An Invitation to Alexandrov Geometry

CAT 0 Spaces Stephanie Alexander

Visit to download the full and correct content document:
https://textbookfull.com/product/an-invitation-to-alexandrov-geometry-cat-0-spaces-st
ephanie-alexander/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Alexandrov geometry preliminary version no 1 Alexander

S.

https://textbookfull.com/product/alexandrov-geometry-preliminary-
version-no-1-alexander-s/

An Invitation to Geomathematics Willi Freeden

https://textbookfull.com/product/an-invitation-to-geomathematics-
willi-freeden/

An Invitation to Health Dianne R. Hales

https://textbookfull.com/product/an-invitation-to-health-dianne-
r-hales/

An Invitation to Feminist Ethics Hilde Lindemann

https://textbookfull.com/product/an-invitation-to-feminist-
ethics-hilde-lindemann/
An Invitation to Computational Homotopy 1st Edition
Graham Ellis

https://textbookfull.com/product/an-invitation-to-computational-
homotopy-1st-edition-graham-ellis/

An Invitation to Model Theory 1st Edition Jonathan

Kirby

https://textbookfull.com/product/an-invitation-to-model-
theory-1st-edition-jonathan-kirby/

Spaces An Introduction to Real Analysis Tom Lindstrøm

https://textbookfull.com/product/spaces-an-introduction-to-real-
analysis-tom-lindstrom/

Essentials of Meteorology: An Invitation to the

Atmosphere C Donald Ahrens

https://textbookfull.com/product/essentials-of-meteorology-an-
invitation-to-the-atmosphere-c-donald-ahrens/

Geometry and Martingales in Banach Spaces 1st Edition

Wojbor A Woyczynski

https://textbookfull.com/product/geometry-and-martingales-in-
banach-spaces-1st-edition-wojbor-a-woyczynski/
SPRINGER BRIEFS IN MATHEMATICS

Stephanie Alexander
Vitali Kapovitch
Anton Petrunin

An Invitation
to Alexandrov
Geometry
CAT(0) Spaces
SpringerBriefs in Mathematics

Series Editors
Nicola Bellomo, Torino, Italy
Michele Benzi, Pisa, Italy
Palle Jorgensen, Iowa City, USA
Tatsien Li, Shanghai, China
Roderick Melnik, Waterloo, Canada
Otmar Scherzer, Linz, Austria
Benjamin Steinberg, New York City, USA
Lothar Reichel, Kent, USA
Yuri Tschinkel, New York City, USA
George Yin, Detroit, USA
Ping Zhang, Kalamazoo, USA

SpringerBriefs in Mathematics showcases expositions in all areas of mathematics

and applied mathematics. Manuscripts presenting new results or a single new result
in a classical field, new field, or an emerging topic, applications, or bridges between
new results and already published works, are encouraged. The series is intended for
mathematicians and applied mathematicians.

More information about this series at http://www.springer.com/series/10030

Stephanie Alexander Vitali Kapovitch
•

Anton Petrunin

An Invitation to Alexandrov
Geometry
CAT(0) Spaces

123
Stephanie Alexander Anton Petrunin
Department of Mathematics Department of Mathematics
University of Illinois Pennsylvania State University
Urbana, IL, USA University Park, PA, USA

Vitali Kapovitch
Department of Mathematics
University of Toronto
Toronto, ON, Canada

ISSN 2191-8198 ISSN 2191-8201 (electronic)

SpringerBriefs in Mathematics
ISBN 978-3-030-05311-6 ISBN 978-3-030-05312-3 (eBook)
https://doi.org/10.1007/978-3-030-05312-3
Library of Congress Control Number: 2018963986

Mathematics Subject Classification (2010): 53C23, 53C20, 53C45, 53C70, 97G10, 51F99, 51K10

© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2019

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

This short monograph arose as an offshoot of the book on Alexandrov geometry we

have been writing for a number of years. The notes were shaped in a number of
lectures given by the third author to undergraduate students at different occasions at
the MASS program at Penn State University and the Summer School “Algebra and
Geometry” in Yaroslavl.
The idea is to demonstrate the beauty and power of Alexandrov geometry by
reaching interesting applications and theorems with a minimum of preparation.
In Chapter 1, we discuss necessary preliminaries.
In Chapter 2, we discuss the Reshetnyak gluing theorem and apply it to a
problem in billiards which was solved by Dmitri Burago, Serge Ferleger, and
Alexey Kononenko.
In Chapter 3, we discuss the Hadamard–Cartan globalization theorem, and apply
it to the construction of exotic aspherical manifolds introduced by Michael Davis.
In Chapter 4, we discuss examples of Alexandrov spaces with curvature
bounded above. This chapter is based largely on work of Samuel Shefel on non-
smooth saddle surfaces.
Here is a list of some sources providing a good introduction to Alexandrov
spaces with curvature bounded above, which we recommend for further informa-
tion; we will not assume familiarity with any of these sources.
• The book by Martin Bridson and André Haefliger [18];
• Lecture notes of Werner Ballmann [13];
• Chapter 9 in the book [20] by Dmitri Burago, Yuri Burago, and Sergei Ivanov.

Early history of Alexandov geometry

The idea that the essence of curvature lies in a condition on quadruples of points
apparently originated with Abraham Wald. It is found in his publication on
“coordinate-free differential geometry” [66] written under the supervision of Karl

v
vi Preface

Menger; the story of this discovery can be found in [43]. In 1941, similar definitions
were rediscovered independently by Alexandr Danilovich Alexandrov; see [7]. In
Alexandrov’s work the first fruitful applications of this approach were given.
Mainly:
• Alexandrov’s embedding theorem—metrics of nonnegative curvature on the
sphere, and only they, are isometric to closed convex surfaces in Euclidean
3-space.
• Gluing theorem, which tells when the sphere obtained by gluing of two disks
along their boundaries, has nonnegative curvature in the sense of Alexandrov.
These two results together gave a very intuitive geometric tool for studying
embeddings and bending of surfaces in Euclidean space, and changed this subject
dramatically. They formed the foundation of the branch of geometry now called
Alexandrov geometry.
The study of spaces with curvature bounded above started later. The first paper
on the subject was written by Alexandrov; it appeared in 1951; see [8]. It was based
on the work of Herbert Busemann, who studied spaces satisfying a weaker con-
dition [24].
Yurii Grigorievich Reshetnyak proved fundamental results about general spaces
with curvature bounded above, the most important of which is his gluing theorem.
An equally important theorem is the Hadamard–Cartan theorem (globalization
theorem). These theorems and their history are discussed in chapters 2 and 3.
Surfaces with upper curvature bounds were studied extensively in the 50s and
60s, and are by now well understood; see the survey [57] and the references therein.

Manifesto of Alexandrov geometry

Alexandrov geometry can use “back to Euclid” as a slogan. Alexandrov spaces are
defined via axioms similar to those given by Euclid, but certain equalities are
changed to inequalities. Depending on the sign of the inequalities, we get
Alexandrov spaces with curvature bounded above or curvature bounded below.
The definitions of the two classes of spaces are similar, but their properties and
known applications are quite different.
Consider the space M4 of all isometry classes of 4-point metric spaces. Each
element in M4 can be described by 6 numbers—the distances between all 6 pairs of
its points, say ‘i;j for 1 6 i \ j 6 4 modulo permutations of the index set (1, 2, 3, 4).
These 6 numbers are subject to 12 triangle inequalities; that is,

‘i;j þ ‘j;k > ‘i;k

holds for all i, j and k, where we assume that ‘j;i ¼ ‘i;j and ‘i;i ¼ 0.
Preface vii

Consider the subset E 4  M4 of all

isometry classes of 4-point metric spaces that
admit isometric embeddings into Euclidean
space. The complement M4 nE 4 has two N4 E4 P4
connected components.

M4
0.0.1. Exercise. Prove the latter statement.

One of the components will be denoted by P 4 and the other by N 4 . Here P and
N stand for positive and negative curvature because spheres have no quadruples of
type N 4 and hyperbolic space has no quadruples of type P 4 .
A metric space, with length metric, that has no quadruples of points of type P 4
or N 4 respectively is called an Alexandrov space with nonpositive or nonnegative
curvature, respectively.
Here is an exercise, solving which would force the reader to rebuild a consid-
erable part of Alexandrov geometry.
0.0.2. Advanced exercise. Assume X is a complete metric space with length
metric, containing only quadruples of type E 4 . Show that X is isometric to a convex
set in a Hilbert space.
In fact, it might be helpful to spend some time thinking about this exercise
before proceeding.
In the definition above, instead of Euclidean space one can take hyperbolic space
of curvature 1. In this case, one obtains the definition of spaces with curvature
bounded above or below by 1.
To define spaces with curvature bounded above or below by 1, one has to take
the unit 3-sphere and specify that only the quadruples of points such that each of the
four triangles has perimeter less than 2  p are checked. The latter condition could
be considered as a part of the spherical triangle inequality.

Urbana, USA Stephanie Alexander

Toronto, Canada Vitali Kapovitch
University Park, USA Anton Petrunin
Acknowledgements

We want to thank David Berg, Richard Bishop, Yurii Burago, Maxime Fortier
Bourque, Sergei Ivanov, Michael Kapovich, Bernd Kirchheim, Bruce Kleiner,
Nikolai Kosovsky, Greg Kuperberg, Nina Lebedeva, John Lott, Alexander Lytchak,
Dmitri Panov, Stephan Stadler, Wilderich Tuschmann, and Sergio Zamora Barrera
for a number of discussions and suggestions.
We thank the mathematical institutions where we worked on this material,
including BIRS, MFO, Henri Poincaré Institute, University of Colone, Max Planck
Institute for Mathematics.
The first author was partially supported by the Simons Foundation grant
#209053. The second author was partially supported by a Discovery grant from
NSERC and by the Simons Foundation grant #390117. The third author was par-
tially supported by the NSF grant DMS 1309340 and the Simons Foundation
#584781.

ix
Contents

1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Metric spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Geodesics, triangles, and hinges . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Length spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Model angles and triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.6 Angles and the first variation . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.7 Space of directions and tangent space . . . . . . . . . . . . . . . . . . . . . 11
1.8 Hausdorff convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.9 Gromov–Hausdorff convergence . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 Gluing theorem and billiards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1 The 4-point condition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Thin triangles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Reshetnyak’s gluing theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Reshetnyak puff pastry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5 Wide corners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.6 Billiards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3 Globalization and asphericity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1 Locally CAT spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Space of local geodesic paths . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Globalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Polyhedral spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Flag complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.6 Cubical complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.7 Exotic aspherical manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.8 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

xi
xii Contents

4 Subsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 Motivating examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.2 Two-convexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3 Sets with smooth boundary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.4 Open plane sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.5 Shefel’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.6 Polyhedral case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.7 Two-convex hulls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.8 Proof of Shefel’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.9 Comments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Appendix: Semisolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Chapter 1
Preliminaries

In this chapter we fix some conventions and recall the main definitions. The chapter
may be used as a quick reference when reading the book.
To learn background in metric geometry, the reader may consult the book of
Dmitri Burago, Yuri Burago, and Sergei Ivanov [20].

1.1 Metric spaces

The distance between two points x and y in a metric space X will be denoted by
|x − y| or |x − y|X . The latter notation is used if we need to emphasize that the
distance is taken in the space X .
The function
dist x : y → |x − y|

is called the distance function from x.

• The diameter of a metric space X is defined as

diam X = sup { |x − y|X : x, y ∈ X } .

• Given R ∈ [0, ∞] and x ∈ X , the sets

B(x, R) = {y ∈ X | |x − y| < R},

B[x, R] = {y ∈ X | |x − y|  R}

are called, respectively, the open and the closed balls of radius R with center x.

© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2019 1

S. Alexander et al., An Invitation to Alexandrov Geometry,
SpringerBriefs in Mathematics, https://doi.org/10.1007/978-3-030-05312-3_1
2 1 Preliminaries

Again, if we need to emphasize that these balls are taken in the metric space X ,
we write
B(x, R)X and B[x, R]X .

A metric space X is called proper if all closed bounded sets in X are compact.
This condition is equivalent to each of the following statements:
1. For some (and therefore any) point p ∈ X and any R < ∞, the closed ball
B[ p, R] ⊂ X is compact.
2. The function dist p : X → R is proper for some (and therefore any) point p ∈ X ;
that is, for any compact set K ⊂ R, its inverse image { x ∈ X : | p − x|X ∈ K }
is compact.
1.1.1 . Exercise. Let K be a compact metric space and

f:K →K

be a distance nondecreasing map. Prove that f is an isometry.

1.2 Constructions

Product space. Given two metric spaces U and V, the product space U × V is
defined as the set of all pairs (u, v) where u ∈ U and v ∈ V with the metric defined
by formula

|(u 1 , v 1 ) − (u 2 , v 2 )|U ×V = |u 1 − u 2 |2U + |v 1 − v 2 |2V .

Cone. The cone V = Cone U over a metric space U is defined as the metric space
whose underlying set consists of equivalence classes in [0, ∞) × U with the equiv-
alence relation “∼” given by (0, p) ∼ (0, q) for any points p, q ∈ U, and whose
metric is given by the cosine rule

|( p, s) − (q, t)|V = s 2 + t 2 − 2 · s · t · cos α,

where α = min{π, | p − q|U }.

The point in the cone V formed by the equivalence class of 0 × U is called the tip
of the cone and is denoted by 0 or 0V . The distance |0 − v|V is called the norm of v
and is denoted by |v| or |v|V .
Suspension. The suspension V = Susp U over a metric space U is defined as the
metric space whose underlying set consists of equivalence classes in [0, π] × U with
the equivalence relation “∼” given by (0, p) ∼ (0, q) and (π, p) ∼ (π, q) for any
points p, q ∈ U, and whose metric is given by the spherical cosine rule

cos |( p, s) − (q, t)|Susp U = cos s · cos t − sin s · sin t · cos α,

1.2 Constructions 3

where α = min{π, | p − q|U }.

The points in V formed by the equivalence classes of 0 × U and π × U are called
the north and the south poles of the suspension.

1.2.1 . Exercise. Let U be a metric space. Show that the spaces

R × Cone U and Cone[Susp U]

are isometric.

1.3 Geodesics, triangles, and hinges

Geodesic. Let X be a metric space and I be a real interval. A globally isometric map
γ : I → X is called a geodesic1 ; in other words, γ : I → X is a geodesic if

|γ(s) − γ(t)|X = |s − t|

for any pair s, t ∈ I.

We say that γ : I → X is a geodesic from point p to point q if I = [a, b] and
p = γ(a), q = γ(b). In this case the image of γ is denoted by [ pq] and with an abuse
of notations we also call it a geodesic. Given a geodesic [ pq], we can parametrize it
by distance to p; this parametrization will be denoted by geod[ pq] (t).
We may write [ pq]X to emphasize that the geodesic [ pq] is in the space X . We
also use the following shortcut notation:

] pq[ = [ pq]\{ p, q}, ] pq] = [ pq]\{ p}, [ pq[ = [ pq]\{q}.

In general, a geodesic between p and q need not exist and if it exists, it need not
be unique. However, once we write [ pq] we mean that we have made a choice of
geodesic.
A metric space is called geodesic if any pair of its points can be joined by a
geodesic.
A geodesic path is a geodesic with constant-speed parametrization by [0, 1]. Given
a geodesic [ pq], we denote by path[ pq] the corresponding geodesic path; that is,

path[ pq] (t) == geod[ pq] (t · | p − q|).

def

A curve γ : I → X is called a local geodesic if for any t ∈ I, there is a neigh-

borhood U of t in I such that the restriction γ|U is a geodesic. A constant-speed

1 Various authors call it differently: shortest path, minimizing geodesic.

4 1 Preliminaries

parametrization of a local geodesic by the unit interval [0, 1] is called a local geodesic
path.
Triangle. For a triple of points p, q, r ∈ X , a choice of a triple of geodesics
([qr ], [r p], [ pq]) will be called a triangle; we will use the short notation
[ pqr ] = ([qr ], [r p], [ pq]).
Again, given a triple p, q, r ∈ X there may be no triangle [ pqr ] simply because
one of the pairs of these points cannot be joined by a geodesic. Also, many different
triangles with these vertices may exist, any of which can be denoted by [ pqr ].
However, if we write [ pqr ], it means that we have made a choice of such a triangle;
that is, we have fixed a choice of the geodesics [qr ], [r p], and [ pq].
The value
| p − q| + |q − r | + |r − p|

will be called the perimeter of the triangle [ pqr ].

Hinge. Let p, x, y ∈ X be a triple of points such that p is distinct from x and y.
A pair of geodesics ([ px], [ py]) will be called a hinge and will be denoted by
[ p xy ] = ([ px], [ py]).
Convex set. A set A in a metric space X is called convex if for every two points
p, q ∈ A, every geodesic [ pq] in X lies in A.
A set A ⊂ X is called locally convex if every point a ∈ A admits an open neigh-
borhood   a in X such that any geodesic lying in  and with ends in A lies
completely in A.
Note that any open set is locally convex by definition.

1.4 Length spaces

A curve is defined as a continuous map from a real interval to a space. If the real
interval is [0, 1], then the curve is called a path.

1.4.1 . Definition. Let X be a metric space and α : I → X be a curve. We define

the length of α as

length α == |α(ti ) − α(ti−1 )|.
def
sup
t0 t1 ...tn i

Directly from the definition, it follows that if a path α : [0, 1] → X connects two
points x and y (that is, if α(0) = x and α(1) = y), then

length α  |x − y|.
1.4 Length spaces 5

Let A be a subset of a metric space X . Given two points x, y ∈ A, consider the

value
|x − y| A = inf {length α},
α

where the infimum is taken for all paths α from x to y in A.2

If |x − y| A takes finite value for each pair x, y ∈ A, then |x − y| A defines a metric
on A; this metric will be called the induced length metric on A.
If for any ε > 0 and any pair of points x and y in a metric space X , there is a path
α connecting x to y such that

length α < |x − y| + ε,

then X is called a length space and the metric on X is called a length metric.
If f : X̃ → X is a covering, then a length metric on X can be lifted to X̃ by
declaring
lengthX̃ γ = lengthX ( f ◦ γ)

for any curve γ in X̃ . The space X̃ with this metric is called the metric cover of X .
Note that any geodesic space is a length space. As can be seen from the following
example, the converse does not hold.

1.4.2 . Example. Let X be obtained by gluing a countable collection of disjoint

intervals {In } of length 1 + n1 , where for each In the left end is glued to p and the
right end to q. Then X carries a natural complete length metric with respect to which
| p − q| = 1, but there is no geodesic connecting p to q.

1.4.3 . Exercise. Give an example of a complete length space for which no pair of
distinct points can be joined by a geodesic.

Let X be a metric space and x, y ∈ X .

(i) A point z ∈ X is called a midpoint between x and y if

|x − z| = |y − z| = 1
2
· |x − y|.

(ii) Assume ε  0. A point z ∈ X is called an ε-midpoint between x and y if

|x − z|, |y − z|  1
2
· |x − y| + ε.

Note that a 0-midpoint is the same as a midpoint.

2 Note that while this notation slightly conflicts with the previously defined notation for distance on

a general metric space, we will usually work with ambient length spaces where the meaning will
be unambiguous.
6 1 Preliminaries

1.4.4 . Lemma. Let X be a complete metric space.

(a) Assume that for any pair of points x, y ∈ X and any ε > 0 there is an
ε-midpoint z. Then X is a length space.
(b) Assume that for any pair of points x, y ∈ X , there is a midpoint z. Then X is
a geodesic space.

Proof. We first prove (a). Let x, y ∈ X be a pair of points.

Set εn = 4εn , α(0) = x and α(1) = y.
Let α( 21 ) be an ε1 -midpoint between α(0) and α(1). Further, let α( 41 ) and α( 43 ) be
ε2 -midpoints between the pairs (α(0), α( 21 )) and (α( 21 ), α(1)), respectively. Apply-
ing the above procedure recursively, on the n-th step we define α( 2kn ), for every
odd integer k such that 0 < 2kn < 1, as an εn -midpoint between the already defined
α( k−1
2n
) and α( k+1
2n
).
In this way we define α(t) for t ∈ W , where W denotes the set of dyadic rationals
in [0, 1]. Since X is complete, the map α can be extended continuously to [0, 1].
Moreover,
∞

length α  |x − y| + 2n−1 · εn 
➊ n=1
ε
 |x − y| + 2
.

Since ε > 0 is arbitrary, we get (a).

To prove (b), one should repeat the same argument taking midpoints instead of
εn -midpoints. In this case ➊ holds for εn = ε = 0. 

Since in a compact space a sequence of n1 -midpoints z n contains a convergent

subsequence, Lemma 1.4.4 immediately implies

1.4.5 . Proposition. A proper length space is geodesic.

1.4.6 . Hopf–Rinow theorem. Any complete, locally compact length space is

proper.

Proof. Let X be a locally compact length space. Given x ∈ X , denote by ρ(x) the
supremum of all R > 0 such that the closed ball B[x, R] is compact. Since X is
locally compact,

➋ ρ(x) > 0 for any x ∈ X .

It is sufficient to show that ρ(x) = ∞ for some (and therefore any) point x ∈ X .

Assume the contrary; that is, ρ(x) < ∞. We claim that

➌ B = B[x, ρ(x)] is compact for any x.

1.4 Length spaces 7

Indeed, X is a length space; therefore for any ε > 0, the set B[x, ρ(x) − ε] is a
compact ε-net in B. Since B is closed and hence complete, it must be compact.
Next we claim that
➍ |ρ(x) − ρ(y)|  |x − y|X for any x, y ∈ X ; in particular ρ : X → R is a con-
tinuous function.

Indeed, assume the contrary; that is, ρ(x) + |x − y| < ρ(y) for some x, y ∈ X .
Then B[x, ρ(x) + ε] is a closed subset of B[y, ρ(y)] for some ε > 0. Then compact-
ness of B[y, ρ(y)] implies compactness of B[x, ρ(x) + ε], a contradiction.
Set ε = min { ρ(y) : y ∈ B }; the minimum is defined since B is compact. From
➋, we have ε > 0.
Choose a finite 10ε -net {a1 , a2 , . . . , an } in B. The union W of the closed balls
B[ai , ε] is compact. Clearly B[x, ρ(x) + 10ε ] ⊂ W . Therefore B[x, ρ(x) + 10ε ] is
compact, a contradiction. 

1.4.7 . Exercise. Construct a geodesic space that is locally compact, but whose
completion is neither geodesic nor locally compact.

1.5 Model angles and triangles

Let X be a metric space and p, q, r ∈ X . Let us define the model triangle [ p̃q̃ r̃ ]
(briefly, [ p̃q̃ r̃ ] = ˜ ( pqr )E2 ) to be a triangle in the plane E2 with the same side
lengths; that is,

| p̃ − q̃| = | p − q|, |q̃ − r̃ | = |q − r |, |r̃ − p̃| = |r − p|.

In the same way we can define the hyperbolic and the spherical model triangles
˜ ( pqr )H2 , ˜ ( pqr )S2 in the hyperbolic plane H2 and the unit sphere S2 . In the latter
case the model triangle is said to be defined if in addition

| p − q| + |q − r | + |r − p| < 2 · π.

In this case the model triangle again exists and is unique up to an isometry
of S2 .
If [ p̃q̃ r̃ ] = ˜ ( pqr )E2 and | p − q|, | p − r | > 0, the angle measure of [ p̃q̃ r̃ ] at p̃
q
will be called the model angle of the triple p, q, r and will be denoted by  ˜ ( p r )E 2 .
q q
In the same way we define  ˜ ( p r )H2 and  ˜ ( p r )S2 ; in the latter case we assume in
addition that the model triangle ˜ ( pqr )S2 is defined.
q
We may use the notation  ˜ ( p r ) if it is evident which of the model spaces H2 , E2 ,
or S is meant.
2
8 1 Preliminaries

x
1.5.1 . Alexandrov’s lemma. Let p, x, y, z be distinct points
in a metric space such that z ∈ ]x y[. Then the following expr-
essions for the Euclidean model angles have the same sign: z
p p
(a) 
˜ (x y ) − 
˜ (x z ), y
p p
˜ (z x ) + 
(b)  ˜ (z y ) − π.
Moreover, p

˜ ( p xy)  
 ˜ ( p zx) + 
˜ ( p zy),

with equality if and only if the expressions in (a) and (b) vanish.
The same holds for the hyperbolic and spherical model angles, but in the latter
case one has to assume in addition that

| p − z| + | p − y| + |x − y| < 2 · π.

Proof. Consider the model triangle [x̃ p̃ z̃] = ˜ (x pz). Take a point ỹ on the extension
of [x̃ z̃] beyond z̃ so that |x̃ − ỹ| = |x − y| (and therefore |x̃ − z̃| = |x − z|).
Since increasing the opposite side in a plane triangle
increases the corresponding angle, the following expre-
ỹ
ssions have the same sign:
p̃ p
(i) [x̃ ỹ ] − 
˜ (x y ),
z˜
(ii) | p̃ − ỹ| − | p − y|,
p̃ p
(iii) [z̃ ỹ ] − 
˜ (z y ).
x̃
Since p˜

p̃ p̃
[x̃ ỹ ] = [x̃ z̃ ] = 
˜ (x zp)

and
p̃
[z̃ ỹ ] = π − [z̃ x̃p̃ ] = π − 
˜ (z xp),

the first statement follows.

For the second statement, construct a model triangle [ p̃ z̃ ỹ ] = ˜ ( pzy)E2 on the
opposite side of [ p̃ z̃] from [x̃ p̃ z̃]. Note that

|x̃ − ỹ |  |x̃ − z̃| + |z̃ − ỹ | =

= |x − z| + |z − y| =
= |x − y|.
1.5 Model angles and triangles 9

Therefore

˜ ( p zx) + 
 ˜ ( p zy) = [ p̃ z̃x̃ ] + [ p̃ z̃ỹ ] =
= [ p̃ x̃ỹ ] 

˜ ( p xy).

Equality holds if and only if |x̃ − ỹ | = |x − y|, as required. 

1.6 Angles and the first variation

Given a hinge [ p xy ], we define its angle as the limit

➊ [ p xy ] == lim 
˜ ( p x̄ȳ)E2 ,
def

x̄, ȳ→ p

where x̄ ∈ ] px] and ȳ ∈ ] py]. (The angle [ p xy ] is defined if the limit exists.)
The value under the limit can be calculated from the cosine law:

| p − x|2 + | p − y|2 − |x − y|2

˜ ( p xy)E2 =
cos  .
2 · | p − x| · | p − y|

The following lemma implies that in ➊, one can use 

˜ ( p x̄ȳ)S2 or 
˜ ( p x̄ȳ)H2 instead
of 
˜ ( p x̄ȳ)E2 .

1.6.1 . Lemma. For any three points p, x, y in a metric space the following inequal-
ities

|˜( p xy)S2 − 
˜ ( p xy)E2 |  | p − x| · | p − y|,
➋
|˜( p xy)H2 − 
˜ ( p xy)E2 |  | p − x| · | p − y|

hold whenever the left-hand side is defined.

Proof. Note that


˜ ( p xy)H2  
˜ ( p xy)E2  
˜ ( p xy)S2 .

Therefore

˜ ( p xy)S2 − 
0 ˜ ( p xy)H2 
˜ ( p xy)S2 + 
 ˜ (x yp)S2 + 
˜ (y xp)S2 − 
˜ ( p xy)H2 − 
˜ (x yp)H2 − 
˜ (y xp)H2 =
= area ˜ ( px y)S2 + area ˜ ( px y)H2 .
10 1 Preliminaries

The inequality ➋ follows since

0  area ˜ ( px y)H2 
 area ˜ ( px y)S2 
 | p − x| · | p − y|.


1.6.2 . Triangle inequality for angles. Let [ px 1 ], [ px 2 ], and [ px 3 ] be three

i
geodesics in a metric space. If all the angles αi j = [ p xx j ] are defined, then they
satisfy the triangle inequality:

α13  α12 + α23 .

Proof. Since α13  π, we may assume that α12 + α23 < π.

Set γi = geod[ px i ] . Given any ε > 0, for all sufficiently small t, τ , s ∈ R+ we
have

|γ1 (t) − γ3 (τ )| |γ1 (t) − γ2 (s)| + |γ2 (s) − γ3 (τ )| <


< t 2 + s 2 − 2 · t · s · cos(α12 + ε) +

+ s 2 + τ 2 − 2 · s · τ · cos(α23 + ε) 

Below we define s(t, τ ) so that for s = s(t, τ ), this

chain of inequalities can be continued as follows:

 t 2 + τ 2 − 2 · t · τ · cos(α12 + α23 + 2 · ε).

Thus for any ε > 0,

t

α13  α12 + α23 + 2 · ε.

23 +ε
Hence the result. =α
To define s(t, τ ), consider three rays γ̃1 , γ̃2 , γ̃3 s
on a Euclidean plane starting at one point, such = α1
2
that (γ̃1 , γ̃2 ) = α12 + ε, (γ̃2 , γ̃3 ) = α23 + ε and +ε
(γ̃1 , γ̃3 ) = α12 + α23 + 2 · ε. We parametrize each
τ

ray by the distance from the starting point. Given

two positive numbers t, τ ∈ R+ , let s = s(t, τ ) be
1.6 Angles and the first variation 11

the number such that γ̃2 (s) ∈ [γ̃1 (t) γ̃3 (τ )]. Clearly s  max{t, τ }, so t, τ , s may
be taken sufficiently small. 

1.6.3 . Exercise. Prove that the sum of adjacent angles is at least π.

More precisely, let X be a complete length space and p, x, y, z ∈ X . If p ∈ ]x y[,
then
[ p zx ] + [ p zy ]  π

whenever each angle on the left-hand side is defined.

p
1.6.4 . First variation inequality. Assume that for a hinge [q x ] the angle
p
α = [q x ] is defined. Then

| p − geod[q x] (t)|  |q − p| − t · cos α + o(t).

Proof. Take a sufficiently small ε > 0. For all sufficiently small t > 0, we have

|geod[q p] (t/ε) − geod[q x] (t)|  t
ε
· 1 + ε2 − 2 · ε · cos α + o(t) 
 t
ε
− t · cos α + t · ε.

Applying the triangle inequality, we get

| p − geod[q x] (t)|  | p − geod[q p] (t/ε)| + |geod[q p] (t/ε) − geod[q x] (t)| 

 | p − q| − t · cos α + t · ε

for any fixed ε > 0 and all sufficiently small t. Hence the result. 

1.7 Space of directions and tangent space

Let X be a metric space with defined angles for all hinges. Fix a point p ∈ X .
Consider the set S p of all nontrivial geodesics that start at p. By 1.6.2, the triangle
inequality holds for  on S p , so (S p , ) forms a pseudometric space; that is, 
satisfies all the conditions of a metric on S p , except that the angle between distinct
geodesics might vanish.
The metric space corresponding to (S p , ) is called the space of geodesic direc-
tions at p, denoted by  p or  p X . Elements of  p are called geodesic directions
at p. Each geodesic direction is formed by an equivalence class of geodesics in S p
for the equivalence relation

[ px] ∼ [ py] ⇐⇒ [ p xy ] = 0.
12 1 Preliminaries

The completion of  p is called the space of directions at p and is denoted by  p

or  p X . Elements of  p are called directions at p.
The Euclidean cone Cone  p over the space of directions  p is called the tangent
space at p and is denoted by T p or T p X .
The tangent space T p could also be defined directly, without introducing the space
of directions. To do so, consider the set T p of all geodesics with constant-speed
parametrizations starting at p. Given α, β ∈ T p , set

|α(ε) − β(ε)|X
➊ |α − β|T p = lim
ε→0 ε

Since the angles in X are defined, ➊ defines a pseudometric on T p .

The corresponding metric space admits a natural isometric identification with the
cone T p = Cone  p . The elements of T p are equivalence classes for the relation

α∼β ⇐⇒ |α(t) − β(t)|X = o(t).

The completion of T p is therefore naturally isometric to T p .

Elements of T p will be called tangent vectors at p, regardless of the fact that T p
is only a metric cone and need not be a vector space. Elements of T p will be called
geodesic tangent vectors at p.

1.8 Hausdorff convergence

It seems that Hausdorff convergence was first introduced by Felix Hausdorff in

[42], and a couple of years later an equivalent definition was given by Wilhelm
Blaschke in [16]. A refinement of this definition was introduced by Zdeněk Frolík
in [35], and later rediscovered by Robert Wijsman in [68]. However, this refinement
takes an intermediate place between the original Hausdorff convergence and closed
convergence, also introduced by Hausdorff in [42]. For this reason we call it Hausdorff
convergence instead of Hausdorff–Blascke–Frolík–Wijsman convergence.
Let X be a metric space and A ⊂ X . We will denote by dist A (x) the distance
from A to a point x in X ; that is,

dist A (x) == inf { |a − x|X : a ∈ A } .

def

1.8.1 . Definition of Hausdorff convergence. Given a sequence of closed sets

(An )∞
n=1 in a metric space X , a closed set A∞ ⊂ X is called the Hausdorff limit
of (An )∞
n=1 , briefly An → A∞ , if
1.8 Hausdorff convergence 13

dist An (x) → dist A∞ (x) as n → ∞

for every x ∈ X .
In this case, the sequence of closed sets (An )∞
n=1 is said to converge in the sense
of Hausdorff.

Example. Let Dn be the disk in the coordinate plane with center (0, n) and radius n.
Then Dn converges to the upper half-plane as n → ∞.

1.8.2 . Exercise. Let An → A∞ as in Definition 1.8.1.

Show that A∞ is the set of all points p such that pn → p for some sequence of
points pn ∈ An .
Does the converse hold? That is, suppose (An )∞ n=1 , A∞ are closed sets such that
A∞ is the set of all points p such that pn → p for some sequence of points pn ∈ An .
Does this imply that An → A∞ ?

1.8.3 . First selection theorem. Let X be a proper metric space and (An )∞ n=1 be a
sequence of closed sets in X . Assume that for some (and therefore any) point x ∈ X ,
the sequence dist An (x) is bounded. Then the sequence (An )∞ n=1 has a convergent
subsequence in the sense of Hausdorff.

Proof. Since X is proper, there is a countable dense set {x1 , x2 , . . .} in X . Note

that the sequence dn = dist An (xk ) is bounded for each k. Therefore, passing to a
subsequence of (An )∞ n=1 , we can assume that dist An (x k ) converges as n → ∞ for
any fixed k.
Note that for each n, the function dist An : X → R is 1-Lipschitz and nonnegative.
Therefore the sequence dist An converges pointwise to a 1-Lipschitz nonnegative
function f : X → R.
Set A∞ = f −1 (0). Let us show that

dist A∞ (y)  f (y)

for any y.
Assume the contrary; that is,

f (z) < R < dist A∞ (z)

for some z ∈ X and R > 0. Then for any sufficiently large n, there is a point z n ∈ An
such that |x − z n |  R. Since X is proper, we can pass to a partial limit z ∞ of z n as
n → ∞.
It is clear that f (z ∞ ) = 0; that is, z ∞ ∈ A∞ . (Note that this implies that A∞ = ∅).
On the other hand,
14 1 Preliminaries

dist A∞ (y)  |z ∞ − y|  R < dist A∞ (y),

a contradiction.
On the other hand, since f is 1-Lipschitz, dist A∞ (y)  f (y). Therefore

dist A∞ (y) = f (y)

for any y ∈ X . Hence the result. 

1.9 Gromov–Hausdorff convergence

1.9.1 . Definition. Let { Xα : α ∈ A } be a collection of metric spaces. A metric ρ on

the disjoint union 
X= Xα
α∈A

is called a compatible metric if the restriction of ρ to every Xα coincides with the

original metric on Xα .

1.9.2 . Definition. Let X1 , X2 , . . . and X∞ be proper metric spaces and ρ be a

compatible metric on their disjoint union X. Assume that Xn is an open set in (X, ρ)
for each n = ∞, and Xn → X∞ in (X, ρ) as n → ∞ in the sense of Hausdorff (see
Definition 1.8.1).
Then we say ρ defines a convergence3 in the sense of Gromov–Hausdorff, and
ρ
write Xn → X∞ or Xn − → X∞ . The space X∞ is called the limit space of the sequence
(Xn ) along ρ.

Usually Gromov–Hausdorff convergence is defined differently. We prefer this

definition since it induces convergence for a sequence of points xn ∈ Xn (Exercise
1.8.2), as well as weak convergence of measures μn on Xn , and so on, corresponding
to convergence in the ambient space (X, ρ).
Once we write Xn → X∞ , we mean that we have made a choice of convergence.
Note that for a fixed sequence of metric spaces (Xn ), it might be possible to construct
ρ ρ
different Gromov–Hausdorff convergences, say Xn − →X∞ and Xn − → X∞ , whose
limit spaces X∞ and X∞ need not be isometric to each other.
For example, for the constant sequence Xn == R0 , there is a convergence with
iso

limit X∞ == R0 ; guess the metric ρ from the diagram.

iso

3 Formally speaking, convergence is the topology induced by ρ on X.

1.9 Gromov–Hausdorff convergence 15

X1

X2
...

X∞

For another metric ρ —also guess it from the diagram—the limit space X∞ is
isometric to the real line.
X1

X2
...


X∞

1.9.3 . Second selection theorem. Let Xn be a sequence of proper metric spaces

with marked points xn ∈ Xn . Assume that for any fixed R, ε > 0, there is
N = N (R, ε) ∈ N such that for each n the ball B[xn , R]Xn admits a finite ε-net with
at most N points. Then there is a subsequence of Xn admitting a Gromov–Hausdorff
convergence such that the sequence of marked points xn ∈ Xn converges.

Proof. From the main assumption in the theorem, in each space Xn there is a sequence
of points z i,n ∈ Xn such that the following condition holds for a fixed sequence of
integers M1 < M2 < . . .
• |z i,n − xn |Xn  k + 1 if i  Mk ,
• the points z 1,n , . . . , z Mk ,n form an k1 -net in B[xn , k]Xn .
Passing to a subsequence, we can assume that the sequence

n = |z i,n − z j,n |Xn

converges for any i and j.

Consider a countable set of points W = {w1 , w2 , . . .} equipped with the pseudo-
metric defined by
|wi − w j |W = lim |z i,n − z j,n |Xn .
n→∞

Let Ŵ be the metric space corresponding to W; that is, points in Ŵ are equivalence
classes in W for the relation ∼, where wi ∼ w j if and only if |wi − w j |W = 0, and
where

|[wi ] − [w j ]|Ŵ == |wi − w j |W .

def
16 1 Preliminaries

Denote by X∞ the completion of Ŵ.

It remains to show that there is a Gromov–Hausdorff convergence Xn → X∞ such
that the sequence xn ∈ Xn converges. To prove this, we need to construct a metric ρ
on the disjoint union of
X = X∞  X1  X2  . . .

satisfying definitions 1.9.1 and 1.9.2. The metric ρ can be constructed as the maximal
compatible metric such that
ρ(z i,n , wi )  m1

for any n  Nm and i < Im for a suitable choice of two sequences (Im ) and (Nm )
with I1 = N1 = 1. 

1.9.4 . Exercise. Let Xn be a sequence of metric spaces that admits two

ρ ρ
convergences Xn −
→ X∞ and Xn −
→ X∞ .
(a) If X∞ is compact, then X∞ == X∞ .
iso

(b) If X∞ is proper and there is a sequence of points xn ∈ Xn that converges in

both convergences, then X∞ == X∞ .
iso
Chapter 2
Gluing theorem and billiards

In this chapter we define CAT(κ) spaces and give the first application, to billiards.
Here “CAT” is an acronym for Cartan, Alexandrov, and Toponogov. It was
coined by Mikhael Gromov in 1987. Originally, Alexandrov called these spaces
“Rκ domain”; this term is still in use.
Riemannian manifolds with nonpositive sectional curvature provide a motivating
example. Specifically, a Riemannian manifold has nonpositive sectional curvature if
and only if each point admits a CAT(0) neighborhood.

2.1 The 4-point condition

Given a quadruple of points p, q, x, y in a metric space X , consider two model

˜ px y)E2 and [q̃ x̃ ỹ] = (q
triangles in the plane [ p̃ x̃ ỹ] = ( ˜ x y)E2 with common side
[x̃ ỹ].
If the inequality
x̃ q̃
| p − q|X  | p̃ − z̃|E2 + |z̃ − q̃|E2
z̃
holds for any point z̃ ∈ [x̃ ỹ], then we say that the quadru-
ple p, q, x, y satisfies CAT(0) comparison.
If we do the same for spherical model triangles
˜ px y)S2 and [q̃ x̃ ỹ] = (q
˜ x y)S2 , then we arrive p̃ ỹ
[ p̃ x̃ ỹ] = (
at the definition of CAT(1) comparison. If one of the spherical model triangles is
undefined,1 then it is assumed that CAT(1) comparison automatically holds for this
quadruple.

1 That is, if

| p − x| + | p − y| + |x − y|  2 · π or |q − x| + |q − y| + |x − y|  2 · π.

© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2019 17

S. Alexander et al., An Invitation to Alexandrov Geometry,
SpringerBriefs in Mathematics, https://doi.org/10.1007/978-3-030-05312-3_2
18 2 Gluing theorem and billiards

We can do the same for the model plane of curvature κ; that is, a sphere if κ > 0,
Euclidean plane if κ = 0 and Lobachevsky plane if κ < 0. In this case we arrive at the
definition of CAT(κ) comparison. However in these notes we will mostly consider
CAT(0) comparison and occasionally CAT(1) comparison; so, if you see CAT(κ),
you can assume that κ is 0 or 1.
If all quadruples in a metric space X satisfy CAT(κ) comparison, then we say
that the space X is CAT(κ). (Note that CAT(κ) is an adjective.)
In order to check CAT(κ) comparison, it is sufficient to know the 6 distances
between all pairs of points in the quadruple. This observation implies the following.

2.1.1 . Proposition. Any Gromov–Hausdorff limit of a sequence of CAT(κ) spaces

is CAT(κ).
In the proposition above, it does not matter which definition of convergence for
metric spaces you use, as long as any quadruple of points in the limit space can be
arbitrarily well approximated by quadruples in the sequence of metric spaces.

2.1.2 . Exercise. Let V be a metric space and U = Cone V. Show that U is CAT(0)
if and only if V is CAT(1).
Analogously, if U = Susp V, then U is CAT(1) if and only if V is CAT(1).

The cone and suspension constructions are defined in Section 1.2.

The following exercise is a bit simpler, but can be proved in essentially the same
way.

2.1.3 . Exercise. Assume U and V are CAT(0) spaces. Show that the product space
U × V is CAT(0).

2.1.4 . Exercise. Show that any complete length CAT(0) space is geodesic.

2.2 Thin triangles

The inheritance lemma 2.2.9 proved below plays a central role in the theory of CAT(κ)
spaces. It will lead to two fundamental constructions: patchwork globalization (3.3.2)
and Reshetnyak gluing (2.3.1), which in turn are used to prove the globalization
theorem (3.3.1).
Recall that a triangle [x 1 x 2 x 3 ] in a space X is a triple of minimizing geodesics
˜ 1 x 2 x 3 )E 2
[x x ], [x 2 x 3 ], and [x 3 x 1 ]. Consider the model triangle [x̃ 1 x̃ 2 x̃ 3 ] = (x
1 2

in the Euclidean plane. The natural map [x̃ x̃ x̃ ] → [x x x ] sends a point

1 2 3 1 2 3

z̃ ∈ [x̃ i x̃ j ] to the corresponding point z ∈ [x i x j ]; that is, z is the point such that
|x̃ i − z̃| = |x i − z|, and therefore |x̃ j − z̃| = |x j − z|.
In the same way, the natural map can be defined for the spherical model triangle
˜ 1 x 2 x 3 )S2 .
(x
2.2 Thin triangles 19

2.2.1 . Definition of thin triangles. A triangle [x 1 x 2 x 3 ] in the metric space X is

˜ 1 x 2 x 3 )E2 → [x 1 x 2 x 3 ] is short (that is, a distance
called thin if the natural map (x
nonincreasing map).
Analogously, a triangle [x 1 x 2 x 3 ] is called spherically thin if the natural map from
˜ 1 x 2 x 3 )S2 to [x 1 x 2 x 3 ] is short.
the spherical model triangle (x
2.2.2 . Proposition. A geodesic space is CAT(0) (CAT(1)) if and only if all its
triangles are thin (respectively, all its triangles of perimeter < 2 · π are spherically
thin).
Proof; “if” part. Apply the triangle inequality and thinness of triangles [ px y]
and [q x y], where p, q, x, and y are as in the definition of CAT(κ) comparison
(Section 2.1).
“Only if” part. Applying CAT(0) comparison to a quadruple p, q, x, y with
q ∈ [x y] shows that any triangle satisfies point-side comparison, that is, the distance
from a vertex to a point on the opposite side is no greater than the corresponding
distance in the Euclidean model triangle.
Now consider a triangle [x 1 x 2 x 3 ], and let y ∈ [x 1 x 2 ] and z ∈ [x 1 x 3 ]. Let ỹ, z̃ be
the corresponding points on the sides of the model triangle (x ˜ 1 x 2 x 3 )E2 . Applying
point-side comparison first to the triangle [x x x ] with y ∈ [x 1 x 2 ], and then to the
1 2 3

triangle [x 1 yx 3 ] with z ∈ [x 1 x 3 ], implies that model angles satisfy

2 y

˜ (x 1 xx 3)E2  
˜ (x 1 x 3)E2  
˜ (x 1 zy)E2 .

Therefore | ỹ − z̃|E2  |y − z|.

The CAT(1) argument is the same. 
2.2.3 . Uniqueness of geodesics. In a proper length CAT(0) space, pairs of points
are joined by unique geodesics, and these geodesics depend continuously on their
endpoint pairs.
Analogously, in a proper length CAT(1) space, pairs of points at distance less
than π are joined by unique geodesics, and these geodesics depend continuously on
their endpoint pairs.
Proof. Given 4 points p 1 , p 2 , q 1 , q 2 in a proper length CAT(0) space U, consider
two triangles [ p 1 q 1 p 2 ] and [ p 2 q 2 q 1 ]. Since both of these triangles are thin, we get

|path[ p1 q 1 ] (t) − path[ p2 q 1 ] (t)|U  (1 − t) · | p 1 − p 2 |U ,

|path[ p2 q 1 ] (t) − path[ p2 q 2 ] (t)|U  t · |q 1 − q 2 |U .

By the triangle inequality,

|path[ p1 q 1 ] (t) − path[ p2 q 2 ] (t)|U  max{| p 1 − p 2 |U , |q 1 − q 2 |U }.

Hence, continuity and uniqueness in the CAT(0) case follow.

The CAT(1) case is done in essentially the same way. 
Another random document with
no related content on Scribd:
single editorial process. Conceivably the work was achieved by a
small body of Levites (see below under Authorship), contemporaries,
sharing the same training and outlook; but it is much more
reasonable to infer the activity of a single writer—the Chronicler. It is
his date which we proceed to consider. The evidence may be
grouped under three heads, of which the last two (B, C) are of chief
importance.

(A) (1) In 1 Chronicles xxix. 7 a sum of money is reckoned in

darics, a Persian coinage introduced by Darius I (521‒486 b.c.).

(2) In 1 Chronicles iii. 19b‒24 (see note ad loc.) according to the

Hebrew the line of David’s family is traced to the sixth generation
after Zerubbabel (circa 520 b.c.). Hence, reckoning a generation as
about 20 to 30 years, this passage would indicate a date not earlier
than 400 b.c. or 340 b.c. The Greek, Syriac, and Latin Versions,
however, differ from the Hebrew by extending the line to the eleventh
generation after Zerubbabel. That would imply a date possibly as
late as 200 b.c. and not earlier than about 300 b.c., but it is very
doubtful whether we can here rely upon the text of these Versions,
and obviously it was easy for a translator or scribe to carry the list on
to his own date. This piece of evidence, therefore, for the later date
cannot be pressed, although it is worthy of notice.

Since, as we have said (§ 2), Ezra‒Nehemiah formed originally

one book with Chronicles, evidence for the date of Chronicles is also
furnished by any indications of date which occur in Ezra‒Nehemiah.

(3) In Nehemiah xii. 22 we find the significant phrase “to the reign
of Darius, the Persian.” Now as long as the Persian empire stood
such a description would have no point when written by a Jewish
writer. For two hundred years down to 332 b.c., when Syria and
Phoenicia fell into the hands of Alexander the Great, the rulers of
Judea were all “Persians.” But from 332 b.c. onward Greek
monarchs were the rulers of Palestine, and nothing is more natural
than that a Jewish chronicler writing under their rule should refer to a
king of the older régime as “the Persian.”
 (4) Further, in Nehemiah xii. 26, 47 the phrase “in the days of
Nehemiah” occurs, implying that for the writer Nehemiah belonged to
the past, but, as one cannot say how near or how distant a past, the
point carries little weight.

(5) Again, in Nehemiah xii. 10, 11 and 22, 23, a list of high-priests
is given, concluding with the name of Jaddua, whom the Chronicler
evidently (and correctly, compare Josephus, Antiquities of the Jews
xi. 7, 8) knew to have been the high-priest about 332 b.c., at the end
of the reign of Darius (Darius III, Codomannus), when the Persian
Empire collapsed before the attack of Alexander the Great.

These details, indicative of the date of composition, are as

numerous as we have any right to expect in a work of the nature of
Chronicles, which deals with past history. Their evidential value can
of course be criticised by supposing that the passages in question
are late interpolations and have therefore no bearing on the date of
the main body of the work. But no solid grounds are adduced for this
objection, and the burden of proof lies upon the objector. The
supposition is extremely improbable, and may be dismissed in view
of the fact that (B) the general character, and (C) the linguistic
peculiarities of Chronicles alike demand a date considerably later
than the period of Ezra‒Nehemiah.

(B) The character of Chronicles has already been referred to, but
in a different connection (§ 2, pp. xvi f.). Here the point to notice is
that throughout the entire work the whole system of law and ritual
found in the Pentateuch is presupposed as existing in its final form.
This system, which may conveniently be described as Priestly (P) in
distinction from the earlier system to which the name Deuteronomic
(D) is applied, and the still earlier standpoint represented by the
Jahvistic and Elohistic writers (J and E), may have been of slow
growth, and no doubt embodies features of law and ritual which are
of relatively high antiquity. But there is overwhelming evidence to
prove that, as an organised and completed system, it cannot be
dated earlier than the period of Nehemiah (circa 425 b.c.). Now in
Chronicles not only is this final code in force; it has evidently been so
long and so firmly established that the Chronicler did not know, or at
least did not believe, that any other earlier system had once ruled
the practice of Israel. He belonged to a period when the
development of the Pentateuch was no more remembered, and
when its origin—in all completeness—had come to be ascribed with
absolute confidence to the remote past, in accordance with that
religious instinct which we have described above on p. xiv.
Manifestly, a considerable lapse of time after Nehemiah’s period
must be allowed for that conviction to have become established.

Another consideration is found in the attitude of the Chronicler

towards the kingdom of Israel. For the apostate Northern Kingdom
the Chronicler has only contempt and hatred and displays no interest
whatever in its fortunes, except that he takes pains occasionally to
indicate the corruption of the North, thus emphasising by contrast the
virtue of the Judeans. His absorption in the affairs of Jerusalem and
his bitter antagonism to the North in all likelihood reflect the anger
felt by the post-exilic Jews of Jerusalem against the Samaritans after
the famous schism between the two communities. Indeed it is
possible, § 6 (1), that his work was directly inspired by the necessity
of combating the religious pretensions of the Samaritans with their
Temple on Mt Gerizim, rivalling Jerusalem. The Samaritan schism is
generally supposed to date from 432 b.c., but there are grounds for
thinking that it was not so early, and possibly the Temple on Mt
Gerizim may have been built, not in 432, but in 332 b.c. If the later
date be correct, we have strong evidence for dating Chronicles not
earlier than the last half of the fourth century b.c.

(C) The late date of Chronicles is finally put beyond all doubt by
the linguistic peculiarities of the book. Excluding, of course, the
passages drawn from earlier Scriptures, the Hebrew of Chronicles is
of such a character that it is impossible to assign anything but a late
post-exilic date for its composition. In every aspect of language—
grammar and syntax and vocabulary—the diction exhibits the
unmistakable characteristics of late Hebrew. It lies beyond the scope
of the present volume to give details of the Hebrew, and reference
may be made to the edition of Chronicles by Curtis and Madsen
(International Critical Commentary), pp. 27 ff., where a list of 136
such peculiarities is given.

The style of Chronicles is disappointing. The Chronicler had

some praiseworthy qualities as a narrator: he displays force and
imagination in the treatment of the material, he knew how to add a
graphic touch, and he was able to revise a story thoroughly while
preserving its internal coherence (e.g. 1 Chronicles xxi.). But he had
not the gift of choice language. In so late a work we could not hope
to find the strength and purity which characterised Hebrew prose of
the “golden age.” It must, however, be confessed that, judged even
by the standard of its own age, the Hebrew of Chronicles is clumsy
and displeasing in many ways.

From this cumulative evidence we infer that the Chronicler was

certainly a post-exilic writer later than the period of Ezra‒Nehemiah
and in all probability not earlier than about 300‒250 b.c. This is a
valuable and definite conclusion, but it is important to observe that it
does not fully answer the problem of the date of the present form of
Chronicles. It remains to ask whether the text as it has reached us
(the Masoretic Hebrew) is precisely the text which left the
Chronicler’s hands, and, if not, what changes have been introduced.
It is safe to say that the Hebrew text has been almost unchanged
since about 150 a.d., but between that date and the time of the
Chronicler is a long and sometimes stormy period. The subject,
though in many ways important, is too intricate to be discussed here
at length: a few remarks must suffice. (1) Like all other books of the
Old Testament, Chronicles has suffered from the usual accidents of
scribal errors in the course of transmission; but the changes due to
this cause, being unintentional, are as a rule unimportant and can
often be detected and corrected (see § 10, Text). (2) More serious
are alterations made by revisers or scribes who were anxious to
bring the narrative of Chronicles into conformity with that of Samuel
and Kings. In the last two chapters of 2 Chronicles the Hebrew text
can be compared with an old Greek Version (1 Esdras—see § 10,
Greek Versions), and the comparison indicates that changes of text
(see notes on 2 Chronicles xxxv. 8, 15) and a harmonisation of
Chronicles with Kings (see note on 2 Chronicles xxxvi. 5; compare
also verse 15) have occurred in that brief section.

Except in these two chapters the old Greek Version has

unfortunately perished, and for all the rest of Chronicles comparison
can only be made with a much later Greek Version, which is a
translation of a Hebrew text almost identical with the present,
Masoretic, form. Even so, differences are found, notably a
substantial passage deleted from the Hebrew in 2 Chronicles xxxv.
19 (where see note). It is a legitimate conjecture that, if the old
Greek Version were extant throughout Chronicles, considerable
variations between the earlier and the present text might be
disclosed. (3) Finally, internal evidence suggests that a few
passages are of a secondary character; i.e. interpolations by a writer
later than the Chronicler: such perhaps are 1 Chronicles vi. 50‒53;
viii. 29‒38; xxiv. 20‒31; 2 Chronicles xv. 16‒19 (see note verse 17);
xx. 33 (see note xvii. 6); xxxi. 17‒19.

Interpolations on a large scale are not likely to have been made.

Yet it must be borne in mind that Chronicles‒Ezra‒Nehemiah were
once a continuous work, and study of Ezra‒Nehemiah shows that
those writings have undergone a complex literary process, involving
serious omissions and transpositions. This heightens the possibility
that Chronicles also, before or after its separation from Ezra‒
Nehemiah, was treated with freedom. Thus “the recurrence of 1
Chronicles ix., Nehemiah xi. [both giving a list of inhabitants of
Jerusalem] in a single work hardly looks like an original feature; like
the more remarkable repetition of Ezra ii., Nehemiah vii., the feature
seems to point to the combination of sources which were primarily
distinct” (Cook, 1 Esdras, in Charles’ Apocrypha, p. 19, but see note
on ix. 17). On the other hand the homogeneity of style and purpose
in Chronicles tells strongly against the probability of large
interpolations, and it is reasonable to believe that in the present text
we have substantially the work produced by the Chronicler.
 (II) Authorship. Chronicles‒Ezra‒Nehemiah contain no hint
whatsoever of the name of their author, and external evidence fails
us equally. From the contents and tone of the work we can infer with
comparative certainty that he belonged to the Levitical order, and in
all probability was a member of one of the Levitical guilds of
musicians and singers (see, e.g. 2 Chronicles xxxiv. 12, note). His
character and conceptions can also be discerned from the nature of
his work. That he was a man of strong intellect and vivid imagination
is shown by his qualities as a narrator (see p. xxi) and by the
consistency and power with which the whole work has been
designed and carried through (see below, §§ 5, 6, 8). Beyond this it
is futile to conjecture.
 § 4. Contents
The books of Chronicles‒Ezra‒Nehemiah give a history of Israel
and its ancestors from Adam down to the conclusion of Nehemiah’s
activity on behalf of the post-exilic community in Jerusalem, circa
432 b.c. Of this history the two books of Chronicles cover the period
from Adam to the fall of Jerusalem in 586 b.c. ¹ Before giving a
detailed analysis, it may be of advantage to call attention to certain
outstanding features. Remark that (1) the traditions of the period
from Adam to Saul’s death have been compressed into a series of
genealogical lists which occupy chapters i.‒ix.; (2) the rest of the two
books gives an account of the history of Judah from the death of
Saul down to the destruction of Jerusalem by the Babylonians in
586, the fortunes of North Israel being ignored, except for some
scornful allusions to its degeneracy in comparison with Judah and
Jerusalem; (3) the keenest interest is displayed by the writer in all
matters connected with the Levites, the Temple, and its worship, so
that an extraordinary amount of space is allotted to those subjects—
note especially 1 Chronicles vi., xxii.‒xxix.; 2 Chronicles iii.‒vii.,
xxix.‒xxxi., xxxiv. 8‒xxxv. 19.

¹ Actually Chronicles concludes with the first three verses of

Ezra i., which relate the famous edict of Cyrus, permitting the
Jews to return to Jerusalem: see § 2 (1).

The following Table gives a general survey of the contents of 1

and 2 Chronicles ¹.
 ¹ The sign “=” in the Table means that the passage in
Chronicles is a transcription, with little or no change, of the
passage given in brackets; the sign “compare” is used where
the account in Chronicles differs considerably from or has
only a faint connection with that of the passage in brackets.

(A) 1 Chronicles i.‒ix. Introductory Genealogies.

i. 1‒4 (compare Genesis v. 3‒32). Genealogy
from Adam to the sons of Noah.
5‒23 (= Genesis x. 2‒29). The descendants of
Japheth, Ham, and Shem.
24‒28 Genealogy from Shem to Ishmael.
29‒31 (= Genesis xxv. 12‒16). Ishmaelite
descendants of Abraham.
32, 33 (= Genesis xxv. 1‒4). Arabian descendants
of Abraham.
34‒37 (compare Genesis xxxvi. 10‒14). Edomite
descendants of Abraham.
38‒42 (compare Genesis xxxvi. 20‒28).
Genealogy of the Horite inhabitants of
Seir.
43‒51a (compare Genesis xxxvi. 31‒39). The early
kings of Edom.
51b‒54 (compare Genesis xxxvi. 40‒43). The
“dukes” of Edom.
ii. 1, 2 (compare Genesis xxxv. 22b‒26). The sons
of Israel.
3‒iv. 23 Genealogies of the tribe of Judah:
ii. 3‒17 Descent of the sons of Jesse.
18‒55 Caleb. Jerahmeel. Caleb.
iii. 1‒9 (= 2 Samuel iii. 2‒5; v. 14‒16).
David’s sons.
10‒24 The Davidic Line before and after
 the Captivity.
iv. 1‒23 Additional genealogies of Judah.
iv. 24‒⁠v. 26 Genealogies of Simeon, Reuben, Gad, and
Manasseh.
vi. 1‒81 The tribe of Levi:
1‒3 Genealogy from Levi to Eleazar.
4‒15 The line of the high-priests to the
Captivity.
16‒30 The three clans of the Levites.
31‒47 The singers.
48‒53 Distinction between the sons of
Aaron and the rest of the
Levites.
54‒81 The cities of the Levites.
vii. 1‒40 Genealogies of Issachar, Zebulun (see note
on vii. 6), Naphtali, Manasseh,
Ephraim, and Asher.
viii. 1‒40 Genealogies of Benjamin:
1‒28 Various Benjamite families.
29‒40 (compare ix. 35‒44). The
genealogy of the house of
Saul.
ix. 1‒17 The heads of the families of Judah,
Benjamin, and Levi, which dwelt in
Jerusalem.
18‒34 The duties of porters and Levites.
35‒44 (= viii. 29‒38). The genealogy of the house
of Saul.
(B) x.‒xxix. David.
x. 1‒14 (= 1 Samuel xxxi. 1‒13). The death of Saul
at the battle of Gilboa.
xi. 1‒9 (= 2 Samuel v. 1‒10). Coronation of David
and capture of Jebus.
 10‒47 (compare 2 Samuel xxiii. 8‒39). David’s
mighty men.
xii. 1‒40 David’s adherents who brought him to the
kingdom.
xiii. 1‒14 (= 2 Samuel vi. 1‒11). The removal of the
Ark from Kiriath-jearim. Death of Uzza.
xiv. 1‒7 (compare 2 Samuel v. 11‒16). David’s sons
born in Jerusalem.
8‒17 (= 2 Samuel v. 17‒25). Two Philistine
attacks repulsed.
xv. 1‒24. Preparations for bringing home the Ark.
25‒⁠xvi. 6 (compare 2 Samuel vi. 12‒20). The Ark
brought into the city of David.
xvi. 7‒36 (= Psalms cv. 1‒15; xcvi. 1‒13; cvi. 1, 47,
48). David’s psalm of praise.
37‒43 Arrangements for daily worship.
xvii. 1‒27 (= 2 Samuel vii. 1‒29). Permission to build
a Temple refused to David.
xviii. 1‒17 (= 2 Samuel viii. 1‒18). David’s foreign
wars. His officials.
xix. 1‒xx. 8 (= 2 Samuel x. 1‒19; xi. 1; xii. 26‒31; xxi.
18‒22). Wars with Ammon, Syria, and
the Philistines.
xxi. 1‒30 (= 2 Samuel xxiv. 1‒25). The census and
the plague.
xxii. 1‒xxix. 20. David’s preparations for the building of the
Temple and for the establishment of its
services:
xxii. The choice of the Temple site. The
charge to Solomon.
xxiii. The organisation of the Levites.
xxiv. The divisions (“courses”) of the
Priests.
 xxv. The divisions of the singers.
xxvi. The divisions of the doorkeepers.
xxvii. Various officers of David.
xxviii. 1‒xxix. David’s charge to Solomon and to
20 all Israel.
xxix. 21‒30 The Epilogue: The Great Rejoicing; the
Anointing of Solomon; Summary of
David’s reign.
(C) 2 Chronicles i.‒ix. Solomon.
i. 1‒6 (compare 1 Kings iii. 4). Solomon’s sacrifice
at Gibeon.
7‒13 (= 1 Kings iii. 5‒15). The Vision and the
prayer for wisdom.
14‒17 (= 1 Kings x. 26‒29). Chariots and horses.
ii. 1, 2, 17, 18 (compare 1 Kings v. 15, 16). Bearers of
burdens and hewers of wood and
stone.
3‒16 (compare 1 Kings v. 2‒11). Negotiations
with Huram (Hiram) king of Tyre.
iii. 1‒v. 1 (compare 1 Kings vi. 1‒vii. 50) The building
and furnishing of the Temple.
v. 2‒14 (= 1 Kings viii. 1‒11). The bringing in of the
Ark and the descent of the cloud.
vi. 1‒11 (= 1 Kings viii. 12‒21). Solomon’s blessing.
12‒42 (= 1 Kings viii. 22‒50). Solomon’s prayer.
vii. 1‒3 The descent of the fire upon the sacrifices.
4‒10 (= 1 Kings viii. 62‒66). The final rejoicings.
11‒22 (= 1 Kings ix. 1‒9). The second Vision and
the acceptance of Solomon’s prayer.
viii. 1‒13, 17, (compare 1 Kings ix. 10‒28). Various Acts
18 of Solomon.
14‒16 Organisation of the Priests and Levites in
the Temple.
 ix. 1‒28 (= 1 Kings x. 1‒27). The Visit of the Queen
of Sheba. Solomon’s greatness.
29‒31 (= 1 Kings xi. 41‒43). The Epilogue.
(D) 2 Chronicles x.‒xxxvi. The Acts of the Kings of Judah.
x. 1‒xi. 4 (= 1 Kings xii. 1‒24). The Revolt of the Ten
Tribes.
xi. 5‒xii. 16 (compare 1 Kings xiv. 21‒31). The Acts of
Rehoboam.
xiii. 1‒22 (compare 1 Kings xv. 1‒8). The Acts of
Abijah (Abijam).
xiv. 1‒xvi. 14 (compare 1 Kings xv. 9‒24). The Acts of
Asa.
xvii. 1‒19 Jehoshaphat’s religious measures. His
captains.
xviii. 1‒34 (= 1 Kings xxii. 1‒37). Jehoshaphat with
Ahab at Ramoth-Gilead.
xix. 1‒xx. 30 Jehoshaphat’s judges. His victory in the
wilderness of Tekoa.
xx. 31‒37 (= 1 Kings xxii. 41‒49). The rest of the Acts
of Jehoshaphat.
xxi. 1‒20 (= 1 Kings xxii. 50; compare 2 Kings viii.
17‒24). Jehoram.
xxii. 1‒9 (= 2 Kings viii. 25‒29; compare ix. 27, 28).
Ahaziah.
10‒xxiii. 21 (= 2 Kings xi. 1‒20). The rise and fall of
Athaliah.
xxiv. 1‒14 (= 2 Kings xii. 1‒16). Restoration of the
Temple under Joash.
15‒22 Apostasy of the princes. Assassination of
the prophet Zechariah.
23‒27 (compare 2 Kings xii. 17‒21). The Syrian
War and the end of Joash.
xxv. 1‒13 (compare 2 Kings xiv. 1‒7). Amaziah. The
 Edomite War. The Ephraimite ravages.
14‒16 Apostasy of Amaziah.
17‒28 (= 2 Kings xiv. 8‒20). Capture of
Jerusalem. Death of Amaziah.
xxvi. 1‒23 (compare 2 Kings xv. 1‒7). Uzziah
(Azariah).
xxvii. 1‒9 (compare 2 Kings xv. 32‒38). Jotham.
xxviii. 1‒27 (compare 2 Kings xvi. 1‒20). Ahaz.
xxix. 1‒xxxi. 21 Hezekiah. Cleansing of the Temple. The
Great Passover. Care for the
priesthood.
xxxii. 1‒23 (compare 2 Kings xviii., xix.). The
deliverance from Sennacherib.
24‒33 (compare 2 Kings xx. 1‒21). Hezekiah’s
sickness. His death.
xxxiii. 1‒20 (compare 2 Kings xxi. 1‒18). Manasseh.
His captivity and repentance.
21‒25 (= 2 Kings xxi. 19‒26). Amon.
xxxiv. 1‒7 (compare 2 Kings xxii. 1, 2; xxiii. 4‒20).
Josiah. Removal of the emblems of
idolatry.
8‒28 (= 2 Kings xxii. 3‒20). Repair of the Temple.
Discovery of the Book of the Law.
29‒33 (= 2 Kings xxiii. 1‒3). Renewal of the
Covenant.
xxxv. 1‒19 (compare 2 Kings xxiii. 21‒23). Josiah’s
Great Passover.
20‒27 (compare 2 Kings xxiii. 28‒30a). The death
of Josiah.
xxxvi. 1‒4 (compare 2 Kings xxiii. 30b‒34). Jehoahaz.
5‒8 (compare 2 Kings xxiii. 36‒xxiv. 6).
Jehoiakim.
9, 10 (compare 2 Kings xxiv. 8‒17). Jehoiachin.
 11‒21 (compare 2 Kings xxiv. 18‒xxv. 21).
Zedekiah. The Captivity of Judah.
22, 23 (= Ezra i. 1‒3a). The decree of Cyrus.

It will be seen at a glance that large portions of earlier canonical

Scripture have been reproduced in Chronicles exactly or very
closely: viz.

Genesis x. 2‒29; xxv. 1‒4, 10‒16; xxxvi., passim.

1 Samuel xxxi.

2 Samuel v.‒viii.; x.; xxiii. 8‒xxiv. 25.

1 Kings iii. 4‒14; v.‒vii. (in part); viii.‒x.; xi. 41‒xii. 24; xiv. 21‒xv.
24 (in part); xxii. (in part).

2 Kings viii. 17‒29; xi., xii.; xiv. 1‒22; xv., xvi. (in part); xxi.‒xxiv.
(in part).

Ezra i. 1‒3.

As the foregoing list shows, Chronicles by no means includes all

the narrative of Samuel and Kings. In particular may be noted the
omission of any account of the early life of David (1 Samuel passim),
the Court History of David (2 Samuel xi.‒xx.), the history of Elijah
and Elisha (1 Kings xvii.‒xxi.; 2 Kings i. 1‒viii. 15), and the affairs of
the Northern Kingdom with a few exceptions.

On the other hand, Chronicles contains a great deal which is

either independent of or not immediately dependent on earlier books
of the Old Testament: note especially the opening nine chapters of
genealogies, the last seven chapters of 1 Chronicles, and many
passages, long and short, in 2 Chronicles x.‒xxxvi. The origin and
significance of this new material will be discussed in the section on
the Sources, § 5.
 § 5. The Sources
From what has been said in § 1 regarding the nature of ancient
historical writings it will be realised that a careful examination of the
material used in the compilation of Chronicles is a necessary
preliminary to the task of estimating the purpose and value of the
work in its final form. Only when the extent of the sources has been
determined can we say whether contributions made by the writer
who combined those sources into the existing work are so great or
so small that we ought to reckon him in the one case a narrator
whose personality must be seriously considered, or in the other a
mere copyist and compiler.

(1) In considering the material of Chronicles, it is convenient to

begin with those passages which seem to be copied or adapted from
earlier books of the Old Testament. That such passages are
numerous, and constitute a very large amount of 1 and 2 Chronicles
will be seen by a glance at the table of contents given in § 4.
Occasionally the Chronicler reproduced the canonical text verbatim,
but generally he introduced alterations, which were sometimes both
numerous and important. The discrepancies thus produced between
Chronicles and other parts of canonical Scripture presented a grave
difficulty to the older commentators, and the theory was put forward
that the Chronicler used, not the canonical books, but the still older
sources from which the canonical books themselves were built up
and to which they frequently refer. It was hoped thus to minimise the
divergences by supposing that the Chronicler had copied somewhat
different portions of these old sources, and had approached them
from a different standpoint. Not only was this hypothesis in the
highest degree improbable, but the reconciliation it was supposed to
effect is now recognised to be for the most part untenable. The
theory is finally discredited by the fact that these sources of the
canonical books always appear in Chronicles combined together in
precisely the same manner in which they are found combined in the
canonical books; i.e. they appear always ‘edited,’ and never in their
original, independent, form. It may be definitely asserted therefore
that for all the passages which are common to Chronicles and other
canonical works the Chronicler was indebted solely to the text of the
canonical books as it appeared in his time.

As for the divergences, real and apparent, between Chronicles

and other canonical Scriptures, it is now recognised that, whilst they
are properly a subject for historical investigation, they do not involve
a religious problem. The old “religious” difficulty is answered by a
deeper comprehension of the nature of Inspiration. The real
inspiration of the Scriptures does not, as was once thought, rest
upon points of historical accuracy: see the article Inspiration by A. E.
Garvie in the Encyclopedia Britannica¹¹, vol. xiv., especially pp. 647
ad fin., 648, with the references there given.

(2) More important and difficult is the problem of the source of the
new material in Chronicles. Nearly one-half of the two books of
Chronicles is material otherwise unknown to us, and not to be
regarded as mere ornamental amplification of the passages drawn
from canonical sources. Rather it is precisely these new parts which
give colour to the whole work, and there can be no doubt that the
Chronicler must have dwelt with special fondness on just these
passages. The question is, Can we discern or infer sources from
which these independent chapters and paragraphs have been
derived, or is the Chronicler himself their only source and origin?

In attempting to answer that question, our first task is to note and

discuss a long list of works to which the Chronicler appeals, either as
authorities for what he says or as sources where fuller information
might (presumably) be expected. They are as follows:

A. Those with specific prophetic titles.

[For the reigns of David and Solomon.]

 (1) The history (literally words, or acts) of Samuel the seer
(1 Chronicles xxix. 29).

(2) The history of Nathan the prophet (1 Chronicles xxix.

29; 2 Chronicles ix. 29).

(3) The history of Gad the seer (1 Chronicles xxix. 29).

(4) The last acts of David (1 Chronicles xxiii. 27). [Perhaps

the same as (5).]

(5) The chronicles (literally acts of the days) of king David

(1 Chronicles xxvii. 24).

(6) The prophecy of Ahijah the Shilonite (2 Chronicles ix.

29).

(7) The visions of Iddo the seer (2 Chronicles ix. 29).

[For the kings of Judah (excepting Jehoram, Ahaziah,

Athaliah, Amon, Jehoahaz, Jehoiachin, and
Zedekiah).]

(8) The histories of Shemaiah the prophet and of Iddo the

seer (2 Chronicles xii. 15).

(9) The commentary (literally Midrash ¹) of the prophet Iddo

(2 Chronicles xiii. 22).

(10) The history of Jehu the son of Hanani which is inserted

in the books of the kings of Israel (2 Chronicles xx.
34).

(11) A writing of Isaiah the prophet (2 Chronicles xxvi. 22).

(12) The vision of Isaiah the prophet the son of Amoz in the
books of the kings of Judah and Israel (2 Chronicles
xxxii. 34).
 (13) ? The history of Hozai (literally the seers) (2 Chronicles
xxxiii. 19).

(14) ? A genealogical register compiled in the time of

Jotham and Jeroboam II (1 Chronicles v. 17).

(15) A collection of “lamentations” (2 Chronicles xxxv. 25).

B. Those with general titles.

(1) A Book of the Kings of Judah and Israel (cited for the
reigns of Asa, Amaziah, Ahaz, and Hezekiah; 2
Chronicles xvi. 11, xxv. 26, xxviii. 26, xxxii. 32).
Compare (12) above.

(2) A Book of the Kings of Israel and Judah (cited for

Jotham, Josiah, and Jehoiakim; 2 Chronicles xxvii.
7, xxxv. 27, xxxvi. 8).

(3) A Book of the Kings of Israel (cited for genealogies, 1

Chronicles xix. 1; for the reign of Jehoshaphat, 2
Chronicles xx. 24; and for Manasseh, 2 Chronicles
xxxiii. 18).

(4) A Midrash ¹ of the Book of Kings (for the reign of Joash,

2 Chronicles xxiv. 27).
 ¹ The noun Midrash is derived from a verb meaning to search
out, explore. The word occurs only in 2 Chronicles xiii. 22,
xxiv. 27 in the Old Testament but is very common as a
description of many works of later Jewish literature. A
Midrash may be defined as an imaginative adaptation of an
idea suggested by Scripture, especially a homiletic exposition
or an edifying religious story. In midrashic writings “numbers
are multiplied, all the details assume huger and more exalted
proportions, right is always richly rewarded and wickedness
signally punished, miracles are common, and prophets and
kings deliver majestic, spiritual addresses, embodying the
best doctrines of later Judaism” (Kent, Student’s O.T. ii. p.
20). Several of the narratives in Chronicles partake of this
character. For later and more obvious examples, compare the
stories of Tobit and Susanna in the Apocrypha.

This great array of authorities dwindles to small proportions on

inspection. Of the fifteen given under A, numbers 13, 14 are
uncertain but of very small importance, whilst number 15 is also
unknown: it is not the canonical book of Lamentations (see the note
on 2 Chronicles xxxv. 25). The rest, numbers 1‒12, almost certainly
were not independent works, but simply sections of some
comprehensive work (see especially numbers 10 and 12), it being
the custom among the Jews to refer to the sections of a large work
by means of distinctive titles—compare Romans xi. 2, “Know ye not
what the Scripture saith in Elijah.” Thus some of these titles, e.g. 1,
2, 3, 6, 7, may refer simply to passages in the canonical books of
Samuel and Kings, numbers 11 and 12 to Isaiah xxxvi.‒xxxix. = 2
Kings xviii. 13‒xx. 19. But the others (and perhaps some also of
those just mentioned) in all probability denote sections of a large
history of a more or less midrashic character; and it is this work
apparently which is meant by the titles given under B. To these we
now turn. It is generally admitted that all four titles mentioned in B
denote one and the same work, a comprehensive history of the
Kingdoms of Judah and Israel. This work was not our canonical
books of Kings, for it is quoted as containing material not found in
those books. Still less was it any of the sources referred to in Kings:
there is not the faintest probability that any of the new material in
Chronicles was derived directly from those very old sources. The
question therefore is whether in this general work to which the
Chronicler appeals he had a source independent or semi-
independent of Kings. Opinion is divided. Some scholars think that it
was essentially dependent on the canonical Kings, merely “a
reconstructed history, embellished with marvellous tales of divine
interposition and prophetic activity.” Others maintain that this
midrashic history had its roots not only in canonical Kings but also in
traditions partly or wholly independent of Kings. The latter opinion is
here preferred, but the reasons for adopting it will be best seen if we
first state and consider two sharply opposing views put forward by
recent writers.

On the one side is Torrey (Ezra Studies, 1910) who argues that
the Chronicler had no source at all other than the canonical books—
all else was the product of his imaginative skill. He describes this
supposed midrashic history of Judah and Israel as “a phantom
source, of which the internal evidence is absolutely lacking, and the
external evidence is limited to the Chronicler’s transparent parading
of authorities.” The strength of Torrey’s contention lies in the fact that
almost all the additional matter in Chronicles is written in one and the
same distinctive style. That style has certain unmistakable
peculiarities. Thus Driver in the Encyclopedia Britannica s.v.
Chronicles, col. 772, writes, “It is not merely that the style of the
Chronicler presents characteristically late linguistic novelties ... but it
has also a number of special mannerisms.... So constant are [these
marks] that there is hardly a single sentence, not excerpted from
Samuel or Kings, in which they are not discernible.” On the other
side we have to consider the attitude adopted in the commentaries of
Benzinger (1901) and Kittel (1902), following up a suggestion made
by Büchler in 1899. These scholars not only believe that non-
canonical sources supplied much of the new material of Chronicles,
but they have attempted to analyse that material minutely into
various contributory elements. According to their view the Chronicler
was essentially a compiler, following his sources closely and
showing such little independence as he exercised chiefly in those
verses and passages where the affairs and interests of the Levites