Experience Seamless Full Ebook Downloads for Every Genre at textbookfull.

com

Statistical Mechanics of Classical and Disordered

Systems Luminy France August 2018 Véronique
Gayrard

https://textbookfull.com/product/statistical-mechanics-of-
classical-and-disordered-systems-luminy-france-
august-2018-veronique-gayrard/

OR CLICK BUTTON

DOWNLOAD NOW

Explore and download more ebook at https://textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

Classical Statistical Mechanics with Nested Sampling 1st

Edition Robert John Nicholas Baldock (Auth.)

https://textbookfull.com/product/classical-statistical-mechanics-with-
nested-sampling-1st-edition-robert-john-nicholas-baldock-auth/

textboxfull.com

Foundations of Classical Mechanics P. C. Deshmukh

https://textbookfull.com/product/foundations-of-classical-mechanics-p-
c-deshmukh/

textboxfull.com

Non equilibrium statistical mechanics Prigogine

https://textbookfull.com/product/non-equilibrium-statistical-
mechanics-prigogine/

textboxfull.com

Statistical Physics Michael V. Sadovskii

https://textbookfull.com/product/statistical-physics-michael-v-
sadovskii/

textboxfull.com
Advanced classical mechanics 1st Edition Bagchi

https://textbookfull.com/product/advanced-classical-mechanics-1st-
edition-bagchi/

textboxfull.com

Structural Syntactic and Statistical Pattern Recognition

Joint IAPR International Workshop S SSPR 2018 Beijing
China August 17 19 2018 Proceedings Xiao Bai
https://textbookfull.com/product/structural-syntactic-and-statistical-
pattern-recognition-joint-iapr-international-workshop-s-
sspr-2018-beijing-china-august-17-19-2018-proceedings-xiao-bai/
textboxfull.com

A modern approach to classical mechanics Iro

https://textbookfull.com/product/a-modern-approach-to-classical-
mechanics-iro/

textboxfull.com

Variational principles in classical mechanics Cline D.

https://textbookfull.com/product/variational-principles-in-classical-
mechanics-cline-d/

textboxfull.com

Renormalization Group Analysis of Nonequilibrium Phase

Transitions in Driven Disordered Systems Taiki Haga

https://textbookfull.com/product/renormalization-group-analysis-of-
nonequilibrium-phase-transitions-in-driven-disordered-systems-taiki-
haga/
textboxfull.com
Springer Proceedings in Mathematics & Statistics

Véronique Gayrard
Louis-Pierre Arguin
Nicola Kistler
Irina Kourkova Editors

Statistical
Mechanics of
Classical and
Disordered Systems
Luminy, France, August 2018
Springer Proceedings in Mathematics &
Statistics

Volume 293
Springer Proceedings in Mathematics & Statistics

This book series features volumes composed of selected contributions from

workshops and conferences in all areas of current research in mathematics and
statistics, including operation research and optimization. In addition to an overall
evaluation of the interest, scientific quality, and timeliness of each proposal at the
hands of the publisher, individual contributions are all refereed to the high quality
standards of leading journals in the field. Thus, this series provides the research
community with well-edited, authoritative reports on developments in the most
exciting areas of mathematical and statistical research today.

More information about this series at http://www.springer.com/series/10533

Véronique Gayrard Louis-Pierre Arguin
• •

Nicola Kistler Irina Kourkova

•

Editors

Statistical Mechanics
of Classical and Disordered
Systems
Luminy, France, August 2018

123
Editors
Véronique Gayrard Louis-Pierre Arguin
Aix-Marseille Universite CNRS Baruch College, CUNY
Institut de Mathematiques de Marseille New York, NY, USA
Marseille, France
Irina Kourkova
Nicola Kistler Laboratoire de Probabilités
Goethe-Universität Frankfurt Pierre and Marie Curie University VI
Frankfurt am Main, Germany Paris, France

ISSN 2194-1009 ISSN 2194-1017 (electronic)

Springer Proceedings in Mathematics & Statistics
ISBN 978-3-030-29076-4 ISBN 978-3-030-29077-1 (eBook)
https://doi.org/10.1007/978-3-030-29077-1
Mathematics Subject Classification (2010): 82-01, 82-02, 82-06, 82B03, 82B05, 82B26, 82B31, 82B41,
82B43, 82B44, 82B80, 82C44, 60F10, 60G07, 60G15, 60G70

© Springer Nature Switzerland AG 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Statistical mechanics aims at understanding the behavior of systems involving very

many interacting components through so-called statistical ensembles, namely,
probability measures over all possible states of the system. Initiated in the late
1800s by Boltzmann and Maxwell, who were motivated by deriving the principles
of thermodynamics of gases from the statistical properties of molecules, the
framework of statistical mechanics was mathematically formalized by Gibbs a few
decades later. With time, statistical mechanics has grown into an extensive and
ever-evolving body of knowledge that encompasses several areas of physics and
probability theory, with deep and far-reaching links with areas as diverse as biology
and computer science. Virtually any area of research dealing with a large number of
agents displaying collective behavior, whether these be molecules, living organisms
or nodes of a network, can benefit from its formalism. The contributions to this
volume, originating from the international conference Advances in Statistical
Mechanics which took place at the CIRM in Luminy in August 2018, nicely reflect,
we believe, the current state of affairs of such a success story.
In fact, all research articles appearing in this volume deal, in one way or another,
with the many facets of modern statistical mechanics, and can be (loosely) grouped
in two major sub-fields: those pertaining to the somewhat classical statistical
mechanics of ordered systems, and those pertaining to the more recent field of
disordered systems.
The paper by Külske et al. and the one by van Enter fall in the first category. The
former addresses, under the lenses of the Widom-Rowlinson model, the “Gibbs vs
non-Gibbs transition”; the latter shows how models from statistical mechanics can
shed light on the delicate issue whether (and to which extent) Markov chains and
Markov fields lead to conceptually equivalent objects. Both issues are currently the
subject of intensive research activities.
The majority of contributions fall into the second category, that of disordered
systems. A further categorization into sub-groups is here possible, as one may
distinguish between “statics vs. dynamics”, i.e., between properties of systems at
equilibrium, as opposed to systems which are yet to relax. Arguin-Persechino study
via large deviations techniques at the level of entropy and free energy the extremes

v
vi Preface

of a GREM in random magnetic field. The paper by Bolthausen provides a new

proof of the replica symmetric solution of the SK-model which relies on a
Morita-type argument, and the TAP equations. Guerra addresses the highly efficient
but mathematically puzzling replica trick, recasting it into the framework of
interpolations. Kersting et al. show how the (R)SB-Parisi solution of the SK-model
emerges via high- temperature expansions from ad hoc Gibbs potentials in finite
volume. The paper by Cerny improves our understanding of the Metropolis
dynamics of Derrida’s REM, insofar it shows that a deterministic normalization
suffices for the convergence of the clock process. Gayrard-Hartung also address
dynamical properties of the REM, and provide a comprehensive analysis of the
phase diagram in the case of random hopping dynamics. The contribution by Wang
et al. addresses the predictability of the zero-temperature Glauber dynamics of Ising
ferromagnets in case of light- or heavy-tailed distributions.
The remaining two contributions are somewhat challenging to characterize in
terms of the aforementioned dichotomy “classical vs. disordered”, but they are
nevertheless deeply rooted in statistical mechanics. The paper by Faggionato pro-
vides a concise review of results establishing large deviation principles and
Gallavotti-Cohen dualities for Markov chains, and which thus provide insights into
the thermodynamics of (bio)molecular motors/pumps. Finally, the paper by Caputo
et al. deals with line ensembles for Brownian polymers: this is a field of probability
theory, which however bears strong connections with the statistical mechanics of
SOS-models.
The international conference, which eventually led to the publication of these
proceedings, would have never been possible without the (financial, logistic and
organizational) support of the CIRM in Luminy, the city of Marseille, the Aix
Marseille Université, the Institut de Mathématiques-Labex Archimède-CARMIN-
FRUMAM (Marseille), the LYSM-LPSM-Groupe de Mathématiques de
l'Aléatoire-Groupe de Modélisation Mathématique (Paris), the CNRS, the NSF, the
DFG, the University of Frankfurt, and the Bonn International Graduate School of
Mathematics. We express our heartfelt gratitude to these institutions, and to the
affiliated people who helped us all along this journey.
The meeting at the CIRM has also been a wonderful opportunity to celebrate the
60th birthday of our friend and colleague Anton Bovier. The works appearing in
these proceedings touch upon a tiny fraction only of his broad scientific interests,
but it seems fair to say that statistical mechanics at large has played a major role in
his distinguished career: we thus dedicate this volume to him, with the best wishes
for many happy returns.

New York, USA Louis-Pierre Arguin

Marseille, France Véronique Gayrard
Frankfurt am Main, Germany Nicola Kistler
Paris, France Irina Kourkova
Contents

Ordered Systems
Gibbs-Non Gibbs Transitions in Different Geometries:
The Widom-Rowlinson Model Under Stochastic Spin-Flip
Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Christof Külske
One-Sided Versus Two-Sided Stochastic Descriptions . . . . . . . . . . . . . . 21
Aernout C. D. van Enter

Disordered Systems
The Free Energy of the GREM with Random Magnetic Field . . . . . . . . 37
Louis-Pierre Arguin and Roberto Persechino
A Morita Type Proof of the Replica-Symmetric Formula for SK . . . . . . 63
Erwin Bolthausen
Concentration of the Clock Process Normalisation
for the Metropolis Dynamics of the REM . . . . . . . . . . . . . . . . . . . . . . . 95
Jiří Černý
Dynamic Phase Diagram of the REM . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Véronique Gayrard and Lisa Hartung
The Replica Trick in the Frame of Replica Interpolation . . . . . . . . . . . . 171
Francesco Guerra
From Parisi to Boltzmann . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Goetz Kersting, Nicola Kistler, Adrien Schertzer and Marius A. Schmidt

vii
viii Contents

Nature Versus Nurture: Dynamical Evolution in Disordered

Ising Ferromagnets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Lily Z. Wang, Reza Gheissari, Charles M. Newman and Daniel L. Stein

Miscellaneous
Tightness and Line Ensembles for Brownian Polymers
Under Geometric Area Tilts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Pietro Caputo, Dmitry Ioffe and Vitali Wachtel
Large Deviations and Uncertainty Relations in Periodically Driven
Markov Chains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
Alessandra Faggionato
Contributors

Louis-Pierre Arguin Department of Mathematics, Baruch College and Graduate

Center, City University of New York, New York, NY, USA
Erwin Bolthausen University of Zurich, Zurich, Switzerland
Pietro Caputo Dipartimento di Matematica e Fisica, Roma Tre University, Rome,
Italy
Jiří Černý Department of Mathematics and Computer Science, University of
Basel, Basel, Switzerland
Alessandra Faggionato Dipartimento di Matematica, Università di Roma ‘La
Sapienza’, Roma, Italy
Véronique Gayrard I2M, CNRS Centrale Marseille, Aix Marseille University,
Marseille, France
Reza Gheissari Courant Institute, New York University, New York, NY, USA
Francesco Guerra Dipartimento di Fisica, Istituto Nazionale di Fisica Nucleare,
Sezione di Roma, Università di Roma “La Sapienza”, Roma, Italy
Lisa Hartung Institut für Mathematik, Johannes Gutenberg-Universität Mainz,
Mainz, Germany
Dmitry Ioffe The Faculty of Industrial Engineering and Management, Technion,
Haifa, Israel
Goetz Kersting J.W. Goethe-Universität, Frankfurt, Germany
Nicola Kistler J.W. Goethe-Universität, Frankfurt, Germany
Christof Külske Fakultät für Mathematik, Ruhr-University of Bochum, Bochum,
Germany

ix
x Contributors

Charles M. Newman Courant Institute, New York University, New York, NY,
USA;
NYU-ECNU Institute of Mathematical Sciences at NYU Shanghai, Shanghai,
China
Roberto Persechino Service des Enseignements Généraux, École de Technologie
Supérieure, Montréal, QC, Canada
Adrien Schertzer J.W. Goethe-Universität, Frankfurt, Germany
Marius A. Schmidt University of Basel, Basel, Switzerland
Daniel L. Stein Department of Physics and Courant Institute, New York
University, New York, NY, USA;
NYU-ECNU Institutes of Physics and Mathematical Sciences at NYU Shanghai,
Shanghai, China;
Santa Fe Institute, Santa Fe, NM, USA
Aernout C. D. van Enter Bernoulli Institute, University of Groningen,
Groningen, The Netherlands
Vitali Wachtel Institut für Mathematik, Universität Augsburg, Augsburg,
Germany
Lily Z. Wang Center for Applied Mathematics, Cornell University, Ithaca, NY,
USA
Ordered Systems
Gibbs-Non Gibbs Transitions in Different
Geometries: The Widom-Rowlinson
Model Under Stochastic Spin-Flip
Dynamics

Christof Külske

Abstract The Widom-Rowlinson model is an equilibrium model for point particles

in Euclidean space. It has a repulsive interaction between particles of different colors,
and shows a phase transition at high intensity. Natural versions of the model can
moreover be formulated in different geometries: in particular as a lattice system or
a mean-field system. We will discuss recent results on dynamical Gibbs-non Gibbs
transitions in this context. Main issues will be the possibility or impossibility of
an immediate loss of the Gibbs property, and of full-measure discontinuities of the
time-evolved models.

Keywords Gibbs measures · Stochastic time-evolution · Gibbs-non Gibbs

transitions · Widom-Rowlinson model

AMS 2000 subject classification 60K57 · 82B24 · 82B44

1 Introduction

Recent years have seen a variety of studies of Gibbs-non Gibbs transitions of mea-
sures which appear as image measures of Gibbs measures, under certain local trans-
formation rules. What is a Gibbs measure? There is a well-defined theory to define
Gibbs measures on lattices, where the probability space is given by the set of all func-
tions from lattice sites to a finite alphabet. The central object is that of a specification
[2, 15]. For other geometries, other but related approaches are adequate, see below.
The unifying idea is that Gibbs measures are measures whose conditional probabil-
ities to see a single symbol at a given site, are nice (continuous) functions of their
conditioning, see below. It has been discovered that Gibbs measures under natural

Collaborations with Benedikt Jahnel, Sascha Kissel, Utkir Rozikov.

C. Külske (B)
Fakultät für Mathematik, Ruhr-University of Bochum, Postfach 102148, 44721 Bochum,
Germany
e-mail: [email protected]
© Springer Nature Switzerland AG 2019 3
V. Gayrard et al. (eds.), Statistical Mechanics of Classical and Disordered Systems,
Springer Proceedings in Mathematics & Statistics 293,
https://doi.org/10.1007/978-3-030-29077-1_1
4 C. Külske

deterministic or stochastic transformations may lead to non-Gibbsian measures [10].

More specifically the study of stochastic time evolutions, even very simple ones,
applied to Gibbs measures, has shown very interesting transition phenomena, the
most prototypical example for this is the Glauber-evolved Ising model in [8]. Indeed,
stochastic time evolutions may destroy the Gibbs property of the image measure at
certain transition times, a phenomenon we call dynamical Gibbs-non Gibbs transi-
tions, see below. The purpose of this note is to take the Widom-Rowlinson model
[39] and variations thereof as a guiding example, apply an independent symmetric
stochastic spin flip dynamics to it, and describe our findings of what may and what
may not happen along the time-evolved trajectory of measures. We treat and com-
pare a hard-core version and a soft-core version of the model in various geometries,
namely in Euclidean space, on the lattice, as a mean-field model, and on a regular
tree. Our aim here is to provide an overview; for detailed statements and proofs we
refer to the original articles.

2 Gibbs on Lattice, Sequentially Gibbs, Marked Gibbs

Point Processes, and the Widom-Rowlinson Model

We start by recalling the notion of an infinite-volume Gibbs measure for lattice

systems. For the purpose of the discussion of the Widom-Rowlinson model and all
measures appearing under time-evolution defined below from it, it is sufficient to
restrict to the local state-space {−1, 0, 1} for particles carrying spins plus or minus,
and holes. Our site space is the lattice Zd . The space of infinite-volume configurations
is  = {−1, 0, 1}Z .
d

2.1 Specifications and Gibbs Measures on the Lattice

The central object in Gibbsian theory on a countable site space which defines the
model is a specification. This covers both cases of infinite lattices and trees. It is a
candidate system for conditional probabilities of an infinite-volume Gibbs measure
μ (probability measure on ) to be defined by DLR equations μ(γ ( f |·)) = μ( f ).
A specification γ is by definition a family of probability kernels γ = (γ )Zd ,
indexed by finite subvolumes , where γ (dω|η) is a probability measure on ,
for each fixed configuration η. It must have the following properties. The first is the
consistency which means that

γ (γ (dω|·)|ω̃) = γ (dω|ω̃) (1)

for all finite volumes  ⊂   Zd . It is suggested by the tower property of condi-

tional expectations.
Gibbs-Non Gibbs Transitions in Different Geometries … 5

The second is the Fc -measurability of γ ( f |·), for any bounded measurable
observable f . Here the sigma-algebra Fc is generated by the spin-variables outside
of the finite volume .
The last property is the properness γ (1 A |·) = 1 A for A ∈ Fc . It means that
the randomization of the kernel takes place only inside of , and an event which
is determined by what is outside of  will indeed be determined by looking at the
boundary condition alone.
An important additional regularity requirement is quasilocality of the specification
which means that the function ω → γ ( f |ω) should be quasilocal for f quasilocal,
and this has to hold for all finite volumes . A quasilocal function is a uniform limit of
local functions, that is of functions which depend only on finitely many coordinates.
More specifically a Gibbsian specification on the infinite-volume state space  =
{−1, 0, 1}Z for an interaction potential  = ( A ) AZd and a priori measure α ∈
d

M1 ({−1, 0, 1}) by definition has probability kernels

1  
γ,,α (ω |ωc ) := e− A∩=∅  A (ω) α(ωi ) (2)
Z  (ωc ) i∈

where Z  (ωc ) is the normalizing partition function. If  is a finite-range potential

(meaning that  A is only nonzero for finitely many A’s), obviously all sums are finite,
when we insist that  A takes finite only values. Finiteness of the sums also holds, if
 is uniformly absolutely convergent. For hard-core models the specification kernels
acquire an indicator, see the example (4) below.
The first statistical mechanics task in this setup is the following. Given a specifi-
cation γ = (γ )Zd in the above sense, find the corresponding Gibbs measures

G(γ) := {μ ∈ M1 (), μγ = μ, for all   Zd } (3)

In general G(γ) may be empty, contain precisely one measure, or more than one
measures. If |G(γ)| > 1 we say that the specification γ has a phase transition. The
Gibbs measures G(γ) form a simplex, meaning that each measure has a unique
decomposition over the extremal elements, called pure states. Pure states can be
recovered as finite-volume limits with fixed boundary conditions.
Existence and extremal decomposition of proper infinite-volume measures becomes
even more involved for systems with random potentials. In general, for systems like
spin-glasses, the construction of infinite-volume states by non-random sequences of
volumes which exhaust the whole lattice is problematic, and for such systems the
higher-level notion of a metastate (a measure on infinite-volume Gibbs measures) is
useful [1, 2, 5, 27, 34].
6 C. Külske

2.2 Hard-Core and Soft-Core Widom-Rowlinson Model on

Lattice and in Mean-Field

We will consider here the version of the hard-core Widom-Rowlinson model on Zd

as in [17]. It has the a priori measure α ∈ M1 ({−1, 0, 1}) as its only parameter. Its
specification kernels are given by

1 
γ,α
hc
(ω |ωc ) := I hc
 (ω  ω  c) α(ωi ), (4)
hc
Z (ωc ) i∈


where the hard-core indicator Ihc (ω) = i∈ I(ωi ω j =−1, ∀ j∼i) forbids +− neighbors
to occur with positive probability. Related hard-core models have been studied on
lattices and trees, see for example [14, 32, 35].
The soft-core Widom-Rowlinson model on Zd has an additional repulsion param-
eter β > 0. In the specification kernels, which are by definition given by

I(ωi ω j =−1) 

1 −β
γ,β,α
sc
(ω |ωc ) := α(ωi ),
b
{i, j}∈E
e (5)
sc
Z (ωc ) i∈

configurations with +− neighbors are suppressed, but not forbidden.

These definitions of a specification immediately extend to any graph with count-
ably infinite vertex set, where each vertex has a finite number of nearest neighbors.
In particular we may study this model on a regular tree with k + 1 neighbors, see
[23].
The mean-field formulation is different, as the model is defined as a whole
sequence of finite-volume Gibbs measures, indexed by the system size N ∈ N. The
elements in the sequence for the Mean-Field soft-core Widom-Rowlinson model
with repulsion parameter β > 0 are the measures

1 β  
N
μ N ,β,α (ω[1,N ] ) := e− 2N 1≤i, j≤N I(ωi ω j =−1)
α(ω j ) (6)
Z N ,β,α j=1

for ω[1,N ] = (ωi )i=1,...,N ∈ {−1, 0, 1} N . For more details see [22].

2.3 Sequential Gibbsianness for Mean-Field (and

Kac-Models on Torus)

There is an intrinsic formulation of the Gibbs property which is suitable also in

situations where a finite-volume Hamiltonian can not be read off directly from the
explicit definition of the measures. It focusses on conditional probabilities instead,
suggested by analogy to the lattice situation [18, 28].
Gibbs-Non Gibbs Transitions in Different Geometries … 7

Take (μ N ) N ∈N a sequence of exchangeable probability measures μ N on the finite-

volume state space {−1, 0, 1} N . The large N -behavior of such a sequence defines our
model. The model is called sequentially Gibbs iff the volume-limit of the single-spin
probabilities in the finite-volume measures

lim μ N (dω1 |ω[2,N ] ) = γ(dω1 |ν) (7)

N ↑∞

exists whenever the empirical distributions of a configuration (ωi )i=2,3,4,... converge,

1 
N
δω → ν. (8)
N − 1 i=2 i

This has to hold for all limiting empirical distributions of conditionings ν ∈

M1 ({−1, 0, 1}).
If there is some ν for which it is possible to obtain different limits for different
boundary conditions (ωi )i=2,3,... , and (ω̄i )i=2,3,... we call this ν a bad empirical mea-
sure. A model fails to be sequentially Gibbs, if there is at least one bad empirical
measure.
As a general consequence of this definition, the sequential Gibbs property implies
that the limiting kernel ν → γ(dω1 |ν) is continuous. In our Widom-Rowlinson case
where ν takes values in a finite-dimensional simplex, all topologies are the same, and
equal to the Euclidean topology. Clearly the mean-field Widom-Rowlinson model
defined above, is sequentially Gibbs.
A similar notion of the sequential Gibbs property can be extended to cover Kac-
models on the torus, and transformed Kac-models which have the same index set.
These models are again described by sequences whose asymptotics one wants to
capture, but they have a spatial structure. As in the mean-field models, there is
again a single-site limiting kernel, however the limiting empirical distribution ν
which appeared as a conditioning in the mean-field model is replaced by a whole
profile of spin densities on the unit torus. For details of these definitions and results,
see [16, 18].

2.4 Marked Gibbs Point Processes in Euclidean Space

Here the good definition of Gibbs measure is in some analogy to the lattice situation
[6, 20, 37]. We restrict again for the sake of our exposition to the specific simple
mark space which covers the Widom-Rowlinson model of point particles in Euclidean
space, and the time-evolved version we will discuss below. In this case the mark space
is {−1, 1}. It does not contain zero. The spatial degrees of freedom are described by
the set  of locally finite subsets of Rd . A marked particle configuration is a pair ω =
(ω − , ω + ) describing the positions of minus-particles (and plus-particles respectively)
where each ω − , ω + ∈ . The configuration space of such marked configurations is
8 C. Külske

. For the measurable structure we need the σ-algebras F, F . These are the σ-
algebras for marked particles generated by the counting variables. They count the
number of plus- and minus particles in Borel sets A in the whole Euclidean space
(or all such sets A ∈  respectively, where  may be any measurable subset of
Euclidean space).
A specification shall become, as on countable graphs, a candidate system for
conditional probabilities of Gibbs measure μ to be defined by DLR equations μγ =
μ for all measurable bounded subsets  in Euclidean space. Hence, one defines a
specification to be a family of proper probability kernels γ = (γ )Rd with the
properties of consistency, that is γ γ = γ for all measurable volumes  ⊂  
Rd . One also needs Fc -measurability of γ ( f |·), for any bounded test observable
f . Properness means here that γ (1 A |·) = 1 A for A ∈ Fc .
We will further assume quasilocality of the specification. This means the compat-
ibility of the kernels γ with the local topology on the space of marked point clouds.
In this topology convergence for a sequence of marked particle clouds means that
the clouds must become constant in each bounded volume.

2.5 Widom-Rowlinson Model in Euclidean Space

We assume spatial dimension d ≥ 2, and fix the two-color local spin space (mark
space) of {−, +}. The model will be obtained as a modification of the base measure
P by which we denote a two-color homogenous Poisson Point Process in the infinite
volume, with intensities λ+ for plus colors and λ− for minus colors.
The (hard-core) Widom-Rowlinson specification is the Poisson-modification with
the specification kernels

1
γ (dω  |ω c ) := χ(ω  ω c )P  (dω  ) (9)
Z  (ω c )

where the indicator χ is one iff the interspecies distance (the distance between points
of different sign) is bigger or equal than 2a, and P  (dω  ) denotes the two-color
Poisson process in the bounded volume . The picture shows a typical configuration
at large λ+ = λ− , in a finite volume.

This picture survives the thermodynamic limit: By results of [3, 4, 36] it is known
that in d ≥ 2, λ+ = λ− large, the continuum WiRo has a phase transition, and how
this is related to percolation of large clusters of overlapping disks.
Gibbs-Non Gibbs Transitions in Different Geometries … 9

In more general models of marked point particles, specifications which are Poisson
modifications may be obtained in terms of exponential factors with finite-volume
Hamiltonians which are formed with potentials. For such potentials one may consider
multibody potentials which are known from statistical physics, but we may also allow
for hyperedge potentials and define

1 
γ (dω  |ω c ) := e− ηω: η∩=∅ (η,ω) P  (dω  ) (10)
Z  (ω c )

A hyperedge potential (η, ω) is by definition allowed to depend on the marked point

cloud ω not only via the particle positions and marks on hyperedge η (which is just a
finite subset of points in the cloud), but also on a whole neighborhood of η, up to some
horizon. This generalization is useful in models of stochastic geometry, involving
e.g. energies depending on the cells of a Voronoi tessellation. Hyperedge potentials
were successfully used in [6] where a general existence theory of infinite-volume
Gibbs measures is developed. We also refer to [21] for representation theorems.
There is it shown how one can go from a measure μ under continuity assumptions of
finite-volume conditional probabilities, to a hyperedge potential . These theorems
are a generalization of Kozlov-Sullivan theorems [25, 38] known on the lattice to
the continuum, and make use in a constructive way of the weak nonlocality allowed
by the hyperedge potential concept.

3 Dynamical Gibbs-Non Gibbs Transitions

We now come to time-evolutions. Consider again the Euclidean space Widom-

Rowlinson model, and fix some cloud of particles carrying the marks plus or minus.
We define a continuous-time stochastic dynamics by the following rule. Particle
locations stay fixed, holes stay fixed. The signs of the particles however change
stochastically, independently of each other, according to a temporal Poisson process
for each particle with rate one. In this way, at every particle location, the probability
to go from + to − in time t is given by

1
pt (+, −) = pt (−, +) = (1 − e−2t ). (11)
2
Starting with the same signed particle configuration shown above, after a small time
t, a fraction of the particles has kept their signs, as shown in the picture (or flipped
back). Of course, there is loss of memory in each fixed bounded volume, which is
exponentially fast in the time t. Interesting things however happen if we consider the
infinite volume, as we will discuss.
10 C. Külske

We will apply the same stochastic dynamics also on the spatially discrete model on
the lattice, and also to the mean-field model, at each finite system size N .
We want to understand better the structural Gibbsian properties of the measures
along a trajectory given by the time-evolution. For the purpose of concreteness we
focus on the Euclidean model. We say the model shows a dynamical Gibbs-non Gibbs
transition if the initial measure μ is Gibbs for a quasilocal specification, and for some
time t the time-evolved measure μt = μPt is not compatible with any quasilocal
specification. Here Pt is the semigroup giving the distribution to find an infinite-
volume configuration after time t when starting with a given initial configuration,
which is integrated over with respect to the starting measure μ. In our example
above Pt is the symmetric independent spin-flip dynamics, and does not involve a
randomization of the spatial degrees of freedom. However it is clear that one would
like to study more generally also dependent dynamics, and also possibly irreversible
dynamics, compare [19, 31]. Such studies have been performed at first for the Ising
model on the lattice, for work on this and related work see [8, 9, 11–13, 26, 28, 30].

3.1 Relation to Disordered Systems

To fix ideas, let us go to the lattice setup. That a time-evolved lattice measure μt (in
our the lattice Widom-Rowlinson model under symmetric spin-flip) is non-Gibbs is
indicated by very long-range dependencies in its conditional probabilities, that is

η → μt (ηi |ηZd \i ) (12)

behaves discontinuously w.r.t. the local topology. More precisely, as the r.h.s. is
only defined up to measure-zero sets, this means that there is no version which is
continuous.
A useful strategy (at least for independent dynamics) is the following. Consider
the two-layer measure, that is the joint distribution of spins and time zero and at in
the future at time t > 0 given by

μ̄t (dω, dη) = μ(dω)Pt (ω, dη). (13)

Analyze hidden phase transitions in first-layer measure constrained on the future

configuration η. By this we mean the measure
Gibbs-Non Gibbs Transitions in Different Geometries … 11

μ̄t (dω|η). (14)

A relation to disordered systems with quenched order appears when we view

the configuration η = (ηi )i∈Zd in the role of a quenched disorder configuration. By
playing with (suitable finite-volume approximations of) conditional probabilities a
picture emerges in which absence of phase transitions in the first-layer model implies
Gibbsian behavior of the time-evolved model. The opposite implication, that the
presence of a phase transition in the first-layer model implies the absence of the
Gibbs property of the time-evolved model is true in many examples, and proved for
a specific class of mean-field systems.
Methods which are different from the two-layer picture are used for depen-
dent dynamics. For mean-field systems and Kac-systems there are also path-large-
deviation principles available which lead to fixed-end-point variational problems for
trajectories of empirical measures. While in an abstract sense this is a solution, the
analytical understanding of the structure of minimizers of such problems can be quite
hard (see however [16]). It is an open challenge to fully develop the analogous theory
on the lattice, with ideas as suggested in [9].

3.2 Results on Dynamical GnG Transitions for Euclidean

Model

A marked infinite-volume configuration ω ∈  (that is a signed point cloud) is called

good for a specification γ iff for any Euclidean ball B we have
 
γ B ( f |ω  c ) − γ B ( f |ω c ) → 0
B B

as ω  ⇒ ω in the sense of local convergence.

We denote by (γ) the set of good configurations for the specification γ. We say
that a specification γ is called quasilocal iff (γ) = , that is if all point clouds are
good.
A measure on signed point clouds μ is called ql (quasilocally Gibbs) iff there exists
a specification γ such that (γ) = . μ is called asql (almost surely quasilocally
Gibbs) iff there exists γ for which at least μ((γ)) = 1, that is the good points are
a full-measure set.
Let us now describe the results on Gibbsian transitions in time and intensity
for μ+ obtained in [20]. The measure μ+ is the measure in the Euclidean Widom-
Rowlinson model obtained as an infinite-volume limit of finite-volume measures
with the maximal boundary condition of overlapping plus-discs. By FKG (stochastic
monotonicity) arguments this measure exists for all choices of parameters. We define

1 λ+ + λ−
tG := log (15)
2 λ+ − λ−
12 C. Külske

for λ+ > λ− . It will serve as a reentrance time into the Gibbsian region. We say
that the model with intensity parameters λ+ , λ− is in the high-intensity (percolating)
regime iff μ+ (B ↔ ∞) > 0 for some ball B (there is positive probability that there
is an infinite cluster of overlapping discs containing B). Then the behavior of the
time-evolved measure μ+ t is summarized in the following table.

λ+ > λ− λ+ = λ−
time 0 < t < tG t = tG tG < t ≤ ∞ 0<t≤∞
high intensity non-asql asql, non-ql ql non-asql
low intensity asq, non-ql asql, non-ql ql asql, non-ql

Main striking features are the immediate loss of the Gibbs property, and the
appearance of full-measure sets of bad configurations (discontinuity points of any
specification). More precise statements and detailed proofs can be found in [20]. The
proofs use cluster representations for conditional probabilities of the time-evolved
measure.
The appearance of typical bad configurations can be heuristically understood: Infi-
nite clusters in the time-zero model, together with the requirement that overlapping
disks have the same sign, gives a strong rigidity in the first-layer model constrained
on the future configuration η at time t. Indeed, conditional on fixed locations in
the percolating cluster, this cluster can only carry uniform plus signs, or unifom
minus signs, at time zero. Keeping locations in a conditioning η fixed and varying
the signs arbitrarily far away provides then a very effective mechanism to induce a
phase transition in the first-layer model. One shows that this implies jumps in certain
conditional probabilities at time t. Hence every percolating point cloud can be a bad
configuration, in the appropriate parameter regimes. As percolation is typical, this
implies full-measure sets of bad configurations.

3.3 Results on Dynamical GnG Transitions for the

Mean-Field Widom-Rowlinson Model

Before we come to dynamics, we need to describe the equilibrium behavior of the

mean-field model. The pressure of the mean-field model can be computed using large-
deviation techniques (Varadhan’s lemma) in terms of a variational formula where
extremal points need to be found in the space of single-site probability distributions.
Denoting by L 1N the fraction of spins with spin 1 at system size N , and using similar
notations for the other spin values, we have
Gibbs-Non Gibbs Transitions in Different Geometries … 13

 
N
1 −1
e−N β L N (ω)L N (ω)
1
p(β, α) = lim log α(dω j )
N →∞ N j=1

= sup (−βν(1)ν(−1) − I (ν|α))

ν∈M1 ({−1,0,1})

where I (ν|α) is the relative entropy of a single-site distribution ν w.r.t. the a priori
measure α. Correspondingly, the empirical distribution satisfies a large deviation
principle with speed N , and rate function given by the negative of the expression
under the sup plus a suitable constant.
A discussion of the variational problem (see [22]) shows: The symmetric model
at any α(1) = α(−1) > 0 has a second-order phase transition driven by repulsion
strength β > 0 at critical repulsion strength given by βc = 2 + e α(0)
α(1)
.
An explicit solution for the mean-field Widom-Rowlinson model at time t = 0
can be obtained as follows. We parametrize the empirical spin distribution ν via
coordinates (x, m) where x plays the role of occupation density, and m plays the role
of the magnetization on the occupied sites, writing
 x x
ν(−1), ν(0), ν(1) = ( (1 − m), 1 − x, (1 + m)) (16)
2 2
Next we parametrize the a priori measure α via coordinates (h, l) where h =
α(1)
1
2
log α(−1) plays the role of an external magnetic field, and l := log 1−α(0)
α(0)
describes a bias on the occupation variables. Using these coordinates, the pressure
can be written as

βx 2
p(β, α) = log α(0) + sup − + x(l − log(2 cosh(h)) − J (x)
0≤x,|m|≤1 4  
part for occupation density
βxm 2
+x + hm − I (m) − log 2
4  
Ising part at occupation-dependent temperature

with an entropy for spins I (m) = 1−m 2

log( 1−m
2
) + 1+m
2
log( 1+m
2
) and an entropy
for occupation variables J (x) = (1 − x) log(3(1 − x)) + x log( 3x2 ). It turns out that
the symmetric antiferromagnetic model (β < 0) has a first-order transition when
crossing a line in β, α(0)-space, where jumps occur in occupation density x, at fixed
zero magnetization m = 0.
We are mostly interested in the ferromagnetic model. In this situation we obtain
that repulsion parameter β > 0, a priori measure α = α(h, l), and typical values
(m, x) of the empirical distribution, are related via the parametrization
14 C. Külske

2   
(I (m) − h)(1 + e−l+log(2 cosh(h))+ m (I (m)−h)−m I (m)+I (m) )
1
β = β(m; α) =
m
 
x = x(m; α) = (1 + e−l+log(2 cosh(h))+ m (I (m)−h)−m I (m)+I (m) )−1
1

We remark as a corollary that the model has mean-field critical exponents: Fix any
α(0) ∈ (0, 1). Let βc be the corresponding critical value for the symmetric model.
Then there are positive constants such that

m(β, h = 0) m(βc , h)
lim 1 = c, lim 1 = c
β↓βc (β − βc ) 2 h↓0 h3

The main point is the study of the dynamical Gibbs-non Gibbs transitions under rate-
one symmetric independent spin-flip, keeping holes fixed, according to transition
probabilities (11).
Recall the notion of sequentially Gibbs and the notion of bad empirical measure,
see (7), (8) and the text below. For the sake of this review let us just present the time
evolution of the set of bad empirical measures in the regime of an inverse temperature
of the time-zero model in the region of β > 3 in a plot (compare [22] for the full
statement of the theorem describing all dynamical transitions). In the plot the inverse
temperature of the time zero model is β = 5 and we are starting from a symmetric
model with α(+) = α(−).
Here time increases from the top left to the top right, then from bottom left
to bottom right. The main features are the following: There is a short-time Gibbs
regime for all β, α. Small repulsion strength β ≤ 2 implies the Gibbs property of the
time-evolved model for all times t. The set of bad empirical measures at given β has
dimension one as a subset of the simplex, in the interior of its existence time-interval.
It can be a union of disconnected curves, a branching curve (which has a Y-shape,
see picture), or a line which is growing with time (growing antenna).

These features can be understood by a bifurcation analysis of a rate function

describing the first-layer model constrained on an empirical distribution. For our
model, this analysis is closely related to that of a time-evolved Ising model in the
Gibbs-Non Gibbs Transitions in Different Geometries … 15

following way. The bad measures α f in the time-evolved mean-field WiRo model
after time t satisfy for any symmetric a priori measure α
  
α f (1) − α f (−1) βα f ({1, −1})
B W i Ro (β, t) = α f ∈ M1 ({−1, 0, 1}), ∈ B I sing ,t .
α f ({1, −1}) 2

where B I sing (β I , t) denotes the set of bad magnetizations for the time-evolved Curie-
Weiss Ising model with initial inverse temperature β I . It is known from [28] that
B I sing (β I , t) turns out to be either empty, contain the magnetization value zero, or
to be given by a symmetric pair.
What can we say about the typicality of bad points in the time-evolved mean-
field Widom-Rowlinson model? Typicality means in the mean-field context that
the minimizers of the large-deviation rate function of the time-evolved model are
contained in the set of bad magnetizations. It is an analytical principle for time-
evolved mean-field Ising models that there is an atypicality of bad configurations.
This follows from the principle of preservation of semiconcavity for time-evolved
rate functions which are defined via integrals over Lagrange densities [26]. In simple
words this regularity statement means that kinks in a rate function can never appear
at local minima.
Our model does not fall in the Ising class, but the corresponding statement can
be proved explicitly. It is very nicely illustrated in our model by the following plot.
The repulsion strength of the model at time zero is β = 4 > 3 is the low-temperature
region. The red Y-shaped set denotes the set of bad empirical measures at a fixed
intermediate time. Its form is independent of the initial a priori distribution α, as
long as this is symmetric. By comparison the typical configurations for any α, after
time t are solid blue. They arise as time-evolution of the dotted blue lines describing
typical measures at time zero.

3.4 Lattice Widom-Rowlinson Model Under Time-Evolution

We describe mainly the lattice soft-core Widom-Rowlinson model. To prove the Gibbs
property of the time-evolved model in appropriate regions, Dobrushin-methods are
useful, as we outline now. Let γ := (γ )Zd be a quasilocal specification on the
lattice. The Dobrushin interdependence matrix, is defined by
16 C. Külske

Ci j (γ) = sup γ{i} (·|ω) − γ{i} (·|η)T V,i (17)

ωZd \{ j} =ηZd \{ j}

for sites i = j. The main theorem, dueto Dobrushin, states: If the Dobrushin condi-
tion holds, namely if c(γ) := supi∈Zd j∈Zd Ci j (γ) < 1, then |G(γ)| = 1. The theory
also allows to control the unique Gibbs measure under perturbations of the specifi-
cation, understand correlation decay in the measure, and derive more useful conse-
quences [15]. We show the Dobrushin region (the region in parameter space for which
c(γ) < 1) for the spatially homogeneous soft-core model on Z2 . The plot shows, for
different values of the repulsion strength β, the Dobrushin region (dark shaded) in the
space of a priori measures α ∈ M({−1, 0, 1}), projected to the α(1), α(−1)-plane.
1.0 1.0 1.0 1.0

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

0.4 0.4 0.4 0.4

0.2 0.2 0.2 0.2

0.0 0.0 0.0 0.0

0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

1
β< 2 β = 0.75 β = 1.05 β=2

It turns out that the boundary of the Dobrushin region consists of finitely many
pieces of ellipses. We have the following theorem for the homogeneous model:
(1) Let 0 ≤ βd < 1. Then for all α ∈ M({−1, 0, 1}) the soft-core Widom-
Rowlinson model satisfies the Dobrushin condition.
(2) For every β > 0 there exists an := (β) > 0 such that the soft-core model
satisfies the Dobrushin condition if dT V (α, δ1 ) < or dT V (α, δ−1 ) < .
With Dobrushin techniques one controls not only the translation-invariant model, but
also the first-layer model constrained on future configurations. With this one may
prove also the Gibbs property for the time-evolved model for small times: Let α ∈
M({−1, 0, 1}), β ≥ 0, and let μ ∈ G(γβ,α ) by any Gibbs measure. Then there exists
a time tc > 0 such that for all t < tc the time-evolved measure μt is a Gibbs measure
for some quasilocal specification γt . For the proof see [24], it extends methods of [29]
to a situation of degenerate time evolutions (where not all transitions are allowed)
to control all first-layer models for possible to control all first-layer models (14) for
possible end-conditionings η. We remark that the method does not make use of the
lattice structure, but gives the same result of short-time Gibbsianness for any graph
with bounded degree, for instance a regular tree.
For the opposite direction we prove: In the soft-core model on the lattice, at
sufficiently large repulsion strength, the maximal measure μ+ t is non-Gibbs, for times
t which are sufficiently large. For the proof it suffices to exhibit one non-removable
bad configuration for the single-site probability of the time-evolved measure. We
may choose in our case a fully occupied checkerboard configuration of alternating
plus- and minus-spins, and show that this configuration is bad, noting that we are
reduced basically to an Ising situation for this conditioning.
For the hard-core lattice model under time-evolution, Dobrushin techniques can
not be applied, as some entries of the Dobrushin matrix will necessarily become equal
Gibbs-Non Gibbs Transitions in Different Geometries … 17

to one. This is not just a shortcoming of the proof. Indeed, we find an immediate loss
of the Gibbs property, as in the Euclidean model, for the proof see [24].

3.5 The Widom-Rowlinson Model on a Cayley Tree

Let us now for our graph consider a Cayley tree, which is the infinite graph which
has no loops, and where each vertex has precisely k + 1 nearest neighbors. The
Widom-Rowlinson model in the hard-core version, and in the soft-core version, is
again defined by the specification kernels of (4) and (5).
We need to start with a good understanding of the Widom-Rowlinson model in
equilibrium. The tree-automorphism-invariant Gibbs measures which are also tree-
indexed Markov chains (also known as tree-invariant splitting Gibbs-measures) are
uniquely described via boundary laws, which appear as solutions of a parameter-
dependent two-dimensional fixed point equation (appearing as a tree recursion).
As a general abstract fact, extremal Gibbs measures for tree models are always
splitting Gibbs measures, the opposite is in general not true [15]. For certain classes
of hard-core models on trees the characterizations of solutions can be found in
[35], at least for low enough degree of the tree. For the equilibrium states of the
ferromagnetic soft-core model on the Cayley tree we find the following [23]. In
the antiferromagnetic model with symmetric intensities there is a transition in the
hole-density, somewhat similar to that in the mean-field model briefly described
above. It can be very explicitly analyzed for any order k, with explicit transition lines
in the interaction-intensity diagram. For the ferromagnetic model with symmetric
intensities, for the trees with 3 and 4 nearest neighbors, the critical lines for the
ferromagnetic phase transition are again explicit, with complete description of all
tree-invariant splitting Gibbs measures. For higher k, there are only bounds on critical
curves, which we conjecture to be sharp, see [23].
What about spin-flip time evolution of these measures on the tree? The Gibbsian
behavior of a time-evolved model can be very different from the behavior in other
geometries. For the Ising model on a Cayley-tree under independent stochastic spin-
flip in [7] the following was proved: The set of bad measures may depend on the
choice of the initial Gibbs measure of the time-evolved state. There can be multiple
transition times in the model with zero external magnetic field, and full-measure sets
of bad measures. For the time-evolved Widom-Rowlinson model on the Cayley tree,
this is an open problem.

Acknowledgements I am very grateful I met Anton, for all the discussions I had with him, and for
all inspiration he gave, during my Ph.D., in later years, until today. I wish him many many more
years, I am looking forward to many more of his contributions, to mathematics and beyond!
18 C. Külske

References

1. Arguin, L.-P., Damron, M., Newman, C., Stein, D.: Uniqueness of ground states for short-range
spin glasses in the half-plane. Commun. Math. Phys. 300, 641–657 (2010)
2. Bovier, A.: Statistical Mechanics of Disordered Systems: A Mathematical Perspective, vol. 18.
Cambridge University Press (2006)
3. Bricmont, J., Kuroda, K., Lebowitz, J.L.: The structure of Gibbs states and phase coexistence for
nonsymmetric continuum Widom–Rowlinson models. Z. Wahrsch. Verw. Gebiete 67, 121–138
(1984)
4. Chayes, J.T., Chayes, L., Kotecký, R.: The analysis of the Widom-Rowlinson model by stochas-
tic geometric methods. Commun. Math. Phys. 172, 551–569 (1995)
5. Cotar, C., Jahnel, B., Külske, C.: Extremal decomposition for random Gibbs measures: from
general metastates to metastates on extremal random Gibbs measures. Electron. Commun.
Probab. 23, 1–12 (2018)
6. Dereudre, D., Drouilet, R., Georgii, H.-O.: Existence of Gibbsian point processes with
geometry-dependent interactions. Probab. Theory Relat. Fields 153, 643–670 (2012)
7. van Enter, A.C.D., Ermolaev, V.N., Iacobelli, G., Külske, C.: Gibbs-non-Gibbs properties for
evolving Ising models on trees. Ann. Inst. Henri Poincare Probab. Stat. 48, 774–791 (2012)
8. van Enter, A.C.D., Fernández, R., den Hollander, F., Redig, F.: Possible loss and recovery of
Gibbsianness during the stochastic evolution of Gibbs measures. Commun. Math. Phys. 226,
101–130 (2002)
9. van Enter, A.C.D., Fernández, R., den Hollander, F., Redig, F.: A large-deviation view on
dynamical Gibbs-non-Gibbs transitions. Moscow Math. J. 10, 687–711 (2010)
10. van Enter, A.C.D., Fernández, R., Sokal, A.D.: Regularity properties and pathologies of
position-space renormalization-group transformations: scope and limitations of Gibbsian the-
ory. J. Stat. Phys. 72, 879–1167 (1993)
11. Enter, A.C.D., Külske, C., Opoku, A.A., Ruszel, W.M.: Gibbs-non-Gibbs properties for n-
vector lattice and mean-field models. Braz. J. Probab. Stat. 24, 226–255 (2010)
12. Ermolaev, V.N., Külske, C.: Low-temperature dynamics of the Curie-Weiss model: periodic
orbits, multiple histories and loss of Gibbsianness. J. Stat. Phys. 141, 727–756 (2010)
13. Fernández, R., den Hollander, F., Martínez, J.: Variational description of Gibbs-non-Gibbs
dynamical transitions for spin-flip systems with a Kac-type interaction. J. Stat. Phys. 147,
1094–1112 (2014)
14. Gallavotti, G., Lebowitz, J.: Phase transitions in binary lattice gases. J. Math. Phys. 12, 1129–
1133 (1971)
15. Georgii, H.-O.: Gibbs Measures and Phase Transitions. De Gruyter, New York (2011)
16. Henning, F., Kraaij, R., Külske, C.: Gibbs-non-Gibbs transition in the fuzzy Potts models with
a Kac-type interaction: closing the Ising gap. Bernoulli J. 25, 2051–2074 (2019)
17. Higuchi, Y., Takei, M.: Some results on the phase structure of the two-dimensional Widom-
Rowlinson model. Osaka J. Math. 41, 237–255 (2004)
18. den Hollander, F., Redig, F., van Zuijlen, W.: Gibbs-non-Gibbs dynamical transitions for mean-
field interacting Brownian motions. Stoch. Process. Appl. 125, 371–400 (2015)
19. Jahnel, B., Külske, C.: Attractor properties for irreversible and reversible interacting particle
systems. Commun. Math. Phys. 366, 139–172 (2015)
20. Jahnel, B., Külske, C.: The Widom-Rowlinson model under spin flip: immediate loss and sharp
recovery of quasilocality. Ann. Appl. Probab. 27, 3845–3892 (2017)
21. Jahnel, B., Külske, C.: Gibbsian representation for point processes via hyperedge potentials
(2017). arXiv:1707.05991
22. Kissel, S., Külske, C.: Dynamical Gibbs-non Gibbs transitions in Curie-Weiss Widom-
Rowlinson models. Markov Process. Relat. Fields 25, 379–413 (2019)
23. Kissel, S., Külske, C., Rozikov, U.: Hard-core and soft-core Widom-Rowlinson models on
Cayley trees. J. Stat. Mech. 4(043204), 22 (2019)
24. Kissel, S., Külske, C.: Dynamical Gibbs-non Gibbs transitions for hard-core and soft-core
Widom-Rowlinson models on the lattice (in preparation)
Gibbs-Non Gibbs Transitions in Different Geometries … 19

25. Kozlov, O.K.: A Gibbs description of a system of random variables. Problemy Peredaci Infor-
macii 10, 94–103 (1974)
26. Kraaij, R., Redig, F., van Zuijlen, W.: A Hamilton-Jacobi point of view on mean-field Gibbs-non
Gibbs transitions (2017). arXiv:1711.03489
27. Külske, C.: Metastates in disordered mean-field models: random field and Hopfield models. J.
Stat. Phys. 88, 1257–1293 (1997)
28. Külske, C., Le Ny, A.: Spin-flip dynamics of the Curie-Weiss model: loss of Gibbsianness with
possibly broken symmetry. Commun. Math. Phys. 271, 431–454 (2007)
29. Külske, C., Opoku, A.A.: The posterior metric and the goodness of Gibbsianness for transforms
of Gibbs measures. Electron. J. Probab. 13, 1307–1344 (2008)
30. Külske, C., Redig, F.: Loss without recovery of Gibbsianness during diffusion of continuous
spins. Probab. Theory Relat. Fields 135, 428–456 (2006)
31. Liggett, T.: Interacting Particle Systems. Springer-Verlag, New York (1985)
32. Mazel, A., Stuhl, I., Suhov, Y.: Hard-core configurations on a triangular lattice and Eisenstein
primes (2018). arXiv:1803.04041
33. Mazel, A., Stuhl, I., Suhov, Y.: A classical WR model with q particle types. J. Stat. Phys. 159,
1040–1086 (2015)
34. Newman, C.M., Stein, D.: Spin Glasses and Complexity. Princeton University Press (2013)
35. Rozikov, U.A.: Gibbs Measures on Cayley Trees. World Sci. Publ, Singapore (2013)
36. Ruelle, D.: Existence of a phase transition in a continuous classical system. Phys. Rev. Lett.
27, 1040–1041 (1971)
37. Ruelle, D.: Statistical Mechanics: Rigorous Results. World Scientific, River Edge, NJ (1999)
38. Sullivan, W.G.: Potentials for almost Markovian random fields. Commun. Math. Phys. 33,
61–74 (1973)
39. Widom, B., Rowlinson, J.S.: New model for the study of liquid-vapor phase transition. J. Chem.
Phys. 52, 1670–1684 (1970)
Discovering Diverse Content Through
Random Scribd Documents
 FOOTNOTES:
[27] Copyright, 1911, by J. B. Lippincott Co., and used by
permission.
 VIII
PSYCHOLOGICAL STUDIES
Markheim.—Robert Louis Stevenson
On the Stairs.—Arthur Morrison
He [the author] can sometimes rouse our intense curiosity
and eagerness by the mere depiction of a psychological
state, as Walter Pater has done in the case of Sebastian
Storck and other personages of his Imaginary Portraits.
The fact that “nothing happens” in stories of this kind may
be precisely what most interests us, because we are
made to understand what it is that inhibits action.—Bliss
Perry, A Study of Prose Fiction.
 PSYCHOLOGICAL STUDIES
A subtle distinction is to be observed between the character-study
and the psychological study, but it will not be supposed that writers
of short-stories plainly label the distinction, or that the two types are
frequently, if ever, found entirely separate. In the character-study
more attention is paid to the true natures of the actors, and the
demonstration of their natures is shown in the action of the story; in
the psychological study more stress is laid upon the actual operation
of thought, feeling and purpose—it is a laboratory study of what goes
on in the human heart, to use a somewhat vague but necessary
term, under stress of crisis.
The psychological study is the most difficult because the most
penetrating of all short-story forms, and in consequence the most
rare in its perfect presentation. To show the processes of reasoning,
the interplay of motive, the power of feeling acting upon feeling, and
the intricate combinations of these, calls for the most clear-sighted
understanding of man, and the utmost skill in literary art, lest the
story be lost in a fog of tiresome analysis and discussion. In
“Markheim” and “On the Stairs,” two master story-tellers are easily at
their best, for they never obtrude their own opinions, but swiftly and
with a firm onward movement the stories disclose the true inward
workings of the unique characters, while from mood, speech, and
action we infallibly infer all the soul-processes by which their
conclusions are reached.

MARKHEIM
BY ROBERT LOUIS STEVENSON

“Yes,” said the dealer, “our wind-falls are of Introduction.

Remarkable
various kinds. Some customers are because it at once
ignorant, and then I touch a dividend on my touches upon the
external crisis of the
superior knowledge. Some are dishonest,” story.
and here he held up the candle, so that the
light fell strongly on his visitor, “and in that case,” he
continued, “I profit by my virtue.”
2. Markheim had but just entered from the daylight streets,
and his eyes had not yet grown familiar with the mingled
shine and darkness in the shop. At these Note double reason.
pointed words, and before the near
presence of the flame, he blinked painfully and looked aside.
3. The dealer chuckled. “You come to me on See how daringly
Christmas-day,” he resumed, “when you the author plays
know that I am alone in my house, put up with the reader
my shutters, and make a point of refusing without arousing
suspicion. Compare
business. Well, you will have to pay for that; Stevenson’s
you will have to pay for my loss of time, reasoning as to the
reader’s suspicions
when I should be balancing my books; you with Dupin’s
will have to pay, besides, for a kind of reasoning in “The
manner that I remark in you to-day very Purloined Letter,”
pp. 91, 92.
strongly. I am the essence of discretion, and
ask no awkward questions; but when a customer can not look
me in the eye, he has to pay for it.” The dealer once more
chuckled; and then, changing to his usual business voice,
though still with a note of irony, “You can Markheim has been
give, as usual, a clean account of how you there before.
came into possession of the object?” he
continued. “Still your uncle’s cabinet? A remarkable collector,
sir!”
4. And the little, pale, round-shouldered dealer stood almost
on tip-toe, looking over the top of his gold spectacles, and
nodding his head with every mark of disbelief. Markheim
returned his gaze with one of infinite pity, and a touch of
horror.
5. “This time,” he said, “you are in error. I Forecast.
have not come to sell, but to buy. I have no
curios to dispose of; my uncle’s cabinet is bare to the
wainscot; even were it still intact, I have done well on the
Stock Exchange, and should more likely add to it than
otherwise, and my errand to-day is simplicity itself. I seek a
Christmas-present for a lady,” he continued, waxing more
fluent as he struck into the speech he had Insincerity evident.
prepared; “and certainly I owe you every
excuse for thus disturbing you upon so small a matter. But the
thing was neglected yesterday; I must produce my little
compliment at dinner; and, as you very well know, a rich
marriage is not a thing to be neglected.”
6. There followed a pause, during which the dealer seemed to
weigh this statement incredulously. The ticking of many clocks
among the curious lumber of the shop, and the faint rushing
of the cabs in a near thoroughfare, filled up the interval of
silence.
7. “Well, sir,” said the dealer, “be it so. You Compare this
are an old customer after all; and if, as you setting, as it is
say, you have the chance of a good gradually unfolded,
with that of Gautier’s
marriage, far be it from me to be an “The Mummy’s
obstacle. Here is a nice thing for a lady Foot.”
now,” he went on, “this hand-glass—fifteenth
century, warranted; comes from a good collection, too; but I
reserve the name, in the interests of my customer, who was
just like yourself, my dear sir, the nephew and sole heir of a
remarkable collector.”
8. The dealer, while he thus ran on in his dry Analyse its nature.
and biting voice, had stooped to take the
object from its place; and, as he had done so, a shock had
passed through Markheim, a start both of hand and foot, a
sudden leap of many tumultuous passions to the face. It
passed as swiftly as it came, and left no trace beyond a
certain trembling of the hand that now received the glass.
9. “A glass,” he said, hoarsely, and then Contributory
paused, and repeated it more clearly. “A incident.
glass? For Christmas? Surely not?”
10. “And why not?” cried the dealer. “Why not a glass?”
11. Markheim was looking upon him with an Forecast.
indefinable expression. “You ask me why
not?” he said. “Why, look here—look in it—look at yourself!
Do you like to see it? No! nor I—nor any man.”
12. The little man had jumped back when Markheim had so
suddenly confronted him with the mirror; but now, perceiving
there was nothing worse on hand, he chuckled. “Your future
lady, sir, must be pretty hard favoured,” said he.
13. “I ask you,” said Markheim, “for a Note the working of
Christmas-present, and you give me this— Markheim’s morbid
this damned reminder of years, and sins and conscience, not yet
understood by
follies—this hand-conscience! Did you mean himself.
it? Had you a thought in your mind? Tell me.
It will be better for you if you do. Come, tell me about yourself.
I hazard a guess now, that you are in secret a very charitable
man?”
14. The dealer looked closely at his First Moral
companion. It was very odd, Markheim did Crisis.
not appear to be laughing; there was
something in his face like an eager sparkle of hope, but
nothing of mirth.
15. “What are you driving at?” the dealer asked.
16. “Not charitable?” returned the other, gloomily. “Not
charitable; not pious; not scrupulous; unloving, unbeloved; a
hand to get money, a safe to keep it. Is that all? Dear God,
man, is that all?”
17. “I will tell you what it is,” began the dealer, with some
sharpness, and then broke off again into a chuckle. “But I see
this is a love match of yours, and you have been drinking the
lady’s health.”
18. “Ah!” cried Markheim, with a strange curiosity. “Ah, have
you been in love? Tell me about that.”
19. “I,” cried the dealer. “I in love! I never Note change of
had the time, nor have I the time to-day for attitude.
all this nonsense. Will you take the glass?”
20. “Where is the hurry?” returned Analyse the forces
Markheim. “It is very pleasant to stand here back of Markheim’s
talking; and life is so short and insecure that parleying.
I would not hurry away from any pleasure—
no, not even from so mild a one as this. We should rather
cling, cling to what little we can get, like a man at a cliff’s
edge. Every second is a cliff, if you think upon it—a cliff a mile
high—high enough, if we fall, to dash us out of every feature
of humanity. Hence it is best to talk pleasantly. Let us talk of
each other; why should we wear this mask? Let us be
confidential. Who knows, we might become friends?”
21. “I have just one word to say to you,” said the dealer.
“Either make your purchase, or walk out of my shop.”
22. “True, true,” said Markheim. “Enough Note how quickly
fooling. To business. Show me something Markheim follows
else.” the unconscious
lead.
23. The dealer stooped once more, this time
to replace the glass upon the shelf, his thin First External
Crisis.
blonde hair falling over his eyes as he did
so. Markheim moved a little nearer, with one hand in the
pocket of his great-coat; he drew himself up and filled his
lungs; at the same time many different Note all these.
emotions were depicted together on his face
—terror, horror and resolve, fascination and a physical
repulsion; and through a haggard lift of his upper lip, his teeth
looked out.
24. “This, perhaps, may suit,” observed the He was prepared for
dealer; and then, as he began to re-arise, the crime.
Markheim bounded from behind upon his
victim. The long, skewerlike dagger flashed and fell. The
dealer struggled like a hen, striking his temple on the shelf,
and then tumbled on the floor in a heap.
25. Time had some score of small voices in First Minor
that shop, some stately and slow as was Climax.
becoming to their great age; others
garrulous and hurried. All these told out the seconds in an
intricate chorus of tickings. Then the passage of a lad’s feet,
heavily running on the pavement, broke in upon these smaller
voices and startled Markheim into the consciousness of his
surroundings. He looked about him awfully. Beginning of the
The candle stood on the counter, its flame internal action. Note
solemnly wagging in a draught; and by that how all external
inconsiderable movement, the whole room things now begin to
play upon the
was filled with noiseless bustle and kept internal man.
heaving like a sea: the tall shadows
nodding, the gross blots of darkness Throughout, note
Stevenson’s rich
swelling and dwindling as with respiration, imagery, and also
the faces of the portraits and the china gods his unusual
changing and wavering like images in water. vocabulary.
The inner door stood ajar, and peered into that leaguer of
shadows with a long slit of daylight like a An unusual word.
pointing finger.
26. From these fear-stricken rovings, Picture.
Markheim’s eyes returned to the body of his
victim, where it lay both humped and sprawling, incredibly
small and strangely meaner than in life. In these poor, miserly
clothes, in that ungainly attitude, the dealer lay like so much
sawdust. Markheim had feared to see it, Evidence of
and, lo! it was nothing. And yet, as he premeditation.
gazed, this bundle of old clothes and pool of
blood began to find eloquent voices. There it Note the interplay of
must lie; there was none to work the the outward picture
cunning hinges or direct the miracle of and Markheim’s
locomotion—there it must lie till it was found. mind. Keep before
you always the
Found! ay, and then? Then would this dead double movement of
flesh lift up a cry that would ring over this study as both
England, and fill the world with the echoes progress side by
side, finally resulting
of pursuit. Ay, dead, or not, this was still the in the predominance
enemy. “Time was that when the brains of the spiritual.
were out,” he thought; and the first word
struck into his mind. Time, now that the deed was
accomplished—time, which had closed for the victim, had
become instant and momentous for the slayer.
27. The thought was yet in his mind, when, first one and then
another, with every variety of pace and voice—one deep as
the bell from a cathedral turret, another ringing on its treble
notes the prelude of a waltz—the clocks began to strike the
hour of three in the afternoon.
28. The sudden outbreak of so many tongues in that dumb
chamber staggered him. He began to bestir himself, going to
and fro with the candle, beleaguered by moving shadows, and
startled to the soul by chance reflections. In many rich
mirrors, some of home designs, some from Venice or
Amsterdam, he saw his face repeated and The old motive
repeated, as it were an army of spies; his reasserts itself.
own eyes met and detected him; and the
sound of his own steps, lightly as they fell, vexed the
surrounding quiet. And still as he continued Plot Incident.
to fill his pockets, his mind accused him,
with a sickening iteration, of the thousand faults of his design.
He should have chosen a more quiet hour; As fear subsides
he should have prepared an alibi; he should craft returns.
not have used a knife; he should have been
more cautious, and only bound and gagged the dealer, and
not killed him; he should have been more bold, and killed the
servant also; he should have done all things otherwise;
poignant regrets, weary, incessant toiling of the mind to
change what was unchangeable, to plan what was now
useless, to be the architect of the A significant
irrevocable past. Meanwhile, and behind all expression.
this activity, brute terrors, like scurrying of
rats in a deserted attic, filled the more remote chambers of his
brain with riot; the hand of the constable Contrast physical
would fall heavy on his shoulder, and his and moral fear.
nerves would jerk like a hooked fish; or he Consider how the
beheld, in galloping defile, the dock, the two are related.
prison, the gallows, and the black coffin.
29. Terror of the people in the street sat Note the primary
down before his mind like a besieging army. use of the word
It was impossible, he thought, but that some “rumor.”
rumor of the struggle must have reached
their ears and set on edge their curiosity; and now, in all the
neighboring houses, he divined them sitting motionless and
with uplifted ear—solitary people, condemned to spend
Christmas dwelling alone on memories of Contrast.
the past, and now startlingly recalled from
that tender exercise; happy family parties, struck into silence
round the table, the mother still with raised finger: every
degree and age and humor, but all, by their own hearths,
prying and hearkening and weaving the rope that was to hang
him. Sometimes it seemed to him he could not move too
softly; the clink of the tall Bohemian goblets rang out loudly
like a bell; and alarmed by the bigness of the ticking, he was
tempted to stop the clocks. And then, again, Study of fear.
with a swift transition of his terrors, the very Impressionism.
silence of the place appeared a source of
peril, and a thing to strike and freeze the passerby; and he
would step more boldly, and bustle aloud among the contents
of the shop, and imitate, with elaborate bravado, the
movements of a busy man at ease in his own house.
30. But he was now so pulled about by An important
different alarms that, while one portion of his observation.
mind was still alert and cunning, another
trembled on the brink of lunacy. One hallucination in particular
took a strong hold on his credulity. The neighbor hearkening
with white face beside his window, the passerby arrested by a
horrible surmise on the pavement—these could at worst
suspect, they could not know; through the brick walls and
shuttered windows only sounds could penetrate. But here,
within the house, was he alone? He knew Note how his
he was; he had watched the servant set reasoning becomes
forth sweethearting, in her poor best, “out for hyper-acute
the day” written in every ribbon and smile.
Yes, he was alone, of course; and yet, in the bulk of empty
house above him, he could surely hear a stir of delicate
footing—he was surely conscious, inexplicably conscious of
some presence. Ay, surely; to every room Forecast.
and corner of the house his imagination
followed it; and now it was a faceless thing, and yet had eyes
to see with; and again it was a shadow of himself; and yet
again behold the image of the dead dealer, reinspired with
cunning and hatred.
31. At times, with a strong effort, he would glance at the open
door which still seemed to repel his eyes. The house was tall,
the skylight small and dirty, the day blind Note force of “blind.”
with fog; and the light that filtered down to
the ground story was exceedingly faint, and showed dimly on
the threshold of the shop. And yet, in that strip of doubtful
brightness, did there not hang wavering a shadow?
32. Suddenly, from the street outside, a very Pseudo crisis.
jovial gentleman began to beat with a staff
on the shop-door, accompanying his blows with shouts and
railleries in which the dealer was continually called upon by
name. Markheim, smitten into ice, glanced Contributory
at the dead man. But no! he lay quite still; he incident.
was fled away far beyond earshot of these
blows and shoutings; he was sunk beneath seas of silence;
and his name, which would once have caught his notice
above the howling of a storm, had become an empty sound.
And presently the jovial gentleman desisted from his knocking
and departed.
33. Here was a broad hint to hurry what remained to be done,
to get forth from this accusing neighborhood, to plunge into a
bath of London multitudes, and to reach, on Note “apparent.”
the other side of day, that haven of safety
and apparent innocence—his bed. One visitor had come: at
any moment another might follow and be more obstinate. To
have done the deed, and yet not to reap the Key.
profit, would be too abhorrent a failure. The
money, that was now Markheim’s concern; and as a means to
that, the keys.
34. He glanced over his shoulder at the Note subsidence of
open door, where the shadow was still acute fears and rise
lingering and shivering; and with no of his true mood.
conscious repugnance of the mind, yet with
a tremor of the belly, he drew near the body of his victim. The
human character had quite departed. Like a suit half-stuffed
with bran, the limbs lay scattered, the trunk doubled, on the
floor; and yet the thing repelled him. Although so dingy and
inconsiderable to the eye, he feared it might have more
significance to the touch. He took the body by the shoulders,
and turned it on its back. It was strangely light and supple,
and the limbs, as if they had been broken, fell into the oddest
postures. The face was robbed of all Carefully consider
expression; but it was as pale as wax, and the question of
shockingly smeared with blood about one Markheim’s sanity,
temple. That was, for Markheim, the one judging only from
the story as thus far
displeasing circumstance. It carried him told.
back, upon the instant, to a certain fair day
in a fisher’s village: a gray day, a piping wind, a crowd upon
the street, the blare of brasses, the booming of drums, the
nasal voice of a ballad singer; and a boy going to and fro,
buried over head in the crowd and divided between interest
and fear, until, coming out upon the chief place of concourse,
he beheld a booth and a great screen with pictures, dismally
designed, garishly colored: Brownrigg with her apprentice; the
Mannings with their murdered guest; Weare in the death-grip
of Thurtell; and a score besides of famous crimes. The thing
was as clear as an illusion; he was once again that little boy;
he was looking once again, and with the same sense of
physical revolt, at these vile pictures; he was still stunned by
the thumping of the drums. A bar of that Reaction.
day’s music returned upon his memory; and
at that, for the first time, a qualm came over him, a breath of
nausea, a sudden weakness of the joints, which he must
instantly resist and conquer.
35. He judged it more prudent to confront than to flee from
these considerations; looking the more hardily in the dead
face, bending his mind to realize the nature and greatness of
his crime. So little a while ago that face had moved with every
change of sentiment, that pale mouth had spoken, that body
had been all on fire with governable energies; and now, and
by his act, that piece of life had been arrested, as the
horologist, with interjected finger, arrests the Key. What caused
beating of the clock. So he reasoned in vain; this benumbed
he could rise to no more remorseful conscience?
consciousness; the same heart which had
shuddered before the painted effigies of crime, looked on its
reality unmoved. At best, he felt a gleam of pity for one who
had been endowed in vain with all those faculties that can
make the world a garden of enchantment, one who had never
lived and who was now dead. But of penitence, no, not a
tremor.
36. With that, shaking himself clear of these Plot Incident.
considerations, he found the keys and
advanced toward the open door of the shop. Outside, it had
begun to rain smartly; and the sound of the shower upon the
roof had banished silence. Like some dripping cavern, the
chambers of the house were haunted by an incessant
echoing, which filled the ear and mingled with the ticking of
the clocks. And, as Markheim approached Forecast of moral
the door, he seemed to hear, in answer to crisis.
his own cautious tread, the steps of another
foot withdrawing up the stair. The shadow still palpitated
loosely on the threshold. He threw a ton’s weight of resolve
upon his muscles, and drew back the door.
37. The faint, foggy daylight glimmered dimly on the bare floor
and stairs; on the bright suit of armor posted, halbert in hand,
upon the landing; and on the dark wood-carvings, and framed
pictures that hung against the yellow panels of the wainscot.
So loud was the beating of the rain through Note harmony of
all the house that, in Markheim’s ears, it setting with tone of
began to be distinguished into many approaching crisis.
different sounds. Footsteps and sighs, the
tread of regiments marching in the distance, the chink of
money in the counting, and the creaking of Compare
doors held stealthily ajar, appeared to Stevenson’s
mingle with the patter of the drops upon the combination of fact
cupola and the gushing of the water in the and fantasy with
Hawthorne’s in “The
pipes. The sense that he was not alone White Old Maid.”
grew upon him to the verge of madness. On
every side he was haunted and begirt by Rise toward crisis.
presences. He heard them moving in the
upper chambers; from the shop, he heard the dead man
getting to his legs; and as he began with a great effort to
mount the stairs, feet fled quietly before him and followed
stealthily behind. If he were but deaf, he Body and spirit.
thought, how tranquilly he would possess
his soul. And then again, and hearkening with every fresh
attention, he blessed himself for that unresisting sense which
held the outposts and stood a trusty sentinel upon his life. His
head turned continually on his neck; his eyes, which seemed
starting from their orbits, scouted on every side, and on every
side were half-rewarded as with the tail of A notable passage.
something nameless, vanishing. The four-
and-twenty steps to the first floor were four-and-twenty
agonies.
38. On that first story, the doors stood ajar, three of them like
three ambushes, shaking his nerves like the throats of
cannon. He could never again, he felt, be sufficiently immured
and fortified from men’s observing eyes; he longed to be
home, girt in by walls, buried among bed- Note the exception.
clothes, and invisible to all but God. And at
that thought he wondered a little, recollecting tales of other
murderers and the fear they were said to entertain of
heavenly avengers. It was not so, at least, with him. He
feared the laws of nature, lest, in their callous and immutable
procedure, they should preserve some damning evidence of
his crime. He feared tenfold more, with a slavish, superstitious
terror, some scission in the continuity of man’s experience,
some willful illegality of nature. He played a game of skill,
depending on the rules, calculating consequence from cause;
and what if nature, as the defeated tyrant overthrew the
chess-board, should break the mold of their succession? The
like had befallen Napoleon (so writers said) when the winter
changed the time of its appearance. The like Note how suspense
might befall Markheim: the solid walls might in the reader is
become transparent and reveal his doings maintained by
like those of bees in a glass hive; the stout disclosing
Markheim’s
planks might yield under his foot like suspense.
quicksands and detain him in their clutch;
ay, and there were soberer accidents that might destroy him;
if, for instance, the house should fall and imprison him beside
the body of his victim; or the house next door should fly on
fire, and the firemen invade him from all sides. These things
he feared; and, in a sense, these things might be called the
hands of God reached forth against sin. But Key.
about God himself he was at ease; his act
was doubtless exceptional, but so were his excuses, which
God knew; it was there, and not among Is this normal?
men, that he felt sure of justice.
39. When he had got safe into the drawing- Note action of auto-
room, and shut the door behind him, he was suggestion.
aware of a respite from alarms. The room
was quite dismantled, uncarpeted besides, and strewn with
packing cases and incongruous furniture; several great pier-
glasses, in which he beheld himself at various angles, like an
actor on the stage; many pictures, framed and unframed,
standing with their faces to the wall; a fine Sheraton
sideboard, a cabinet of marquetry, and a great old bed, with
tapestry hangings. The windows opened to the floor; but by
great good fortune the lower part of the shutters had been
closed, and this concealed him from the neighbors. Here,
then, Markheim drew in a packing case before the cabinet,
and began to search among the keys. It was a long business,
for there were many; and it was irksome, besides; for, after
all, there might be nothing in the cabinet, and time was on the
wing. But the closeness of the occupation sobered him. With
the tail of his eye he saw the door—even glanced at it from
time to time directly, like a besieged commander pleased to
verify the good estate of his defenses. But in Remarkable relief in
truth he was at peace. The rain falling in the suspense period.
street sounded natural and pleasant.
Presently, on the other side, the notes of a piano were
wakened to the music of a hymn, and voices of many children
took up the air and words. How stately, how comfortable was
the melody! How fresh the youthful voices! Markheim gave
ear to it smilingly, as he sorted out the keys; and his mind was
thronged with answerable ideas and images; church-going
children and the pealing of the high organ; children afield,
bathers by the brookside, ramblers on the brambly common,
kite-flyers in the windy and cloud-navigated sky; and then, at
another cadence of the hymn, back again to church, and the
somnolence of summer Sundays, and the high genteel voice
of the parson (which he smiled a little to recall) and the
painted Jacobean tombs, and the dim lettering of the Ten
Commandments in the chancel.
40. And as he sat thus, at once busy and absent, he was
startled to his feet. A flash of ice, a flash of fire, a bursting
gush of blood, went over him, and then he Powerful contrast.
stood transfixed and thrilling. A step
mounted the stair slowly and steadily, and Approach of moral
presently a hand was laid upon the knob, crisis.
and the lock clicked, and the door opened.
41. Fear held Markheim in a vice. What to Markheim perceives
expect he knew not, whether the dead man only a physical
walking, or the official ministers of human danger. Note how
long he remains
justice, or some chance witness blindly dead to any moral
stumbling in to consign him to the gallows. judgment of himself.
But when a face was thrust into the
aperture, glanced round the room, looked at him, nodded and
smiled as if in friendly recognition, and then Here is a real
withdrew again, and the door closed behind though
it, his fear broke loose from his control in a unrecognized moral
hoarse cry. At the sound of this the visitant crisis. Fear
eventually leads to
returned. his moral triumph.

42. “Did you call me?” he asked, pleasantly, Note the symbolism
and with that he entered the room and of the closed door.
closed the door behind him.
43. Markheim stood and gazed at him with Key.
all his eyes. Perhaps there was a film upon
his sight, but the outlines of the newcomer seemed to change
and waver like those of the idols in the wavering candle-light
of the shop; and at times he thought he knew him; and at
times he thought he bore a likeness to himself; and always,
like a lump of living terror, there lay in his This states the
bosom the conviction that this thing was not problem.
of the earth and not of God.
44. And yet the creature had a strange air of the common-
place, as he stood looking on Markheim with a smile; and
when he added: “You are looking for the money, I believe?” it
was in the tones of everyday politeness.
45. Markheim made no answer.
46. “I should warn you,” resumed the other, “that the maid has
left her sweetheart earlier than usual and will soon be here. If
Mr. Markheim be found in this house, I need not describe to
him the consequences.”
47. “You know me?” cried the murderer.
48. The visitor smiled. “You have long been a favorite of
mine,” he said; “and I have long observed and often sought to
help you.”
49. “What are you?” cried Markheim: “the devil?”
50. “What I may be,” returned the other, This is an important
“cannot affect the service I propose to passage.
render you.”
51. “It can,” cried Markheim; “it does! Be Forecast of
helped by you? No, never; not by you! You Markheim’s struggle
do not know me yet, thank God, you do not with his better self.
know me!”
52. “I know you,” replied the visitant, with a sort of kind
severity or rather firmness. “I know you to the soul.”
53. “Know me!” cried Markheim. “Who can Does Markheim
do so? My life is but a travesty and slander really know himself?
on myself. I have lived to belie my nature.
All men do; all men are better than this disguise that grows
about and stifles them. You see each dragged away by life,
like one whom bravos have seized and muffled in a cloak. If
they had their own control—if you could see their faces, they
would be altogether different, they would shine out for heroes
and saints! I am worse than most; myself is more overlaid; my
excuse is known to me and God. But, had I the time, I could
disclose myself.”
54. “To me?” inquired the visitant.
55. “To you before all,” returned the Note the author’s
murderer. “I supposed you were intelligent. I name for Markheim.
thought—since you exist—you would prove
a reader of the heart. And yet you would propose to judge me
by my acts! Think of it; my acts! I was born Seek a cause for
and I have lived in a land of giants; giants such reasoning.
have dragged me by the wrists since I was
born out of my mother—the giants of circumstance. And you
would judge me by my acts! But can you not look within? Can
you not understand that evil is hateful to me? Can you not see
within me the clear writing of conscience, never blurred by
any willful sophistry, although too often disregarded? Can you
not read me for a thing that surely must be common as
humanity—the unwilling sinner?”
56. “All this is very feelingly expressed,” was Note the distinction
the reply. “But it regards me not. These between the final
points of consistency are beyond my importance of cause
province, and I care not in the least by what and effect.
compulsion you may have been dragged away, so as you are
but carried in the right direction. But time flies; the servant
delays, looking in the faces of the crowd and Contrast.
at the pictures on the hoardings, but still she
keeps moving nearer; and remember, it is as if the gallows
itself was striding toward you through the Christmas streets!
Shall I help you; I, who know all? Shall I tell Minor Moral
you where to find the money?” Crisis.

57. “For what price?” asked Markheim. A test of Markheim’s

consistency.
58. “I offer you the service for a Christmas
gift,” returned the other.
59. Markheim could not refrain from smiling with a kind of
bitter triumph. “No,” said he, “I will take nothing at your hands;
if I were dying of thirst, and it was your hand that put the
pitcher to my lips, I should find the courage to refuse. It may
be credulous, but I will do nothing to commit myself to evil.”
60. “I have no objection to a death-bed repentance,” observed
the visitant.
61. “Because you disbelieve their efficacy!” Markheim cried.
62. “I do not say so,” returned the other; “but I look on these
things from a different side, and when the life is done my
interest falls. The man has lived to serve Key.
me, to spread black looks under color of
religion, or to sow tares in the wheat-field, as you do, in a
course of weak compliance with desire. Now Is this irony?
that he draws so near to his deliverance, he
can add but one act of service—to repent, to die smiling, and
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com