PDF Statistical Mechanics of Classical and Disordered Systems Luminy France August 2018 Veronique Gayrard Ebook Full Chapter

Download as pdf or txt
Download as pdf or txt
You are on page 1of 53

Statistical Mechanics of Classical and

Disordered Systems Luminy France


August 2018 Véronique Gayrard
Visit to download the full and correct content document:
https://textbookfull.com/product/statistical-mechanics-of-classical-and-disordered-syst
ems-luminy-france-august-2018-veronique-gayrard/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

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/

Foundations of Classical Mechanics P. C. Deshmukh

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

Non equilibrium statistical mechanics Prigogine

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

Statistical Physics Michael V. Sadovskii

https://textbookfull.com/product/statistical-physics-michael-v-
sadovskii/
Advanced classical mechanics 1st Edition Bagchi

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

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/

A modern approach to classical mechanics Iro

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

Variational principles in classical mechanics Cline D.

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

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/
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)
Another random document with
no related content on Scribd:
"Good-bye, ma'am," said Jane. "God bless you!"

"Good-bye, mother, dear, dear mother!" cried Mavis, trying to smile.

Then, as the gig moved off, she waved her hand, and continued to do so till it was out of
sight. After that, she found it impossible to keep her composure any longer, and burst into
a flood of tears. Her aunt and Jane were both very kind and sympathetic, but she begged
them to let her be by herself. And, running into the house, she sought refuge in her own
room, where she sobbed out her grief undisturbed for some time. By-and-by, however,
Jane arrived, duster in hand. And Mavis, who had now passed the first keen pangs of
sorrow, bathed her tear-stained face, and inquired where she would find her aunt.

"She's downstairs, miss; you'll find her either in the kitchen or the back garden. Monday's
always a busy day with us, for it's washing-day. A woman from the village, Mrs. Long,
comes to wash. She's worked for Mrs. John for years."

"You mean Aunt Lizzie when you speak of Mrs. John, don't you, Jane?"

"Yes; most folks call her Mrs. John, for master's mother was living when he married. Your
mother is Mrs. Grey now, you know, for your father was the elder son. He might have had
the mill, if he had liked, but he preferred to be a clergyman. I knew both your father and
your uncle when they were boys. I lived here as servant when they were growing up, so
you see I've been with the family a great many years."

Mavis went downstairs and found her way to the kitchen, beyond which was a big scullery,
and outside that a wash-house, where a stout, rosy-cheeked woman was hard at work at a
wash-tub, up to elbows in soapsuds, and enveloped in a cloud of steam.

"Good morning, missie," she said to Mavis, smiling at her good-temperedly, and with
sympathy in her glance; for she knew the little girl's mother had left that morning, and
guessed that was the cause of her sorrowful face.

"Good morning," Mavis replied, returning her smile.

She went out into the kitchen-garden, where she found her aunt hanging various garments
on the clothes lines, which extended the whole length of the garden.

"Do let me help you, Aunt Lizzie," she said. "Isn't there something I can do?"

"You might spread these handkerchiefs on the hedge to bleach, they're Bob's. See what a
dreadful colour they are, and no wonder, for I caught him dusting his boots with one of
them and cleaning his slate with another! Boys make no end of work."

Mavis did as she was desired.

But her aunt had nothing more for her to do, so she found her way out of the garden by a
gate in the hedge into the meadow beyond, and strolled along the bank of the river.

She was still within sight of the house, when she was startled by a big, black, formidable-
looking dog, which came up and sniffed at her more out of curiosity—as she was quick
enough to discern—than with any idea of intimidating her. Mavis was unaccustomed to
dogs, but she was no coward, so she extended her hand to the great animal and spoke to
him, whereupon he was so overcome with her condescension that he quite lost his head,
and circled around her in delight, whilst she laughed heartily.

"Well, little maiden!"


Mavis turned at the sound of an amused voice addressing her, and found herself face to
face with the Vicar. She recognized his elderly, clean-shaven countenance, with its sweet-
tempered mouth and clear grey eyes, immediately.

"You are not afraid of my dog, I perceive," he proceeded, with a smile. "You need not be,
for he is very quiet. His name is Max. I do not know you, do I?" he asked doubtfully, as he
saw recognition in her glance.

"Oh no!" she responded, quickly.

"I think you must be the miller's little niece," he said, after a minute's reflection, during
which he had noticed the traces of recent tears on her face. "Mr. Grey told me he expected
you, and explained the circumstances under which you were to be left with him. Is your
mother gone?"

"She went this morning, not long ago," Mavis replied, with quivering lips. "But she does
not sail till Thursday. She is going ever so far away—to Australia—and I shan't see her for
months and months," she added, mournfully.

"Meanwhile, I hope you will be very happy at the Mill House with your relatives. Come
here, Max."

The dog obeyed his master's call, and allowed Mavis to pat his great head, after which, he
licked her hand, and she felt she had made a friend.

"He is a Newfoundland," Mr. Moseley said, "and he will fetch anything out of the water.
See!"

He picked up a stone, showed it to the dog, and flung it into the river. Max dived after it
immediately, and presently, reappearing, swam ashore and laid the same stone at his
master's feet. Mavis was delighted, and the performance was repeated several times for
her benefit.

"What a clever dog he is!" she cried, enthusiastically. "Aren't you very fond of him?"

"Yes," the Vicar replied, "Max and I are great friends; we understand each other. How do
you think you will like life in the country?" he inquired.

"I should like it, if mother was here, but I don't think I can be happy anywhere without
her," was the doleful response.

"Oh yes, you can," he said decidedly; "you must try to be happy; that is the duty of every
one. Life is hard for most of us at some time or other; it brings pain and separation. But
we ought not to become gloomy and sad. If there were no partings, there would be no
happy meetings. What is your name, my dear?"

"Mavis Grey."

"Mavis is a very pretty name. A little girl with that name should be as happy as a bird!"

Mavis smiled. She thought she would like to tell him how much she had enjoyed his
sermon last night. And, after a brief hesitation, she did so.

He appeared greatly pleased.


"It was just as though you knew how lonely I was feeling," she said eagerly, "just as
though you were preaching to me. I shall remember all you said—always, I hope. I'm not
going to let my heart be troubled—not more than I can help. And I'm going to try not to be
afraid, though there are so many things which might happen to mother!"

"There are many things which might happen to all of us. But we must trust ourselves and
those we love to our Father in Heaven." Mr. Moseley paused for a minute, then proceeded,
"I did not know my sermon last night was appropriate to any one in particular, but God
knew, we may be sure of that."

"Did He tell you what to say?" Mavis inquired. "Oh," she cried, as her companion assented,
"I was certain He did."

Mavis lingered a short while longer in conversation with her new acquaintance. Then she
remembered that her aunt might be wondering where she was, and, having said good-bye
to Mr. Moseley and bestowed a farewell pat on Max, she retraced her footsteps the way she
had come. Her drooping spirits were reviving, for her talk with the Vicar had done her
good. She looked up into the clear, blue sky overhead, and a glow of happiness crept into
her heart. Then she glanced across the sweep of meadow-lands, and began to sing in a
soft undertone—

"The Lord is only my support, and He that doth me feed;


How can I then lack anything whereof I stand in need?
In pastures green He feedeth me, where I do safely lie;
And after leads me to the streams which run most pleasantly."

Meanwhile, Mr. Grey had returned from Oxford, and was in the kitchen-garden talking to
his wife, who had brought out another basket of clothes. He was telling her, that he had
seen his sister-in-law off from the railway-station, and was inquiring what had become of
Mavis, when the little girl appeared at the garden gate.

"Why, she's singing!" he exclaimed, in surprise.

"I don't understand her," his wife replied, looking puzzled. "She seemed very cut up after
her mother had gone, and wept most bitterly, but I think her feelings must be all on the
surface—they can't go very deep."

At that moment Mavis caught sight of them. Her voice suddenly ceased, and she ran up to
her uncle, to learn that her mother had really gone.

"What was that you were singing, Mavis?" he inquired, curiously.

"The twenty-third psalm, Uncle John," she answered; "seeing the green meadows put me
in mind of it."

Then, observing he looked bewildered, she continued eagerly, "Don't you understand? 'He
maketh me to lie down in green pastures.'"

"Oh, now I follow your train of thought," he replied, with a smile, glancing at his wife.
"Fancy a child like you thinking of that!"
CHAPTER V
PETTY JEALOUSY

THE morning following Mrs. Grey's departure from the Mill House, Mavis accompanied Rose
to school. Their way led past the church and the vicarage, and through the village. And as
they went, Mavis looked about her with interested eyes, admiring the picturesque creeper-
covered cottages with their trim gardens and thatched roofs.

"They seem to sell everything here," she remarked, with an amused smile, as she paused
before the one shop of the place, which was also the post-office, "groceries, brushes,
notepaper, and medicines too, I declare!"

"We must not dawdle," said Rose, as her cousin lingered, peering into the shop window,
"or we shall be late, and that won't do.'

"Would Miss Matthews be angry?" inquired Mavis. "Is she very strict?"

"Yes," nodded Rose; "mother says it's right she should be. If we were late, she would keep
us in after twelve o'clock, and most likely give us each an imposition—though perhaps you
would go unpunished, as it is your first day. You have never been to school before, have
you?"

"Never. Mother taught me to read and write. And then a young lady, a governess who
lodged at the same house that we did, used to teach me in the evenings. How many girls
are there at Miss Matthews' school?"

"About a dozen—most of them are boarders. Here we are. You see it takes us quite a
quarter of an hour to walk to school."

Miss Matthews' house was at the far end of the village. It was a modern red-brick villa with
bow windows, over the under-blinds of one of which Mavis saw the heads of several girls.
Rose led the way into the house by a side door. And, having shown her cousin where to
leave her hat and jacket on one of a row of pegs in the passage, piloted her to the
schoolroom, and introduced her to her schoolfellows. A few minutes later, Miss Matthews
herself appeared upon the scene, followed by a young governess called Miss Forbes.

"So this is my new pupil," observed Miss Matthews, her eyes scrutinizing Mavis very kindly
as she shook hands with her. "You are called Mavis, I hear," she proceeded; "it is an
uncommon name and a very pretty one."

Miss Matthews was a dark, middle-aged woman with a plain, clever face, and Mavis' first
impression of her was that she was very ugly, but the moment a smile lit up her
countenance, she decided that she was really quite good-looking.

By-and-by, the new pupil was handed over to the governess, who classed her with several
little girls varying from ten to twelve years of age. And so her school life began.

It was soon discovered that though Mavis' education had been carried on in a somewhat
desultory fashion, she was by no means backward. She owned an excellent memory, and
was quick to learn, taking after her father, as her uncle remarked when Rose told him how
easily her cousin mastered her lessons.
"She has inherited Rupert's clever brain," he said to his wife. "You can look in her face and
see she's as sharp as a needle. I hope she'll brisk up our Rose, who's one of the slow,
plodding sort—like myself," he added, with a laugh.

This speech did not please his hearer, though she recognized its truth. Rose had many
excellent qualities, but she was not in the least clever, as far as book-learning was
concerned, and found lessons great drudgery.

Mavis was very soon on the best of terms with all the inmates of the Mill House, with the
exception of her aunt, whom she found it impossible to like as well as the others, though
she could not have told why she did not feel at home with her, if she had been asked for a
reason. The fact was, Mrs. John failed to understand Mavis, who was naturally of a light-
hearted, joyous disposition, and she was confirmed in her impression, as time went on,
that the child's nature was a superficial one. When, on the morning subsequent to the day
on which the 'Nineveh' had sailed from Plymouth, Mavis had received a farewell letter from
her mother, over which she had shed tears, she had had her aunt's full sympathy. But
when, a few hours later, she had returned from school with Rose, apparently in good
spirits, her aunt had privately dubbed her a heartless little thing, being quite unaware of
the brave fight the child had made against depression.

"I advise you not to make too much of Mavis," Mrs. John remarked to her husband on one
occasion, after he had taken the little girl for a drive. "You will spoil her if you're not
careful."

"Oh, nonsense, my dear," he replied; "there's small danger of my doing that. She's had
few pleasures in her life, poor child, and our young folks have had a great many. Bob
wanted to accompany me to-day—he said it was his turn—but I told him he must give up
his place in the gig to his cousin."

"That was hard on the boy, John."

"Not at all. I don't see it."

But Bob himself considered that it had been very hard, for he was unaccustomed to self-
sacrifice, and he liked nothing so well as driving with his father. So when, after tea, Mavis
commenced telling him and Rose of the delightful time she had had, he listened in
somewhat sullen silence.

"It was so kind of Uncle John to take me to Oxford," Mavis said happily. "I think it is such a
lovely place, with those beautiful virginian creepers growing all over the colleges."

"The leaves will soon be off the creepers after the first frost," remarked Rose. "I'm glad
you've seen them, Mavis. Some people think Oxford prettier in the autumn than at any
time, but I like it in the spring, when the hawthorn and lilacs and laburnums are in flower."

"Uncle put up the horse at an inn, and took me to see T—, that was father's college, you
know, and he pointed out the rooms that were father's once, and I saw the chapel and the
lime-walk, and he told me such a lot about father, how clever he was, and that he won
scholarships, and in that way more than half paid for his own education. Oh, how I wish
mother could have been with me to-day!"

"You'll be able to tell her all about it some time," said Rose, as she noticed a shade of
sadness cross her cousin's face.

"Oh yes; but not for a long, long time. How did you and Bob spend the afternoon?" Mavis
asked, glancing from the sister to the brother.
"We went blackberry gathering," answered Rose.

"It was slow work," observed Bob, joining in the conversation at last. "You had the best of
it, Mavis," he added, grudgingly.

"Indeed I think I did," Mavis agreed, with a smile. Then, becoming aware by the
expression of Bob's face that he was displeased, she inquired, "Would you have liked to
have gone to Oxford instead of me this afternoon?"

"Rather!" he exclaimed. "Father took you for a drive last Saturday, too!" he reminded her,
in a way which showed he resented the fact.

"Bob!" exclaimed Rose, in an expostulating tone.

"Well, it's not fair that father should make more of Mavis than of us," he grumbled, "I
know mother thinks so too."

"For shame!" cried Rose, her kind heart touched by the hurt expression on her cousin's
face. "Don't take any notice of what he says, Mavis."

"I didn't know he wanted to go this afternoon," murmured Mavis, looking distressed, "but I
suppose one of you would have gone, if I had not been here. I—I am very sorry."

"There is nothing to be sorry about," Rose replied. "There, I believe you've made her cry,"
she said, turning upon her brother with indignation, as Mavis slipped quietly out of the
room. "I saw tears in her eyes. You are very selfish, and you ought to be ashamed of
yourself, Bob. If father makes much of Mavis, it's only because he wants her to be happy
with us. It's so sad that her mother should have had to go away and leave her."

"I didn't mean to make her cry," Bob answered. "I call her a great baby!"

"She's quite a little girl, of course," said Rose, who, being two years her cousin's senior,
felt almost grown up in comparison to her, "but I don't consider her at all babyish. See how
little fuss she made when her mother left!"

"I don't believe she cared—not much, heard mother say so to Jane; she said Mavis was
singing a few hours after Aunt Margaret had gone, and people don't sing if they're sad."

Rose did not argue the point. Instead, she went upstairs in search of Mavis, whom she
found in her bedroom, sitting on a chair by the window in the dark.

"What are you doing, Mavis?" she asked.

"Nothing," Mavis answered, "only thinking, and—and wishing that mother had left me in
London with Miss Tompkins. I never guessed Bob wanted to go this afternoon, I never
thought that I was taking his place!"

"It was very selfish and unkind of him to speak as he did, and father would be very angry
if he knew he had done so. Think no more of it, Mavis. You haven't been crying, have
you?"

"No," Mavis answered. "I've been praying," she added, after a minute's pause.

"Praying?" Rose was surprised. "But it isn't bedtime," she said, "I always say my prayers
night and morning, don't you?"
"Yes, and odd times besides, whenever I feel I want to. It—it comforts me. It's so nice to
think Jesus is always near to hear one, isn't it, and to remember He understands what
other people can't? I expect Bob thinks it was very selfish of me to go to Oxford with Uncle
John—"

"I shall tell father how he spoke to you!" Rose broke in, impetuously.

"Oh, don't, please don't!" implored Mavis. "Don't let us say any more about it. Promise you
won't."

Rose gave the required promise, and the two little girls went downstairs together. Bob,
who was now ashamed of the jealous spirit he had exhibited, found an opportunity during
the evening of telling Mavis he was sorry if he had seemed unkind, and that he was really
glad that she had enjoyed the afternoon.

September was nearly out now, but the fine weather continued, so that the young folks
were able to spend their spare time out-of-doors. They had several blackberrying
expeditions, from which they returned laden with luscious fruit, which Jane converted into
bramble jelly. Mavis soon knew the prettiest walks around W—, and learnt the dangerous
places in the river, where the water was deep and swift.

Sometimes in their rambles, the children came upon Mr. Moseley, who generally stopped
and talked to them. Rose and Bob, like many country children, were shy, and had little to
say for themselves. But Mavis, on the contrary, was always ready to further a
conversation.

"You should have heard Mavis chattering to the Vicar this afternoon," said Bob to his
mother, one Saturday evening. "I should think she talked to him for quite half an hour."

"Oh, quite!" agreed Mavis.

"I hope he did not think you a forward little girl, Mavis," said her aunt gravely, with a note
of rebuke in her voice.

Mavis coloured indignantly, and a quick retort rose to her lips, but she refrained from
uttering it, and kept silence.

"Mr. Moseley asked Mavis what she thought of Oxford, and she told him," explained Rose.
"And then he questioned her about Aunt Margaret, and when the 'Nineveh' was due to
arrive at Sydney, and of course, she had to answer him."

"Yes," assented her mother, "that was quite right. But little girls must not be too ready
with their tongues."

"Rose isn't," said Bob, with a mischievous glance at his sister. "She scarcely spoke a word
to Mr. Moseley this afternoon."

"I-I don't know Mr. Moseley very well," stammered poor Rose, "and I was shy, I suppose.
So were you, Bob, for that matter."

"Bob is three years your junior, Rose," said her mother. "At your age, you ought not to be
shy. Why could you not talk to the Vicar as well as Mavis?"

Rose hung her head and made no response, whilst her cousin felt very uncomfortable.
Mavis was fully conscious, by this time, that her aunt did not approve of her, that she
regarded her with critical eyes, and that she was always displeased if any one noticed her
more than her cousins. And these facts prevented her from being as happy as she
otherwise would have been at the Mill House. She was never quite at her ease in her
aunt's presence, and certainly never at her best. And yet, Mrs. John had no intention of
being otherwise than just and kind to her little niece, and was vexed when she observed
that Mavis' affection for her uncle was deepening day by day, whilst she held more and
more aloof from herself.

"Aunt Lizzie doesn't like me," Mavis thought frequently, and she would wonder if she could
have possibly done anything to evoke her aunt's displeasure. "I try to please her, but I see
she doesn't care for me, and I'm afraid I don't care for her—much."

CHAPTER VI
ROSE IN TROUBLE

AUTUMN had given place to winter, a wet, depressing winter with rain and westerly gales,
and the flat country between W— and Oxford was flooded. There had been almost
incessant rain for weeks now. And Mavis, as she sat at the little table by her bedroom
window engaged in writing a letter to her mother, which she was taking great pains to spell
correctly, considering every word, glanced at the leaden sky every now and again, in the
hope of seeing a break in the clouds.

"I don't think we ever had such bad weather in London," she reflected. "But perhaps I
didn't notice it so much there. What will happen, I wonder, if the floods go on increasing?
We seem almost surrounded by water as it is."

By-and-by, Rose came to the door, wanting to know if her cousin had nearly finished her
letter.

"Yes," Mavis replied, "I'm ending up now. Come in, Rosie."

So Rose came in. The two little girls had become very friendly by this time, for, though
there was a difference of two years in their ages, in many ways Mavis appeared as old if
not older than her cousin, no doubt because she had always been to a great extent in her
mother's confidence. Rose had lived her twelve years in a home where she had had every
comfort. Whilst Mavis had known times when she and her mother could not have told from
what source the wherewithal was to come to provide them with the necessaries of life, and
yet God had never allowed them to want, He had given them always sufficient for their
needs.

"What is the matter, Rose?" said Mavis, as, having put away her writing materials, she
turned her attention to her cousin, who stood at the window with an expression of gloom
on her face.

"Nothing more than usual," Rose answered, in a tone which implied that she generally had
much to bear. "Mother's been scolding me," she proceeded, as Mavis continued to look at
her inquiringly. "She says she's most dissatisfied with the progress I'm making at school,
that if I'm not careful you'll soon get ahead of me, and—and I can't help it, if you do. I try
to learn, Mavis, but I'm so slow, and—oh, you mustn't think that I'm jealous of you, for I'm
not!"

"Of course I don't think that, Rosie," Mavis replied, greatly distressed at the sight of her
cousin's tearful eyes and quivering lips. "It wouldn't be true, if you did. I told mother that
Miss Matthews said it was quite likely you would be raised into my class next term. I
thought she'd be pleased; but, instead, she was angry with me, and called me a dunce.
Perhaps I am a dunce," Rose admitted, with a sob. "If I am, I can't help it."

Mavis did not know what to say. She was aware that Rose learnt with difficulty, and that
her mother was frequently impatient with her for being so slow, which seemed to her very
unkind. And she had looked forward to being in the same class as her cousin, because she
thought it would be pleasant for them to do their lessons together.

"Miss Matthews knows that I try to get on," Rose continued, in the same aggrieved tone.
"She never complains of me, and I don't consider mother ought to have scolded me, just
because I'm not so sharp as other girls. She doesn't worry Bob about his lessons like she
does me. Bob's her favourite, and he can do nothing amiss. I declare I won't try to learn
any more, for mother's sure to find fault with me, anyway! It's most unjust."

"I don't suppose Aunt Lizzie understands how hard you try to learn," Mavis said, putting
her arms around Rose and kissing her flushed cheek. "Don't be unhappy about it, dear. You
do your best, I'm sure."

"That's why I feel it's so hard mother should be cross with me, Mavis. I don't idle my time
away, like some of the girls at school do, and—and she says I must, or I shouldn't be so
backward." Rose brushed away an angry tear, and choked back a sob. "Let us talk of
something else," she said. "You've been writing to Aunt Margaret, haven't you?"

"Yes. Aunt Lizzie said I might enclose a letter with one she has written. I wonder how long
this rainy weather will last, Rosie."

"Father thinks we shall have a change soon, for the wind is getting more northerly; it's
been due west for weeks. If we get frost now the floods are out, we shall have fine
skating; you will like that, Mavis?"

"I can't skate," Mavis answered. "I never tried."

"Oh, we will soon teach you. I am looking forward to a long spell of frost, like we had last
winter."

"Are you? We thought that frost was dreadful in London, because it made things so hard
for the poor—they don't feel the wet so much as long as it's mild, but when it's cold and
frosty, the distress is terrible. Last winter, not far from where we lived, a poor old woman
was found dead on a doorstep; I couldn't sleep for nights afterwards for thinking of her."

"How shocking!" exclaimed Rose, in an awe-struck voice. "Had she no home?"

"No. There are hundreds and hundreds of people in London without homes. Mother knew a
great many poor people, and it used to make her so sad when she couldn't help them.
Often they'd come and tell her their troubles, because, you see, being poor herself, she
could understand better than if she had been rich," Mavis explained, with a wisdom beyond
her years.

"Were you, then, very poor, Mavis?" Rose inquired, wonderingly.


"Yes," nodded Mavis; "but we always had enough to eat, though sometimes it was only
bread-and-butter. Once we couldn't pay our rent, and mother was in great trouble about
that, but Miss Tompkins was very kind, and said she would willingly wait for it. And then,
mother had a good engagement to nurse a rich old gentleman for a few weeks, so Miss
Tompkins hadn't to wait long."

"Why didn't your mother write to father for some money, Mavis?" asked Rose. "I am sure
he would have been very pleased to send her some."

"I am sure he would, too, now I know him. I don't know why she didn't write to him;
perhaps she did not like to bother him, yes, that must have been it."

"Father says Aunt Margaret has been a wonderful woman to do as she has done, with no
one to help her," remarked Rose.

"God helped her," said Mavis, simply; "mother says He helps all who trust in Him."

"You tell Him all your troubles, don't you, Mavis?" Rose asked. Then, as her cousin nodded
assent, she said, "I don't mean only great troubles, but little ones?"

"Oh yes."

"Well, I don't. I just say my regular prayers twice a day, and that's all. I don't feel God's
my Friend, like you do."

"Don't you? Oh, but you should. I'd tell Him everything, if I were you."

"What, that mother's vexed with me for being slow at school, for instance?"

"Yes. God knows you do your best, Rosie, if Aunt Lizzie doesn't. I'd ask Him to make her
believe it, if I were you. Mother says when we've told God our trouble, we oughtn't to
worry about it any more, but just leave it to Him, and He'll be sure to put it right."

"Do you really think that?"

"Of course I do. He has promised to bear our troubles, and you know it says in the Bible,
'there has not failed one word of all His good promise.' I'll show you the verse, if you like."

"It astonishes me that you should be so religious," Rose observed, after a few minutes'
thought, "because you're such a merry little soul as a rule, always singing about the place
and ready for any fun, and I thought religious people were generally very solemn."

"Oh, do you think so? Mother says religion ought to make people joyous and happy, and
that it's mistrusting God to be gloomy and sad. That's why I've tried not to trouble about
her leaving me; but sometimes I haven't been able to help crying when I've thought how
far she's gone away, and then I've felt so bad about it afterwards."

At that moment heavy footsteps, easily recognizable as Mr. Grey's, were heard ascending
the stairs, and a minute later came a knock at the door, and a voice outside called—

"Mavis, I've news for you, my dear."

"News?" Mavis sprang to the door and flung it open wide. "Oh, Uncle John," she cried, as
she saw her uncle standing smiling at her, "do you mean that you have news of mother?
But no, it cannot be that!"
"Yes, it is, child," Mr. Grey responded, "and good news, too. The 'Nineveh' has arrived at
Sydney. Your mother's in first-rate health, and Miss Dawson is better. I've had a letter from
Mr. Dawson, and he has had a cablegram from your mother. It seems, he promised her he
would let us know as soon as he heard from her. It is very good news, isn't it?"

"Splendid!" cried Mavis, her face aglow with happiness. "How kind of Mr. Dawson to write
to you, Uncle John! I never guessed we should get news so soon, did you?"

"Well, I thought it just possible," Mr. Grey admitted, "but I didn't say a word about it, for
fear you should be disappointed. Mr. Dawson is evidently in high spirits, judging from the
tone of his letter. And I don't believe your mother would raise his hopes about his
daughter, if she had not good cause for doing so. The change of climate may really set up
the poor young lady's health, after all, and I sincerely hope God may spare her life. Her
father has my sympathy. I know how grieved I should feel if I had to send my little girl
away from me, especially if she was ill."

And Mr. Grey smiled affectionately at Rose as he spoke.

She threw her arms around his neck and hugged and kissed him again and again,
exclaiming the while—

"Oh, you dear father! You do love me, don't you, just as much as though I was clever?"

"Bless the child, yes," he replied, with his hearty laugh. "Why, Rosie, there are tears in
your eyes. What's the meaning of that?"

She would not tell him, however, and he was wiser than to press the question.

Meanwhile, Mavis had gone in search of the other members of the household to impart her
news to them. Her heart was singing with joy, and her aunt thought she had never seen a
brighter, happier pair of eyes than those which peeped around the kitchen door to see if
she was there.

"Such good news, Aunt Lizzie!" cried Mavis, and she proceeded to explain what it was.

"Thank God they have made the journey in safety!" exclaimed Jane, who had stopped in
the midst of the important business of stove-cleaning when the little girl had appeared.

"I do thank Him," Mavis answered, softly.

"How relieved in mind Mr. Dawson must be!" exclaimed Mrs. John. She was rubbing the
contents of the plate-basket with a piece of chamois leather, her usual task on a Saturday
afternoon. "Poor man, I sympathize with him greatly," she continued, "especially as Miss
Dawson is his only child. How grateful he will be to your mother, Mavis, if she brings him
back her patient restored to health!"

"Mother will take good care of her," Mavis responded. "She's a capital nurse, every one
who knows her says that; but only God can make Miss Dawson well."

"That's so," agreed Jane, with a nod, as she returned to her stove. "I don't believe that
child ever forgets that God's above all," she remarked, as Mavis left the kitchen to look for
Bob, to impart her news to him. "It's to be hoped she'll always remember it."

"She's a strange little thing," her mistress answered, "so very childish in some ways, and
in others thoughtful beyond her years."
CHAPTER VII
A GREAT GIFT

THE heavy rains had ceased, and there had been several nights of hard frost, which had
covered the flooded meadows surrounding W— with a thick coating of ice.

"The ice will bear to-day," remarked Mr. Grey, one morning at breakfast. "I believe we're in
for a spell of dry weather. You must look to your skates, children, for, if all's well, you'll get
some skating now."

By the following morning—a Saturday—the ice was in splendid condition, and the young
people of the village spent nearly the whole day on it, as well as many of their elders.

Mavis had been supplied with an old pair of skates which had belonged to Rose, who had
bought a new pair the previous season. And during the morning, her uncle gave her, her
first lesson in skating; but he was called away on business in the afternoon, and she was
left to her own resources. She got on by herself fairly well, and managed to keep her feet
unaided; but it was slow work, and she grew tired and cold long before her cousins were
ready to leave the ice. By-and-by she divested herself of her skates, and declared to Rose
her intention of going home.

SHE WAS LEFT TO HER OWN RESOURCES.


"What, already?" cried Rose. "Why, how cold you look! I'm most beautifully warm. You
don't want me to go with you, do you?" she asked.

"Oh no," Mavis replied, "certainly not. I don't think I've got on badly. But I can't skate fast
enough to keep warm like you, and my feet are so cold, there's no feeling in them.
Otherwise I should like to stay and look on."

Rose nodded, and skimmed away over the ice. Whilst Mavis left the meadow by the
gateway, and turned into the road leading to the village, through which she had to pass on
her way home, walking briskly to get herself warm.

It was a perfect winter afternoon. The sun was sinking rosily in the western sky, and the
keen, frosty air was most invigorating. Mavis had enjoyed the day; but she sighed, and her
pretty face grow grave as she thought of those to whom frost meant only added misery,
and she felt glad that there were no extremely poor people in W—.

"I don't suppose there's any one in the place who hasn't enough to eat," she reflected.
"For I heard uncle say last night that the villagers were very well looked after; they get
coal-tickets, and they belong to blanket clubs, and they have good homes."

She had reached the village by this time, and was passing the post-office when the Vicar
came out, followed by his dog. A smile lit up Mr. Moseley's kindly countenance as his eyes
fell on the little girl.

"All alone?" he said. "How is that?"

"I got so very cold on the ice that I thought I'd go home," Mavis explained, as she patted
Max. "I only began to learn to skate to-day, and I grew very tired."

"Naturally. You are warmer now?"

"Oh yes, thank you. I have been walking fast. Max is very pleased to see me, isn't he?"

"Very. He counts you as a friend, there is no doubt about that. I am glad to hear you have
had good news of your mother. I saw your uncle a few days ago, and he mentioned that
Mrs. Grey had arrived at Sydney safely."

"Oh yes. I shall be having a letter from her from Sydney before Christmas, I expect. Did
uncle tell you that Miss Dawson is better, Mr. Moseley?"

"Miss Dawson is the young lady your mother is nursing, I suppose? No, your uncle did not
tell me that; but I am very glad."

"She is such a pretty young lady, and so rich; but she is very delicate, though you wouldn't
think it to look at her. I never saw her but once, and then she was very kind to me. She
gave me a beautiful gold locket and chain for a keepsake. I had nothing of the kind before,
and we got quite friendly, though we were only together for a little while. Isn't it odd how
quickly one gets friendly with some people?"

He smiled and assented. They were walking in the direction of the vicarage, which Mavis
had to pass on her way home.

"I think it's very strange," the little girl continued, knitting her brows thoughtfully. "Now,
Rose and I are great friends, and, on the whole, I get on well with Bob, and I'm very fond
of Uncle John; but, do you know—" she dropped her voice confidentially as she spoke—"I
never can quite like Aunt Lizzie. I do hope it isn't very wrong of me."
"Why can't you like her?" the Vicar asked, looking surprised.

"I don't know," Mavis admitted, shaking her head.

"She's kind to you, I'm sure."

"Oh yes, yes!" The little girl grew red, and hesitated. "Please, Mr. Moseley, what is it to be
superficial?" she asked, by-and-by. "Is it something one ought not to be?"

"To be superficial is to be all on the surface—shallow," he replied. "But why do you ask
that?"

"Aunt Lizzie says I'm superficial," Mavis explained. "I heard her tell Uncle John so. But he
said no, I was not. Indeed, I was not trying to listen," she proceeded quickly. "I was
coming downstairs, and they were in the hall. I didn't know they were talking about me.
Then it's nothing very bad if one is superficial, Mr. Moseley?"

"No," he answered, with an involuntary smile; "and it's nothing for you to trouble about.
But I agree with your uncle. I think, perhaps, your aunt is mistaken; she probably does not
understand you, and you evidently do not understand her. No doubt you will get to know
each other better by-and-by. I am coming to see your aunt about you one day soon."

"About me?" Mavis exclaimed, questioningly.

"Yes. I am going to get up a concert—not just yet, during Christmas week—and invite all
the villagers to attend. It will be held in the schoolroom, and I think you can help me, if
your aunt will permit it."

"I!" cried the little girl in amazement. "What can I do?"

"You can sing. I have heard you on several occasions when you have been with your
cousins in the woods, though you have not known I have been listening. Once I heard you
sing a most beautiful version of the twenty-third psalm, and that is what I should like you
to sing at my concert."

"Oh, Mr. Moseley, I don't think I could—before a lot of strange people!"

"Not if it gave them pleasure?" he inquired, with a smile.

"I should be so nervous," faltered the little girl.

"Perhaps, at first, but you would very soon get over that."

"But I have never learnt to sing properly—not with music, I mean. I couldn't sing with a
piano; it would put me out."

"I should like you to sing without an accompaniment, as you have been accustomed. God
has given you a groat gift, my dear, don't you think He expects you to use it for the benefit
of others?"

"Do you think that?" Mavis asked earnestly.

"Certainly I do. The poet Longfellow says that God sent His singers upon earth—"

"'That they might touch the hearts of men,


And bring them back to heaven again.'"

"Those lines recurred to my memory when I heard you sing that beautiful, comforting
psalm."

Mavis' face broke into a sudden, radiant smile. In imagination, she heard Miss Dawson's
well-remembered voice saying: "You have given me comfort, and reminded me that I am
not setting out on a long journey without support from God." She knew the sick girl had
referred to the words she had previously sung—

"The Lord is only my support, and He that doth me feed;


How can I then lack anything whereof I stand in need?"

Had God indeed given her a great gift, expecting her to use it for the benefit of others?
She had never thought of her voice in that light before; she had sung instinctively, like the
bird after which she had been named.

"Mr. Moseley, I will sing at your concert gladly, if Aunt Lizzie will let me," she said, at
length.

"Thank you, my dear. I felt sure you would. Well, I shall call at the Mill House shortly. You
will soon run home from here?"

"Oh yes," assented Mavis.

They had reached the vicarage gate, and, having shaken hands with her companion, and
put her arms around Max's neck and given him an affectionate hug, she hastened on. She
felt very light-hearted, and hummed a little tune happily to herself as she tripped along.
But her voice suddenly ceased as she neared the Mill House and caught sight of a man's
figure ahead of her, clad in a ragged suit of clothes. A pang of pity shot through her
sympathetic heart.

"I suppose he's a tramp or a beggar," she thought, "he looks dreadfully poor."

The man turned at the sound of her light footsteps, and looked at her. She saw his face
was pinched and blue with the cold, and that it wore a very wretched, dispirited
expression. As she caught up to him, he spoke.

"Have you a penny you could spare me, missie?" he said, in a voice which sounded weak,
she thought. He was quite a young man, tall and broad-shouldered, but extremely thin.

"No, I haven't," she replied, regretfully. "Oh, I'm so sorry! You do look miserably cold."

"Aye, I'm cold," he agreed, with a short, bitter laugh, "cold and hungry, too."

"Hungry? Oh dear, how dreadful! Do you live here—at W— I mean?"

"No. I'm on the look-out for work—have been for weeks—but it's very scarce. I'm not a
beggar from choice. I've been hanging around the mill in hopes of seeing the miller,
thinking that he might give me a job. They told me in the village that he wanted a man to
drive a waggon."
"I believe he does," said Mavis. "But he's not at home this afternoon. I live at the Mill
House; Mr. Grey's my uncle. Go round to the back door, I'm sure my aunt will give you
something to eat."

The man looked at her doubtfully, but he did as he was told.

Meanwhile Mavis passed through the wicket-gate, ran up the garden path, and entered the
house by the front door. She found her aunt in the parlour, engaged in darning stockings,
and immediately informed her that there was a poor starving man outside.

"Do give him some food, please, Aunt Lizzie, and let him warm himself by the kitchen fire,"
she said, pleadingly. "He wants to see Uncle John, to ask him if he can give him work.
Mayn't he come in and wait? I have sent him around to the back door."

"Really, Mavis, you take too much upon yourself!" cried her aunt, irascibly. "I never
encourage tramps; the workhouse is always open to them. I must send this man off at
once."

"Aunt Lizzie, you don't mean it!" exclaimed Mavis, aghast. "Oh, you won't be so cruel! He
is hungry, I am certain he is, and, oh, it will be unkind if you don't give him something to
eat—if only a slice of bread!"

The little girl repented having spoken so hastily the moment after the words had passed
her lips, and she hung her head and commenced a stammering apology. Her aunt did not
stay to listen to it, however, but hurried to the kitchen. The man was already at the back
door, and Jane was speaking to him.

"I dare say missus will give you a bit of bread and meat," she was remarking, as Mrs.
John, closely followed by Mavis, entered the kitchen.

"How often have I warned you not to encourage tramps, Jane!" said her mistress, severely.
"Go away, or I'll report you to the police for begging," she declared, imperatively motioning
to the man to depart.

"The little lady thought you'd give me something to eat, ma'am," he said. "She told me to
come."

"Yes, I did," asserted Mavis, nearly weeping. "Oh, Aunt Lizzie, don't, don't send him away
hungry."

Mrs. John wavered. She looked scrutinizingly at the man, and saw he was evidently
wretchedly cold and inadequately clothed, and her heart was stirred with pity. So she went
into the larder and cut some bread and meat, which she gave him.

"Thank you, ma'am," he said, his tone evincing real gratitude. But though he addressed
the donor of the food, his glance went past her to Mavis. He was not allowed to say more,
however, for at that moment the door was shut in his face, and he had no choice but to go
away.

"I am never knowingly unkind to any one, Mavis," Mrs. John said, as she turned her
attention to her little niece, who was furtively wiping her eyes, "and I am greatly
astonished that you should have spoken to me in such an unbecoming manner."

"It was very wrong of me, and I'll never do so again, Aunt Lizzie," Mavis responded, in a
tremulous voice. "But you didn't seem to understand that the poor man was really hungry.
Won't you forgive me? I did not mean to be rude. I spoke without thinking."
"Yes, I forgive you. But never presume to dictate to me again. Why have you returned
before the others?"

Mavis explained, and went on to repeat the conversation she had held with the Vicar, to
which her aunt listened with an expression of disapproval on her countenance, afterwards
remarking—

"You are too young, in my opinion, to sing in public. However, I will hear what Mr. Moseley
has to say, and consult your uncle upon the matter. The idea of a child like you singing at a
concert. It is preposterous to think of it!"

Mavis made no answer, for she saw Mrs. John was greatly displeased. She thought it was
because of the manner in which she had addressed her in reference to the hungry
stranger, and was quite unaware that, added to that, her aunt was jealous on her
children's account. Truth to tell, Mrs. John was annoyed that the Vicar should seek to bring
Mavis into notice. She would not admit, even to herself, that the little girl had a
wonderfully beautiful voice.

CHAPTER VIII
LOOKING FORWARD TO CHRISTMAS

THE miller's thrifty wife was not by any means a hard woman, but she lacked that quick
sympathy for others which is generally the outcome of a vivid imagination joined to a
kindly heart. She never realized the sorrows of her fellow-creatures without they were set
plainly before her. And though she was in many ways much shrewder than her husband,
she often made mistakes of which he would have been incapable.

She had been a worker all her life, and consequently entertained a great contempt for
idlers. And she believed that if people worked they always got on well, for that had been
her own experience. She had lived rather a narrow life, with few interests outside her own
family.

Had it been otherwise, doubtless she would have known that sometimes God denies
success—as the world counts success—to those who do their best and work their hardest.
Thus it was, that rarely did any one come to her for help or sympathy, whilst many were
the tales of woe which were poured into her husband's ears.

When Mavis had hastened to her aunt requesting food for a hungry man, Mrs. John had
immediately jumped to the conclusion that he must be a professional beggar, and
therefore a dangerous character. Her eyes had shown her the real misery of his condition,
however, so she had fed him. But she had not been possessed of sufficient discernment to
notice that he was not an ordinary mendicant, so that when her husband informed her, a
few days later, at dinner-time, that he had engaged the man she had so unwillingly
assisted, to drive one of his waggons, she was greatly astonished.

"You cannot mean it!" she explained. "Why, he was literally in rags! John, surely you are
very unwise."
"That remains to be proved, my dear," responded her husband, gravely, "but I hope I am
not. The man is accustomed to horses; he has been in the employ of a farmer living near
Woodstock, and I see no reason why he should not suit me, if he keeps the promises he
has made me to be honest and steady. His name is Richard Butt, and he's twenty-five
years of age, and has a young wife, who is at present living with her parents at
Woodstock."

"Then he doesn't support her? He has been out of work some time, I suppose? Why did he
leave his last place?"

"Well, he got himself into trouble, my dear; he was very frank about it, and I have made
inquiries, and find he told me the truth. Remember, children," Mr. Grey proceeded,
addressing the three young folks, who were present and listening with great interest, "this
is not to go beyond our own household, you are not to speak of it to outsiders."

"We will not!" they agreed eagerly.

"Well, he was caught poaching, and sent to prison for six weeks," Mr. Grey went on to
explain. "No doubt he deserved his punishment. Of course, when he was released from
jail, he found his master had filled his place and had no work for him. And his young wife,
unable to pay the rent of their cottage, had been obliged to give it up and return to her
own people. For several weeks now, he has been tramping the district for miles around in
search of employment, without any success, ashamed to return to Woodstock, where he is
well-known, to be a disgrace to his relations."

"On Saturday, he heard I wanted a waggoner, so he waited about the place till he could
see me, which was on Monday. I believe he slept two nights in the cattle shed in Brimley
meadow, and I'm certain he's been half starved."

"Oh, how terrible!" cried Rose.

Whilst Mavis, a little paler than usual, glanced at her aunt, who was listening with an
impatient frown on her forehead.

"This cold, frosty weather, too!" exclaimed Bob. "To think of us all with plenty to eat, a fire
to warm ourselves by, and comfortable beds to lie on, and some one close to our house in
that old tumble-down shed!"

"You must look over my stock of clothes, Lizzie," said the miller, "and see if I can't spare
the poor fellow a suit. You don't approve of my having engaged him to work for me, I see."

"How can I approve of your befriending a man of that class? Do you expect a poacher to
be honest? He'll rob you for a certainty."

"I trust not; but if he does, I shall get rid of him at once. And at any rate, I shall have
given him a chance to redeem his character. I've written to his late master, who informs
me that Richard Butt is a strong, willing young fellow, and that he believes he took to
poaching for the love of sport. I don't know about that, I'm sure; but I don't fancy he'll
attempt anything of that kind again. Mind you, I'm not making excuses for him. As I've
told him, a man who poaches a rabbit is as much a thief as a man who robs a poultry-
yard, the principle is the same. But I can't help being sorry for him, and I wouldn't have it
on my conscience for anything that I might have assisted a fellow-creature and hadn't
done it. It's the right thing, I take it, to give a helping hand where one can."

"Perhaps God sent him to you on purpose, Uncle John," said Mavis; "because you could
give him work, I mean."
"May be so, my dear," agreed the miller.

"When does he commence to work for you?" inquired his wife, still looking dubious.

"Next Monday. He has found lodgings in the village."

"How has he managed about money?"

"Well, I have advanced him a little," Mr. Grey admitted. "I offered to do so, and I believe
he is very grateful to me."

"It is to be hoped he will keep faith with you, John," his wife remarked drily.

"You will look-out a suit of clothes for him, won't you, Lizzie?"

"I will, if he turns up on Monday as arranged. But it would not astonish me, if he does not.
We shall see."

Richard Butt did keep faith with the kind man who had befriended him, however. And
proved himself quite equal to the task he had undertaken, to drive the big waggon with its
pair of fine horses.

Mrs. John duly presented him with a suit of her husband's clothes, and various other
articles of clothing which she thought might be useful to him. And she took the opportunity
to question him about his wife, whom he said he was going to send for as soon as he could
make a home for her.

"She's been able to save a few bits of our furniture," he explained, "and if I give Mr. Grey
satisfaction, and find he's willing to keep me on, I shall look-out for a cottage in the
village. Meantime, my wife will stay with her parents, she's no expense to them, for she
earns enough for herself by doing plain needlework."

The children, Mavis especially, took great interest in the new waggoner. And they were
careful not to tell any one the circumstances which had brought him to the deplorable
position he had been in when Mr. Grey had taken pity upon him. He had paid the penalty
for his sin, and was starting life afresh.

The severe frost continued for more than a fortnight, and Mavis learnt to skate very nicely.
Many happy hours did she spend with her cousins and her schoolfellows on the ice; and
deeply regretful were the young people when a thaw set in, and weather-wise folks began
to prophesy a mild Christmas.

Rose and Bob were looking forward to Christmas with much eagerness, for the season had
always been a very happy one for them. But Mavis was anticipating the coming festival
with very sober thoughts. Hitherto, she and her mother had been together at Christmas;
now they were divided by thousands and thousands of miles of land and sea. She listened
somewhat half-heartedly to her cousins' plans for making the most of the holidays, until
Rose gave her a look of wonder and reproach, and said—

"What is the matter? You don't seem yourself, Mavis."

"I've been thinking of last Christmas, and that has made me rather sad," Mavis answered.
"Mother and I were together then," she added, the tears rushing to her hazel eyes as she
spoke.

"Oh!" cried Rose, comprehendingly, whilst Bob inquired—


"Did you have a lot of presents?"

"No; I only had two—a work-basket from mother, and a story-book from Miss Tompkins.
But it was a lovely Christmas! Mother and I were so happy together! On Christmas Eve, it
was fine, and we went out and had a good look at the shops. We enjoyed seeing all the
pretty things, and thinking what we would buy, if we were rich. We spent Christmas Day
quite by ourselves. In the morning, we went to the service at a little mission church where
father used to preach sometimes—the people who go there are mostly poor people, some
of them so poor that they wouldn't like to go and sit with those who are well-dressed. And
in the afternoon, after dinner, we sat by the fire and talked, and never, never dreamed that
we should be so far apart from each other when Christmas came again."

And Mavis heaved a deep sigh.

"Oh, you mustn't get sad," said Rose, earnestly. "I'm sure your mother wouldn't like you to
be that. We want you to have a very happy Christmas. We shall break up at school about
the twentieth of December, and then we shall be very busy at home, making mincemeat
and puddings, and preparing our Christmas presents."

The time had arrived now when Mavis might expect her mother's first letter from Sydney.
It was delivered one afternoon whilst she was at school, and given to her immediately on
her return. She ran upstairs to her bedroom to enjoy it undisturbed, and her heart
throbbed with happiness as she read that her mother was well, and that Miss Dawson was
continuing to improve in health.

"You are constantly in my thoughts and prayers, little daughter,"


Mrs. Grey had written, "and ever in my heart. God bless and keep you,
my darling child. I enclose a money-order for a pound, for you to spend
as you like; doubtless you will find it useful at Christmas."

"Oh, how nice!" exclaimed Mavis, delighted beyond measure. "Now I shall be able to give
presents to every one! A whole pound! I never had more than half a crown in my life
before!"

She finished reading her mother's letter, then went back to the beginning, and read it right
through to the end again before she looked into the envelope for the money-order. There it
was safe enough, and a half-sheet of notepaper, on which were written a few lines in an
unfamiliar handwriting—

"DEAR LITTLE MAVIS,"

"This is to wish you a very happy Christmas. I am really better, and I


believe God is going to allow me to get well. Often I have been very low-
spirited and sad since we left England, but when I have thought of the
Good Shepherd, of whom you sang to me so sweetly, I have felt better.
'The Lord is only my support,' dear Mavis, and I am learning to trust in
Him more and more. Your mother is so good to me, so patient, so kind;
we have become great friends."

"I have written to my father asking him to send you a present from
London for me; you may expect to get it a few days before Christmas. It
will be my Christmas-box to you, and please accept it with my love.

You might also like