Radiative Transfer An Introduction To Exact And

Asymptotic Methods Hlne Frisch download

https://ebookbell.com/product/radiative-transfer-an-introduction-
to-exact-and-asymptotic-methods-hlne-frisch-51985332

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Radiative Transfer In Coupled Environmental Systems An Introduction To

Forward And Inverse Modeling 1st Edition Knut Stamnes

https://ebookbell.com/product/radiative-transfer-in-coupled-
environmental-systems-an-introduction-to-forward-and-inverse-
modeling-1st-edition-knut-stamnes-5318032

An Introduction To Radiative Transfer 1st Annamaneni Peraiah

https://ebookbell.com/product/an-introduction-to-radiative-
transfer-1st-annamaneni-peraiah-1460988

Remote Sensing Of The Environment And Radiation Transfer An

Introductory Survey 1st Edition Anatoly Kuznetsov

https://ebookbell.com/product/remote-sensing-of-the-environment-and-
radiation-transfer-an-introductory-survey-1st-edition-anatoly-
kuznetsov-2519296

Radiative Transfer In Participating Media Rahul Yadav C Balaji

https://ebookbell.com/product/radiative-transfer-in-participating-
media-rahul-yadav-c-balaji-45334314
Radiative Transfer In The Atmosphere And Ocean Second Knut Stamnes

https://ebookbell.com/product/radiative-transfer-in-the-atmosphere-
and-ocean-second-knut-stamnes-9950782

Radiative Transfer In Stellar And Planetary Atmospheres Lucio

Crivellari

https://ebookbell.com/product/radiative-transfer-in-stellar-and-
planetary-atmospheres-lucio-crivellari-10679910

Radiative Transfer In The Atmosphere And Ocean Gary E Thomas

https://ebookbell.com/product/radiative-transfer-in-the-atmosphere-
and-ocean-gary-e-thomas-1659948

3d Radiative Transfer In Cloudy Atmospheres Physics Of Earth And Space

Environments 1st Edition Alexander Marshak

https://ebookbell.com/product/3d-radiative-transfer-in-cloudy-
atmospheres-physics-of-earth-and-space-environments-1st-edition-
alexander-marshak-2115158

Heat Transfer Volume 2 Radiative Transfer Michel Ledoux Abdelkhalak El

Hami

https://ebookbell.com/product/heat-transfer-volume-2-radiative-
transfer-michel-ledoux-abdelkhalak-el-hami-46563032
Hélène Frisch

Radiative
Transfer
An Introduction to Exact
and Asymptotic Methods
Radiative Transfer
Hélène Frisch

Radiative Transfer
An Introduction to Exact and Asymptotic
Methods
Hélène Frisch
Observatoire de la Côte d’Azur, CNRS,
Laboratoire Lagrange
Université Côte d’Azur
Nice Cedex 04, France

ISBN 978-3-030-95246-4 ISBN 978-3-030-95247-1 (eBook)

https://doi.org/10.1007/978-3-030-95247-1

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2022
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, expressed 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
To Uriel, Anne, and Thomas
Preface

Transport of particles such as photons, neutrons, electrons, and molecules through

a host medium usually takes place by multiple (repeated) scatterings. This book
presents methods for constructing exact solutions of stationary radiative transfer
problems related to the transport of photons in a stellar or planetary atmosphere
or any foggy medium. When the scatterings do not modify the physical properties
of the host medium, the equations describing the transport of particles are linear.
In spite of this, neither the construction of exact solutions nor numerical solutions
are straightforward, because of the nonlocal character of multiple scattering. Exact
solutions provide trustful descriptions of the physical phenomenon and can serve as
testing ground for numerical methods of solutions.
Exact solutions, that is solutions which can be expressed as known functions
or quadratures, can be constructed for a few, but widely used, types of scattering
processes and geometries of the host medium. Scattering processes allowing exact
solutions discussed in this book are monochromatic scattering, complete frequency
redistribution, and Rayleigh scattering. Monochromatic scattering and Rayleigh
scattering (which accounts for the linear polarization and blue color of the sky)
govern the formation of continuous spectra. Photons change directions at each
scattering but keep the same frequency. Complete frequency redistribution (an
idiomatic expression of the astrophysical literature) governs the formation of
most atomic spectral lines. Photons do undergo changes of direction, but also of
frequencies, in an uncorrelated way, within the frequency range of the spectral line.
Concerning the host medium, the spreading of a source of light by a large number
of scatterings can be studied by assuming a medium of infinite extension. A standard
Fourier transformation can then provide the relevant solution. In astrophysics, and
also in nuclear reactors, a very important question is how a radiation field will
be reflected and transmitted through a scattering medium, such as a stellar or
planetary atmosphere. A closely related question is how photons created inside these
atmospheres will escape the medium, and what will be their observable distribution
in directions and frequencies. A reasonably good model for a stellar atmosphere is
a semi-infinite medium. Finding exact solutions is possible but rather hard. The first
exact solution of a transport problem in a semi-infinite plane-parallel medium was

vii
viii Preface

constructed by (Wiener and Hopf 1931) for the Milne problem, which describes the
temperature profile of a stellar atmosphere in radiative equilibrium. Their method
of solution, which now carries their name, is one the greatest achievements in
mathematical physics in the first half of this century. It relies on the properties of
analytic functions in the complex plane. In contrast, when the question is posed to
find the distribution of photons that have been reflected or transmitted through a
scattering medium of finite extension, a slab for example, then there is no explicit
expression for the radiation field, inside or outside the medium. However, there
are many interesting exact relations and integral equations, which can be solved
numerically. This topic is only briefly mentioned in this book.
Following the work of Wiener and Hopf, several methods have been developed
to tackle semi-infinite medium problems. One of the best known method in
astrophysics is the construction of a nonlinear integral equation for the famous H -
function, a sort of special function for half-space transport problems. Proposed for
the first time by Ambartsumian (1942), it largely avoids complex plane analysis and
relies on invariance properties of the radiative transfer process. Plasma physicists
became strongly interested in the subject in relation to the transport of neutrons
in nuclear reactors. New methods were developed, such as the Case singular
eigenfunction expansion method (Case 1960), which leads to linear singular integral
equations with Cauchy-type kernels. Methods of solutions for these equations,
also based on complex plane analysis, rely largely on a technique introduced by
Carlemann (1922) and developed by Muskhelishvili (1953). These methods lead
to boundary value problems in the complex plane, known as Riemann–Hilbert
problems, and provide explicit expressions for the H -function, as does the Wiener–
Hopf method.
A key motivation for writing this book was to present in the same volume several
of the methods leading to exact results for semi-infinite media, usually presented
in separate books or articles, for both scalar and polarized radiation fields, and also
connections, that are not always clearly apparent, between the various methods.
When the number of scatterings in the host medium is very large, asymptotic
techniques may be employed to analyze the large-scale behavior of the random
walk of the photons and to explain, for example, why monochromatic scattering has
all the characteristics of an ordinary diffusion process, while complete frequency
redistribution of spectral lines has in contrast those of a Lévy walk. This asymptotic
approach is presented in Part III of the book, while exact methods are presented in
Parts I and II. Part I deals with unpolarized radiation and Part II with monochro-
matic radiation, polarized by Rayleigh scattering, and spectral lines, polarized by
resonance scattering and the Hanle effect, which is a modification of resonance
polarization by a weak magnetic field. Details on the organization of each part are
given in Chap. 1.
The intended readership for the book ranges from first-year graduate students
to professional scientists. Astrophysicists can find in this book exact methods of
solutions used in radiative transfer, but also applied in distant fields, such as financial
mathematics. A major effort has been made in the organization of the material
to give a synthetic view on exact and asymptotic methods in radiative transfer.
Preface ix

Although oriented towards methods of solutions, this book also provides exact
expressions for the radiation field intensity and polarization for a number of standard
problems.
I am deeply grateful to many colleagues for their help and encouragement in
this undertaking. My sincere thanks go particularly to V. Bommier, V. V. Ivanov,
B. Rutily, and P. Zweifel, who have read significant parts of the book and have
taken the time to share with me their intimate knowledge of the field. I have adopted
many of their suggestions. My thanks go also to my family, especially to my mother
D. Piron-Lévy, for lasting encouragements; to my husband, Uriel, for reading the
full manuscript and playing the role of the non-specialist; and to L. Anusha, E. Lega,
and M. Sampoorna for their very generous help in the preparation of the figures. This
book would not exist without the continuous support of Springer teams in Europe
and India. I owe them a thousands thanks.

Nice Cedex 04, France Hélène Frisch

References

Ambartsumian, V.A.: Light scattering by planetary atmospheres. Astron. Zhurnal

19, 30–41 (1942)
Carleman, T.: Sur la résolution de certaines équations intégrales. Ark. Mat. Astr.
Fys. 16, 1–19 (1922)
Case, K.M.: Elementary solutions of the transport equation and their applications.
Ann. Phys. (New York) 9, 1–23 (1960)
Muskhelishvili, N.I.: Singular Integral Equations. Noordhoff, Groningen (Based on
the Second Russian Edition Published in 1946) (1953); Dover Publications
(1991)
Wiener, N., Hopf, D.: Über eine Klasse singulärer Integralgleichungen, Sitzungs-
berichte der Preussichen Akademie, Mathematisch-Physikalische Klasse 3. Dez.
1931, vol. 31, pp. 696–706 (ausgegeben 28. Januar 1932) (1931); English
translation: In: Paley, R.C., Wiener, N., Fourier transforms in the Complex
Domain. Am. Math. Soc. Coll. Publ., vol. XIX, pp. 49–58 (1934)
Contents

1 An Overview of the Content .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1

1.1 Part I: Scalar Radiative Transfer Equations . . . . .. . . . . . . . . . . . . . . . . . . . 1
1.2 Part II: Scattering Polarization .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 4
1.3 Part III: Asymptotic Properties of Multiple Scattering . . . . . . . . . . . . . 6
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 8

Part I Scalar Radiative Transfer Equations

2 Radiative Transfer Equations . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 11
2.1 The Integro-Differential Radiative Transfer Equations .. . . . . . . . . . . . 12
2.1.1 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 17
2.1.2 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 18
2.1.3 The Diffuse Radiation Field. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 22
2.2 Integral Equations for the Source Function . . . . .. . . . . . . . . . . . . . . . . . . . 23
2.2.1 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 23
2.2.2 The Milne Problem . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 25
2.2.3 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 27
2.3 Neumann Series Expansion .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 30
2.4 The Green Function and Associated Functions .. . . . . . . . . . . . . . . . . . . . 32
2.4.1 Some Definitions . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 32
2.4.2 A Lemma on Wiener–Hopf Equations.. . . . . . . . . . . . . . . . . . . . 34
2.4.3 The Source Function and the Resolvent Function . . . . . . . . 35
2.4.4 Some Properties of the Green Function . . . . . . . . . . . . . . . . . . . 36
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 37
3 Exact Methods of Solution: A Brief Survey . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 41
3.1 The Infinite Medium and the Dispersion Function . . . . . . . . . . . . . . . . . 42
3.2 Exact Methods for a Semi-Infinite Medium . . . .. . . . . . . . . . . . . . . . . . . . 43
3.2.1 The Wiener–Hopf Method . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 43
3.2.2 Traditional Real-Space Methods .. . . . . .. . . . . . . . . . . . . . . . . . . . 44
3.2.3 The Singular Integral Equation Approach .. . . . . . . . . . . . . . . . 45
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 47

xi
xii Contents

4 Singular Integral Equations . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 51

4.1 The Inverse Laplace Transformation . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 52
4.1.1 The Half-Space Case . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 52
4.1.2 The Full-Space Case. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 54
4.2 The Direct Laplace Transformation . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 55
4.3 The Hilbert Transform Method of Solution .. . . .. . . . . . . . . . . . . . . . . . . . 57
4.3.1 A Special Case . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 58
4.3.2 The General Case . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 59
4.3.3 Application to Radiative Transfer.. . . . .. . . . . . . . . . . . . . . . . . . . 61
Appendix A: Hilbert Transforms .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
A.1 Definition and Properties.. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 63
A.2 The Plemelj Formulae .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 64
A.3 The Plemelj Formulae for a Dirac Distribution .. . . . . . . . . . . . . . . . . . . . 66
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 67
5 The Scattering Kernel and Associated Auxiliary Functions . . . . . . . . . . . 69
5.1 The Kernel and Its Inverse Laplace Transform .. . . . . . . . . . . . . . . . . . . . 69
5.2 The Dispersion Function .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 73
5.2.1 Symmetries and Analyticity Properties .. . . . . . . . . . . . . . . . . . . 73
5.2.2 The Zeroes of the Dispersion Function .. . . . . . . . . . . . . . . . . . . 78
5.2.3 The Index κ . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
5.3 The Half-Space Auxiliary Function X(z) . . . . . . .. . . . . . . . . . . . . . . . . . . . 80
5.4 Some Properties of X(z) . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
5.4.1 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 83
5.4.2 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 84
5.4.3 Some Properties of the H -Function . . .. . . . . . . . . . . . . . . . . . . . 85
Appendix B: Properties of the Half-Space Auxiliary Function .. . . . . . . . . . . 88
B.1 Factorization Relation .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 89
B.2 Identities for the Boundary Values of X(z) . . . . .. . . . . . . . . . . . . . . . . . . . 89
B.3 Integral Equations for X(z) . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
B.4 Values of X(z) and H (z) . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 92
B.5 Moments of X(z) and H (z) . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 94
6 The Surface Green Function and the Resolvent Function . . . . . . . . . . . . . 97
6.1 Infinite Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 98
6.1.1 The Fourier Transform Method .. . . . . . .. . . . . . . . . . . . . . . . . . . . 98
6.1.2 The Inverse Laplace Transform Method .. . . . . . . . . . . . . . . . . . 101
6.2 Semi-infinite Medium. The Inverse Laplace
Transform Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 102
6.2.1 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 103
6.2.2 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 105
6.3 The H -Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 108
Appendix C: The Direct Laplace Transform Method . .. . . . . . . . . . . . . . . . . . . . 110
C.1 Complete Frequency Redistribution .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 111
C.2 Monochromatic Scattering .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 113
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 114
Contents xiii

7 The Emergent Intensity and the Source Function . .. . . . . . . . . . . . . . . . . . . . 115

7.1 The Emergent Intensity . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 115
7.2 The Source Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 118
7.2.1 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 119
7.2.2 Monochromatic Scattering. A Fourier Inversion .. . . . . . . . . 123
7.3 Some Standard Results . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 126
7.3.1 Uniform Primary Source Term .. . . . . . . .. . . . . . . . . . . . . . . . . . . . 126
7.3.2 Exponential Primary Source Term . . . . .. . . . . . . . . . . . . . . . . . . . 128
7.3.3 Diffuse Reflection . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 130
Appendix D: The Monochromatic Source Function . . .. . . . . . . . . . . . . . . . . . . . 134
D.1 The Inverse Laplace Transform Method . . . . . . . .. . . . . . . . . . . . . . . . . . . . 135
D.2 The Direct Laplace Transform Method . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 139
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 142
8 Spectral Line with Continuous Absorption . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 143
8.1 The Radiative Transfer Equation . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 144
8.2 Properties of the Auxiliary Functions . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 147
8.2.1 The Dispersion Function . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 147
8.2.2 The Half-Space Auxiliary Function . . .. . . . . . . . . . . . . . . . . . . . 149
8.3 Some Exact Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 149
8.3.1 The Resolvent Function . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 150
8.3.2 The Line Source Function .. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 151
8.3.3 The Emergent Intensity.. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 154
Appendix E: Kernels for Spectral Lines with a Continuous
Absorption .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 155
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 158
9 Conservative Scattering: The Milne Problem . . . . . . .. . . . . . . . . . . . . . . . . . . . 159
9.1 The Monochromatic Milne Problem . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 160
9.1.1 The Auxiliary Functions . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 162
9.1.2 The Source Function and the Emergent Intensity .. . . . . . . . 163
9.1.3 The Hopf Function . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 165
9.2 The Milne Problem for a Spectral Line . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 167
Appendix F: The Conservative Auxiliary Functions. Complete
Frequency Redistribution . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 169
F.1 Asymptotic Behavior of the Kernel . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 170
F.2 The Dispersion Function and the Half-Space Auxiliary
Function Near the Origin . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 171
Appendix G: The Milne Problem with the Inverse Laplace
Transform Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 174
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 176
10 The Case Eigenfunction Expansion Method. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 179
10.1 The Eigenfunctions and Eigenvalues .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 180
10.1.1 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 180
10.1.2 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 182
xiv Contents

10.2 The Diffuse Reflection Problem .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 184

10.2.1 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 184
10.2.2 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 188
10.3 Further Applications of the Case Method .. . . . . .. . . . . . . . . . . . . . . . . . . . 190
10.3.1 A Complete Redistribution Full-Space Problem .. . . . . . . . . 190
10.3.2 A Half-Space Problem with Internal Sources . . . . . . . . . . . . . 192
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 194
√
11 The -law √and the Nonlinear H -Equation . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 195
11.1 The -law .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 195
11.1.1 A Quadratic Approach . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 197
11.1.2 A Green Function Approach . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 198
11.2 Construction of the Nonlinear H -Equation . . . . .. . . . . . . . . . . . . . . . . . . . 200
11.2.1 Scattering Coefficient Method . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 200
11.2.2 Exponential Primary Source Method . .. . . . . . . . . . . . . . . . . . . . 204
11.3 Some Properties of the H -Equation .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 206
11.3.1 A Factorization Relation . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 206
11.3.2 Uniqueness and Alternative Nonlinear H -Equations . . . . . 207
Appendix H: An Elementary Construction of the H -Function . . . . . . . . . . . . 211
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 213
12 The Wiener–Hopf Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 215
12.1 An Overview of the Wiener–Hopf Method . . . . .. . . . . . . . . . . . . . . . . . . . 216
12.2 Full Description of the Wiener–Hopf Method . .. . . . . . . . . . . . . . . . . . . . 220
12.2.1 Decomposition Formula .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 220
12.2.2 Analyticity of Complex Fourier Transforms . . . . . . . . . . . . . . 222
12.2.3 Properties of the Dispersion Function V (z) . . . . . . . . . . . . . . . 224
12.2.4 Factorization of the Dispersion Function .. . . . . . . . . . . . . . . . . 224
12.2.5 Decomposition of the Inhomogeneous Term .. . . . . . . . . . . . . 228
12.2.6 Fourier Transform of the Source Function . . . . . . . . . . . . . . . . 229
12.3 The Resolvent Function .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 231
12.4 The Milne Problem .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 233
12.5 The Wiener–Hopf Method for Spectral Lines . .. . . . . . . . . . . . . . . . . . . . 237
12.5.1 Factorization of V (z) . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 238
12.5.2 The Fourier Transform of the Source Function . . . . . . . . . . . 239
12.5.3 The Emergent Intensity.. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 240
12.6 The Sommerfeld Half-Plane Diffraction Problem . . . . . . . . . . . . . . . . . . 241
Appendix I: The Auxiliary Function Vu (z) . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 246
I.1 Explicit Expression of Vu (z). . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 247
I.2 The Auxiliary Functions Vu (z) and X(z) . . . . . . .. . . . . . . . . . . . . . . . . . . . 248
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 250

Part II Scattering Polarization

13 The Scattering of Polarized Radiation . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 255
13.1 Description of a Polarized Radiation .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 256
Contents xv

13.2 The Rayleigh Scattering Phase Matrix . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 261

13.2.1 The Scattering of a Collimated Radiation Beam .. . . . . . . . . 262
13.2.2 The Rayleigh Phase Matrix in a Fixed
Reference Frame .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 264
13.3 Resonance Scattering of Spectral Lines . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 271
13.3.1 Populations and Coherences . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 272
13.3.2 The Scattering Phase Matrix . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 274
13.4 The Hanle Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 277
13.4.1 Some General Properties . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 277
13.4.2 The Hanle Phase Matrix .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 279
Appendix J: Spectral Details of Resonance Scattering .. . . . . . . . . . . . . . . . . . . . 282
Appendix K: Spectral Details of the Hanle Effect.. . . . .. . . . . . . . . . . . . . . . . . . . 285
K.1 Magnetic Field Reference Frame . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 286
K.2 Atmospheric Reference Frame . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 289
K.3 Approximations for the Redistribution Matrix ... . . . . . . . . . . . . . . . . . . . 289
Appendix L: The Hanle Effect with the Classical Harmonic
Oscillator Model.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 290
L.1 The Harmonic Oscillator Model .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 290
L.2 Polarization Direction and Polarization Rate . . .. . . . . . . . . . . . . . . . . . . . 293
L.3 Frequency Redistribution . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 295
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 300
14 Polarized Radiative Transfer Equations . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 305
14.1 Rayleigh Scattering. The Radiative Transfer Equation .. . . . . . . . . . . . 307
14.2 Conservative Rayleigh Scattering . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 308
14.2.1 The Polarized Milne Problem .. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 309
14.2.2 The Diffuse Reflection Problem . . . . . . .. . . . . . . . . . . . . . . . . . . . 311
14.3 Rayleigh Scattering. A (KQ) Expansion of the Stokes
Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 312
14.3.1 Cylindrically Symmetric Radiation Field . . . . . . . . . . . . . . . . . 312
14.3.2 Azimuthally Dependent Radiation Field . . . . . . . . . . . . . . . . . . 316
14.4 Resonance Polarization of Spectral Lines . . . . . . .. . . . . . . . . . . . . . . . . . . . 321
14.4.1 One-Dimensional Medium .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 323
14.4.2 Multi-Dimensional Medium . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 327
14.5 The Hanle Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 330
14.5.1 The Hanle Redistribution Matrix . . . . . .. . . . . . . . . . . . . . . . . . . . 331
14.5.2 One-Dimensional Medium .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 334
14.5.3 Multi-Dimensional Medium . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 338
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 339
√
15 The -Law, the Nonlinear H-Equation, and Matrix
Singular Integral Equations . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 341
15.1 The Source
√ Vector: Its Derivative and Matrix Representation . . . . . 342
15.2 The -Law .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 344
15.2.1 Rayleigh Scattering and Resonance Polarization . . . . . . . . . 345
15.2.2 The Hanle Effect .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 347
xvi Contents

15.3 The Green Matrix and the Resolvent Matrix .. . .. . . . . . . . . . . . . . . . . . . . 349

15.3.1 Rayleigh Scattering and Resonance Polarization . . . . . . . . . 349
15.3.2 The Hanle Effect .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 351
15.4 Construction of the H-Equation . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 352
15.4.1 Rayleigh Scattering and Resonance Polarization . . . . . . . . . 353
15.4.2 The Hanle Effect .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 355
15.5 Alternative Definitions of the H-Matrix .. . . . . . . .. . . . . . . . . . . . . . . . . . . . 356
15.5.1 Exponential Primary Source . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 357
15.5.2 Uniform Primary Source . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 357
15.6 Singular Integral Equations .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 359
15.6.1 Rayleigh Scattering and Resonance Polarization . . . . . . . . . 360
15.6.2 The Hanle Effect .. . . . . .√ . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 361
15.7 Factorization Relations and the -Law .. . . . . . .. . . . . . . . . . . . . . . . . . . . 362
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 364
16 Conservative Rayleigh Scattering: Exact Solutions.. . . . . . . . . . . . . . . . . . . . 365
16.1 Radiative Transfer Equations . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 366
16.2 The Dispersion Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 369
16.3 Construction of the Auxiliary X-matrix . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 370
16.4 The Resolvent Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 373
16.5 The H-matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 377
16.6 The Polarized Milne Problem.. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 378
16.6.1 Some Physical Properties . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 379
16.6.2 The Source Vector . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 381
16.6.3 The Generalized Hopf Functions . . . . . .. . . . . . . . . . . . . . . . . . . . 382
16.6.4 The Emergent Radiation Field . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 385
16.6.5 The Stokes Parameters I and Q . . . . . . .. . . . . . . . . . . . . . . . . . . . 386
16.7 The Diffuse Reflection Problem .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 389
16.7.1 The Emergent Radiation Field . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 390
16.7.2 The Source Vector and the Radiation Field at Infinity .. . . 392
Appendix M: Properties of the Auxiliary Functions . . .. . . . . . . . . . . . . . . . . . . . 394
M.1 The Functions L1 (z) and L2 (z) . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 395
M.2 Some Properties of the Matrix L(ν) . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 397
M.3 The Auxiliary Functions X1 (z) and X2 (z) . . . . . .. . . . . . . . . . . . . . . . . . . . 398
M.3.1 Construction of X1 (z) and X2 (z) . . . . . .. . . . . . . . . . . . . . . . . . . . 398
M.3.2 Factorizations Relations . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 399
M.3.3 Nonlinear Integral Equations.. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 400
M.3.4 The Functions Hl (z) and Hr (z) . . . . . . . .. . . . . . . . . . . . . . . . . . . . 401
M.3.5 Moments of Hl (μ) and Hr (μ) . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 402
Appendix N: The Milne Integral Equation . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 404
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 409
17 Scattering Problems with No Exact Solution I: The Auxiliary
Matrices.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 411
17.1 Non-Conservative Rayleigh Scattering.. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 412
17.1.1 The Dispersion Matrix . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 413
17.1.2 The Dispersion Function . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 414
Contents xvii

17.1.3 The Auxiliary X-Matrix . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 417

17.1.4 A Factorization of the Dispersion Matrix . . . . . . . . . . . . . . . . . 419
17.2 Resonance Polarization . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 421
17.2.1 The Dispersion Matrix . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 422
17.2.2 The Auxiliary X-Matrix . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 423
17.3 The Hanle Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 424
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 426
18 Scattering Problems with No Exact Solution II: The
Resolvent Matrix, the H-Matrix, and the I-Matrix... . . . . . . . . . . . . . . . . . . . 427
18.1 Non-Conservative Rayleigh Scattering.. . . . . . . . .. . . . . . . . . . . . . . . . . . . . 428
18.1.1 The Resolvent Matrix . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 428
18.1.2 The H-Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 432
18.1.3 The H-Equation in the Complex Plane . . . . . . . . . . . . . . . . . . . . 434
18.1.4 Uniqueness of the Solution of the H-Equation .. . . . . . . . . . . 436
18.1.5 An Alternative H-Equation .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 437
18.1.6 The Emergent Radiation Field and the I-Matrix . . . . . . . . . . 439
18.2 Resonance Polarization . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 442
18.2.1 The Resolvent Matrix . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 443
18.2.2 The H-Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 444
18.2.3 Nonlinear Integral Equations.. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 445
18.3 The Hanle Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 447
18.3.1 The H-Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 448
18.3.2 Nonlinear Integral Equations.. . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 449
Appendix O: The Singular Integral Equation for G̃(p) . . . . . . . . . . . . . . . . . . . . 451
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 455

Part III Asymptotic Properties of Multiple Scattering

19 Asymptotic Properties of the Scattering Kernel K(τ ) .. . . . . . . . . . . . . . . . . 459
19.1 The Monochromatic Kernel . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 461
19.2 The Complete Frequency Redistribution Kernel.. . . . . . . . . . . . . . . . . . . 461
19.2.1 Asymptotic Behavior at Large Optical Depths .. . . . . . . . . . . 462
19.2.2 The Inverse Laplace Transform Near the Origin .. . . . . . . . . 463
19.2.3 The Fourier Transform Near the Origin . . . . . . . . . . . . . . . . . . . 464
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 466
20 Large Scale Radiative Transfer Equations . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 467
20.1 Asymptotic Analysis of the Source Function . . .. . . . . . . . . . . . . . . . . . . . 468
20.1.1 Monochromatic Scattering . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 470
20.1.2 Complete Frequency Redistribution . . .. . . . . . . . . . . . . . . . . . . . 470
20.1.3 Higher Order Terms . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 472
20.2 Asymptotic Analysis in Fourier Space . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 473
20.3 The Thermalization Length .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 474
20.4 Ordinary and Anomalous Diffusion .. . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 477
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 478
xviii Contents

21 The Photon Random Walk . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 479

21.1 The Mean Displacement After n Steps . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 481
21.1.1 Mean Displacement: Method I . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 481
21.1.2 Mean Displacement: Method II . . . . . . . .. . . . . . . . . . . . . . . . . . . . 486
21.1.3 Application to Normal Diffusion and Lévy Walks . . . . . . . . 487
21.2 The Mean Positive Maximum After n Steps . . . .. . . . . . . . . . . . . . . . . . . . 491
21.2.1 A Wiener–Hopf Integral Equation .. . . .. . . . . . . . . . . . . . . . . . . . 491
21.2.2 A Precise Asymptotics √ for the Mean Maximum . . . . . . . . . . 494
21.3 A Probabilistic Proof of the -Law.. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 499
Appendix P: Universality of Escape from a Half-Space . . . . . . . . . . . . . . . . . . . 502
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 505
22 Asymptotic Behavior of the Resolvent Function . . . .. . . . . . . . . . . . . . . . . . . . 507
22.1 The Infinite Medium Resolvent Function . . . . . . .. . . . . . . . . . . . . . . . . . . . 507
22.1.1 Infinite Medium with a Plane Source ... . . . . . . . . . . . . . . . . . . . 508
22.1.2 Infinite Medium with a Point Source . .. . . . . . . . . . . . . . . . . . . . 511
22.2 Semi-Infinite Medium .. . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 513
Reference .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 514
23 The Asymptotics of the Diffusion Approximation . .. . . . . . . . . . . . . . . . . . . . 515
23.1 The Rescaled Equation .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 516
23.2 The Interior Radiation Field . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 517
23.3 An Improved Diffusion Equation .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 520
23.4 The Boundary Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 522
23.4.1 Boundary Conditions for the Interior Solution .. . . . . . . . . . . 524
23.4.2 The Emergent Intensity.. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 525
23.5 The Eddington Approximations . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 526
23.6 Determination of the Critical Size . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 529
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 531
24 The Diffusion Approximation for Rayleigh Scattering .. . . . . . . . . . . . . . . . 533
24.1 The Radiative Transfer Equation . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 534
24.2 The Interior Radiation Field Expansion .. . . . . . . .. . . . . . . . . . . . . . . . . . . . 536
24.3 The Boundary Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 539
24.3.1 The Boundary Layer as a Diffuse Reflection Problem.. . . 539
24.3.2 Matching of the Interior and Boundary Layer Fields . . . . . 541
24.4 The Emergent Radiation Field . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 543
24.4.1 The Polarization Rate. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 544
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 548
25 Anomalous Diffusion for Spectral Lines . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 549
25.1 Interior and Boundary Layer Expansions . . . . . . .. . . . . . . . . . . . . . . . . . . . 550
25.2 The Radiation Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 552
25.3 Resonance Polarization . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 553
25.4 Scaling Laws for a Slab .. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 556
25.5 Mean Number of Scatterings and Mean Path Length . . . . . . . . . . . . . . 558
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 562
Contents xix

26 Asymptotic Results for Partial Frequency Redistribution . . . . . . . . . . . . . 563

26.1 RI Asymptotic Behavior . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 565
26.2 RIII Asymptotic Behavior .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 568
26.3 RV Asymptotic Behavior . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 569
26.4 RII Asymptotic Behavior.. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 572
26.4.1 Diffusion Equation for the Source Function . . . . . . . . . . . . . . 574
26.4.2 Diffusion Equation for the Radiation Field.. . . . . . . . . . . . . . . 576
26.4.3 RII Scaling Laws .. . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 578
References .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 580

Author Index.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 583

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 587
List of Figures

Fig. 2.1 Left panel: the space variable τ and the inclination angle
θ of a ray with direction n. The axis z is normal to the
surface of the medium. Right panel: the heliocentric
interpretation of the angle θ . The Sun is viewed from the
right. Observations at disk center correspond to θ = 0 and
those at the limb to θ = π/2 . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 14
Fig. 2.2 The Doppler profile, and several Voigt profiles with
different values of the parameter a. The left panel shows
ϕ(x) and the right panel log ϕ(x). In the right panel, the
Lorentzian wing regime characterized by ϕ(x) ∼ 1/x 2
can be observed for |x| larger than (− ln a)1/2 .. . . . . . . . . . . . . . . . . . . . 21
Fig. 2.3 Complete frequency redistribution. The integration
domains (ξ, x) and (x, ξ ), with ξ = μ/ϕ(x) . . .. . . . . . . . . . . . . . . . . . . . 29
Fig. A.1 Integration contour for the Plemelj formulae. The
contribution from the singular point is obtained by letting
ρ → 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 65
Fig. 5.1 The function g(ξ ) for the Doppler profile and the Lorentz
profile, 1/(π(1 + x 2 )), in linear scales in the upper panel
and in log-log scales in the lower panel. The value of
g(ξ ) is constant for ξ ≤ 1/ϕ(0). The algebraic behavior
of g(ξ ) for ξ → ∞ given in Eqs. (5.8) and (5.9) can be
observed in the lower panel. The Voigt profile has the
same algebraic behavior as the Lorentz profile .. . . . . . . . . . . . . . . . . . . . 71
Fig. 5.2 The function k(ν) for the Doppler profile and the Lorentz
profile, 1/(π(1 + x 2 )), in linear scales in the upper panel
and in log-log scales in the lower panel. The angular
point is located at ν = ϕ(0). For the Doppler and Lorentz
profiles, k(ν) tends to zero as 1/ν for ν → ∞. For
ν → 0, one can observe in the lower panel the algebraic
behaviors given in Eqs. (5.8) and (5.9). The Voigt profile
has the same algebraic behaviors as the Lorentz profile . . . . . . . . . . . 72

xxi
xxii List of Figures

Fig. 5.3 Branch cuts of the monochromatic dispersion function.

The solid black lines on the real axis show the branch cuts
of L(z) . For complete frequency redistribution, L(z) is
singular along the full real axis, except at the origin . . . . . . . . . . . . . . . 74
Fig. 5.4 Monochromatic scattering. The function λ(ν), real
part of L± (ν), defined in Eq. (5.28), is shown here for
 = 1/8 = 0.125. It becomes infinite for ν = ±1, takes
the value  at ν = 0 and has two zeroes in the interval
[−1, +1] at ±ν0 . They coalesce into a double zero at the
origin for  = 0. The zeroes at ±ν0 correspond to the two
zeroes of L(z) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
Fig. 5.5 Monochromatic scattering. Phase diagrams of L+ (ν) and
of its complex conjugate L− (ν) for ν between 1 and
+∞. The change of sign of [L+ (ν)] corresponds to the
change of sign of λ(ν) in the interval ν ∈ [1, ∞[. The
angle θ is the argument of L+ (ν). It varies between −π
and 0 for ν ∈ [1, ∞[. The destruction probability  has
the value 0.125 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 77
Fig. 5.6 Complete frequency redistribution. Sketch of the phase
diagrams of L+ (ν) and L− (ν) for a Lorentz profile and
 = 0. The phase diagram lies entirely to the right of the
imaginary axis since [L+ (ν)] = λ(ν) has no zero .. . . . . . . . . . . . . . . 77
Fig. 5.7 Contour for the determination of the number of zeroes
of the dispersion function L(z). For monochromatic
scattering, the end point of the cuts are at ν = ±1. For
complete frequency redistribution, they are at ν = 0 .. . . . . . . . . . . . . . 78
Fig. 5.8 The monochromatic function H (μ) for isotropic
scattering, with different √ values of . For
 = 0, H (μ) √ = 3μ as μ → ∞. For  = 0,
H (μ)
√ = 1/  as μ → ∞. For the chosen values of ,
1/  = 6.3, 3.2, 2.2, 1.6. The numerical data are from
Chandrasekhar (1960, p. 125) . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 87
Fig. B.1 Complete frequency redistribution. Closed contour for the
construction of the nonlinear integral equation satisfied
by X(z) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
Fig. 6.1 Contours for the determination of the infinite medium
Green function by application of the Fourier inversion
formula. Left panel: complete frequency redistribution;
right panel: monochromatic scattering . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 99
Fig. 6.2 Contours for the calculation of G̃(p), the Laplace
transform of the surface Green function G(τ ). Left
panel: complete frequency redistribution. Right panel:
monochromatic scattering . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 108
List of Figures xxiii

Fig. C.1 Integration contours for the application of the Fourier

inversion formula. Left panel: calculation of the resolvent
function (τ ) for complete frequency redistribution.
Right panel: calculation of the source function S(τ ) for
monochromatic scattering . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 112
Fig. 7.1 The source function for a uniform primary source Q∗ = 
and several choices of . The absorption profile is a Voigt
−2
profile with parameter √ a = 10 . The surface value S(0)
follows the exact -law . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
Fig. 7.2 The source function for an exponential primary source
Q∗ (τ ) = exp(−10−4 τ ) and  = 10−6 , for a Voigt profile
and a Doppler profile. For the latter, the straight line
shows the asymptotic behavior of the boundary layer
solution (see also Fig. 9.1) . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 129
Fig. E.1 The function kβ (ν, β) for the Doppler profile and
different values of the parameter β, ratio of the continuous
absorption coefficient to the line absorption coefficient.
The function kβ (ν, β) is zero up to ν = β. It is defined in
Eq. (8.17) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 157
Fig. 9.1 Large τ asymptotic behavior of the source function for
conservative scattering. The function S(τ ) is calculated
with a uniform primary source  = 10−6 for a Doppler
profile and a Voigt profile. The straight lines show the
corresponding asymptotic behaviors, τ 1/2 (ln τ )1/4 and
τ 1/4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 169
Fig. F.1 Complete frequency redistribution. Sketch of the phase
diagrams of L+ (ν) and L− (ν) for  = 0. At the point
ν = 0, L± (0) = L(0) = 0 and the phase diagrams are
tangent to the vertical axis in the case of the Doppler
profile and and make an π/4 angle with this axis for a
Voigt profile. For ν → ∞, L± (∞) = L(∞) = 1 and the
phase diagrams are tangent to the vertical axis .. . . . . . . . . . . . . . . . . . . . 173
Fig. 12.1 Integration contour for the decomposition formula. The
vertical lines with an arrow-head indicate that f + (z) is
analytic in an upper half-plane and that f − (z) is analytic
in a lower half-plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 221
Fig. 12.2 The analyticity domains of V (z), Su (z), the Fourier
transform of S + (τ ), and of Sl (z), the Fourier transform
of S − (τ ). The dashed line a has been placed arbitrarily
below the line +1. The vertical lines with two arrows
indicate a strip of analyticity, those with a simple arrow
indicate a half-plane analyticity. The functions V (z)
and Su (z) have a common strip of analyticity defined
by a < (z) < 1. The function V (z) is analytic in
−1 < (z), +1 and has two zeroes at ±i ν0 . . . .. . . . . . . . . . . . . . . . . . . . 223
xxiv List of Figures

Fig. 12.3 The figure shows the strips of analyticity of V (z), the
dispersion function, and of V ∗ (z) defined in Eq. (12.34).
It shows also the analyticity half-planes of Vu (z) and
Vl (z), which satisfy V (z) = Vu (z)/Vl (z), and of Vu∗ (z)
and Vl∗ (z), which satisfy V ∗ (z) = Vu∗ (z)/Vl∗ (z) . . . . . . . . . . . . . . . . . . 225
Fig. 12.4 Analyticity half-planes of the inhomogeneous term
Q̂∗ (z) and of the functions Gu (z) and Gl (z) defined in
Eq. (12.47). The thick solid line indicates the branch cut
of Vl (z) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
Fig. 12.5 The analyticity half-planes of the functions in the left
and right hand-sides of Eq. (12.16) and their common
analyticity strip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 230
Fig. 12.6 The Wiener–Hopf method for the Milne problem. The
half-plane analyticity domains of the functions in the left
and right hand-sides of Eq. (12.73) and their common
strip of analyticity. The function Vu (z) has a double zero
at the origin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 235
Fig. 12.7 The Sommerfeld diffraction problem: (a) the geometry;
(b) the boundary conditions for the field ϕ(x, y) and
for ϕ (x, 0± ) = ∂ϕ(x, 0± )/∂y, the normal derivative of
ϕ(x, y) on the axis y = 0. The notation 0± means that
y → 0 by positive or negative values. The continuity
across the axis y = 0 is indicated by a vertical line with
two arrows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 241
Fig. 12.8 The Sommerfeld diffraction problem: the factorization of
γ = (k − λ)1/2 (k + λ)1/2 . The principal determination of
(k + λ)1/2 is analytic in (k) > 0 and that of (k − λ)1/2
in (k) < 0 .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 245
Fig. I.1 Monochromatic scattering. (a) Contours for obtaining an
explicit expression of Vu (z) in terms of the logarithm
of the dispersion function. The point z lies in the upper
half-plane (z) > δ > −ν0 . (b) Contour for obtaining the
relation between Vu (z) and X(z). The point z lies in the
upper half-plane (z) > −1 . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 247
Fig. 13.1 Ellipse described by the tip of the electric field vector in
the plane orthogonal to the direction of propagation. The
angle  (0 ≤  ≤ π) specifies the orientation of the
ellipse and the angle χ (−π/4 ≤ χ ≤ π/4) the ellipticity
and the sense in which the ellipse is being described . . . . . . . . . . . . . . 257
Fig. 13.2 Rotation of the reference direction by an angle α . . . . . . . . . . . . . . . . . . 259
Fig. 13.3 Geometry of a single scattering. The components Ir
and Il are respectively perpendicular and parallel to the
scattering plane (,  ) . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 262
List of Figures xxv

Fig. 13.4 The atmospheric reference frame. The axis Oz is along

the direction normal to the surface. A direction  is
defined by its colatitude θ and azimuth χ. In the plane
perpendicular to , ea is the reference direction and eb is
perpendicular to ea . γ is the angle between the tangent to
the meridian plane and ea . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 264
Fig. 13.5 Sketch of the Zeeman displacement of the magnetic
sublevels for a normal Zeeman triplet. The horizontal axis
is the magnetic field intensity B. The Hanle effect can be
observed as long as the Zeeman sublevels are overlapping
each other. The magnetic field intensity Btyp corresponds
to a Hanle factor H of order unity . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 277
Fig. L.1 The spherical unit vectors e0 , e+1 , and e+1 . The vector e0
is aligned with the magnetic field vector B . . . . .. . . . . . . . . . . . . . . . . . . . 292
Fig. L.2 Motion of the damped classical oscillator in the plane
perpendicular to the direction of the magnetic field.
The parametric plot shows the time evolution of the
component xy (t) versus the component xx (t), defined in
Eq. (L.14). The left panel shows a rosette with a Hanle
parameter H = 1 and the right one a daisy with H = 9.
The curves have been calculated with γ = 1, 2πν0 = 50,
and a maximum value t = 2.5 for the rosette and t = 1.5
for the daisy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 294
Fig. 14.1 The elements of the kernel matrix K(τ ) for the Doppler
profile, with the depolarization parameter set equal to 1.
The left panel shows the elements Kij (τ ) and the right
panel shows log K11 (τ ) (left vertical axis) and the ratios
K12 (τ )/K11 (τ ) and K22 (τ )/K11 (τ ) (right vertical axis) . . . . . . . . . . . . 327
Fig. 16.1 The functions Hl (μ) and Hr (μ) for Rayleigh scattering,
and the function H (μ) for conservative monochromatic
scattering.
√ For μ → ∞, √ Hl (μ) increases at infinity
as 5μ and H (μ) as 3μ, while Hr (μ) tends to the
constant value 0.1414. The linear growth of Hl (μ) and
H (μ) is already installed for μ around one. The data are
from Chandrasekhar (1960, p. 125 and p. 248) . . . . . . . . . . . . . . . . . . . . . 373
Fig. 20.1 The thermalization length for monochromatic scattering
and complete frequency redistribution. The figure shows
the source function S(τ ) for a uniform primary source 
with  = 10−6 . The thermalization length, distance from
the surface at which the source√ function S(τ ) reaches its
saturation value, is of order 1/  for monochromatic
scattering, 1/ for the Doppler profile, and a/( 2) for the
Voigt profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 475
xxvi List of Figures

Fig. 20.2 Ordinary diffusion and anomalous diffusion. The left

panel, shows a diffusive-type random walk. All the
random steps have the same constant value. The right
panel shows an anomalous random walk. The random
steps are calculated with a probability distribution having
the form 1/(1 + |x|β+1 ), with β = 1, mimicking the
Doppler frequency redistribution . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 476
Fig. 21.1 A typical discrete random walk on a line. The vertical
axis represents the position on the line of a random walker
starting at 0. After n steps, the random walker has the
position xn .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 480
Fig. 21.2 Examples of Lévy walks. The probability distribution has
the form 1/(1 + |x|β+1). In the left panel, β = 2 and in
the right panel β = 1.5, which is a model for a Holtsmark
distribution. The case β = 1 is shown in the right panel
of Fig. 20.2. The particles are starting at the origin (0,0).
Each graphic shows 104 steps . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 490
Fig. 21.3 A typical random walk on a line, for a random walker
starting at the point x. Mn indicates the maximum
position reached while performing n steps. The first step
brings the random walker from x to x1 , x1 ≤ y . . . . . . . . . . . . . . . . . . . . 492
Fig. 22.1 Geometry for establishing the relation between of the
point source and the plane source resolvent functions . . . . . . . . . . . . . 512
Fig. 23.1 At each point r b of the surface, the boundary layer is
a plane-parallel, semi-infinite medium extending from
s = 0 to s = −∞. The surface of the boundary layer is
tangential to the surface of the medium at r b . . .. . . . . . . . . . . . . . . . . . . . 520
Fig. 24.1 The surface polarization rate (in percent). A comparison
between the asymptotic predictions of Eq. (24.60) (the
dashed lines) and numerical solutions of the polarized
transfer equation, for different values of  (the solid
lines). The √ model is a slab with an optical thickness
larger than , an unpolarized and uniform primary
source and no incident radiation. From top to bottom,
 = 10−5 , 10−3 , 10−2 , 10−1 . The asymptotic result is
not plotted for  = 10−1. The Chandrasekhar limit is
graphically indistinguishable from the  = 10−5 results . . . . . . . . . . . 546
Fig. 26.1 The RI partial redistribution function. This figure show
the ratio r̄I (x, x )/ϕ(x) as a function of x for different
values of x; x and x are respectively the incident and
scattered dimensionless frequencies . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 566
List of Figures xxvii

Fig. 26.2 The RV partial redistribution function. This figure shows

the partial redistribution function rV (ξ, ξ ) in the atomic
rest frame given in Eq. (26.22) for the damping parameter
γ1 = 0.05, the incident frequency ξ = 2, and three
different values of the ratio γ2 /γ1 . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 569
Fig. 26.3 The partial RII redistribution function. This figure shows
the ratio r̄II (x, x )/ϕ(x) as a function of x for different
values of the frequency x; x and x are respectively the
incident and scattered dimensionless frequencies . . . . . . . . . . . . . . . . . . 573
List of Tables

Table 5.1 Monochromatic scattering and complete frequency

redistribution. Real part, imaginary part, and argument
of L+ (ν) for  = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 76
Table B.1 The functions X(z), X∗ (z) = (ν0 − z)X(z), and H (z)
for  = 0: values at z = 0 and behavior at infinity. For
monochromatic scattering, ν0 ∈ [0, 1] is the positive
√ dispersion function L(z). For small values of
root of the
, ν0 3. The behavior at infinity of H (z) for  = 0
is discussed in Chap. 9. . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 93
Table 6.1 Infinite medium. G∞ (τ ) is the infinite medium Green
function and ∞ (τ ) the resolvent function. Columns
(2) and (3) correspond to the Fourier transform and
inverse Laplace transform of the functions in column
(1). Column (4) gives the primary source term for the
convolution integral equation satisfied by the functions
in column (1). Columns (5) and (6) give the Fourier and
inverse Laplace transforms of the functions in column
(4). The notation—stands for undefined functions . . . . . . . . . . . . . . . 98
Table 6.2 Semi-infinite medium. G(τ ) is the surface Green
function, and (τ ) the resolvent function. Columns (2)
and (3) give the direct and inverse Laplace transforms
of the functions in column (1). Column (4) gives
the primary source term for the convolution integral
equation satisfied by the functions in column (1).
Columns (5) and (6) give the direct and inverse
Laplace transforms of the functions in column (4). The
notation—stands for nonexistent functions . . .. . . . . . . . . . . . . . . . . . . . 103

xxix
xxx List of Tables

Table 7.1 Semi-infinite medium. S(τ ) denotes the source function,

Q∗ (τ ) the primary source term and K(τ ) the kernel of
the Wiener–Hopf integral equation for S(τ ). Columns
(2) and (3) correspond to the direct and inverse Laplace
transforms of the functions in column (1). Column (4)
lists examples of primary source terms and columns (5)
and (6) their direct and inverse Laplace transforms . . . . . . . . . . . . . . . 119
Table 8.1 The line source function at the surface and at infinity.
The effects of a continuous absorption and of a
continuous emission. β is the ratio of the continuous to
line absorption coefficient, ¯ the effective destruction
probability, and ρB the continuous emission. The
parameters β, , ρ and B are constants . . . . . . . .. . . . . . . . . . . . . . . . . . . . 153
Table F.1 Complete frequency redistribution. Real part, imaginary
part, and argument of L+ (ν) for  = 0 . . . . . . . .. . . . . . . . . . . . . . . . . . . . 172
Table 15.1 Notation for inverse Laplace transforms of vectors and
matrices depending on the optical depth τ . . . .. . . . . . . . . . . . . . . . . . . . 359
Table 16.1 The Rayleigh Milne problem. The surface value and
behavior at infinity of the two components of the source
vector for different representation of the polarized
radiation field. Sl and Sr correspond to the (Il , Ir )
representation and SI and SQ correspond to the (I, Q)
one. For comparison the first line shows the source
function S(τ ) for the scalar Milne problem. For the
scalar √case, s1 = 3F /(4π) and for Rayleigh scattering,
s1R = 6F /(8π) with F the total flux, defined for
Rayleigh scattering in Eq. (16.77). We recall that
c = β/α = 0.873 and q = 2/α = 0.690 . . . . . .. . . . . . . . . . . . . . . . . . . . 388
Table M.1 Real parts, imaginary parts, and arguments of L+ 1 (ν)
and L+ 2 (ν) along the cut [1, ∞[ . . . . . . . . . . . . . . .. . .................. 396
Table M.2 The functions L1,2 (z), X1,2 (z) and Hl,r (z) at
three special points z = 0, 1, ∞. We recall that
zX1 (−1/z) = Hl (z) and X2 (−1/z) = Hr (z). The
values of Hl,r (1) are from Chandrasekhar. The values of
X1,2 (1) are derived from the values of Hl,r (1) and the
factorization relations in Eq. (M.24) . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 399
Table M.3 Moments of Hl (μ) and Hr (μ) and some numerical
constants. The two first moments of Hl (μ) and Hr (μ),
the constants q and c, and the values of Hl (1) and Hr (1)
are from (Chandrasekhar 1960, p. 248), rounded to 3
significant digits. This reference provides also tables for
Hl (μ) and Hr (μ), μ ∈ [0, 1], which we have used to
calculate numerically the second and third-moments . . . . . . . . . . . . . 402
List of Tables xxxi

Table 18.1 The matrices G̃(z), H(z), G̃R (z), and HR (z) at the origin
and at infinity. The matrix R = GR (∞) is a rotation
matrix, which can be calculated by solving numerically
the nonlinear integral equations for G̃R (z) or HR (z). The
matrix E = diag[, Q ] is equal to L(0). The I-matrix is
defined by I(0, z) = HR (z) . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 439
Table 19.1 The asymptotic behaviors of K(τ ) for τ → ∞, g(ξ ) for
ξ → ∞, k(ν) for ν → 0, and K̂(k) for k → 0. The
kernel is an even function of τ . Here τ , ξ , and ν are
positive. For monochromatic scattering, k(ν) is zero in
the interval [0, 1[ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 465
Table 19.2 The index α for the Doppler and Voigt profiles and the
asymptotic expressions of f (y) and f (y) for y large.
The function f (y) is the inverse function of the profile
ϕ(x) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 466
Table 22.1 Infinite medium. Behavior at large optical depths of the
resolvent function for small values of , the destruction
probability per scattering. The constant a is the Voigt
parameter of the line. The nearly conservative zone
corresponds to 1 τ < τeff and the strong absorption
zone to τ > τeff . For monochromatic scattering τeff ∼ ν0 . . . . . . . . 509
Table 22.2 Semi-infinite medium. Asymptotic behavior of the
resolvent function in the nearly conservative and strong
absorption zones. The thermalization lengths τeff are
given in Table 22.1. For monochromatic scattering
τeff ∼ ν0 . For  = 0, the monochromatic √ resolvent
function has exactly the value 3 at infinity . .. . . . . . . . . . . . . . . . . . . . 513
Table 24.1 Moments of Hl (μ) and Hr (μ) and other numerical
constants. The two first moments of Hl (μ) and Hr (μ),
the constants q and c, and the values of Hl (1) and Hr (1)
are from Chandrasekhar (1960, p. 248). This reference
provides also tables for Hl (μ) and Hr (μ), μ ∈ [0, 1],
used to calculate numerically the third and fourth order
moments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 544
Table 25.1 Scaling laws for complete frequency redistribution
with a Doppler profile. In the first column, the small
expansion parameter; in row 1: β = 0 and T = ∞, in
row 2:  = 0 and T = ∞, and in row 3: β =  = 0. The
four subsequent columns show τeff , the characteristic
scale of variation of the radiation field, xc , the
characteristic frequency defined by τeff ϕ(xc ) ∼ 1, the
mean number of scatterings N, and the mean path
length L .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 559
xxxii List of Tables

Table 25.2 Scaling laws for complete frequency redistribution with

a Voigt profile. a Voigt parameter of the line. Same
presentation as in Table 25.1 . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 559
Table 26.1 Scaling laws for the RII partial frequency redistribution.
a parameter of the line profile. The expansion parameter
is  for an infinite medium and 1/T for a slab with total
optical thickness T . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 579
Chapter 1
An Overview of the Content

The book is organized in three parts: Part I, is devoted to the presentation of

exact methods of solution for scalar radiative transfer problems, and Part II to their
generalization to vector and matrix problems for linearly polarized radiation fields.
In Part I we deal only with the radiation field intensity, whereas in Part II we also
consider the linear polarization created by the scattering of photons. In Part III,
we use asymptotic methods to examine the structure of polarized and unpolarized
radiation fields when photons undergo a very large number of scatterings.

1.1 Part I: Scalar Radiative Transfer Equations

In Part I we consider two different scattering processes. One of them is monochro-

matic scattering. At each scattering, photons undergo a change of direction but keep
the same frequency (energy). This process is relevant to the formation of continuous
spectra. The other process is relevant to the formation of spectral lines. The photons
undergo, at each scattering, a change of direction and also of frequency, within the
width of the spectral line. The frequency of the absorbed and scattered photons
are assumed to be fully uncorrelated, a situation known as complete frequency
redistribution or noncoherent scattering in the astronomical literature. It is a
frequently valid assumption. The general case, with frequency correlations between
the absorbed and scattered photons is known as partial frequency redistribution.
For monochromatic scattering and complete frequency redistribution, we assume
that the scattering in direction is isotropic. Although fairly different, monochromatic
scattering and complete frequency redistribution have a common property, namely
that the local emission of photons, which is the sum of a scattering term and of a
primary source of photons, depends only on the position inside the medium. This
local emission term is known in the astronomical literature as the source function,
an expression used throughout this book.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

H. Frisch, Radiative Transfer, https://doi.org/10.1007/978-3-030-95247-1_1
2 1 An Overview of the Content

For a one-dimension medium, the source function, satisfies a convolution integral

equation, the kernel of which contains the scattering properties of the transport
process. For monochromatic scattering, the kernel has an exponential tail and for
complete frequency redistribution an algebraic one, due to changes in frequency. In
all the examples that are considered, the kernel is translation invariant, otherwise no
exact solutions would not exist. For semi-infinite media, these integral equations are
often referred to as Wiener–Hopf1 integral equations, a custom, which we follow.
Wiener–Hopf integral equations for the source function and for its associated Green
function are starting points for many exact methods of solution for radiative transfer
problems.
Scattering processes are also characterized by a destruction probability per
scattering, henceforth denoted . When  = 0, all the absorbed photons are
reemitted and the single scattering albedo, (1 − ), is equal to one. It will be said
that the scattering, or the medium, are conservative. The term non-conservative is
employed when  = 0; the single scattering albedo is then less than one.
Part I is organized as follows. In Chap. 2 we introduce the radiative transfer
equations for monochromatic scattering and complete frequency redistribution, the
source function, the Green2 function, and the resolvent function, which is the regular
part of the surface value of the Green function. We also introduce the convolution
integral equations satisfied by these functions and discuss some of their properties.
We present in Chap. 3 a summary of each of the exact methods of solution
considered in the subsequent chapters. Their presentation follows, more or less, a
chronological order, while their detailed description and application follows, also
more or less, an inverse chronological order, ending with the Wiener–Hopf method.
A common property of exact methods of solution for radiative transfer problems
in semi-infinite media is that exact solutions, when they exist, ultimately require
the solution of a nonlinear problem, although the starting point is a linear radiative
transfer equations or a convolution integral equation. Examples will be found in the
subsequent chapters.
Exact methods of solutions often start with the transformation of a radiative
transfer equation, or a Wiener–Hopf integral equation, into another linear equation,
to which can be applied more or less standard mathematical tools. In Chap. 4, we
show how to transform a Wiener–Hopf integral equation into a singular integral
equation with a Cauchy-type3 kernel, by a direct or an inverse Laplace transform.
We then describe the main lines of a Hilbert transform method of solution, which
transforms singular integral equations into boundary layer problems in the complex
plane, known as Riemann–Hilbert problems.4

1 Norbert Wiener: 1894–1964 (Columbia, USA-Stockholm); Eberhard Hopf: 1902–1983

(Salzburg-Bloomington, USA).
2 George Green: 1793–1841 (Sneiton-Nottingham, GB).
3 Augustin-Louis Cauchy: 1789–1857 (Sceaux-Paris).
4 Bernhard Riemann: 1826–1866 (Jameln, Germany-Verbenia, Italy); David Hilbert: 1862–1943

(Königsberg-Göttingen).
1.1 Part I: Scalar Radiative Transfer Equations 3

In Chap. 5, we show how to solve Riemann–Hilbert problems arising from

radiative transfer. We show in particular that for semi-infinite media, the solution
requires the introduction of an auxiliary function, solution of a homogeneous
Riemann–Hilbert problem. The latter is a nonlinear problem of the type mentioned
above. The auxiliary function is essentially the function, known nowadays as the
H -function.
The singular integral equation approach, associated to a Hilbert transform
method is employed in Chap. 6 to construct an exact expression of the resolvent
function for non-conservative scattering. The resolvent function is chosen as first
example for the application of an exact method of solution. It depends exclusively
on the scattering process and has a simple expression in terms of the H -function.
The relation between the resolvent function and the H -function can be generalized
to polarized radiation fields, as will be shown in Part II.
In the three subsequent chapters, we give other examples of the same singular
integral equation method. In Chap. 7 we construct exact expressions for the source
function and the emergent intensity, for non-conservative scattering, considering
both monochromatic scattering and complete frequency redistribution. In Chap. 8,
we treat the case of a spectral line embedded in a continuum, and in Chap. 9
conservative scattering. In this latter chapter, we consider the Milne5 problem, the
diffusion reflection problem, and their generalized versions for complete frequency
redistribution.
We then describe three alternative methods of solution: the Case eigenfunction
expansion method, the traditional resolvent method, and, last but not least, the
Wiener–Hopf method. The Case expansion method is presented in Chap. 10. Its
starting point is the radiative transfer equation for the radiation field intensity. An
eigenfunction expansion of the radiation field leads to singular integral equations
with Cauchy-type kernels similar to the singular integral equations introduced in
Chap. 4. In Chap. 11, we present a traditional approach, sometimes referred to
as the resolvent method. Based mainly on invariance properties, it is a real-space
method, which does not require complex plane analysis. It leads to a nonlinear
integral equation for the H -function, in brief referred to as the H-equation. The
resolvent method is another example of how a linear radiative transfer problem is
transformed into a nonlinear one. We briefly discuss the properties of H -equations,
in particular the non-uniqueness of the solution for monochromatic scattering. √ In
the same chapter we give a simple algebraic proof of the famous exact -law,
applicable in a semi-infinite medium with a uniform primary source.
The Wiener–Hopf method is presented in Chap. 12. It is a generalization
to convolution integral equations on a semi-infinite line of the standard Fourier
transform method for convolution integral equations on the full line. It makes a
systematic use of the analyticity properties of Fourier transforms in the complex
plane. We describe it for monochromatic scattering and then for complete frequency
redistribution. In the latter case, it becomes a Riemann–Hilbert problem of the type

5 Edward Arthur Milne: 1896–1950 (Kingston upon Hull-Dublin).

4 1 An Overview of the Content

solved in Chap. 5. One of the key ingredient of the Wiener–Hopf method makes
its applicable to partial differential equations. We give, with a wave diffraction
problem, an example in Sect. 12.6.

1.2 Part II: Scattering Polarization

The second Part is devoted to radiative transfer problems for fields linearly
polarized by scattering processes sharing the same Rayleigh polarization phase
matrix, namely the Rayleigh scattering of monochromatic radiation, the resonance
polarization of spectral lines, formed with complete frequency redistribution, and
the associated Hanle6 effect, which takes into account the effects of a weak magnetic
field.
An important feature of the Rayleigh polarization phase matrix is that it can be
factorized as the product of two matrices depending separately on the directions
of the incident and scattered beams. We make use of this property to introduce
a representation of the polarized radiation field based on the irreducible spherical
tensors for polarization introduced by Landi Degl’Innocenti (1984). With this
representation, which can be introduced for one-dimensional and multi-dimensional
media, the scattering term (the integral over the directions of the incident beams
weighted by a polarization matrix) in the radiative transfer equation depends only
on the position inside the medium. The radiative transfer equations have thus the
same structure as those of Part I, except that the radiation field and the source term
are now vectors. For a one-dimensional medium, the convolution equation for the
source term has also the same structure as in the scalar case, but its kernel is a matrix.
The Green function, the resolvent function, the H -function also become matrices.
The scalar Riemann–Hilbert problem becomes a matrix problem and the nonlinear
H -equation becomes a nonlinear matrix equation. The matrix structure makes the
situation significantly more complicated and it is in general impossible to construct
explicit solutions for these equations.
In some very special cases, homogeneous matrix Riemann–Hilbert problems
have an explicit solution, conservative monochromatic Rayleigh scattering is
one of them. The reason is that the corresponding matrix problem can actually
be transformed into two scalar ones. This property was first demonstrated by
Chandrasekhar7 (1946), by a detailed analysis of the properties of the matrix kernel
and has allowed him to construct exact solutions for the Milne and the diffuse
reflection problems in terms of two functions Hl and Hr with exact expressions.
There is no such decomposition for non-conservative Rayleigh scattering, however
an exact solution was obtained by Siewert et al. (1981, see also Aoki and Cercignani
1985) with very advanced mathematics, out of the scope of this book. It is a

6 Wilhem Hanle: 1901–1993 (Mannheim-Giessen).

7 Subramayan Chandrasekhar: 1910–1995 (Lahore-Chicago).
1.2 Part II: Scattering Polarization 5

real breakthrough but the solution that has been obtained has a very complicated
form, not easily amenable to a numerical solution. Here we treat non-conservative
Rayleigh scattering in the same way as the resonance polarization of spectral lines
and the Hanle effect. For spectral lines, there is no exact solution, even when are
formed with complete frequency redistribution and the scattering is conservative.
We show, by different methods, how to construct a nonlinear matrix H-equation,
which can be solved numerically.
Part II is organized as follows. Chapter 13 is devoted to various aspects of
scattering physics. After recalling the classical electromagnetic description of
a polarized beam of light, we present polarization matrices for the Rayleigh
scattering,8 the resonance polarization of spectral lines and the Hanle effect. For
spectral lines, the polarization matrix depends on the frequencies of the incident
and scattered photons and will be referred to as the redistribution matrix. Wandering
from our main subject, we described in some details, for a simple-two level atom,
the spectral profile of the scattered radiation. The purpose is to place the assumption
of complete frequency redistribution in a broader context and also to offer a brief
introduction to the literature on the scattering polarization of spectral lines.
In Chap. 14, we show how the radiative transfer equations for the Stokes
parameters of a field linearly polarized by one of the three mechanisms mentioned
above, can be transformed into new radiative transfer equations, in which the source
term depends only on the position in the medium, as in the scalar case. For a one-
dimensional medium, these new equations can then be recast as vector or matrix
Wiener–Hopf integral equations and used as starting points to generalize some of
the methods of solution described in the scalar case. In Chap. 15 we generalize to
a polarized radiation√ the real-space methods described in Chap. 11 for the scalar
case. We construct -laws for Rayleigh scattering, resonance polarization, and the
Hanle effect, and also the corresponding nonlinear matrix H-equations. We also
introduce matrix singular integral equations that will be used in the subsequent
chapters.
In the remaining three chapters we show how vector and matrix singular integral
equations with Cauchy-type kernels can be treated with Hilbert transform methods.
For conservative Rayleigh scattering, we explain in Chap. 16 how the matrix
Riemann–Hilbert problem can be transformed into two scalar problems and how the
polarized Milne and the diffuse reflection problems can be solved exactly in terms of
the two scalar functions Hl and Hr . For polarization problems with no exact solution
(non-conservative Rayleigh scattering, resonance polarization, and Hanle effect), we
establish in Chap. 17 homogeneous matrix Riemann–Hilbert problems and discuss
in detail the asymptotic behavior of their solutions at infinity (in the complex plane).
Although they contain undetermined constants, these solutions can be used, as we
show in Chap. 18, to recover the nonlinear H-equations, established in Chap. 15
with a real-space method. This second method has the advantage of proving that
the existence of solutions to these nonlinear equations. We also discuss in the same

8 The expression Rayleigh scattering always implies that the scattering is monochromatic.
6 1 An Overview of the Content

chapter a non-uniqueness problem for the monochromatic nonlinear H-equation and

present alternative nonlinear integral equations, well suited to numerical solutions.
We present in particular the equation for an I-matrix, constructed with the emergent
radiation field, which has been introduced by Ivanov (1996).

1.3 Part III: Asymptotic Properties of Multiple Scattering

We describe in Part III asymptotic methods based on the existence of small or large
parameters such as , the destruction probability per scattering, or T the optical
thickness of the medium. They are applicable when photons undergo a very large
number of scatterings and greatly contribute to the physical understanding of the
formation of a radiation field by multiple scatterings.
For example, it will be shown that for monochromatic scattering photons perform
an ordinary diffusion random walk. In particular, that the averaged distance travelled
after n scatterings (n large) goes as n1/2 . For complete frequency redistribution, the
random walk of photons has an anomalous behavior, performing a so-called Lévy9
walk, because of the changes of frequency within the spectral line. The averaged
distance travelled by photons varies as n1/α with α smaller than 2. The difference
between ordinary and anomalous diffusion has its origin in the asymptotic behavior
of the kernel. These scaling laws are derived in Part III with an asymptotic analysis
of the integral equation for the source function and also with a statistical analysis of
the random walk of the photons.
Asymptotic methods can address radiative transfer problems, which have no
exact solution and are usually solved numerically. They provide equations that
are simpler than the original equations. Their solutions are approximations to the
original problem, but their domain of validity and accuracy are controlled by
the small expansion parameter. We give examples with non-conservative Rayleigh
scattering and partial frequency redistribution of spectral lines.
Another strength of asymptotic methods is that they provide spatial and fre-
quency scales of variation for the radiation field, which describe the spreading of
photons in space and frequency (in the case of spectral lines). For example the
spreading in space, which is known in the astronomical literature as thermalization
length, henceforth denoted τeff , can be identified with the characteristic scale for the
spatial variations of the radiation field.
Part III is organized as follows. The somewhat technical Chap. 19 contains a
detailed analysis of the asymptotic behavior of the kernel K(τ ) for large values of τ
and of its Fourier transform K̂(k) for k → 0. In Chap. 20 we show how to perform a
large scale asymptotic analysis of the integral equation for the source function in an
infinite medium, by a proper rescaling of the optical depth. This analysis provides
scaling laws for the thermalization length τeff for monochromatic scattering and

9 Paul Pierre Lévy: 1886–1971 (Paris-Paris).

1.3 Part III: Asymptotic Properties of Multiple Scattering 7

complete frequency redistribution. We show in particular that the ordinary diffusion

is encountered when the second order moment of the kernel is finite, that is when
 +∞
τ 2 K(τ ) dτ < ∞. (1.1)
−∞

In the following Chap. 21 we present a statistical analysis of a random walk of

photons, progressing with random steps on an infinitely long line. We examine
the mean displacement and the mean maximum displacement after n steps. We
show how they are related to the thermalization length, how the discrete description
is related to the continuous description with radiation fields, and how anomalous
diffusion is related to stable probability distributions. For complete frequency
redistribution, the analysis of the random walk of the photons is further developed in
Chap. 22, where we analyze the propagation of photons away from a source, using
exact expressions of the Green function in an infinite medium established in Part I.
The two subsequent chapters are devoted to monochromatic scattering in a
medium with boundaries, the scalar case in Chap. 23 and the Rayleigh scattering in
Chap. 24. Starting from the radiative transfer equation, we show that the radiation
field can be decomposed into an interior field, which satisfies a diffusion equation,
and a boundary layer field, solution of a semi-infinite conservative radiative transfer
equation. The two fields are described by equations, which are simpler than the
original transfer equation, and which can be asymptotically matched to obtain a
solution valid in the full medium. For the Rayleigh scattering, we obtain corrections
to the Chandrasekhar’s polarization laws for conservative Rayleigh scattering.
Chapter 25 is devoted to an asymptotic analysis of spectral lines formed with
complete frequency redistribution. We show that the radiation field can also be
decomposed into an interior field and a boundary layer field and that they can be
matched together to construct a solution valid in the whole medium. The asymptotic
expansion is performed on the integral equation for the source function as in
Chap. 20, the anomalous diffusion of the photons preventing an asymptotic analysis
of the radiative transfer equation itself. We also examine the case of resonance
polarization and present various scaling laws concerning the mean number of
scatterings and the mean pathlength, an accumulated distance travelled by photons.
Finally, we present in Chap. 26 some asymptotic properties of spectral lines
formed with partial frequency redistribution. We consider the four elementary
partial redistribution functions introduced in Hummer (1962). We show that three
of them (RI , RIII , RV ) are behaving as complete frequency redistribution and that
RII , which describes a coherent scattering in the atomic frame, is of the ordinary
diffusion type at large frequencies.
8 1 An Overview of the Content

References

Aoki, K., Cercignani, C.: On the matrix Riemann–Hilbert problem relevant to Rayleigh scattering.
J. Appl. Math. Phys. (ZAMP) 36, 61–69 (1985)
Chandrasekhar, S.: On the radiative equilibrium of a stellar atmosphere. X. Astrophys. J. 103,
351–370 (1946)
Hummer, D.G.: Non-coherent scattering I. The redistribution functions with Doppler broadening.
Mon. Not. R. Astr. Soc. 125, 21–37 (1962)
Ivanov, V.V.: Generalized Rayleigh scattering. III. Theory of I-matrices, Astron. Astrophys. 307,
319–331 (1996)
Landi Degl’Innocenti, E.: Polarization in Spectral lines III : Resonance polarization in the non-
magnetic, collisionless regime. Sol. Phys. 91, 1–26 (1984)
Siewert, C.E., Kelley, C.T., Garcia, R.D.M.: An analytical expression for the H-Matrix relevant to
Rayleigh scattering. J. Math. Anal. Appl. 84, 509–518 (1981)
 Part I
Scalar Radiative Transfer Equations
Chapter 2
Radiative Transfer Equations

In this chapter, we present the scalar radiative transfer equations used in Part I to
illustrate exact method of solutions for radiative transfer equations in semi-infinite
media. We also present different types of integral equations that can be derived
from the integro-differential equations. All the equations are time-independent and
one-dimensional. They describe the formation of continuous spectra and spectral
lines. For continuous spectra, there is no change in frequency at each scattering.
We refer to it as monochromatic scattering. For a spectral line, photons change
their frequency at each scattering. They can be fully decorrelated, a situation known
as complete frequency redistribution, or partially correlated. One says that there is
partial frequency redistribution. An example of partial frequency redistribution is
presented in Appendix J of Chap. 13 and some asymptotic results on the large scale
behavior of the radiation field in Chap. 26. Otherwise, we consider only spectral
lines formed with complete frequency redistribution (also known as noncoherent
scattering) for which exact results can be obtained.
The integro-differential radiative transfer equations are presented in Sect. 2.1
for monochromatic scattering and complete frequency redistribution. Section 2.2
is devoted to the source function, sum of the scattered radiation and of a primary
source of photons. With our assumptions on the frequency redistribution, it depends
only on the position in the medium. We show that it satisfies an integral equation
of the convolution type, with a translation invariant kernel. Section 2.4 is devoted to
the Green function and some associated functions, such as the resolvent function,
regular part of the surface Green function. These functions, which characterize the
scattering process and the geometry of the medium, independently of the primary
sources, obey integral equations identical to those of the source function, except for
inhomogeneous terms. In this same Sect. 2.4, we prove some fundamental properties
of the Green function and of convolution type integral equations for semi-infinite
media.

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 11

H. Frisch, Radiative Transfer, https://doi.org/10.1007/978-3-030-95247-1_2
12 2 Radiative Transfer Equations

2.1 The Integro-Differential Radiative Transfer Equations

Radiative transfer equations are linear Boltzmann equations describing the propaga-
tion of an ensemble of photons inside a host medium and their interactions with the
particles of the medium. Each photon is treated as a particle of zero mass, moving
with the velocity of light c. At an instant t, photons are characterized by a position r,
a direction of motion n, and an energy E = hν, with h the Planck constant and ν the
frequency. A systematic derivation of radiative transfer equations from the Maxwell
equations is outside the scope of this book. The electro-magnetic description of the
radiation field is introduced briefly in Part II when dealing with polarized radiation
fields. A systematic electro-magnetic description of multiple scattering of light can
be found in, e.g., Mishchenko et al. (2006) and Mishchenko (2008) (see also Landi
Degl’Innocenti and Landolfi 2004, Chap. 5).
In this book we consider exclusively interaction processes relevant to the forma-
tion of spectral lines and continuous spectra. These processes are (i) creation, (ii)
absorption followed by a destruction and (iii) absorption followed by a reemission.
The third type of interaction is referred to as a scattering process. Creations
correspond to a radiative decay of a level (bounded or unbounded) that has been
first excited by a collision, say, with electrons. Destruction is the opposite process:
a level excited by photon absorption is deexcited by collision with some particles of
the medium. These three processes, together with the photons entering or escaping
the medium participate in the formation of the radiation field.
The time evolution of an ensemble of photons can be described by a distribution
function f (t, r, n, ν) dV d dν, which is the number of photons at time t in an
infinitesimal volume dV around r, a solid angle d around n and a frequency
interval dν around ν. In Astronomy, the custom is to work with the specific intensity,
I (t, r, n, ν), related to the photon distribution function by

I (t, r, n, ν) = c h νf (t, r, n, ν). (2.1)

The specific intensity is proportional to the density of energy of the radiation field.
In the traditional radiometric point of view, it is defined in terms of the amount
of radiant energy dE in a frequency interval dν around ν, transported across an
element of area dσ in an infinitesimal solid angle d around a direction n, during a
time dt. By definition

dE = I (t, r, n, ν, t) cos θ dν dσ d dt, (2.2)

where θ is the angle between the direction n and the outward normal to dσ . The
specific intensity is a positive quantity. The description of the radiation field with
the specific intensity, or with a scalar distribution function, is sufficient as long as
the polarization of the radiation is ignored. In Astronomy, polarized radiation is
usually described by its four Stokes parameters (see Chap. 13).
2.1 The Integro-Differential Radiative Transfer Equations 13

A radiative transfer equation has the general form:

1 ∂I (t, r, n, ν)
+ n · ∇I (t, r, n, ν) =
c ∂t
− [χd (t, r, n, ν) + χs (t, r, n, ν)]I (t, r, n, ν)
 ∞
d
+ (t, r, n, ν, n , ν )I (t, r, n , ν ) dν
0 4π
+ Q∗ (t, r, n, ν), (2.3)

with
 ∞ d
χs (t, r, n, ν) = (t, r, n, ν, n , ν ) dν . (2.4)
0 4π

The right hand side in Eq. (2.3) describes the sinks and sources of photons. The
coefficient χd corresponds to an absorption followed by a destruction and the
coefficient and χs to an absorption followed by a reemission. The integral term
describes the scattering of an incident beam of radiation with direction n and
frequency ν into a scattered beam with direction n and frequency ν. The integration
is over all possible directions and frequencies of the incident beam. The last
term Q∗ (t, r, n, ν) represents a source of primary photons. An analysis of the
interaction between the radiation field and the matter of the host medium is needed
to determine the coefficients of the radiative transfer equation. For problems relevant
to Astronomy and more specifically to the formation of stellar spectra, this subject
is developed in several text books such as Aller (1963), Jefferies (1968), Mihalas
(1978), Ivanov (1973), Landi Degl’Innocenti and Landolfi (2004), and Hubeny and
Mihalas (2015).
The equations considered in this book are much simpler than Eq. (2.3) but are
not mathematical oddities. In particular they can describe the formation of spectral
lines and other spectral features such as atomic continua in a stellar atmosphere. The
simplifications come from several assumptions:
(i) The radiation field is assumed to be independent of time. The time dependence
plays a role only for very fast evolving phenomena, namely those with a
characteristic time of variation t such that t l/c, with l the scale
of variation in the physical space.
(ii) The medium is assumed to be one-dimensional. This assumption holds for
atmospheres with a thickness small compared to the radius of the star, such as
the solar atmosphere. The one-dimensional hypothesis implies that the physical
properties of the atmosphere (temperature, density, chemical composition,
and so on) have no horizontal variations. The one-dimensional variable, here
denoted z, is measured along the normal to the medium (see Fig. 2.1). This
one-dimensional assumption is raised for some problems considered in Parts II
and III.
14 2 Radiative Transfer Equations

+∞ z
z θ
n
n z
θ n
θ
τ n
z

−∞
+∞

Fig. 2.1 Left panel: the space variable τ and the inclination angle θ of a ray with direction n. The
axis z is normal to the surface of the medium. Right panel: the heliocentric interpretation of the
angle θ. The Sun is viewed from the right. Observations at disk center correspond to θ = 0 and
those at the limb to θ = π/2

(iii) In Part I, we also assume that the radiation field has a cylindrical symmetry.
The direction of the field is fully defined by the colatitude θ ∈ [0, π], i.e.
inclination of the ray with respect to the z-axis (see Fig. 2.1) or with μ = cos θ ,
μ ∈ [−1, +1]. For solar observations, n being the direction of the line-of-
sight, then θ is the so-called heliocentric angle. Observations at disk center
correspond to θ = 0 (μ = 1) and those at the limb to θ = π/2 (μ = 1) (see
Fig. 2.1).
Radiation fields without a cylindrical symmetry are considered in Part II,
when we discuss the polarization of radiation fields.
(iv) The absorption coefficient is assumed to be independent of the direction n. It
may then be written as

χ(t, r, n, ν) = kd (r)f (ν). (2.5)

This form of coefficient holds for continuous spectra and spectral lines. The
coefficient kd has the dimension of the inverse of a length. It is proportional to
the number of scattering atoms per unit volume multiplied by a cross-section.
The function f (ν) gives the probability density that a photon is absorbed in a
frequency interval [ν, ν + dν].
(v) Concerning the scattering process, we consider monochromatic scattering and
complete frequency redistribution.
For monochromatic scattering

(t, r, n, ν, n , ν ) = ks (r)δ(ν − ν )f (ν), (2.6)

where δ(ν − ν ) is the Dirac distribution. This coefficient describes scatterings

of photons without change of frequency and a spherical (isotropic) indicatrix.
It applies to the formation of continuous spectra.
2.1 The Integro-Differential Radiative Transfer Equations 15

For complete frequency redistribution,

(t, r, n, ν, n , ν ) = ks (r)f (ν )f (ν). (2.7)

This coefficient is encountered in the formation of spectral lines. The atomic

transitions are occurring between bounded atomic levels, broadened by radia-
tive decay (natural broadening) and by collisions with particles such as
electrons or neutral hydrogen atoms. Because of the broadening of the levels,
absorbed and scattered photons have in general different frequencies. When
there is a high rate of collisions, these frequencies can be fully uncorrelated
and the scattering coefficient may be written as above. Complete frequency
redistribution is the term used for this situation.
In this book, we consider only isolated spectral line. The function f (ν), is
referred to as the absorption profile and following a well establish custom,
it will be henceforth denoted ϕ(ν). It takes its maximum value at the line
center frequency νc and typically has the form of a Gaussian or of a Voigt
function, which is the convolution product of a Gaussian and a Lorentzian
(see Sect. 2.1.2). All the absorption profiles, which will be considered, are
normalized to unity, that is,
 ∞
ϕ(ν) dν = 1. (2.8)
0

Equation (2.4), which expresses the detail balance between the absorptions and
emissions takes then the form

χs (t, r, n, ν) = ks (r)ϕ(ν). (2.9)

(vi) For spectral lines and continuous spectra, the primary source of photons often
comes from the radiative decay of an atomic level (bound or free) that has been
excited by a collision. When the colliding particles have a Maxwellian velocity
distribution, the primary source has the form

Q∗ (r, n, ν) = kd (r)ϕ(ν)Q∗ν (r), (2.10)

where Q∗ν (r) stands for the Planck function at the point r and at the
frequency ν.
With the assumptions listed above, Eq. (2.3) becomes

∂
μ I (z, ν, μ) = −κ(z)ϕ(ν)I (z, ν, μ)
∂z
 
1 +1 ∞
+ ks (z) R(ν, ν )I (z, ν , μ ) dν dμ + kd (z)ϕ(ν)Q∗ν (z), (2.11)
2 −1 0
16 2 Radiative Transfer Equations

where

κ(z) ≡ kd (z) + ks (z). (2.12)

In the scattering integral,

R(ν, ν ) = δ(ν − ν )ϕ(ν ), or R(ν, ν ) = ϕ(ν)ϕ(ν ), (2.13)

respectively for monochromatic scattering and complete frequency redistribution.

We introduce the optical depth τ , defined by

dτ = −κ(z) dz. (2.14)

The use of the optical depth as space variable is quite popular in radiative transfer
because it provides a scale for the extinction of the radiation field. For a semi-infinite
medium, it is positive inside the medium with τ = 0 at the surface (see Fig. 2.1).
Equation (2.11) can thus be rewritten as

∂
μ I (τ, ν, μ) = ϕ(ν)I (τ, ν, μ)
∂τ
 
1 +1 ∞
− [1 − (τ )] R(ν, ν )I (τ, ν , μ ) dν dμ − (τ )ϕ(ν)Q∗ν (τ ), (2.15)
2 −1 0

where
kd (τ ) kd (τ )
(τ ) = = . (2.16)
κ(τ ) kd (τ ) + ks (τ )

The coefficient [1 − (τ )] gives the probability that a photon that has been absorbed
will be reemitted and (τ ) that it will be destroyed (its energy turned into thermal
energy). The notation  for the destruction probability per scattering can already by
found in Eddington (1929) and has been continuously in use (e.g. Thomas 1957).
For continuous spectra (monochromatic scattering), kd (τ ) is due to bound-free
and free-free transitions and ks (τ ), the scattering coefficient, is due to, e.g., Rayleigh
or Thomson scattering. For a spectral line, kd (τ ) is the absorption coefficient from
the lower level of the transition and ks (τ ) is related to the spontaneous radiative
deexcitation from the upper level of the transition. For a two-level atom, the
destruction coefficient may be written as

C21 (τ )
(τ ) = , (2.17)
A21 + C21 (τ )

where A21 is the Einstein coefficient for spontaneous emission, typically around
107 − 108 s−1 , and C21 (τ ) a rate of de-population of the excited level by inelastic
2.1 The Integro-Differential Radiative Transfer Equations 17

collisions, mainly with electrons. It depends on local physical conditions (tempera-

ture and density). For resonance lines1  can be very small, of order 10−3 or less.
In the following we assume that (τ ) is a constant independent of τ , simply
denoted . Exact solutions exist only under this condition. When  = 0, all
the absorbed photons are scattered. The scattering (or the medium) is said to be
conservative. When  = 1, there is no scattering, hence no long range coupling
between different regions. All the sources and sinks of photons are local, and the
radiative transfer equation becomes an ordinary differential equation. In an infinite
medium, the mean number of scattering events that a photon can suffer before its
energy is returned to thermal energy is of order 1/. Radiative transfer equations are
often written in terms of the single scattering albedo 1 − , usually denoted λ or 
or a.
We now rewrite the transfer equation in Eq. (2.15) separately for monochromatic
scattering and complete frequency redistribution, assuming, as everywhere in this
book, that  is a constant.

2.1.1 Monochromatic Scattering

Monochromatic scattering was first formulated by Schwarzschild (1906) as a model

for the radiative equilibrium of a stellar atmosphere with a grey absorption coeffi-
cient, i.e., independent of frequency (see Sect. 2.2.2). Monochromatic scattering
is also a reasonable approximation for the transport of fast and slow neutrons
(Kourganoff 1963; Case and Zweifel 1967; Ivanov 1973).
The radiative transfer equation can be treated frequency by frequency, since
R(ν, ν ) = δ(ν − ν )ϕ(ν ). Introducing the monochromatic optical depth τν , defined
by dτν = −ϕ(ν)κ(z) dz, henceforth simply denoted τ , and the notation I (τ, μ) for
the radiation field, Eq. (2.15) may be written as
 +1
∂I 1
μ (τ, μ) = I (τ, μ) − (1 − ) I (τ, μ ) dμ − Q∗ (τ ). (2.18)
∂τ 2 −1

Here Q∗ (τ ) stands for Q∗ν (τ ). Equation (2.18) with  = 0 describes the radiative
equilibrium of a grey stellar atmosphere (see Sect. 2.2.2).
It is convenient to rewrite the radiative transfer equation as

∂I
μ (τ, μ) = I (τ, μ) − S(τ ), (2.19)
∂τ

1 For a resonance line, the lower level is the atomic fundamental level.
18 2 Radiative Transfer Equations

where the so-called source function S(τ ) is defined by

 +1
1
S(τ ) ≡ (1 − ) I (τ, μ) dμ + Q∗ (τ ). (2.20)
2 −1

The first term expresses the contribution of the photons that have been scattered at
least once and the second term yields the primary source of photons. The source
function depends only on the space variable and satisfies an integral equation of the
convolution type (see Sect. 2.2). For this reason, it plays a fundamental role in the
solution of radiative transfer equations.

2.1.2 Complete Frequency Redistribution

As pointed out by Eddington2 (1929), monochromatic scattering, or coherent

scattering as it used to be called, cannot be employed for the study of spectral lines
because it ignores the natural damping of the energy levels. It also ignores broaden-
ing by collisions with electrons and other species and the thermal movements of the
atoms (see Woolley and Stibbs 1953; Hummer 1962, for a history of the subject).
The opposite assumption, namely complete frequency redistribution, was found to
be adequate for a large majority of spectral lines formed in stellar atmospheres,
the main cause for the absence of correlation being the perturbation of the excited
levels of the transitions by collisions with passing atoms. The first investigations on
multiple scattering with complete frequency were made independently by Holstein
(1947) and Biberman (1947).
For very strong lines, this assumption will break down. Because of the very short
mean life time of the excited level, photons can be reemitted without changing
their frequency and there is, as one says partial frequency redistribution. In this
case, radiative transfer problems have in general no exact solutions. Attempts have
been made by Hemsh and Ferziger (1972) and by Yengibarian and Khachatrian
(1991) at finding exact solutions for the resonant scattering of gamma-ray quanta
by Mösbauer nuclei, a process which can be described by a linear combination of
monochromatic scattering and complete frequency redistribution. The authors could
show that the problem can be reduced to the construction of a function depending
only on the optical depth.
When dealing with one single isolated spectral line, as done here, it is convenient
to introduce a dimensionless frequency variable,

ν − νc
x= , (2.21)
νD

2 Arthur Eddington (Sir): 1882–1944 (Kendal-Cambridge, UK).

2.1 The Integro-Differential Radiative Transfer Equations 19

where νc is the frequency of the line center and νD , the Doppler width. The
Doppler width measures the broadening of the absorption line profile by small scale
velocity fields, thermal and turbulent. It has the form
νc 2 2
νD = (v + vturb )1/2 , (2.22)
c th

with vturb the turbulent velocity and vth = (2kT /M)1/2, the thermal velocity. We
recall that k is the Boltzmann constant, T the temperature of the medium, M the
mass of the scattering atom and c the speed of light. The Doppler width varies
in general slowly inside a stellar atmosphere. Here we take it as a constant. The
frequency νc being very large for spectral lines in the optical domain (νc 1015 −
1016 Hz), we make the usual assumption that x varies from minus infinity to plus
infinity, with x = 0 at line center.
Introducing now R(ν, ν ) = ϕ(ν)ϕ(ν ) into Eq. (2.15) and the frequency variable
x, we can write the radiative transfer equation as

∂I
μ (τ, x, μ) = ϕ(x)I (τ, x, μ)
∂τ
 +∞ +1
1
−(1 − ) ϕ(x) ϕ(x )I (τ, x , μ ) dμ dx − ϕ(x)Q∗x (τ ). (2.23)
2 −∞ −1

Over the width of a spectral line, the frequency variation of the Planck function
is in general very small and the value of Q∗x (τ ) can be taken at the line center
frequency. Here we assume that Q∗x (τ ) is independent of frequency and denote it
Q∗ (τ ). For the standard two-level atom model (see e.g. Mihalas 1978),  is given
by Eq. (2.16) and may take very small values, say, 10−4 or less. The normalization
of the absorption profile in Eq. (2.8) becomes
 +∞
ϕ(x) dx = 1. (2.24)
−∞

The optical depth τ is known as the frequency integrated or frequency averaged line
optical depth.
For spectral lines in the optical range, the absorption profile is in general a Voigt
function. A Voigt function is the convolution of a Lorentz function and of a Doppler
function, namely
 +∞ e−y
2
a
U (x, a) = √ dy. (2.25)
π π −∞ a 2 + (x − y)2
20 2 Radiative Transfer Equations

√
The function H (x, a) = π U (x, a) is the so-called Voigt function. In contrast to
U (x, a) it is not normalized to unity. Here we refer to U (x, a) as the Voigt function.
The Lorentz function,

1 a
ϕL (x) = , (2.26)
π a + x2
2

takes into account the natural width of the energy levels and their broadening by
collisions with electrons and other species. The parameter a is a positive constant,
called the Voigt parameter of the line, defined by a = /4πνD , where , measured
in s −1 , is the line damping rate (see, e.g., Landi Degl’Innocenti and Landolfi 2004,
p. 590). In stellar atmospheres, typical values for a are 10−1 –10−3 . The Doppler
function,

1
ϕD (x) = √ e−x ,
2
(2.27)
π

takes care of the velocity distribution of the atoms, assumed to be a Maxwellian

distribution. A Voigt function has a Gaussian core and Lorentzian wings for
frequencies greater than (− ln a)1/2. Several approximations for the Voigt function
can be found in the literature (see, e.g., Hummer 1965, Hubeny and Mihalas 2015,
p. 320, Landi Degl’Innocenti and Landolfi 2004, p. 167). A simple one valid for
small values of a and x = 0 is

1 a 3
√ e−x +
2
U (x, a) [1 + 2 + . . .]. (2.28)
π πx 2 2x

Voigt profiles for three different values of the parameter a smaller than one are
shown in Fig. 2.2. The Doppler profile corresponds to a = 0.3
The exact methods of solutions presented in this book hold if ϕ(x) is an even
function of x, is decreasing to zero monotonously as |x| → ±∞ and is integrable,
that is, satisfies some normalization condition such as Eq. (2.24).
For complete frequency redistribution, Eq. (2.23) can also be written as

∂I
μ (τ, x, μ) = ϕ(x)[I (τ, x, μ) − S(τ )]. (2.29)
∂τ
The source function S(τ ), which contains the primary sources of photons and the
contribution from the scattered photons, is defined by
 +1  +∞
1
S(τ ) ≡ (1 − ) ϕ(x)I (τ, x, μ) dμ dx + Q∗ (τ ). (2.30)
2 −1 −∞

3 Voigt profiles for a larger than one, can be found in https://en.wikipedia.org/wiki/Voigt_profile.

2.1 The Integro-Differential Radiative Transfer Equations 21

Fig. 2.2 The Doppler profile, and several Voigt profiles with different values of the parameter a.
The left panel shows ϕ(x) and the right panel log ϕ(x). In the right panel, the Lorentzian wing
regime characterized by ϕ(x) ∼ 1/x 2 can be observed for |x| larger than (− ln a)1/2

Because of the assumption of complete frequency redistribution S(τ ) depends only

on the optical depth.
The definition of the source function, given in Eq. (2.30) for a spectral line
formed with complete frequency redistribution and in Eq. (2.20) for monochromatic
scattering, is independent of the description of the atomic structure of the radiating
atom. From an atomic point of view, the source function for a simple two-level
atom model, assuming complete frequency redistribution, and neglecting stimulated
emission, has the form

2hνc3 gl nu
S(τ ) = , (2.31)
c2 gu nl

where νc is the frequency at line center, h the Planck constant, c the speed of
light, gl , nl and gu , nu the statistical weights and the populations of the lower
and upper levels of the transition. The populations of the levels satisfy statistical
equilibrium equations, which describe the balance between population and de-
population by radiative and collisonal processes. When collisions dominate over
radiative processes, the populations of the levels follow the Saha–Boltzmann law
and the source function reduces to the Planck function. This situation is known
in Astronomy as Local Thermodynamic Equilibrium (LTE). In contrast, when
radiative processes have a dominant role, which is the case for strong resonance
lines, the source function will depend on the radiation field. For simple atomic
models it will take the form written in Eq. (2.30), where Q∗ (τ ) is the Planck
function at the frequency of the line center. This physical picture, known as Nonlocal
Thermodynamic Equilibrium (NLTE) was stressed in the late fifties by J. T. Jefferies
and R. N. Thomas (for historical references, see Hubeny and Mihalas 2015, p. 13).
For spectral lines formed in NLTE, the determination of the radiation field demands
that statistical equilibrium equations and radiative transfer equations be solved
simultaneously (see, e.g., Landi Degl’Innocenti and Landolfi 2004; Hubeny and
Mihalas 2015).
22 2 Radiative Transfer Equations

2.1.3 The Diffuse Radiation Field

Primary photons can be created inside the medium, but can also come from outside.
It is a typical situation for planetary atmospheres, for which the main and usually
only source of energy comes from an outside illumination. The radiation field inside
the atmosphere and the reemitted fraction, known as the albedo, can be derived
from the radiative transfer equation. There are two different ways of handling the
incident radiation: as a boundary condition for the radiative transfer equation, or by
introducing a diffuse field. We describe this method, for a one-dimensional medium,
say a semi-infinite one, first for monochromatic scattering and then for complete
frequency redistribution. Diffuse fields are used in several chapters, for unpolarized
and polarized radiation.
We assume that a field I inc (μ) is incident on the surface at τ = 0. The idea of
the method is to separate the field in two parts: a direct field and a diffuse field. The
diffuse field denoted I d (τ, μ) is defined by

I d (τ, μ) ≡ I (τ, μ) − I inc (μ) e−τ/|μ| , μ < 0,

I d (τ, μ) ≡ I (τ, μ), μ > 0. (2.32)

With our convention, μ < 0 for the radiation incident on the surface and μ > 0
for the radiation emerging from the medium. The diffuse field is zero at the surface
for incoming directions. It is equal to the total field for the outgoing directions and
at τ = 0 yields the emerging radiation. Introducing Eq. (2.32) into Eq. (2.18), one
obtains for I d (τ, μ) the equation
 +1
∂I d 1
μ (τ, μ) = I d (τ, μ) − (1 − ) I d (τ, μ ) dμ − Q∗t (τ ), (2.33)
∂τ 2 −1

with
 1
1
Q∗t (τ ) ∗
= Q (τ ) + (1 − ) I inc (−μ)e−τ/μ dμ. (2.34)
2 0

The second term in the right-hand side describes the scattering of the incident field
inside the medium. The important point is that it depends only on τ . The source
function for the diffuse field is
 +1
1
S(τ ) = (1 − ) I d (τ, μ) dμ + Q∗t (τ ). (2.35)
2 −1

For complete frequency redistribution, one introduces in a similar way a diffuse

field I d (τ, x, μ) defined by

I d (τ, x, μ) = I (τ, x, μ) − I inc (x, μ)e−τ ϕ(x)/|μ| , μ < 0, (2.36)

2.2 Integral Equations for the Source Function 23

I d (τ, x, μ) = I (τ, x, μ), μ > 0. (2.37)

It satisfies Eq. (2.23) with the internal primary source Q∗ (τ ) replaced by
 1  +∞
1
Q∗t (τ ) = Q∗ (τ ) + (1 − ) ϕ(x)I inc (x, −μ) e−τ ϕ(x)/μ dx dμ.
2 0 −∞
(2.38)

The second term in the right-hand side describes the scattering of the incident
photons and depends only on τ , just as with monochromatic scattering.
Radiative transfer problems with no internal source but an incident radiation
are usually referred to as diffuse reflection problems. Thanks to the introduction
of a diffuse field, any exact method of solution applicable to a problem with an
internal primary source, can also be employed when the the medium is illuminated
from outside. With the Case singular eigenfunction expansion method described in
Chap. 10, it is actually preferable to work with the original radiation field and the
associated boundary condition.

2.2 Integral Equations for the Source Function

We now construct integral equations of the convolution type for the monochromatic
and for the complete redistribution source functions. The method is quite simple.
It suffices to introduce the solution of the transfer equations, expressed in terms of
S(τ ), into the defining equations for S(τ ). Monochromatic scattering and complete
frequency redistribution lead to similar integral equations.

2.2.1 Monochromatic Scattering

Solving Eq. (2.19) as if S(τ ) were a known function, we can write


T −τ T τ − τ dτ
I (τ, μ) = I (T , μ) exp(− )+ S(τ ) exp(− ) , (2.39)
μ τ μ μ

where T and τ are two points inside the medium. When T > τ , Eq. (2.39) holds for
μ > 0 and when T < τ , it holds for μ < 0. To calculate the radiation field we need
boundary conditions. They depend on the geometry of the medium.
For an infinite medium, physically meaning full solutions of the radiative transfer
equation should grow less rapidly than an exponential at plus and minus infinity.
This condition, applied at T = +∞ for μ > 0 and at T = −∞ for μ < 0, implies
that the first term in the right-hand side of Eq. (2.39) is zero. Hence, in an infinite
Random documents with unrelated
content Scribd suggests to you:
The Project Gutenberg eBook of Studies in
Greek Scenery, Legend and History
 This ebook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this ebook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.

Title: Studies in Greek Scenery, Legend and History

Author: James George Frazer

Release date: November 19, 2017 [eBook #56002]

Most recently updated: October 23, 2024

Language: English

Credits: Produced by KD Weeks, Turgut Dincer and the Online

Distributed Proofreading Team at http://www.pgdp.net
(This
file was produced from images generously made
available
by The Internet Archive)

*** START OF THE PROJECT GUTENBERG EBOOK STUDIES IN

GREEK SCENERY, LEGEND AND HISTORY ***
 Transcriber’s Note:
Footnotes have been collected at Marginal notes
the end of each chapter, and are appear
linked for ease of reference. like this.
Minor errors, attributable to the printer, have been
corrected. Please see the transcriber’s note at the end of
this text for details regarding the handling of any textual
issues encountered during its preparation.
The cover image has been created based on title page
information, and is placed in the public domain.
Any corrections are indicated using an underline
highlight. Placing the cursor over the correction will
produce the original text in a small popup.
 STUDIES
IN
GREEK SCENERY, LEGEND
AND HISTORY
 BY THE SAME AUTHOR
Pausanias’ Description of Greece. Translated by Sir J. G. Frazer,
D.C.L., Litt.D. With commentary, illustrations, and maps. Second
Edition. 6 vols. 8vo. 126s. net.
The Golden Bough. A Study in Magic and Religion. Third
Edition, revised and enlarged. 8vo.
Part 1. The Magic Art and the Evolution of Kings. 4th
Impression. Two vols. 20s. net.
Part 2. Taboo and the Perils of the Soul. 3rd Impression.
One vol. 10s. net.
Part 3. The Dying God. 3rd Impression. One vol. 10s. net.
Part 4. Adonis, Attis, Osiris. Two vols. Third Edition. 20s. net.
Part 5. Spirits of the Corn and of the Wild. 2nd Impression.
Two vols. 20s. net.
Part 6. The Scapegoat. One vol. 10s. net.
Part 7. Balder the Beautiful: The Fire-Festivals of Europe,
and the Doctrine of the External Soul. 2nd Impression.
Two vols. 20s. net.
Vol. XII. Bibliography and General Index. 20s. net.
Lectures on the Early History of the Kingship. 8vo. 8s. 6d. net.
Totemism and Exogamy. A Treatise on Certain Early Forms of
Superstition and Society. With Maps. 4 vols. 8vo. 50s. net.
The Belief in Immortality and the Worship of the Dead. Vol. I.
The Belief among the Aborigines of Australia, the Torres Straits
Islands, New Guinea, and Melanesia. 8vo. 10s. net.
Psyche’s Task. A Discourse concerning The Influence of
Superstition on the Growth of Institutions. Second Edition, revised
and enlarged, to which is added “The Scope of Social
Anthropology.” 8vo. 5s. net.
Essays of Joseph Addison. Chosen and edited, with a Preface and
a few Notes, by Sir J. G. Frazer. 2 vols. Gl. 8vo. 10s. net.
Letters of William Cowper. Chosen and edited, with Memoir and
Notes, by Sir J. G. Frazer. 2 vols. Gl. 8vo. 10s. net.
LONDON: MACMILLAN AND CO., Ltd.
 STUDIES
IN

GREEK SCENERY, LEGEND

AND HISTORY
SELECTED FROM HIS COMMENTARY ON
PAUSANIAS’ ‘DESCRIPTION OF GREECE’

BY

SIR JAMES GEORGE FRAZER

FELLOW OF TRINITY COLLEGE, CAMBRIDGE

AUTHOR OF ‘THE GOLDEN BOUGH’

MACMILLAN AND CO., LIMITED

ST. MARTIN’S STREET, LONDON
1917
First published under the title, “Pausanias and other Greek Sketches,” 1900
Reprinted 1917
 PREFACE

The Englishman in Greece who pays any heed to the remains of

classical antiquity is apt, if he be no scholar, to wonder who a certain
Pausanias was whose authority he finds often quoted on questions
of ancient buildings and sites. The first of the following sketches may
do something to satisfy his curiosity on this head. It has already
served as an introduction to a version of Pausanias’s Description of
Greece which I published with a commentary two years ago. The
account of Pericles was contributed to the ninth edition of the
Encyclopaedia Britannica. I desire to thank Messrs. A. and C. Black
for their courteous permission to republish it. The other sketches are
reprinted, with some small changes and adjustments of detail, from
my commentary on Pausanias. References to authorities have been
omitted as needless in a book which is not specially addressed to the
learned. Any one who wishes to pursue the subject further will find
my authorities amply cited in the original volumes. Among works
from which I have borrowed both outlines and colours for some of
my sketches of Greek landscape I will here mention only two—the
Erinnerungen und Eindrücke aus Griechenland of the Swiss scholar
W. Vischer, and the Peloponnes of the German geologist Mr. A.
Philippson. Slight and fragmentary as these sketches are, I am not
without hope that they may convey to readers who have never seen
Greece something of the eternal charm of its scenery. To such as
already know and love the country they will yet be welcome, if here
and there they revive some beautiful or historic scene on those
tablets of the mind from which even the brightest hues so quickly
fade.
J. G. F.
Cambridge, March 30, 1900.
 CONTENTS

PAGE
1. Pausanias and his Description of Greece 1
2. Oropus 160
3. Rhamnus 163
4. Marathon 165
5. Prasiae 174
6. Mount Hymettus 178
7. Mount Pentelicus 182
8. Phyle 187
9. The Port of Athens 191
10. The Sacred Way 209
11. The Hall of Initiation at Eleusis 214
12. Eleutherae 216
13. Megara 219
14. The Scironian Road 220
15. The Isthmus of Corinth 223
16. The Bath of Aphrodite 226
17. The Prospect from Acro-Corinth 227
18. The Capture of Corinth by Aratus 228
19. Sicyon 232
20. Phliasia 233
21. Nemea 237
22. The Pass of the Tretus 238
23. Mycenae 242
24. The End of the Mycenaean Age 245
25. Mount Arachnaeus 248
26. Epidaurus 249
27. The Temple in Aegina 251
28. The Sanctuary of Poseidon in Calauria 252
29. Troezen 253
30. From Troezen to Epidaurus 255
31. Methana 260
32. Nauplia 261
33. The Springs of the Erasinus 263
34. The Lernean Marsh 266
35. The Anigraean Road 269
36. The Battlefield of Sellasia 270
37. Sparta 271
38. Mistra 274
39. On the Road from Sparta to Arcadia 278
40. Cape Malea 279
41. Monemvasia 281
42. Maina 282
43. Pharae and the Messenian Plain 284
44. Messene 285
45. On the Road to Olympia 287
46. Olympia 290
47. Phidias’s Image of Olympian Zeus 292
48. The Hermes of Praxiteles 293
49. Lasion 296
50. The Erymanthus 299
51. The Monastery of Megaspeleum 300
52. The Gulf of Corinth 301
53. On the Coast of Achaia 303
54. Pellene 304
55. The Road from Argos to Arcadia 306
56. Mantinea 308
57. The Road to Stymphalus 310
58. The Lake and Valley of Stymphalus 312
59. The Lake of Pheneus 315
60. From Pheneus to Nonacris 320
61. The Fall of the Styx 324
62. The Valley of the Aroanius 330
63. The Springs of the Ladon 331
64. The Gorge of the Ladon 333
65. Aliphera 336
66. Dimitsana 337
67. Gortys 338
68. The Plain of Megalopolis 342
69. The Cave of the Black Demeter 343
70. The Temple of Apollo at Bassae 345
71. The Temple of Artemis at Aulis 346
72. Glaucus’s Leap 347
73. Evening on the Euripus 349
74. The Copaic Lake 349
75. The Great Katavothra 355
76. The Vale of the Muses 357
77. Hippocrene 358
78. Lebadea 359
79. The Boeotian Orchomenus 361
80. The Plain of Chaeronea 364
81. Panopeus 364
82. Near Hyampolis 365
83. Tithorea 366
84. From Amphissa to Gravia 369
85. Daulis 371
86. The Cleft Way 373
87. Delphi 374
88. Aeschines at Delphi 378
89. The Pythian Tune 379
90. The Lacedaemonian Trophy at Delphi 380
91. The Gods in Battle 382
92. The Sibyl’s Wish 384
93. Orpheus in Hell 386
94. The Acheron 387
95. A Ride across Parnassus 389
96. Pericles 392

PAUSANIAS | AND OTHER GREEK SKETCHES

I. Pausanias and his Description of Greece. Greece in
the second
—It may be reckoned a peculiar piece of good century
fortune that among the wreckage of classical A.D.
literature the Description of Greece by Pausanias
should have come down to us entire. In this work we possess a
plain, unvarnished account by an eye-witness of the state of Greece
in the second century of our era. Of no other part of the ancient
world has a description at once so minute and so trustworthy
survived, and if we had been free to single out one country in one
age of which we should wish a record to be preserved, our choice
might well have fallen on Greece in the age of the Antonines. No
other people has exerted so deep and abiding an influence on the
course of modern civilisation as the Greeks, and never could all the
monuments of their chequered but glorious history have been
studied so fully as in the second century of our era. The great age of
the nation, indeed, had long been over, but in the sunshine of peace
and imperial favour Greek art and literature had blossomed again.
New temples had sprung up; new images had been carved; new
theatres and baths and aqueducts ministered to the amusement and
luxury of the people. Among the new writers whose works the world
will not willingly let die, it is enough to mention the great names of
Plutarch and Lucian.
It was in this mellow autumn—perhaps rather the Indian summer
—of the ancient world, when the last gleanings of the Greek genius
were being gathered in, that Pausanias, a contemporary of Hadrian,
of the Antonines, and of Lucian, wrote his description of Greece. He
came in time, but just in time. He was able to describe the stately
buildings with which in his own lifetime Hadrian had embellished
Greece, and the hardly less splendid edifices which, even while he
wrote, another munificent patron of art, Herodes Atticus, was
rearing at some of the great centres of Greek life and religion. Yet
under all this brave show the decline had set in. About a century
earlier the emperor Nero, in the speech in which he announced at
Corinth the liberation of Greece, lamented that it had not been given
him to confer the boon in other and happier days when there would
have been more people to profit by it. Some years after this imperial
utterance Plutarch declared that the world in general and Greece
especially was depopulated by the civil brawls and wars; the whole
country, he said, could now hardly put three thousand infantry in the
field, the number that formerly Megara alone had sent to face the
Persians at Plataea; and in the daytime a solitary shepherd feeding
his flock was the only human being to be met with on what had
been the site of one of the most renowned oracles in Boeotia. Dio
Chrysostom tells us that in his time the greater part of the city of
Thebes lay deserted, and that only a single statue stood erect
among the ruins of the ancient market-place. The same picturesque
writer has sketched for us a provincial town of Euboea, where most
of the space within the walls was in pasture or rig and furrow, where
the gymnasium was a fruitful field in which the images of Hercules
and the rest rose here and there above the waving corn, and where
sheep grazed peacefully about the public offices in the grass-grown
market-place. In one of his Dialogues of the Dead, Lucian represents
the soul of a rich man bitterly reproaching himself for his rashness in
having dared to cross Cithaeron with only a couple of men-servants,
for he had been set upon and murdered by robbers on the highway
at the point where the grey ruins of Eleutherae still look down on the
pass; in the time of Lucian the district, laid waste, he tells us, by the
old wars, seems to have been even more lonely and deserted than it
is now. Of this state of things Pausanias himself is our best witness.
Again and again he notices shrunken or ruined cities, deserted
villages, roofless temples, shrines without images and pedestals
without statues, faint vestiges of places that once had a name and
played a part in history. To the site of one famous city he came and
found it a vineyard. In one neglected fane he saw a great ivy-tree
clinging to the ruined walls and rending the stones asunder. In
others nothing but the tall columns standing up against the sky
marked the site of a temple. Nor were more sudden and violent
forces of destruction wanting to hasten the slow decay wrought by
time, by neglect, by political servitude, by all the subtle indefinable
agencies that sap a nation’s strength. In Pausanias’s lifetime a horde
of northern barbarians, the ominous precursor of many more,
carried fire and sword into the heart of Greece, and the Roman
world was wasted by that great pestilence which thinned its
population, enfeebled its energies, and precipitated the decline of
art.
The little we know of the life of Pausanias is gathered entirely
from his writings. Antiquity, which barely mentions the writer, is
silent as to the man.
Fortunately his date is certain. At the beginning Date of
of his description of Elis he tells us that two Pausanias.
hundred and seventeen years had elapsed since the restoration of
Corinth. As Corinth was restored in 44 B.C., we see that Pausanias
was writing his fifth book in 174 A.D. during the reign of Marcus
Aurelius. With this date all the other chronological indications in his
book harmonise. Thus he speaks of images which were set up in 125
A.D. as specimens of the art of his day. Again, he gives us to
understand that he was a contemporary of Hadrian’s, and he tells us
that he never saw Hadrian’s favourite, Antinous, in life. Now Hadrian
died in 138 A.D., and the mysterious death of Antinous in Egypt
appears to have fallen in 130 A.D. It is natural to infer from
Pausanias’s words that though he never saw Antinous in life, he was
old enough to have seen him; from which we conclude that our
author was born a good many years before 130 A.D., the date of
Antinous’s death. The latest historical event mentioned by him is the
incursion of the Costobocs into Greece, which seems to have taken
place some time between 166 A.D. and 180 A.D., perhaps in 176 A.D.
From these and a few more hints we may draw Dates of
some conclusions as to the dates when the the various
various books that make up the Description of books.
Greece were written. In the seventh book Pausanias tells us that his
description of Athens was finished before Herodes Atticus built the
Music Hall in memory of his wife Regilla. As Regilla appears to have
died in 160 or 161 A.D. and the Music Hall was probably built soon
afterwards, we may suppose that Pausanias had finished his first
book by 160 or 161 A.D. at latest. There is, indeed, some ground for
holding that both the first and the second book were composed
much earlier. For in the second book Pausanias mentions a number
of buildings which had been erected in his own lifetime by a Roman
senator Antoninus in the sanctuary of Aesculapius at Epidaurus. If,
as seems not improbable, the Roman senator was no other than the
Antoninus who afterwards reigned as Antoninus Pius, we should
naturally infer that the second book was published in the reign of
Hadrian, that is, not later than 138 A.D., the year when Hadrian died
and Antoninus succeeded him on the throne. With this it would
agree that no emperor later than Hadrian is mentioned in the first or
second book, or indeed in any book before the eighth. Little weight,
however, can be attached to this circumstance, for in the fifth book
Hadrian is the last emperor mentioned although that book was
written, as we have seen, in the reign of Marcus Aurelius, thirty-six
years after Hadrian’s death. A much later date has been assigned to
the second book by Mr. W. Gurlitt in his valuable monograph on
Pausanias. He points out that when Pausanias wrote it the sanctuary
of Aesculapius at Smyrna had already been founded, and that if
Masson’s chronology of the life of the rhetorician Aristides is right
the sanctuary was still unfinished in 165 A.D. Hence Mr. Gurlitt
concludes that the second book of Pausanias was written after 165
A.D. Even the first book, according to him, must be dated not earlier
than 143 A.D. His reason is that when Pausanias wrote this book the
stadium at Athens had already been rebuilt of white marble by
Herodes Atticus, and that the reconstruction cannot, if Professor C.
Wachsmuth is right, have been begun before 143 A.D. or a little
earlier. With regard to the other books, the evidence, scanty as it is,
is less conflicting. The fifth book, as we have seen, was composed in
the year 174 A.D. The eighth book, in which mention is made of the
victory of Marcus Antoninus over the Germans, must have been
written after 166 A.D., the year when the German war broke out, and
may have been written in or after 176 A.D., the year in which the
emperor celebrated a triumph for his success. In the tenth book
occurs the reference to the inroad of the Costobocs; hence the book
was written between 166 and 180 A.D. Further, the references which
Pausanias makes both forwards and backwards to the several parts
of his work show that the books were written in the order in which
they now stand. Hence books six to ten cannot have been composed
earlier, and may have been composed a good deal later, than 174
A.D., the year in which our author was engaged on his fifth book.
Thus the composition of the work extended over a period of at least
fourteen years and probably of many more. That Pausanias spent a
long time over it might be inferred from a passage in which he
explains a change in his religious views. When he began his work, so
he tells us, he looked on some Greek myths as little better than
foolishness, but when he had got as far as his description of Arcadia
he had altered his opinion and had come to believe that they
contained a kernel of deep wisdom under a husk of extravagance.
Such a total change of attitude towards the religious traditions of his
country was more probably an affair of years than of weeks and
months.
That the first book was not only written but published before the
others seems clear. Amongst the proofs of this the The first
strongest is the writer’s statement in the seventh book
book, that when he wrote his description of written and
published
Athens the Music Hall of Herodes Atticus had not before the
yet been built. This implies that when he wrote rest.
the seventh book the first was already published;
otherwise he could easily have incorporated a notice of the Music
Hall in its proper place in the manuscript. Again, in the eighth book
he expressly corrects a view which he had adopted in the first; this
also he might have done in the manuscript of the first book if he still
had it by him. In other places he tacitly adds to statements and
descriptions contained in the first book. Further, the narrative of the
Gallic invasion in the first book is superseded by the much fuller
narrative given in the tenth book, and would hardly have been
allowed to stand if it had been in the author’s power to cut it out.
More interesting are the passages in which we seem to discover
references to criticisms which had been passed on his first book.
Thus in the third book he repeats emphatically the plan of work
which he had laid down for himself in the first, adding that the plan
had been adopted after mature deliberation, and that he would not
depart from it. This sounds like a trumpet-blast of defiance to the
critics who had picked holes in the scheme of his first book.
Elsewhere he seems conscious that some of their strictures were not
wholly undeserved. In speaking of the descendants of Aristomenes
he is sorely tempted to go into the family history of the Diagorids,
but pulls himself up sharply with the remark that he passes over this
interesting topic “lest it should appear an impertinent digression.”
Clearly the arrows of the reviewers had gone home. The tedious
historical dissertations with which he had sought to spice the plain
fare of Athenian topography were now felt by the poor author
himself to savour strongly of impertinent digressions. Again, old
habit getting the better of him, the sight of a ruined camp of King
Philip in a secluded Arcadian valley sets him off rambling on the
divine retribution that overtook that wicked monarch and his
descendants and the murderers of his descendants and their
descendants after them, till, his conscience smiting him, he suddenly
returns to business with the half apology, “But this has been a
digression.” That Pausanias had the fear of the critics before his eyes
is stated by himself in the plainest language. He had made, he tells
us, careful researches into the vexed subject of the dates of Homer
and Hesiod, but refrained from stating the result of his labours,
because he knew very well the carping disposition of the professors
of poetry of his own day. Little did he foresee the disposition of
certain other professors who were to sit in judgment on him some
seventeen hundred years later. Had he done so he might well have
been tempted to suppress the Description of Greece altogether, and
we might have had to lament the loss of one of the most curious
and valuable records bequeathed to us by antiquity.
The birthplace of Pausanias is less certain than Birthplace
his date, but there are good grounds for believing of
that he was a Lydian. For after saying that in his Pausanias.
country traces were still to be seen of the abode of Pelops and
Tantalus, he mentions some monuments and natural features
associated with the names of these ancient princes on and near
Mount Sipylus. This is nearly a direct affirmation that the region
about Mount Sipylus in Lydia was his native land. The same thing
appears, though less directly, from the minute acquaintance he
displays with the district and from the evident fondness with which
he recurs again and again to its scenery and legends. He had seen
the white eagles wheeling above the lonely tarn of Tantalus in the
heart of the hills; he had beheld the stately tomb of the same hero
on Mount Sipylus, the ruined city at the bottom of the clear lake, the
rock-hewn throne of Pelops crowning the dizzy peak that overhangs
the cañon, and the dripping rock which popular fancy took for the
bereaved Niobe weeping for her children. He speaks of the clouds of
locusts which he had thrice seen vanish from Mount Sipylus, of the
wild dance of the peasantry, and of the shrine of Mother Plastene,
whose rude image, carved out of the native rock, may still be seen
in its niche at the foot of the mountain. From all this it is fair to
surmise that Pausanias was born and bred not far from the
mountains which he seems to have known and loved so well. Their
inmost recesses he may have explored on foot in boyhood and have
drunk in their old romantic legends from the lips of woodmen and
hunters. Whether, as some conjecture, he was born at Magnesia, the
city at the northern foot of Mount Sipylus, we cannot say, but the
vicinity of the city to the mountain speaks in favour of the
conjecture. It is less probable, perhaps, that his birthplace was the
more distant Pergamus, although there is no lack of passages to
prove that he knew and interested himself in that city. As a native of
Lydia it was natural that Pausanias should be familiar with the
western coast of Asia Minor. There is indeed no part of the world
outside of Greece to which he refers so often. He seizes an
opportunity to give us the history of the colonisation of Ionia, and
dwells with patriotic pride on the glorious climate, the matchless
temples, and the natural wonders of that beautiful land.
Some scholars have identified our author with a Other
sophist of the same name who was born at writers of
Caesarea in Cappadocia, studied under Herodes the same
name.
Atticus, and died an old man at Rome, leaving
behind him many declamations composed in a style which displayed
a certain vigour and some acquaintance with classical models. But,
quite apart from the evidence that our author was a Lydian, there
are strong reasons for not identifying him with his Cappadocian
namesake. Neither Suidas nor Philostratus, who has left us a short
life of the Cappadocian Pausanias, mentions the Description of
Greece among his works; and on the other hand our Pausanias,
though he often mentions Herodes Atticus, nowhere speaks of him
as his master or of any personal relations that he had with him.
Further, the author of the Description of Greece is probably to be
distinguished from a writer of the same name who composed a work
on Syria to which Stephanus of Byzantium repeatedly refers. It is
true that our Pausanias evidently knew and had travelled in Syria,
but this in itself is no reason for supposing that he was the author of
a work to which in his extant writings he makes no allusion. The
name Pausanias was far too common to justify us in identifying all
the authors who bore it, even when we have grounds for believing
them to have been contemporaries.
That Pausanias had travelled widely beyond the limits of Greece
and Ionia is clear from the many allusions he lets fall to places and
objects of interest in foreign lands. Some of them he expressly says
that he saw; as to others we may infer that he saw them from the
particularity of his description. In Syria he had seen the Jordan
flowing through the Lake of Tiberias and falling into the Dead Sea,
and had gazed at the red pool near Joppa in which Perseus was said
to have washed his bloody sword after slaying the sea-monster. He
describes a tomb at Jerusalem, the door of which by an ingenious
mechanical contrivance opened of itself once a year at a certain
hour, and he often alludes to Antioch which for its vast size and
wealth he ranked with Alexandria. In Egypt he had seen the
Pyramids, had beheld with wonder the colossal statue of Memnon at
Thebes, and had heard the musical note, like the breaking of a lute-
string, which the statue emitted at sunrise. The statue still stands,
and many inscriptions in Greek and Latin carved by ancient visitors
on its huge legs and base confirm the testimony of Pausanias as to
the mysterious sound. From Egypt our author seems to have
journeyed across the desert to the oasis of Ammon, for he tells us
that in his time the hymn which Pindar sent to Ammon was still to be
seen there carved on a triangular slab beside the altar. Nearer home
he admired the splendid fortifications of Rhodes and Byzantium.
Though he does not describe northern Greece, he had visited
Thessaly, and had seen the blue steaming rivulet rushing along at
the foot of the rugged forest-tufted mountains that hem in like a wall
the pass of Thermopylae on the south. He appears to have visited
Macedonia, and perhaps, too, Epirus; at least he speaks repeatedly
of Dodona and its oracular oak, and he mentions the sluggish
melancholy rivers that wind through the dreary Thesprotian plain
and that gave their names to the rivers in hell. He had crossed to
Italy and seen something of the cities of Campania and the wonders
of Rome. The great forum of Trajan with its bronze roof, the Circus
Maximus—then probably the most magnificent building in the world
—and the strange beasts gathered from far foreign lands, seem to
have been the sights which most impressed him in the capital of the
world. In the Imperial Gardens he observed with curiosity a tusk
which the custodian assured him had belonged to the Calydonian
boar; and he noticed, doubtless with less pleasure, the great ivory
image of Athena Alea which Augustus had carried off from the
stately temple of the goddess at Tegea. In the neighbourhood of
Rome the bubbling milk-white water of Albula or Solfatara, as it is
now called, on the road to Tibur, attracted his attention, and beside
the sylvan lake of Aricia he appears to have seen the grim priest
pacing sword in hand, the warder of the Golden Bough. The absurd
description he gives of the beautiful and much-maligned Strait of
Messina would suffice to prove that he never sailed through it.
Probably like most travellers coming from the East he reached Italy
by way of Brundisium. Of Sardinia he has given a somewhat full
description, but without implying that he had visited it. Sicily, if we
may judge by a grave blunder he makes in speaking of it, he never
saw.
The aim that Pausanias had in writing his Aim of
Description of Greece is nowhere very fully or Pausanias’s
clearly stated by him. His book has neither head work.
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com