the curve γ, will be

denoted by V = ξ1 e1 + · · · + ξm em . Taking the fact (5.5) into account, we
deduce that

0 = g(∇γ̇ V, e1 ) = ξ̇1 + ξ2 g(∇γ̇ e2 , e1 ) + · · · + ξk g(∇γ̇ ek , e1 ) ;
..

 .
0 = g(∇γ̇ V, ek ) = ξ̇k + ξ1 g(∇γ̇ e1 , ek ) + · · · + ξk−1 g(∇γ̇ ek−1 , ek ) .
The crucial remark which has to be made is that the functions ξk+1 , . . . , ξm
do not enter in these equations because of (5.5), whence a system of first-
order differential equations in the functions ξ1 , . . . , ξk with initial condi-
tions ξ1 (0) = 0, . . . , ξk (0) = 0 is obtained. The conclusion is that ξ1 ≡
· · · ≡ ξk ≡ 0.
It has thus been shown that the action of the local holonomy
group leaves the subspace span ek+1 (p), . . . , em (p) of T p M invariant.

Now the local de Rahm theorem can be applied, which gives

us that a neighbourhood of p ∈ (M , g) is isometric to a Cartesian product
of Riemannian manifolds. This is in contradiction with the assumption that
(M , g) has non-vanishing sectional curvature.
5.3. The Intrinsic Geometry of the Second Funda-

mental Form.
The starting point of our investigation of the intrinsic geometry of the sec-
ond fundamental form shall be the corresponding scalar curvature SII . The
calculation of this curvature proceeds entirely in the same manner as out-
lined in [Aledo–Romero 2003, Aledo–Alías–Romero 2005, Aledo–Haesen–
Romero 2007, Haesen 2007].
5.3.1. The Scalar Curvature of the Second Fundamental Form.

The Riemann–Christoffel curvature tensor of the second fundamental form
will be denoted by RII . Let Q 1 and Q 2 stand for the tensors of type (3, 1),
98
5.3. THE INTRINSIC GEOMETRY OF THE SECOND FUNDAMENTAL. . . 5.3. 99
carrying a triple of vector fields X , Y, Z ∈ X(M ) onto


 Q 1 (X , Y )Z = ∇ II
Y L (X , Z) − ∇II
X L (Y, Z) ;
Q 2 (X , Y )Z = L X , L(Y, Z) − L Y, L(X , Z) .

Some rote computation results in
RII (X , Y )Z = R(X , Y )Z + Q 1 (X , Y )Z + Q 2 (X , Y )Z .
The tensors Q
f1 and Q
f2 are obtained from Q 1 and Q 2 by taking the trace:

 Q 1 (X , Y ) = trace{Z 7→ Q 1 (X , Z)Y } ;
f
f2 (X , Y ) = trace{Z 7→ Q 2 (X , Z)Y } .

Q
Since also Ric and RicII are obtained from R and RII as a trace

 Ric(X , Y ) = trace{Z 7→ R(X , Z)Y } ;
RicII (X , Y ) = trace{Z 7→ RII (X , Z)Y } .


there immediately results
RicII (X , Y ) = Ric(X , Y ) + Q
f1 (X , Y ) + Q
f2 (X , Y ) .
Lemma 5.9.
The II-trace of the tensor Q
f1 is given by
f1 = − divII Z .
trII Q
Proof of Lemma 5.9.

X
f1 =
trII Q II Q 1 (Vi , Vj )Vi , Vj κi κ j
ij
99
X X
= II ∇IIV L (Vi , Vi ) , Vj κi κ j − II ∇IIV L (Vj , Vi ) , Vj κi κ j
j i
ij ij
If the above expression is expanded so that lemma 5.3 can be applied, there
results
¨
X h t i t
= Vj −g R(Vi , N )Vi , Vj − αg R(Vi , N ) ∇IIV Vj , Vi
j
ij
«
t t
− αg R(∇IIV Vj , N )Vi , Vj − αg R(Vj , N )Vi , ∇IIV Vj κi κ j α ,
i i
which, with a little work, can be rewritten as
= − divII Z .
Lemma 5.10.
The II-trace of the tensor Q
f2 satisfies
1
f2 = II(L, L) − II(trII L, trII L) − trace ` − II(B, B) + II(Z , Z ) .
trII Q
2
Proof of Lemma 5.10. The different definitions give us, together with the
symmetry of L, that
m
X
f2 =
trII Q II Q 2 (Vi , Vj )Vi , Vj κi κ j
i j=1
m
X

= II L(Vi , L(Vj , Vi )) , Vi − II L(Vj , L(Vi , Vi )) , Vj κi κ j
i j=1
Xm
= II L(Vi , Vj ) , L(Vj , Vi ) − α g R(Vi , N )L(Vj , Vi ) , Vj
i j=1

− II L(Vj , Vj ) , L(Vi , Vi ) + α g R(Vj , N )L(Vi , Vi ) , Vj κi κ j
= II(L, L) − II(trII L, trII L)
100
m
X

+α g R(Vi , N )Vj , L(Vj , Vi ) − g R(Vj , N )Vj , L(Vi , Vi ) κi κ j .
i j=1
By means of (5.3), this can be written as the following expression, which

can be reworked to the desired formula with help of equation (5.1), the
second formula of lemma 5.4, and equation (5.4):
= II(L, L) − II(trII L, trII L)

m
1X

− II B(Vi , Vj ) , B(Vi , Vj ) + 2 II B(Vj , Vj ) , L(Vi , Vi ) κi κ j .
2 i j=1
Theorem 5.11 (The Scalar Curvature of the Second Fundamental

Form).
The scalar curvature of the second fundamental form is given by
SII = trII Ric − divII Z + II(L, L) − II(trII L, trII L) − trace `

1
+II(Z , Z ) − II(B, B) .
2
Proof of Theorem 5.11. This is an easy consequence of the last two lem-
mata.
Is is noteworthy to point out that the above formula simplifies

considerably in several special cases.
For hypersurfaces in the Euclidean space Em+1 , this becomes
II(gradII det A, gradII det A)

SII = m(m − 1)H + II(L, L) − , (5.6)
4(det A)2
in accordance with the formula on page 29.

A short calculation, starting from the above theorem, shows
that the scalar curvature of the second fundamental form of a hypersur-
face M in a semi-Riemannian space form (M , g) of dimension m + 1 and
101
sectional curvature C, satisfies
II(gradII det A, gradII det A)

SII = α(m−1) α trace A+C trace A← +II(L, L)−

.
4(det A)2
(5.7)
This has been mentioned already in [Aledo–Alías–Romero 2005, Aledo–
Romero 2003].
Further, the formula simplifies in case the dimension of the
surrounding space is equal to three. In this case the formula for the scalar
curvature SII = 2KII of the second fundamental form can be simplified due
to lemma 5.5:
HK 1 1 1 1
KII = − divII Z + II(L, L) − II(trII L, trII L) − trace ` (5.8)
det A 2 2 2 2
in accordance with [Aledo–Haesen–Romero 2007, Haesen 2007].
Finally, if the hypersurface is a curve in a two-dimensional
semi-Riemannian manifold, the expected formula SII = 0 is found.
5.3.2. Applications for Surfaces in the Hyperbolic Space.
It has been noticed already in [Aledo–Romero 2003] that many of the char-
acterisations of Euclidean spheres among ovaloids, in which the curvatures
of their second fundamental form are involved, can be generalised to sur-
faces in the de Sitter space. As an application of the formula for the scalar
curvature of the second fundamental form, these theorems will be adapted
similarly to surfaces in the three-dimensional hyperbolic space H3 . In this
way, some new theorems are obtained (see also [Verpoort 0000b]).
As can be inferred from the contracted Gauss equation det A =
K + 1, a unit normal vector field can always globally be chosen on a sur-
face in H3 with strictly positive Gaussian curvature, in such a way that the
second fundamental form becomes positive-definite. This will implicitly be
assumed for surfaces satisfying K > 0.
Further, the expression for the Gaussian curvature of the sec-
ond fundamental form reduces to

HK 1 II gradII K , gradII K
KII = + II(L, L) − .
K +1 2 8(K + 1)2
102
Cartan’s formula can also be adapted for surfaces in H3 with a positive-

definite second fundamental form [Klotz 1975]:
λ2 λ1

HK 1
KII = + e1 [λ2 ] e1 log + e2 [λ1 ] e2 log .
K +1 4(K + 1) λ1 λ2
We also remark that Erard’s formula (Q) of page 29 can be adapted for
surfaces in H3 with a positive-definite second fundamental form:
H2 H2

det A 1
II gradII , gradII H − I grad , grad det A (Q)
2 det A 4 det A
HK

= 2H KII − (H 2 − det A) .
det A
It is straightforward that the above equation is satisfied at the umbilical

points of the surface. The equation can be established in a region which
is free of umbilical points, by expanding all terms which occur in the II-
orthonormal basis {V1 , V2 }.
We will make use of the fact that every compact, oriented sur-
face in the hyperbolic space H3 has a point where the Gaussian curvature
is strictly positive (see [Aledo–Alías–Romero 2005], p. 1820).
The following alternative proof of a Liebmann–Hilbert theo-
rem is only a small adaption of a result which has been given in [Aledo–
Alías–Romero 2005].
Theorem 5.12.
Every compact surface in H3 with constant Gaussian curvature is an
extrinsic sphere.
Proof of Theorem 5.12. According to a previous remark, the constant

K = det A − 1 is strictly positive. Thus the second fundamental form with
the appropriate unit normal is a positive-definite form with curvature
HK 1 HK K
KII = + II(L, L) ¾ ¾p .
det A 2 det A det A
103
Thus the function KII − p K is non-negative. On the other

det A
hand, the Gauss–Bonnet theorem gives that
Z Z Z
K
KII dΩII = K dΩ = p dΩII .
det A
This is only possible if KII = p K , and this clearly implies that the differ-
det A
ence tensor L vanishes. Lemma 5.8 finishes the proof.
Theorem 5.13.
Let M be a compact surface in the hyperbolic space H3 , and assume
that the second fundamental form of M is positive-definite. The
Gaussian curvature of the second fundamental form is constant if and
only if M is an extrinsic sphere.
Proof of Theorem 5.13. The proof proceeds along similar lines as the
proof we gave for Schneider’s theorem. If a surface in H3 is given, which
satisfies the mentioned conditions and most notably has a second funda-
mental form of constant curvature KII , choose a point p+ where K attains
its maximal value. In particular, K|(p+ ) is strictly positive, such that every
q ∈ M satisfies
     
H K K K
KII (q) = KII (p ) ¾  p p  ¾ p  ¾ p 
+
det A K + 1 K +1 det A
(p+ ) (p+ ) (q)
(where we have made use of the fact that the function ] −1 , +∞ [ → R :

x
x 7→ p1+x is increasing). We conclude that
Z Z p Z Z
KII dΩII = KII det A dΩ ¾ K dΩ = KII dΩII .
K
This is only possible if the equality KII = p1+K is satisfied on M . In partic-
ular, K is constant, and the result follows from the previous theorem.
Conversely, it is plain that the function KII is constant for every
extrinsic sphere.
104
The following theorem, which is reminiscent to [Koutroufiotis 1974], has

already been mentioned in [Haesen 2007] as prop. 13.
Theorem 5.14.
Assume M ⊆ H3 is a compact surface in the hyperbolic space which
has strictly positive Gaussian curvature. Then C K = KII (for some
C ∈ R) if and only if M is an extrinsic sphere.
Proof of Theorem 5.14. First of all, it should be remarked that C > 0.

This is an immediate consequence of the Gauss–Bonnet theorem.
Let now p− be a point of M where the function det A = K + 1
attains its minimal value. The given condition implies that

C K|(p− ) = KII (p ) = + II(L, L) −
− K +1 2 8(K + 1)2 (p− )
HK

¾ .
K + 1 (p− )
Due to the fact that K|(p− ) is strictly positive, it follows that

p
C (K + 1) ¾ H|(p− ) ¾ K + 1 ,

(p− ) (p− )
and we conclude that every q ∈ M satisfies

p p
C K + 1 ¾ C K + 1 ¾ 1.

(q) (p− )
p p
This means that KII det A = C K K + 1 ¾ K, and by integration there
results Z Z Z
p
KII dΩII = KII det A dΩ ¾ K dΩ .
p
Thus the equality C KII K + 1 = C K = KII is valid, which can only be the
case if K is a constant. This finishes the proof.
The next lemma will enable us to generalise a result of [Simon 1976] in
the subsequent theorem.
105
Lemma 5.15.
Let M ⊆ H3 be a compact surface with positive-definite second funda-
mental form. If a point p can be found where KII has a global mini-
mum while K has a global maximum, then M is an extrinsic sphere.
Proof of Lemma 5.15. It is known that K(p) is strictly positive, and it

follows in this way that every point q ∈ M satisfies
H K K K

KII (q) ¾ KII (p) ¾ p
p ¾ p ¾ p .
det A K + 1 (p) K + 1 (p) K + 1 (q)
A twofold application of the Gauss–Bonnet theorem thus gives us that the

integral of a non-negative function is zero:
Z
K

0= KII − p dΩII .
K +1
Consequently, this integrand vanishes identically, and every inequality in

the above reasoning is an equality. This means that KII is constant, such
that M is an extrinsic sphere.
Theorem 5.16.
Let M ⊆ H3 be a compact surface with a positive-definite second fun-
damental form. Assume that the condition F (K, KII ) = 0 is fulfilled
on M for a function F : R2 → R : (u, v) 7→ F (u, v) which satisfies the
following requirements:
¨ ¨
Fu > 0 ; Fu ¾ 0 ;
or
Fv ¾ 0 , Fv > 0 .
Then M is an extrinsic sphere.
(In particular, this condition is satisfied if K = f (KII ) or KII =

f (K) for a decreasing function f .)
106
Proof of Theorem 5.16. Suppose that a surface M ⊆ H3 satisfies the

conditions as formulated in the theorem for some function F , including
the first set of requirements on Fu and F v . Let q ∈ M be a point where
K achieves its global maximum, and r ∈ M a point where KII achieves a
global minimum. Assume that M is not an extrinsic sphere. According to
the previous lemma, there necessarily holds K(q) > K(r) , and hence

0 = F K(q) , KII |(q) > F K(r) , KII |(q) ¾ F K(r) , KII |(r) = 0 .
This is clearly a contradiction. The case in which the second set of condi-
tions is satisfied is similar.
The following result, which is similar to [Koutroufiotis 1974], also follows
from prop. 14 of [Haesen 2007].
Theorem 5.17.
A compact surface M ⊆ H3 with strictly positive Gaussian curvature is
K
either an extrinsic sphere, or the function KII − p changes sign.
det A
Proof of Theorem 5.17. (First Version). Assume that this function does
not change sign on a compact surface M (with K > 0) in the hyperbolic
K
space. Then there obviously holds KII = p . Let the function ϕ be
det A
defined as
H

ϕ : M → R : p 7→ p −1 . (5.9)
K +1 (p)
If M is not an extrinsic sphere, then the point p+ where this function

reaches its maximum is a non-umbilical point satisfying ϕ(p+ ) > 0. This
implies that we can describe a neighbourhood of the point p+ with lines-
of-curvature co-ordinates (u, v), in which the fundamental forms can be
written as (
I = E du2 + G dv 2 ;
II = E du2 + G dv 2 .
107
In this co-ordinate system, the intrinsic curvatures of these forms satisfy

 (    )
−1 ∂ 1 ∂E ∂ 1 ∂G
K = p p + p  ;
∂v ∂v ∂u ∂u

2 EG EG EG




   ) (5.10)
 (
−1 ∂ 1 ∂E ∂ 1 ∂G


K = p

p + p  .
II
2 EG ∂ v EG ∂ v ∂u EG ∂ u
K
Our assumption KII = p implies that
det A
p p
EG KII = EG K . (5.11)
The Codazzi equations can be written as


∂E ∂E
 H = ;
 ∂v ∂v


(5.12)
 ∂G ∂G
=

H
 .
∂u ∂u
If first the Codazzi equations (5.12) and the definition (5.9) of ϕ are used,
and then a combination of (5.10) and (5.11), it follows that
   
∂ ∂v E ∂ ∂u G
p ϕ + p ϕ
∂v EG ∂u EG
   « ¨    «
∂ ∂v E ∂ ∂u G ∂ ∂v E ∂ ∂u G
¨
=− p + p  + p  + p 
∂v EG ∂u EG ∂v EG ∂u EG
= 0.
This in turn implies that
2 E G K ϕ = (∂ v E) ∂ v ϕ + (∂u G) ∂u ϕ .
The left-hand side of the above equality is strictly positive at the point p+ ,
whereas the right-hand side disappears. This contradiction establishes that
M cannot have non-umbilical points.
108
Proof of Theorem 5.17. (Second Version). A contradiction at the point

p+ , which is chosen as in the previous proof, can be obtained in another
way. Namely, formula (Q) reduces at the point p+ to
 
H K

0 = 2H H 2 − det A KII −  .
det A
(p+ )
The same assumptions on the point p+ as in the previous proof, imply that
the right-hand side is strictly negative.
Corollary 5.18.
A compact surface M ⊆ H3 with strictly positive Gaussian curvature is
a sphere as soon as any of the following conditions is satisfied:

K
 (i.) KII ¾ p ;
K +1






 K
(ii.) KII ¶ p ;


 K +1


 (iii.) K ¶ K .


II
H
The following theorem generalises theorem 29 of [Aledo–Ro-

mero 2003].
Theorem 5.19.
Let M ⊆ H3 be a compact surface with strictly positive Gaussian cur-
vature. Assume real numbers C, r and s can be found, subject to the
conditions 0 ¶ s ¶ 1, r ¶ 1, and 2r + s ¶ 1, such that the equation
KII = C H s K r
is satisfied. Then M is an extrinsic sphere.
109
Proof of Theorem 5.19. Let ϕ stand for the function
x 1−r
ϕ : ] 0 , +∞ [ → R : x 7→ s+1
.
(1 + x) 2
It should be remarked that ϕ 0 (x) ¾ 0 for all x ∈ ] 0 , +∞ [. Furthermore,

the constant C has to be strictly positive. Let K assume its maximum in a
point p+ . Then we conclude

(C H s K r )(p+ ) = KII p
+
= + II(L, L) −
K +1 2 8(K + 1)2 (p+ )
HK

¾ .
K + 1 (p+ )
s−1
Since s − 1 ¶ 0, it follows that (K + 1) 2 ¾ H s−1 . Consequently,
1

s−1
C (K + 1) 2 K r−1
¾ CH s−1
K r−1
(p+ )
¾
(p+ ) K +1 (p+ )
and hence for all q ∈ M

   
K 1−r K 1−r
C ¾ s+1
 = ϕ(K)(p )
¾ ϕ(K)(q) =  s+1
 .
+
(K + 1) 2
(p+ ) (K + 1) 2
(q)
Thus there holds

p p Hs K
KII K + 1 = C Hs K r K +1¾ s ¾K.
(K + 1) 2
(In the last step we needed s ¾ 0.) The theorem follows by virtue of corol-
lary 5.18.
The lemma below is a preparation for theorem 5.21, which

generalises a theorem of [Hasanis 1982].
110
Lemma 5.20.
Assume a surface M ⊆ H3 has positive-definite second fundamental
HK

form. If p ∈ M is a critical point of H, then KII (p) ¾
.
det A (p)
Proof of Lemma 5.20. Let thepoint pbe such as in the lemma and suppose
HK
that the inequality KII (p) < holds. Since the gradient of H
det A (p)
vanishes at the point p, formula (Q) implies the following at the point p:
H2
¨ «

0 ¶ I grad det A, grad det A
4(det A)2 (p)
H K

= 2 H (H 2 − det A) KII − .
det A (p)
Furthermore, 2 H (H 2
− det A) is non-negative. The above inequality
can
only be valid if H (p) = det A|(p) , and consequently grad det A (p) vanishes.
2
Thus, we deduce the following contradiction:
   
HK 1 HK
KII (p) =  + II(L, L) ¾   .
det A 2 det A
(p) (p)
Theorem 5.21.
vature. Assume real numbers C, r and s can be found, subject to the
condition −1 ¶ r ¶ −1
2
, such that the equation
KII = C H s (K + 1) r K
is satisfied. Then M is an extrinsic sphere.
Proof of Theorem 5.21. It will first be shown that
1 ¶ C H s+2r+1 . (5.13)
111
1. First Case: s + 2r + 1 ¾ 0. Let p− be a point where H achieves its

minimum, and choose an arbitrary point q ∈ M . It is known that
 
det A ¦ ©
H(p− ) ¶  KII  = C H s (K + 1)1+r (p )
K −
(p− )
and hence
¦ ©
1 ¶ C H s−1 (K + 1)1+r (p )
−
¦ ©
s−1+2+2r s+2r+1
¶ C H (p )
¶ C H (q)
.
−
2. Second Case: s + 2r + 1 ¶ 0. Let p+ be a point where H achieves its

maximum, and choose an arbitrary point q ∈ M . It is known that
 
det A ¦ ©
H(p+ ) ¶  KII  = C H s (K + 1)1+r (p )
K +
(p+ )
and hence
¦ ©
1 ¶ C H s−1 (K + 1)1+r (p )
+
¦ ©
s−1+2+2r s+2r+1
¶ C H (p )
¶ C H (q)
.
+
It follows from (5.13) that

−1−2r 1
(K + 1) 2 ¶ H −1−2r = ¶ C Hs ,
H 1+2r
and hence also
p p
K ¶ C H s (K + 1) r K K + 1 = KII K + 1.
The result now follows from corollary 5.18.
Lemma 5.22.
Let M ⊆ H3 be a surface with positive-definite second fundamental
112
H
form. If p• ∈ M is a critical point of det A
, there holds
 
HK
KII (p ) ¾   .
• det A
(p• )
H
Proof of Lemma 5.22. Since p• is a critical point of det A
, the formula
H2

• H
gradII = gradII H
det A det A
•
and similar ones are valid. (The symbol = indicates that both sides of an
equation should be evaluated at the point p• .) In this way, formula (Q) can
be rewritten as follows at the point p• :
1 HK

←
•
2
I grad H , {2Hid − A} ◦ {A } grad H = 2H H − det A KII −

.
4 det A
Since both operators between curly brackets in the above formula are positive-
definite, we deduce
HK

•
2
0 ¶ 2H H − det A KII − .
det A
•
HK
Assume first that the inequality KII ¾ det A
is not satisfied. In regard of
•
the above inequality, this can only be the case if H 2 = det A. But now the
• •
rewritten formula (Q) reveals that grad H = 0 and hence also gradII det A =
0. Consequently,
• HK 1 • HK
KII = + II(L, L) ¾ ,
det A 2 det A
which is in contradiction with our assumption. This finishes the proof.
The next two theorems generalise results of [Baikoussis–Kou-

fogiorgos 1987a].
113
Theorem 5.23.
vature. If the equality
KII = C H s (K + 1) r K
−1
is satisfied for real numbers C, s and r, subject to the condition 2
¶
r + s ¶ 0, then is M an extrinsic sphere.
H
Proof of Theorem 5.23. We first remark that, for a critical point p• of det A
,
• H −1−s−2r
there holds 1 ¶ C det A . Namely, an application of the previous
lemma gives that
• • HK
C H s (K + 1) r K = KII ¾
K +1
and hence
•
1 ¶ C H s−1 (K + 1) r+1
•
= C H s−1 (K + 1) r+1+(s+r) (K + 1)−(s+r)
•
¶ C H s−1−2(s+r) (K + 1) r+1+(s+r)
•
= C H −1−s−2r (K + 1)1+s+2r
−1−s−2r
H

•
= C .
K +1
It is now an easy consequence that the equality
−1−s−2r
H

1¶C (5.14)
det A
holds on the entire surface M .
H
1. First Case: 1 + s + 2r ¶ 0. Let now p− be a point where det A
assumes
its minimum. There holds, for every point q ∈ M ,
−1−s−2r −1−s−2r
H H

1¶C ¶C .
K + 1 (p− ) K + 1 (q)
114
H
2. Second Case: 1+s+2r ¾ 0. Let now p+ be a point where det A
assumes
its maximum. There holds, for every point q ∈ M ,
−1−s−2r −1−s−2r
H H

1¶C ¶C .
K +1 (p+ ) K +1 (q)
It now follows from (5.14) that
K ¶ C H −1−s−2r (K + 1)1+s+2r K
1
p
= C H s (K + 1) r K H −1−2s−2r (K + 1) 2 +s+r K +1
2 − 1 −s−r
H 2 p
= KII K +1
det A
p
¶ KII K + 1 .
The theorem follows from corollary 5.18.
Lemma 5.24.
K
Let p be a point of a compact surface M ⊆ H3 with K > 0 where KII
H
achieves its absolute minimum and det A
achieves its absolute maxi-
mum. Then M is an extrinsic sphere.
Proof of Lemma 5.24. Under the imposed assumptions, every point q of

M satisfies
KII K H H
¾ II ¾ ¾ .
K (q) K (p) det A (p) det A (q)
The result follows at once from corollary 5.18.
Theorem 5.25.
H K
vature. Assume that the condition F ( det , II ) = 0 is fulfilled for a
A K
115
function F : R2 → R : (u, v) 7→ F (u, v) which satisfies the following

requirements:
¨ ¨
Fu > 0 ; Fu ¾ 0 ;
or
Fv ¾ 0 , Fv > 0 .
Then M is an extrinsic sphere.
Proof of Theorem 5.25. Similar as the proof of theorem 5.16.
We finish with a theorem on surfaces in the hyperbolic space,

of which the second fundamental form is a Lorentzian metric.
Theorem 5.26.
The second fundamental form of a minimal surface in the hyperbolic
space (with K < −1) is a locally flat Lorentzian metric.
Proof of Theorem 5.26. This follows at once from the adaption of Cartan’s
formula, which remains valid if the second fundamental form is Lorentz-
ian.
5.3.3. Further Applications.
We now generalise the setting, in the sense that the enveloping space is not
assumed to be a space form any more. The following theorem agrees with
theorem 4.1. of [Aledo–Haesen–Romero 2007].
Theorem 5.27.
Let M be a compact Riemannian surface in a three-dimensional semi-
116
5.4. THE EXTRINSIC GEOMETRY OF THE SECOND. . . 5.4. 117
Riemannian manifold (M , g). Assume that the second fundamental

form of M is positive-definite and that K vanishes nowhere. Under
these conditions,
Z n 
o
II(L, L) − II trII L, trII L − trace ` dΩII ¾ 0 (if K > 0) 


······ ¶ 0 (if K < 0) 


if and only if M is totally umbilical.
Proof of Theorem 5.27. Assume first that the integral inequality is satis-
fied for K > 0. It can be seen by means of formula (5.8) on page 102 that
this integral inequality has the following consequence:
Z Z
HK 1
KII dΩII = + II(L, L) − II trII L, trII L − trace `

dΩII
det A 2
Z Z Z Z
H K K
¾ p p dΩII ¾ p dΩII = K dΩ = KII dΩII .
det A det A det A
This is only possible if M is totally umbilical.
p
Similarly, if K < 0, the fact H ¾ det A implies that H K ¶
p
K det A. In this way, it can be concluded from the above integral inequality
that
Z Z Z Z
H K K
KII dΩII ¶ p p dΩII ¶ p dΩII = K dΩ .
det A det A det A
Thus M is totally umbilical.
Conversely, if M is totally umbilical, the result follows from
Z
HK

0= KII − dΩII
K +1
Z
1 n o
= II(L, L) − II(trII L , trII L) − trace ` dΩII .
2
117
5.4. The Extrinsic Geometry of the Second Funda-

mental Form.
This section is devoted to the introduction of the “mean curvature of the

second fundamental form” for hypersurfaces in a semi-Riemannian space,
of which both the first and the second fundamental form are genuine semi-
Riemannian metrics.
(As such, some of the results which are described below are
contained in [Haesen–Verpoort 0000] and [Haesen–Verpoort–Verstraelen
0000].)
The mean curvature of a hypersurface of a semi-Riemannian

manifold describes the instantaneous response of the area functional with
respect to deformations of the hypersurface. Similarly, a measure for the
rate of change of the total area of a hypersurface, as surveyed in the geom-
etry of its second fundamental form, will be introduced. This variational
problem, in which the setting of § 2.4 is generalised, has already been stud-
ied by F. Dillen and W. Sodsiri for surfaces in E31 .
The letter E will designate the set of all hypersurfaces in a se-

mi-Riemannian manifold (M , g) of which the first as well as the second
fundamental form is a semi-Riemannian metrical tensor. Our first objec-
tive is to determine the critical points of the area functional of the second
fundamental form
Z
AreaII : E → R : M 7→ AreaII (M ) = dΩII .
M
Remark. Even for hypersurfaces M of (M , g) with infinite II-area, the ex-

pression δAreaII (M ) is well-defined for a deformation µ of M , by virtue of
requirement (ii) on p. 72.
118
5.4. THE EXTRINSIC GEOMETRY OF THE SECOND. . . 5.4. 119
Theorem 5.28.
Let M be a hypersurface in a semi-Riemannian manifold (M , g) of
which the first as well as the second fundamental form are semi-Rie-
mannian metrical tensors. Let µ be a deformation of M through hy-
persurfaces which all belong to the class E . Assume that the deforma-
tion vector field has normal component f N . The variation of the area
functional associated to the second fundamental form is given by

Z m
1 X
δAreaII (M ) = −α f · m H − g R(Vi , N )Vi , N κi
2 i=1

α
+ ∆II log |det A| − α divII Z  dΩII .
2
p
Proof of Theorem 5.28. The variation of det A along the given deforma-
tion can be calculated by means of theorem 4.1 on p. 72:
p 1
δ det A = p δ (det A)
2 det A
1p
= det A trace (δA ◦ A← ) .
2
This trace can be calculated in the II-orthonormal frame field {V1 , . . . , Vm }.

Moreover, the fact that the trace of the composition of two operators is
independent of their order will be used. There results:
(
m
p 1p X
δ det A = det A f II A← R(N , Vi )N , Vi κi + f trace A
2 i =1
)
m
X
+α II A← Hs f (Vi ) , Vi κi
i =1
(
m
αp X
= det A f g R(Vi , N )Vi , N κi + f m H
2 i =1
119
)
+ α trII Hess f .
If we make use of the fact that δdΩ = − f α m H dΩ, we find

Z p Z p Z p
δ det AdΩ = δ det A dΩ + det A (δdΩ)
 
m
α
Z Z
X 1
= f g R(Vi , N )Vi , N κi − m H dΩII +
 trII Hess f dΩII .
2 i =1
2
An application of lemmata 5.7 and 5.4, together with the divergence theo-
rem, concludes the proof.
Definition 5.29.
Let M be an m-dimensional hypersurface in a semi-Riemannian man-
ifold (M , g), of which both the first and the second fundamental form
are semi-Riemannian metrical tensors. The mean curvature of the
second fundamental form HII is defined by
 
m
1 X α
HII = m H − g(R(Vi , N )Vi , N ) κi + ∆II log |det A| − α divII Z  .
2 i=1
2
(5.15)
If HII = 0, the hypersurface will be called II-minimal.
In consequence of theorem 5.28 and definition 5.29, we obtain

the following formulae for the variation of the classical area and of the area
of the second fundamental form:
∂
 Z
Area(µ t (M )) = −mα f H dΩ ;
 ∂ t t=0



 ∂

 Z
AreaII (µ t (M )) = −α f HII dΩII .

∂ t t=0
Remark. The expression for HII can be rewritten in an alternative way at

a point p ∈ M where the frame fields can be chosen such that
120
5.5. SOME CHARACTERISATIONS OF EXTRINSIC HYPERSPHERES. . . 5.5. 121
¦ ©
• the I-orthonormal basis e1 (p), . . . , em (p) of T p M is composed of
eigenvectors of the shape operator (principal directions) at p:
A(ei (p)) = λi (p) · ei (p) (i = 1, . . . m) ;
¦ ©
• the II-orthonormal basis V1 (p), . . . , Vm (p) of T p M consists of the
rescaled principal directions at p:
1
Vi (p) = p ei (p) (i = 1, . . . m) .
|λi (p)|
The following expression for the mean curvature of the second fundamen-
tal form holds at the point p:
(
m
m 1X 1
(p) = K(ei , N )

HII H− (5.16)
2 2 i=1
λi
α α
«
+ ∆II log |det A| − divII Z .
4 2 (p)
Remark. With help of the contracted Gauss equation (4.2), yet another
expression for the mean curvature of the second fundamental form can be
derived:
α
¨
HII = − trII Ric − trII Ric + α(m2 − 2m)H (5.17)
2
«
1
− ∆II log |det A| + divII Z .
2
Example. The standard embedding of Sm ( p12 ) in Sm+1 (1) is II-minimal.

Furthermore, the standard embedding of Sk ( p12 ) × Sm−k ( p12 ) in Sm+1 (1)
(see, e.g., [Lawson 1969]) is a II-minimal hypersurface (k = 1, . . . , m − 1).
These assertions can be proved with ease when one takes the fact that these
hypersurfaces are parallel (in the sense that ∇II = 0) into account. z
121
5.5. Some Characterisations of Extrinsic Hyper-

spheres, in Which the Curvatures of the Second
Fundamental Form are Involved.
Below, the mean curvature of the second fundamental form will be inves-
tigated for hypersurfaces in space forms, in an Einstein space, in a three-
dimensional manifold, and finally for curves in a surface. It will be shown
that only extrinsic hyperspheres can satisfy certain inequalities, in which
the mean curvature of the second fundamental form is involved. (Most of
the described results are contained in [Haesen–Verpoort 0000] or [Haesen–
Verpoort–Verstraelen 0000].)
5.5.1. Hypersurfaces in a Space Form.

m+1
We shall denote M 0 (C) for the following Riemannian manifolds of di-
mension m + 1:
the Euclidean hypersphere Sm+1 ( p1 ) (for C > 0) ;


 C
the Euclidean space Em+1 (for C = 0) ;
Hm+1 ( p1 ) (for C < 0) .

the hyperbolic space
−C
m+1
We shall denote M 1 (C) for the following Lorentzian manifolds of dimen-
sion m + 1:
S1m+1 ( p1 ) (for C > 0) ;


the de Sitter space
 C
the Minkowski space E1m+1 (for C = 0) ;
m+1 p1
the anti-de Sitter space H1 ( ) (for C < 0) .

−C
Any of the above semi-Riemannian manifolds has constant sectional curva-

ture C. For C = ±1, abbreviations such as Sm+1 instead of Sm+1 (1) will be
adopted.
Lemma 5.30.
Let M be a compact semi-Riemannian hypersurface in a semi-Rie-
122
mannian manifold (M , g) of constant sectional curvature C and di-

mension m + 1 (with m ¾ 2). Assume that the second fundamental
form of M is positive-definite. The inequality
 
SII ¶ 2α(m − 1) HII + CtrA←  (5.18)
is satisfied if and only if the Levi-Civita connections of the first and

the second fundamental form coincide.
Proof of Lemma 5.30. The following expressions are valid for the curva-
tures which are involved in the above inequality:
  
1 α ∆II det A
 HII = α trace A − C trace A←  +
2 4 det A





α II(gradII det A, gradII det A)



− ;


4 (det A)2




  

SII = α(m − 1) α trace A + C trace A←  + II(L, L)










 1 II(gradII det A, gradII det A)
− .
4 (det A)2
The above inequality (5.18) is equivalent with
(m − 1) ∆II det A (2m − 3) II(gradII det A, gradII det A)

0¶ − − II(L, L)
2 det A 4 (det A)2
and this implies
det A = constant and ∇ = ∇II .
Conversely, if ∇ = ∇II , it follows that ∇II vanishes. Consequently, det A is a

constant and the inequality is satisfied.
123
Theorem 5.31.
Let M be a compact Riemannian hypersurface in the space form
m+1
M e (C), with m ¾ 2. Assume that the second fundamental form
of M is positive-definite. The inequality
 
SII ¶ 2α(m − 1) HII + CtrA←  (5.19)
is satisfied if and only if M is an extrinsic hypersphere.
Proof of Theorem 5.31. Three cases will be treated separately.

m+1
1. M e (C) is a Riemannian space form. It has already been shown
that the inequality (5.19) implies that M is parallel, in the sense
that ∇II vanishes. Such hypersurfaces were classified in theorem 4
of [Lawson 1969]. If C ¾ 0, the only hypersurfaces which appear in
this classification, of which the second fundamental form is positive-
definite, are the extrinsic hyperspheres. If C < 0, the extrinsic hyper-
spheres are the only compact hypersurfaces in the classification.
m+1
2. M e (C) is a Lorentzian space form with C ¶ 0. It follows from the
Gauss equation that (M , g) has strictly negative sectional curvature.
The result follows from lemmata 5.8 and 5.30.
m+1
3. M e (C) is the de Sitter space. It follows from (5.19) that ∇A van-
ishes. Consequently, M has constant mean curvature and an applica-
tion of theorem 4 of [Montiel 1988] concludes the proof.
5.5.2. Hypersurfaces in an Einstein Space.
Theorem 5.32.
Let (M , g) be a Riemannian Einstein manifold of dimension m + 1
124
(with m ¾ 3) with strictly positive scalar curvature S. A compact

hypersurface M ⊆ M with positive-definite second fundamental form
satisfies r
m−2 1

HII + m S ¾ trII Ric (5.20)
m+1 2
q
S
if and only if it is an extrinsic sphere with A = (m−2)(m+1) id. In this
case, q
its mean curvature of the second fundamental form is given by
S
HII = (m−2)(m+1) .
S
Proof of Theorem 5.32. Since Ric = m+1
g, we deduce that trII Ric =
S
m+1
trA← . Define β and ρ by
r È
m−2 S

β= S and ρ= .
m+1 (m − 2)(m + 1)
Furthermore, the principal curvatures will be denoted by λi (i = 1, . . . , m).

It follows now from (5.17) and the assumption (5.20) that
Z  m 
ρ λi
Z
X
trII Ric dΩII = 2HII + β +  dΩII
i=1
λ i ρ
Z   Z
¾ 2 HII + mβ dΩII ¾ trII Ric dΩII .
 
This is only possible if all principal curvatures are equal to ρ. That HII = ρ
follows immediately from equation (5.15).
The converse is straightforward.
125
5.5.3. Surfaces in a three-dimensional semi-Riemannian Mani-

fold.
In case (M , g) is a three-dimensional semi-Riemannian manifold, equa-

tion (5.17) can be rewritten:
HK α α α
HII = α − trII Ric + ∆II log |det A| − divII Z . (5.21)
det A 2 4 2
If (M , g) has moreover constant sectional curvature C, the following ex-

pression for HII is valid:
α

K −2C
HII = H + ∆II log K − C . (5.22)
K−C 4
The Gaussian curvature of (M , g) will be denoted by K, whereas

K stands for the sectional curvature of M in (M , g). As a consequence of
the Gauss equation, we have det A = α(K − K).
Theorem 5.33.
Let M be a compact surface in a three-dimensional semi-Riemannian
manifold, such that the Gaussian curvature K of M is strictly positive.
Assume both the first and the second fundamental form are positive-
definite metrical tensors. If HII is constant, then
Z
1
α HII AreaII (M ) + trII Ric dΩII ¾ 4π ,
2
and equality occurs if and only if M is totally umbilical.
Proof of Theorem 5.33. This result is easily deduced once the expression
(5.21) is integrated, with the following result:
Z
HK 1

α HII AreaII (M ) = − trII Ric dΩII .
det A 2
126
Theorem 5.34.
Let M be a compact surface in a three-dimensional semi-Riemann-
ian manifold (M , g) and suppose that the first as well as the second
fundamental form of M is positive-definite. Suppose that the Gaussian
curvature K of M is nowhere zero. M is totally umbilical if and only
if
1 1

KII ¾ α HII + trII Ric − ∆II log det A (if K > 0); 
2 4
(5.23)
KII ¶ ······ (if K < 0).

Proof of Theorem 5.34. By means of equations (5.8) and (5.21), the

inequality (5.23) is equivalent with a condition on the sign of
II(L, L) − II trII L, trII L − trace ` ,

and the result follows from theorem 5.27.
Theorem 5.35.
Let M be a compact surface in a three-dimensional semi-Riemannian
manifold (M , g) and suppose that the first as well as the second funda-
mental form of M is positive-definite. Assume the Gaussian curvature
K of M is strictly positive. M is an extrinsic sphere if and only if
1
KII ¾ α HII + trII Ric . (5.24)
2
Proof of Theorem 5.35. Assume first (5.24) is satisfied. Since M is com-

pact, we immediately see
Z
HK 1

0= −α HII + − trII Ric dΩII
det A 2
127
Z
1 K p
= −α HII + KII − trII Ric + H− det A dΩII .
2 det A
Since the condition (5.24) is satisfied, we deduce

Z
K p
0¾ H− det A dΩII .
det A
p
This is only possible if there holds H = det A on the entire of M , which
means that M is totally umbilical. Moreover, equality has to occur in (5.24).
On the other hand, the previous theorem gives us that equality (5.23) has
to be satisfied as well, which means that det A is constant. Hence M is an
extrinsic sphere.
The converse follows from theorem 5.34.
Theorem 5.36.
3
Let M be a compact Riemannian surface in the space form M 0 (C)
(with C ∈ R) or the de Sitter space. Assume that the second funda-
mental form of M is positive-definite and that the Gaussian curvature
of (M , g) is strictly positive. Then either
CH
HII − αKII + 2 (5.25)
K−C
changes sign or M is an extrinsic sphere.
Proof of Theorem 5.36. Assume that the function, given in (5.25), does
not change sign. It will be shown that M is an extrinsic sphere. First we
remark that 2 C H = α2 trII Ric.
K−C
A first possibility would be that
1
KII − trII Ric ¾ α HII ,
2
but it can be immediately concluded from theorem 5.35 that M is an ex-
trinsic sphere.
128
On the other hand, if
1
KII − trII Ric ¶ α HII ,
2
it can be concluded directly from theorem 5.31 that M is an extrinsic

sphere.
5.5.4. Curves in a semi-Riemannian Surface.
An investigation of the quantity HII is of particular interest for curves, since

the length of the second fundamental form of a curve γ
Z
p
LengthII (γ) = |κ| ds
is a modification of the classical bending energy

Z
κ2 ds
which was studied already by D. Bernoulli and L. Euler.

We will denote an arc-length parameter by s, a II-arc-length pa-
rameter by sII , and κ for the geodesic curvature of an arcwise parametrised
time-like or space-like curve
γ : ]a, b[ → (M , g) : s 7→ γ(s)
in a semi-Riemannian surface. The unit tangent vector γ0 along γ will be

denoted alternatively by T . It will be supposed that g(∇ T T, ∇ T T ) vanishes
nowhere. By virtue of this property, γ is sometimes called a Frenet curve.
On the other hand, this requirement means precisely that II is a semi-Rie-
mannian metrical tensor on γ. Let {T, N } be the Frenet frame field along
γ:
1
T = γ0 , N = Æ ∇T T .
g(∇ T, ∇ T )
T T
129
Further, we set β = g(T, T ) = ±1 and α = g(N , N ) = ±1. The geodesic

curvature κ of γ in (M , g) is determined by the Frenet-Serret formula:
    
∇
 T =T 0 βκ  T  .
∇T N −ακ 0 N
This geodesic curvature κ is equal to the mean curvature of γ ⊆ (M , g). The

functional LengthII , which was formerly denoted by AreaII but for curves
measures their lengths with respect to the second fundamental form, can
be computed as the integral
Z
p
LengthII (γ) = |κ| ds .
γ
Let K denote the Gaussian curvature of (M , g). A calculation shows

 00 
1 −αK αβ κ (κ0 )2
HII =  +κ+ 2 2 −3 3 , (5.26)
2 κ 4 κ κ
where ϕ 0 is defined by ϕ 0 = T ϕ for ϕ ∈ F(γ).

A curve γ (with κ > 0) in E2 is II-minimal if and only if the

curvature κ, which is seen as a function of the arc-length, satisfies
−4κ4 − 2κκ00 + 3(κ0 )2 = 0 .
Moreover, the formula

A
κ(s) = A ∈ ]0, +∞[ , Q ∈ R
A2 (s + Q)2 +1
describes the general solution of this differential equation. Such a curve

has been depicted in figure 5.3. ItR follows that all inextendible II-minimal
curves in E2 have total curvature γ κ ds = π. z
Remark. It can be asked as well, whether a curve in E2 can be found

which minimises LengthII along all curves with κ > 0 joining two given
points. This requirement is stronger than merely II-minimality of γ, and a
130
1
Figure 5.3: A II-minimal curve in E2 . Its curvature function is κ(s) = s2 +1
.
simple argument shows that no such minimum exists: if γR is an arc of a

circle of radius R which joins the two given points, there holds
lim LengthII (γR ) = 0 .

R→∞
For curves on the unit sphere, the equation HII = 0 can be

rewritten as
4κ2 − 4κ4 − 2κ κ00 + 3(κ0 )2 = 0 .
This is equation (4) of [Arroyo–Garay–Mencía 2003], if the length func-

tional of the second fundamental form is interpreted as so-called curva-
ture energy functional. As is proved and beautifully illustrated in [Arroyo–
Garay–Mencía 2003], there exists a discrete family of closed, immersed, II-
minimal curves on the unit sphere. An embedded “II-minimal” curve which
belongs to this family is S 1 ( p12 ) ⊆ S 2 (1). This curve is, as is remarked in
[Arroyo–Garay–Mencía 2003], actually a local maximum of LengthII .
Theorem 2.19 can be adapted for plane curves.
131
Theorem 5.37.
A simple closed plane curve with strictly positive curvature is a circle
p
if and only if HII = C κ for some constant C.
Proof of theorem 5.37. This follows immediately from the reasoning be-
low, in which some straightforward modifications of formulae from the
second chapter to plane curves have been applied:
Z
π Length ¶ HII ds Length
Z Z
p
=C κ ds Length = C dsII Length
Z
= C LengthII Length = C LengthII κ 〈−N , P〉 ds
Z Z
p
= LengthII C κ 〈−N , P〉 dsII = LengthII HII 〈−N , P〉 dsII
Z 2
1 2 1 p
= LengthII = κ ds
2 2
Z Z
1
¶ κ ds ds
2
= π Length .
Theorem 5.38 (Variational Rewording of Theorem 5.37).

Among the simple closed plane curve γ with strictly positive curva-
ture, the circles are the only critical points of the length functional
with respect to deformations under which the length, as measured by
means of the second fundamental form, is preserved.
132
5.6. CALCULATION OF THE CURVATURES FOR SOME SPECIAL. . . 5.6. 133
5.6. Calculation of the Curvatures for Some Special

Classes of Hypersurfaces
In this section the reader will be provided with some examples of hyper-
surfaces, for which the various curvatures have been explicitly calculated.
In § 5.6.1, we will turn our attention to surfaces of revolu-
3
tion in E , for which the second fundamental form is Lorentzian. This
is an obvious generalisation of the formulae which have been presented on
page 52 ff.
In the following § 5.6.2, we will give some expressions for the
curvatures of the second fundamental form which are valid for a certain
class of surfaces in the three-dimensional Minkowski space.
Finally, in § 5.6.3 it will be shown that all slices of a certain
generalised Robertson–Walker space are II-minimal.
5.6.1. Calculation of the Curvatures for Some Surfaces of Rev-

olution.
Consider the surface of revolution constructed from the graph of a strictly
positive function f as on page 52 where now f 00 > 0. In this case, the
second fundamental form is a Lorentzian metric on the surface of revolu-
tion. We choose the unit normal vector field which points towards the axis
of revolution. The expressions for the curvatures which were given there
remain valid.
Example. We calculate the curvatures of the catenoid, which is obtained

from f (x) = cosh x.
−1
K= ; KII = 0 ;
(cosh x)4
1
H = 0; HII = .
(cosh x)2
p
In particular, we notice that HII is proportional to |K|. Thus to the
catenoid, which was characterised already variationally as the only mini-
mal surface of revolution, can be attributed the additional property of being
133
a critical point of the AreaII –functional with respect to compactly supported

deformations under which the total mean curvature is preserved. z
Example. The curvatures of the one-sheeted

p hyperboloid, which is con-
structed from the function f (x) = 1 + x , are given by
2

−1 2x 2 + 3 x 2
K= 2 ; KII = 3 ;
1 + 2x 2 2x 2 + 1 2
x2 4x 4 + 3x 2 + 2
H= 3 ; HII = 3 . z
2x 2 + 1 2 2x 2 + 1 2
5.6.2. Calculation of the Curvatures for Some Surfaces in the

Minkowski Space.
We will study the curvature of surfaces in the three-dimensional Minkowski

space E31 which are invariant under a one-parameter group of isometries
which preserve an isotropic vector.
Let {u0 , u1 , u2 } be an orthonormal basis of E31 , with respect to
which the metric g is given by
 
−1 0 0
g =
 

 0 1 0.

0 0 1
We will study certain surfaces which are invariant under the one-parameter
group of isometries
 
s2 2
 1+ 2
− s2 s 
 
Ω(s) =  s2 2
1 − s2
 
 s 

 2
 

s −s 1
for which Ω(s) u0 + u1 = u0 + u1 . In order to simplify the calculations,

134
5.6. CALCULATION OF THE CURVATURES FOR SOME SPECIAL. . . 5.6. 135
the following basis of E31 is introduced:


` = p12 u0 + p1 u1 ;
 0
 2
 `1 = p12 u0 − p1
2
u1 ;

`2 = u2 .
We will denote the co-ordinates of a vector with respect to {u0 , u1 , u2 }

as (t, x, y), whereas the co-ordinates with respect to {`0 , `1 , `2 } will be
denoted by ((v, w, y)) . A similar notation will be adopted for tensors. For
example, we have

 
 
 p 

0 −1 0 1 s2 2s
g = and Ω(s) = 

 
 
 



 −1 0 0 

 

 0 1 0 .


p

 
 
 


0 0 1 0 2s 1
Let f : ] 0 , +∞ [ → R be a function which satisfies f 0 < 0. The surface

M ⊆ E31 under consideration is parametrised by
µ : ] 0 , + ∞[ × R → M ⊆ E31 : (a, s) 7→ Ω(s) f (a), a, 0

.
The surface can be described implicitly as the set of points in the Minkowski
space which satisfy
y2
v = f (w) + (with w > 0) .
2w
This surface will be said to be a surface of revolution with isotropic axis

in the Minkowski space. Since f 0 < 0, the surface M is a Riemannian
surface. We choose the following unit normal vector field:
1 p
Nµ(a,s) = p s2 − f 0 , 1, 2 s .
−2 a f 0
We signal that here as well as in the sequel we write merely f instead of

f (a), and this convention applies to the derivatives of f as well. It will
be assumed that either f 00 > 0 or f 00 < 0 which means that the second
fundamental form is either a Riemannian or a Lorentzian metrical tensor.
135
In both cases, the following formulae hold true:
− f 00
K = ;
4a f 02
f 00 1
H = p 3
+ p p ;
4 2(− f 0 ) 2 2 2a − f 0
n o
2 f 02 f 00 + 2a f 02 f 000 + a f 0 f 002 + a2 f 0 f 00 f 000 − 2a2 f 003
KII = p 3
;
4 2 a2 (− f 0 ) 2 f 002
1

HII = p 3
− 2a f 02 f 00 f 000 − f 02 f 002 + 3a2 f 02 f 0002
4 2 a2 (− f 0 ) 2 f 003

2 0 002 000 2 02 00 0000 2 004
+ 2a f f f − 2a f f f − 3a f .
(Here the left-hand side has to be evaluated at a point µ(a, s) of the surface,
while the value a has to be substituted in the derivatives of the functions f
which appear in the right-hand side.)
1
Example. If we specify f by f (a) = 2a , the above surface is an open part
¦ ©
of the hyperbolic plane H = ξ ∈ E1 | g(ξ, ξ) = −1 and we recover the
2 3
expressions K = −1, H = 1, KII = −1, HII = 1. z
Example. We calculate the curvatures of the surface obtained from the

function f (a) = a12 . There results
p
−3a −5 a
K= ; KII = ;
8 8
p p
5 a a
H= ; HII = . z
8 2
Example. For f (a) = −a3 we obtain a minimal surface, which shares with
136
5.7. GEODESIC HYPERSPHERES IN A RIEMANNIAN MANIFOLD 5.7. 137
the catenoid the property of being a critical point for the II-area under total
mean curvature constraint:
1
K= 4; KII = 0 ;
6a
1
H = 0; HII = p . z
6a2
Remark. It can be seen that the surfaces originating from the function
f (a) = W a t (where the constants W and t satisfy t 6= 1 and W t < 0)
satisfy −H = KII = t−3
4
HII .
5.6.3. The Slices of a Generalised Robertson–Walker Space.

We will study the slices of a generalised Robertson–Walker space M =
J ×h F . Here J is an open interval of R equipped with the metrical tensor
αg R and (F, g F ) is a connected two-dimensional Riemannian manifold of
constant Gaussian curvature. The warping function h is a positive smooth
function on J. Explicitly, the metric of M can be written as
g = π∗ (αg R ) + (h ◦ π)2 σ∗ (g F ) ,
where π : M → J and σ : M → F are the standard projections. The
standard tangent vector field ∂ t on J ⊆ R will be lifted to a unit normal
vector field of the slices F t = π−1 ({t}). The second fundamental form of
such a slice is given by
h0 (t)
II(F t ) = −α I(F t ) .
h(t)
2
h0 (t)

Consequently, det(A(F t )) = h(t)
and a short calculation shows
00
h (t) h0 (t)

HII (F t ) = −α + .
h0 (t) h(t)
This shows that every slice F t in the generalized Robertson–Walker space
J ×h F with warping function
(P ∈ R+ , Q ∈ R)
p
h(t) = αP t + Q
is a II-minimal surface. In particular has every slice F t the same II-volume.
137
5.7. Geodesic Hyperspheres in a Riemannian Mani-

fold.
As a final example we shall investigate the (sufficiently small) geodesic
hyperspheres in a Riemannian manifold, since these provide us with a nat-
urally defined class of hypersurfaces with a positive-definite second funda-
mental form. Herefore, we will use the method of power series expansions
which was applied extensively by A. Gray, and also by B.-Y. Chen and L.
Vanhecke [Gray 1973, Gray–Vanhecke 1979, Chen–Vanhecke 1981] (see
also [Gray 2004]).
(The main outcome of this section of the thesis is described in
the last part of [Haesen–Verpoort 0000].)
It will first be shown that a Riemannian space, of which the
value of HII agrees for every geodesic hypersphere in any of its points with
the corresponding value for a hypersphere in a Euclidean space, has to be
locally flat.
The geodesic hypersphere of radius r and centre n will be de-
noted by Gn (r). It was asked in [Gray–Vanhecke 1979] whether the Rie-
mannian geometry of the ambient manifold (M , g) is fully determined by
the area functions
M × ]0, +∞[ → R : (n, r) 7→ Area(Gn (r)) (r sufficiently small)
of the geodesic hyperspheres. It appears that a decisive answer has not

been given yet. We ask similarly whether a Riemannian manifold of which
every geodesic hypersphere has the same II-area as a Euclidean hypersphere
of the same radius is locally flat. In analogy with [Gray–Vanhecke 1979],
we were only able to give an affirmative answer if additional hypotheses
are made. For example, the question is answered in the affirmative if the
dimension of the ambient manifold does not exceed five.
5.7.1. On the Technique of Power Series Expansions for Global

Invariants of Geodesic Hyperspheres.
Below, a brief and personal sketch of some aspects of the technique of
power series expansions, as applied to the determination of global geomet-
138
ric invariants of geodesic hyperspheres, will be given without presenting

the enormous technical difficulties. It should be stressed that our descrip-
tion of the method is in this respect not entirely faithful. As such, the
reader whishing a more systematic and profound account of the method is
referred to the original works.
The previous notational conventions will be prolonged in the
sense that the ambient manifold will be denoted by (M , g), and is assumed
to have dimension m + 1. However, our approach differs from before in
two aspects: primo, the introduction of a co-ordinate system seemed de-
sirable in order to explicitly evaluate certain functions and their integrals;
secundo, tensor index notation with respect to this co-ordinate system will
be adopted.
The application of the technique which we envisage is an in-
vestigation of the geodesic hyperspheres of M which are centred around a
selected fixed point n ∈ M , and in particular of their areas, as measured by
means of their second fundamental form. In order to calculate this area as
an integral over Gn (r), the integrand will be expanded as a power series
in r. Herefore the following theorem is needed, which gives the expected
relation between a given tensor W on a M -neighbourhood of the point n
and the value W(n) which this tensor takes at the point n, together with
k
all the higher-order tensorial derivatives ∇ W . By means of parallel
(n)
translation with respect to the manifold (M , g), a tensor Z(n) on Tn M can

be extended to a tensor field Z(n) y along a geodesic ray γ emanating

from the point n.
Theorem 5.39 (Taylor’s Theorem for Tensor Fields).

Let W be a tensor which is defined on a neighbourhood of the point
n of M . For any arcwise parametrised geodesic segment γ : ]−`, `[ →
M , where γ0 (0) = ξ ∈ Tn M , there holds
h−1 k
X r k

Wγ(r) = ∇ξ···ξ W y + O (r h) .

k=0
k! (γ(r))
139
Proof of Theorem 5.39. Let u = e0 , e1 , · · · , em be an orthonormal basis

of Tn M , and denote by {E0 , · · · , Em } the vector fields which are obtained by

parallel translation of these vectors along the geodesic segment. The above
theorem reduces to the usual theorem of Taylor, applied to the component
functions of W with respect to this frame field.
We now choose an orthonormal basis {e0 , . . . , em } of Tn M and

consider the associated normal co-ordinate system x = (x 0 , . . . , x m ) of
(M , g) at n:
!!
m
X
x exp j
t ej = (t 0 , . . . , t m ) . (5.27)
s=0
Tensor indices which refer to this co-ordinate system will be denoted with a
bar. The coefficients of the Riemann–Christoffel curvature tensor of (M , g)
are determined by
Rı u v e = g(R(∂ ı , ∂ u )∂ v , ∂ e ) (ı, u, v, e = 0, . . . , m) .
The letter r will designate also the distance function with respect to the
point n which has been chosen as origin of the normal co-ordinate system.
It should be noted that the co-ordinate vector fields are not

necessarily parallel along geodesic rays passing through the distinguished
origin n, which makes a precise translation of Taylor’s theorem in terms
of the co-ordinate coefficients of a tensor by no means an easy task. The
answer has been provided in [Gray 1973].
140
Theorem 5.40 (Gray).

Let W be a tensor of type (0, r) on (M , g). The normal co-ordinate
coefficients of W satisfy
m
X
Wα1 ··· αr = Wα1 ··· αr (n) + ∇ı Wα1 ... αr (n) x ı
ı= 0
!
m m Xr
1 X 2 1 X
+ ∇ı Wα1 ... αr − Rı αa  s Wα1 ··· αa−1 s αa+1 ··· αr xıx 
2 ı =0 3 s=0 a=1 (n)
m m X
r
1 X 3
X
+ ∇ı  p Wα1 ... αr − Rı αa  s ∇ p Wα1 ··· αa−1 s αa+1 ···αr
6 ı  p=0 s=0 a=1
!
m X r
1X
xı x xp

− ∇ı R  αa p s Wα1 ··· αa−1 s αa+1 ··· αr
2 s=0 a=1
(n)
m
1 X 4
+ ∇ı  p q Wα1 ··· αr
24 ı  p q=0
m X
X r 2
−2 Rı αa  s ∇ p q Wα1 ··· αa−1 s αa+1 ··· αr
s=0 a=1
m X
X r
−2 ∇ı R  αa p s ∇q Wα1 ··· αa−1 s αa+1 ··· αr
s=0 a=1
m X r
3 X 2

− ∇ı R p αa q s Wα1 ··· αa−1 s αa+1 ··· αr
5 s=0 a=1
m X r
1 X
+ Rı αa  s R p s q u Wα1 ··· αa−1 u αa+1 ··· αr
5 s u=0 a=1
m X
r
2 X
+ Rı αa  s R p αc q u Wα1 ... αa−1 s αa+1 ... αc−1 u αc+1 ··· αr
3 s u=0 a c=1
a<c
!
x ı x  x p x q + O (r 5 ) .
(n)
141
It should be remarked that the above theorem cannot be di-

rectly applied to tensors which are ill-defined at the co-ordinate centre.
An example of such a tensor, which is of primary geometric interest in the
study of a family of concentric geodesic hyperspheres, is the shape operator
of this family. In [Chen–Vanhecke 1981] a power series expansion has been
given for the shape operator of these geodesic hyperspheres with respect to
an orthonormal frame field. Their and similar methods either make use of
the Jacobi differential equation
2
∇γ0 γ0 X + R(γ0 , X ) γ0 = 0 ,
which is a necessary condition for the existence of a deformation through

geodesics with deformation vector field X of the geodesic γ, or, which is
well-nigh equivalent, of the formula for the unit normal deformation of the
shape operator of the geodesic hyperspheres (theorem 4.1 on page 72)
δN A = A2 + R(N , ·)N .
In the following pages, a power series expansion for the co-

ordinate coefficients of this shape operator will be given.
The next lemma, which is needed in order to integrate over the
geodesic hyperspheres, is similar to lemma 3.2 of [Gray–Vanhecke 1979].
By Sm we will denote the unit hypersphere of Em+1 which is centred at the
origin; the area of this unit hypersphere will be denoted by αm .
Lemma 5.41.
Let h : Em+1 → Tn M be a homothety with factor r, which sends the
origin of Em+1 to the origin of Tn M . Let the area elements of (M , g),
of the geodesic hypersphere Gn (r), and of the Euclidean hypersphere
Sm be written as dΩ = ω dx 0 · · · dx m , dΩ, and dΩSm , respectively. Let
f : Gn (r) → R be any function whatsoever. There holds:
Z Z
f dΩ = r m
f ◦ exp ◦ h ω ◦ exp ◦ h dΩSm .

Gn (r) Sm
142
The last two results reduce the determination of AreaII (Gn (r))
to the problem of integrating a power series, in which quantities of a tenso-
rial character at the point n occur, over the unit hypersphere of Tn M . The
formulae, as given in [Gray 1973, Gray–Vanhecke 1979, Chen–Vanhecke
1981], which express how such an integration of tensors on the unit hy-
persphere of the tangent space can be performed, are of a purely algebraic
character in the sense that they can be presented for linear tensors on the
(m + 1)-dimensional Euclidean vector space Em+1 endowed with the stan-
dard scalar product.
Lemma 5.42.
Let Q be a linear tensor of type (0, k) on the Euclidean vector space
Em+1 . The integral of the function Sm → R : ξ 7→ Q(ξ , · · · , ξ) is given
by the following formula:

 0 (k odd) ;





αm


(k = 2) ;

tr Q
Z 
(m + 1)

Q(ξ , · · · , ξ) dΩ =
Sm



  
 tr12 tr34 Q
αm


 + tr tr Q 
 (k = 4) .

  
 (m + 1)(m + 3)
  13 24 
+ tr14 tr23 Q
(The indices in the last expression refer to the slots on which the trace
should be taken.)
Historical Remark. The above formula most naturally finds its application
within classical differential geometry, if the second fundamental form of a
surface M ⊆ E3 is substituted for the tensor Q (which now becomes a tensor
on a tangent plane of the surface). Let us recall that the normal curvature
assigns to a unit vector ξ ∈ S1 ⊆ Tn M , tangent to the surface at a point
n, the curvature of the normal section determined by ξ and the normal Nn
to the surface. By an application of the above formula, the important fact
143
that the mean value of this normal curvature function ρ is exactly the mean
curvature H, is easily recovered.
This fact also immediately follows from Euler’s well-known
formula, although the concept of mean curvature was only introduced by
Poisson and Germain after Euler’s death.
Further, it seems that Dupin was the first to notice that the sum
of the normal curvatures of two orthogonal normal sections is independent
of the choice of these. This theorem has then been extended by Babinet to
k normal sections which are separated by equal angles:
1
ρ(θ1 ) + · · · + ρ(θk ) = H

k
By letting k → ∞ we obtain the aforementioned fact. See [Reich 1973, p.
287].
5.7.2. The Mean Curvature of the Second Fundamental Form of

the Geodesic Hyperspheres.
We will now develop the function HII (seen for a family of concentric hy-
perspheres centred at n) as a power series in the variable r by means of the
techniques of [Gray 1973, Gray–Vanhecke 1979, Chen–Vanhecke 1981].
The normal co-ordinate system x = (x 0 , . . . , x m ) on an (M , g)–
neighbourhood of n has already been introduced in equation (5.27). In
order to calculate quantities which are associated to the geodesic hyper-
spheres Gn (r) (for r > 0 a variable small value), a co-ordinate system on a
region of these hyperspheres Gn (r) will be introduced.
For any fixed r, a co-ordinate system on Gn (r) is given by x =
(x , . . . , x ) in a Gn (r)-neighbourhood of the point exp(r e0 ). Let further γ
1 m
be the geodesic satisfying γ(0) = n and γ0 (0) = e0 , such that the relation
x(γ(r)) = (r, 0, . . . , 0) holds true.
Our first purpose is to determine the first few terms in the
power series expansion (in the variable r > 0) for the value HII (γ(r)) which
the mean curvature of the second fundamental form of the geodesic hyper-
sphere Gn (r) assumes at the point γ(r). It will be supposed throughout that
r > 0 is sufficiently small, in order that everything below is well-defined.
144
Figure 5.4: A simplified drawing for the co-ordinate systems x and x. The co-
ordinate grid on (M , g) of x is displayed in gray.
It should be noticed that the co-ordinate vector fields ∂ j of M

and ∂ j of Gn (r) are related by ( j = 1, . . . , m)
xj
∂j = ∂ j − ∂0,
x0
and in particular, there holds ∂ j = ∂ j along γ. (See also figure 5.4.) Over-
lined tensor indices will refer to the co-ordinate system x, whereas ordi-
nary tensor indices refer to the co-ordinate system x of the geodesic hyper-
145
spheres with centre n.

In [Chen–Vanhecke 1981], the expansion for the mean curva-
ture H of the geodesic hyperspheres was given at the point γ(r):
1 r r2
H (γ(r)) = − Ric0 0 (n) − ∇0 Ric0 0 (n)
r 3m 4m !
3 m
r 1 2 1 X 2
+ − ∇0 0 Ric0 0 − R0 a 0 e + O (r 4 ) .
m 10 45 a e=0
(n)
Thus the following theorem is an immediate consequence of the calcula-

tions of [Chen–Vanhecke 1981].
Theorem 5.43.
The locally flat spaces are the only Riemannian manifolds of which all
geodesic hyperspheres have a constant mean curvature which is equal
to the inverse of their radius.
The formula of theorem 5.40 simplifies considerably when ap-

plied to the metric tensor g. There results:
m m
1 X 1 X
g ı  = δı  − a c
Ra ı c  (n) x x − ∇a Rc ı e  (n) x a x c x e
3 a c=0 6 a c e=0
(5.28)
m m
1 X 2 16 X
+ − 6∇a c Re ı u  + Ra ı c s Re  u s x a x c x e x u + O (r 5 ) .
120 a c e u=0 3 (n)
s=0
This formula, which is valid for ı,  = 0, . . . , m and holds on the normal

neighbourhood of n, implies the following one:
r2 r3
gı  (γ(r))
= δı  − R0 ı 0  (n) − ∇0 R0 ı 0  (n)
3 6 !
m
r4 2 16 X
+ −6∇0 0 R0 ı 0  + R0 ı 0 s R0  0 s + O (r 5 ) .
120 3 s=0
(n)
It has been mentioned already that formula (3.5) of [Chen–

Vanhecke 1981] gives us the components of the shape operator with respect
146
to an orthonormal frame field. As a consequence of their formula (3.5), we

have the following expression:
−1 −1

2 3
log det A (γ(r)) + m log(r) = r +r

Ric0 0 ∇0 Ric0 0
3 (n) 4 (n)
!
n
−7 X 2 1 2
+r 4 R0 a 0 c − ∇ Ric0 0 + O (r 5 ) . (5.29)
90 a c=0 10 0 0
(n)
This, in turn, establishes the following theorem.
Theorem 5.44.
The locally flat spaces are the only Riemannian manifolds of which
all geodesic hyperspheres have a constant Gauss–Kronecker curvature
which is equal to the inverse of the m-th power of their radius.
In order to find an expression for the co-ordinate coefficients of

the shape operator of Gn (r), we will compute the Christoffel symbols of
(M , g). Partial derivatives will be denoted with a vertical bar | in tensor
components. From (5.28) we see that the following expression holds true
(e, ı,  = 0, . . . , m):
−r
g ı |e (γ(r))
= Re ı 0  + R0 ı e  (n)
3
r2
− ∇e R0 ı 0  + ∇0 Re ı 0  + ∇0 R0 ı e  (n)
6
r3 2 2 2 2
+ − 6∇e 0 R0 ı 0  − 6∇0 e R0 ı 0  − 6∇0 0 Re ı 0  − 6∇0 0 R0 ı e 
120
m m
16 X 16 X
+ R e ı 0 s R0  0 s + R0 ı e s R0  0 s (5.30)
3 s=0 3 s=0
m m

16 X 16 X
+ R0 ı 0 s Re  0 s + R0 ı 0 s R0  e s + O (r 4 ) .
3 s=0 3 s=0 (n)
The inverse components of the metric are given by: (ı,  = 0, . . . , m)

r2
gı 

= δı  + R0 ı 0  (n) (5.31)
(γ(r)) 3
147
r3
+ ∇0 R0 ı 0  (n) + O (r 4 ) .
6
Remark. According to the Gauss lemma, the matrix (g ı ) has the following
structure at the point γ(r):
   
1 0 ··· 0 1 0 ··· 0
   
 0 g1 1 · · · g1 m   0 g1 1 · · · g1 m 

  
 
(g ı )(γ(r)) =  = .
 
 .. .. 

 
 .. .. 

0 . .  0 . . 

  
 

0 g m 1 · · · g m m (γ(r)) 0 g m 1 · · · g m m (γ(r))
Consequently, the same holds for the inverse matrix. This means that (for
ı,  = 1, . . . , m) formula (5.31) gives also the inverse components
g ı  (γ(r)) = g ı  (γ(r))

of the metrical tensor g of Gn (r), at a point on the curve γ.


The Christoffel symbols Γ0 ı of (M , g) with respect to the co-ordinate system
x can be computed by means of equations (5.30) and (5.31) at a point of
the geodesic ray γ.
On the other hand, the inward pointing unit normal vector
field N of Gn (r) is given by
m
−1 X
N= x v∂ v .
r v=0

Since r|ı (γ(r)) = 0 for ı = 1 . . . m, we obtain (for r > 0)
!
m
1 X
A( ∂ ) = A( ∂ ı ) = − ∇∂ (N ) = ∇∂ v
x ∂v

ı (γ(r)) (γ(r)) ı (γ(r)) r ı
v=0

(γ(r))
! !
m m
1 X s 1 X s
= ∂ı+ x v Γv ı ∂s = ∂ı + r Γ0 ı ∂s .
r s v=0
r
(γ(r)) s=0 (γ(r))
s
Consequently, there holds 1r δı s + Γ0 ı = Ası at the point γ(r). In this way,
we obtain the following expression for the shape operator of Gn (r) at γ(r):
(ı,  = 1, . . . , m)
1 r r2
Ası (γ(r)) = δı s − R0 ı 0 s (n) − ∇0 R0 ı 0 s (n) (5.32)
r 3 4
148
!
m
3
−1 2 1 X
+r ∇0 0 R0 ı 0 s − R0 ı 0 w R0 w 0 s + O (r 4 ) .
10 45 w=0
(n)
Finally, we can compute the components of the second fundamental form

in the following way (ı,  = 1, . . . , m):
1 2r 5r 2
IIı  (γ(r)) = g ı  (n) − R0 ı 0  (n) − ∇0 R0 ı 0  (n) (5.33)
r 3 12 !
m
−3 2 2 X
+ r3 ∇0 0 R0 ı 0  + R0 ı 0 s R0 s 0  + O (r 4 ) .
20 15 s=0
(n)
The above equation is only valid at the single point γ(r) = exp(r e0 ) of
Gn (r), and hence needs to be rewritten in order to compute the leading
term of IIı |e at γ(r). A more general expression for IIı , which is valid at
any point p = exp(rξ) with co-ordinates (x 0 , . . . , x m ) (for a unit vector
ξ ∈ Tn M , as in figure 5.4), is obtained by

 xı
∂ı (γ(r)) ∂ = ∂ − ∂

0
ı (p) ı
 (p) x0 (p)

 


 

 
substitution of ∂ ı by ∂

− xı
∂

(n) ı x 0 0
(n) (n)

 


 

 
 Pm
e0 ξ = 1r a=0 x a ea
in the previous formula. The result is
1 xıx  2 Xm 2 X m xı
IIı  = δı  + − R a ı c  (n) x a c
x + R a c
a 0 c  (n) 0 x x
r (x 0 )2 3 a c=0 3 a c=0 x
x xı x  a c
!
m m
2 X 2 X
+ a c
Ra ı c 0 (n) 0 x x − Ra 0 c 0 (n) 0 2 x x + O (r 2 ) ,
3 a c=0 x 3 a c=0 (x )
where the function x 0 can be expressed in the co-ordinate system x on

Gn (r) by p
x 0 = r 2 − (x 1 )2 − . . . − (x m )2 .
Consequently, there holds (ı, , e = 1, . . . , m):
−2
IIı |e (γ(r)) = Re ı 0  + R0 ı e  (n) + O (r) . (5.34)
3
149
In this way, we obtain an expression for the leading term of the Christoffel
symbols of the second fundamental form at γ(r) with respect to the co-
ordinate system x of the geodesic hyperspheres (ı, , s = 1, . . . , m):
2r
ΓII sı  = Rs ı 0  + R0 ı s  (n) + O (r 2 ) . (5.35)
(γ(r)) 3
After some work, it can be concluded from the foregoing equations (5.29),
(5.33) and (5.35) that
−2r
∆II log det A(γ(r)) = S − (m + 1)Ric0 0 (n)
3
3

2
+r −S |0 + (m + 2)∇0 Ric0 0
4 (n)
m

−16 X
+r 3 R0 v 0 w Ric v w
45 v w=0
m
14 X 2
+ (3 + m) R0 v 0 w
45 v w=0
m
7 X 2 3
− Rı v 0 w − Hess(S)0 0
15 ı v w=0 5
m
(6 + 2m) 2 22 X 2
+ ∇0 0 Ric0 0 + Ric0 v
5 45 v=0

4 2 1
− Ric0 0 − ∆ Ric0 0 + O (r 4 ) .
9 5 (n)
We will not give the details of the further calculations which can be ob-
tained in a similar way. The II-divergence of the vector field Z is given
by:

divII Z (γ(r)) = r (m + 1) Ric0 0 − S (n)
3

2
+r (m + 2)∇0 Ric0 0 − S |0
2 (n)
n
(m + 3) 2

−1 X
+r 3 R0 ı 0 Ricı  + ∇0 0 Ric0 0
3 ı =0 2
m 2 (m + 3) X m
2X 2
+ Ric0 v + R0 ı 0 
3 v=1 3 ı =0
150
m

1 X 2
−Hess(S)0 0 − Ra c e 0 + O (r 4 ) .
2 a c e=0 (n)
The II-trace of the Ricci tensor can be calculated as

trII Ric = r S − Ric0 0 (n) + r 2 S |0 − ∇0 Ric0 0 (n)

(γ(r))
 
m
3
1X 1 2 1
+r R0 ı 0 Ricı  − ∇0 0 Ric0 0 + Hess(S)0 0  + O (r 4 ) .
3 ı =0 2 2
(n)
The II-trace of the Ricci tensor satisfies

m(m − 1) (m + 5)

trII Ric (γ(r)) =
+r S− Ric0 0
r 3 (n)
(m + 7)

+r 2 S |0 − ∇0 Ric0 0
4 (n)
m
1 X (m + 9) 2
+r 3 R0 ı 0 Ricı  − ∇0 0 Ric0 0
3 ı =0 10
m
(m + 14) X

2 1
− R0 ı 0  + Hess(S)0 0 + O (r 4 ) .
45 ı =0
2 (n)
Gathering the foregoing results and formula (5.17) together,

an expression for the mean curvature of the second fundamental form of
the geodesic hyperspheres is obtained:
m r
HII (γ(r)) = + S − (m + 3)Ric0 0 (n) (5.36)
2r 3
1 (20 + 5m)

2
+r S |0 − ∇0 Ric0 0
2 16 (n)
m
(15 + 3m) 2

7 X
+r 3 R0 ı 0 Ricı  − ∇0 0 Ric0 0
90 ı =0 20
m 2 (20 + 4m) X m
19 X 2
− Ric0 v − R0 ı 0 
90 v=1 45 ı =0
m
7 2 X 2
+ Hess(S)0 0 + Ra c e 0
20 15 a c e=0
151

1 2 1
+ Ric0 0 − ∆ Ric0 0 + O (r 4 ) .
90 20 (n)
Theorem 5.45.
A Riemannian manifold (of dimension m + 1) is locally flat if and
only if the mean curvature of the second fundamental form of every
m
geodesic hypersphere is equal to the constant 2r (where r stands for
the radius of the geodesic hypersphere).
Proof of Theorem 5.45. Suppose that (M , g) is a Riemannian manifold

m
such that the relation HII = 2r holds for every geodesic hypersphere. Then
for any choice of n ∈ M and e0 ∈ Tn M , the coefficients of the positive
powers of r in formula (5.36) vanish. An analysis of the equation
∀n ∈ M ∀ ξ ∈ Tn M with kξk = 1, (m + 3) Ric(ξ, ξ) = S (n)
gives that M is Ricci flat. The fact that the coefficient of r 3 vanishes, implies
that for each point n ∈ M and for each unit vector ξ ∈ Tn M , there holds
m m
(20 + 4m) X 2 2 X 2
Rξ ı ξ  = Ra c e ξ .
45 ı =0
15 a c e=0
Both sides of the above equation can be integrated over the unit hyper-
sphere of Tn M with help of lemma 5.42. By means of the resulting equa-
tion, it can be concluded that R vanishes.
5.7.3. The Area of the Geodesic Hyperspheres, as Measured by

Means of the Second Fundamental Form.
A calculation which makes use of lemmata 5.41–5.42 and formula (5.29),
gives

m S
AreaII (Gn (r)) = r 2 αm 1 − r 2 (5.37)
3(m + 1) (n)
152
m

4
1 1 1 X 2
+r (S)2 + Ricı 
(m + 1)(m + 3) 18 15 ı =0
m

1 X 2 3
− Ra c e s − ∆S + O (r 5 ) .
15 a c e s=0 20 (n)
Thus, the following adaption of theorem 4.1 of [Gray–Vanhecke 1979] can

be obtained.
Theorem 5.46.
Let (M , g) be a Riemannian manifold of dimension m + 1, and sup-
pose that the area of every geodesic hypersphere of M , as seen in the
m
geometry of its second fundamental form, is equal to r 2 αm (where r
stands for the radius of the geodesic hypersphere). Then there holds
¨
S = 0;
(5.38)
kRk2 = kRick2 .
It can be concluded that M is locally flat if any of the following addi-

tional hypotheses is made:
(i) dim M ¶ 5;
(ii) M is conformally flat and dim M 6= 6;
(iii) the Ricci tensor of M is positive or negative semi-definite (in

particular if M is Einstein);
(iv) M is a Kähler manifold of complex dimension ¶ 5;
(v) M is a Bochner flat Kähler manifold of complex dimension 6= 6;
(vi) M is a product of surfaces (with an arbitrary number of factors).
Proof of Theorem 5.46. The first part of the theorem follows immediately
from the given power series expansion (5.37). Assume (5.38) is satisfied.
(i) Suppose that M has dimension ¶ 5 (i.e., m ¶ 4). The trivial case
153
m = 1 should be excluded in the reasoning below. Let W denote the

Weyl conformal curvature tensor of (M , g). There holds
0 ¶ kWk2
4 2 2
= kRk2 − kRick2 + S
m−1 m(m − 1)
m−5
= kRk2 ¶ 0 ,
m−1
and consequently, 0 = R.
(ii) The case where dim M ¶ 5 has already been proved. So assume
M is a conformally flat Riemannian manifold which satisfies (5.38),
dim M ¾ 7 (i.e., m ¾ 6) and 0 6= kRk. The fact that 0 = kWk2 implies
(m − 1)kRk2 = 4kRick2 = 4kRk2 < (m − 1)kRk2 ,
which is clearly a contradiction.
(iii) If εRic is positive semi-definite, for ε = ±1, then 0 ¶ εtrRic = εS = 0

and consequently Ric = 0 and R = 0.
(iv) and (v) can be proved similarly to the two previous cases by an analy-
sis of the squared norm of the Bochner curvature tensor. (vi) can be proved
in the same way as in [Gray–Vanhecke 1979].
Remark. For a given r > 0 and n ∈ M , the collection concentric geodesic

hyperspheres Gn (r + s) can be seen as a deformation of Gn (r) with de-

formation vector field −N . An application of theorem 5.28 gives that the

relation
∂
Z
AreaII (Gn (r)) = HII dΩII
∂r G (r)
n
holds. It can indeed be checked that the first terms in the power series
expansion of both functions agree.
154
Bibliographical Notes for § 5.7. Although theorem 5.39 is straightforward, the result is
seldom mentioned in a text on differential geometry. Notable exceptions are, e.g., [Her-
mann 1968, lemma 23.5 p. 310 ; Chen–Vanhecke 1981, lemma 4.1].
In [Kowalski–Vanhecke 1986] another expression for the shape operator of
the Gn (r) is given. We also refer to [Gray 2004].
155
156
Chapter 6.
A LITERATURE OVERVIEW ON THE

GEOMETRY OF THE SECOND FUNDAMENTAL
FORM AND THE GEOMETRY OF OVALOIDS.
6.1. Introduction.
This final chapter is intended to provide the reader with an overview on

the literature of the geometry of the second fundamental form and the
geometry of ovaloids.
It seems fair to state that, besides the earlier work of J. Wein-
garten, G. Darboux and E. Cartan in which the connection and the cur-
vature of the second fundamental form appeared, the systematic study of
the second fundamental form as a metrical tensor was stimulated by initial
work of P.J. Erard in the late sixties. Some years later an overview on the
subject already appeared [Glässner–Simon 1973]. In the present chapter,
this is brought together with a selection of the substantial number of results
which have been added in the subsequent thirty-five years.
In the first paragraph, the adopted notation is explained. Some
first results concerning the second fundamental form as a metric have been
included in § 6.3.
In § 6.4 we approach the second fundamental form from the
point of relative differential geometry. The fact that all relative metrics on a
hypersurface are conformal, motivates the study of the conformal geometry
157
158 6. A LITERATURE OVERVIEW. . . 6.
of the second fundamental form (§ 6.5). In § 6.6, we turn our attention to

II-minimal surfaces.
In the subsequent § 6.7, congruence theorems in which the
shape operator, the second or the third fundamental form occur, are given.
These theorems give a sufficient condition on a mapping between two oval-
oids to be a congruence. The infinitesimal model of a one-parameter family
of mappings is an infinitesimal deformation, and similar theorems for in-
finitesimal deformations are listed in § 6.8.
The following § 6.9 describes classical facts about the geome-
try of ovaloids. For example, some basic inequalities between geometrical
quantities of an ovaloid such as its area, its volume, and its area as mea-
sured in the geometry of its second fundamental form are listed. That the
spheres are the only ovaloids achieving equality in some of these inequal-
ities gives us a characterisation of them as the unique minimum of certain
functionals. The weaker condition that the functional is stationary with re-
spect to deformations of the ovaloid precisely means that the corresponding
Euler-Lagrange equation, in which the curvatures of the ovaloid can occur,
is satisfied. Thus one arrives in § 6.10 at characterisations of the spheres
as the only ovaloids satisfying various pointwise conditions on their cur-
vatures, and similar characteriations of totally umbilical hypersurfaces in
curved semi-Riemannian spaces.
The next paragraph covers miscellaneous related topics: con-
vexity (§ 6.11.1), submanifolds of a higher codimension (§ 6.11.2), and
non-smooth convex surfaces (§ 6.11.3).
Finally, some open problems are listed in § 6.12.
Some accessible, general and useful textbooks are [Blaschke
1916, Blaschke 1923, Efimow 1957, Chern 1967, Huck et al 1973, Hopf
1983, Montiel–Ros 2005].
6.2. Notation and Preliminaries.
6.2.1. Surfaces in E3 .
On the very first pages of this dissertation, the definition of the first fun-
damental form, the shape operator, the Gaussian and the mean curvature
158
6.3. THE SECOND FUNDAMENTAL FORM AS A METRICAL TENSOR. 6.3. 159
of a surface M ⊆ E3 has been given already. It should be remarked that

the Gaussian curvature is determined by the first fundamental form only
and that the critical points of the area functional are characterised by the
vanishing of the mean curvature.
An ovaloid is a compact surface in the three-dimensional Euclid-
ean space, which has strictly positive Gaussian curvature. It will always be
assumed that N denotes the inward pointing unit normal vector field on
an ovaloid, such that the second fundamental form II is a positive-definite
bilinear form. The Gauss map, which sends a point p of an ovaloid M to
its unit normal vector Np which is seen as an element of the unit sphere, is
a diffeomorphism G : M → S2 [Hadamard 1897, Hadamard 1903].
Similarly, a hyperovaloid is a compact hypersurface M ⊆ Em+1 ,
the second fundamental form of which is positive-definite.
A congruence is an isometry of E3 ; hence, a reflection will
also be called a congruence. A similarity is a homothety of E3 .
6.2.2. Hypersurfaces in semi-Riemannian Manifolds.

The adopted notation for a hypersurface M in a semi-Riemannian manifold
(M , g) has already been introduced on p. 62 ff and p. 70 ff. It should be re-
called that the following sign for the Riemann–Christoffel curvature tensor
has been adopted: if X , Y, Z ∈ X(M ), then
R(X , Y )Z = ∇[X ,Y ] Z − ∇X ∇Y Z + ∇Y ∇X Z .
We finish with the following reminder: Hypersurfaces will al-

ways be supposed to be connected and embedded. Generally speaking, the
precise regularity or differentiability conditions which are imposed have not
been stated explicitly in this text.
6.3. The Second Fundamental Form as a Metrical

Tensor and the Curvatures of the Second Fun-
damental Form.
Let M ⊆ E3 be a surface with positive-definite second fundamental form.

The intrinsic (Gaussian) curvature of the second fundamental form
159
will be denoted by KII . Several formulae for this curvature KII (or, more
generally, the scalar curvature of the second fundamental form of a
hypersurface in a semi-Riemannian space with a non-degenerate second
fundamental form) can be found in, a.o., [Cartan 1943, Rozet 1946a,
Rozet 1946b, Grove 1957, Hicks 1965, Wolf 1966, Erard 1968, Gardner
1972a, Schneider 1972, Glässner–Simon 1973, Klotz 1975, Baikoussis–
Koufogiorgos 1987a, Aledo–Romero 2003, Aledo–Haesen–Romero 2007,
Haesen 2007]. (See also p. 28 ff and p. 98 ff.)
The critical points of the area functional of the second funda-
mental form are precisely the surfaces for which the the mean curvature of
the second fundamental form HII vanishes [Glässner–Simon 1973, Gläss-
ner 1974]. The corresponding variational problem has been studied more
generally for semi-Riemannian hypersurfaces [Sodsiri 2005, Dillen–Sodsiri
2005a, Haesen–Verpoort–Verstraelen 0000, Haesen–Verpoort 0000]. (See
also [Arroyo–Garay–Mencía 2003] for a variational study of this functional
for curves in the unit sphere.) (See also p. 30 ff and p. 118 ff.)
The classical relations
1
∇II = ∇ + ∇III ∇IIIV W = A← ∇V (AW )

and (6.1)
2
for the connections of the fundamental forms are named after Wein-
garten [Bianchi 1894], § 85. They remain valid for hypersurfaces in space
forms. (See also [Hicks 1965, Simon 1972] and lemma 5.2.)
For the connection and curvature of the third fundamental
form of a hypersurface in a space form, see also [Baikoussis–Koufogiorgos
1988a], formula (3) and [Liu–Simon–Verstraelen–Wang 1999].
The I-geodesics on a hypersurface agree with the II-geodesics
(as unparametrised curves) on a hypersurface in a Riemannian space form
with positive-definite second fundamental form if and only if M is to-
tally umbilical [Hicks 1965, Simon 1972]. Similarly, the I-geodesic hyper-
spheres agree with the II-geodesic hyperspheres on a hypersurface M if and
only if M is totally umbilical.
As is known, the Gauss map G : (M , I) → (S2 , η) of a surface
in E with K 6= 0 is conformal if and only if the surface is minimal or part
3
of a sphere. (Here we denoted η for the standard metric on the sphere.)

It is easy to see that the Gauss map G : (M , II) → (S2 , η) is conformal if
160
6.3. THE SECOND FUNDAMENTAL FORM AS A METRICAL TENSOR. 6.3. 161
and only if M is part of a sphere. Furthermore, a hypersurface in Em+1 has

constant mean curvature if and only if the Gauss map G : (M , I) 7→ (Sm , η)
is a harmonic map [Ruh–Vilms 1970].
[Galvez–Martinez 2000]. Let M ⊆ Em+1 be an orientable hypersurface

with non-degenerate second fundamental form. M has con-
stant Gauss–Kronecker curvature if and only if the Gauss map
G : (M , II) 7→ (Sm , η) is a harmonic map.
(Since G : (M , III) 7→ (Sm , η) is locally an isometry, it is auto-

matically a harmonic map.) The theorem has been adapted for time-like
surfaces in, a.o., E31 (see the articles of M. Toda).
We list some theorems on the curvatures of the second funda-
mental form.
Darboux. (1 ) Suppose M ⊆ E3 is a surface, for which α + β H + γK = 0

(with α, β, γ ∈ R). If αI + βII + γIII is a non-degenerate metric,
then its Gaussian curvature is zero.
Corollary. (2 ) On a minimal surface with nowhere vanishing Gaussian

curvature, there holds KII = 0. Thus there exists a special co-
ordinate system on a minimal surface (Study’s parameters3 )
in which the coefficients of the second fundamental form are
constant. However, there exist surfaces which satisfy KII = 0
but are not minimal.
[Kühnel 1981]. A surface of revolution (with K nowhere vanishing) sat-

isfies KII = H if and only if the ratio of the principal curvatures
is constant.
[Blair–Koufogiorgos 1992]. A ruled surface (with K nowhere vanishing)

in E3 such that aKII + bH is constant along each rule for some
a, b ∈ R, 2a+ b 6= 0, must be a piece of a helicoid. The classical
right conoids satisfy KII = 2H.
1
In [Klotz 1983] it is asserted that Wolf has informed that Darboux has observed a
similar fact in [Darboux 1915] Vol. 2.
2
This theorem is also mentioned in [Rozet 1946b, Hicks 1965, Wolf 1966, Blair–Koufo-
giorgos 1992] et al.
3
[Blaschke–Leichtweiss 1973] Sect. 118; [Glässner 1974], Thm. 7.
161
[Becker–Kühnel 1996]. Let a straight line ρ and a plane π containing

ρ in Em+1 be given. Suppose γ ⊆ π is a plane curve. The
hypersurface which is described by the action on γ of the set
of isometries of Em+1 which preserve ρ, is called a rotation-
ally symmetric hypersurface. Assume M is a rotationally sym-
metric hypersurface M in Em+1 (m ¾ 2) with strictly negative
Gauss-Kronecker curvature.
— If m = 2 and KII < 0, then M does not meet the axis of

revolution.
— If (M , II) is complete and flat, then m = 2 and M is the
catenoid (up to similarity).
— If m ¾ 3 and M is complete (either w.r.t. the first or the
second fundamental form), then the second fundamental
form can impossibly have constant sectional curvature.
[Baikoussis–Koufogiorgos 1997]. Call a surface helicoidal if it is invari-

ant under a one-parameter family of “screwing” congruences.
A helicoidal surface (with K nowhere vanishing) satisfies KII =
H if and only if the ratio of the principal curvatures is constant.
Ruled surfaces in three-dimensional Euclidean or Lorentzian

space with a condition on the curvatures of the second fundamental form
have also been studied in, a.o., [Manhart 1982, Stamou–Magkos 2004,
Dillen–Sodsiri 2005a, Dillen–Sodsiri 2005b, Sodsiri 2005, Kim–Yoon 2004,
Kim–Yoon 2006, Yoon 2006b, Kim–Yoon 2007, Altın 2007]. For ruled sur-
faces in E3 whose mean curvature vector is an eigenvector of ∆II , see [Yoon
2006a] and some of the references therein. For translation surfaces in E31
with KII = 0, see [Goemans–Van de Woestyne 2007].
6.4. The Geometry of the Second Fundamental Form

and Relative Differential Geometry.
Athough relative differential geometry has not been touched upon, it seems
relevant to include the subject in this chapter since the geometry of the
162
6.4. RELATIVE DIFFERENTIAL GEOMETRY. 6.4. 163
second fundamental form can be seen as a special case thereof. We refer,

for example, to [Simon–Schwenk-Schellschmidt–Viesel 1991] for a system-
atic treatment of this subject. Initiating original articles which could be
mentioned here are written by W. Blaschke and his school, A. Duschek, E.
Müller, W. Süss, et al. From the point of view of convex geometry, we refer
to the 1903 article of H. Minkowski; see also, e.g., [Hadwiger 1955].
In the classical differential geometry of surfaces in E3 , the ex-
trinsic geometry of a surface M is described by means of structures induced
by the unit normal vector field. The notions of a relative normal vector field
and a relative co-normal vector field generalise respectively the vector field
N and the field of one-forms 〈N , ·〉 along M . If these require a compat-
ibility condition similar to 〈N , N 〉 = 1, the pair will be called a relative
normalisation.
It will be supposed that the affine space A3 is endowed with a
fixed origin and a fixed volume element Det. The standard connection of
A3 will be denoted by D.
Assume M is a surface in the affine space A3 . A vector field
y ∈ X(M ) which is transversal to M (i.e., y is nowhere tangent to M ) such
that, for all V ∈ X(M ), the vector field DV y is tangent to M , and which
satisfies an additional regularity condition which will be mentioned below,
is called a (regular) relative normal field. The operator
A( y) : X(M ) → X(M ) : V 7→ −DV y
is called the relative shape operator of M with respect to y. The relative

mean curvature H( y) is then defined as half of the trace of A( y) . Similarly
we define the relative Gauss–Kronecker curvature H2 ( y) as the determi-
nant of A( y) .
A possible way in which such a relative normal field y arises
is if, instead of the standard sphere, a “gauge ovaloid” M is defined with
respect to the chosen origin. The mapping P which sends a point m ∈ M
to the point P (m) ∈ M such that Tm M is parallel to TP (m) M , is the Pe-
terson mapping. If ym denotes the vector of Tm A3 which is parallel to the
position vector of the point P (m), a relative normal field is obtained. Up
to identification of these parallel tangent planes there holds dP = −A( y) .
163
A relative normal field y ∈ X(M ) determines a connection

( y)
∇ and a symmetric bilinear form h( y) by means of the relative (nor-
mal) Gauss equation: (for V, W ∈ X(M ))
( y)
DV W = ∇V W + h( y) (V, W ) y .
| {z } | {z }
tangent to M proportional to y
In this text, it will always be supposed that the relative normal field y
satisfies the additional regularity condition that h( y) is a semi-Riemannian
metrical tensor, the so-called relative metric.
The relative normal field y defines a volume form dΩ( y) =
y ù Det which is parallel with respect to the connection ∇( y) .
A (regular) relative co-normal field Y along M ⊆ A3 is a
field of one-forms of the surrounding affine space, Y ∈ Λ(M ), satisfying
Ker(Yp ) = T p M for all p ∈ M , as well as an additional regularity condition.
A symmetric bilinear field is now determined by
G(Y ) : X(M ) × X(M ) : (V, W ) 7→ (DV Y )(W ) .
The regularity condition which will be imposed on Y is that G(Y ) is a semi-

Riemannian metrical tensor.
A relative co-normal field Y ∈ X(M ) determines a connection
(Y )
∇ and a symmetric bilinear form Sb(Y ) by means of the relative (co-
normal) Gauss equation: (for V, W ∈ X(M ))
(Y )

DdY (V ) dY (W ) = dY ∇V W − Sb(Y ) (V, W ) Y .

| {z } | {z }
tangent to Im(Y ) proportional to Y
(In the above equation, the relative co-normal field is seen as a mapping
∗
Y : M → A3 of M in the dual space, and D also denotes the standard
connection on the dual space.)

A pair Y, y of a relative co-normal field and a relative normal
field is called a relative normalisation if Y ( y) = 1. In this case it follows
that G(Y ) = −h( y) .
All connections on X(M ) which have been introduced, can be
extended to a connection on one-forms in the standard manner. These will
be denoted with the same symbol.
164
6.4. RELATIVE DIFFERENTIAL GEOMETRY. 6.4. 165
Let Φ : X(M ) → Λ(M ) stand for the metrical equivalence with

respect to the relative metric h. Relation (6.1) is valid for an arbitrary
relative normalisation Y, y [Nomizu–Simon 1992]:

1 ( y) (Y )

( y)

∇(h( y) ) = ∇ + ∇(Y )

and ∇V W = Φ← ∇V (ΦW ) .
2
Here ∇(h( y) ) stands for the Levi-Civita connection of the semi-Riemannian
metric h( y) .
The relative area of M is defined as
Z
Area( y) (M ) = dΩ( y) .
M
If y is constructed from the gauge ovaloid M , this relative area will also be
written as AreaM (M ). In [Minkowski 1903] this relative area is denoted
as 3 V (M , M , M ). (This relative area can also be seen as an anisotropic
area.) The critical points of the area functional relative to a gauge ovaloid
M are exactly the surfaces of which the relative mean curvature vanishes.
All what has been said is straightforward for the Euclidean
normalisation {〈N , ·〉 , N } of a surface M ⊆ E3 . In this case, h(N ) is the
The study of the area of the second fundamental form is also
related with the relative differential geometry stemming from another rel-
ative normalisation, according to the following result.
[Manhart 1983]. On a surface M ⊆ E3 (with nowhere vanishing Gauss-

ian curvature), there exists a unique relative normalisation
YII , NII such that dΩII = dΩ(NII ) .

In particular, there holds AreaII (M ) = Area(NII ) (M ). This II-normalisation

is given by 
p p
 NII = |K| N − gradII |K| ;
 YII = p1 〈N , ·〉 .
|K|
The relative metric is h(NII ) = p1 II. The relative mean curva-

|K|
ture is given by p
H(NII ) = |K| HII . (6.2)
165
In particular, HII = 0 if and only if H(NII ) = 0. Since both of these equa-

tions are the Euler-Lagrange equations of a variational problem, we find
the following result.
[Manhart 1983]. Let a surface M ⊆ E3 (with K nowhere vanishing) be

given. Assume the surface in E3 which is described by the
vector field NII is part of an ovaloid M . Then M is a critical
point of the area functional of the second fundamental form if
and only if M is a critical point of the area functional relative
to the gauge ovaloid M .
In [Simon 1967] (see also the references of T. Kubota, W. Süss,

W. Scherrer and K.P. Grotemeyer therein) the Jellett-Minkowski formulae
have been generalised to the relative setting. Specialising these formulae
to the II-normalisation, the integral formula (2.11) is recovered:
 Z Z

 dΩII = HII 〈−N , P〉 dΩII ;


Z Z
 H pK dΩ = H2 (NII )



II II p 〈−N , P〉 dΩII .
K
Moreover, similar to the results of the first chapter, a combi-

nation of both these integral formulae gives some progress on a problem
stated on page 45:
Corollary of the Relative Jellett–Minkowski Integral Formulae. If an

p
ovaloid in E3 satisfies HII K = C for some C ∈ R, then the
relative shape operator A(NII ) is a multiple of the identity. This
means
p p
K A + ∇ gradII K = C id . (6.3)
It is not clear whether this implies that M is a Euclidean sphere.
166
6.6. THE CONFORMAL GEOMETRY OF THE SECOND. . . 6.6. 167
6.5. The Conformal Geometry of the Second Funda-

mental Form.
A study of the conformal geometry of the second fundamental form is mo-

tivated since a change in a relative normalisation leads to a conformal
change in the relative metric. For applications of the conformal structure
of the second fundamental form, see the articles of T. Klotz.
A corollary of the results described in [Galvez–Martinez 2000]
is the following.
[Galvez–Martinez 2000]. An ovaloid in E3 is determined up to similarity

by the third fundamental form and the conformal structure of
6.6. II-minimal Surfaces.
In view of § 6.4, results on relative minimal surfaces can be specialised to

II-minimal surfaces. In this way the following result can be found:
“A Partial II-Bernstein Theorem” [Reilly 1976] Thm. 1. There does not

exist a surface M ⊆ E3 satisfying:
— M is the graph of a function defined over the entire R2 ;

— M has nowhere vanishing Gaussian curvature;
— M is II-minimal;
— The gauge surface M described by Manhart’s II-normal
field is bounded, convex, has an injective, non-singular
2
HM
Gauss-mapping and satisfies KM
< C for some constant
C.
The following theorems hold true for ruled surfaces4 .

4
The results on II-minimal ruled surfaces of Glässner are not correct.
167
[Manhart 1982]. A ruled surface with nowhere vanishing Gaussian cur-

vature is a II-minimal surface if and only if it is a piece of a
helicoid.
Furthermore, similar to theorem 2.31, every simply connected

domain of a II-minimal surface admits a non-trivial isometric deformation.
Information about the second variation of AreaII should be in-
cluded a.o. in the dissertations of Glässner (1972) and in [Arroyo–Garay–
Mencía 2003] for curves in the unit sphere. Information about the second
variation of Area(NII ) should be included a.o. in Wiehe’s dissertation (1998).
6.7. Congruence Theorems.
6.7.1. Some Classical Congruence Theorems.
We owe the classical congruence theorem of surfaces in E3 to P.O. Bonnet:
[Bonnet 1867a, Bonnet 1867b]. A surface M ⊆ E3 is determined up to

congruence by the first and the second fundamental form.
Conversely, if I and II satisfy the Gauss and Codazzi equations
det II

 det I = intrinsic (Gaussian) curvature of (M , I) ;

∇II is totally symmetric;
and I is positive-definite, such a surface can be found.
Bonnet himself began to study additional hypotheses which re-

duce the possible isometries of surfaces to congruences of the surrounding
Euclidean space. This research led him, for example, to the construction
of pairs of surfaces in E3 , between which an isometry can be found which
168
6.7. CONGRUENCE THEOREMS. 6.7. 169
preserves either the principal directions or the principal curvatures5 . Fur-

thermore, Cohn-Vossen has proved that ovaloids do not admit non-trivial
isometries.
Remark. We shall say that the first fundamental form is preserved un-
der a mapping µ : M → M e if for all vector fields V, W on M , there holds
eI(dµ(V ), dµ(W )) = I(V, W ). The same convention applies to all other ten-
sors.
“The Rigidity of the ovaloid.” (Cohn-Vossen 1927) (6 ) An ovaloid in E3

is determined up to congruence by its first fundamental form,
in the following sense: a diffeomorphism between two oval-
oids which preserves the first fundamental form is a congru-
ence.
The theorem (together with its infinitesimal version, which

will be discussed below) goes back to Euler, Lagrange, Cauchy, Minding
and others7 . We refer to [Struik 1933b] p. 165 and [Efimow 1957] p. 223
for some historical comments, and to [Efimow 1957, Pogorelov 1973] for
various extensions of the theorem.
A certain generalisation of Cohn-Vossen’s theorem has been
given for surfaces in three-dimensional space forms in [Hsiung–Liu 1977].
A different modification of this theorem is the following result.
[Hsü 1960]. A diffeomorphism between two ovaloids in E3 which pre-

serves K I is a similarity.
5
Many of these classical results make use of the following theorem, which seems note-
worthy to mention in this context. The “Middle Surface” (Mittelsfläche) of two isometric
surfaces (in E3 ) is described by the midpoints of the segments which connect corresponding
points. In this case, the second fundamental form of the “Mittelsfläche” is the arithmetic
mean of the second fundamental forms of the surfaces.
6
See, a.o., [Cohn-Vossen 1927, Blaschke 1923, Hopf–Samuelson 1938, Herglotz 1943,
Efimow 1957, Klingenberg 1978, Hopf 1983, Chern 1967, Montiel–Ros 2005].
7
[Cauchy 1813], [Minding 1838a] p. 368.
169
Remark. Similar to lemma 5.8 on p. 96, it can easily be shown that a

diffeomorphism between ovaloids in Em+1 which preserves the Levi-Civita
connection of the first fundamental form is a similarity. Hsü’s theorem
extends this result, since it is only asked that the Ricci tensors agree.
It could be asked whether another generalisation of this re-
sult, which states that an ovaloid is determined by its Levi-Civita connec-
tion, holds: “Suppose a diffeomorphism between two ovaloids preserves the
geodesics (as unparametrised curves). Is the diffeomorphism a similarity?”
The answer is in the negative, see [Simon–Vrancken–Wang–Wiehe 2000]
and the references of V.S. Matveev, P.J. Topalov, S. Tabachnikov and K. Voss
therein.
For hypersurfaces in a Euclidean space of higher dimension,

we have the following result.
Theorem of Beez (1876) – Killing (1885). (8 ) A hypersurface in Em+1 of

which the shape operator has rank ¾ 3 is determined up to
congruence by its first fundamental form.
We mention the following theorems by Voss and Schur–Ros.
[Voss 1956], p. 196. (9 ) Assume the function F : { (u, v) ∈ R2 : u2 ¾

v } ⊆ R2 → R satisfies
2 2
1 ∂F ∂F ∂F ∂F

+u +v > 0.
4 ∂u ∂u ∂v ∂v
Suppose an orientation-preserving mapping between two com-
pact surfaces of E3 is a parallel mapping, i.e., the lines con-
necting a point on the first surface with its image on the other
surface are parallel. This mapping is a translation as soon as it
preserves F (H, K).
In particular, a parallel mapping between ovaloids which pre-
serves either H or K is a translation.
8
See, e.g., [Spivak 1979], Vol 5, Chap. 12., Thm. 1.
9
See also [Spivak 1979], volume V, Chapter 12, p. 302.
170
Theorem of Schur and Ros (1988). (10 ) Let ψ : M → Em+1 be a com-

pact hypersurface with non-negative mean curvature H. Let
ψe : M → Em+n be another isometric immersion with mean
e If |H|
curvature vector H. e ¶ H, then both immersions are con-
gruent.
6.7.2. Congruence Theorems Involving the Second Fundamen-

tal Form.
It seems natural to ask whether Cohn-Vossen’s theorem can be adapted to
the geometry of the second fundamental form:
[Schneider 1972, Glässner–Simon 1973]. Is an ovaloid determined up

to congruence by the second fundamental form?
As far as I know, only partial results have been obtained. First

of all, theorem 2.7 (p. 29) of Schneider implies that a (sufficiently smooth)
II-isometry between an ovaloid M and the standard sphere S2 is a congru-
ence. Furthermore, various extra assumptions force a II-isometry between
ovaloids to be a congruence.
[Grove 1957]. An ovaloid in E3 is determined up to congruence by the

second fundamental form and the Gaussian curvature.
[Erard 1968] p. 31. A surface which is II-isometric with a part of a sphere

S 2 ( 1r ) and has the same Gaussian curvature r12 , is locally con-
gruent to that sphere.
[Vortisch–Walden 1970]. An ovaloid in E3 is determined up to congru-

ence by the second fundamental form and the mean curvature.
[Walden 1971]. (11 ) Let N , M ⊆ E3 be ovaloids. If there exists a II-

isometry between N and M which preserves any of the fol-
lowing quantities, then N and M are congruent:
(a) H;
10
[Chern 1967] p. 36, [Ros 1988] p. 216.
11
See also [Glässner–Simon 1973] for a generalisation of this theorem.
171
(b) H/K;
(c) K;
(d) (λ1 2 +λ2 2 ), where λ1 and λ2 are the principal curvatures;
(e) (R1 2 + R2 2 ), where R1 and R2 are the radii of principal
curvature.
[Gardner 1972a, Gardner 1972b]. A II-isometry between hyperovaloids

in Em+1 which preserves any of the following quantities is a
congruence:
(a) the Gauss–Kronecker curvature;

(b) the volume element;
(c) the Ricci tensor.
[Glässner–Simon 1973], [Huck et al 1973] § 2.3.(a).B. Assume a real-

valued function F : R+ × R+ 7→ R : (u, v) 7→ F (u, v) of two
variables satisfies
∂F ∂F ∂F ∂F
¾0 and + > 0. (6.4)
∂u ∂v ∂u ∂v
Every II-isometry between two ovaloids in E3 which preserves
either F (H, K) or F ( HK , K1 ), is a congruence.
It is known that certain surfaces of constant Gaussian curva-

ture belong to a one-parameter class of II-isometric surfaces.
[Erard 1968] p. 34; [Voss 1970]. A surface with strictly positive constant
Gaussian curvature, free of ombilical points, belongs to a one-
parameter family of surfaces all of which have the same second
fundamental form and the same principal curvatures.
(attributed to S. Lie 1870). (12 ) Every surface M ⊆ E3 with constant

negative Gauss curvature has a one-parameter family of defor-
mations preserving the second fundamental form, the Gauss-
ian curvature, and the angle between the asymptotic lines.
12
The fact is described in [Toda 2002], p. 10.
172
6.7.3. Congruence Theorems Involving the Third Fundamental

Form.
Suppose M ⊆ E3 is an ovaloid. In consequence of the definition of the third

fundamental form, the Gauss map is an isometry G : (M , III) → (S2 , η).
Since the congruences of E3 are the only isometries of S2 , the following
information concerning the abstract surface M is equivalent:
— the third fundamental form of M ;
— the Gauss map G : M → S2 ⊆ E3 (up to composition with a congru-

ence).
In consequence, all congruence theorems involving the third

fundamental form can be formulated in two ways. For example, a well-
known theorem of Christoffel is13
[Christoffel 1865]. Version a: An ovaloid in E3 is determined up to trans-

lation by HK (viewed as function on its spherical image through
G ).
[Christoffel 1865]. Version b: An ovaloid in E3 is determined up to con-

gruence by HK and the third fundamental form.
The previous congruence result is, like so many others, of a

truly global nature. This is illustrated by the theorem below.
[Erard 1968], p. 39. A surface M which is free of umbilics, which has

nowhere vanishing Gaussian curvature, and for which HK is a
non-zero constant, belongs to a one-parameter family of sur-
faces all of which have the same third fundamental form and
the same principal curvatures.
Further classical results concerning the third fundamental form are:
See [Christoffel 1865] and also, e.g., [Hurwitz 1902, Blaschke 1923, Walden–Wegner
13
1972] and [Spivak 1979] Vol. V, Chapter 12, p. 299.
173
Theorem of Liebmann–Minkowski (1899). (14 ) An ovaloid in E3 is de-

termined up to congruence by the third fundamental form and
the Gaussian curvature.
Theorem of Liebmann (1899). (15 ) An ovaloid in E3 is determined up

to congruence by the third fundamental form and the mean
curvature.
[Aleksandrov 1939]. Let F : R+ +

0 ×R0 → R : (u, v) 7→ F (u, v) be a function
which satisfies
∂F ∂F
> 0.
∂u ∂v
An ovaloid in E3 is determined up to congruence by the third
fundamental form and F ( HK , K1 ).
[Bianchi 1894], § 86. A surface (with nowhere vanishing Gaussian cur-

vature) M ⊆ E3 is determined up to congruence by the second
and the third fundamental form. Conversely, if III has a con-
stant Gaussian curvature equal to one and the second Codazzi
equation
∇III II is totally symmetric
is satisfied, such a surface can be found. The surface M ⊆ E3

is determined (up to congruence) by the second and the third
fundamental form.
6.7.4. Congruence Theorems Involving the Shape Operator.

In a similar way it can be asked to which amount an ovaloid is determined
by its shape operator. Such a study can be seen as a natural extension of
Bonnet’s Ansatz which was mentioned on p. 169.
Question. Assume that a diffeomorphism between two ovaloids (with

nowhere dense ombilics) in Euclidean three-dimensional space
preserves the shape operator (in the sense of the remark on p.
169). Is this mapping a congruence?
14
[Liebmann 1899b] p. 136; [Minkowski 1903] p. 481.
15
[Liebmann 1899b] p. 136.
174
We mention some partial results (see also [Simon–Voss–Vrancken–Wiehe

2002]).
[Simon–Vrancken–Wang–Wiehe 2000]. Let a diffeomorphism between

two ovaloids in E3 with nowhere dense umbilics be given. If
the shape operator as well as the spherical area element dΩIII
is preserved under the mapping, then it is a congruence.
[Simon–Vrancken–Wang–Wiehe 2000]. Let a diffeomorphism between

two ovaloids in E3 which preserves the shape operator be given.
If one of the two ovaloids is an ovaloid of revolution with
nowhere dense umbilics, then the mapping is a congruence.
Mappings between surfaces which preserve the shape oper-

ator have also been described in [Finikoff 1933, Finikoff–Gambier 1933,
Ferapontov 2001, Bryant 2001, Voss 2001]. However, most attention has
been paid to the local situation: the surfaces are often assumed to be home-
omorphic to a square and/or free of umbilics.
An important observation which has been made is that when-
ever a mapping between two surfaces can be found which preserves the
shape operator, a one-parameter family of such surfaces joining the two
initial surfaces exists.
6.7.5. The Weyl and Minkowski Problem.
Weyl posed the question whether an ovaloid can be constructed for any
given metric [Weyl 1916].
The Weyl embedding problem. Assume a metric g with positive Gauss-

ian curvature K is given on the sphere S2 . Can the abstract
Riemannian manifold (S2 , g) be realised as an ovaloid in E3 ?
This question has been solved in the affirmative16 . Similarly,

the following question can be posed.
See [Weyl 1916, Weyl 1917, Lewy 1938b, Nirenberg 1953, Efimow 1957, Heinz 1962,
16
Pogorelov 1973].
175
The Minkowski embedding problem. Assume a positive function K is

given on the sphere S2 . Can an ovaloid M ⊆ E3 be found, of
which the Gaussian curvature is K ◦ G ?
Under a suitable integrability condition, the question has been

solved in [Minkowski 1903, Lewy 1938a].
The theorem on existence and uniqueness of ovaloids (Weyl’s
embedding theorem and Cohn-Vossen’s theorem) is adapted for surfaces in,
a.o., hyperbolic space in [Pogorelov 1973], p. 321). The solution of Weyl’s
embedding problem is generalised for certain closed surfaces in a general
Riemannian space in [Pogorelov 1973], p. 369 and p. 413.
Remark. We signal that no congruence theorems for surfaces in space

forms have been included in this overview.
6.8. Infinitesimal Deformations.

The results which are listed below are tightly connected with the preceding
ones, since an infinitesimal isometric deformation is the infinitesimal model
of a family of isometries of surfaces. Hence most congruence results have
an infinitesimal counterpart. However, since both the premises and the
conclusions in these counterparts are weaker (as compared to the original
theorems), neither are these results an immediate consequence of these
earlier-mentioned results, nor do they directly imply them.
The classical theorem in this field, which was found by Lieb-
mann, is known as the “infinitesimal rigidity of the ovaloid17 .”
“The Infinitesimal Rigidity of the Ovaloid.” (Liebmann 1900). (18 )

The infinitesimal congruences are the only infinitesimal iso-
metric deformations of an ovaloid.
17
Although the theorem is often simply referred to as the “rigidity of the ovaloid.”
18
See [Liebmann 1899a], p. 45 (for the sphere), and [Liebmann 1900], p. 83 (for an
ovaloid).
176
6.8. INFINITESIMAL DEFORMATIONS. 6.8. 177
Several proofs of this result and various related theorems have

been given by, amongst others, A. Voss, H. Weyl and W. Blaschke19 .
This is a global result, since we have:
[Cohn-Vossen 1927]. (20 ) An ovaloid, an arbitrarily small (open) part of

which has been removed, admits a non-trivial isometric defor-
mation.
[Cohn-Vossen 1930]. Construction of surfaces of revolution of genus 0 en

1, which are not infinitesimally rigid.
The infinitesimal rigidity is given for certain closed surfaces in

a general Riemannian space in [Pogorelov 1973], p. 369 and p. 413.
6.8.1. Infinitesimal Deformations Which Preserve the Second

Fundamental Form.
We do not wish to mention the manifold extensions of these infinitesimal
rigidity theorems which exist [Efimow 1957, Pogorelov 1973], but shall
turn our attention instead to the study of infinitesimal II-isometric deforma-
tions. The main problem is to solve the infinitesimal version of a question
which was mentioned on p. 171, namely:
Question. Is every infinitesimal II-isometric deformation of an ovaloid an

infinitesimal congruence?
As far as I am aware, only partial results have been obtained21 :
[Erard 1968] p. 63. Let a surface M ⊆ E3 be given which either has non-
zero constant Gaussian curvature or is a surface of revolution
satisfying K(H 2 − K) 6= 0. There exist locally non-trivial in-
finitesimal deformations under which
δII = 0 , δH = 0 and δK = 0 .
[Simon 1971]. (22 ) An infinitesimal II-isometric deformation of an oval-

19
See also [Liebmann 1900, Liebmann 1902, Blaschke 1912], [Blaschke 1916] p. 163–
164, and [Efimow 1957].
20
See also [Süss 1927].
21
See also [Švec 1977, Švec 1981, Zhou 1987, Yang 1989, Zhou 1990, Chen 1991].
22
See also [Glässner–Simon 1973] for a generalisation of this result.
177
oid which satisfies any of the additional hypotheses below, is

an infinitesimal congruence:
(a) δH ≡ 0;
(b) δ(H/K) ≡ 0;
(c) δK ≡ 0;
(d) δ(λ1 2 + λ2 2 ) ≡ 0;
(e) δ(R1 2 + R2 2 ) ≡ 0, where R1 and R2 are the radii of prin-
cipal curvature.
[Voss 1970]. The sphere in E3 does not admit a non-trivial infinitesimal

II-isometric deformation of class of differentiability C 3 . (This
result is not valid for C 2 [Becker–Kühnel 1996]. See also
[Huck et al 1973].)
[Roitzsch 1971, Glässner–Simon 1973, Huck et al 1973]. Suppose a

function F : R+ × R+ 7→ R : (u, v) 7→ F (u, v) of two variables
satisfies (6.4) on p. 172. Every infinitesimal II-isometric defor-
mation of an ovaloid which satisfies either δ (F (H, K)) = 0 or
δ F ( HK , K1 ) = 0, is an infinitesimal congruence.

6.8.2. Infinitesimal Deformations Which Preserve the Third Fun-

damental Form.
[Walden–Wegner 1972, Huck et al 1973]. Let M ⊆ E3 be an ovaloid. An

H 1 III-isometric deformation for which δ (F (H, K)) or
infinitesimal
δ F ( K , K ) vanishes, where the function F satisfies (6.4) on
p. 172, is an infinitesimal congruence.
6.8.3. Infinitesimal Deformations Which Preserve the Shape Op-

erator.
Of course, the flow of a tangent vector field with a support which is com-
posed of umbilical points, is an infinitesimal deformation which preserves
the shape operator.
178
6.9. ON THE GLOBAL GEOMETRY OF OVALOIDS. 6.9. 179
We make the following simple related observation. Let f :

S → R be the signed distance to a plane through the centre of S2 . The non-
2
trivial infinitesimal deformation of the sphere, determined by the normal

deformation vector field f N , preserves the shape operator (i.e., δA = 0).
See also [Simon–Voss–Vrancken–Wiehe 2002] for one-param-
eter families of non-compact surfaces of revolution preserving the shape
operator.
6.9. On the Global Geometry of Ovaloids.
6.9.1. Isoperimetric Inequalities and Related Integral Inequal-

ities and Equalities.
The sphere can be characterised amongst all compact surfaces by a sim-

ple relation between two quantities of a global and primitive nature. This
might be described as one of the most old and profound characteristic prop-
erties of the sphere.
(Some general references for this § are [Blaschke 1916, Blasch-
ke 1923, Hadwiger 1955, Eggleston 1977, Osserman 1978, Hopf 1983,
Gardner 2002, Montiel–Ros 2005].)
Isoperimetrical Characterisation of the Sphere. A compact surface M ⊆

E3 satisfies
(Area(M ))3 ¾ 36π (Vol(M ))2
and equality is achieved if and only if M is a sphere.
Minkowski was able to deduce this isoperimetrical characteri-

23
sation starting from an inequality in which the volume of the sum of two
convex bodies is involved. If Ψ and Φ are convex bodies in R3 , the body
¦ ©
sΨ + tΦ = sψ + tϕ ψ ∈ Ψ and ϕ ∈ Ψ
is still convex. There holds:

23
For an older proof (Steiner’s symmetrisation), see [Blaschke 1916].
179
The Brunn–Minkowski inequality (1888).

p p p
Vol (Ψ + Φ)) ¾ Vol (Ψ)) + Vol (Φ)) .
3 3 3
Minkowski obtained further profound results with help of the last-mentioned

inequality.
The Minkowski Inequalities (1903). (24 ) For a convex surface, we have

the inequality Z
p
4π Area(M ) ¶ H dΩ
with equality if and only if M is a sphere. Furthermore,

Z
3 Vol(M ) H dΩ ¶ (Area(M ))2 .
Equality does not force M to be a sphere, as Minkowski has

noted.
The Integral Formulae of Jellett–Minkowski (25 ) have been

mentioned already on p. 13. Besides the adaptions of the Jellett-Minkowski
formulae to the geometry of the second fundamental form (cf. supra), the
following result is noteworthy.
[Grove 1957] p. 783. Every ovaloid M ⊆ E3 satisfies

Z Z
H(2KII − H) dΩ = K dΩ . (6.5)
On p. 44 ff we have mentioned already two geometric inequal-

ities in which the area of the second fundamental form was involved. One
of these was the “II-isoperimetric Inequality” [Manhart 1989b].
The inequality of Wirtinger–Sobolev–Poincaré–Blaschke is (at
least for curves) related to the isoperimetric inequality [Osserman 1978].
This inequality has been generalised recently in the following way.
24
See [Minkowski 1903] and also, e.g., [Blaschke 1916], p. 101–103, the references in
[Blaschke 1916], [Osserman 1978],[Hadwiger 1955].
25
See [Jellett 1853, Bonnet 1853, Minkowski 1903, Blaschke 1916, Yano 1970,
Osserman 1978, Klingenberg 1978, Montiel–Ros 2005].
180
6.10. CHARACTERISATIONS OF SPHERES OR . . . 6.10. 181
[Colesanti 2007]. Let M ⊆ Em+1 R be a hyperovaloid. For any function

f : M → R satisfying f dΩ = 0, there holds
Z Z
2
m H f dΩ ¶ II(gradII f , gradII f ) dΩ .
The following theorem can in particular be applied to the do-

main M bounded by a hyperovaloid M ⊆ Em+1 .
[Ros 1987]. Suppose (M , g) is be compact Riemannian (m + 1)-dimen-

sional manifold with smooth boundary M and non-negative
Ricci curvature. Suppose that the mean curvature H of the
boundary M is strictly positive. There holds
Z
1
dΩ ¾ (m + 1)Vol(M ),
M
H
and equality occurs if and only if (M , g) is isometric to a Euclid-

ean ball.
(In the same article, Ros obtained an extension of the char-

acterisation of the Euclidean spheres as critical points of the isoperimetric
functional. See [Ros 1987], Thm. 3.)
6.9.2. Existence of Umbilics and Closed Geodesics.
The following conjecture has been proved for analytic ovaloids.
Carathéodory’s conjecture. (26 ) There exist at least two umbilics on any

ovaloid.
Theorem of Poincaré (1905). (27 ) Every ovaloid has at least three closed
geodesics.
26
See also Spivak, Vol 3, Ch. 4.
27
See [Poincaré 1905], and also, e.g., [Blaschke 1923], § 97. Also Birkhoff, Herglotz,
Lyusternik, Schnirelmann, Klingenberg et al have worked on similar theorems.
181
6.10. Characterisations of Totally Umbilical Hyper-

surfaces.
The fact that the sphere is the only minimum of the area functional, which
has been restricted to the class of compact surfaces in E3 of a certain fixed
volume, has been mentioned already. Besides area and volume, other quan-
tities such as the total mean curvature were involved in further unicity re-
sults of the sphere.
The infinitesimal behaviour of these quantities, which depend
on the global geometry of the surface, can be studied as well; and the
question arises whether the sphere is even the only ovaloid the area of
which is not affected by infinitesimal, volume-preserving distortions. Thus,
the surface is required to be a critical point of the area functional under a
volume constraint and has to satisfy the constant mean curvature equation,
which is the Euler-Lagrange equation of the variational problem.
In this manner, we turn our attention to characterisations of
the sphere amongst the ovaloids by means of pointwise relations between
their curvatures. Some, but perhaps not all, of these equations can be
interpreted as the Euler-Lagrange equation of a variational problem.
Further relations in which the curvatures of the second funda-
mental form are involved can be satisfied by spheres only.
Some generalisations for hypersurfaces of semi-Riemannian
manifolds will be considered.
6.10.1. Characterisations of Spheres in E3 .
• Some Useful Theorems.—Before we list several further characteristic

properties of the sphere, the following important results of Hilbert and
Weyl are mentioned.
Lemma of Hilbert (1901). (28 ) Let M ⊆ E3 be a surface where K > 0.

Suppose p ∈ M satisfies λ1 (p) > λ2 (p). It is impossible that
λ1 has a local maximum at p and λ2 has a local minimum at
p.
28
See, e.g., [Hilbert 1901]; [O’Neill 1966], Lemma 3.6; [Hopf 1983], p. 126.
182
Lemma of Weyl (1916). (29 ) Let M ⊆ E3 be a surface where K > 0. It

is impossible that H has a local maximum and K has a local
minimum at the same non-umbilical point.
Chern has employed these ideas to obtain the following result.
Corollary of Hilbert’s Lemma. (30 ) The spheres are the only ovaloids
which satisfy λ1 = f (λ2 ) for a decreasing function f .
Corollary of Weyl’s Lemma. The spheres are the only ovaloids which sat-
isfy F (H, K) = 0 for a function F : R+ × R+ → R which is
increasing in both variables and strictly increasing in at least
one of its variables.
In this sense the articles [Hilbert 1901, Chern 1945] have had
considerable influence on the results which will be described in this section.
For example, in the articles [Simon 1976] and [Baikoussis–Koufogiorgos
1987a], adaptions of these results to curvatures, associated to the second
fundamental form, have been employed. (Compare also with § 5.3.2.)
• The Constant Mean Curvature Condition.—The first characterisation of

the sphere which will be mentioned, is Liebmann’s famous theorem. Jellett
and Bonnet have also worked on the same problem.
Theorem of Jellett–Liebmann (1899). (31 ) The spheres are the only oval-
oids in E3 with constant mean curvature.
Aleksandrov’s Sphere Theorem (∼1950). (32 ) The spheres are the only
compact surfaces in E3 of constant mean curvature. (As else-
where in this text, a surface cannot have self-intersections.)
29
The lemma follows from the fact that a point p where the mean curvature of a sur-
face M ⊆ E3 with strictlypositiveGaussian curvature achieves a local maximum, satis-
∆K
2
fies the inequality H(p) ¶ K− 4K
. See [Weyl 1916], [Chern 1945], [Efimow 1957],
(p)
[Hopf 1983] p. 126.
30
[Chern 1945], [Hopf 1983] p. 126.
31
[Liebmann 1899a] p. 52; see also [Jellett 1853, Bonnet 1853, Liebmann 1900].
32
[Aleksandrov 1962]; see also [Reilly 1973, Reilly 1977, Reilly 1982, Hopf 1983,
Li 1983, Ros 1987, Ros 1988] and others.
183
Theorem of Hopf (∼1950). (33 ) The only constant mean curvature im-
mersion of the sphere in E3 is the round sphere.
In view of the previous theorems, the following result is really

remarkable.
[Wente 1986]. (34 ) There exists a contant mean curvature immersion of

a torus in E3 .
• Three Unicity Results For the Sphere.—We mention only briefly the
following three unicity results for the sphere:
— The Theorem of Liebmann (1899)–Hilbert (1901), (35 ) which was

mentioned on p. 15;
— It has already been mentioned that the spheres can be variationally

characterised among the ovaloids in E3 as the only critical
points of several other functionals (see p. 16 ff);
— Of course, a number of congruence theorems have characterisations of

the sphere as a corollary.
• Characterisations of Spheres in E3 by Means of Curvature Equa-

tions.—Let us now list several characteristic properties of the sphere.
[Nakajima 1926]. If K = C H 2 for some C ∈ R0 on an ovaloid M ⊆ E3 ,

then M is a sphere.
[Nakajima 1926]. An ovaloid is a sphere if
aK + bH + c = 0 ,
for non-zero constants a, b and c.

33
See, e.g., [Hopf 1951, Hopf 1983] and [Montiel–Ros 2005].
34
A precursor of this result is [Hsiang–Teng–Yu 1982].
35
See [Liebmann 1899a] p. 44; [Hilbert 1901].
184
[Grove 1957]. (36 ) The spheres are the only ovaloids satisfying KII = H.
[Schneider 1972]. If KII is constant on an ovaloid M ⊆ E3 , then M is a

sphere.
[Koutroufiotis 1974]. If KII = C K (with C ∈ R) on an ovaloid M ⊆ E3 ,

then M is a sphere.
p
[Koutroufiotis 1974]. If KII = C K (with C ∈ R) on an ovaloid M ⊆ E3 ,
then M is a sphere.
[Simon 1976]. Suppose F (K, KII ) = 0 on an ovaloid M ⊆ E3 , where the

function F : R+ 0 × R → R is increasing in both variables and
strictly increasing in at least one of its variables. Then M is a
sphere.
[Koufogiorgos–Hasanis 1977]. Let M ⊆ E3 be an ovaloid and C, s, r ∈ R

(where 0 ¶ s ¶ 1). If KII = C H s K r , then M is a sphere.
[Stamou 1978]. Let a function F : R+

0 ×R → R : (u, v) 7→ F (u, v) be given,
which satisfies
∂F F ∂F
0¶ ¶ and ¶0 (on R+
0 × R).
∂u u ∂v
The spheres are the only ovaloids which satisfy F (H, KII ) = 0.
[Stamou 1978]. Let a function F : R+

0 ×R → R : (u, v) 7→ F (u, v) be given,
which satisfies
∂F F F ∂F
0¶ ¶ and ¶ (on R+
0 × R).
∂u u 2v ∂v
The spheres are the only ovaloids which satisfy F (H, KII ) = 0.
[Stamou 1978]. If F (H, KII ) = 0 on an ovaloid M ⊆ E3 , where the func-

tion F : R+
0 × R → R is increasing in both variables and strictly
increasing in at least one of its variables, then M is a sphere.
36
See the last line of [Grove 1957]. The result is an immediate consequence of (6.5) on
p. 180.
185
[Koufogiorgos 1979]. If KII = C H K r (with C, r ∈ R) on an ovaloid M ⊆

E3 , then M is a sphere.
p
[Hasanis 1980]. If KII = C H K r (with C, r ∈ R) on an ovaloid M ⊆ E3 ,
then M is a sphere.
[Klotz 1980] fact 5, p. 369. (37 ) Let M ⊆ E3 be an ovaloid. Suppose that

∇II I is totally symmetric. Then M is a sphere.
[Stamou 1981]. If any of the assumptions
• HII = H f (K) ;
• HII = H + f (K) ;
holds for an increasing function f on an ovaloid M ⊆ E3 , then

M is a sphere.
[Stamou 1981]. If HII = H on an ovaloid M ⊆ E3 , then M is a sphere.
[Stamou 1981]. If KII = HII on an ovaloid M ⊆ E3 , then M is a sphere.
[Hasanis 1982]. If KII = C H s K r (with C, r, s ∈ R and 0 ¶ r ¶ 12 ) on an

ovaloid M ⊆ E3 , then M is a sphere.
[Stamou 1987]. Suppose KII = C H s K r on an ovaloid M ⊆ E3 , for C, r, s ∈

R which satisfy one of the following conditions:
• r(s − 1) > 0 ;
• 2r + s − 1 = 0 and s < 1 .
Then M is a sphere.
[Stamou 1987]. Let M ⊆ E3 be an ovaloid. Let a function F : R×R+ 0 →R

be given which is increasing in one variable and decreasing in
the other, and is strictly monotonic in at least one of its vari-
ables. M is a sphere as soon as one of the following conditions
is satisfied:
• F (HII − H, K) = 0 ;
37
See also [Saban 1982].
186
H
• F ( HII , K) = 0 ;
• F (HII − KII , K) = 0 ;
H
• F ( K II , K) = 0 and KII > 0 .
II
[Baikoussis–Koufogiorgos 1987a]. Let M ⊆ E3 be an ovaloid and C, s, r ∈

R (where 21 ¶ r + s ¶ 1). If KII = C H s K r , then M is a sphere.
[Baikoussis–Koufogiorgos 1987a]. Let a function F : R×R → R be given

which is increasing in both variables and strictly increasing in
at least one of its variables. The spheres are the only ovaloids
K
satisfying F ( HK , KII ) = 0.
[Baikoussis–Koufogiorgos 1988b]. Let a function F : R+ +

0 × R0 → R be
given such that
∂F ∂F F
0¶ and ¶ (on R+ +
0 × R0 ),
∂u ∂v 2v
and x 7→ F (x, x 2 ) is monotonic. The spheres are the only oval-
oids satisfying KII = F (H, K).
[Baikoussis–Koufogiorgos 1988b]. Let a function F : R+ +

0 × R0 → R be
given such that
∂F F
0¶ ¶ (on R+ +
0 × R0 ),
∂u u
and x 7→ F (x, x 2 ) is monotonic. The spheres are the only oval-
oids satisfying KII = F (H, K).
Theorem 2.19 on p. 42. The spheres are the only ovaloids satisfying HII =
p
C K for some C ∈ R.
Theorem 2.21 on p. 43. The spheres are the only ovaloids satisfying HII =
C KII for some C ∈ R.
• Characterisations of Spheres in E3 by Means of Curvature Inequali-

ties.—The fact that some curvature invariants are signed, characterises the
spheres.
187
p
[Koutroufiotis 1974]. On an ovaloid M ⊆ E3 either KII − K changes sign
or M is a sphere.
[Singley 1975]. The spheres are the only ovaloids satisfying KII − H ¾ 0.
[Stamou 1981]. On an ovaloid M ⊆ E3 either HII − H changes sign or M

is a sphere.
[Manhart 1989a, Stamou 2003]. On an ovaloid M ⊆ E3 either KII − HII

changes sign or M is a sphere.
6.10.2. Characterisations of Hyperspheres in Em+1 .

Throughout this paragraph it will be assumed that m ¾ 2, and we denote
αm for the total (m-dimensional) area of the unit hypersphere Sm (1).
Liebmann’s theorem (1899). (38 ) The hyperspheres are the only com-
pact hypersurfaces in Em+1 with constant Gauss-Kronecker cur-
vature.
[Süss 1929b, Hsiung 1954]. The hyperspheres are the only hyperoval-
oids M ⊆ Em+1 with constant s-th mean curvature (1 ¶ s ¶
m).
See also the other articles of W. Süss for related characterisa-

tions of the hyperspheres among the hyperovaloids. Also [Hsiung 1954,
Voss 1956, Chern 1959], et al, are of interest. A significant result of A. Ros
states that the convexity condition in the last-mentioned theorem can be
omitted [Ros 1987].
Let us now focus on characterisations of the hypersphere, in
which the curvatures of the second fundamental form occur.
[Gardner 1972a] Cor. 15. (39 ) A hyperovaloid M ⊆ Em+1 satisfies

Z Z
trI RicII dΩ ¾ S dΩ ,
with equality if and only if M is a hypersphere.

38
See [Liebmann 1899a] p. 53; [Liebmann 1900].
39
For m = 2 this immediately follows from Grove’s integral formula (6.5) on p. 180.
188
[Gardner 1972a]. Let M ⊆ Em+1 be a hyperovaloid. If the Ricci tensor of

the second fundamental form II coincides with the Ricci tensor
of the first fundamental form I, then M is a hypersphere.
[Schneider 1972]. The hyperspheres are the only hyperovaloids M ⊆

Em+1 the second fundamental form of which has constant sec-
tional curvature.
[Schneider 1972]. The hyperspheres are the only hyperovaloids M ⊆

Em+1 for which the scalar curvature SII of the second funda-
mental form satisfies
2
αm

m
SII ¶ m(m − 1) .
AreaII (M )
[Hasanis 1984]. Let a hyperovaloid M ⊆ Em+1 be given. There holds

Z
1
II gradII H, gradII det A dΩ ¾ 0
det A
and equality occurs if and only if M is a hypersphere.
[Hasanis 1984]. Let a hyperovaloid M ⊆ Em+1 be given. There holds

Z
II gradII trA← , gradII det A dΩ ¶ 0

and equality occurs if and only if M is a hypersphere.
[Baikoussis–Koufogiorgos 1987a]. Let M ⊆ Em+1 be a hyperovaloid. Sup-

pose that for any tangent plane σ of M there holds
2
AreaII (M )

m
KII (σ) ¶ K(σ)
αm
(where K and KII stand for the sectional curvature with respect
to the first and the second fundamental form, respectively).
Then M is a hypersphere.
[Manhart 1989b]. The hyperspheres are the only hyperovaloids in Em+1

which satisfy
2(m − 1)HII ¾ SII .
189
6.10.3. Characterisations of Extrinsic Spheres in Space Forms.

Since [Goddard 1977] formulated a conjecture on Riemannian complete
hypersurfaces in the de Sitter space, a number of articles on this subject
has appeared. The most well-known theorem is the result (found inde-
pendently by Ramanathan and Akutagawa40 in 1987) which characterises
the extrinsic spheres as the only compact, Riemannian surfaces in the three-
dimensional de Sitter space with constant mean curvature H (with H 2 < 1).
The large number of extensions of this theorem will not be

covered here41 . However, the characterisations of extrinsic spheres in S31
by means of the geometry of the second fundamental form will be described
(compare also our § 5.3.2 on p. 102).
[Aledo–Romero 2003]. Let M ⊆ S13 be a compact Riemannian surface

with Gaussian curvature K < 1 (i.e., II is positive-definite). The
following assertions are equivalent:
• K is constant;
• KII is constant;
• F (K, KII ) = 0 for a function F : R × R → R which is in-
creasing in both variables (strictly monotonic in at least
one of them);
• M is an extrinsic sphere.

K
with Gaussian curvature 0 < K < 1. Then either KII − p1−K
changes sign or M is an extrinsic sphere.
40
See also [Montiel 1988].
41
See, e.g., the following list of references, which includes some related characterisa-
tions of totally geodesic or minimal surfaces in which the norm of the second funda-
mental form plays a rôle: [Simons 1968, Lawson 1969, Nomizu–Smyth 1969, Chern–
do Carmo–Kobayashi 1970, Cheng–Yau 1977, Barbosa–do Carmo 1984, Walter 1985,
Akataguwa 1987, Ramanathan 1987, Baikoussis–Koufogiorgos 1988a, Barbosa–do Carmo–
Eschenburg 1988, Montiel 1988, Coghlan–Itokawa 1990, Alencar–Colares 1993, Rosenberg
1993, Alencar–do Carmo 1994, Barbosa–Colares 1995a, Barbosa–Colares 1995b, Zheng
1995a, Zheng 1996, Hou 1997, Li 1997, Vlachos 1997, Cheng–Ishikawa 1998, Aledo–
Alías–Romero 1999, Ximin 2001], et al.
190

with Gaussian curvature 0 < K < 1. If KII = C H s K r on M ,
where C, 0 ¶ s ¶ 1 and r ¶ 1 are constants, then M is an
extrinsic sphere.
We conclude with a result on the third fundamental form geometry.
[Liu–Simon–Verstraelen–Wang 1999]. Let M ⊆ (M , g) be a hypersur-

face in a complete connected semi-Riemannian manifold of
constant curvature C = ±1 and dimension m+1 (with m ¾ 3).
If III is a positive-definite Einstein metric on M , then M is an
open part of one of the following hypersurfaces:
• S1m (r) ⊆ Sm+1 (1) (with 0 < r < 1);

• S1k ( p12 ) × Sm−k ( p12 ) ⊆ Sm+1 (1) (with 0 < k < m);
• Sm (r) ⊆ Hn+1 (1) (with 0 < r);
• Hm (r) ⊆ Hn+1 (1) (with 1 < r);
• the horosphere Fm ⊆ Hm+1 (1).
6.10.4. Characterisations of Totally Umbilical Hypersurfaces of

semi-Riemannian Manifolds.
[Aledo–Haesen–Romero 2007]. (42 ) Let (M , g) be a compact Riemann-
ian surface in a connected three-dimensional time-oriented Lo-
rentzian manifold (M , g). Suppose that the second fundamen-
tal form II is positive-definite and that the Gaussian curvature
K of (M , g) is strictly positive. Let L be the difference tensor
between the Levi-Civita connections of the first and the second
fundamental form. There holds:
• M is totally umbilical if and only if trII L = 0.

• If KII is constant, Z = 0, and the sectional curvature of
any tangent plane to M in (M , g) does not depend on the
chosen plane, then M is totally umbilical.
42
Some of the described results are unpublished.
191
• If Z = 0, and KII = C K for some C ∈ R+

0 , then M is
totally umbilical.
(See also [Haesen 2007] and the current chapter 5.)
6.11. Miscellaneous Related Topics.
6.11.1. Convexity.
A compact hypersurface in Em+1 with a positive-definite second fundamen-

tal form bounds a convex domain. We refer to the following works for a
discussion of convexity of subsets in Riemannian geometry: [Bishop 1974,
Burago–Zalgaller 1977, Berger 1990, Gruber 1993].
6.11.2. Submanifolds in a Euclidean Space of Arbitrary Codi-

mension.
[do Carmo–Lima 1969]. (43 ) Let M ⊂ En+m be a compact, oriented sub-

manifold of dimension n. Suppose that the second fundamen-
tal form in the direction a p
T p M × T p M → R : (vp , w p ) 7→ 〈II(vp , w p ), a p 〉
(II stands for the vector-valued second fundamental form) is

semi-definite for for all p ∈ M and all vectors a p ∈ T p En+m .
Suppose that, for some p ∈ M , this form is definite in at least
one direction a p . Then M is the boundary of a convex body in
a linear subspace of dimension n + 1 in En+m .
[Gardner 1975]. Suppose a diffeomorphism Ψ between compact, oriented

submanifolds N , M ⊆ En+p preserves the mean curvature vec-
tors H in the sense that (for all q ∈ N ) H q and H ψ(q) are iden-
tical up to translation. Then Ψ is a translation.
43
See also [Gardner 1975].
192
6.11. MISCELLANEOUS RELATED TOPICS. 6.11. 193
6.11.3. Non-smooth Convex Surfaces.
The geometry of convex sets in E3 , and in particular of convex polyhedrons,

have been studied since antiquity. We mention only the following three
theorems, which may be compared with the previously stated results on
convex surfaces.
“Rigidity of the convex, compact polyhedron.” (44 ) A convex, compact

polyhedron in E3 is up to congruence determined by the faces
and the pairs of edges which are to be identified.
Theorem of Euler. The number of vertices v, the number of edges e, and

the number of faces f of a convex, compact polyhedron satisfy
v − e + f = 2.
Theorem of Steinitz. Suppose v, e and f are strictly positive integers sat-

isfying v − e + f = 2. There exists a convex, compact polyhe-
dron with v vertices, e edges and f faces.
Principally Russian geometers (Aleksandrov, Pogorelov, Topo-

nogov, Tychonov) have done considerable work on the geometry of non-
smooth surfaces45 . For example, Toponogov introduced the concept of
non-negative curvature for a metric space. The Cohn-Vossen rigidity the-
orem has been generalised for these non-smooth surfaces of non-negative
curvature46 .
Furthermore, results similar to Weyl’s embedding problem
have been obtained. In order to do so, a procedure in which convex sur-
faces are studied as limit of a sequence of convex polyhedra has been devel-
oped. This idea can be read already in [Minkowski 1903]. In this respect,
the following result (which is due to Blaschke) is of considerable impor-
tance.
44
This theorem was mentioned already by Euclid, although it appears that Cauchy pro-
vided the first proof of the theorem in 1812. See also [Liebmann 1900].
45
Compare, e.g., [Efimow 1957, Pogorelov 1973, Burago–Zalgaller 1992].
46
See also [Spivak 1979] Vol. 5, p. 310.
193
“The Blaschke Selection Theorem.” (47 ) Let the space of all compact, con-
vex subsets in En+1 be endowed with the Hausdorff metric.
Then each bounded sequence contains a convergent subse-
quence.
6.12. Ten Interesting Open Problems.
In conclusion of this work, ten open problems which are related with the
second fundamental form of a surface in the three-dimensional Euclidean
space, are stated.
1. Is an ovaloid in E3 determined up to congruence by the second fun-

damental form?
2. Is an ovaloid (free of totally umbilical open regions) determined up

to congruence by its shape operator?
3. Prove or disprove: “An ovaloid does not admit non-trivial infinitesimal

II-isometric deformations.”
4. Prove or disprove: “An ovaloid does not admit non-trivial infinitesimal

deformations satisfying δA = 0, unless it contains a totally umbilical
open region.”
5. Are the spheres the only critical points (among ovaloids in E3 ) of the
II-area functional under volume constraint?
6. Are the spheres the only critical points (among ovaloids in E3 ) of the
II-area functional under area constraint?
7. Give an explicit example of a II-minimal surface of revolution in E3 .
8. Which surfaces in E3 are II-minimal and minimal?
9. Which surfaces in E3 are II-minimal and affine-minimal?

The result originally appeared in [Blaschke 1916]. See [Eggleston 1977] Ch. 4 (or
47
[Busemann 1958] Ch. 2) as well.
194
6.12. TEN INTERESTING OPEN PROBLEMS. 6.12. 195
10. It would be interesting if the last condition in the statement of Reilly’s

“partial II-Bernstein theorem” (p. 167) could be removed or, at least,
weakened.
195
196
BIBLIOGRAPHY.
[Aeppli 1959] A. Aeppli, Einige Ähnlichkeits- und Symmetriesätze für differenzierbare

Flächen im Raum, Commentarii Mathematici Helvetici 33 (1959) 174–195.
[Akutagawa 1987] K. Akutagawa, On space-like hypersurfaces with constant mean curvature
in the de Sitter space, Mathematische Zeitschrift 196 (1987), 13–19.
[Aledo–Alías–Romero 1999] J.A. Aledo, L.J. Alías, A. Romero, Integral formulas for com-
pact space-like hypersurfaces in de Sitter space: Application to the case of constant higher
order mean curvature, Journal of Geometry and Physics 31 (1999), 195–208.
[Aledo–Alías–Romero 2005] J.A. Aledo, L.J. Alías, A. Romero, A new proof of Liebmann
Classical Rigidity Theorem for Surfaces in Space Forms, Rocky Mountain Journal of
Mathematics 35 (2005) 6, 1811–1824.
[Aledo–Haesen–Romero 2007] J.A. Aledo, S. Haesen and A. Romero, Spacelike surfaces
with positive definite second fundamental form in 3-dimensional spacetimes, Journal of
Geometry and Physics 57 (2007), 913–923.
[Aledo–Romero 2003] J.A. Aledo and A. Romero, Compact spacelike surfaces in 3-
dimensional de Sitter space with non-degenerate second fundamental form, Differental
Geometry and its Applications 19 (2003), 97–111.
[Aleksandrov 1939] A.D. Aleksandrov, Sur les théorèmes d’unicité pour les surfaces fermées,
Comptes Rendus (Doklady) URSS 22 (1939), 99–102.
[Aleksandrov 1948] A.D. Aleksandrov, Intrinsic Geometry of Surfaces, (Russian) 1948; Ger-
man (Die innere Geometrie der konvexen Flächen, Akademie Verlag, 1955); English
(Intrinsic Geometry of Surfaces, Translations of Mathematical Monographs Volume
Fifteen, Providence, Rhode Island 1967.)
[Aleksandrov 1962] A.D. Aleksandrov, Uniqueness theorems for Surfaces in the Large, I–V,
American Mathematical Society Translations 21 (1962) 2, 341–416.
[Alencar–Colares 1993] H. Alencar, M. do Carmo and A.G. Colares, Stable hypersurfaces
with constant scalar curvature, Mathematische Zeitschrift 213 (1993) 1, 117–131.
[Alencar–do Carmo 1994] H. Alencar and M. do Carmo, Hypersurfaces with Constant Mean
Curvature in Spheres, Proceedings of the American Mathematical Society 120 (1994)
4, 1223–1229.
[Alías–Romero–Sánchez 1995] L.J. Alías, A. Romero and M. Sánchez, Uniqueness of com-
plete spacelike hypersurfaces of constant mean curvature in generalized Robertson-
Walker spacetimes, General Relativity and Gravitation 27 (1995) 1, 71–84.
197
198 BIBLIOGRAPHY.
[Alías–Romero–Sánchez 1997a] L.J. Alías, A. Romero and M. Sánchez, Spacelike hyper-

surfaces of constant mean curvature in certain spacetimes, Proceedings of the Second
World Congress of Nonlinear Analysts, Part 1 (Athens, 1996), 655–661.
[Alías–Romero–Sánchez 1997b] L.J. Alías, A. Romero and M. Sánchez, Spacelike hyper-
surfaces of constant mean curvature and Calabi-Bernstein type problems, The Tohoku
Mathematical Journal. Second Series 49 (1997) 3, 337–345.
[Altın 2007] A. Altın, On the inner curvature of the second fundamental form of ruled sur-
faces in 3-dimensional Minkowski space, International Journal of Mathematics and
Statistics 1 (2007), 23–30.
[Amur 1971] K. Amur, On a Characterization of the 2-sphere, The American Mathematical
Monthly 78 (1971) 4, 382–383.
[Arroyo–Garay–Mencía 2003] J. Arroyo, O.J. Garay and J.J. Mencía, Closed generalized
elastic curves in S2 (1), Journal of Geometry and Physics 48 (2003), 339–353.
[Baikoussis–Defever–Koufogiorgos–Verstraelen 1996] C. Baikoussis, F. Defever, T. Koufo-
giorgos and L. Verstraelen, Bonnet’s Theorem in Lorentzian Space Forms, Rendiconti
del Seminario Matematico di Messina. Serie II 19 (1996/7) 4, 1–10.
[Baikoussis–Koufogiorgos 1987a] C. Baikoussis and T. Koufogiorgos, On convex hypersur-
faces in Euclidean space, Archiv der Mathematik 49 (Feb 1987) 4, 337–343.
[Baikoussis–Koufogiorgos 1987b] C. Baikoussis and T. Koufogiorgos, Integral Inequalities
for Ovaloids in Euclidean Space, Mathematika 34 (1987), 77–81.
[Baikoussis–Koufogiorgos 1988a] C. Baikoussis and T. Koufogiorgos, On convex hypersur-
faces in a space form, Monatshefte für Mathematik 106 (1988) 257–264.
[Baikoussis–Koufogiorgos 1988b] C. Baikoussis and T. Koufogiorgos, Ovaloids with pre-
scribed curvature of the second fundamental form, Colloquium Mathematicum 55
(1988) 2, 245–248.
[Baikoussis–Koufogiorgos 1997] C. Baikoussis and T. Koufogiorgos, On the inner curvature
of the second fundamental form of helicoidal surfaces, Archiv der Mathematik 68 (Feb
1997) 2, 169–173.
[Barbosa–Colares 1995a] J. Barbosa and A.G. Colares, Soap bubbles in space forms, IX
School of Differential Geometry (Vitória, 1994). Matemática Contemporânea 9
(1995), 23–40.
[Barbosa–Colares 1995b] J. Barbosa and A.G. Colares, Spheres are the only r-stable con-
stant r-mean curvature compact hypersurfaces in a space form, Anais da Academia
Brasileira de Ciências 67 (1995) 2, 147–151.
[Barbosa–do Carmo 1984] J. Barbosa and M. do Carmo, Stability of hypersurfaces with
constant mean curvature, Mathematische Zeitschrift 185 (1984) 3, 339–353.
[Barbosa–do Carmo–Eschenburg 1988] J. Barbosa, M. do Carmo and J.H. Eschenburg,
Stability of hypersurfaces of constant mean curvature in Riemannian manifolds, Mathe-
matische Zeitschrift 197 (1988) 1, 123–138.
[Becker–Kühnel 1996] M. Becker and W. Kühnel, Hypersurfaces with constant inner curva-
ture of the second fundamental form, and the non-rigidity of the sphere, Mathematische
Zeitschrift 223 (1996) 4, 693–708.
[Berger 1970] M. Berger, Quelques formules de variation pour une structure Riemannienne,
Annales Scientifiques de l’École Normale Supérieure. Quatrième Série 3 (1970), 285–
294.
[Berger 1990] M. Berger, Convexity, The American Mathematical Monthly 97 (1990) 8
(Special Geometry Issue), 650–678.
198
BIBLIOGRAPHY. 199
[Besse 1987] A. Besse, Einstein Manifolds, Springer-Verlag, Berlin 1987.

[Bianchi 1894] L. Bianchi, Lezioni di geometria differenziale I, Spoerri, Pisa, 1894.
[Bianchi 1922] L. Bianchi, Lezioni di geometria differenziale II, Spoerri, Pisa, 1922.
[Bishop 1974] R.L. Bishop, Infinitesimal convexity implies local convexity, Indiana Univer-
sity Mathematics Journal 24 (1974/75), 168–172.
[Blair–Koufogiorgos 1992] D.E. Blair and T. Koufogiorgos, Ruled surfaces with vanishing
second Gaussian curvature, Monatshefte für Mathematik, 113 (1992) 3, 177–181.
[Blair 2000] D.E. Blair, Spaces of metrics and curvature functionals, in: F.J.E. Dillen and
L.C.A. Verstraelen (editors), Handbook of Differential Geometry, Vol. I, 153–185,
North-Holland, Amsterdam 2000.
[Blaschke 1912] W. Blaschke, Ein Beweis für die Unverbiegbarkeit geschlossener kon-
vexer Flächen, Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen,
Mathematisch-Physikalische Klasse (1912), 607–610.
[Blaschke 1916] W. Blaschke, Kreis und Kügel, Verlag, Leipzig 1916.
[Blaschke 1921] W. Blaschke, Ueber affine Geometrie XXIX: Die Starrheit der Eiflächen,
Mathematische Zeitschrift 9 (1921), 142–146.
[Blaschke 1923] W. Blaschke, Vorlesungen über Differentialgeometrie. I, Elementare Differ-
entialgeometrie, Springer, Berlin 1923.
[Blaschke–Leichtweiss 1973] W. Blaschke and K. Leichtweiss, Elementare Differentialge-
ometrie, Die Grundlehren der mathematischen Wissenschaften, Band 1. Springer-
Verlag, Berlin-New York, 1973.
[Bonnet 1853] P.O. Bonnet, Sur une propriété de maximum relative à la sphère, Nouvelles
Annales de Mathématiques 12 (1853), 433–453.
[Bonnet 1867a] P.O. Bonnet, Mémoire sur la théorie des surfaces applicables sur une surface
donnée, Journal de l’Ecole Polytechnique 25 (1867), 31–151.
[Bonnet 1867b] P.O. Bonnet, Mémoire sur la théorie des surfaces applicables, Journal de
l’Ecole Polytechnique 42 (1867), 72–92.
[Borisenko 2005] A.A. Borisenko, Compact Spacelike Surfaces in the 3-dimensional de Sitter
Space, arXiv:math/0511292 .
[Brasil–Colares–Palmas 2003] A. Brasil, A.G. Colares and O. Palmas, Complete spacelike
hypersurfaces with constant mean curvature in the de Sitter space: A gap theorem,
Illinois Journal of Mathematics 47 (2003) 3, 847–866.
[Bryant 2001] R. Bryant, On surfaces with prescribed shape operator, Results in Mathemat-
ics 40 (2001) 1-4, 88–121.
[Burago–Shefel’ 1992] Y.D. Burago and S.Z. Shefel’, The Geometry of Surfaces in Euclidean
Spaces, Chapter I of: Y.D. Burago and V.A. Zalgaller (Eds.), Geometry III, Encyclopae-
dia of Mathematical Sciences, Volume 48. Springer-Verlag, Berlin 1992.
[Burago–Zalgaller 1977] Y.D. Burago and V.A. Zalgaller, Convex Sets in Riemannian Spaces
of Non-Negative Curvature, Russian Mathematical Surveys 32 (1977) 3, 1–57.
[Busemann 1958] H. Busemann, Convex Surfaces, Interscience Tracts in Pure and Applied
Mathematics Number 6, Interscience Publishers, New York, 1958.
[Cai–Xu 2006] K. Cai and H. Xu, Spacelike hypersurfaces in de Sitter space with constant
higher-order mean curvature, International Journal of Mathematics and Mathematical
Sciences (2006), 1–6.
199
200 BIBLIOGRAPHY.
[Cartan 1942] E. Cartan, Sur les couples de surfaces applicables avec conservation des cour-
bières principles, Bulletin des Sciences Mathématiques (2e série) 66 (1942), 55–85.
[Cartan 1943] E. Cartan, Les surfaces qui admettent une seconde forme fondamentale don-
née, Bulletin des Sciences Mathématiques (2e série) 67 (1943), 8–32.
[Cauchy 1813] A. Cauchy, Sur les polygones et les polyèdres, Journal de l’École Polytech-
nique 16 (1813), 87–98.
[Chen 1973] B.-Y. Chen, On a variational problem of hypersurfaces, Journal of the London
Mathematical Society. Second Series 6 (1973), 321–325.
[Chen–Vanhecke 1981] B.-Y. Chen and L. Vanhecke, Differential Geometry of Geodesic
Spheres, Journal für die Reine und Angewändte Mathematik (Crelle’s Journal) 325
(1981), 28–67.
[Chen–Yano 1978] B.-Y. Chen and K. Yano, On the theory of normal variations, Journal of
Differential Geometry 13 (1978) 1, 1–10.
[Chen 1991] Fuliang Chen, On infinitesimal II-isometry of convex surfaces (Chinese), Jour-
nal of Jiangxi Normal University. Natural Sciences Edition 15 (1991) 1, 16–21.
[Cheng–Ishikawa 1998] Q.M. Cheng and S. Ishikawa, Spacelike hypersurfaces with constant
scalar curvature, manuscripta mathematica 95 (1998), 499–505.
[Cheng–Yau 1977] S.Y. Cheng and S.T. Yau, Hypersurfaces with constant scalar curvature,
Mathematische Annalen 225 (1997) 3, 195–204.
[Chern 1945] S.S. Chern, Some new characterizations of the Euclidean sphere, Duke Math-
ematical Journal 12 (1945), 279–290.
[Chern 1959] S.S. Chern, Integral Formulas for hypersurfaces in Euclidean space and their
applications to uniqueness theorems, Journal of Mathematics and Mechanics 8 (1959)
6, 947–955.
[Chern 1967] S.S. Chern, Curves and Surfaces in Euclidean Space, in: S.S. Chern (ed.),
Studies in Global Geometry and Analysis, Studies in Mathematics 4, The Mathematical
Association of America, Englewood Cliffs, N.J., Prentice-Hall, 1967.
[Chern 1985] S.S. Chern, Deformation of surfaces preserving principal curvatures. In: I.
Chavel and H. M. Farkas, Differential geometry and complex analysis (A volume dedi-
cated to the memory of Harry Ernest Rauch), Springer, Berlin 1985.
[Chern–do Carmo–Kobayashi 1970] S.S. Chern, M. do Carmo and S. Kobayashi, Minimal
submanifolds of a Sphere with Second Fundamental Form of Constant Length, pp. 59–75
in: F. Browder (editor), Functional Analysis and Related Fields, Springer-Verlag, Berlin
1970.
[Christoffel 1865] E.B. Christoffel, Über die Bestimmung der Gestalt einer krummen Ober-
fläche durch lokale Messungen auf derselben, Journal für die reine und angewandte
Mathematik (Crelle’s Journal) 64 (1865), 193–209.
[Christoffel 1867] E.B. Christoffel, Über einige allgemeine Eigenschaften der Minimums-
flächen, Journal für die reine und angewandte Mathematik (Crelle’s Journal) 67
(1867), 218–228.
[Coghlan–Itokawa 1990] L. Coghlan and Y. Itokawa, On the Sectional Curvature of Compact
Hypersurfaces, Proceedings of the American Mathematical Society 109 (1990) 1, 215–
221.
[Cohn-Vossen 1927] S. Cohn-Vossen, Zwei Sätze über die Starrheit der Eiflächen,
Nachrichten der Akademie der Wissenschaften in Göttingen. Mathematisch-
Physikalische Klasse 1 (1927), 125–134.
200
BIBLIOGRAPHY. 201
[Cohn-Vossen 1930] S. Cohn-Vossen, Unstarre geschlossene Flächen, Mathematische An-

nalen 102 (1930) 1, 10–29.
[Colesanti 2007] A. Colesanti, From the Brunn-Minkowski inequality to a class of Poincaré
inequalities, arXiv:math/070358 .
[Darboux 1915] G. Darboux, Leçons sur la théorie générale des surfaces, Gauthiers-Villars,
Paris 1915.
[Dillen–Sodsiri 2005a] F. Dillen and W. Sodsiri, Ruled surfaces of Weingarten type in
Minkowski 3-space, Journal of Geometry 83 (2005), 10–21.
[Dillen–Sodsiri 2005b] F. Dillen and W. Sodsiri, Ruled surfaces of Weingarten type in
Minkowski 3-space. II, Journal of Geometry 84 (2005), 34–44.
[do Carmo–Lima 1969] M. do Carmo and E. Lima, Isometric immersions with semi-definite
second fundamental forms, Archiv der Mathematik 20 (1969), 173–175.
[Dorfmeister–Inoguchi–Toda 2002] J. Dorfmeister, J. Inoguchi and M. Toda, Weierstraß-
type representation of timelike surfaces with constant mean curvature, Differential Ge-
ometry and Integrable Systems, Contemporary Mathematics AMS 308 (2002).
[Duschek 1926] A. Duschek, Über relative Flächentheorie, Sitzungsberichte der Akademie
der Wissenschaften in Wien. Mathematisch-naturwissenschaftliche Klasse. Abteilung
2 A 135 (1926), 1–8.
[Duschek 1927] A. Duschek, Über relative Flächentheorie. II, Sitzungsberichte der
Akademie der Wissenschaften in Wien. Mathematisch-naturwissenschaftliche Klasse.
Abteilung 2 A 136 (1927), 265–270.
[Duschek 1929] A. Duschek, Die Starrheit der Eiflächen, Monatshefte für Mathematik und
Physik 36 (1929), 191–134.
[Efimow 1957] N.W. Efimow, Flächenverbiegung im Großen, mit einem Nachtrag von E.
Rembs und K.P. Grotemeyer, Akademie-Verlag, Berlin 1957.
[Eggleston 1977] H.G. Eggleston, Convexity, Cambridge Tracts in Mathematics and Math-
ematical Physics 47, Cambridge University Press, Cambridge 1977.
[Eisenhart 1909] L.P. Eisenhart, A Treatise on the Differential Geometry of Curves and Sur-
faces, Ginn and Company, Boston 1909.
[Eisenhart 1926] L.P. Eisenhart, Riemannian geometry, Princeton University Press, Prince-
ton (N.J.) 1926.
[Elsgolc 1962] L.E. Elsgolc, Calculus of variations, London, Pergamon Press/Reading,
Mass., Addison-Wesley Publishing Co. 1962.
[Erard 1968] P.J. Erard, Über die zweite Fundamentalform von Flächen im Raum, Abhand-
lung zur Erlangung der Würde eines Doktors der Mathematik der Eidgenössischen
Technischen Hochschule Zürich. Diss. No. 4234, ETH Zürich, 1968.
[Feeman–Hsiung 1959] G. F. Feeman and C.C. Hsiung, Characterizations of Riemannian
n-spheres, American Journal of Mathematics 81 (1959), 691–708.
[Ferapontov 2001] E.V. Ferapontov, Surfaces in 3-space possessing nontrivial deformations
which preserve the shape operator, arXiv:math/0107122 .
[Finikoff 1933] S. P. Finikoff, Surfaces dont les lignes de courbure se correspondent avec égal-
ité des rayons de courbure principaux homologues, Comptes Rendus de l’Académie des
Sciences (Paris) 197 (1933), 984–986.
[Finikoff–Gambier 1933] S.P. Finikoff et B. Gambier, Surfaces dont les lignes de courbure se
correspondent avec égalité des rayons de courbure principaux, Annales Scientifiques de
l’Ecole Normale Supérieure, III. Ser. 50 (1933), 319–370.
201
202 BIBLIOGRAPHY.
[Frank 1969] H. Frank, Invarianten isometrischer Flächen in Räumen konstanter

Krümmung, Wissenschaftliche Zeitschrift der Humboldt-Universität zu Berlin.
Mathematisch-Naturwissenschaftliche Reihe 18 (1969), 805–811.
[Galvez–Martinez 2000] J.A. Galvez and A. Martinez, The Gauss map and Second Funda-
mental From of Surfaces in R3 , Geometriae Dedicata 81 (2000), 181–192.
[Gardner 1972a] R.B. Gardner, Subscalar pairs of metrics and hypersurfaces with a nonede-
generate second fundamental form, Journal of Differential Geometry 6 (1972), 437–
458.
[Gardner 1972b] R. Gardner, Rigidity Theorem for Convex Hypersurfaces, Notices of the
American Mathematical Society, 19 (1972) 1, A195–A195.
[Gardner 1975] R.B. Gardner, Geometry of Submanifolds of Euclidean Spaces, Proceedings
of Symposia in Pure Mathematics, Vol XXVII (1973) part I, 125–134. American Math-
ematical Society, Providence, R.I., 1975.
[Gardner 2002] R.B. Gardner, The Brunn-Minkowski inequality, Bulletin of the American
Mathematical Society 39 (2002) 3, 355–405.
[Glässner 1972] E. Glässner, Über die Minimalflächen der zweiten Fundamentalform, PhD.
Dissertation, Stuttgart, 1972.
[Glässner 1974] E. Glässner, Über die Minimalflächen der zweiten Fundamentalform,
Monatshefte für Mathematik 78 (1974), 193–214.
[Glässner–Simon 1973] E. Glässner and U. Simon, Zur Geometrie der zweiten Grundform,
Überblicke Mathematik 6 (1973), 197–235.
[Goddard 1977] A.J. Goddard, Some remarks on the existence of spacelike hypersurfaces of
constant mean curvature, Mathematical Proceedings of the Cambridge Philosophical
Society 82 (1977) 3, 489–495.
[Goemans–Van de Woestyne 2007] W. Goemans and I. Van de Woestyne, Translation Sur-
faces with Vanishing Second Gaussian Curvature in Euclidean and Minkowski 3-space,
in: F. Dillen and I. Van de Woestyne, Pure and Applied Differential Geometry (PADGE
2007), Shaker Verlag, Aachen 2007.
[Goldstein–Ryan 1975] R.A. Goldstein and P.J. Ryan, Infinitesimal Rigidity of Submanifolds,
Journal of Differential Geometry 10 (1975), 49–60.
[Gray 1973] A. Gray, The Volume of a Small Geodesic Ball of a Riemannian Manifold, Michi-
gan Mathematical Journal 20 (1973), 329–344.
[Gray 2004] A. Gray, Tubes, Progress in Mathematics, 221. Birkhäuser Verlag, Basel 2004.
[Gray–Vanhecke 1979] A. Gray and L. Vanhecke, Riemannian Geometry as Determined by
the Volumes of Small Geodesic Balls, Acta Mathematica 142 (1979) 3-4, 157–198.
[Grove 1957] V.G. Grove, On closed convex Surfaces, Proceedings of the American Mathe-
matical Society 8 (1958), 382–388.
[Gruber 1993] P.M. Gruber, History of Convexity, in: P.M. Gruber and J.M. Wills, Handbook
of convex Geometry, Elsevier, Amsterdam, 1993.
[Hadamard 1897] J. Hadamard, Sur certaines propriétés des trajectoires en dynamique,
Journal de Mathématiques Pures et Appliquées 5 (1897) 3, 331–387.
[Hadamard 1903] J. Hadamard, Sur les surfaces à courbure positive, Bulletin de la Société
Mathématique de France 31 (1903), 300–301.
[Hadwiger 1955] H. Hadwiger, Altes und Neues über Konvexe Körper, Birkhäuser Verlag,
Basel 1955.
202
BIBLIOGRAPHY. 203
[Haesen 2007] S. Haesen, Some Characterizations of totally umbilical surfaces in threee-

dimensional warped product spaces, Accepted for publication, Monatshefte für Mathe-
matik 152 (2007), no. 4, 303–314.
[Haesen–Verpoort 0000] S. Haesen and S. Verpoort, The Mean Curvature of the Second
Fundamental Form of a Hypersurface, pre-print KULeuven 2007. arXiv:0709.2107 .
[Haesen–Verpoort–Verstraelen 0000] S. Haesen, S. Verpoort and L. Verstraelen, The Mean
Curvature of the Second Fundamental Form, Accepted for publication, Houston Journal
of Mathematics.
[Hamilton 1833] W.R. Hamilton, On a new method of investigating the relations of surfaces
to their normals, with results respecting the curvatures of ellipsoids, Dublin University
Review (July 1833), 583–584.
[Hartman 1978] P. Hartman, On Complete Hypersurfaces of Nonnegative Sectional Curva-
tures and Constant m-th Mean Curvature, Transactions of the American Mathematical
Society 245 (1978), 363–374.
[Hasanis 1980] T. Hasanis, A new characterization of the Sphere in R3 , Annales Polonici
Mathematici 38 (1980), 47–49.
[Hasanis 1982] T. Hasanis, Characterizations of the sphere by the curvature of the second
fundamental form, Colloquium Mathematicum 46 (1982), 41–44.
[Hasanis 1983] T. Hasanis, Submanifolds and the length of the second fundamental tensor,
Australian Mathematical Society. Journal. Series A. Pure Mathematics and Statistics48
34 (1983), no. 1, 1–6.
[Hasanis 1984] T. Hasanis, On convex hypersurfaces in En+1 , Colloquium Mathematicum
48 (1984) 1, 49–55.
[Hasanis 1988] T. Hasanis, Ovaloids with prescribed curvature of the second fundamental
form, Colloquium Mathematicum 55 (1988) 2, 245–248.
[Hasanis–Koutroufiotis 1984] T. Hasanis and D. Koutroufiotis, Applications of the Gauss
mapping for hypersurfaces of the sphere, Global Differential Geometry and Global Anal-
ysis 1984 (Berlin, 1984), 180–193.
[Hasegawa–Abe 1995] I. Hasegawa and T. Abe, A characterization for spheres, Third Inter-
national Conference on Differential Geometry and its Applications (Athens, 1994).
Tensor (N.S.) 56 (1995) 1, 75–78.
[Heinz 1962] E. Heinz, On Weyl’s embedding problem, Journal of Applied Mathematics and
Mechanics 11 (1962) 3, 421–454.
[Herglotz 1943] G. Herglotz, Über die Starrheit der Eiflächen, Abhandlungen aus dem
Mathematischen Seminar der Hansischen Universität Hamburg 15 (1943), 127–129.
[Hermann 1968] R. Hermann, Differential Geometry and the Calculus of Variations, Aca-
demic Press, New York 1968.
[Hicks 1965] N. Hicks, Linear perturbations of connexions, The Michigan Mathematical
Journal 12 (1965), 398–397.
[Hilbert 1901] D. Hilbert, Über Flachen von Constanter Gaussscher Krummung, Transac-
tions of the American Mathematical Society 2 (1901) 1, 87–99.
[Hilbert–Cohn-Vossen 1952] D. Hilbert and S. Cohn-Vossen, Geometry and the Imagination,
Chelsea Publishing Company 1952.
48
Continued as Journal of the Australian Mathematical Society.
203
204 BIBLIOGRAPHY.
[Hopf 1951] H. Hopf, Über Flächen mit einer Relation zwischen den Haupt-Krümmungen,
Mathematische Nachrichten 4 (1951), 232–249.
[Hopf 1983] H. Hopf, Differential Geometry in the Large, Lecture Notes in Mathematics
1000, Springer-Verlag, Berlin 1983.
[Hopf–Samuelson 1938] H. Hopf und H. Samelson, Zum Beweis des Kongruenzsatzes für
Eiflächen, Mathematische Zeitschrift 43 (1938) 1, 749–766.
[Hou 1997] Zhong Hua Hou, Hypersurfaces in a Sphere with Constant Mean Curvature,
Proceedings of the American Mathematical Society 125 (1997) 4, 1193–1196.
[Hsiang–Teng–Yu 1982] W.-Y. Hsiang, Z.-H. Teng and W.-C. Yu, Examples of constant mean
curvature immersions of the 3-sphere into Euclidean 4-space, Proceedings of the Na-
tional Academy of Sciences of the United States of America 79 (1982) 3931–3932.
[Hsiung 1954] C.C. Hsiung, Some integral formulas for closed hypersurfaces, Mathematica
Scandinavica 2 (1954), 286–294.
[Hsiung–Liu 1977] C.C. Hsiung and J.D. Liu, A Generalization of the Rigidity Theorem of
Cohn-Vossen, Journal of the London Mathematical Society (Second Series) 15 (1977),
557–565.
[Hsü 1960] Chin-shui Hsü, Generalization of Cohn-Vossen’s theorem, Proceedings of the
American Mathematical Society 11 (1960), 845–846.
[Huck et al 1973] H. Huck, R. Roitzsch, U. Simon, W. Vortisch, R. Walden, B. Wegner and
W. Wendland, Beweismethoden der Differentialgeometrie im Großen, Lecture Notes in
Mathematics 335, Springer-Verlag, Berlin 1973.
[Hurwitz 1902] A. Hurwitz, Sur quelques applications géometriques des séries de Fourier,
Annales scientifiques de l’École normale supérieure (Paris) 3 (1902) 19, 357–408.
[Inoguchi–Toda 2004] J. Inoguchi and M. Toda, Timelike Minimal Surfaces via Loop
Groups, Acta Applicandae Mathematicae 83 3 (II) (2004), 313–355.
[Jellett 1853] J.-H. Jellett, Sur la surface dont la courbure moyenne est constante, Journal
de Mathématiques Pures et Appliquées, Sér. I 18 (1853), 163-167.
[Kim–Yoon 2004] Y.H. Kim and D.W. Yoon, Classification of ruled surfaces in Minkowski
3-space, Journal of Geometry and Physics 49 (2004), 37–44.
[Kim–Yoon 2006] Nam Gil Kim and Dae Won Yoon, Mean curvature of non-degenerate sec-
ond fundamental form of ruled surfaces, Honam Mathematical Journal 28 (2006) 4,
549–558.
[Kim–Yoon 2007] Young Ho Kim and Dae Won Yoon, On non-developable ruled surfaces in
Lorentz–Minkowski 3-spaces, Taiwanese Journal of Mathematics 11 (2007) 1, 197–
214.
[Klingenberg 1978] W. Klingenberg, A Course on Differential Geometry, Graduate Texts in
Mathematics 51, Springer-Verlag, New York-Heidelberg, 1978.
[Klotz 1963a] T. Klotz (Milnor), Some uses of the second conformal structures on strictly
convex surfaces, Proceedings of the American Mathematical Society 14 (1963), 793–
799.
[Klotz 1963b] T. Klotz (Milnor), Some geometric consequences of conformal structure, Trans-
actions of the American Mathematical Society 108 (1963), 38–53.
[Klotz 1963c] T. Klotz (Milnor), Another conformal structure on immersed surfaces of nega-
tive curvature, Pacific Journal of Mathematics 13 (1963), 1281–1288.
[Klotz 1975] T. Klotz (Milnor), The curvature of αI + βII + γIII on a surface in a 3-manifold
of constant curvature, The Michigan Mathematical Journal 22 (1975) 3, 247–255.
204
BIBLIOGRAPHY. 205
[Klotz 1977] T. Klotz (Milnor), The curvatures of some skew fundamental forms. Proceed-
ings of the American Mathematical Society 62 (1977) 2, 323–329.
[Klotz 1980] T. Klotz (Milnor), Abstract Weingarten surfaces, Journal of Differential Geom-
etry 15 (1980) 3, 365–380.
[Klotz 1981] T. Klotz (Milnor), Codazzi Pairs on Surfaces. In: I. Bivens, J.-P. Bourguignon,
A. Derdziński, D. Ferus, O. Kowalski, T. Klotz (Milnor), V. Oliker, U. Simon, W.
Strübing, K. Voss, Discussion on Codazzi-tensors. In: Global differential geometry and
global analysis (Berlin, 1979), Lecture Notes in Mathematics 838, Springer, Berlin-
New York, 1981.
[Klotz 1983] T. Klotz (Milnor), Surfaces in Minkowski 3-space on which H and K are linearly
related, The Michigan Mathematical Journal 30 (1983), 309–315.
[Koh 1998] S.E. Koh, A characterization of round spheres, Proceedings of the American
[Koufogiorgos 1979] T. Koufogiorgos, Some characteristic properties of the sphere related
to the curvature of the second fundamental form, Revue Roumaine de Mathématiques
Pures et Appliquées 24 (1979) 4, 611–614.
[Koufogiorgos–Hasanis 1977] T. Koufogiorgos and T. Hasanis, A Characteristic Property of
the Sphere, Proceedings of the American Mathematical Society 67 (1977) 2, 303–305.
[Koutroufiotis 1973] T. Koutroufiotis, On a conjectured characterization of the Sphere,
Mathematische Annalen 205 (1973), 211–217.
[Koutroufiotis 1974] T. Koutroufiotis, Two Characteristic Properties of the Sphere, Proceed-
ings of the American Mathematical Society 44 (1974), 176–178.
[Kowalski–Vanhecke 1986] O. Kowalski and L. Vanhecke, A new formula for the shape op-
erator of a geodesic sphere and its applications, Mathematische Zeitschrift 192 (1986)
4, 613–625.
[Kühnel 1981] W. Kühnel, Zur inneren Krümmung der zweiten Grundfurm, Monatshefte für
Mathematik 91 (1981), 241–251.
[Kühnel 1988] W. Kühnel, Conformal Transformations betwwen Einstein Spaces, pp. 105–
146 in: R.S. Kulkarni and U. Pinkall (Editors), Conformal Geometry (Aspects of Math-
emtaics: E 12), Vieweg, Wiesbaden, 1988.
[Kühnel 2002] W. Kühnel, Differential geometry (Curves—surfaces—manifolds), Student
Mathematical Library, 16. American Mathematical Society, Providence, RI 2002.
[Kühnel–Steller 2005] W. Kühnel and M. Steller, On Closed Weingarten Surfaces, Monats-
hefte für Mathematik 146 (2005), 113–126.
[Lafontaine 1982] J. Lafontaine, Courbure de Ricci et fonctionnelles critiques, Comptes Ren-
dus de l’Académie des Sciences. Série I. Mathématique 295 (1982) 12, 687–690.
[Lawson 1969] H.B. Lawson Jr., Local rigidity theorems for minimal hypersurfaces, Annals
of Mathematics 89 (1969), 187-197.
[Leichtweiß 1991] K. Leichtweiß, On the History of the Affine Surface Area for Convex Bod-
ies, Results in Mathematics 20 (1991), 650–656.
[Leichtweiß 1993] K. Leichtweiß, Convexity and differential geometry, in: P.M. Gruber and
J.M. Wills, Handbook of convex Geometry, Elsevier, Amsterdam, 1993.
[Levy 1926] H. Levy, Tensors determined by a hypersurface in Riemann space, Transactions
of the American Mathematical Society 28 (1926) 4, 671–694.
[Lewy 1938a] H. Lewy, On differential Geometry in the large, I (Minkowski’s problem),
Transactions of the Amerian Mathematical Society 43 (1938), 258–270.
205
206 BIBLIOGRAPHY.
[Lewy 1938b] H. Lewy, On the existence of a closed convex surface realizing a given Riemann-
ian metric, Proceedings of the National Academy of Sciences of the United States of
America 24 (2) (1938), 104–106.
[Li 1983] An-Min Li, Some integral Formulas for hypersurfaces and a generalization of the
Hilbert-Liebmann theorem, Proceedings of the American Mathematical Society 88
(1983) 2, 326–329.
[Li 1997] H. Li, Global Rigidity Theorems for hypersurfaces, Arkiv för Matematik 35 (1997),
327–351.
[Lichnerowicz 1961] A. Lichnerowicz, Propagateurs et commutateurs en relativité générale,
Institut des Hautes Études Scientifiques. Publications Mathématiques 10 (1961),
293–344.
[Liebmann 1899a] H. Liebmann, Eine neue Eigenschaft der Kugel, Gött. Nachr. (Kgl. Ges. d.
Wiss. Nachrichten. Math.-phys.) 1899, 44–55.
[Liebmann 1899b] H. Liebmann, Beweis zweier Sätze über die Bestimmung von Ovaloï-
den durch das Krümmungsmaß oder die mittlere Krümmung für jede Normalrichtung,
Nachrichten von der Königlicher Gesellschaft der Wissenschaften zu Göttingen 1899,
134–142.
[Liebmann 1900] H. Liebmann, Über die Verbiegung der geschlossenen Flächen positiver
Krümmung, Mathematische Annalen 53 (1900) 1-2, 81–112.
[Liebmann 1901] H. Liebmann, Neuer Beweis des Satzes, dass eine geschlossene convexe
Fläche sich nicht verbiegen lässt, Mathematische Annalen 54 (1901) 4, 505–517.
[Liebmann 1903] H. Liebmann, Neuer Beweis des Mindingschen Satzes, Jahresbericht der
Deutschen Mathematiker-Vereinigung (1903), 540–555.
[Liebmann 1919a] H. Liebmann, Die Verbiegung von geschlossenen und offenen Flächen pos-
itiver Krümmung, Münchener Berichte (1919), 267–291.
[Liebmann 1919b] H. Liebmann, Die Verbiegung analytischer Eiflächen, Mathematische
Zeitschrift 5 (1919) 1-2, 132–136. (1922).
[Liu–Simon–Verstraelen–Wang 1999] H.L. Liu, U. Simon, L. Verstraelen and C.P. Wang,
The third fundamental form metric for hypersurfaces in nonflat space forms, Journal of
Geometry 65 (1999), 130–142.
[Manhart 1982] F. Manhart, Die II-Minimalregelflächen, Anzeiger der Österreichische
Akademie der Wissenschaften. Mathematisch-Naturwissenschaftliche Klasse 119
(1982), 157–160.
[Manhart 1983] F. Manhart, Zur Differentialgeometrie der 2. Grundform, Second Austrian
geometry colloquium in Rein (Rein, 1983), Ber. No. 219, 7 pp.
[Manhart 1989a] F. Manhart, Relativgeometrische Kennzeichnungen euklidischer Hyper-
sphären, Geometriae Dedicata, 29 (1989) 2, 193–207.
[Manhart 1989b] F. Manhart, Kennzeichnungen Euklidischer Hypersphären durch
isoperimetrische Ungleichungen, Glasnik Matematički. Serija III 24 (44) (1989)
4, 541–555.
[Minding 1838a] F. Minding, Über die Biegung gewisser Flächen, Journal für die reine und
angewandte Mathematik (Crelle’s Journal) 18 (1838), 297–302.
[Minding 1838b] F. Minding, Über die Biegung krummer Flächen, Journal für die reine und
angewandte Mathematik (Crelle’s Journal) 18 (1838), 365–368.
[Minkowski 1901] H. Minkowski, Sur les surfaces convexes fermées, Comptes Rendus Heb-
domadaires des Séances de l’Académie des Sciences (Paris) 132 (1901) 21–24.
206
BIBLIOGRAPHY. 207
[Minkowski 1903] H. Minkowski, Volumen und Oberfläche, Mathematische Annalen 57

(1903), 447–495.
[Montiel 1988] S. Montiel, An integral inequality for compact spacelike hypersurfaces in de
Sitter space and applications to the case of constant mean curvature, Indiana University
Mathematics Journal 37 (1988), 909–917.
[Montiel–Ros 2005] S. Montiel and A. Ros, Curves and Surfaces, Graduate Studies in Math-
ematics Volume 69, American Mathematical Society and Real Sociedad Matemática
Española 2005.
[Müller 1921] E. Müller, Relative Minimalflächen, Monatshefte für Mathematik und Physik
31 (1921), 3–19.
[Münzner 1967] H.W. Münzner, Eine Bemerkung zum Kongruenzsatz von Grove, Archiv der
Mathematik 18 (1967), 525–528.
[Muto 1975a] Y. Muto, Critical Riemannian Metrics, Tensor N.S. 29 (1975), 125–133.
[Muto 1975b] Y. Muto, Curvature and critical Riemannian metric, Differential geometry
(Proc. Sympos. Pure Math., Vol. XXVII, Stanford Univ., Stanford, Calif., 1973), Part 1,
pp. 97–100. Amer. Math. Soc., Providence, R.I. 1975.
[Nagano 1967] T. Nagano, A problem on the existence of an Einstein metric, Journal of the
Mathematical Society of Japan 19 (1967), 30–31.
[Nakagawa–Yokote 1972] H. Nakagawa and I. Yokote, On Hypersurfaces with constant
scalar curvature in a Riemannian manifold of constant curvature, Kōdai Mathemati-
cal Seminar Reports 24 1972, 471–481.
[Nakajima 1926] S. Nakajima, Über charakteristische Eigenschaften der Kugel, Tohoku
Mathematical Journal 26 (1926), 361–364.
[Nirenberg 1953] L. Nirenberg, The Weyl and Minkowski problems in differential geometry
in the large, Communications on Pure and Applied Mathematics 6 (1953), 337–394.
[Nishikawa 2002] S. Nishikawa, Variational Problems in Geometry, Translations of Mathe-
matical Monographs 205, AMS, Providence, Rhode Island, 2002.
[Nitsche 1993] J.C.C. Nitsche, Boundary Value Problems for Variational Integrals Involving
Surface Curvatures, Quarterly of Applied Mathematics 11 (1993) 2, 363–387.
[Nomizu–Simon 1992] K. Nomizu and U. Simon, Notes on conjugate connections, pp. 152–
173 in: Geometry and topology of submanifolds, IV (Leuven, 1991), World Scientific
Publishing, River Edge, NJ, 1992.
[Nomizu–Smyth 1969] K. Nomizu and B. Smyth, A formula of Simons’ type and hyper-
surfaces with constant mean curvature, Journal of Differential Geometry 3 (1969),
367–377.
[Okumura 1970] M. Okumura, Sur l’hypersurface localement convexe dans une variété à
courbure constante, Comptes rendus de l’Académie des sciences (Paris) 270 (1970),
541–542.
[Okumura 1973] M. Okumura, Submanifolds and a Pinching Problem on the Second Fun-
damental Tensor, Transactions of the American Mathematical Society 178 (1973),
285–291.
[Okumura 1974] M. Okumura, Hypersurfaces and a Pinching Problem on the Second Fun-
damental Tensor, American Journal of Mathematics 96 (1974), 207–213.
[O’Neill 1966] B. O’Neill, Elementary Differential Geometry Academic Press, New York
1966.
207
208 BIBLIOGRAPHY.
[O’Neill 1983] B. O’Neill, Semi-Riemannian Geometry (with Applications to Relativity), Aca-

demic Press, New York 1983.
[Osserman 1978] R. Osserman, The Isoperimetric Inequality, Bulletin of the American
[Osserman 1990] R. Osserman, Curvature in the Eighties, The American Mathematical
Monthly 97 (1990) 8 (Special Geometry Issue), 731–756.
[Pogorelov 1961] A.V. Pogorelov, Topics in the theory of surfaces in elliptic space (With a
foreword by R. Sacksteder), Russian tracts on advanced mathematics and physics 1,
Gordon and Breach, New York 1961.
[Pogorelov 1973] A.V. Pogorelov, Extrinsic geometry of convex surfaces, Jerusalem: Israel
program for scientific translations, 1973.
[Pogorelov 1978] A.V. Pogorelov, The Minkowski multidimensional problem, Scripta series
in mathematics, Washington (D.C.): Winston 1978.
[Pogorelov 1988] A.V. Pogorelov, Bendings of surfaces and stability of shells, Translations
of mathematical monographs 72, Providence (R.I.): American Mathematical Society,
1988.
[Poincaré 1905] H. Poincaré, Sur les Lignes Géodésiques des Surfaces Convexes, Transactions
of the American Mathematical Society 6 (1905) 3, 237–274.
[Ramanathan 1987] J. Ramanathan, Complete Spacelike Hypersurfaces of Constant Mean
Curvature in the de Sitter Space, Indiana University Mathematics Journal 36 (1987),
343–359.
[Reich 1973] K. Reich, Die Geschichte der Differentialgeometrie von Gauß bis Riemann
(1828–1868), Archive for History of Exact Sciences 11 (1973) 4, 273–382.
[Reilly 1973] R.C. Reilly, Variational Properties of Functions of The Mean Curvatures for
Hypersurfaces in Space Forms, Journal of Differential Geometry 8 (1973), 465–477.
[Reilly 1976] R.C. Reilly, The Relative Differential Geometry of Nonparametric Hypersurfaces,
Duke Mathematical Journal 43 (1976) 4, 705–721.
[Reilly 1977] R.C. Reilly, Applications of the Hessian Operator in a Riemannian Manifold,
Indiana University Mathematics Journal 26 (1977) 3, 459–472.
[Reilly 1982] R.C. Reilly, Mean Curvature, The Laplacian, And Soap Bubbles, The American
Mathematical Monthly 89 (1982) 3, 180–188 and 197–198.
[Rembs 1952a] E. Rembs, Zur Verbiegung von Flächen im Großen, Mathematische
Zeitschrift 56 (1952), 271–279.
[Rembs 1952b] E. Rembs, Integralformeln der Verbiegungstheorie, Mathematische
Nachrichten 7 (1952), 61–64.
[Roitzsch 1971] R. Roitzsch, Die dritte Fundamentalform von Flächen und globale Kongruen-
zprobleme, Diplomarbeit FB 3, TU Berlin (1971).
[Ros 1987] A. Ros, Compact hypersurfaces with constant higher order mean curvatures, Re-
vista Matemática Iberoamericana 3 (1987) 3-4, 447–453.
[Ros 1988] A. Ros, Compact hypersurfaces with constant scalar curvature and a congruence
theorem, Journal of Differential Geometry 27 (1988), 215–220.
[Rosenberg 1993] H. Rosenberg, Hypersurfaces of constant curvature in space forms, Bul-
letin des Sciences Mathématiques 117 (1993) 2, 211–239.
[Rosso–Virga 1999] R. Rosso and E.G. Virga, Adhesive Borders of Lipid Membranes, Pro-
ceedings of the Royal Society A 455 (1999), 4145–4168.
208
BIBLIOGRAPHY. 209
[Rozet 1946a] O. Rozet, Sur certaines formes différentielles d’une surface. Bulletin de la So-
ciété Royale des Sciences de Liège 15 (1946), 117–119.
[Rozet 1946b] O. Rozet, Sur certaines formes différentielles d’une surface. II, Bulletin de la
Société Royale des Sciences de Liège 15 (1946), 188–191.
[Ruh–Vilms 1970] E. Ruh and J. Vilms, The tension field of the Gauss map, Transactions of
the American Mathematical Society 149 (1970), 567–573.
[Saban 1982] G. Saban, A Characteristic Property of the Sphere, Proceedings of the Ameri-
can Mathematical Society 86 (1982) 1, 123–125.
[Sacksteder 1962] R. Sacksteder, The Rigidity of Hypersurfaces, Journal of Mathematics
and Mechanics 11 (1962), 929–939.
[Scherrer 1940] W. Scherrer, Eine Kennzeichnung der Kugel, Vierteljahresschrift der natur-
forschenden Gesellschaft in Zurich, Beiblatt (Festschrift Rudolf Feuter) 89 (1940) 3,
40–463.
[Schneider 1967a] R. Schneider, Zur affinen Differentialgeometrie im Großen. I, Mathema-
tische Zeitschrift 101 (1967), 375–406.
[Schneider 1967b] R. Schneider, Zur affinen Differentialgeometrie im Großen. II. Über eine
Abschätzung der Pickschen Invariante auf Affinsphären, Mathematische Zeitschrift 102
(1967), 1–8.
[Schneider 1972] R. Schneider, Closed convex hypersurfaces with second fundamental form
of constant curvature, Proceedings of the American Mathematical Society 35 (1972)
230–233.
[Shiohama 1982] K. Shiohama, Convexity in Riemannian Geometry, in: S.S. Chern and Wu
Wen-Tsün, Proceedings of the 1980 Beijing Symposium on Differential Geometry and
Differential Equations, vol 3, Science Press, Beijing, China 1982.
[Simon 1967] U. Simon, Minkowskische Integralformeln und ihre Anwendungen in der Dif-
ferentialgeometrie im Großen, Mathematische Annalen 173 (1967), 307–321.
[Simon 1968] U. Simon, Kennzeichnungen von Sphären, Mathematische Annalen 175
(1968), 81–88.
[Simon 1971] U. Simon, II-Verbiegungen von Eiflächen, Archiv der Mathematik 23 (1971),
319–324.
[Simon 1972] U. Simon, On the inner geometry of the second fundamental form, The Michi-
gan Mathematical Journal 19 (1972), 129–132.
[Simon 1973] U. Simon, Characterization of Relative Spheres and Ellipsoids, Archiv der
Mathematik 24 (1973) 1, 100–104.
[Simon 1976] U. Simon, Characterizations of the sphere by the curvature of the second fun-
damental form, Proceedings of the American Mathematical Society 55 (1976), 382–
384.
[Simon–Schwenk-Schellschmidt–Viesel 1991] U. Simon, A. Schwenk-Schellschmidt and
H. Viesel, Introduction to the Affine Differential Geometry of Hypersurfaces, Lecture
Notes, Science University of Tokyo, 1991.
[Simon–Vrancken–Wang–Wiehe 2000] U. Simon, L. Vrancken, Changping Wang and M.
Wiehe, Intrinsic and extrinsic geometry of ovaloids and rigidity, in: W. H. Chen, C. P.
Wang, A.-M. Li, U. Simon, M. Wiehe and L. Verstraelen, Geometry and topology of
submanifolds, X, World Scientific Publishing, River Edge, N.J., 2000.
[Simon–Voss–Vrancken–Wiehe 2002] U. Simon, K. Voss, L. Vrancken and M. Wiehe, Sur-
faces with prescribed Weingarten operator, in: PDEs, submanifolds and affine differential
geometry (Warsaw, 2000), p. 171–178, Banach Center Publications 57, Polish Acad.
Sci., Warsaw, 2002.
209
210 BIBLIOGRAPHY.
[Simon–Weinstein 1972] U. Simon and A. Weinstein, Anwendungen der de Rhamschen Wer-

legung auf Probleme der lokalen Flächentheorie, Manuscripta Mathematica 1 (1969),
139–146.
[Simons 1968] J. Simons, Minimal varieties in Riemannian manifolds, Annals of Mathe-
matics 2 (1968) 88, 62–105.
[Singley 1975] D. Singley, Pairs of metrics on parallel hypersurfaces and ovaloids, Proceed-
ings of Symposia in Pure Mathematics 27, Part 1 (1975), 237–243.
[Sodsiri 2005] W. Sodsiri, Ruled Surfaces of Weingarten Type in Minkowski 3-space, PhD.
Thesis K.U.Leuven 2005.
[Spivak 1979] M. Spivak, A comprehensive introduction to differential geometry (second
ed.), Publish or Perish, Berkeley 1979.
[Stamou 1978] G. Stamou, Global Characterizations of the Sphere, Proceedings of the
American Mathematical Society 68 (1978) 3, 328–330.
[Stamou 1981] G. Stamou, Characterizations of the Sphere by the mean II Curvature, Bul-
letin of the Australian Mathematical Society 23 (1981) 2, 249–253.
[Stamou 1987] G. Stamou, Characterizations of the Sphere by the II-curvature and the mean
II-curvature, Revue Roumaine de Mathématiques Pures et Appliquées 32 (1987), 555–
559.
[Stamou 2002] G. Stamou, Kennzeichnungen Euklidischer Hypersphären im Rahmen der
relativen Hyperflächentheorie, Studia Scientiarum Mathematicarum Hungarica 39
(2002) 3-4, 377–384.
[Stamou 2003] G. Stamou, Kennzeichnungen Euklidischer Hypersphären durch relativge-
ometrische Grössen, Studia Scientiarum Mathematicarum Hungarica 40 (2003) 3,
349–358.
[Stamou–Magkos 2004] G. Stamou and A. Magkos, Regelflächen relativgeometrisch behan-
delt, Beiträge zur Algebra und Geometrie 45 (2004) 1, 209–215.
[Steiner 1840] J. Steiner, Ueber parallele Flächen, Monatsbericht der Akademie der
preussischen Wissenschaften zu Berlin (1840), 114–118.
[Struik 1933a] D.J. Struik, Outline of a History of Differential Geometry (I), Isis 19 (1933)
1, 92–120.
[Struik 1933b] D.J. Struik, Outline of a History of Differential Geometry (II), Isis 20 (1933)
1, 161–191.
[Struik 1950] D.J. Struik, Lectures on Classical Differential Geometry, Addison-Wesley Pub-
lishing Company, Inc., Reading, Massachusetts, USA 1950.
[Süss 1926] W. Süss, Eine elementare Eigenschaft der Kugel, the Tôhoku Mathematical Jour-
nal 26 (1926), 125–127.
[Süss 1927] W. Süss, Zur relativen Differentialgeometrie, III: Ueber Relativ-Minimalflächen
und Verbiegung, Japanese Journal of Mathematics 4 (1927), 203–207.
[Süss 1928a] W. Süss, Zur relativen Differentialgeometrie II: Relative Rotationsflächen, the
Tôhoku Mathematical Journal 28 (1929), 237–249.
[Süss 1928b] W. Süss, Zur relativen Differentialgeometrie IV: Ein Vierscheitelsatz bei
geschlossenen Raumkurven, the Tôhoku Mathematical Journal 28 (1929), 359–362.
[Süss 1929a] W. Süss, Kennzeichnende Eigenschaften der mehrdimensionalen Relativsphären
und Ellipsoide, the Tôhoku Mathematical Journal 29 (1929), 90–95.
[Süss 1929b] W. Süss, Zur relativen Differentialgeometrie V: über Eihyperflächen in Rn+1 ,
the Tôhoku Mathematical Journal 30 (1929), 202–209.
210
BIBLIOGRAPHY. 211
[Süss 1932] W. Süss, Ein Satz von P. Urysohn über mehrdimensionale Eikörper, the Tôhoku
Mathematical Journal 35 (1932), 326–328.
[Süss 1933] W. Süss, Bestimmung einer geschlossenen konvexen Flächen durch die Summe
ihrer Hauptkrümmungsradien, Mathematische Annalen 108 (1933) 1, 143–148.
[Süss 1937] W. Süss, Relative Minimalflächen und Parallelismus der affinen Flächentheorie,
the Tôhoku Mathematical Journal 43 (1937), 233–235.
[Süss 1948] W. Süss, Eine Kennzeichnung der Kugel, Archiv der Mathematik 1 (1948), 190–
191.
[Süss 1950] W. Süss, Bestimmung einer Fläche durch die dritte Grundform und die Summe
der Hauptkrümmungsradien, Archiv der Mathematik 2 (1950), 103–104.
[Süss 1953] W. Süss, Eine elementare kennzeichnende Eigenschaft des Ellipsoids,
Mathematisch-Physikalische Semesterberichte 3 (1953), 57–58.
[Süss 1957] W. Süss, Eindeutige Bestimmung von Eihyperflächen durch die Summe ihrer
Hauptkrümmungsradien, Archiv der Mathematik 8 (1957), 352–354.
[Švec 1975] A. Švec, Several new Characterizations of the Sphere, Czechoslovak Mathemat-
ical Journal 25 (100) (1975).
[Švec 1976] A. Švec, Several Characterizations of the Sphere, Periodica Mathematica Hun-
garica 7 (1976) 3-4, 313–317.
[Švec 1977] A Švec, Contributions to the Global Differential Geometry of Surfaces,
Rozpravy Československé Akad. Věd Řada Mat. Přírod. Rozpravy Československé
Akademie Věd Řada Matematických a Přírodních Věd 87 (1977), no. 1.
[Švec 1978] A. Švec Contributions to the Theory of Infinitesimal II-isometries, Mathematis-
che Nachrichten 83 (1978) 1, 73–82.
[Švec 1981] A. Švec, Global Differential Geometry of Surfaces, D. Reidel Publishing Com-
pany, Dordrecht 1981.
[Terng 1987] C.-L. Terng, Submanifolds with flat Normal Bundle, Mathematische Annalen
277 (1987), 95–111.
[Toda 2002] M. Toda, Weierstrass-type Representation of Weakly Regular Pseudospherical
Surfaces in Euclidean Space, Balkan Journal of Geometry and Its Applications 7 (2002)
2, 87–136.
[Toda 2005] M. Toda, Initial Value Problems of the Sine-Gordon Equations and Geometric
Solutions, arXiv:math/0307270 .
[Veeravalli 2000] A.R. Veeravalli, On a characterization of geodesic spheres, Indian Journal
of Pure and Applied Mathematics 31 (2000) 8, 979–982.
[Verpoort 0000a] S. Verpoort, The Mean Curvature of the Second Fundamental Form of an
Ovaloid, pre-print K.U.Leuven 2007. arXiv:0709.1644 .
[Verpoort 0000b] S. Verpoort, On the Inner Curvature of the Second Fundamental Form of a
Surface in the Hyperbolic Space, pre-print K.U.Leuven 2008. arXiv:0804.0271 .
[Vlachos 1997] T. Vlachos, A characterization for geodesic spheres in space forms, Geome-
triae Dedicata 68 (1997) 1, 73–78.
[Vortisch–Walden 1970] W. Vortisch und R. Walden, Ein Kongruenzsatz für Eiflächen,
Manuscripta Mathematica 3 (1970), 315–319.
[Voss 1956] K. Voss, Einige differentialgeometrische Kongruenzsätze für geschlossene Flächen
und Hyperflächen, Mathematische Annalen 131 (1956), 73.
211
212 BIBLIOGRAPHY.
[Voss 1970] K. Voss, Isometrie von Flächen bezüglich der zweiten Fundamentalform,
Nachrichten Der Österreichischen Mathematischen Gesellschaft, Sonderheft Nr. 91
(1970), 78–78.
[Voss 2001] K. Voss, On the shape operator of surfaces in space forms, Results in Mathemat-
ics 40 (2001) 1-4, 310–320.
[Walden 1971] R. Walden, Eindeutigkeitssätze für II-isometrische Eiflächen, Mathematische
Zeitschrift 120 (1971), 143–147.
[Walden–Wegner 1972] R. Walden and B. Wegner, Eine Bemerkung zur Integralformelmeth-
ode, Archiv der Mathematik (Basel) 23 (1972), 324–328.
[Walter 1985] R. Walter, Compact hypersurfaces with a constant higher mean curvature func-
tion, Mathematische Annalen 270 (1985) 1, 125–145.
[Weiner 1978] J. Weiner, On a problem of Chen, Willmore, et al, Indiana University Mathe-
matics Journal 27 (1978), no. 1, 19–35.
[Weingarten 1887] J. Weingarten 1887, Ueber die Deformationen einer biegsamen unaus-
dehnbaren Fläche, Journal für die Reine und Angewändte Mathematik (Crelle’s Jour-
nal) 100 (1887), 296–310.
[Wente 1986] H.C. Wente, Counterexample to a conjecture of H. Hopf, Pacific Journal of
Mathematics 121 (1986) 1, 193–243.
[Weyl 1916] H. Weyl, Über die Bestimmung einer geschlossenen konvexen Fläche durch ihr
Linienelement, Vierteljahresschrift der naturforschenden Gesellschaft 61 (1916), 40–
72.
[Weyl 1917] H. Weyl, Über die Starrheit der Eiflächen und konvexe Polyeder, Königlich
Preussische Akademie des Wissenschaften zu Berlin, Sitzungsberichte (1917), 250–
266.
[Weyl 1939] H. Weyl, On the Volume of Tubes, American Journal of Mathematics 61 (1939)
2, 461–472.
[Wiehe 1998] M. Wiehe, Deformations in Affine Hypersurface Theory, PhD. Thesis, TU
Berlin 1998; Shaker Verlag, Aachen 1999.
[Willmore–Jhaveri 1972] T.J. Willmore and C.S. Jhaveri, An extension of a result by Bang-
Yen Chen, The Quarterly Journal of Mathematics. Oxford. Second Series 23 (1972),
319–323.
[Wolf 1966] J.A. Wolf, Exotic metrics on immersed surfaces, Proceedings of the American
Mathematical Society 17 (1966), 871–877.
[Ximin 2001] L. Ximin, Space-like Hypersurfaces in the de Sitter Spaces, Journal of Mathe-
matical Physics 42 (2001) 8, 3965–3972.
[Yang 1989] Wenmao Yang, On infinitesimal Bonnet II-isometric deformations of surfaces in
E3 (Chinese), Journal of Mathematics. Wuhan University 9 (1989) 2, 217–228.
[Yano 1949] K. Yano, Sur la théorie des déformations infinitésimales, Journal of the Faculty
of Sciences of the University of Tokyo. Section 1 6 (1949), 1–75.
[Yano 1970] K. Yano, Integral Formulas in Riemannian geometry, Pure and applied mathe-
matics 1, Dekker, New York 1970.
[Yano 1978] K. Yano, Infinitesimal Variations of Submanifolds, Kodai Mathematical Journal
1 (1978), 30–44.
[Yoon 2006a] Dae Woon Yoon, Ruled surfaces whose mean curvature vector is an eigenvector
of the Laplacian of the second fundamental form, International Mathematical Forum.
Journal for Theory and Applications 1 (2006) 33-36, 1783–1788.
212
BIBLIOGRAPHY. 213
[Yoon 2006b] Dae Woon Yoon, Ruled surfaces in Lorentz-Minkowski 3-spaces with non-
degenerate second fundamental form, Acta Mathematica Academiae Paedagogicae
Nyíregyháziensis. New Series 22 (2006) 3, 275–284.
[Zheng 1996] Y. Zheng, Space-like hypersurfaces with constant scalar curvature in the de
Sitter spaces, Differential Geometry and its Applications 6 (1996) 1, 51–54.
[Zheng 1995a] Y. Zheng, On space-like hypersurfaces in the de Sitter space, Annals of Global
Analysis and Geometry 13 (1995), 317–321.
[Zhou 1987] Jia Zhu Zhou, On infinitesimal II-isometry of closed convex surfaces, Journal of
Mathematics. Shuxue Zazhi (Wuhan) 7 (1987) 2, 181–188.
[Zhou 1990] Jia Zhu Zhou, Some results of infinitesimal II-isometry of closed surfaces, Acta
Mathematica Sinica. New Series 6 (1990) 4, 327–333.
213
214 BIBLIOGRAPHY.
214
NOTATION INDEX.
(The numbers refer to the page where the corresponding symbol has been
introduced.)
End of a proof.
z End of an example.
End of a remark or a bibliographical note.
Chapter 1. Variational Formulae for Surfaces in the Three-Dimen-

sional Euclidean Space.
¦ ©
2 X(M ) = E3 -valued vector fields on a surface M .
¦ ©
3 F(M ) = real-valued functions on M .
¦ ©
3 X(M ) = vector fields on M .
3 I = first fundamental form.
×) = µ∗ B(µ ) .

3 B(µ t t t
5 δB = variation of the tensor B.
5 D = standard connection of E3 .
5 ei = principal direction.
5 λi = principal curvature.
215
216 NOTATION INDEX.
5 A = shape operator.
5 H = mean curvature.
5 K = Gaussian curvature.
6 L Z (B) = Lie derivative of B with respect to Z.
8 f ← = inverse of a bijection f : A → B.
8 ∇ = Levi-Civita connection.
t
10 vp = tangent part of the vector vp .
n
10 vp = normal part of the vector vp .
12 P = position vector field.
21 ψ = distance function to a surface.
Chapter 2. The Geometry of the Second Fundamental Form of an

Ovaloid in the Three-Dimensional Euclidean Space.
20 II = second fundamental form.
24 A sub- or superscript II refers to the geometry of the second

fundamental form.
e
24 Vi = pi (II-normalised principal directions).
λi
ei
24 Wi = λi
(III-normalised principal directions).
24 III = third fundamental form.
24 ∇II = connection of the second fundamental form.
24 ∇III = connection of the third fundamental form.

¦ ©
26 Λ(M ) = differential forms on M .
27 div B = divergence of a symmetric (1, 1)-tensor B.
29 KII = Gaussian curvature of the second fundamental form.
216
NOTATION INDEX. 217
33 HII = mean curvature of the second fundamental form.
47 τ + 1 = generalised fundamental form.
58 Y = rotation vector field of an infinitesimal isometric defor-

mation.
59 ω = Weingarten’s characteristic function.
Chapter 3. Variational Formulae for Abstract semi-Riemannian Mani-

folds.
62 ∇ = Levi-Civita connection of (M , g).
62 R = Riemann–Christoffel curvature tensor.
62 Ric = Ricci tensor.
62 Rc = Ricci operator.
62 S = scalar curvature.
63 δB = variation of the tensor B.
64 X = variation tensor of the connection.
64 ∆B = Laplacian of the tensor B.
67 Hs f = Hessian operator of f .
Chapter 4. The Variation of the Shape Operator of a Hypersurface in

a semi-Riemannian Manifold.
t n
(The definition of the previously introduced symbols ∇, vp , vp , A, I, II,
III, H and δA is extended to a hypersurface (M , g) ⊆ (M , g) on p. 70–72.)
70 A bar refers to the ambient space (M , g).
70 ∇ = Levi-Civita connection of (M , g).
71 K(σ) = sectional curvature of a plane σ tangent to (M , g).
217
218 NOTATION INDEX.
71 K(σ) = sectional curvature of a plane σ tangent to (M , g).
72 µ = (arbitrary) deformation.
73 ξ = unit normal deformation.
73 N = unit vector field, normal to µ t (M ) everywhere.
73 U = unit vector field, normal to ξ t (M ) everywhere.
76 ¹· , ·º = Lie bracket.
81 H2 = det A (Gauss–Kronecker curvature of a surface in a

three-dimensional semi-Riemannian manifold).
83 S31 = three-dimensional de Sitter space.
Chapter 5. The Geometry of the Second Fundamental Form of a Hy-

persurface in a semi-Riemannian Manifold.
¨ «
t
←
90 B(V, W ) = A R(V,N )W .
90 Z = trII B .
91 L = difference tensor between ∇ and ∇II .
96 ` =ZùL.
101 SII = scalar curvature of the second fundamental form.
102 H3 = hyperbolic space.
120 HII = mean curvature of the second fundamental form.

m+1
122 M 0 (C) = Riemannian space form of sectional curvature C.
m+1
122 M 1 (C) = Lorentzian space form of sectional curvature C.
129 {T, N } = Frenet frame field along a curve.
130 κ = geodesic curvature of a curve.
138 Gn (r) = geodesic hypersphere of radius r and centre n.
218
NOTATION INDEX. 219

139 Z(n) y = extension of a tensor Z(n) along a geodesic by means of

parallel translation.
140 x = normal co-ordinate system on (M , g).
140 Rı u v e = co-ordinate coefficients of the curvature tensor.
142 αm = area of the unit hypersphere Sm .
144 x = co-ordinate system on Gn (r).
Chapter 6. A Literature Overview on the Geometry of the Second

Fundamental Form and the Geometry of Ovaloids.
163 A( y) = relative shape operator of M ⊆ A3 with respect to a

relative normal field y.
1
163 H( y) = 2
tr A( y) (relative mean curvature).
163 H2 ( y) = det A( y) (relative Gauss–Kronecker curvature).
163 P = Peterson mapping.
164 ∇( y) = relative connection induced by a relative normal field

y.
164 h( y) = relative metric induced by a relative normal field y.
164 dΩ( y) = y ù Det.
164 G(Y ) = relative metric induced by a relative conormal field Y .
164 ∇(Y ) = relative connection induced by a relative conormal field

Y.
164 Sb(Y ) = symmetric bilinear form induced by a relative conormal
field Y .
165 {YII , NII } = Manhart’s II-normalisation.
219
220 NOTATION INDEX.
220
AUTHOR AND SUBJECT INDEX.
(This index does not refer to authors of articles which are mentioned in foot-
notes or the bibliography.)
affine space, 163 C. Carathéodory, 181

K. Akutagawa, 190 E. Cartan, 29, 103, 157, 160
J.A. Aledo, 98, 102, 103, 109, 116, 160, catenoid, 133
190–191 A.-L. Cauchy, 169
A.D. Aleksandrov, 174, 183, 193 B.-Y. Chen, 138–147, 155
L.J. Alías, 98, 102, 103 S.S. Chern, 18, 158, 183, 188
A. Altın, 162 E.B. Christoffel, 15, 17, 18, 38, 42, 43, 173,
K. Amur, 18 see also curvature tensor
anisotropic area, 165 Codazzi equation, 8, 71, 168
J. Arroyo, 131, 160, 168 second, 174
S. Cohn-Vossen, 66, 169, 171, 176–177, 193
J. Babinet, 144 congruence, 159
C. Baikoussis, 29, 83, 113, 160, 162, 183, infinitesimal, 58
187, 189 congruence theorems, 168–176
M. Becker, 30, 162, 178 conjugate directions, 24
Beez, 170 connection, 8, 62, see also second, third fun-
E. Beltrami, 34, 35 damental form
M. Berger, 192 Euclidean, 5
D. Bernoulli, 129 variation tensor of , 64
A. Besse, 68 convexity, 192
L. Bianchi, 24, 59, 160, 174 curvature, see also mean, Gaussian, Gauss–
R.L. Bishop, 192 Kronecker, geodesic, principal, scalar,
D. Blair, 67, 161 sectional
W. Blaschke, 17, 18, 158, 163, 177, 179, curvature tensor, 62, 98
180, 193 variation, 64
P.O. Bonnet, 13, 82, 168, 174 curvature vector of a curve, 20
Bonnet transformation, 83
branched surface, 83 G. Darboux, 157, 161
H. Brunn, 180 G. de Rahm, 98
Brunn–Minkowski inequality, 180 de Sitter space, 83, 122
R. Bryant, 175 F. Defever, 83
Y.D. Burago, 192 deformation, 1, 72
221
222 AUTHOR AND SUBJECT INDEX.
infinitesimal, 57–60 generalised fundamental form, 47

isometric, 58 geodesic curvature, 130
trivial, 58 geodesic hyperspheres, 138–155, see also sec-
linear, 6 ond fundamental form
normal, 2 Gauss–Kronecker curvature, 147
tangent, 2 mean curvature, 146
unit normal, 15, 73 mean curvature of the second funda-
deformation vector field, 2, 72 mental form, 144–152
deforming mapping, 1 second fundamental form, 149
difference tensor, 25, 91 shape operator, 148
squared norm, 27 geodesic umbrella, 89
symmetry, 25, 92 S. Germain, 144
trace, 25, 26, 93 E. Glässner, 31, 45, 157, 160, 168, 171, 172,
F. Dillen, 118, 160, 162 178
distance function, 21, 77 A.J. Goddard, 190
Hessian, 22, 77 W. Goemans, 162
divergence theorem, 10 A. Gray, 138–155
M. do Carmo, 192 K.P. Grotemeyer, 166
C. Dupin, 24, 144 V.G. Grove, 160, 171, 180
A. Duschek, 163 P.M. Gruber, 192
N.W. Efimow, 60, 158, 177 J. Hadamard, 159

H.G. Eggleston, 179 H. Hadwiger, 163, 179
Einstein space, 124 S. Haesen, 98, 102, 116, 160, 191
ellipsoid, 55 W.R. Hamilton, 13
embedding problem, see Minkowski and T. Hasanis, 110, 185, 186, 189
Weyl helicoid, 167
equidistant surface, 22, 73, 82 helicoidal surface, 162
P.J. Erard, 29, 103, 157, 160, 171–173, 177 R. Hermann, 155
Euclidean normalisation, 165 Hessian, 27, see also distance function
L. Euler, 129, 144, 169, 193 N. Hicks, 96, 160
extrinsic hypersphere, 89 D. Hilbert, 15, 18, 30, 103, 182–184
H. Hopf, 158, 179, 184
E.V. Ferapantov, 175 C.C. Hsiung, 169, 188
S.P. Finikoff, 175 C.-S. Hsü, 169
first fundamental form H. Huck, 158, 172, 178
variation, 76 hyperbolic space, 102, 122, 176
J.F. Frenet, 58, 129 hyperboloid, 54, 134
functional
variation, 12 infinitesimal deformations, 176–179
isoperimetric inequality, 179–181, see also
J.A. Galvez, 161, 167 second fundamental form
B. Gambier, 175 isotropic axis, 135
O.J. Garay, 131, 160, 168
R. Gardner, 96, 160, 172, 179, 188, 189, C.G.J. Jacobi, 93
192 Jacobi equation, 142
gauge ovaloid, 163 J.H. Jellett, 13, 15, 18, 35, 41, 51, 166, 180,
Gauss equation, 71, 168 183
Gauss map, 159–161, 173
Gauss–Kronecker curvature W. Killing, 170
variation, 81 N.G. Kim, 162
Gaussian curvature, 5 Y.H. Kim, 162
variation, 10, 81 W. Klingenberg, 18
222
AUTHOR AND SUBJECT INDEX. 223
T. Klotz (Milnor), 83, 103, 160, 167, 186 abstract, 66

Koszul formula, 64
T. Koufogiorgos, 29, 83, 113, 160–162, 183, Palatini equation, 68
185–187, 189 paraboloid, 54
T. Koutroufiotis, 40, 105, 107, 185, 188 parallel mapping, 170
O. Kowalski, 155 Peterson mapping, 163
T. Kubota, 166 A.V. Pogorelov, 176, 177, 193
W. Kühnel, 30, 68, 161, 162, 178 H. Poincaré, 180, 181
S.D. Poisson, 11, 144
J.-L. Lagrange, 169 position vector field, 12
Laplacian, 10, 64 Laplacian, 34
H.B. Lawson, 121, 124 principal curvature, 5
T. Levi-Civita, see connection variation, 10
H. Lewy, 66, 193 principal direction, 5
A.-M. Li, 46
A. Lichnerowicz, 68 J. Ramanathan, 190
S. Lie, 59, 172 K. Reich, 144
Lie bracket, 76 R.C. Reilly, 17, 167, 195
Lie derivative, 6 relative area, 165
H. Liebmann, 15, 18, 30, 59, 103, 174, 176, relative co-normal field, 164
183, 184, 188 relative differential geometry, 162–166
E. Lima, 192 relative Gauss equation, 164
H.L. Liu, 160, 191 relative Gauss–Kronecker curvature, 163
J.D. Liu, 169 relative mean curvature, 163
relative metric, 164
A. Magkos, 162 relative normal field, 163
F. Manhart, 39, 44, 162, 165, 168, 180, 188, relative normalisation, 164
189 relative shape operator, 163
A. Martinez, 161, 167 Ricci operator, 62
V.S. Matveev, 170 Ricci tensor, 62
mean curvature, 5, 71 variation, 65
variation, 10, 81 B. Riemann, see curvature tensor
variational characterisation, 31 rigidity, 169, 193
J.J. Mencía, 131, 160, 168 infinitesimal, 176
J.B. Meusnier, 20 Robertson–Walker space, 137
F. Minding, 169 R. Roitzsch, 178, see also H. Huck
H. Minkowski, 13, 17, 35, 41, 51, 163, 165, A. Romero, 98, 102, 103, 109, 116, 160,
166, 174, 179–180, 193, see also 190–191
Brunn A. Ros, 18, 158, 171, 179, 181, 188
Minkowski embedding problem, 175–176 R. Rosso, 17
Minkowski inequalities, 180 rotation vector field, 58
Minkowski space, 122, 134 O. Rozet, 160
S. Montiel, 18, 124, 158, 179 E. Ruh, 161
E. Müller, 163 ruled surface, 161, 162
S. Nakajima, 18, 184 scalar curvature, 62, 71

J.C.C. Nitsche, 17 variation, 65, 66
K. Nomizu, 165 W. Scherrer, 166
normal curvature function, 89 R. Schneider, 29, 104, 160, 171, 185, 189
F. Schur, 171
B. O’Neill, 25, 27 A. Schwenk-Schellschmidt, 163
R. Osserman, 17, 179–181 second fundamental form, 20, 70
ovaloid, 15, 23, 159 II-normalisation, 165
223
224 AUTHOR AND SUBJECT INDEX.
area, 41, 160 variation of a , 3, 63, 72

geodesic hyperspheres, 152–154 C.-L. Terng, 83
area element, 24 third fundamental form, 24, 70
Bernstein problem, 167 connection, 24, 160
conformal geometry, 167 Einstein metric, 191
connection, 24, 90, 160 mean curvature of , 48
curvature tensor of , 98 M. Toda, 161
extrinsic geometry of , 30–38, 118– P.J. Topalov, 170
121 V.A. Toponogov, 193
Gaussian curvature of , 28–30, 102, totally umbilical, 89
160 translation surfaces, 162
unit normal variation, 36 A.N. Tychonov, 193
geodesic hyperspheres w.r.t. , 160
geodesics, 160 I. Van de Woestyne, 162
geometrical interpretation, 20, 76 L. Vanhecke, 138–155
intrinsic geometry of , 23–30, 98– variation, 3, 5
117 variation-operator, 4
isoperimetric inequalities, 44–46, 180 L. Verstraelen, 83, 160, 191
Jellett–Minkowski formula, 41, 51 H. Viesel, 163
Laplacian, 28, 31, 34, 95, 162 J. Vilms, 161
mean curvature of , 30–34, 120, E.G. Virga, 17
160 W. Vortisch, 171, see also H. Huck
geodesic hyperspheres, 144–152 A. Voss, 177
unit normal variation, 36 K. Voss, 17, 170, 172, 175, 178, 179, 188
minimal surface, 30, 116 L. Vrancken, 170, 175, 179
Ricci tensor of , 99
scalar curvature of , 98–102, 160 R. Walden, 171, 178, see also H. Huck
variation, 72 C.P. Wang, 160, 170, 175, 191
sectional curvature, 71 B. Wegner, 178, see also H. Huck
variation, 81 J. Weingarten, 24, 59, 90, 157, 160
J.A. Serret, 129 W. Wendland, see H. Huck
shape operator, 5, 70 H.C. Wente, 184
divergence, 27 H. Weyl, 66, 175, 177, 182–183
variation, 9, 69, 72, 142 Weyl embedding problem, 66, 175–176, 193
similarity, 159 M. Wiehe, 168, 170, 175, 179
U. Simon, 45, 56, 96, 105, 157, 160, 163, W. Wirtinger, 180
165, 166, 170–172, 175, 177–179, J. Wolf, 160
183, 185, 191, see also H. Huck
D. Singley, 188 K. Yano, 68
S.L. Sobelev, 180 D.W. Yoon, 162
W. Sodsiri, 118, 160, 162
G. Stamou, 38–40, 43, 162, 185, 186, 188 V.A. Zalgaller, 192
E. Steinitz, 193
strictly convex, 23
Study’s parameters, 161
support function, 13
surface of revolution, 52–57, 133–135, 161–
162
W. Süss, 163, 166, 188
S. Tabachnikov, 170
B. Taylor, 139
tensor
224

The Geometry of The Second Fundamental Form: Curvature Properties and Variational Aspects

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Geometry of The Second Fundamental Form: Curvature Properties and Variational Aspects

Uploaded by

Copyright:

Available Formats

KATHOLIEKE UNIVERSITEIT LEUVEN

The Geometry of the

Proefschrift voorgedragen tot het

The Geometry of the

(De Meetkunde van de Tweede Fundamentaalvorm:

Proefschrift voorgedragen tot het

The broad subject of this dissertation can be described as “the geometry of

included. To this category the attempts towards a solution of the problems

Nederlandstalige Samenvatting (Dutch Summary). xv

1. Variational Formulae for Surfaces in the Three-Dimensional

1.1. Deformations of Surfaces. . . . . . . . . . . . . . . . . 1

1.2. Applications to Integral Formulae and Unicity Results for

2. The Geometry of the Second Fundamental Form of an Oval-

2.2. Four Geometrical Interpretations of the Second Funda-

2.3. The Second Fundamental Form as a Metric and the Cor-

2.4. The Extrinsic Geometry of the Second Fundamental

2.4.1. The Mean Curvature of the Second Fundamental Form. 30

2.5. Some Characterisations of Euclidean Spheres, in

2.5.1. Some First Results.. . . . . . . . . . . . . . . . . . 38

2.6. Calculation of the Curvatures for Some Surfaces of Revol-

2.7. Infinitesimal Deformations of Surfaces in E3 . . . . . . . . 57

3. Variational Formulae for Abstract semi-Riemannian Mani-

3.1. Deformations of Metrics. . . . . . . . . . . . . . . . . . 62

3.2. Variational Problems for Abstract Ovaloids. . . . . . . . 66

4. The Variation of the Shape Operator of a Hypersurface in a

4.2. Hypersurfaces in a semi-Riemannian Manifold. . . . . . 70

4.2.1. Tensors and Curvature.. . . . . . . . . . . . . . . . 70

4.3. The Variation of the Shape Operator. . . . . . . . . . . . 72

4.4. Applications of the Formula for the Variation of the Shape

4.4.1. The Variation of the Curvatures of a Surface in a Three-

5. The Geometry of the Second Fundamental Form of a Hyper-

5.2. Some Technical Results. . . . . . . . . . . . . . . . . . 89

5.2.1. The Tensor B and the Vector Field Z . . . . . . . . . . 90

5.3. The Intrinsic Geometry of the Second Fundamental

5.3.1. The Scalar Curvature of the Second Fundamental Form. 98

5.4. The Extrinsic Geometry of the Second Fundamental

5.5. Some Characterisations of Extrinsic Hyperspheres, in

5.6. Calculation of the Curvatures for Some Special Classes of

5.7. Geodesic Hyperspheres in a Riemannian Manifold. . . . 138

5.7.1. On the Technique of Power Series Expansions for Global

6. A Literature Overview on the Geometry of the Second Fun-

6.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . 157

6.2. Notation and Preliminaries. . . . . . . . . . . . . . . . 158

6.2.1. Surfaces in E3 .. . . . . . . . . . . . . . . . . . . . 158

6.3. The Second Fundamental Form as a Metrical Tensor and

6.4. The Geometry of the Second Fundamental Form and Rel-

6.5. The Conformal Geometry of the Second Fundamental

6.6. II-minimal Surfaces. . . . . . . . . . . . . . . . . . . . . 167

6.7. Congruence Theorems. . . . . . . . . . . . . . . . . . . 168

6.7.1. Some Classical Congruence Theorems. . . . . . . . . 168

6.8. Infinitesimal Deformations. . . . . . . . . . . . . . . . . 176

6.8.1. Infinitesimal Deformations Which Preserve the Second

6.9. On the Global Geometry of Ovaloids. . . . . . . . . . . . 179

6.10. Characterisations of Totally Umbilical Hypersurfaces. . . 182

6.12. Ten Interesting Open Problems. . . . . . . . . . . . . . 194

Notation Index. 215

Author and Subject Index. 221

In this dissertation, hypersurfaces with a non-degenerate second funda-

As such, the geometrical properties of the hypersurface can not

The most well-studied example of hypersurfaces fulfilling the

First of all, the intrinsic geometry of the second fundamental

But a second point of view can be adopted: it could be asked

By means of preparation for our investigations from the second

= −DA(Wp ) Z − DWp (δN ) .