Structure Formation In Astrophysics Chabrier G Ed

buallipuchol 3 views 89 slides May 21, 2025
Slide 1
Slide 1 of 89
Slide 1
1
Slide 2
2
Slide 3
3
Slide 4
4
Slide 5
5
Slide 6
6
Slide 7
7
Slide 8
8
Slide 9
9
Slide 10
10
Slide 11
11
Slide 12
12
Slide 13
13
Slide 14
14
Slide 15
15
Slide 16
16
Slide 17
17
Slide 18
18
Slide 19
19
Slide 20
20
Slide 21
21
Slide 22
22
Slide 23
23
Slide 24
24
Slide 25
25
Slide 26
26
Slide 27
27
Slide 28
28
Slide 29
29
Slide 30
30
Slide 31
31
Slide 32
32
Slide 33
33
Slide 34
34
Slide 35
35
Slide 36
36
Slide 37
37
Slide 38
38
Slide 39
39
Slide 40
40
Slide 41
41
Slide 42
42
Slide 43
43
Slide 44
44
Slide 45
45
Slide 46
46
Slide 47
47
Slide 48
48
Slide 49
49
Slide 50
50
Slide 51
51
Slide 52
52
Slide 53
53
Slide 54
54
Slide 55
55
Slide 56
56
Slide 57
57
Slide 58
58
Slide 59
59
Slide 60
60
Slide 61
61
Slide 62
62
Slide 63
63
Slide 64
64
Slide 65
65
Slide 66
66
Slide 67
67
Slide 68
68
Slide 69
69
Slide 70
70
Slide 71
71
Slide 72
72
Slide 73
73
Slide 74
74
Slide 75
75
Slide 76
76
Slide 77
77
Slide 78
78
Slide 79
79
Slide 80
80
Slide 81
81
Slide 82
82
Slide 83
83
Slide 84
84
Slide 85
85
Slide 86
86
Slide 87
87
Slide 88
88
Slide 89
89

About This Presentation

Structure Formation In Astrophysics Chabrier G Ed
Structure Formation In Astrophysics Chabrier G Ed
Structure Formation In Astrophysics Chabrier G Ed


Slide Content

Structure Formation In Astrophysics Chabrier G
Ed download
https://ebookbell.com/product/structure-formation-in-
astrophysics-chabrier-g-ed-2041774
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.
Structure Formation In The Universe Proceedings Of The Nato Advanced
Study Institute On Structure Formation In The Universe Cambridge Uk 26
July 6 August 1999 1st Edition Martin J Rees Auth
https://ebookbell.com/product/structure-formation-in-the-universe-
proceedings-of-the-nato-advanced-study-institute-on-structure-
formation-in-the-universe-cambridge-uk-26-july-6-august-1999-1st-
edition-martin-j-rees-auth-4206908
Structure Formation In Solution Ionic Polymers And Colloidal Particles
1st Edition Ikuo Sogami Norio Ise
https://ebookbell.com/product/structure-formation-in-solution-ionic-
polymers-and-colloidal-particles-1st-edition-ikuo-sogami-norio-
ise-977712
Structure Formation In Modified Gravity Cosmologies 1st Edition
Alexandre Barreira Auth
https://ebookbell.com/product/structure-formation-in-modified-gravity-
cosmologies-1st-edition-alexandre-barreira-auth-10313982
Structure Formation In Polymeric Fibers Salem David R
https://ebookbell.com/product/structure-formation-in-polymeric-fibers-
salem-david-r-12068492

Adsorption Aggregation And Structure Formation In Systems Of Charged
Particles From Colloidal To Supracolloidal Assembly 1st Edition
Bhuvnesh Bharti Auth
https://ebookbell.com/product/adsorption-aggregation-and-structure-
formation-in-systems-of-charged-particles-from-colloidal-to-
supracolloidal-assembly-1st-edition-bhuvnesh-bharti-auth-4931352
Structure Phenomenology Preconscious Formation In The Epistemic
Disclosure Of Reality Herbert Witzenmann
https://ebookbell.com/product/structure-phenomenology-preconscious-
formation-in-the-epistemic-disclosure-of-reality-herbert-
witzenmann-48792272
Multiscale Structure Formation And Dynamics In Cosmic Plasmas 1st
Edition Andre Balogh
https://ebookbell.com/product/multiscale-structure-formation-and-
dynamics-in-cosmic-plasmas-1st-edition-andre-balogh-5353444
Snow Crystals A Case Study In Spontaneous Structure Formation Kenneth
G Libbrecht
https://ebookbell.com/product/snow-crystals-a-case-study-in-
spontaneous-structure-formation-kenneth-g-libbrecht-46707950
Crystallization Modalities In Polymer Melt Processing Fundamental
Aspects Of Structure Formation 1st Edition Hermann Janeschitzkriegl
Auth
https://ebookbell.com/product/crystallization-modalities-in-polymer-
melt-processing-fundamental-aspects-of-structure-formation-1st-
edition-hermann-janeschitzkriegl-auth-4194396

This page intentionally left blank

STRUCTURE FORMATION IN ASTROPHYSICS
Understanding the formation of objects at all scales in the universe, from galaxies
to stars and planets, is a major issue in modern astrophysics, and one of the most
exciting challenges of twenty−first century astronomy. Even though they are charac−
terized by different scales, the formation of planets, stars and galaxies share many
common physical processes and are rooted in the same underlying domains of
physics.
This unique reference for graduate students and researchers in astrophysics is the
first to cover structure formation on various scales in one volume. This book gath−
ers together extensive reviews written by world experts in physics and astrophysics
working in planet, star and galaxy formation, and related subjects. It addresses
current issues in these fields and describes the recent observational status and
state−of−the−art theoretical and numerical methods aimed at understanding these
problems.
G
ILLESCHABRIERis Head of the Theoretical Astrophysics group at the Ecole
Normale Supérieure de Lyon. He has recently won the Johann Wempe prize and
the silver medal of the Centre National de la Recherche Scientifique (CNRS
His research interests include dense matter physics, compact objects, stellar and
planetary physics, as well as galactic implications.

STRUCTURE FORMATION
IN ASTROPHYSICS
Edited by
GILLES CHABRIER
Ecole Normale Supérieure, Lyon

CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
First published in print format
ISBN-13 978-0-521-88779-3
ISBN-13 978-0-511-51769-3
© Cambridge University Press 2009
2009
Information on this title: www.cambrid
g
e.or
g
/9780521887793
This publication is in copyright. Subject to statutory exception and to the
provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
Cambridge University Press has no responsibility for the persistence or accuracy
of urls for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
eBook
(
NetLibrar
y)
hardback

Contents
List of contributors pagevii
Preface ix
Part I Physical Processes and Numerical Methods Common
to Structure Formations in Astrophysics 1
1 The physics of turbulence 3
E. Lévêque
2 The numerical simulation of turbulence 20
W. Schmidt
3 Numerical methods for radiation magnetohydrodynamics in
astrophysics 37
R. Klein and J. Stone
4 The role of jets in the formation of planets, stars and galaxies84
R. E. Pudritz, R. Banerjee and R. Ouyed
5 Advanced numerical methods in astrophysical fluid dynamics110
A. Hujeirat and F. Heitsch
Part II Structure and Star Formation in the Primordial
Universe 131
6 New frontiers in cosmology and galaxy formation:
challenges for the future 133
R. Ellis and J. Silk
7 Galaxy formation physics 159
T. Abel, G. Bryan and R. Teyssier
8 First stars: formation, evolution and feedback effects 180
V. Bromm, A. Ferrara and A. Heger
v

vi Contents
Part III Contemporary Star and Brown Dwarf Formation 203
9 Diffuse interstellar medium and the formation of molecular clouds205
P. Hennebelle, M.-M. Mac Low and E. Vazquez-Semadeni
10 The formation of distributed and clustered stars in molecular clouds228
S. T. Megeath, Z.-Y. Li and Å. Nordlund
11 The formation and evolution of prestellar cores 254
P. André, S. Basu and S. Inutsuka
12 Models for the formation of massive stars 288
M. R. Krumholz and I. A. Bonnell
Part IV Protoplanetary Disks and Planet Formation 321
13 Observational properties of disks and young stellar objects323
G. Duchene, F. Menard, J. Muzerolle and S. Mohanty
14 Structure and dynamics of protoplanetary disks 350
C. P. Dullemond, R. H. Durisen and J. C. B. Papaloizou
15 Planet formation and evolution: theory and observations 378
Y. Alibert, I. Baraffe, W. Benz, G. Laughlin and S. Udry
16 Planet formation: assembling the puzzle 410
G. Wurm and T. Guillot
Part V Summary 425
17 Open issues in small− and large−scale structure formation427
R. S. Klessen and M.-M. Mac Low
18 A final word 441
E. E. Salpeter

Contributors
Tom Abel, KIPAC, Stanford University, USA
Yann Alibert, Physikalisches Institut University of Benn, Switzerland
Philippe André, Service d’Astrophysique, CEA Saclay, France
Robi Banerjee, ITA, University of Heidelberg, Germany
Isabelle Baraffe, Ecole Normale Supérieure de Lyon, CRAL, France
Shantanu Basu, Department of Physics and Astronomy, University of Western
Ontario, Canada
Willy Benz, Physikalisches Institut University of Bern, Switzerland
Ian Bonnell, SUPA, School of Physics and Astronomy, University of St Andrews,
United Kingdom
Volker Bromm, Department of Astronomy, University of Texas at Austin, USA
Greg Bryan, Department of Astronomy, Columbia University, USA
Gaspard Duchêne, Laboratoire d’Astrophysique de Grenoble, France
Cornelis Dullemond, Max Planck Institut für Astronomie, Heidelberg, Germany
Richard Durisen, Astronomy Department, Indiana University, USA
Richard Ellis, Department of Astrophysics, Oxford, UK
Andrea Ferrara, SISSA/International School for Advanced Studies, Trieste, Italy
Tristan Guillot, OCA, France
Alexander Heger, Los Alamos National Laboratory, USA; Department of
Astronomy and Astrophysics, UCSC, USA; School of Physics, University of
Minnesota
Fabian Heitsch, Department of Astronomy, University of Michigan, Ann Arbor,
USA
Patrick Hennebelle, Ecole Normale Supérieure, Paris, France
Ahmad Hujeirat, ZAH – Center for Astronomy, University of Heidelberg,
Germany
Shu−Ichiro Inutsuka, Department of Physics, Graduate School of Science, Kyoto
University, Japan
vii

viii Contributors
Richard Klein, Department of Astronomy, University of California, Berkeley and
Lawrence Livermore National Laboratory, USA
Ralf Klessen, ZAH/ITA, Universität Heidelberg, Universität, Germany
Mark Krumholz, Department of Astrophysical Sciences, Princeton University,
USA
Gregory Laughlin, University of California, Santa−Cruz, USA
Emmanuel Lévêque, Laboratoire de physique, Ecole normale supérieure de Lyon,
France
Zhi−Yun Li, University of Virginia, USA
Mordecai−Mark Mac Low, American Museum of Natural History, New York,
USA
Thomas Megeath, Department of Physics, University of Toledo, USA
Francois Menard, Laboratoire d’Astrophysique de Grenoble, France
Subhanjoy Mohanty, Harvard University, USA
James Muzerolle, Steward Observatory, University of Arizona, USA
Aake Nordlund, Niels Bohr Institute, Denmark
Rachid Ouyed, Department of Physics and Astronomy, University of Calgary,
Canada
John Papaloizou, DAMT, Centre for Mathematical Sciences, University of
Cambridge, United Kingdom
Ralph Pudritz, Physics and Astronomy Department, McMaster University,
Hamilton, Canada
Edwin Salpeter, Space Sciences, Cornell University, USA
Wolfram Schmidt, Institut für Theoretische Physik und Astrophysik, Universität
Würzburg, Germany
Joseph Silk, Department of Physics, University of Oxford, United Kingdom
James Stone, Department of Astrophysical Sciences, Princeton University, USA
Romain Teyssier, Service d’Astrophysique CEA Saclay, France
Stephane Udry, Observatoire de Genéve, Switzerland
Enrique Vazquez−Semadeni, Universidad Nacionál Autónoma de México, Mexico
Gerhard Wurm, Institut für Planetologie, Münster, Germany

Preface
Understanding the formation of gravitationally bound structures at all scales in the
universe is one of the most fascinating challenges of modern astronomy. It is now
realized that the initial building blocks of galaxies were small collapsing dark mat−
ter halos, produced by the primordial fluctuations. These blocks then merged and
were assembled into progressively larger galaxies, a scheme generally described as
the hierarchical model of galaxy formation. The modern understanding of star for−
mation involves large−scale turbulent motions producing local overdensities which
eventually collapse and form prestellar cores under the action of gravity. The most
likely scenario for planet formation is the collapse of a vast gaseous envelope
onto a central dense core formed from the aggregation of millimetre−size grains
in the original protoplanetary nebula, although disk fragmentation could remain an
alternative scenario in some situations. The detailed processes responsible for the
formation of these structures, however, remain poorly understood. Many impor−
tant issues remain unsettled, so the robustness of these general paradigms is still
ill determined. All these scenarios for the formation of galaxies, stars and planets,
although involving vastly different scales, share many underlying physical mecha−
nisms. They all involved hydrodynamical processes, generally leading to turbulent
motions, but the very nature of these motions and their real role in structure for−
mation remains unclear. The role of magnetic fields, in the collapse itself and in
the generation of winds and jets, remains one of the major unknowns in the for−
mation of structures. As a structure starts to collapse, its ability to cool determines
the final bound objects and thus radiation, or more precisely radiation hydrody−
namics is a key process in galaxy, star and planet formation. How the ubiquitous
presence of gravity modifies the impact of all these processes remains a major
issue. Only a comprehensive description of these complex physical mechanisms,
and of their interplay, will enable us to fully assess the validity of the afore−
mentioned scenarios for the formation of gravitationally bound structures in the
universe.
ix

x Preface
It was indeed the aim of this conference, organized at the foot of Mont Blanc,
in Chamonix, to bring together world experts in galaxy, star and planet formation,
in order to address these issues and to share their expertise and problems over the
course of a week. Since ‘the star is always right’, all the sessions started with an
observational review, presenting our current understanding of the problem and the
main questions to be answered. The theoretical talks addressed in detail the var−
ious physical problems encountered in the astrophysical situations of interest, as
well as the numerical methods and challenges in the description of these complex
processes. The talks were followed by lively discussions between the participants
during the daily poster sessions, the lunches taken together on the site and the
long morning and afternoon breaks, promoting interactions between the various
communities.
I am deeply indebted to the different speakers for their outstanding reviews and
for the remarkable review chapters they have written, which are included in the
present volume. Bringing together various authors to write a common, genuine
review was a challenge and I am really thankful to all the contributors for having
accepted this task. I believe this has generated a volume which will be useful to
young and less young researchers interested in or working in the exciting field
of structure formation in astrophysics. My profound gratitude to all the mem−
bers of the Scientific Organizing Committee (SOC
1
whose input was essential
in choosing the various topics and speakers, and in shaping up the scientific back−
bone of the conference. I am also deeply indebted to Cathy Meot and to the whole
team at the Hotel Majestic for their kindness, their efficiency and their remark−
able professionalism. Their role was essential in making this conference a success.
Special thanks also to Jimmy Paillet, our talented artistic graduate student, who
designed the logo of the conference, and to the city and the mayor of Chamonix,
for hosting the conference in this unique site. Finally, it is important to mention
that this conference could not have been organized without the essential financial
support of various sponsors, namely the Ecole Normale Supérieure de Lyon, the
Centre National de la Recherche Scientifique, the Ministère de la Recherche et
de la Technologie, the Institut National des Sciences et de l’Univers, the Centre
de Recherche Astrophysique de Lyon, the Région Rhône−Alpes and the Conseil
Général de Haute−Savoie. My last thoughts will be for the IAU and the ‘experts’
of the various Divisions/Commissions, that the proposal had been submitted to, for
decidingnotto support the conference, for the general reason that it was unlikely
that people coming from different scientific communities would be persuaded to
listen to each other. This certainly strengthened my wish to take up the challenge!
1
The SOC of the conference was composed of: Ph. André (Paris), I. Baraffe (Lyon), P. Bodenheimer (UCSB
Bromm (Austin
Mayor (Geneva), C. McKee (Berkeley), Ph. Myers (Harvard), J. Papaloizou (Cambridge
F. Shu (UCSD

Part I
Physical Processes and Numerical Methods Common
to Structure Formations in Astrophysics

1
The physics of turbulence
E. Lévêque
Turbulence in fluids is a topic of great interest. First and foremost, most flows
in nature are turbulent and this is particularly true in the astrophysical context
(Kritsuk & Norman 2004). Also, turbulence leads to very peculiar mechanics
that still escapes to a great extent from our understanding. Since the pioneer−
ing works conducted by Osborne Reynolds at the end of the nineteenth century
(around 1895), turbulence in fluids has become a rich and challenging research
subject in which scientists from engineering, theoretical and experimental physics
have been involved with many different perspectives. There is no doubt that
bridging ideas from one field to another, and therefore stimulating new interdis−
ciplinary approaches, should provide a fruitful means of gaining understanding on
turbulence in the future.
In this chapter, thebackgroundphysics of turbulence will be discussed spon−
taneously at a (very) basic level, i.e. without getting into details or precise
formulation. The discussion will be limited to incompressible hydrodynamics
governed by the Navier–Stokes (NS
bulence (as a statistical−mechanical problem) will be made. Then, I shall attempt
to provide some hints (rather than definite answers) to a series of questions: What
is generally the source of turbulence? What are the main statistical features of tur−
bulence? How to deal with turbulence? Much more elaborated developments and
references may be sought in the following books (among many others) dealing with
turbulence:
•a reference book on the physics of turbulence:A first course in turbulenceby
H. Tennekes and J. L. Lumley, MIT Press, Cambridge, USA (1972)
•a reference book on turbulence as a statistical−mechanical problem:Turbulence: The
Legacy of A. N. Kolmogorovby U. Frisch, Cambridge University Press, Cambridge, UK
(1995)
•a reference textbook on the modelling of turbulence:Turbulent flowsby S. Pope,
Cambridge University Press, Cambridge, UK (2000
Structure Formation in Astrophysics. ed. G. Chabrier. Published by Cambridge University Press.
cCambridge University Press 2009.

4 The physics of turbulence
•a reference book on the numerical modelling of turbulence:Large-eddy simulation
for incompressible flows – An introductionby P. Sagaut, Springer−Verlag, Scientific
Computation series (2005
1.1 General comments on turbulence
Turbulenceis employed to label flows with the common characteristics of com−
plexity and disorder.Complexityrefers to the complicated swirling motion of the
fluid, the ability to distort material fluid elements into complex convoluted geome−
tries.Disorderis related to this dynamics being random (or unpredictable). In
this respect, turbulence should be approached from the viewpoint of statistical
mechanics (Monin & Yaglom 1975).
Complexityin turbulence has to do with the essential non−linearity arising from
the advection term in the dynamical equations:

tu(x,t)+(u(x,t)·∇)u(x,t)=forces by unit mass.
u(x,t)denotes the velocity field. By recasting this term in the Fourier space (and
assuming local homogeneity), one gets

t∂ui(k,t)=−ik j
η
p+q=k
∂ui(p,t)∂u j(q,t)+...,
where the summation is over all allowed wavevectorspandq. Thus, the time evo−
lution of modekis a priori driven by the triadic interactions with all modes such
thatp+q=k. A specific feature of turbulence (as a dynamical system) lies in the
impossibility to reduce this interaction to a restricted set of interacting modespand
q. On the contrary,strong interactionswith all triads of modes must be considered.
Furthermore, long−range (phase
play an essential role; it is heuristically connected to the concentration of vorticity
into intense thin fluid structures such as vortex filaments (Figure1.1).
Disorderin turbulence has to do with a strong departure from absolute statistical
equilibrium. From a theoretical viewpoint, the statistical problem of turbulence is
a priori well posed if at initial timet
0the mean velocityU i(x,t0)and the two−point
correlation function
R
ij(x,t0;x
ν
,t0)εζu i(x,t0)uj(x
ν
,t0)ρ
are prescribed for allxandx
ν
and if it is assumed that the (multivariate) distri−
bution of the turbulent (fluctuating u
i(x,t0)is normal. However, at
timest>t
0, it is observed that the multivariate distribution ofu i(x,t)strongly
deviates from the normal distribution due to statistical correlations generated by

1.2 What is the source of turbulence? 5
Fig. 1.1. Vortex filaments in a turbulent jet visualized by micro−bubbles. Courtesy
of Olivier Cadot, École nationale supérieure de techniques avancées, Paris.
the hydrodynamical forces. Analytically, this departure from the normal distribu−
tion is governed by the entire infinite sequence of statistical equations (deduced
from the NS equations) for all multi−point multi−time correlation functions
R
ijk...(x,t;x
ν
,t
ν
;x
νν
,t
νν
;...)εζu i(x,t)u j(x
ν
,t
ν
)uk(x
νν
,t
νν
)...ρ.
An identified fundamental issue resides in finding an appropriate closure condi−
tion to convert this infinite hierarchy into a closed (low−dimensional) subset of
equations and eventually solve it.
In practice, turbulence is often investigated from a phenomenological standpoint,
i.e. starting from hypotheses motivated by experimental and numerical observa−
tions. This line of study has yielded very fruitful results during the past half century
and continues to expand nowadays (in absence of any successful statistical theory).
1.2 What is the source of turbulence?
Turbulence is generally triggered by the inhomogeneities of the flow (by the
mean velocity field not being uniform in space). Boundary conditions are often
responsible for such inhomogeneities (Figure1.2).
In order to dissect this mechanism, let us consider the flow over a (solid
plate. Because of the no−slip condition at the boundary, the velocity field neces−
sarily decreases to zero in the vicinity of the plate. This implies a strong gradient
of the mean velocity in the direction perpendicular to the plate. From kinematic
considerations, this strong gradient may be viewed as a vorticity sheet attached to
the plate. This sheet is generically unstable and generates streaky structures that

6 The physics of turbulence
Fig. 1.2. Streamlines of a turbulent flow (from left to right) around an obstacle. It
appears that turbulence originates in the vicinity of the obstacle.
Streaky stru ctures
y
U: bulk velocity
x
u|
wall=0
Turbulence
Vorticity sheet
Fig. 1.3. A solid boundary may be viewed as a source of vorticity which is generically unstable, detaches from the boundary and sustains turbulence in the bulk.
eventually detach, interact and contribute to sustain turbulence in the bulk of the
flow (Figure1.3).
As rule of thumb, one may claim that a key ingredient involved in the gen−
eration of turbulence is the inhomogeneity of the mean flow, i.e. strong mean
velocity gradients, and the instability of these gradients. This mechanism also tells
us that it is important to learn about the mean flow before getting to the turbulent
fluctuations.

1.3 What are the main statistical features of turbulence?7
1.3 What are the main statistical features of turbulence?
This section is devoted to the statistical features of turbulent motions, usually
viewed as the net result of the interaction of agas of turbulent eddies. By analogy
with a molecule, an eddy may be seen as a glob of fluid of a given size, or (spa−
tial) scale, that has a certain structure and life history of its own. This interaction
is unpredictable in detail; however, statistically distinct properties can be identi−
fied and profitably examined. The need for a statistical description arises from
both theintrinsiccomplexity of individual solutions and the instability of these
solutions to infinitesimal perturbations (in the initial and boundary conditions).
This makes it natural to examine ensemble of realizations rather than each individ−
ual realization and seek for robust statistical features insensitive to the details of
perturbations.
It is characteristic of turbulence that turbulent eddies are distributed over a wide
range of size scales and associated turn−over timescales. This range spreads from
theintegral scale L
S, which refers to the size of the largest eddies in the flow
(triggered by the inhomogeneities of the mean flow as seen previously), to theele-
mentary scaleη, which nails down the size of the smallest eddies. The macroscale
L
Scan be estimated by equalling the timescale related to the local mean veloc−
ity gradient, 1/δη Uδ, and the turn−over time of the largest eddies,L
S/u, whereu
denotes the root−mean−squared turbulent velocity. Let us note that the norm of the
mean velocity gradient is often referred to as theshearin the literature, andL
Sis
called theshear length scale. The elementary microscaleηis the viscous cut−off
scale. Formally, it is estimated by equalling the viscous timescaleη
2
/ν, whereνis
the kinematic viscosity of the fluid, and the turn−over time of the smallest eddies.
This latter must be modelled (see p. 9).
The range of excited scales
L
S∼
u
δηUδ
≥r≥η
is called theinertial range. In that range, interactions between turbulent eddies
result in an effective transfer of kinetic energy from the large scales (comparable
toL
S), where energy is fed into turbulence, to the small scales (comparable toη),
where this energy is dissipated by molecular viscosity. In 1941, Kolmogorov envis−
aged to describe this mechanism by a self−similar cascade of kinetic energy, which
is local in scale and in which all statistical information concerning the large scales
is lost (except for the mean energy−cascade rateε). Kolmogorov’s theory yields the
celebrateduniversal lawfor the kinetic energy spectrum (in wavenumberk):
E(k)=Cε
2/3
k
−5/3
,

8 The physics of turbulence
whereCis a universal constant (Kolmogorov 1941). It is worth noting that this
law does not include any characteristic scale; this refers to the idea that the energy
cascade is a self−similar process in scale.
The Kolmogorov’s energy spectrum represents a form for the inertial range of
wavenumbers, in which energy is transferred from small to large wavenumbers by
the process of vortex stretching. Indeed, the chaotic nature of turbulence tends to
separate any two fluid elements initially near to each other. Consequently, there is
a tendency to stretch initial vorticity into ever−elongated and thinning structures,
until viscosity stops the thinning (Figure1.4). By Kelvin’s theorem, if the cross
section of a vortex structure decreases under stretching, the fluid in the vortex
must spin faster. The combination of stretching and spin−up means a transfer of
energy from lower to higher wavenumbers. The viscous cut−off scaleη, which
identifies the bottom scale of the energy cascade, is given (from dimensional
arguments) by
R
λ
 ~ 120
Fig. 1.4. Snapshot of high−enstrophy isosurfaces from a numerical simulation of
three−dimensional turbulence; the local enstrophy is defined by|∇×u(x,t)|
2
.
The swirling activity of the flow concentrates into very localized elongated and
thin fluid structures: thevortex filaments(E. Lévêque).

1.3 What are the main statistical features of turbulence?9
η=
ν
ν
3
ε

1/4
.
A large body of experimental and numerical measurements corroborate the
Kolmogorov’s energy spectrum (Figure1.5). However, higher−order statistical
correlations are not universal in the sense of Kolmogorov’s hypothesis. These dis−
crepancies are rooted in the fact that the cascade of energy is actually a highly
non−uniform process in space and time (Figure1.6). This feature is usually referred
to asintermittencyin the literature. From the viewpoint of statistical mechanics,
intermittency implies that the macroscopic parameterεis not sufficient to char−
acterize the energy−cascade state of turbulence, but fluctuations ofε(x,t)should
be taken into account. Once Kolmogorov’smean field theoryis abandoned, a Pan−
dora’s box of possibilities is opened and a specific contact with the dynamics of
turbulence (solution of the NS equations) must be achieved. Current models have
not succeeded to establish this contact. They essentially rely on plausible hypothe−
ses but fail to relate themselves to the actual dynamics. More recent works attempt
to correlate turbulent high−order velocity correlations with the presence of highly
10
0
10
1
10
2
10
3
10
4
10
5
10
0
10
–2
10
–4
10
–6
10
–8
10
–10
k (m
–1
)
E(k) (m
3
 s
−2

Kolmogorov ís theory
 
Fig. 1.5. The density (in wavenumberk) of turbulent kinetic energyE(k)exhibits
a universalk
−5/3
decrease (obtained from a laboratory experiment, courtesy
of C. Baudet and S. Ciliberto, E
NS−Lyon, France) in agreement with the Kol−
mogorov’s theory. Turbulent motions are strongly damped by molecular viscosity
at very large wavenumbers. The energy is not equally distributed among Fourier
modes:E(k)∼k
2
would be expected at statistical thermodynamic equilibrium.
Turbulence is a far−from−equilibrium system.

10 The physics of turbulence
120
100
80
60
40
20
0
0
20 40 60 80 100
120
1
0.9
0.8
0.7
0.5
0.4
0.3
0.2
0.1
0.6
Fig. 1.6. Slice of a snapshot of the energy−dissipation rateε(x,t)obtained from
a numerical simulation of the NS equations (E. Lévêque). The magnitude of
ε(x,t)is represented by a grayscale bar ranging from 0 to 1. Energy dissipation
is concentrated on fine structures; it is not uniformly distributed.
coherent dynamical structures, the so−calledShe–Lévêque model(She & Lévêque
1994), for instance.
High−order multi−point correlations of turbulent velocity fluctuations are com−
monly investigated through the velocity structure functions, defined by
S
p(r)≡
ζ
|u(x,t)−u(x+r,t)|
p
ρ
forp=1,2,... and the separation scalerwithin the inertial range. It is observed
both experimentally and numerically that theS
p(r)’s exhibit power−law scalings:
S
p(r)∼r
ζp
and the scaling exponentsζ pare found to be universal. The She–Lévêque model
relates the set of scaling exponentsζ
pto the presence of vortex filaments and yields
a formula without any adjustable parameter:
ζ
p=
p
9
+2
δ
1−
ν
2
3

p/3

.
This model, which is found in very good agreement with experimental and numer−
ical data, establishes a concrete link between the dynamics and the statistics of

1.4 How to deal with turbulence? 11
turbulence. Furthermore, its formulation is very general and can apply to a vast
class of turbulent systems. In astrophysics, it has been successfully employed to
relate the statistics to coherent dynamical structures in the interstellar medium, in
the solar wind or in cosmic rays.
1.4 How to deal with turbulence?
1.4.1 The Reynolds-averaged Navier–Stokes equations
How to account for turbulence in the mean flow? This fundamental question was
raised by Osborne Reynolds more than a century ago (Reynolds 1895) and remains
mostly open today.
In the turbulent regime, the mean flow is solution of the NS equations comple−
mented by a force which encompasses the exchange of momentum between the
mean flow and the turbulent agitation:
ρ
dU
i
dt
=NS(U) −ρ
∂uiuj
∂xj
.
−ρuiujis termed theReynolds stress(the overbar represents the statistical mean
value). These equations are commonly called theReynolds-averaged Navier–
Stokes(RANS
the mean velocity field in order to close the RANS equations.
1.4.1.1 The concept of turbulent viscosity
The idea behind the introduction of a turbulent viscosity is to treat the deviatoric (traceless
−ρ
uiuj+
1
3
ρukukδij=2ρν turbSij,
whereν
turbis the (kinematic
Sijis the mean rate−of−strain
tensor (the symmetric part of the velocity gradient tensor). The turbulent viscosity ν
turb(x,t)depends on the positionxin the flow and timet; it is a property of the flow
not of the fluid. The introduction of a turbulent viscosity relies on the hypothesis of an internal friction (related to the turbulent agitation of the fluid) responsible for
the diffusive transport of momentum from the rapid to the slow mean−flow regions.
Dimensionally,ν
turbis equivalent to the product of a velocity and a length. This
suggests that (by analogy to the kinetic theory for gases)
ν
turb=u ,
whereuand would represent the characteristic velocity and length of the turbu−
lent agitation. In themixing-lengthmodel,uand are specified on the basis of the

12 The physics of turbulence
geometry of the flow. In two−equation models (the so−called(k−)model being
the prime example),uand are related to the turbulent kinetic energy and the tur−
bulent dissipation, for which modelled constitutive equations are explicated. See
Wilcox (1993
1.4.1.2 The mixing-length model (zero-equation model)
In 1925, the so−calledmixing-lengthmodel has been proposed by Prandtl for the
two−dimensional flow over an horizontal wall. The turbulent viscosity reads
ν
turb=
2
m




∂U
x
∂y




,
whereyis the coordinate perpendicular to the wall (along thexaxis).
The turbulent length is the mixing length and the velocity scale is locally
determined by the mean velocity gradient:
=
mandu= m




∂U
x
∂y




.
Here,uis zero wherever the mean velocity gradient is zero. This may be prob−
lematic for bounded flows in some particular circumstances (on the centreline of a
channel flow, for instance).
In practice, the mixing length
m(x,t)must be specified accordingly to the
geometry of the flow. This specification is usually empirical; it is derived from
experiment and observation rather than theory. The mixing−length model, which
is arguably the simplest modelling of turbulence, appears to be valuable in sim−
ple flows such as shear−layer or pressure−driven flows, but cannot account for the
transport effects of turbulence. Indeed, it implies that the local level of turbulence
depends on the local generation and dissipation rates. In reality, turbulence may
be carried or diffused to locations where no turbulence is actually being generated
at all.
1.4.1.3 One-equation model: the k-model
In 1942, independently, Kolmogorov and Prandtl suggested to base the turbulent
velocity scale on the mean turbulent kinetic energy:u∝
k
1/2
. If the turbulent
length scale is taken as the mixing length, the turbulent viscosity then becomes
ν
turb= m
k
1/2
.
In that case, a transport equation forkis required.

1.4 How to deal with turbulence? 13
The (exact) dynamical equation fork(deduced from the NS equations) is as
follows:
Dk
Dt
=−

∂xj





1
2
uiuiuj+
1
ρ
ujp
τ γ
turbulent transport

∂k
∂xj





−uiujSij
τγ
production:P
−ν
ν
∂u
i
∂xj
ε
2
τγ
dissipation:ε
•−uiuj≈2νturbSij−2/3kδijwithν turb= mk
1/2
according to the turbulent−viscosity
hypothesis.
•1/2(uiuiuj)+1/ρ(ujp)≈(ν turb)/σk(∂k)/∂x j. This stems from the gradient−diffusion
hypothesis. The empirical coefficientσ
kmay be viewed as a turbulent Prandtl number
(of order unity) for the kinetic energy.
•From a dimensional argument, the mean dissipation rate (per unit mass)ε=C
k
3/2
/ m,
whereCis a model constant.
Finally, the closed transport equation for the mean turbulent kinetic energy is as
follows:
Dk
Dt
=−

∂xj

ν
ν+
ν
turb
σk
ε
∂k
∂xj

+P−ε
withν
turb= m
k
1/2
andε=Ck
3/2
/ mtogether with the turbulent−viscosity
hypothesis for−uiujand the specification of the mixing length m.
Thek−model does allow for the transport of turbulence into regions where
there is locally no generation. It is therefore inherently capable of simulating
some phenomena more realistically than the mixing−length model. However, the
mixing length remains an empirical parameter and knowledge is almost totally
absent for recirculating and three−dimensional flows. This model is found useful in
boundary−layer flows, where the mixing length is fairly well known.
1.4.1.4 Two-equation model: the (k–ε) model
The (k–ε) model belongs to the class of two−equation models, which are more
efficient in the case of flows in complex geometry and higher Reynolds numbers
(Jones & Launder 1972).
The two relevant macroscopic parameters in the (k –ε) model are the turbulent
mean kinetic energy and mean energy dissipation. From dimensional analysis, one
gets that the turbulent length scale ∼k
3/2
/εand the turbulent velocity scale
u∼k
1/2
. It follows that
ν
turb=Cμ
k
2
ε
,
whereC
μis an empirical constant.

14 The physics of turbulence
The standard model equation forεis viewed as

Dt
=

∂xj
νν
ν+
ν
turb
cσε

∂ε
∂xj

−2c
1
νturbεk
Sij
∂ui
∂xj
−c2
ε
2
k
.
The five empirical coefficients (C
μ,c1,c2,σk,σε) are fixed by considering
experimental results for simple−geometry flows.
1.4.1.5 A general comment
In principle, the concept of viscosity is properly applicable only when there is a
significant separation between the length scale of inhomogeneity of the mean field
and the mixing length of the agitation. This condition is not satisfied in turbulent
flows where turbulent motions display a continuous distribution of scale sizes (tur−
bulent motions do not occur at some characteristic mixing length). Consequently,
the concept of a local transport has not much justification here, except in a very
crude qualitative sense. A more realistic approach, mainly devoted to the numerical
simulation of turbulent flows, consists of the large−scale modelling of turbulence
dynamics (see Section 1.4.2).
1.4.2 The Large-eddy Simulation of turbulent flows
1.4.2.1 The general context
The prohibitive cost of thedirect numerical simulation(DNS
(that is, without calling to any simplification of the dynamics) motivates the elab−
oration of reduced models, requiring less computation but still relevant (to some
degree) for reproducing the large−scale dynamics of the flow (Lesieur 1997).
Indeed, on the one hand, the elementary scaleηfixes the required grid spacing
of the simulation. On the other hand, the integration domain must encompass the
largest scales of motion, comparable to the macroscaleL
S. Therefore, the number
of grid points required to suitably resolve all spatial scales of motion is given by
ν
L
S
η

3
∼Re
9/4
,
where
Re=
uL

denotes theReynolds number. Here, it has been assumed that ε∼u
3
/LS.
Furthermore, when the Courant–Friedrich–Levy time−step constraint is factored in,
uδtη
<1,

1.4 How to deal with turbulence? 15
one ends up with a computational cost for the DNS which grows as Re
3
.Inreal
flows, the encountered Reynolds numbers are typically very large (Re∼10
10
in
astrophysics), which definitively rules out this direct numerical approach. Reduced
models of turbulence must be considered.
Roughly speaking, large−scale motions transport most of the kinetic energy
of the flow. Their strength makes them the most efficient carriers of conserved
quantities (momentum, heat, mass, etc.). On the contrary, small−scale motions are
primarily responsible for the dissipation while they are weaker and contribute little
to transport. From mechanical aspects, the large−scale (energy−carrying) dynam−
ics are thus of particular importance, and the costly computation of small−scale
dynamics should be avoided. Furthermore, while large−scale motions are strongly
dependent on the external flow conditions, small−scale motions are expected to
behave more universally. Hence, there is a hope that numerical modelling can be
feasible and/or requires few adjustments when applied to various flows.
The RANS approaches (introduced in Section 1.4.1) lead to the estimation of
mean quantities but do not resolve the turbulent fluctuations; they are based on
questionable closure conditions and often appeal to numerous empirical parame−
ters. But their computational cost is unbeatable. Thelarge-eddy simulation(LES)
offers a compromise between the DNS and the RANS approaches. In a LES, the
large−scale dynamics of turbulence are explicitly integrated in time and the interac−
tion with the unsolved small−scale motions is modelled. In order to separate the
small−scale component and the large−scale component of the turbulent velocity
field, an explicit filtering procedure is widely used.
A spatial filtering is conceptually introduced as
κφ(x,t)=
χ
φ(x
ν
,t)G (x−x
ν
)dx
ν
,
where the filter widthfixes the size of the smallest scales of variation retained
in the flow variableφ(x,t). In practical applications,is chosen much larger than
the cut−off spatial scale ofφ(x,t)so thatκφ(x,t)may be properly considered as the
large−scale component ofφ(x,t). Applying the previous filtering procedure to the
NS equations yields
∂κu
i
∂t
+κu
j
∂κui
∂xj
+
∂τ
ij
∂xj
=−
∂κp
∂xi


2
κui
∂xk∂xk
with
∂κu
i
∂xi
=0,
whereκu
i(x,t)andκp(x,t)represent the large−scale velocity and pressure, respec−
tively, andνis the kinematic viscosity of the fluid.
These equations are amenable to numerical discretization with a grid spacing
comparable to,sinceκu(x,t)is expected to vary smoothly over the distance.
τ
ij(x,t)≡ ∂u i(x,t)u j(x,t)−κu i(x,t)κu j(x,t)

16 The physics of turbulence
is named thesubgrid-scale(SGS)stress tensorand encompasses all interactions
between the grid−scale component and the unresolved SGS component ofu(x,t).
In LES,τ
ij(x,t)needs to be expressed in terms of the grid−scale velocity field
˘u(x,t)only, which is a difficult problem.
Eddy−viscosity models parameterize the SGS stress tensor as
τ
ij−
1
3
τ
kkδij=−2ν T
˘S
ij,
whereν
T(x,t)is the scalar eddy−viscosity and
˘S
ij(x,t)≡
1
2
ı
∂˘u
i
∂xj
(x,t)+
∂˘u
j
∂xi
(x,t)

is the resolved rate−of−strain tensor. This empirical modelization is rooted in the
idea that SGS motions are primarily responsible for a diffusive transport of momen−
tum from the rapid to the slow grid−scale flow regions. The theoretical basis for
the introduction of an eddy−viscosity is rather insecure; however, it appears to be
workable in practice. The eddy−viscosityν
T(x,t)is then primarily designed to
ensure the correct mean drain of kinetic energy from the grid−scale flow to the SGS
motions. Another important feature is thatν
T(x,t)must vanish in laminar−flow
regions (where there is no turbulence).
1.4.2.2 The Smagorinsky model
The Smagorinsky model is certainly the simplest and most commonly used
eddy−viscosity model of turbulence (Smagorinsky 1963). The prescription for
ν
T(x,t)is
ν
T(x,t)=(C S)
2
δ˘S(x,t)δ,
whereδ˘Sδε(2˘S
ij
˘S
ij)
1/2
represents the magnitude of the resolved rate−of−strain
andC
Sis a non−dimensional coefficient (called theSmagorinsky constant). The
major merits of the Smagorinsky model are its manageability, its computational
stability and the simplicity of its formulation (involving only one adjusted param−
eter). All this makes it a very valuable tool for many applications. However, while
this model is found to give acceptable results in the LES of homogeneous and
isotropic turbulence (withC
S≈0.2), it is found to be too dissipative with respect to
the resolved motions in strongly non−homogeneous flow regions, due to an exces−
sive eddy−viscosity arising from the mean shear. Furthermore, the eddy−viscosity
predicted by Smagorinsky is non−zero in laminar−flow regions; the model intro−
duces spurious dissipation which has the effect of damping the growth of small
perturbations and thus restrain the transition to turbulence.

1.4 How to deal with turbulence? 17
1.4.2.3 The shear-improved Smagorinsky model
Recently, a shear−improved Smagorinsky model (SISM
(Lévêqueet al.2007). In this refined modelling, it is proposed to subtract the mag−
nitude of the shear from the magnitude of the instantaneous resolved rate−of−strain
in the eddy−viscosity:
ν
T(x,t)=(C S)
2
˙
δ→S(x,t)?S(x,t)

.
S(x,t)denotes the shear at the positionxand timet;C
Sis the Smagorinsky con−
stant and=(xyz)
1/3
is the local grid spacing. It is worth noting that
the eddy−viscosity involved in the SISM implies some information concerning the
mean flow: the shearS(x,t). It is assumed that the flow is well enough resolved in
the direction of the shear, so that
S(x,t)δζ→S(x,t)ρδ. (1.1)
Angle bracketsζρdenote statistical (ensemble
age over homogeneous directions and/or time average should be considered by
invoking ergodic properties of the flow).
The main concerns of the SISM are, first, to take into account shear effects
in the exchanges of momentum to the SGS motions (without any kind of adjust−
ment) and, second, to make the eddy−viscosity automatically vanish in laminar−flow
regions (without any ad hoc damping function). The SISM stems from analytical
developments derived from the NS equations.
In flow regions where the fluctuating part of the rate−of−strain is much larger
than the shear, turbulence can be considered as homogeneous and isotropic and
the SISM naturally reduces to the original Smagorinsky model (which is known to
perform reasonably well). In flow regions where shear effects are significant, the
SISM yields a transfer of energy to SGS motions fully consistent with the exact
energy budget obtained from the NS equations.
A stringent test of the SISM has been performed recently in the backward−facing
step geometry (Toschiet al.2006) (Figure1.7). In addition to wall effects, the flow
over a backward−facing step is strongly affected by the detached shear layer and
the recirculating motions behind the step. An adequate resolution of the instability
of the shear layer is required to predict correctly the location of the reattachment
point. The computation of the skin−friction coefficient is displayed in Figure1.8.
The skin−friction coefficient changes sign from negative to positive at the reattach−
ment point (indicated by the arrow). We observe that the prediction is in very good
agreement with experimental and DNS results (at comparable Reynolds number).
We also see that the SISM performs well with respect to thedynamic Smagorinsky
model(Germanoet al.1991) (a reference eddy−viscosity model) but with the great
advantage of being much less demanding in computational resources.

18 The physics of turbulence
Re−attachment point
–0.3 1.1
Fig. 1.7. Configuration of the backward−facing step flow.
–0.004
–0.002
0
 0.002
 0.004
0 5 10 15 20
SISM
DSM
DNS  data – Le et al. (1997)
Exp. data – Jovic and Driver (1994)
Fig. 1.8. For the backward−facing step flow atRe=4800 (based on the step
height), the prediction of the skin−friction coefficient, here displayed as a function
of the distance to the step, is in good agreement with experimental data, DNS data
and with a LES based on thedynamic Smagorinsky model(DSM
Reynolds number. Notice that the strength of the secondary recirculation bubble
near the corner (a sensitive diagnostic) is also correctly predicted. (Courtesy of
Federico Toschi.)

References 19
First tests indicate that the SISM possesses a good predictive capacity with
a computational cost and a manageability comparable to the original Smagorin−
sky model. The implementation of the SISM in the astrophysics code ENZO
(http://lca.ucsd.edu/software/enzo/) is currently under development and the first
results are encouraging (Wolfram Schmidt, personal communication).
References
Germano, M., Piomelli, U., Moin, P. and Cabot, W. H., 1991,Phys. Fluids A3, 1760
Jones, W. O. and Launder, B. E., 1972,Int. J. Heat Mass Tran.15, 301
Jovic, S. and Driver, D., 1994,NASA Tech. Mem. 108807
Kolmogorov, A. N., 1941,Dokl. Akad. Nauk SSSR30, 299
Kritsuk, A. G. and Norman, M. L., 2004,ApJ.601,L55
Le, H., Moin, P. and Kim, J., 1997,J. Fluid Mech.330, 349–374
Lesieur, M., 1997,Turbulence in Fluids(Kluwer Academic Publishers Dordrecht,
Boston, London)
Lévêque, E., Toschi, F., Shao, L. and Bertoglio, J.−P., 2007,J. Fluid Mech.570, 491
Monin, A. S. and Yaglom, A. M., 1975,Statistical Fluid Mechanics(MIT Press,
Cambridge)
Reynolds, O., 1895,Phil. Trans. Roy. Soc. Lond. Ser. A186, 123
She, Z.−S. and Lévêque, E., 1994,Phys. Rev. Lett.72, 334
Smagorinsky, J., 1963,Mon. Weather Rev.91,99
Toschi, F., Kobayashi, H., Piomelli, U. and Iaccarino, G., 2006,Proceedings of the
Summer Program 2006, Center for Turbulence Research, Stanford, USA
Wilcox, D. C., 1993,Turbulence Modeling for CFD(DWC Industries Inc., La Canada,
USA)

2
The numerical simulation of turbulence
W. Schmidt
Turbulence is a remarkable subject in physics. The underlying equations, which
are in their simplest formulation the Euler equations, were published 250 years
ago (Euler 1757). Yet a theoretical grasp of the phenomenology emerging from
these equations had not been achieved before the mid−twentieth century, when
Heisenberg(1923)and Kolmogorov(1941) obtained their first analytical results.
Eventually, it took the capabilities of modern supercomputers to obtain a full appre−
ciation of the complexity that is inherent to the Euler equations. Astrophysics
is now at the very frontier of numerical turbulence modelling. Among the addi−
tional ingredients for making turbulence in astrophysics even more complex are
supersonic flow, self−gravity, magnetic fields and radiation transport. In contrast,
terrestrial turbulence is mostly incompressible or only weakly compressible. Exter−
nal gravity is, of course, an issue in the computation of atmospheric processes on
Earth. Self−gravity, however, is only encountered on large, astrophysical scales. The
dynamics of turbulent plasma has met vivid attention in research related to nuclear
fusion reactors but, otherwise, is not encountered under terrestrial conditions.
In this chapter, I give an overview of the various approaches towards the
numerical modelling of turbulence, particularly, in the interstellar medium (ISM
The discussion is placed in a physical context, i.e. computational problems are
motivated from basic physical considerations. Presenting selected examples for
solutions to these problems, I introduce the basic ideas of the most commonly used
numerical methods. For detailed methodological accounts, the reader is invited to
follow the references. Some important results and astrophysical implications are
briefly outlined. Since turbulence, in a strict sense, is genuinely three−dimensional,
I almost exclusively consider three−dimensional simulations. Furthermore, turbu−
lence is a multi−scale phenomenon, and, in order to capture its properties correctly,
sufficient numerical resolution is essential. This is why newer, higher−resolution
computations are generally preferred as examples (as regards self−gravitating
Structure Formation in Astrophysics. ed. G. Chabrier. Published by Cambridge University Press.
cCambridge University Press 2009.

2.1 Fundamentals 21
turbulence, however, high−resolution simulations are appearing just now and are
not included here).
To begin with, I briefly discuss the equations which are numerically solved
in astrophysics and related theoretical aspects. The following sections deal
with supersonic turbulence, self−gravitating turbulence and magnetohydrodynamic
(MHD
numerical simulations are placed depending on their main objectives. Finally, I give
a resume of what has been achieved and which challenges are likely to be met.
2.1 Fundamentals
In the following, we consider self−gravitating inviscid fluid subject to the action
of external force fields and, possibly, magnetic fields. The evolution of the mass
densityρ, the velocityvand the specific energyeof the fluid is given by the
compressible Euler equations:
D
Dt
ρ=−ρ∇v, (2.1)
ρ
D
Dt
v=−∇P+ρ(f+g), (2.2)
ρ
D
Dt
e+∇·vP=−+ρv·(f+g), (2.3)
where the Lagrangian time derivative is defined by
D
Dt
=

∂t
+v·∇. (2.4)
The total energy per unit mass is given by
e=
1
2
v
2
+
P
(γ−1)ρ
, (2.5)
whereγis the adiabatic exponent and the pressurePis related to the mass density
and the temperatureTvia the ideal gas law:
P=
ρk
BT
μmH
. (2.6)
The constantsk
B,μandm Hdenote, respectively, the Boltzmann constant, the mean
molecular weight and the mass of the hydrogen atom. The energy budget can be altered by heating () and cooling (), by non−gravitational forces (ρ f), as well
as by gravity (ρg). Generally, heating and cooling processes are important for the
dynamics of the ISM (Sections2.3and2.4).
Non−gravitational forces supplying energy to the flow can be mechanical or mag−
netic. An example of a mechanical force would be the random driving force that

22 The numerical simulation of turbulence
is commonly used in turbulence simulations. This is an external force, i.e. it is
independent of the dynamics of the fluid (Section2.2). Quite the opposite holds
for conducting fluids in the presence of magnetic fields. In the case of ideal mag−
netohydrodynamics, the fluid is dragged by the Lorentz forceρf=J×B, where
the currentJ=∇×B. The magnetic fieldB, in turn, depends on the flow via
Faraday’s law,
∂B
∂t
=∇×(v×B). (2.7)
The interaction between turbulent flow and magnetic fields alters properties of turbulence such as the fragmentation behaviour (Section2.4).
The gravitational accelerationg=−∇·φarises, in part, from the self−gravity
of the fluid and, depending on the boundary conditions, the gravitational field of external sources. In the case of periodic boundary conditions, gravity is solely pro− duced by density fluctuations with respect to the global mean (Section2.3). In this
case, the gravitational potential is determined by the Poisson equation

2
φ=4πG(ρ?ρρ), (2.8)
whereGis Newton’s constant. The mean mass densityζρρ=ρ
0is a constant
because of mass conservation.
Letting the curl operator∇×act upon Eq. (2.2), an evolutionary equation for the
vorticity of the flow,ω=∇×v, is obtained:
D
Dt
ω=S·ω−dω+
1
ρ
2
∇ρ×∇P+∇×f. (2.9)
The rate−of−strain tensorSis the symmetrized velocity gradient and the divergence
d=∇·v. Apart from the stirring of fluid due to rotational force components,
vorticity can be generated by the baroclinic term∇ρ×∇P, if the gradient of the
mass density and the pressure gradient are not aligned. With the ideal gas Eq. (2.6),
it follows that baroclinic vorticity generation occurs in non−isothermal gas. The termS·ω−dωaccounts for the stretching of vortices caused by strain, in addition
to the expansion or contraction of vortices due to the compressibility of the fluid.
The time evolution of the divergencedis given by
D
Dt
d=
1
2
˙
ω
2
−|S|
2


1
ρ

2
P
+
1
ρ
2
∇ρ·∇P−4πG(ρ−ρ 0)+∇·f. (2.10)
This equation also follows from the conservation law (2.2) by applying the operator
∇and substituting the Poisson equation for the gravitational potential (2.8). The
norm of the rate of strain is defined by|S|=(2 S
ijSij)
1/2
. From the first term, one

2.1 Fundamentals 23
can see that vorticity contributes to positive divergence, while strain is associated
with negative divergence, i.e. converging flow. In the incompressible case, the first
term on the right−hand side of Eq. (2.10) is cancelled by the Laplacian of the pres−
sure divided by the density. Whereas vorticity cannot be generated by a gradient
field such as gravity, an increasing rate of convergence will be caused by gravity,
if the term 4π G(ρ−ρ
0)dominates. In this case, gravitational collapse of the gas
ensues.
The numerical simulation of astrophysical turbulence has to account for all
fluid dynamical processes which become manifest in the evolutionary equations
of the vorticity (2.9) and the divergence (2.10). In many terrestrial applications, on
the other hand, incompressible turbulence without gravity and magnetic fields is
considered. Then Eqs (2.9)and(2.10) reduce to
D
Dt
ω=S·ω+∇×f, (2.11)
D
Dt
d=0. (2.12)
The only direct mechanism generating vorticity in this case is to stir the fluid by solenoidal(rotational
large−scale eddies produced by stirring into thin vortex filaments. This is the very essence of turbulent fluid dynamics. Here, the question arises whether the gen− eration of turbulence in the ISM resembles the scheme of stirring, stretching and folding at all. Interestingly, this presumption underlies the application of solenoidal
driving forces in most numerical simulations.
Incompressible fluid dynamics serves as an important reference case for which
scaling properties of isotropic turbulence in the ensemble average can be derived
analytically.Kolmogorov(1941) showed that the root mean square (RMS
fluctuationsv
ν
( )at a certain length scale obey the so−called 2/3−law,
v
ν
( )
2

2/3
, (2.13)
provided that is small compared to the integral length scalesLat which energy is
supplied to the flow. The 2/3−law implies that the rate of kinetic energy dissipation
is asymptotically constant, as∼v
ν
( )
3
/ τconst. in the limit /L1. This
result is known as the law of positive energy dissipation (Frisch 1995). Both the
2/3−law and the law of positive energy dissipation appear to be robust properties of
turbulence as long as the flow becomes isotropic and nearly incompressible towards
small length scales. This is not always the case. Magnetic fields, for instance, intro−
duce small−scale anisotropy. Moreover, there is no consensus yet as to what extent
the scaling properties of incompressible turbulence carry over to supersonic flow.

24 The numerical simulation of turbulence
2.2 Supersonic turbulence
Turbulence is called supersonic if the RMS Mach number,M
RMS=ζ(v/c s)
2
ρ
1/2
,
exceeds unity. Kinetic energy dissipation in supersonic turbulence caused by shock
fronts propagating through the fluid dominates viscous dissipation. Since the result−
ing dissipative heating raises the speed of sound to a value comparable to the typical
flow velocity within a dynamical time, supersonic turbulence cannot be maintained
if the excess heat is not efficiently removed from the fluid. As an idealization, heat−
ing is exactly balanced by cooling such that the temperature remains constant. In
this case, continuously driven turbulent flow settles into a steady state in which the
RMS Mach number is asymptotically constant. In isothermal gas, solenoidal forc−
ing is used to produce turbulence because the density gradient is aligned with the
pressure gradient and, consequently,∇ρ×∇P=0. In this case, it can be seen from
Eq. (2.9) that non−zero vorticity either results from discontinuities (shock fronts)
or it is generated by the curl of the force field. As a consequence of the constant
speed of sound, the properties of driven supersonic turbulence in isothermal gas
depend on a single parameter only, namely, the RMS Mach number (provided that
the forcing is purely solenoidal (Schmidt & Federrath 2008).
A lot of effort has gone into the numerical simulation of supersonic flow
by means of finite−volume schemes of higher order, in particular, the piecewise
parabolic method (PPMColella & Woodward 1984). The applicability of PPM
to turbulence was tested in simulations of adiabatic turbulence (Sytine et al.2000)
and nearly isothermal transonic turbulence. An example for the latter is the driven
turbulence simulation byPorteret al.(2002). They utilized an explicit cooling
function and, thereby, maintained an RMS Mach number≈1.0. InSchmidtet al.
(2006), the equation of state of degenerate electron gas was applied in simulations
of forced transonic turbulence with PPM. Degeneracy entails an extremely large
heat capacity. Since the isothermal case corresponds to the limit of infinite heat
capacity, the gas in these simulations is effectively isothermal. One of the conclu−
sions was that the numerical dissipation of PPM acts as an implicit subgrid scale
model in the weakly compressible regime (at small scales).
A new approach is the application of adaptive mesh refinement (AMR
bulence simulations (Kritsuket al.2006; Schmidtet al.2008). AMR is based on
the finite−volume approach with a hierarchy of grid patches of different resolution.
The basic idea is to represent relatively smooth flow regions on coarse grids only,
whereas steep gradients of the velocity, density or any other field are computed
with higher accuracy by dynamically inserting grids of higher resolution (Berger &
Oliger 1984; Berger & Colella 1989). In astrophysics, AMR has been successfully
used for the treatment of strong shocks or gravitational collapse in a variety of prob−
lems (O’Shea et al.2004). On the contrary, the simulation of developed turbulence

2.2 Supersonic turbulence 25
with AMR has been regarded as infeasible, because, in the picture of the turbulent
cascade of eddies, small−scale features of the flow would be space−filling (Frisch
1995). For a particular flow realization, however, turbulent eddies occupy an ever
decreasing fraction of the flow domain towards smaller length scales at any given
instant of time. This follows from the intermittency of turbulence (Frisch 1995).
Picking up the intermittent picture of turbulence, Kritsuket al.proposed that
AMR would offer a computational advantage compared to static grids of fixed
resolution (Kritsuket al.2006). This can be motivated from the Landau estimate
of the number of degrees of freedomLandau and Lifshitz(1987),
N∼Re
9/4
, (2.14)
whereReis the Reynolds number of the flow. Present−day supercomputers manage
N∼10
10
, which allows for Re∼10
4
. Note that fully developed turbulence is
known to set in at Reynolds numbers of a few thousand. On the contrary, invoking
intermittency models of turbulence such as theβ−model (Frisch 1995), we have
N∼Re
3D/(D+1)
, (2.15)
whereDis interpreted as the fractal dimension of dissipative structures. Assum−
ing a value ofD≈2 as in the Burger model of supersonic turbulence, it follows
thatN∼10
8
. This figure is smaller than the Landau estimate (corresponding to a
fixed−resolution grid) by two orders of magnitude. Of course, there is substantial
computational overhead with AMR in comparison to static grids. Apart from that,
more sophisticated intermittency models and numerical studies suggest thatD>2
in supersonic turbulence (Boldyrev 2002;Boldyrevet al.2002). Nevertheless,
AMR is expected to pay off if several levels of refinement are used. In particular,
this applies to scenarios including gravitational collapse (see Section2.3).
InKritsuket al.(2007), the effective resolution (i.e. the resolution correspond−
ing to the most refined grids) of an AMR simulation of supersonic turbulence was
raised to the yet unprecedented value of 2048
3
. Nearly isothermal gas was main−
tained by setting the adiabatic exponentγto a value that differs only by a small
fraction from unity. This means that the internal energyE
int=P/(γ−1)ρis
artificially increased by a huge factor such thatE
int
1
2
ρv
2
, and, hence, it takes
many dynamic timescales to heat the gas significantly by kinetic energy dissipa−
tion. In combination with a 1024
3
static grid simulation, a great wealth of results
was obtained from the analysis of the numerical data. Figure2.1shows the pro−
jected mass density in logarithmic scaling for one snapshot of the AMR simulation.
One can see voids in between high−density regions which display intricate turbu−
lent structure. This is a tell−tale sign of the pronounced intermittency of supersonic
turbulence.

26 The numerical simulation of turbulence
Fig. 2.1. Projected mass density in an AMR simulation of supersonic turbulence
with effective resolution 2048
3
by Kritsuket al.(2007
regions contain gas of high density, while under−dense gas appears dark.
Further indications of intermittent properties come from the probability density
function of the time−averaged mass density, which follows very closely a log−
normal distribution (Padoan & Nordlund 2007), and the scaling exponents of the
velocity structure functions. The definition of structure functions is based on spatial
correlations of the velocity field:
S
p( )=ζ|v(x+∇)−v(x)|ρ
p
. (2.16)
The square brackets denote the average over the whole domain of the flow.
Structure functions probe the correlations of the velocity field at different spa−
tial positions depending on the separation . It is both a theoretical prediction
and an experimentally well−established fact that the structure functions of isotropic
turbulence obey power laws
S
p( )∝
ζp
. (2.17)
The exponentsζ
pare called the scaling exponents. For incompressible turbulence,
Kolmogorov found thatζ
p=p/3. In the case of the second−order structure func−
tions (p=2), this result corresponds to the 2/3−law for the velocity fluctuations

2.3 Self-gravitating turbulence 27
mentioned in Section2.1. Calculating the scaling exponents from their numeri−
cal data, Kritsuket al.were able to demonstrate significant deviations form the
Kolmogorov relation in accordance with predictions of the intermittency model
proposed byBoldyrevet al.(2002), which is a generalization of the She–Lévêque
model for incompressible turbulence (She & Lévêque 1994). In particular, it was
found thatζ
2≈0.76, whereasζ 2=2/3 in the Kolmogorov theory. Intriguingly,
the analysis also revealed that the scaling exponents applicable to incompressible
turbulence are obtained if the statistics is computed for the density−weighted vari−
ableρ
1/3
vin place ofv, a result that is not fully understood yet, but might bear
important implications on the nature of supersonic turbulence.
2.3 Self−gravitating turbulence
According to the linear stability analysis of the Euler equations byJeans(1902), a
perturbation in isothermal gas of temperatureTand uniform densityρ
0becomes
unstable against gravitational collapse, if its size exceeds the Jeans length
λ
J=cs
Þ
π
Gρ0
, (2.18)
wherec
s∝T
1/2
is the isothermal speed of sound. Beginning with Chandrasekhar’s
proposition to substitutec
2
s
by
c
2
eff
=c
2
s
+
1
3
ζv
2
ρ (2.19)
in the above expression for the Jeans length, several attempts have been made to extend the Jeans stability analysis to turbulent gas of RMS velocityζv
2
ρ
1/2
(Chandrasekhar 1951;Bonazzolaet al.1987, 1992). Equation (2.19) implies that
the effective gas pressure is given by the sum of the thermal pressurePand the
turbulence pressureP
turb=(2/3)E kin, whereE kinis the mean kinetic energy
density.
However, even the most advanced analysis put forward so far (Bonazzola et al.
1992) suffers from severe constraints. A perturbation analysis can be carried out for a statistically stationary equilibrium state only. Thus, it has to be assumed that the free−fall timescaleT
ff∼(Gρ 0)
−1/2
is much greater than the dynamical timescale
of turbulent flow. From the divergence Eq. (2.10), one can see that this assumption
implies that the self−gravity term remains small compared with the other terms at all times. Allowing for gravitational collapse, however, self−gravity eventually dominates and
D
Dt
d∼−Gρ. (2.20)

28 The numerical simulation of turbulence
A non−perturbative theory of the regime in which the free−fall timescale and the
dynamical timescale of turbulence are comparable was suggested byBiglari and
Diamond(1989). Combining scaling relations from theβ−model of turbulence
(Frisch 1995) with the assumption of energy equipartition between gravity and
turbulence at all scales, they derived an intermittent hierarchical cloud model of
self−gravitating turbulence.
At present, however, there is no general theory that would fully encompass the
highly non−linear and non−stationary interplay between gravity and turbulence in
the ISM. Numerical simulations, on the contrary, can aid in the understanding of
self−gravitating turbulence, although one has to resort to artificial mechanisms of
producing turbulence or imposing more or less arbitrary initial conditions. One
technique is to apply a driving force as mentioned in Section2.2. Alternatively,
some random initial velocity field might be assumed as initial condition. This
results in decaying turbulence. Yet another option is to consider gas in an unstable
state and apply small perturbations. The most prominent example is the thermal
instability (Kritsuk & Norman 2002; Elmegreen & Scalo 2004).
Numerical simulations of self−gravitating driven or decaying supersonic tur−
bulence were initially performed with smoothed particle hydrodynamics (SPH
because of its ability to tackle variations of the mass density over many orders
of magnitude (Monaghan 1992). SPH makes use of the Lagrangian framework
of fluid mechanics and represents fluid parcels by particles. Examples for the
application of SPH to self−gravitating turbulence are the simulations byKlessen
et al.(2000), and Klessen(2000, 2001). In these simulations, the influence of
self−gravity on probability density functions (Klessen 2000) and the fragmen−
tation properties of the gas were investigated both for driven and for decaying
supersonic turbulence (Klessenet al.2000; Klessen 2001). Sink particles were
introduced to capture collapsing regions beyond a certain density threshold in
order to prevent the mass density from growing indefinitely. Figure2.2illustrates
that sink particles are mostly formed within filamentary structures (which are also
seen in cosmological simulations). Varying the Mach number, it was found that
higher turbulence intensity slows down considerably the rate at which sink parti−
cles are produced (Figure2.3). The major conclusion drawn by Klessenet al.was
that supersonic turbulence inhibits gravitational collapse globally, as one would
suspect from Eq. (2.19). The strong gas compression caused by shocks, on the
contrary, can trigger gravitational collapse locally and initiate the formation of
stars.
Jappsenet al.(2005) studied the mass spectrum produced by turbulent fragmen−
tation in non−isothermal gas. They defined the state variables by piecewise poly−
tropic relations, i.e.P∝ρ
γefffor a certain range of densities. Ifγ eff<1, the gas

2.3 Self-gravitating turbulence 29
Fig. 2.2. Filamentary structure with sink particles in an SPH simulation of self−
gravitating turbulence (courtesy of R. Klessen). In this plot, over−dense regions
appear bright.
1.0
0.8
0.6
M
*
k = 1..2
k = 7..8
k = 3..4
k = 7..8
k = 39..40
k = 3..4
k = 15..16
k = 1..2
k = 3..4
1
2
3
1
2
3
5
2
4
0.4
0.2
0.0
–2 0 2 4
<M
J
>
turb
 = 0.6
<M
J
>
turb
 = 3.2
<M
J
>
turb
 = 18.2
6
t
81012
Fig. 2.3. Total mass fraction captured by sink particles as a function of time
in SPH simulations with different Mach numbers and varying driving scale of
turbulence (Klessenet al.2000).

30 The numerical simulation of turbulence
cools with increasing density. This effect enhances the compressibility of the gas
and therefore supports gravitational collapse (negative∇ρ·∇Tcontribution adds
to the gravity term in the divergence Eq. (2.10)). At some threshold density, how−
ever, the cooling regime ceases andγ
effbecomes greater than unity. From various
simulations, it was found that the transition from the cooling regime to isothermal
or nearly isothermal gas sets a characteristic mass scale of turbulent fragmentation.
This mass scale was interpreted to correspond to the peak of the observed initial
mass function.Vàzquez−Semadeniet al.(2007) included explicit heating and cool−
ing in SPH simulations of colliding gas streams. They made use of a model for the
cooling functionin the energy Eq. (2.3) that was proposed byKoyama and Inut−
suka(2002). Due to the thermal instability (compression causes the gas to cool), the
gas at the collision interface undergoes gravitational collapse and becomes increas−
ingly turbulent. This is interpreted as possible mechanism of molecular cloud
formation. Remarkably, approximate equipartition of gravitational and kinetic
energy was found, although clearly no state of virial equilibrium was approached
by the collapsing gas. Further applications of SPH, addressing problems such as
the formation of brown dwarfs, binary star systems and stellar clusters via turbu−
lent fragmentation, as well as the origin of the initial mass function, were presented
byBateet al.(2002a,b), Bate and Bonnell(2005), andBonnellet al.(2003).
Despite the numerous important contributions to the understanding of the role
of turbulence in star formation, the adequacy of treating turbulence with SPH has
been a matter of debate. On the one hand, numerical studies of self−gravitating sys−
tems indicate basic agreement between AMR and SPH (Commerconet al.2007).
However, comparisons with grid−base codes suggest that SPH dissipates small−
scale velocity fluctuations significantly stronger (Padoanet al.2007; Kitsionas
et al., 2008). In particular, it was demonstrated that the steeper power spectra
obtained from SPH simulations are accompanied by mass distributions which are
not consistent with observations. Apart from that, it appears to be rather difficult to
accommodate magnetohydrodynamics in the SPH formalism. Although there are
ongoing attempts to overcome this shortcoming (Maron & Howes 2003 ), MHD has
been treated with great success using finite−volume discretization.
2.4 Magnetohydrodynamic turbulence
In purely hydrodynamical (HD
tion for small−scale velocity fluctuation. The randomization of the flow due to
non−linear turbulent interactions produces statistical isotropy towards small scales
which can be interpreted as a symmetry of the ensemble average (Frisch 1995).
This symmetry is broken by a magnetic fieldBin turbulent conducting fluid if the
ratio of the thermal to the magnetic pressure,

2.4 Magnetohydrodynamic turbulence 31
β=
P
PM
=
8πP
B
2
, (2.21)
is comparable to unity or less. If the curl of the Lorentz force,∇×(J×B),inthe
vorticity Eq. (2.9) is sufficiently strong, then the vorticity will mainly grow in the
direction ofB. In developed MHD turbulence, this effect causes eddies to shrink
perpendicular to the field lines. As a further implication, the anisotropy of MHD
turbulence increases towards smaller length scales (Biskamp 2003).
The numerical simulation of MHD turbulence is rather challenging. Godunov−
based schemes of higher order such as PPM become very complex upon including
MHD. For this reason, simpler schemes have been adopted for MHD. Examples are
the Zeus code (Stone & Norman 1992), the Stagger code (Nordlund & Galsgaard
1995) and the Ramses code which also features AMR (Teyssier 2002;Fromang
et al.2006). As an illustration, Figure2.4shows a simulation of the gravitational
collapse of a magnetized cloud with up to six levels of refinement. One of the major
problems that needs to be addressed when solving the MHD equations numeri−
cally is to keep the magnetic field divergence−free, i.e.∇·B=0. In Ramses
t = 0.0158
9
8
7
6
log of density (cm
–2
)
5
4
3
–0.2
–0.2
0.0
0.0
x (pc)
y (pc)
0.2
0.2
Fig. 2.4. Simulation of a collapsing gas cloud of 1000 solar masses with Ramses
(courtesy of P. Hennebelle). The magnetic energy of the clouds is about the same
as its kinetic energy.

32 The numerical simulation of turbulence
and other codes, the constrained transport algorithm is employed to ensure this
constraint (Evans & Hawley 1988).
A comparison of the properties of driven HD and MHD turbulence in high−
resolution simulations was presented byPadoanet al.(2007). Whereas the
energy spectrum functions, irrespective of magnetic fields, were found to follow
closely the power lawk
−2
in the inertial subrange, the fragmentation properties
of the gas appear to be markedly different for HD and MHD turbulence. In plots
of the projected density fields for both cases (Figures2.5and2.6), one can see that
the gas is more concentrated in pronounced filaments under the action of magnetic
fields. Utilizing a clump−find algorithm, it was demonstrated that a significantly
steeper mass spectrum of dense cores (defined by a density threshold based on
the Bonnor–Ebert mass) resulted from purely HD turbulence. Furthermore, they
were able to reproduce the Chabrier initial mass function (Chabrier 2003 )inthe
MHD case, although this result must be considered with care given the ambiguity
of calculating mass spectra.
Fig. 2.5. Projected mass density in a simulation of hydrodynamic turbulence with
the Stagger code (Padoanet al.2007). For comparison with MHD turbulence, see
Figure2.6.

2.5 Perspectives 33
Fig. 2.6. Projected mass density in a simulation of MHD turbulence with the
Stagger code (Padoan et al.2007). For comparison with purely HD turbulence,
see Figure2.5.
Glover and Mac Low(2007) performed simulations of self−gravitating MHD
turbulence including various thermal and chemical processes. They were able to
show that the fast formation of molecular hydrogen within a few Myr, which is
typically observed in molecular clouds, can be explained as an effect of turbulence.
Hennebelle and Audit(2007) investigated thermally bistable MHD turbulence and
found mass spectra similar to those inferred from CO observations of molecular
clouds. Although first computed in only two dimensions, the setup was generalized
to include self−gravity and computed in three dimensions as well. The ansatz by
Hennebelleet al.is different from simulations of isothermal MHD turbulence as it
does not rely on a driving force or an initial turbulent velocity field.
2.5 Perspectives
Up to now, most approaches to the numerical simulation of astrophysical tur−
bulence have focused on particular aspects, be it a thorough understanding of
supersonic turbulence, the study of the dynamics of self−gravitating turbulent gas or

34 The numerical simulation of turbulence
an emphasis on the effects of magnetic fields. While impressive progress has been
made in the area of isotropic turbulence, both in the HD and in the MHD case,
more complex scenarios remain elusive. The treatment of thermal and chemical
processes is still fairly approximate. Our understanding of self−gravitating tur−
bulence remains poor despite many ongoing efforts. Radiation HD and MHD in
combination with adaptive methods are just beginning.
There is one aspect that should be highlighted whenever we are talking about
the simulation of turbulence in the ISM. The huge range of length scales from
the galactic disk down to proto−stellar cores does not allow for coverage by a sin−
gle numerical simulation. In general, there are two numerical cutoffs: one towards
larger scales and the other towards smaller scales. The former poses the question of
initial as well as boundary conditions and whether simple models such as periodic
boxes and random forcing can account for those in a self−consistent fashion. The
small−scale cutoff would, in general, necessitate closures. These are considered as
essential in atmospheric sciences but have met very little attention in astrophysics.
As regards the large scales, it certainly will not be feasible in the near future to
run a simulation of a disk galaxy or even a piece of a galactic disks including all
processes of turbulence production, the different phases of the ISM, the formation
of molecular clouds and the collapse of gas giving rise to the birth of stars. If energy
is injected via random forcing in a simulation of turbulence over some subrange of
scales in the ISM, one faces the question of what the influence of the forcing might
be. That the outcome can vary depending on the applied forcing is illustrated by a
simple numerical experiment. Figure2.7shows projected mass densities obtained
from 128
3
simulations, in which purely solenoidal (left
forcing was applied. Although there is no fully developed turbulence at such low
resolution, the plots suggest that the flow morphology is genuinely different, with
markedly higher density contrasts and more pronounced intermittency in the case
Fig. 2.7. Projected mass density in 128
3
simulations of turbulence driven by a
solenoidal and a compressive force field, respectively. (Courtesy of C. Federrath.)

References 35
of compressive forcing. Indeed, this conclusion is confirmed by high−resolution
simulations (Schmidtet al.2008).
For the high−fidelity treatment of small scale effects, AMR emerges as the most
promising tool, simply following the maxim to go to very high levels of refinement
wherever the critical events take place. This was most impressively executed by
Abelet al.(2002) in AMR simulations of the formation of the first stars in the
Universe. However, circumstances are not that favourable when it comes to galac−
tic star formation and it remains to be seen whether AMR by itself will hold its
promises. An alternative approach was suggested byNiemeyeret al.(2005). The
basic idea is to combine the techniques of AMR, which has been brought to great
success in astrophysics, and subgrid scale modelling, which is commonly used in
engineering. Furthermore, the incorporation of additional physics such as thermal
processes, complex chemistry as well as radiation transport in AMR simulations
are essential for realistic models of turbulence in the ISM.
References
Abel, T., Bryan, G. L. and Norman, M. L., 2002,Science295,93
Ballesteros−Paredes, J., Gazol, A., Kim, J., Klessen, R. S., Jappsen, A.−K. and Tejero, E.,
2006,Astrophys. J.637(1
Bate, M. R. and Bonnell, I. A., 2005,Mon. Not. R. Astron. Soc.356(4
Bate, M. R., Bonnell, I. A. and Bromm, V., 2002a,Mon. Not. R. Astron. Soc.332,L65
Bate, M. R., Bonnell, I. A. and Bromm, V., 2002b,Mon. Not. R. Astron. Soc.336, 705
Berger, M. J. and Colella, P., 1989,J. Comput. Phys.82,64
Berger, M. J. and Oliger, J., 1984,J. Comput. Phys.53, 484
Biglari, H. and Diamond, P. H., 1989,Physica D. 37, 206
Biskamp, D., 2003,Magnetohydrodynamic Turbulence, Cambridge: Cambridge
University Press
Boldyrev, S., 2002,Astrophys. J.569, 841
Boldyrev, S., Nordlund, A. and Padoan, P., 2002,Astrophys. J.573, 678
Bonazzola, S., Falgarone, E., Heyvaerts, J., Perault, M. and Puget, J. L., 1987,Astron.
Astrophys.172, 293
Bonazzola, S., Perault, M., Puget, J. L., Heyvaerts, J., Falgarone, E. and Panis, J. F., 1992,
J. Fluid Mech.245,1
Bonnell, I. A., Bate, M. R. and Vine, S. G., 2003,Mon. Not. R. Astron. Soc.343(2
Chabrier, G., 2003,PASP115, 763
Chandrasekhar, S., 1951,Proc. R. Soc. Lond. A210,26
Colella, P. and Woodward, P. R., 1984,J. Comput. Phys.54, 174
Commercon, B., Hennebelle, P., Audit, E., Chabrier, G. and Teyssier, R., 2007, to appear
inProceedings of SF2A-2007: Semaine de l’Astrophysique Francaise(J. Bouvier,
A. Chalabaev and C. Charbonnel, eds.) eprint arXiv:0709.2450
Elmegreen, B. G. and Scalo, J., 2004,Ann. Rev. Astron. Astrophys.42(1
Euler, L., 1757,Principes généraux du mouvement des fluides. Berlin: Königliche
Akademie der Wissenschaften
Evans, C. and Hawley, J., 1988,Astrophys. J.332, 659

36 The numerical simulation of turbulence
Frisch, U., 1995,Turbulence, Cambridge: Cambridge University Press
Fromang, S., Hennebelle, P. and Teyssier, R., 2006,Astron. Astrophys.457, 371
Glover, S. C. O. and Mac Low, M.−M., 2007,Astrophys. J.659, 1317
Heisenberg, W., 1923,Über Stabilität und Turbulenz von Flüssigkeitsströmen, Doctoral
Dissertation, Ludwigs−Maximillian−Universität München
Hennebelle, P. and Audit, E., 2007,Astron. Astrophys.65(2
Jappsen, A. K., Klessen, R. S., Larson, R. B., Li, Y. and Mac Low, M.−M., 2005,Astron.
Astrophys.435(2
Jeans, J. H., 1902,Philos. Trans. R. Soc. Lond. A199,1
Klessen, R. S., 2000,Astrophys. J.535(2
Klessen, R. S., 2001,Astrophys. J.556(2
Klessen, R. S., Heitsch, F. and Mac Low, M.−M., 2000,Astrophys. J.535(2
Kitsionaset al., 2008, submitted toAstron. Astrophys.
Kolmogorov, A., 1941,Dokl. Akad. Nauk SSSR30, 301
Koyama, H. and Inutsuka, S.−I., 2002,Astrophys. J.564(2
Kritsuk, A. G. and Norman, M. L., 2002,Astrophys. J. Lett.569, L127
Kritsuk, A. G., Norman, M. L. and Padoan, P., 2006,Astrophys. J. Lett.638,L25
Kritsuk, A. G., Norman, M. L., Padoan, P. and Wagner, R., 2007,Astrophys. J.665, 416
Landau, L. D. and Lifshitz, E. M., 1987,Course of Theoretical Physics Vol. 6: Fluid
Mechanics, 2nd edition, Butterworth−Heinemann: Pergamon Press
Mac Low, M.−M. and Klessen, R. S., 2004,Rev. Mod. Phys.76(1
Maron, J. L. and Howes, G. G., 2003,Astrophys. J.595(1
Monaghan, J. J., 1992, inAnn. Rev. Astron. Astrophys.30, 543
Niemeyer, J., Schmidt, W. and Klingenberg, C., 2005, inRingberg Proceedings on
Interdisciplinary Aspects of Turbulence, MPA/P15, 175
Nordlund, Å. and Galsgaard, K., 1995,A 3D MHD Code for Parallel Computers.
Technical report, Astronomical Observatory, Copenhagen University
O’Shea, B. W., Bryan, G., Bordner, J., Norman, M. L., Abel, T., Harkness, R. and Kritsuk,
A., 2004,Adaptive Mesh Refinement – Theory and Applications(T. Plewa, T. Linde,
and V. G. Weirs, eds.) Springer Lecture Notes in Computational Science and
Engineering
Padoan, P. and Nordlund, Å., 2007,Astrophys. J.576, 870
Padoan, P., Nordlund, Å., Kritsuk, A. G., Norman, M. L. and Li, P. S., 2007, inTriggered
Star Formation in a Turbulent ISM(B.G.Elme
green and J. Palous, eds.) Cambridge:
Cambridge University Press, 283
Porter, D., Pouquet, A. and Woodward, P., 2002,Phys. Rev. E.66, 026301
She, Z.−S. and Lévêque, E., 1994,Phys. Rev. E.72, 336
Schmidt, W., and Klessen, R.S., Federrath, C., 2008, submitted toPhys. Rev. Lett.,
arXiv:0810.1397
Schmidt, W., Federrath, C., Hupp, M., Maier, A. and Niemeyer, J., 2008, submitted to
Astron. Astrophys., preprint arXiv:0809.1321
Schmidt, W., Hillebrandt, W. and Niemeyer, J. C., 2006,Comp. Fluids35, 353
Stone, J. M. and Norman, M. L., 1992,Astrophys. J. Suppl.80, 791
Sytine, I. V., Porter, D. H., Woodward, P. R., Hodson, S. W. and Winkler, K., 2000,J.
Comput. Phys.158, 225
Teyssier, R., 2002,Astron. Astrophys.385, 337
Vàzquez−Semadeni, E., Gómez, G. C., Jappsen, A. K., Ballesteros−Paredes, J., González,
R. F. and Klessen, R. S., 2007,Astrophys. J.657(2

3
Numerical methods for radiation
magnetohydrodynamics in astrophysics
R. Klein and J. Stone
Abstract
We describe numerical methods for solving the equations of radiation mag−
netohydrodynamics (MHD
essential for the investigation of the time−dependent and multidimensional
dynamics of a variety of astrophysical systems, although our particular
interest is motivated by problems in star formation. Over the past few
years, the authors have been members of two parallel code development
efforts, and this review reflects that organization. In particular, we dis−
cuss numerical methods for MHD as implemented in the Athena code,
and numerical methods for radiation hydrodynamics as implemented in
the Orion code. We discuss the challenges introduced by the use of adap−
tive mesh refinement (AMR
directions for future developments.
3.1 Introduction
The dynamics of astrophysical systems described by the equations of radiation
magnetohydrodynamics (MHD
ter regimes, from the interiors of stars (Kippenhahn & Weigert 1994), to accretion
disks around compact objects (Turneret al.2003), to dusty accretion flows around
massive protostars (Krumholzet al.2005, 2007a), to galactic−scale flows onto
AGN (Thompsonet al.2005). All of these systems have in common that mat−
ter, radiation and magnetic fields are strongly interacting and that the energy and
momentum carried by the radiation field is significant in comparison to that carried
by the gas. Thus, an accurate treatment of the problem must include analysis of
both the matter and the radiation, as well as the magnetic fields, and their mutual
interaction.
Structure Formation in Astrophysics. ed. G. Chabrier. Published by Cambridge University Press.
cCambridge University Press 2009.

38 Numerical methods for radiation magnetohydrodynamics in astrophysics
Numerical methods are essential to generate time−dependent and multidimen−
sional solutions to the non−linear equations of radiation MHD. In fact, numerical
methods for both MHD and radiation hydrodynamics are in, and of, themselves
active areas of development, let alone for the combined system of radiation MHD.
Our goal in this chapter is to discuss the current status of numerical methods for
radiation MHD, with emphasis on the challenges and areas where further devel−
opment is needed. However, in order to clarify the discussion, we will describe
methods for MHD and those for radiation hydrodynamics in separate sections,
using examples drawn from two separate efforts. In particular, we will use the
MHD algorithms implemented in the Athena code (developed by Stone and col−
laborators) to demonstrate the issues and challenges associated with higher−order
Godunov methods for MHD using adaptive mesh refinement (AMR
will use the radiation hydrodynamics algorithms implemented in the Orion code
(developed by Klein and collaborators) to demonstrate the challenges associated
with higher−order Godunov methods for radiation hydrodynamics using AMR.
Our focus is on a discussion of the fundamental issues for numerical algorithms
in each of these areas, rather than a step−by−step description of the actual codes
(for a thorough discussion of the algorithms in Athena, see Gardiner and Stone
(2005, 2008) and Stoneet al.(2008 et al.
(1998 et al.(2005 et al.
(2007b)).
In addition to a discussion of methods for MHD (in 3.2) and radiation hydrody−
namics (in 3.3), we shall also discuss the issues associated with the implementation
of AMR (in 3.4), as well as directions for future research (in 3.5).
3.2 Magnetohydrodynamic algorithms: the Athena code
3.2.1 Introduction
There are a wide range of algorithms that have been developed to solve the equa−
tions of compressible MHD. For example, the ZEUS code (Stone & Norman 1992)
implements methods based on operator splitting a non−conservative formulation
of the equations of motion. However, operator split methods based on the non−
conservative formulation are not suitable for use with AMR, and therefore in recent
years there has been a surge in the development of higher−order Godunov methods
for MHD.
In this section, we describe directionally unsplit, higher−order Godunov methods
for MHD as implemented in the Athena code (Gardiner & Stone 2005, 2008; here−
after GS05 and GS07, respectively). Other codes that implement similar methods

Other documents randomly have
different content

pahasti, että se vei kaikki hänen voimansa, ja ensi hetkenä hän luuli
olevansa kuoleman oma. — — —
Boo oli sillä välin, hänestä erottuaan, kulkenut eteenpäin, koko
ajan ajatellen Ingmaria. Alkumatkasta hän kiirehti aika lailla, sillä
hän olisi mielellään tahtonut ensimmäisenä ennättää vuorelle, mutta
vähän ajan kuluttua hän hiljensi kulkuaan. Hän naurahti alakuloisesti
omaa itseään. Tiedänhän sen, että menisinpä minä millaista kyytiä
tahansa, en minä sittenkään joudu yhtä pian kuin Ingmar. En ole
ikänäni tavannut toista niin hyväonnista miestä, joka niin varmasti
aina kykenisi saamaan tahtonsa perille. En totta tosiaan voi muuta
odottaa, kuin että hän lopultakin vie Gertrudin kotiin Taalainmaahan.
Nytkin viimeisenä puolena vuonna on hänen tahtonsa ollut täällä
siirtokunnan kesken kaikki kaikessa.
Mutta saavuttuaan Öljymäelle sovittuun paikkaan Boo ei
tavannutkaan Ingmaria siellä odottamassa, ja hän tuli vallan
hyvilleen. Hän ryhtyi työhön ja jatkoi sitä kotvan aikaa. Onpa hän nyt
edes kerran saanut huomata valinneensa väärän tien, Boo ajatteli.
Päivä alkoi sitten sarastaa, ja kun Ingmaria ei vieläkään kuulunut,
alkoi Boo käydä levottomaksi, sillä kaiketi hänelle oli sattunut jokin
vahinko. Hän alkoi laskeutua vuorta alaspäin etsiäkseen häntä. Se
Ingmar on sentään omituinen, hän ajatteli, sillä vaikk'ei minun
oikeastaan kannata hänestä pitää, niin luultavasti minun kumminkin
olisi paha mieli, jos hänelle on sattunut tapaturma. Kunnon mies hän
on, ja täällä Jerusalemissa hän on suurenmoisesti auttanut meitä.
Joll'ei Gertrudia olisi meitä erottamassa, niin pian meistä voisi tulla
täysin hyvätkin ystävät.
Aamu valkeni nopeasti, ja Josafatin laaksoon asti ehdittyään Boo
piankin löysi Ingmarin pitkänään parin hautapaaden välissä.

Ingmarin kädet olivat siteissä, ja hän makasi liikahtamatta, mutta
kuultuaan Boon raskaat askelet hän kohotti päätään. "Boo, sinäkö se
olet?" hän sanoi. — "Niin olen", Boo vastasi, "mitenkäs sinun laitasi
on?" Sitä sanoessaan hän näki Ingmarin kasvot; molemmat silmät
olivat ummessa, toinen niistä turvoksissa, ja silmäkulmasta tihkui
verta. "Kuulehan nyt, miten sinun oikein on käynyt?" Boo kysyi ja
hänen äänensä oli omituisen sumea. — "Jouduin tappeluun noiden
haudanavaajien kanssa", Ingmar sanoi, "kompastuin sitten yhteen
heistä, ja hänellä oli kädessään veitsi, joka osui suoraan silmääni."
Boo laskeutui polvilleen Ingmarin viereen ja alkoi irroittaa siteitä
hänen käsistään. — "Miten sinä haudanavaajien kanssa jouduit
tappeluun?" Boo sanoi. — "Kuljinhan laakson kautta, ja silloin minä
kuulin heidän kaivamisensa." — "Ja sittenkö et voinut suvaita, että
se ruumis olisi tänäkin yönä heitetty pois haudasta?" — "Juuri niin",
Ingmar sanoi, "sitä en voinut suvaita." — "Siinä sinä teit miehen
työn", Boo sanoi. — "Enkä tehnyt", sanoi Ingmar, "tyhmää se oli,
mutta en malttanut olla sitä tekemättä." — "Mutta minä sanon
sinulle", Boo sanoi, "että vaikka se tyhmää oli, niin nyt minä olen
sinun ystäväsi koko ikäni, sen vuoksi että tämän teit."

ÖLJYMÄELLÄ
Ingmar sai hoitajakseen erään lääkärin englantilaisten suuresta
silmäsairaalasta. Tämä kävi joka päivä siirtokunnan talolla
muuttamassa käärettä. Ingmarin silmä parani pian hyväksi, ja hän
tunsi jo kohta olevansa niin terve, että voi nousta vuoteelta ja olla
jalkeilla.
Mutta eräänä aamuna lääkäri huomasi, että hänen terve silmänsä
rupesi verestämään ja näytti turvonneelta. Hän kävi hyvin
levottomaksi ja alkoi heti antaa ohjeita tämän silmän hoitamiseen.
Sitten hän kääntyen Ingmarin puoleen sanoi hänelle suoraan, että
hänen oli parasta lähteä pois Palestiinasta heti ensi tilassa. "Pelkään,
että te olette saanut itämaiden vaarallisen silmätaudin", hän sanoi.
"Minä nyt koetan auttaa teitä niin hyvin kuin voin, mutta teidän
ainoa silmänne ei nyt ole kyllin vahva torjumaan tartuntaa, jota
täällä kaikkialla on ilmassa. Jos jäätte tänne, tulee teistä parissa
viikossa sokea aivan auttamattomasti."
Tämä herätti suurta surua siirtokunnassa, eikä ainoastaan
Ingmarin sukulaisten, vaan kaikkien muidenkin kesken. He kaikki
sanoivat, että Ingmar oli tehnyt heille mitä suurimman hyväntyön,
kun hän oli houkutellut heitä ansaitsemaan leipänsä hiessä päin

muiden ihmisten lailla, ja että sellaista miestä ei siirtokunnasta
koskaan pitäisi päästää. Mutta kaikki tunsivat, että Ingmarin oli
pakko lähteä, ja mrs Gordon sanoi heti, että jonkun veljistä pitäisi
heti varustautua saattamaan häntä, kun hänen nyt oli vaikea
matkustaa yksin.
Ingmar kuunteli kauan ääneti sitä puhetta, että hänen piti
matkustaa. Viimein hän sanoi: "Eihän se nyt niin varmaa ole, että
minä tulen sokeaksi, jos tänne jään." Mrs Gordon kysyi, mitä hän
ajatteli niin sanoessaan. — "En vielä ole saanut ajetuksi sitä asiaa,
jota varten tänne tulin", hän sanoi hitaasti. — "Aiotteko te siis
kieltäytyä matkustamasta?" mrs Gordon kysyi. — "Sitä minä
ajattelen", sanoi Ingmar, "kovin katkeraltahan tuntuisi tällainen
pakollinen kotiinlähtö kesken asiansa."
Nyt nähtiin, kuinka suuressa arvossa mrs Gordon Ingmaria piti,
sillä hän meni Gertrudin luokse ja kertoi hänelle, ettei Ingmar
tahtonut lähteä pois, vaikka hän oli vaarassa tulla sokeaksi, jos jäi.
"Tiedät kyllä, kuka siihen on syypää, ettei hän tahdo lähteä", sanoi
mrs Gordon. — "Tiedän kyllä", Gertrud vastasi.
Gertrud katsoi ihmetellen mrs Gordoniin, ja tämä ei sen enempää
puhunut. Mrs Gordon ei voinut suoraan kehoittaa häntä rikkomaan
siirtokunnan keskeisiä sääntöjä, mutta Gertrud ymmärsi, että hänen
kyllä suvaittaisiin tehdä Ingmarin hyväksi, mitä hän vain tahtoi.
Olisipa joku muu kuin minä tässä asemassa, niin varmaan mrs
Gordon olisi vähemmän myöntyväinen, hän hiukan harmissaan
ajatteli. Mutta kun he eivät minua pidä oikein selväpäisenä, niin
taitaisivat olla iloissaankin, jos lähtisin tieheni.
Pitkin päivää tuli yksi toisensa jälkeen Gertrudin luo puhumaan
Ingmarista. Kukaan ei rohjennut suoraan sanoa hänelle, että hänen

pitäisi lähteä Ingmarin seurassa kotiin, mutta ruotsalaiset talonpojat
istuivat hänen viereensä ja puhuivat siitä sankarista, joka taisteli
vainajan puolesta Josafatin laaksossa, ja sanoivat, että nyt Ingmar
oli näyttänyt olevansa oikean vanhan sukupuun oksa. "Suuri vahinko
olisi, jos hänenlaisensa mies tulisi sokeaksi."
"Minä katselin Ingmaria sinä päivänä, jona Ingmarilassa pidettiin
huutokauppaa", sanoi Gabriel, "ja minä sanon sinulle, että jos olisit
hänet silloin nähnyt, et ikinä olisi voinut häneen suuttua."
Mutta Gertrudista tuntui aivan kuin olisi koko päivän ponnistellut
sellaista unta vastaan, josta tahtoisi rientää pois, mutta ei voi päästä
hievahtamaan. Hän olisi tahtonut auttaa Ingmaria, mutta ei tiennyt,
mistä siihen saisi voimaa. Miten minä voin tehdä tämän Ingmarille,
kun en enää rakasta häntä? hän kysyi itseltään. Ja miten minä voin
olla sitä tekemättäkään, kun tiedän, että hän tulee sokeaksi? sitäkin
hän kysyi itseltään.
Illalla Gertrud seisoi talon edustalla suuren sykomoripuun alla ja
ajatteli lakkaamatta sitä samaa, että hänen pitäisi lähteä Ingmarin
mukaan, mutta ettei hänen voimansa siihen päätökseen riittänyt.
Boo tuli silloin ulos hänen luokseen.
"Sattuu niitä omituisia kohtauksia", Boo sanoi, "kuten se, että
ihminen voi ilahtua onnettomuudestaan ja taas ruveta onneansa
murehtimaan."
Gertrud kääntyi suoraan häntä kohti ja katsoi häneen säikähtynein
silmin. Hän ei sanonut mitään, mutta Boo arvasi hänen ajattelevan:
"Tuletko sinäkin nyt minua ärsyttämään ja vainoamaan?"

Boo puraisi huultaan, ja hänen kasvonsa hiukan vääristyivät,
mutta heti sen jälkeen hän kuitenkin sanoi, mitä oli aikonut.
"Kun on pitänyt jostakin ihmisestä koko ikänsä", hän sanoi, "silloin
ei voi olla pelkäämättä hänen menettämistään. Mutta kaikkein
enimmän pelottaa sellainen menettäminen, joka johtuu siitä, että
huomaa hänet niin kovasydämiseksi, ettei hän voi antaa anteeksi
eikä unohtaa."
Boo lausui nämä kovat sanansa hyvin lempeällä äänellä, eikä
Gertrud suuttunut, vaan herahti itkuun. Hänen mieleensä muistui
kerran näkemänsä uni, jossa hän oli puhkaissut Ingmarin silmät.
Nythän näkyy, että tämä uni oli totta ja että minä tosiaan olen niin
kovasydäminen ja kostonhimoinen kuin silloin unissani olin, hän
ajatteli. Varmaankin Ingmar nyt menettää näkönsä minun tähteni.
Hän tuli hyvin surulliseksi, mutta hänen painajaisen tapainen
voimattomuutensa ei siitä väistynyt, eikä hän vielä yölläkään maata
ruvetessaan ollut lopullisesta päätöksestä selvillä.
Aamulla hän nousi hyvin varhain ja lähti tapansa mukaan
kiipeämään kukkulain yli Öljymäelle.
Koko matkan hän ponnisteli päästäkseen irti tuosta mielensä
raukeudesta. Hän näki kyllä, mitä hänen piti tehdä, mutta hänen
tahtonsa oli uuvuksissa eikä jaksanut voittaa sitä, minkä kahleissa
hän oli.
Hän muisti kerran nähneensä tervapääskysen, joka oli pudonnut
maahan ja hietikossa räpisteli siipiään saamatta niin paljon ilmaa,
että olisi voinut heittäytyä niiden varaan. Juuri samoin hänkin tunsi
turhaan pyristelevänsä pääsemättä paikaltaan.

Mutta saavuttuaan Öljymäen huipulle, samalle paikalle, jossa
hänen tapansa oli odottaa auringonnousua, hän nyt näki, että se
Jeesuksen näköinen dervishi oli siellä jo ennen häntä. Hän istui
maassa, jalat allaan ristissä, ja hänen suuret silmänsä katselivat
Jerusalemia.
Gertrud ei hetkeksikään unohtanut, että mies vain oli
dervishiparka, jossa ei ollut muuta mainittavaa kuin että hän yllytti
seuralaisiaan kiihkeämpään tanssiin kuin kukaan muu. Mutta kun
hän näki hänen kasvonsa ja tuskaisen piirteen hänen suunsa
ympärillä, silloin koko hänen ruumiinsa vavahti. Hän jäi seisomaan
kädet ristissä aivan miehen lähelle ja katsomaan häntä.
Hän ei ensinkään uneksinut, eikä hän nähnyt näkyjä, tuo suuri
yhdennäköisyys vain vaikutti, että hän luuli hänessä näkevänsä
jumalvoimaisen ihmisen.
Nyt hän uudestaan uskoi, että jos tämä mies vain tahtoi esiintyä
ihmisille, silloin nähtäisiin, että hän on syventynyt kaiken tiedon
syntyihin saakka. Hän uskoi, että myrskyt ja laineet tottelivat tämän
miehen käskyjä, että hän oli tyhjentänyt kaiken kärsimyksen kalkin,
uskoi että hänen ajatuksensa käsittivät tuntemattomankin, sen mitä
tutkimaan kukaan muu ei pystynyt.
Hän tiesi, että jos hän olisi ollut sairas, niin häntä katsoessaan hän
olisi parantunut.
Tämä ei voi olla tavallinen ihminen, hän ajatteli. Minä tunnen
olevani taivasten autuudessa, kun vain saan häntä katsella.
Hän oli seisonut kauan aikaa dervishin vieressä nähtävästi tämän
huomaamatta, mutta yht'äkkiä dervishi kääntyi suoraan häneen päin.

Gertrud lyyhistyi kokoon noiden silmien edessä, sillä tuota katsetta
hän ei voinut kestää.
Mies katseli häntä ääneti ja tyynesti ainakin koko minuutin ja
sitten ojensi hänelle kätensä suudeltavaksi niinkuin hänen
seuralaistensa tapana oli tehdä. Ja Gertrud suuteli nöyrästi hänen
kättään.
Sen jälkeen mies ystävällisen totisena viittasi hänet nyt menemään
matkaansa, jottei hän enää häiriintyisi.
Gertrud kääntyi kuuliaisena hänestä pois ja käveli hitaasti
alamäkeen. Tuo miehen hyvästijättötapa oli hänestä niin omituinen,
että varmaankin se jotakin merkitsi. Tuntui aivan kuin hän olisi sillä
sanonut: Jonkin aikaa sinä olet ollut minun omani ja palvellut minua,
mutta nyt minä päästän sinut. Elä nyt maailmassa kanssaihmistesi
hyväksi!
Kuta lähemmäs siirtokuntaa hän tuli, sitä enemmän tuo suloinen
huumaus häipyi. "Nyt minä tiedän, ettei hän ole Kristus. Enkä minä
uskokaan häntä Kristukseksi", hän nyt uudestaan sanoi.
Mutta suuren muutoksen tuo miehen näkeminen hänessä sai
aikaan. Olihan hän ainakin ollut Gertrudin silmissä ikään kuin elävä
Kristuksen kuva, ja jo sen vuoksi hänestä nyt jok'ainoa maan kivi
tuntui kertaavan niitä pyhiä oppeja, joita hän muinoin oli tässä
maassa julistanut, ja kukkaset tuntuivat kuuluttavan, kuinka suloista
oli vaeltaa hänen teitään.
Heti talolle palattuaan Gertrud meni suoraan Ingmarin
huoneeseen. "Nyt,
Ingmar, minä lähden sinun mukanasi kotiin", hän sanoi.

Pari syvää henkäystä kohosi Ingmarin rinnasta. Näkyi, että hän
tunsi suurta huojennusta.
Hän otti Gertrudin kädet omiensa väliin ja puristi niitä. "Nyt Jumala
on minulle ollut kovin hyvä", hän sanoi.

"KERRAN TOISEMME
KOHDATAHAN"
Siirtokunnassa oli erinomainen kiire. Taalain talonpojilla oli kullakin
omassa huoneessaan niin paljon puuhaa, etteivät he joutaneet
ulkotöilleen pelloille eikä viinitarhoihin, ja ruotsalaiset lapset olivat
kiireitten kotitöiden vuoksi saaneet luvan koulusta.
Ingmarin ja Gertrudin lähtö oli määrätty tästä päivästä
ylihuomiseksi, ja siitä syystä nyt oli niin kiire laittaa kuntoon kaikki,
mitä kenenkin teki mieli lähettää heidän muassaan kotipuoleen.
Nyt oli sopiva tilaisuus lähettää pieni muistolahja
lapsuudenaikaisille koulutovereille ja vanhoille ystäville, jotka koko
ikänsä olivat pysyneet uskollisina. Nythän saattoi ilmaista, että
toisinaan vielä ystävällisesti ajateltiin monta kotona olevaa entistä
toveria, joiden kanssa kiihkeimpänä alkuaikana oli lakattu
seurustelemasta, ja vanhaa viisasta väkeä, joiden neuvoista matkalle
lähtiessä oli pahastuttu. Nythän sopii jollain tavalla ilahduttaa
vanhempia, mielitiettyä, kirkkoherraa ja koulumestaria, joka oli
heidät kaikki kasvattanut.

Ljungo Björn ja Kolåsan Gunnar istuivat koko päivän kynä
kourassa kankein sormin piirtäen kirjeitä ystävilleen ja sukulaisille, ja
Gabriel taas seisoi sorvauspenkin ääressä laittaen pieniä
öljypuukuppeja, ja Katri Ingmarintytär pani moneen eri kääreeseen
suuria valokuvia Getsemanesta ja Pyhän haudan kirkosta, siitä
uhkeasta talosta, jossa he asuivat, ja heidän kauniista
kokoussalistaan.
Lapset kokivat tushipensselillä maalata kuvia ohuille
öljypuuliuskoille, jota he olivat oppineet amerikkalaisten koulussa, tai
sommittelivat valokuvakehyksiä, joihin koristeiksi liimattiin Itämaalla
kasvavia tähkäpäitä ja siemeniä ja hedelmänsydämiä.
Martta Ingmarintytär leikkasi kangaspuista hienon palttinansa ja
rupesi kirjailemaan nimiä pyyhkeisiin ja ruokaliinoihin, jotka olivat
aiotut hänen langolleen ja kälylleen. Ja hän hymyili ajatellessaan,
että nyt he kotipuolessa saavat nähdä hänen vielä muistavan entisen
taitonsa kutoa hienoa ja tiivistä kangasta, vaikka olikin
Jerusalemissa.
Molemmat Amerikassa olleet Ingmarintyttäret sitoivat
päällyspapereita aprikoosi- ja persikkahillopurkkeihin, ja niiden
pohjiin he kirjoittivat rakkaita nimiä, joita he eivät voineet muistella
muuten kuin vesissä silmin.
Israel Tuomaanpojan vaimo leipoi mesileipätaikinaa, ja sitä paitsi
hänellä oli uunissa piirakka silmällä pidettävänä. Piirakka oli aiottu
Ingmarin ja Gertrudin evääksi, mutta mesileivät, jotka voivat säilyä
kuinka kauan tahansa, heidän piti koskematta viedä Myckelsmyran
eukolle, joka oli siistinä ja pyhäpuvussaan seisonut tienvieressä, kun
Jerusalemin matkalaiset olivat ajaneet siitä ohi, ja Eeva
Gunnarintyttärelle, joka kerran oli kuulunut heidän pariinsa.

Sitä myöten kuin pikku kääröt valmistuivat, ne vietiin Gertrudille,
ja hän pani ne kaikki paikalleen suureen matka-arkkuun.
Mutta jollei Gertrud olisi ollut niin perehtynyt syntymäpitäjäänsä, ei
hän olisi voinut ottaa toimittaakseen näitä kaikkia eri lähetyksiä
niiden oikeille omistajille, sillä joidenkuiden osoitteet olivat hyvin
hullunkurisia. Hän sai pariinkin kertaan ajatella, ennen kuin arvasi,
mistä hän löytäisi "Ranssun, joka asui tienhaarassa", ja "Liisan, joka
oli Petter Laurinpojan sisar", ja "Erkin, joka toissa vuonna oli
lautamiehen renkinä".
Ljungo Björninpojan poika Gunnar toi isoimman käärön. Se oli
osoitettu "Katrille, joka oli vieruskumppanini koulussa ja oli salokyliltä
kotoisin". Sukunimeä hän ei muistanut; mutta tälle Katrille hän oli
suutaroinut parin kiiltonahkakenkiä, joissa oli korkeat, käyristetyt
korot. Hän tiesi tunnossaan, että sen kauniimpaa kenkätyötä
siirtokunnassa ei ikinä ollut tehty. "Ja vie hänelle terveiset, että tulee
tänne minun luokseni niinkuin kotoa lähtiessä sanottiin", hän sanoi
jättäessään käärön Gertrudin huostaan.
Mutta suurtalolliset tulivat Ingmarin luokse ja jättivät hänelle
kirjeitä ja tärkeitä toimia. "Ja sitten sinä käyt kirkkoherran ja
lautamiehen ja koulumestarin luona", he lopuksi aina sanoivat, "ja
kerro heille, että olet omin silmin nähnyt meidän olevan hyvissä
oloissa ja että me asumme oikeassa talossa emmekä maakuopissa,
ja että me teemme työtä ja saamme kelvollista ravintoa ja että me
elämme siivosti oikeiden ihmisten lailla."
Siitä lähtien kun Boo oli tavannut Ingmarin Josafatin laaksossa,
olivat heidän välinsä olleet hyvin ystävälliset, ja Boo oli kaikkina
joutohetkinään istunut Ingmarin luona, kun tämä taudin aikana oli
muutettu erityiseen vierashuoneeseen. Mutta sinä päivänä, jolloin

Gertrud oli Öljymäeltä tultuaan luvannut lähteä Ingmarin mukana
Taalainmaahan, ei Boo ollenkaan pistäytynyt sairashuoneeseen.
Ingmar kyseli häntä monta kertaa, mutta kukaan ei ollut Boota
tavannut.
Boon poissaolo alkoi Ingmaria yhä enemmän vaivata, kuta
pitemmälle päivä kului. Ensi hetkenä hän oli ollut Gertrudin
myöntymisestä hyvillään ja onnellisena. Hän oli silloin ollut vain
kiitollinen saadessaan luvan viedä hänet pois tästä vaarallisesta
maasta, jonne Gertrudin oli hänen syystään ollut melkein pakko
lähteä. Ja tosin hän yhä vieläkin oli siitä iloissaan, mutta vähitellen
hänessä selveni selvenemistään se tunne, että hän ikävöi vaimoansa.
Viimein hän ei enää voinut pitää mahdollisena koko tämän aikeensa
toteuttamista. Väliin hänestä tuntui olevan pakko kertoa Gertrudille
koko elämänsä, mutta tarkemmin mietittyään hän ei sitä kuitenkaan
rohjennut. Saatuaan tietää hänen pitävän toisesta Gertrud varmaan
heti ensimmäiseksi kieltäytyisi lähtemästä hänen kanssaan
kotimatkalle. Eihän hän edes tiennyt, kenestä Gertrud piti, hänestä
itsestäänkö vai jostakin muusta. Välistä hän oli luullut Boota siksi,
mutta nyt viime aikoina hänen oli täytynyt huomata, ettei Gertrud
koko siirtokunnassa oloaikanaan ollut rakastanut ketään muuta kuin
sitä, jota hän kävi Öljymäellä odottamassa. Ja nyt hänen maailmaan
palatessaan kenties vanha rakkaus Ingmariin herää hänessä
uudelleen. Ja jos niin kävisi, silloin Ingmarin kai olisi parempi mennä
hänen kanssaan naimisiin ja koettaa tehdä hänet onnelliseksi, kuin
koko elämänsä ajan kaipailla toista, jota ei koskaan voisi saada
omakseen.
Mutta vaikka hän näin koetti omaa itseään vastustaa, hän
kuitenkin yhä sitkeämmin kiintyi tuohon vastenmielisyyteen, joka
häntä kiusasi. Kääre silmillä istuessaan hän näki alati edessään

vaimonsa kuvan. Ihan nähtävästihän minut on määrätty kuulumaan
hänen kanssaan yhteen, hän ajatteli. Ei kenelläkään muulla ole
minuun mitään valtaa.
Tiedän jo, mistä syystä minä oikeastaan rohkenin ryhtyä tähän
yritykseen, hän jatkoi. Tahdoin olla yhtä kunnon mies kuin isä.
Samoin kuin hän nouti äidin vankilasta kotiin, minä aioin noutaa
Gertrudin Jerusalemista. Mutta nyt minä käsitän, ettei se voi minulta
käydä niinkuin isältä. Keskitiehen minä jään, sillä minähän pidän
toisesta.
Iltapuolella Boo viimein tuli Ingmarin luo. Hän jäi ovensuuhun
seisomaan, ikään kuin aikoen heti lähteä tiehensä. "Kuulin sinun
kyselleen minua", hän sanoi. — "Niin kyselin", sanoi Ingmar, "nyt
minä näet olen lähtöaikeissa." — "Tiedän jo, että se nyt on
päätetty", Boo ihan lyhyesti virkkoi.
Ingmarin silmät olivat kääreessä. Hän käänsi päätään sinnepäin,
missä Boo seisoi, ikään kuin olisi toivonut hänet näkevänsä.
"Puheestasi kuuluu, että sinulla on kiire", hän sanoi. — "Onhan
minulla yhtä toista askarta." Boo astui askelen oveen päin. — "Olisin
muuten tahtonut kysyä sinulta hiukan."
Boo kääntyi takaisin huoneeseen, ja Ingmar alkoi uudestaan:
"Mitenkähän on, Boo, etköhän sinä tahtoisi lähteä pariksi
kuukaudeksi kotona käymään? Luulen, että äidistäsi olisi hyvin
mieluista saada sinua nähdä." — "En käsitä, miten sinun päähäsi
sellaista johtuu", sanoi Boo. — "Minä kyllä kustantaisin matkan, jos
mielesi tekisi nyt lähteä mukaan", Ingmar jatkoi. — "Vai niin", Boo
sanoi. — "Niin kyllä", sanoi Ingmar yhä enemmän innostuen,
"äitisihän on minun ainoa tätini, ja tahtoisin häntä ilahduttaa sillä,
että hän saisi kerran vielä nähdä sinua ennen kuolemaansa." —

"Sinä kai veisit mukaasi vaikka koko siirtokunnan", Boo sanoi melkein
ivallisesti.
Ingmar vaikeni kerrassaan. Muuta hän ei enää ollut osannut
toivoakaan kuin että Boo lähtisi hänen mukanaan kotiin. Nyt minä
uskon, että Gertrud ennemmin voi ruveta pitämään hänestä kuin
minusta, kun hän vain lähtisi mukaamme, hän ajatteli. Hän on aina
ollut Gertrudille uskollinen, niin, ja tottahan sekin jotakin vaikuttaa,
kun hän itse pitää Gertrudista.
Parin silmänräpäyksen päästä Ingmar kuitenkin alkoi uudelleen
toivoa. Taisin vain tulla tyhmästi pyytäneeksi, hän ajatteli, oma syyni
se kai on, ettei hän suostu. "No niin", hän sanoi ääneen, "totta
puhuen minä tällä tarkoitin enimmäkseen vain omaa etuani." Boo ei
vastannut. Ingmar kuulosti hetken aikaa vastausta, mutta jatkoi
sitten, kun sitä ei tullut: "En oikein ymmärrä, mitenkä minä tulen
Gertrudin kanssa toimeen tällä rasittavalla matkalla. Jos minun on
kuljettava näin kääre silmillä, niin on aina vaarallista astua
veneeseen tai veneestä pois, kun meitä soudetaan laivaan. Eikä ole
helppoa kavuta köysitikapuissa ja muissa sellaisissa. Aina on hätä
käsissä, että jospa sattuisi astumaan harhaan ja putoamaan mereen.
Kun olisi miehinen mies mukana, tuntisi olevansa turvassa." — "Kyllä
kai sinä siinä oikeassa olet." — "Eikä Gertrud taitaisi osata ostaa
meille pilettejäkään." — "Sitä mieltä minäkin olen, että sinun pitäisi
ottaa joku saattomies." — "Niin", sanoi Ingmar iloisesti, "kyllähän
sinä sen ymmärrät, että minun on pakko saada joku täältä mukaan."
— "Kysyisit Gabrielia. Hänen isänsä varmaan olisi iloissaan, kun saisi
häntä nähdä."
Ingmar vaikeni taas. Aloittaessaan sitten uudestaan hän oli hyvin
alakuloisena. "Olin ajatellut saada sinut lähtemään mukaani." — "Ei,

älä veikkonen pyydä minua mihinkään sellaiseen", Boo sanoi. "Minun
on niin hyvä olla täällä siirtokunnassa. Voithan sinä saada saattajaksi
kenen muun tahansa." — "On siinä ero sillä kenenkä saa, sinähän
olet matkustellut paljon enemmän kuin kukaan muu." — "En minä
sittenkään voi lähteä."
Ingmar kävi yhä enemmän rauhattomaksi. "Tässä minä sitten olen
pahasti pettynyt", hän sanoi. "Olisin luullut sinun jotakin tarkoittavan
sillä, kun sanoit tahtovasi olla minun ystäväni." Boo katkaisi äkkiä
hänen puheensa. — "Kiitoksia vain tarjouksestasi, mutta en luule
sinun voivan keksiä sellaista syytä, joka muuttaisi mieleni, niin että
nyt minä lähden pois omille askareilleni." Heti sen sanottuaan hän
kääntyi ympäri ja meni tiehensä, ennen kuin Ingmar ennätti lisätä
sanaakaan.
Kim Boo oli eronnut Ingmarista, ei hänellä olisi voinut huomata
olevan mitään sellaista kiirettä, josta hän puhui. Hän meni vallan
hitaasti portista ulos ja istui maahan suuren sykomoripuun juurelle.
Oli jo ilta, päivänvalo oli jäljettömiin häipynyt, mutta tähdet ja pieni
tuikkiva nousukuu loistivat kauniisti.
Tuskin viittä minuuttia hän oli siellä istunut, kun portti hiljaa
aukeni ja Gertrud tuli ulos. Hän jäi hetkiseksi katselemaan
ympärilleen, mutta sitten hän huomasi Boon. "Boo, sinäkö se olet?"
hän sanoi ja tuli istumaan maahan hänen viereensä.
"Arvasinpa oikein, että täällä ulkona minä sinut tapaan", hän
sanoi. — "Niin, täällähän me olemme istuneet monena iltana", sanoi
Boo. — "Niin olemme", Gertrud sanoi, "mutta tämä kai on
viimeinen." — "Niin, viimeinenhän se on."

Boo istui hyvin suorana ja jäykkänä, hänen äänensä oli kylmä ja
kova, ikään kuin hän ei olisi paljon viitsinyt käydä koko
puheenaiheeseen.
"Ingmar kertoi aikovansa pyytää sinua meille matkatoveriksi." —
"Niin, jo hän sitä minulta kysyi", Boo sanoi, "mutta minä en
suostunut." — "Tiesinhän minä sen, ettei sinua haluta lähteä
mukaan", sanoi Gertrud.
He istuivat kauan aikaa aivan sanattomina, ikään kuin olisivat
olleet toisilleen ventovieraita, mutta Gertrud käänsi vähän väliä
kasvonsa Boohon päin ja katseli häntä. Boo istui pää hiukan jäykästi
kenossa ja katsoi taivaalle.
Äänettömyyttä oli kauan kestänyt, kun Boo viimein sanoi
kääntämättä katsettaan tähdistä tai muuten liikahtamatta: "Eikö
sinua palella, kun istut ulkona näin kauan?" — "Lähdenkö minä sitten
pois, sitäkös sinä tahdot?" sanoi Gertrud. Pään nyökkäyksellä Boo
sen myönsi, mutta kaiketi hän ei luullut Gertrudin sitä pimeässä
näkevän. Ääneen hän sanoi: "Kyllä minusta on hauskaa, että sinä
täällä istut."
"Tulin tänne siitä syystä tänä iltana", Gertrud sanoi, "kun ei ole
varmaa, tapaammeko enää toisiamme näin yksin ennen poislähtöäni.
Ja ajattelin, että nyt minulla on tilaisuus kiittää sinua niistä kerroista,
jolloin sinä aamuisin saatoit minua Öljymäelle." — "En minä sitä
tehnyt kuin omaksi huvikseni", Boo sanoi. — "Kiittäisin sinua
myöskin siitä kerrasta, jolloin kävit hakemassa vettä
Paratiisinkaivosta", sanoi Gertrud hymyillen. Boo näytti aikovan
vastata, mutta sanojen sijaan tuli vain nyyhkytyksen tapainen ääni.

Boon koko olento tuntui Gertrudista tänä iltana sanomattoman
liikuttavalta, ja hän sääli häntä sydämensä pohjasta. Surku minun
tulee häntä, kun hän ei koskaan enää saa minua nähdä, hän ajatteli.
Hän on niin lujatahtoinen, ettei valita, ja kuitenkin hän on, sen
tiedän, rakastanut minua koko ikänsä. Kun edes osaisin jotakin
sanoa lohduttaakseni häntä! Voi, kun löytäisin sellaiset sanat, joita
hän voisi mielensä huviksi muistella iltaisin istuessaan yksinään tässä
puun juurella!
Mutta tätä ajatellessaan Gertrud tunsi, että hänen omakin
sydämensä ikään kuin rusentui surun kourissa ja että outo,
kangistava virtaus valui hänen ruumistaan pitkin. Onhan totta, että
minäkin vastedes monesti kaipaan Boota, hän ajatteli, meillä on
viime aikoina ollut niin paljon puhumista. Minä olen nyt tottunut
näkemään hänen seestyvän ja ilahtuvan, aina kun olemme yksiin
sattuneet, ja oikein hyvää on tehnyt, kun on lähellä sellainen, joka
aina on tyytyväinen minuun, vaikka millä lailla menettelen.
Hän istui ääneti vähän aikaa. Hän tunsi kaipauksen kasvavan aivan
kuin äkkinäinen tauti. Mitä tämä on, mitä tämä on? hän ajatteli. Ei
suinkaan Boosta eroaminen voi tuottaa minulle näin raskasta surua.
Yht'äkkiä Boo rupesi puhumaan. "Ajattelen tässä erästä asiaa",
hän sanoi, "joka koko iltana ei ole väistynyt silmieni edestä." —
"Kerro minulle mitä se on!" pyysi Gertrud innostuneena. Tuntui kuin
hänen sydämensä olisi keventynyt heti kun Boo herkesi olemasta
vaiti. — "Niin", Boo sanoi, "Ingmar mainitsi kerran jotakin siitä
sahasta, joka hänellä on Ingmarilan likellä. Hän taisi sillä tarkoittaa,
että minun pitäisi lähteä hänen mukanaan ja vuokrata se." —
"Kovinhan Ingmar näyttääkin tulleen sinun ystäväksesi", sanoi
Gertrud, "sitä sahaa hän pitää arvokkaimpana koko

omaisuudestaan." — "Nyt sen sahan pärinä on soinut korvissani
koko illan", Boo sanoi. "Koski pauhaa, ja sirkkeli helisee, ja pölkyt
virrassa jyskivät kolhien toinen toistaan. Etpä usko, kuinka kauniilta
ne äänet sointuvat. Ja sitten minä tässä ajattelen, miltä tuntuisi
tehdä omintakeista työtä, olla jossakin oma isäntänsä eikä aina
alistua tällaiseen siirtokuntaelämään."
"Vai niin, vai sellaista sinä näin äänettömyydessä ajatteletkin",
Gertrud sanoi hyvinkin kylmästi, sillä hän tunsi tavalla tai toisella
pettyneensä Boon puheesta. "Eihän sinun sitä tarvitse kauan
uneksia, et kuin lähdet vain saattamaan Ingmaria kotiin."
"On minulla sentään muutakin mielessä", sanoi Boo. "Ingmar
näetkös mainitsi vielä, että hänellä on siellä sahan lähellä valmiina
hirret mökin rakennusta varten. Hän sanoi valinneensa sen paikaksi
kosken yläpuolella parin suuren koivun välissä olevan mäentörmän.
Ja nyt minä olen sitä mökkiä kuvitellut kaiken iltaa. Näen sen ihan
selvästi sekä ulkoa että sisältä. Näen oven edessä vihreät
kuusenhavut ja näen takassa palavan pystyvalkean. Ja kun sitten
tulen sahalta kotia, näen jonkun seisovan ovensuussa minua
vartomassa."
"Kuulehan Boo, nyt alkaa tulla kylmä", sanoi Gertrud keskeyttäen.
"Eikö sinustakin ole parasta nyt lähteä sisään?"
"Vai sinun nyt tekee mielesi sisään!" sanoi Boo.
Mutta he eivät silti nousseet paikaltaan, vaan istuivat vielä kauan
aikaa vieretysten, hyvin harvoin mitään virkkaen.
Kerran Gertrud keskeytti äänettömyyden sanoen: "Luulin sinun,
Boo, rakastavan siirtokuntaa enemmän kuin mitään muuta, enkä olisi

uskonut sinun millään maailman mahdilla siitä eroavan."
"Ei niin", Boo sanoi, "kyllä minä sentään jonkin vuoksi siitä
luopuisin." Gertrud istui uudelleen mietteissään ja kysyi viimein:
"Etkö tahtoisi minulle sanoa, mitä se olisi?"
Boo ei vastannut heti, vaan vasta pitkän aikaa arveltuaan ja
melkein tukahtuvalla äänellä: "Voin mielellänikin sen sinulle sanoa.
Silloin sen tekisin, jos se tyttö, josta minä pidän, tulisi sanomaan,
että hän minua rakastaa."
Gertrud jäi niin äänettömäksi, että tuskin uskalsi hengittää.
Mutta vaikk'ei yhtään sanaa lausuttu, oli Boo kuitenkin
kuulevinaan Gertrudin sanovan jotakin sellaista, että hän rakastaa
häntä, tai saman asian joillakin muilla sanoilla, sillä hän rupesi heti
jatkamaan vallan sujuvasti: "Saat nähdä, Gertrud, että sinun
rakkautesi Ingmariin vielä kerran herää.
"Sinä olet ollut jonkin aikaa hänelle suutuksissasi siitä, että hän
sinut petti, mutta nyt, kun olet antanut hänelle anteeksi, sinä taas
alat pitää hänestä samoin kuin ennen." Hän pysähtyi odottamaan
vastausta, mutta Gertrud oli vaiti. "Hirveätä olisikin jollet sinä pitäisi
hänestä", Boo sanoi. "Muistahan, mitä kaikkea hän on tehnyt
saadakseen sinut takaisin! Hän olisi tahtonut vaikka tulla sokeaksi
ennen kuin lähteä ilman sinua kotiin." — "Niin, hirveätä se olisi,
jollen minä hänestä pitäisi", Gertrud myönsi miltei sammuvalla
äänellä. Hänelle selveni, että hänen sielunsa sisimmässä oli tähän
päivään muuttumatta säilynyt se ajatus, ettei hän voinut pitää
kenestäkään muusta kuin Ingmarista.

"En minä voi, Boo hyvä, nyt tänä yönä sitä päättää", sanoi
Gertrud. "En tiedä, mikä minussa oikein on. Mutta älä nyt puhu
minulle Ingmarista."
Ja sitten he taas vuorotellen ottivat puheeksi, että heidän pitäisi
mennä sisälle, mutta jäivät kuitenkin alalleen, kunnes Katri
Ingmarintytär tuli ulos ja huusi heitä. "Ingmar pyytää teitä molempia
luokseen sisälle", hän sanoi.
Sillä aikaa kun Gertrud puheli Boon kanssa oli Katri nimittäin
sattunut olemaan Ingmarin luona. Katri oli puhunut siitä, mitä
terveisiä hän tahtoi lähettää Ingmarin mukana kotiin. Hän pitkitti
tahallaan keskustelua, nähtävästi sen vuoksi, että hänellä oli
Ingmarille jotakin sanomista, joka tuntui hänestä vaikealta lausua.
Viimein hän sanoi niin hitaasti ja välinpitämättömällä äänellä, että
jokainen Ingmarien suvun tuntija olisi arvannut hänen tulleen
pääasiaansa: "Ljungo Björn on äsken saanut kirjeen Petter
veljeltään." — "Vai niin", sanoi Ingmar. — "Niin, nyt näen, että minä
kohtelin sinua väärin silloin, kun kerran juteltiin minun kamarissani,
kohta sinun tultuasi", Katri sanoi. — "Et ollenkaan", sanoi Ingmar,
"sinä vain sanoit, mitä pidit oikeana." — "Ei niin, sillä nyt tiedän
sinulla olleen pätevät syyt erota Barbroosta", Katri sanoi. "Ljungo
Petter kirjoittaa, ettei koko ihminen ole mikään häävi." — "Minä en
ikinä ole Barbroosta puhunut pahaa sanaa", sanoi Ingmar. —
"Nythän kerrotaan, että sillä on lapsi siellä Ingmarilassa." — "Miten
vanha se lapsi on?" Ingmar kysyi. — "Tässä elokuussa se kuuluu
syntyneen." — "Se nyt ainakin on valhe", sanoi Ingmar iskien
kädellään pöytään. Lyönti oli vähällä sattua Katrin käteen, joka oli
pöydällä. — "Lyömäänkös nyt rupeat?" hän sanoi. — "En nähnyt,
että kätesi oli siinä", Ingmar sanoi.

Katri puhui samasta asiasta vielä vähän aikaa, ja Ingmar alkoi heti
rauhoittua. "Ymmärräthän, ettei minun ole sellaista hauska kuulla",
hän sanoi. "Menisitkö sinä nyt viemään Ljungo Björnille minun
terveiseni, ettei hän päästä sitä huhua liikkeelle, ennen kuin
nähdään, onko se ihan totta." — "Kyllä minä pidän huolta, että hän
pysyy vaiti", Katri sanoi. — "Ja sitten olisi hyvä, jos voisit toimittaa
Boon ja Gertrudin käymään täällä minun luonani", Ingmar sanoi.
Kun Gertrud ja Boo tulivat sairashuoneeseen, istui Ingmar
kyyryssä selin pimeässä nurkassa. Alussa he tuskin häntä
huomasivatkaan. "Mitä sinä haluat, Ingmar?" sanoi Boo. — "Sitä
teille tahdoin sanoa, että minä olen sitoutunut tehtävään, joka on yli
minun voimaini", Ingmar sanoi. Hän istui huojutellen edestakaisin
ruumistaan. — "Ingmar", sanoi Gertrud käyden lähelle häntä,
"kertoisit nyt suoraan huolesi minulle! Emmehän me koskaan,
lapsuudesta saakka, ole mitään toisiltamme salanneet." Ingmar istui
yhä tuskissaan. Gertrud tuli ihan hänen luokseen ja laski kätensä
hänen päänsä päälle. "Minäpä luulen arvaavani, mikä sinua vaivaa",
hän sanoi.
Ingmar nousi äkkiä pystyyn. "Etkä tiedä, Gertrud, äläkä suotta
arvaile", hän sanoi. Samassa hän otti lompakkonsa taskusta ja antoi
sen Gertrudille. "Löydätkö siitä suurta kirjettä, joka on kirkkoherralle
osoitettu?" — "Löydän", Gertrud sanoi, "tässä se on." — "Nyt minä
pyytäisin sinua sen lukemaan", sanoi Ingmar. "Tai lukekaa se Boon
kanssa yhdessä. Kirjoitin sen heti alussa tänne tultuani, mutta silloin
minussa oli vielä miestä jättää se lähettämättä."
Boo ja Gertrud istuutuivat nyt pöydän ääreen lukemaan. Ingmar
jäi istumaan nurkkaan. Siellä hän kuunteli, kun he käänsivät kirjeen
lehtiä. Nyt he lukevat sitä kohtaa, hän ajatteli, ja nyt taas sitä. Nyt

he ovat päässeet siihen asti, missä Barbroo kertoo minulle Birger
Sven Petterinpojan juonesta, jolla hän sai meidät mieheksi ja
vaimoksi. Nyt he ovat siinä kohdassa, kun hän osti hopeapikarin
takaisin, ja nyt he ovat ennättäneet Stigge Börjenpojan
kertomukseen asti. Ja nyt Gertrud saa tietää, etten minä enää pidä
hänestä, nyt hän oikein näkee, minkälainen raukka minä olen.
Huoneessa oli hievahtamaton hiljaisuus. Gertrud ja Boo eivät
liikahtaneet muulloin kuin lehteä kääntäessään. He tuskin näyttivät
uskaltavan hengähtää.
Ja miten Gertrud voi käsittää sitä, että minä juuri tänään huomaan
sen mahdottomaksi, että minun nyt, juuri hänen myönnyttyään, on
mahdotonta olla hänelle sanomatta pitäväni Barbroosta? Ingmar
ajatteli.
Ja ymmärränkö minä edes itse itseäni: mitenkä minun juuri silloin,
kun kuulin Barbroosta parjauksia, tuntui mahdottomalta olla toiseen
sidottuna? En ymmärrä enää, minkä vallan ohjissa kuljen; tokko
edes enää lienen terve koko ihminen?
Hän kuunteli hartaasti, odotti lakkaamatta, että toiset edes jotakin
virkkaisivat, mutta kuuli vain paperinlehtien ratinaa.
Viimein hän ei kestänyt enää kauempaa, vaan kohotti hyvin hiljaa
siteen siltä silmältä, jolla vielä näki.
Sitten hän katsahti Boohon ja Gertrudiin. He lukivat yhä, ja
molempain päät olivat niin liki toisiaan, että he istuivat miltei poski
poskessa, ja Boo oli kiertänyt käsivartensa Gertrudin vyötäisille.

Ja mitä pitemmälle he lehti lehdeltä luvussaan ennättivät, sitä
lähemmäksi toisiaan he painautuivat. Heidän poskensa olivat
hehkuvan punaiset, he nostivat väliin päänsä kumarasta ja katsoivat
syvästi toisiaan silmiin, ja heidän silmänsä olivat tavallista
tummemmat ja säteilevämmät.
Kun he vihdoin olivat päässeet viime arkin loppuun, näki Ingmar
Gertrudin painautuvan ihan Boon rintaa vasten, ja niin he istuivat
syleillen toisiaan hyvin liikutuksen valtaamina ja juhlallisina. Tuskin
he olivat muuta käsittäneetkään kaikesta lukemastaan, kuin että nyt
ei mikään enää estänyt heidän rakkauttaan. Ja Ingmar liitti hiljaa
ristiin suuret kätensä, jotka näyttivät vanhan, kärsineen ihmisen
käsiltä, ja kiitti Jumalaa. Ja kauan kesti ennen kuin yksikään heistä
kolmesta liikahti.
* * * * *
Siirtolaiset kokoontuivat suureen saliin pitämään aamurukousta. Se
oli viimeinen tässä talossa pidetty hartaushetki, johon Ingmar ottaisi
osaa. Hänen ja Gertrudin ja Boon piti parin tunnin perästä lähteä
maitse Jaffaan.
Boo oli edellisenä päivänä kertonut mrs Gordonille ja parille muulle
siirtokunnan johtajalle, että hän aikoi Ingmarin mukana lähteä
kotipitäjään ja jäädä sinne. Hänen oli silloin täytynyt kertoa kaikki
Ingmarin vaiheet. Mrs Gordon istui kauan miettien mitä oli kuullut ja
sanoi sen jälkeen: "En luule kenenkään voivan ottaa
omalletunnolleen, jos teemme Ingmarin vielä entistä
onnettomammaksi. Siksi en tahdo estää sinua lähtemästä kotiin
hänen kanssaan. Mutta minusta tuntuu, että tämmöinen lähtö saa
sinut Gertrudin kanssa kerran vielä palaamaan luoksemme. Olen
varma siitä, ettette koskaan ole muualla oikein tyytyväisiä."

Mutta jottei Ingmarin ja toisten lähdön edellä mitenkään hyvä
sopu rikkoutuisi, päätettiin siirtokunnan muille jäsenille ilmoittaa
vain, että Boo lähti Ingmaria ja Gertrudia saattamaan ollakseen
apuna vaivalloisella matkalla.
Juuri ennen aamurukousten alkua talutettiin Ingmar kokoussaliin.
Mrs Gordon nousi silloin ylös ja kävi häntä vastaan. Hän tarttui
Ingmarin käteen ja talutti hänet viereensä istumaan. Hän oli
asettanut mukavan tuolin hänelle ja hyvin huolellisesti auttoi häntä
istuutumaan siihen.
Sitten miss Young, joka istui urkujen ääressä, aloitti aamuvirren, ja
sitä seurasi tavallinen aamurukous.
Mutta kun mrs Gordon oli lopettanut lyhyen raamatunselityksen,
joka hänen oli tapana pitää joka aamu, nousi vanha miss Hoggs ylös
ja rukoili, että Jumala antaisi siunauksensa Ingmarin matkalle ja soisi
hänen onnellisesti päästä kotiin. Sitten nousi amerikkalaisia ja
syyrialaisia toinen toisensa jälkeen rukoilemaan, että Jumala
johdattaisi Ingmarin totuuden oikeaan valoon.
Muutamat puhuivat hyvin kauniin sanoin. He lupasivat rukoilla joka
päivä Ingmarin, rakkaimman veljensä, puolesta ja toivottivat hänelle
jälleen terveyttä. Ja kaikki toivoivat, että hän palaisi Jerusalemiin.
Vieraiden puhuessa olivat ruotsalaiset äänettöminä. He istuivat
vastapäätä Ingmaria ja katselivat häntä.
Ingmaria katsellessaan he ehdottomasti johtuivat ajattelemaan,
miten turvallista ja rehellistä ja hyvin järjestettyä heidän vanhassa
maassaan kaikki oli. Niin kauan kuin Ingmar oli täällä heidän
joukossaan, he olivat olleet huomaavinaan jotakin siitä täälläkin.

Mutta nyt Ingmarin lähtiessä alkoi turvattomuudentunne heitä
ahdistaa. He tunsivat ikään kuin eksyneensä laittomaan maahan
kaikkien niiden keskeen, jotka säästämättä ja armotta taistelivat
keskenään ihmissieluista.
Ja sitten he alkoivat murhemielin ajatella kotia. He näkivät koko
seudun ja sen talot ja vainiot. Ja ihmiset kulkivat rauhallisina ja hiljaa
teillä, kaikki oli turvallista, päivät kuluivat vaihteluitta, ja toinen vuosi
oli niin toisensa kaltainen, ettei niissä huomannut minkäänlaista
eroa.
Mutta talonpoikain juuri muistellessa kotiseudun suurta hiljaisuutta
johtui heidän mieleensä, miten huumaavan suurta nyt oli ollut joutua
elämään, saada elämälleen oikea päämäärä ja päästä irtautumaan
päivien harmaasta yksitoikkoisuudesta.
Ja eräs heistä kohotti äänensä ja alkoi rukoilla ruotsiksi ja sanoi:
"Minä kiitän sinua, Jumala, että olet antanut minun tulla
Jerusalemiin."
Sitten he nousivat toinen toisensa jälkeen ja kiittivät kaikki
Jumalaa siitä, että hän oli johdattanut heidät Jerusalemiin.
He kiittivät häntä rakkaasta siirtolasta, joka oli heidän suuri ilonsa.
He kiittivät häntä siitä, että heidän lapsensa jo pienuudesta asti
saivat oppia elämään yhteishengessä kaikkien ihmisten kanssa. He
odottivat, että nuorista tulisi paljon täydellisempiä kuin he itse olivat.
He kiittivät vainoista ja kärsimyksistä, he kiittivät siitä ihanasta
opista, jota heidät oli kutsuttu harjoittamaan.
Ei yksikään istuutunut todistamatta suuresta onnestaan. Ja Ingmar
ymmärsi heidän sanoneen sen hänen takiaan, jotta hän saisi sen

käsityksen ja niin kotonakin kertoisi, että he kaikki olivat onnellisia.
Heitä kuunnellessaan Ingmar oikaisihe tuolissaan hiukan
suorempana istumaan. Hän kohotti päänsä pystympään, ja jäykkä
piirre hänen suunsa ympärillä kävi selvemmäksi.
Viimein, kun todistusten tulva oli asettunut, miss Young viritti
virren, ja sen jälkeen kaikki luulivat juhlan olevan lopussa ja nousivat
lähteäkseen. Mutta silloin mrs Gordon sanoi: "Tänään laulamme
myöskin ruotsalaisen laulun."
Silloin ruotsalaiset alkoivat saman laulun, jonka he olivat
kotimaastaan lähtiessään laulaneet. "Kerran toisemme kohdatahan",
he lauloivat, "loistoss' armahan Eedenin maan."
Ja laulun kaikuessa kaikkien mieli heltyi, ja kyynelet tulvivat
useimpain silmiin. Sillä nyt he taasen ajattelivat kaikkia niitä, joita
heidän täytyi kaivata ja joita he eivät saaneet ennen tavata kuin
taivaassa.
Mutta samassa kun laulu loppui nousi Ingmar ja koetti sanoa pari
lausetta. Hän tahtoi lausua heille parisen sanaa ikään kuin viestinä
siitä maasta, jonne hän taas lähti. "Minä ajattelen, että te täällä
tuotatte suurta kunniaa meille kotona oleville", hän sanoi. "Minä
ajattelen, että kaikki ilahtuvat tavatessaan teidät missä tahansa,
taivaassa tai maan päällä. Ajattelen, ettei mikään ole niin kaunista
kuin nähdä ihmisiä, jotka suurin uhrauksin harjoittavat
vanhurskautta."

PYHIINVAELLUKSELTA PALATTUA
Nyt on kerrottava mitä Barbroo Svenintyttärelle tapahtui Ingmarin
lähdettyä Jerusalemiin.
Kun Ingmar oli ollut poissa kuukauden päivät, huomasi Ingmarilan
Vanha-Liisa, että Barbroo oli ja pysyi levottomana. Hänen silmissään
oli eukon mielestä aina ihmeellisen hurja katse. Enpä lainkaan
ihmettelisi, hän ajatteli, jos hän jonakin päivänä tulisi mielenvikaan.
Eräänä iltapuhteena hän alkoi tutkistella Barbroota. "Minua
ihmetyttää, mikä sinua oikein vaivaa", hän sanoi. "Pienenä tyttönä
ollessani näin eräänä talvena Ingmarilan emännän silmissä
samanlaisen tuikkeen kuin sinulla nyt." — "Oliko se se lapsensa
surmaaja?" kysyi Barbroo vilkastuen. — "Oli", sanoi eukko, "ja nyt
minusta alkaa tuntua kuin sinulla olisi sama mielessäsi." Barbroo ei
vastannut tähän suoraan. — "Kun minä olen kuullut tuota vanhaa
tapausta kerrottavan", hän sanoi, "olen kummastellut siitä vain yhtä
seikkaa." Vanha-Liisa kysyi, mikä se sitten olisi. — "Sitä, että miksei
hän lopettanut itseään myöskin."
Vanha-Liisa oli istunut kehräämässä. Hän laski kätensä pyörälle
pysäyttääkseen sen ja loi silmänsä Barbroohon. "Ei ole ihme että

sinun on ikävä odotella pienokaista nyt, kun miehesi on sinut
jättänyt", hän sanoi hitaasti. "Hänellä ei tietysti lähtiessään ollut
aavistustakaan siitä?" — "Emme sitä tienneet kumpikaan, ei hän
enkä minä", sanoi Barbroo matalalla äänellä, ikään kuin suru olisi
häntä niin painostanut, ettei hän kyennyt puhumaan. — "Mutta nyt
sinä varmaankin kirjoitat hänelle." — "En", sanoi Barbroo, "ainoana
lohtunani on juuri se, että hän on poissa." Eukko laski kauhuissaan
kätensä rukin pyörältä. — "Sekö sitten muka on lohtua?" hän
huudahti. Barbroo seisoi ikkunan luona tuijottaen suoraan eteensä.
— "Etkö tiedä, missä kiroissa minä olen?" sanoi hän koettaen tehdä
äänensä vakavaksi ja rauhalliseksi. — "Tiedän kyllä, ainahan jotakin
selville saa, kun väenpirtissä käänteleikse", eukko sanoi. "Olen
kuullut, että olet Surumäen sukua."
He olivat hetkisen vaiti. Vanha-Liisa polkea suristi rukkiaan, vähän
väliä katsahtaen Barbroohon, joka seisoi vielä ikkunan ääressä,
silloin tällöin vavahdellen ikään kuin horkassa. Noin viiden minuutin
päästä eukko keskeytti työnsä ja kävi ovea kohti. "Mihin menet?"
Barbroo kysyi. — "Saatanhan tuon sanoa. Menen etsimään jotakin
ihmistä, joka kirjoittaisi Ingmarille."
Barbroo asettui ihan hänen tielleen. — "Jätä se tekemättä", hän
sanoi.
"Ennen kuin se kirje on kirjoitettu, olen minä jo Långforsin
koskessa."
He seisoivat nyt vastatusten katsellen toisiaan. Barbroo oli kookas
ja väkevä. Vanha-Liisa luuli hänen aikovan väkisin häntä pidättää.
Mutta äkkiä Barbroo purskahti nauruun ja väistyi tieltä. "Kirjoita
vain", hän sanoi, "ei se minua haittaa. Silloin minun on vain tehtävä
loppu tästä ennemmin kuin aioin." — "Ei sentään", sanoi eukko, joka

huomasi parhaaksi olla varuillaan Barbroon kanssa, kun hän oli niin
toivottomana. "En minä kirjoita. En tahdo pakottaa sinua
maltittomiin tekoihin." — "Ei, kirjoita vain!" sanoi Barbroo. "En minä
sillä pidä väliä. Ymmärräthän, että minun kuitenkin täytyy lopettaa
itseni. Ja kenellähän olisi tuntoa pitkittää tällaista kurjuutta
iänkaiken."
Eukko meni takaisin rukin ääreen ja istuutui kehräämään. "Etkös
lähde
kirjettä toimittamaan?" sanoi Barbroo kulkien hänen jäljessään. —
"Saisikohan sinun kanssasi puhua pari järkevää sanaa", sanoi
Vanha-Liisa. — "Miks'ei", Barbroo sanoi, "puhu pois vain."
"Ajattelen sillä lailla", Vanha-Liisa sanoi, "että lupaisin pitää kaikki
salassa, jos vain sinä suostut jättämään lapsen vahingoittamatta ja
itsesi myöskin, kunnes nähdään, käykö niinkuin sinä luulet." Barbroo
jäi seisoalleen miettimään. — "Lupaatko, että sitten saan tehdä, mitä
tahdon?" — "Lupaan", sanoi eukko, "sitten saat tehdä mitä haluat,
sen minä lupaan." — "Olisi minun mielestäni sen saman vallan hyvin
voinut tehdä yksin tein", sanoi Barbroo välinpitämättömän
näköisenä. — "Olen viime aikoina luullut sinun ennen kaikkea
toivoneen, että Ingmar saisi sovittaa, mitä on rikkonut", sanoi eukko,
"mutta tuskin siitä nyt mitään tulee, jos hän saa kuulla sellaisia
uutisia." Barbroo säpsähti ja koetti sydäntään. — "Käyköön sitten
sinun tahtosi mukaan", hän sanoi, "mutta se on raskas lupaus. Sinun
tulee tarkkaan muistaa, ettet petä minua."
He pitivätkin hyvin sopimuksensa. Vanha-Liisa ei ilmaissut asiaa
kenellekään, ja Barbroo oli sen jälkeen niin varuillaan, ettei kukaan
aavistanut mitä hänelle piti tapahtua. Hänen onnekseen tuli sinä
vuonna aikainen kevät. Lumi suli metsistä jo maaliskuussa. Barbroo

antoi heti ensimmäisten vihreiden korsien maasta kohottua ajaa
osan elukoita karjamökille, joka oli kaukana yksinäisellä salolla. Hän
itse lähti Vanhan-Liisan kanssa sinne niitä hoitamaan.
Sitten toukokuun lopulla lapsi syntyi. Se oli tälläkin kertaa poika,
mutta paljon huonomman näköinen kuin se, joka Barbroolle syntyi
edellisenä keväänä. Se oli pieni ja heikko ja kirkui alituiseen. Kun
Vanha-Liisa näytti lasta Barbroolle, hymyili tämä katkerasti. "Suotta
sinä pakotit minun elämään tuon lapsen takia", hän sanoi. — "Ei
kukaan niin pienestä näe, mikä siitä vielä voi tulla", eukko sanoi. —
"Muista nyt lupauksesi, että annat minun tehdä, mitä tahdon", sanoi
Barbroo lujasti. — "Kyllä muistan", eukko sanoi, "mutta ensin minun
täytyy tietää, että se on sokea." — "Älä ole olevinasi niin tietämätön,
kun kuitenkin näet, minkälainen tämä lapsi on", sanoi Barbroo.
Barbroo itse oli huonompana kuin viime kerralla. Koko ensi viikon
hän oli niin heikko, ettei voinut nousta sängystä. Lapsi ei maannut
tuvassa, vaan eukko oli kätkenyt sen pieneen latoon, joka oli
karjamökin tanhualla. Hän hoiti sitä yötä päivää, juotti sille
vuohenmaitoa ja sai vaivaa nähden sen pysymään hengissä. Pari
kertaa päivässä hän toi sen sisään tupaan. Silloin Barbroo aina
kääntyi seinään päin, jottei tarvitsisi nähdä sitä.
Eräänä päivänä Vanha-Liisa seisoi karjamökin pienessä ikkunassa
ja katseli ulos. Hän piteli lasta käsivarsillaan, se kirkui tapansa
mukaan, ja eukko säälitteli itsekseen sen pienuutta ja kurjuutta.
"Kas, kas", sanoi hän äkkiä ja kumartui eteenpäin paremmin
nähdäkseen, "tuolta tulee oikein kylävieraita." Hetken perästä hän
toi pojan Barbroolle. "Ota sinä lapsi siksi aikaa. Menen tästä
tulijoiden luokse ja kiellän heitä sisään tulemasta, kun sinä olet
sairas." Hän laski lapsen vuoteelle, ja Barbroo antoi sen maata

koskematta siihen. Se kirkui koko ajan voimainsa takaa. Vanha-Liisa
tuli hetken päästä takaisin. "Tuon lapsen kirkuna kuuluu yli koko
salon", hän sanoi. "Jollet saa sitä vaikenemaan, niin mahdotonta sitä
on pitää ihmisiltä salassa." Sitten hän lähti uudelleen, eikä Barbroo
tiennyt muuta keinoa kuin tarjota lapselle rintaa.
Eukko oli kauan aikaa ulkona. Hänen takaisin tullessaan lapsi
nukkui, ja Barbroo makasi kyljellään ja katseli sitä. "Älä ole
peloissasi", sanoi Vanha-Liisa. "Eivät he mitään kuulleet ja lähtivätkin
ihan toiseen suuntaan kulkemaan." Barbroo katsahti häneen
raskaasti. — "Nyt luulet varmaan tehneesi hyvin", hän sanoi. "Etkö
luule minun ymmärtävän, ettei siellä ulkona ketään ollut, peloitit vain
minua saadaksesi minut ottamaan lapsen." — "Voin viedä sen
uloskin, jos niin on", eukko sanoi. — "Maatkoon nyt tuossa niin
kauan, kunnes herää."
Illemmällä eukko taas tahtoi viedä pojan pois. Se oli silloin ääneti
ja hyvällä päällä ja aukoi ja sulki pikku käsiään. "Missä sinä pidät
häntä öisin?" Barbroo kysyi. — "Se makaa tuolla ulkona
heinäladossa." — "Annatko sinä hänen maata siellä kuin
kissanpoikasen?" — "En luullut tämän lapsen olosta niin suurta väliä
pidettävän. Mutta saa se olla sisällä tuvassakin, jos niin tahdot."
Kun poika oli kuusipäiväinen, istui Barbroo vuoteella katsellen,
miten eukko kapaloi sitä. "Sinähän pitelet sitä vallan kehnosti",
Barbroo sanoi, "ei ole ihmekään, että hän huutaa niin kovin." —
"Olen minä ennenkin lapsia hoitanut", eukko sanoi. "Luulen minä
ymmärtäväni ne asiat yhtä hyvin kuin sinäkin." Barbroo oli hetken
vaiti, mutta ajatteli itsekseen, ettei hän ollut ikinä nähnyt kenenkään
pitelevän lasta niin pahasti. — "Sinähän kolhit poikaa niin, että sen
kasvot ovat ihan mustansiniset", hän sanoi kärsimättömästi. —

"Pitäisikö muka mokomaa vaihdokasta vaalia kuin prinssiä ikään!"
sanoi eukko vihoissaan, "mutta kosk'en minä kelpaa, niin koeta itse."
— Ja tämän sanottuaan hän lennätti lapsen Barbroolle ja lähti ulos.
Barbroo otti lapsen vastaan. Hän kapaloi sen ja sai pojan pian
vaikenemaan ja rauhoittumaan. "Näetkös, että nyt se on vaiti", sanoi
hän ylpeän näköisenä, kun Liisa tuli takaisin. — "Aina toki ennen
minun lapsenhoitoani on kiitetty, minkä minä olen kuullut", sanoi
eukko uudelleen ja oli pitkän aikaa nyreillä nenin.
Tämän jälkeen Barbroo kuitenkin otti lapsen omaan hoitoonsa.
Eräänä päivänä, kun hän vielä oli makuulla, pyysi hän Liisalta
puhdasta kapaloriepua. Eukko vastasi, ettei hänellä ollut yhtään.
Kaikki olivat juuri pesussa. Barbroo punehtui, ja kyynelet tulvivat
hänen silmiinsä. "Tätä lasta hoidetaan kuin kerjäläisakan kakaraa",
hän sanoi pikaisuudessa. — "Sinun olisi itsesi pitänyt sitä hiukkasen
ajatella", sanoi eukko. "Mutta sanopa, millä ihmeen tavalla olisit
toimeen tullut, jollen minä olisi keräillyt kaikenlaisia vaatekappaleita
ja tuonut mukanani." Kaikki muistui taas Barbroon mieleen. Hänen
menneentalvinen synkkä epätoivonsa valtasi nyt hänet, niin että hän
uudestaan kävi ankaraksi. — "Parempi tämän lapsen olisi ollut
vaikkapa ilman mokomaa hoitoa", hän sanoi.
Seuraavana päivänä Barbroo nousi jalkeille. Hän otti esiin neulan
ja lankaa ja alkoi leikellä lakanaa kappaleiksi ommellakseen siitä
vaatteita pojalle. Hetken aikaa sitä tehtyään hän joutui uudelleen
synkkiin ajatuksiinsa. Eipä oikein kannata ruveta sille tällaisia
laittelemaan. Parasta olisi, kun heittäytyisin suohon sen kanssa;
sinne meidän molempain kumminkin on mentävä.
Hän lähti ulos katsomaan Vanhaa-Liisaa, joka lypsi lehmiä, ennen
kuin niitä ajettiin metsään. "Tiedätkö sinä, Liisa, kuinka pitkän ajan

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